EXPERT COMMENTARY

Both the Centers for Disease Control and Prevention and the World Health Organization currently recommend that patients with trichomoniasis be treated with a single 2-g oral dose of metronidazole.¹ Following treatment, the reported rates of repeat infection or persistent infection range from 5% to 31%. Repeat infection rates may be even higher in HIV-infected patients.

Repeat infections presumably result from a failure to treat the patient’s sexual partner(s) or from the patient’s exposure to a new partner. Persistent infections, however, may be the result of inadequate primary therapy, even though inherent resistance of the organism to metronidazole is quite rare. To date, no single study has shown that single-dose therapy is inferior to multidose therapy, but most of these studies lack sufficient power to completely exclude the possibility of a type-2 statistical error.² To compare single-dose with multidose therapy for trichomoniasis in a more systematic manner, Howe and Kissinger conducted a meta-analysis, which was recently published in Sexually Transmitted Diseases.

Related article:
2016 Update on infectious disease

Details of the study

The investigators conducted a comprehensive literature search using Embase, Medline, and ClinicalTrials.gov; 6 articles were included in the final results, 4 of which were randomized controlled trials. Approximately 1,300 participants were included in the 6 trials. All of the patients in the single-dose treatment arms received a 2-g oral dose of metronidazole. In the multidose treatment arms for 2 studies the participants received metronidazole 250 mg orally 3 times daily for 7 days, and for 2 studies the dose was 200 mg 3 times daily for 7 days. The fifth study employed a 500-mg oral dose of metronidazole twice daily for 7 days. The final study used a 400-mg oral dose twice daily for 5 days. The key study end point was treatment failure.

Howe and Kissinger demonstrated that women who received the single 2-g dose were 1.87 times (95% CI, 1.23−2.82; P<.01) more likely to experience a treatment failure compared with women who received a multidose regimen. When the one study that focused only on HIV-infected women was excluded from analysis, the results were similar. The relative risk of treatment failure was 1.80 (95% CI, 1.07−3.02; P<.03).

Related article:
Preventing infection after cesarean delivery: Evidence-based guidance

Study limitations

The results of this meta-analysis are interesting and provocative. However, the analysis has several important limitations. Five of the 6 studies were published many years ago (1971, 1972, 1979, 1980, and 1982). The most recent study was published in 2010. The investigators used 4 different multidose regimens, with metronidazole doses ranging from 200 mg to 500 mg and duration of therapy ranging from 5 to 7 days. Four of the six investigations used saline microscopy as the definitive diagnostic test of treatment failure. Compared with culture or DNA testing, microscopy is not as accurate. Moreover, the timing of retesting varied in the studies, and some apparent treatment failures actually may have been due to reinfection. In addition, the studies did not consistently track the adequacy of treatment of the sexual partner.

Share your thoughts! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

EXPERT COMMENTARY

Both the Centers for Disease Control and Prevention and the World Health Organization currently recommend that patients with trichomoniasis be treated with a single 2-g oral dose of metronidazole.¹ Following treatment, the reported rates of repeat infection or persistent infection range from 5% to 31%. Repeat infection rates may be even higher in HIV-infected patients.

Repeat infections presumably result from a failure to treat the patient’s sexual partner(s) or from the patient’s exposure to a new partner. Persistent infections, however, may be the result of inadequate primary therapy, even though inherent resistance of the organism to metronidazole is quite rare. To date, no single study has shown that single-dose therapy is inferior to multidose therapy, but most of these studies lack sufficient power to completely exclude the possibility of a type-2 statistical error.² To compare single-dose with multidose therapy for trichomoniasis in a more systematic manner, Howe and Kissinger conducted a meta-analysis, which was recently published in Sexually Transmitted Diseases.

Related article:
2016 Update on infectious disease

Details of the study

The investigators conducted a comprehensive literature search using Embase, Medline, and ClinicalTrials.gov; 6 articles were included in the final results, 4 of which were randomized controlled trials. Approximately 1,300 participants were included in the 6 trials. All of the patients in the single-dose treatment arms received a 2-g oral dose of metronidazole. In the multidose treatment arms for 2 studies the participants received metronidazole 250 mg orally 3 times daily for 7 days, and for 2 studies the dose was 200 mg 3 times daily for 7 days. The fifth study employed a 500-mg oral dose of metronidazole twice daily for 7 days. The final study used a 400-mg oral dose twice daily for 5 days. The key study end point was treatment failure.

Howe and Kissinger demonstrated that women who received the single 2-g dose were 1.87 times (95% CI, 1.23−2.82; P<.01) more likely to experience a treatment failure compared with women who received a multidose regimen. When the one study that focused only on HIV-infected women was excluded from analysis, the results were similar. The relative risk of treatment failure was 1.80 (95% CI, 1.07−3.02; P<.03).

Related article:
Preventing infection after cesarean delivery: Evidence-based guidance

Study limitations

The results of this meta-analysis are interesting and provocative. However, the analysis has several important limitations. Five of the 6 studies were published many years ago (1971, 1972, 1979, 1980, and 1982). The most recent study was published in 2010. The investigators used 4 different multidose regimens, with metronidazole doses ranging from 200 mg to 500 mg and duration of therapy ranging from 5 to 7 days. Four of the six investigations used saline microscopy as the definitive diagnostic test of treatment failure. Compared with culture or DNA testing, microscopy is not as accurate. Moreover, the timing of retesting varied in the studies, and some apparent treatment failures actually may have been due to reinfection. In addition, the studies did not consistently track the adequacy of treatment of the sexual partner.

Share your thoughts! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

EXPERT COMMENTARY

Both the Centers for Disease Control and Prevention and the World Health Organization currently recommend that patients with trichomoniasis be treated with a single 2-g oral dose of metronidazole.¹ Following treatment, the reported rates of repeat infection or persistent infection range from 5% to 31%. Repeat infection rates may be even higher in HIV-infected patients.

Repeat infections presumably result from a failure to treat the patient’s sexual partner(s) or from the patient’s exposure to a new partner. Persistent infections, however, may be the result of inadequate primary therapy, even though inherent resistance of the organism to metronidazole is quite rare. To date, no single study has shown that single-dose therapy is inferior to multidose therapy, but most of these studies lack sufficient power to completely exclude the possibility of a type-2 statistical error.² To compare single-dose with multidose therapy for trichomoniasis in a more systematic manner, Howe and Kissinger conducted a meta-analysis, which was recently published in Sexually Transmitted Diseases.

Related article:
2016 Update on infectious disease

Details of the study

The investigators conducted a comprehensive literature search using Embase, Medline, and ClinicalTrials.gov; 6 articles were included in the final results, 4 of which were randomized controlled trials. Approximately 1,300 participants were included in the 6 trials. All of the patients in the single-dose treatment arms received a 2-g oral dose of metronidazole. In the multidose treatment arms for 2 studies the participants received metronidazole 250 mg orally 3 times daily for 7 days, and for 2 studies the dose was 200 mg 3 times daily for 7 days. The fifth study employed a 500-mg oral dose of metronidazole twice daily for 7 days. The final study used a 400-mg oral dose twice daily for 5 days. The key study end point was treatment failure.

Howe and Kissinger demonstrated that women who received the single 2-g dose were 1.87 times (95% CI, 1.23−2.82; P<.01) more likely to experience a treatment failure compared with women who received a multidose regimen. When the one study that focused only on HIV-infected women was excluded from analysis, the results were similar. The relative risk of treatment failure was 1.80 (95% CI, 1.07−3.02; P<.03).

Related article:
Preventing infection after cesarean delivery: Evidence-based guidance

Study limitations

The results of this meta-analysis are interesting and provocative. However, the analysis has several important limitations. Five of the 6 studies were published many years ago (1971, 1972, 1979, 1980, and 1982). The most recent study was published in 2010. The investigators used 4 different multidose regimens, with metronidazole doses ranging from 200 mg to 500 mg and duration of therapy ranging from 5 to 7 days. Four of the six investigations used saline microscopy as the definitive diagnostic test of treatment failure. Compared with culture or DNA testing, microscopy is not as accurate. Moreover, the timing of retesting varied in the studies, and some apparent treatment failures actually may have been due to reinfection. In addition, the studies did not consistently track the adequacy of treatment of the sexual partner.

Share your thoughts! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

The proportion of rectal cancer cases diagnosed in people younger than 55 years doubled over the past 2 decades, according to a report published online in the Journal of the National Cancer Institute.

In contrast, the proportion diagnosed in people older than 55 years has decreased over the last 4 decades, said Rebecca L. Siegel, MPH, strategic director of surveillance information services of surveillance and health services research at the American Cancer Society and her associates.

Courtesy Wikimedia Commons/nephron/Creative Commons License

AGA Resource

The AGA Colorectal Cancer Clinical Service Line provides tools to help you become more efficient, understand quality standards and improve the process of care for patients: http://www.gastro.org/patient-care/conditions-diseases/colorectal-cancer

The proportion of rectal cancer cases diagnosed in people younger than 55 years doubled over the past 2 decades, according to a report published online in the Journal of the National Cancer Institute.

In contrast, the proportion diagnosed in people older than 55 years has decreased over the last 4 decades, said Rebecca L. Siegel, MPH, strategic director of surveillance information services of surveillance and health services research at the American Cancer Society and her associates.

Courtesy Wikimedia Commons/nephron/Creative Commons License

AGA Resource

The AGA Colorectal Cancer Clinical Service Line provides tools to help you become more efficient, understand quality standards and improve the process of care for patients: http://www.gastro.org/patient-care/conditions-diseases/colorectal-cancer

The proportion of rectal cancer cases diagnosed in people younger than 55 years doubled over the past 2 decades, according to a report published online in the Journal of the National Cancer Institute.

In contrast, the proportion diagnosed in people older than 55 years has decreased over the last 4 decades, said Rebecca L. Siegel, MPH, strategic director of surveillance information services of surveillance and health services research at the American Cancer Society and her associates.

Courtesy Wikimedia Commons/nephron/Creative Commons License

AGA Resource

The AGA Colorectal Cancer Clinical Service Line provides tools to help you become more efficient, understand quality standards and improve the process of care for patients: http://www.gastro.org/patient-care/conditions-diseases/colorectal-cancer

The proportion of rectal cancer cases diagnosed in people younger than 55 years doubled over the past 2 decades, according to a report published online Feb. 28 in the Journal of the National Cancer Institute.

In contrast, the proportion diagnosed in people older than 55 years has decreased over the last 4 decades, said Rebecca L. Siegel, MPH, strategic director of surveillance information services of surveillance and health services research at the American Cancer Society and her associates.

Courtesy Wikimedia Commons/nephron/Creative Commons License

The proportion of rectal cancer cases diagnosed in people younger than 55 years doubled over the past 2 decades, according to a report published online Feb. 28 in the Journal of the National Cancer Institute.

In contrast, the proportion diagnosed in people older than 55 years has decreased over the last 4 decades, said Rebecca L. Siegel, MPH, strategic director of surveillance information services of surveillance and health services research at the American Cancer Society and her associates.

Courtesy Wikimedia Commons/nephron/Creative Commons License

The proportion of rectal cancer cases diagnosed in people younger than 55 years doubled over the past 2 decades, according to a report published online Feb. 28 in the Journal of the National Cancer Institute.

In contrast, the proportion diagnosed in people older than 55 years has decreased over the last 4 decades, said Rebecca L. Siegel, MPH, strategic director of surveillance information services of surveillance and health services research at the American Cancer Society and her associates.

Courtesy Wikimedia Commons/nephron/Creative Commons License

Endometriosis is a common gynecologic problem in adolescents and women. It often presents with pelvic pain, an ovarian endometrioma, and/or subfertility. In a prospective study of 116,678 nurses, the incidence of a new surgical diagnosis of endometriosis was greatest among women aged 25 to 29 years and lowest among women older than age 44.¹ Using the incidence data from this study, the calculated prevalence of endometriosis in this large cohort of women of reproductive age was approximately 8%.

Although endometriosis is known to be a very common gynecologic problem, many studies report that there can be long delays between onset of pelvic pain symptoms and the diagnosis of endometriosis (Figure 1).²⁻⁶ Combining the results from 5 studies, involving 1,187 women, the mean age of onset of pelvic pain symptoms was 22.1 years, and the mean age at the diagnosis of endometriosis was 30.7 years. This is a difference of 8.6 years between the age of symptom onset and age at diagnosis.²⁻⁶

What factors contribute to the diagnosis delay?

Both patient and physician factors contribute to the reported lengthy delay between symptom onset and endometriosis diagnosis.^7,8 Differentiating dysmenorrhea due to primary and secondary causes is difficult for both patients and physicians. Women may conceal the severity of menstrual pain to avoid both the embarrassment of drawing attention to themselves and being stigmatized as unable to cope. Most disappointing is that many women with endometriosis reported that they asked their clinician if endometriosis could be the cause of their severe dysmenorrhea and were told, “No.”^7,8

Of interest, the reported delay in the diagnosis of endometriosis is much shorter for women who pre-sent with infertility than for women who present with pelvic pain. In one study from the United States, the delay to diagnosis was 3.13 years for women who presented with infertility and 6.35 years for women who presented with severe pelvic pain.³ This suggests that clinicians and patients more rapidly pursue the diagnosis of endometriosis in women with infertility, but not pelvic pain.

Related article:
Endometriosis: Expert answers to 7 crucial questions on diagnosis

Initial treatment of pelvic pain with NSAIDs and estrogen—progestin contraceptives

Many women with undiagnosed endometriosis present with pelvic pain symptoms including moderate to severe dysmenorrhea. These women are often empirically treated with nonsteroidal anti-inflammatory drugs (NSAIDs) and combination estrogen−progestin contraceptives in either a cyclic or continuous manner.^9,10 Since many women with endometriosis will have a reduction in their pelvic pain with NSAID and contraceptive treatment, diagnosis of their endometriosis may be delayed until their disease progresses years after their initial presentation. It is important to gently alert these women to the possibility that they have undiagnosed endometriosis as the cause of their pain symptoms and encourage them to report any worsening pain symptoms in a timely manner.

Sometimes women with pelvic pain are treated with NSAIDs and contraceptives but no significant reduction in pain symptoms occurs. For these women, speedy consideration should be given to offering a laparoscopy to determine the cause of their pain.

Related article:
Avoiding “shotgun” treatment: New thoughts on endometriosis-associated pelvic pain

Diagnosing endometriosis relies on identifying flags in the patient’s history

The gold standard for endometriosis diagnosis is surgical visualization of endometriosis lesions, most often with laparoscopy, plus histologic confirmation of endometriosis on a tissue biopsy.^9,10 A key to reducing the time between onset of symptoms and diagnosis of endometriosis is identifying adolescents and women who are at high risk for having the disease. These women should be offered a laparoscopy procedure. In women with moderate to severe pelvic pain of at least 6 months duration, medical history, physical examination, and imaging studies can be helpful in identifying those at increased risk for endometriosis.

Items from the patient history that might raise the likelihood of endometriosis include:

• abdominopelvic pain, dysmenorrhea, menorrhagia, subfertility, dyspareunia and/or postcoital bleeding¹¹
• symptoms of dysmenorrhea and/or dyspareunia that are not responsive to NSAIDs or estrogen−progestin contraceptives¹²
• symptoms of dysmenorrhea and/or dyspareunia associated with absenteeism from school or work¹³
• multiple visits to the emergency department for severe dysmenorrhea
• endometriosis in the patient’s mother or sister
• subfertility with regular ovulation, patent fallopian tubes, and a partner with a normal semen analysis
• urinary frequency, urgency, and/or pain on urination
• diarrhea, constipation, nausea, dyschezia, bowel cramping, abdominal distention, and early satiety.

A daunting clinical challenge is that symptoms of endometriosis overlap with other gynecologic and nongynecologic problems including pelvic infection, adhesions, ovarian cysts, fibroids, irritable bowel syndrome, inflammatory bowel disease, interstitial cystitis, myofascial pain, depression, and history of sexual abuse.

Diagnosing endometriosis relies on identifying flags on physical exam

Physical examination findings that raise the likelihood that the patient has endometriosis include:

• fixed and retroverted uterus
• adnexal mass
• lesions of the cervix or posterior fornix that visually appear to be endometriosis
• uterosacral ligament abnormalities, including tenderness, thickening, and/or nodularity^14,15
• lateral displacement of the cervix (FIGURE 2)^16,17
• severe cervical stenosis.

Illustration: Marcia Hartsock for OBG Management

In one study of 57 women with a surgical diagnosis of endometriosis, uterosacral ligament abnormalities, lateral displacement of the cervix, and cervical stenosis were observed in 47%, 28%, and 19% of the women, respectively.¹⁷ In this same study 22 women had none of these findings, but 8 had a complex ovarian mass consistent with endometriosis.

The possibility of endometriosis increases as the number of history and physical examination findings suggestive of endometriosis increase.

Related article:
Endometriosis and pain: Expert answers to 6 questions targeting your management options

When transvaginal ultrasound can aid diagnosis

Most women with endometriosis have normal transvaginal ultrasonography (TVUS) results because ultrasound cannot detect small isolated peritoneal lesions of endometriosis present in Stage I disease, the most common stage of endometriosis. However, ultrasound is useful in detecting both ovarian endometriomas and nodules of deep infiltrating endometriosis (DIE).¹⁸ TVUS has excellent sensitivity (>90%) and specificity (>90%) for the detection of ovarian endometriomas because these cysts have characteristic, homogenous, low-level internal echoes.^19,20 For the diagnosis of DIE of the uterosacral ligaments and rectovaginal septum, TVUS has fair sensitivity (>50%) and excellent specificity (>90%).²¹ In most studies, magnetic resonance imaging performs no better than TVUS for imaging ovarian endometriomas and DIE. Hence, TVUS is the preferred imaging modality for detecting endometriosis.²²

This is a practice gap we can close

Clinicians take great pride in accurately solving patient problems in a timely and efficient manner. Substantial research indicates that we can improve the timeliness of our diagnosis of endometriosis. By acknowledging patients’ pain symptoms and recognizing the myriad symptoms and physical examination and imaging findings that are associated with endometriosis, we will close the gap and make this diagnosis with greater speed.

Share your thoughts! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Endometriosis is a common gynecologic problem in adolescents and women. It often presents with pelvic pain, an ovarian endometrioma, and/or subfertility. In a prospective study of 116,678 nurses, the incidence of a new surgical diagnosis of endometriosis was greatest among women aged 25 to 29 years and lowest among women older than age 44.¹ Using the incidence data from this study, the calculated prevalence of endometriosis in this large cohort of women of reproductive age was approximately 8%.

Although endometriosis is known to be a very common gynecologic problem, many studies report that there can be long delays between onset of pelvic pain symptoms and the diagnosis of endometriosis (Figure 1).²⁻⁶ Combining the results from 5 studies, involving 1,187 women, the mean age of onset of pelvic pain symptoms was 22.1 years, and the mean age at the diagnosis of endometriosis was 30.7 years. This is a difference of 8.6 years between the age of symptom onset and age at diagnosis.²⁻⁶

What factors contribute to the diagnosis delay?

Both patient and physician factors contribute to the reported lengthy delay between symptom onset and endometriosis diagnosis.^7,8 Differentiating dysmenorrhea due to primary and secondary causes is difficult for both patients and physicians. Women may conceal the severity of menstrual pain to avoid both the embarrassment of drawing attention to themselves and being stigmatized as unable to cope. Most disappointing is that many women with endometriosis reported that they asked their clinician if endometriosis could be the cause of their severe dysmenorrhea and were told, “No.”^7,8

Of interest, the reported delay in the diagnosis of endometriosis is much shorter for women who pre-sent with infertility than for women who present with pelvic pain. In one study from the United States, the delay to diagnosis was 3.13 years for women who presented with infertility and 6.35 years for women who presented with severe pelvic pain.³ This suggests that clinicians and patients more rapidly pursue the diagnosis of endometriosis in women with infertility, but not pelvic pain.

Related article:
Endometriosis: Expert answers to 7 crucial questions on diagnosis

Initial treatment of pelvic pain with NSAIDs and estrogen—progestin contraceptives

Many women with undiagnosed endometriosis present with pelvic pain symptoms including moderate to severe dysmenorrhea. These women are often empirically treated with nonsteroidal anti-inflammatory drugs (NSAIDs) and combination estrogen−progestin contraceptives in either a cyclic or continuous manner.^9,10 Since many women with endometriosis will have a reduction in their pelvic pain with NSAID and contraceptive treatment, diagnosis of their endometriosis may be delayed until their disease progresses years after their initial presentation. It is important to gently alert these women to the possibility that they have undiagnosed endometriosis as the cause of their pain symptoms and encourage them to report any worsening pain symptoms in a timely manner.

Sometimes women with pelvic pain are treated with NSAIDs and contraceptives but no significant reduction in pain symptoms occurs. For these women, speedy consideration should be given to offering a laparoscopy to determine the cause of their pain.

Related article:
Avoiding “shotgun” treatment: New thoughts on endometriosis-associated pelvic pain

Diagnosing endometriosis relies on identifying flags in the patient’s history

The gold standard for endometriosis diagnosis is surgical visualization of endometriosis lesions, most often with laparoscopy, plus histologic confirmation of endometriosis on a tissue biopsy.^9,10 A key to reducing the time between onset of symptoms and diagnosis of endometriosis is identifying adolescents and women who are at high risk for having the disease. These women should be offered a laparoscopy procedure. In women with moderate to severe pelvic pain of at least 6 months duration, medical history, physical examination, and imaging studies can be helpful in identifying those at increased risk for endometriosis.

Items from the patient history that might raise the likelihood of endometriosis include:

• abdominopelvic pain, dysmenorrhea, menorrhagia, subfertility, dyspareunia and/or postcoital bleeding¹¹
• symptoms of dysmenorrhea and/or dyspareunia that are not responsive to NSAIDs or estrogen−progestin contraceptives¹²
• symptoms of dysmenorrhea and/or dyspareunia associated with absenteeism from school or work¹³
• multiple visits to the emergency department for severe dysmenorrhea
• endometriosis in the patient’s mother or sister
• subfertility with regular ovulation, patent fallopian tubes, and a partner with a normal semen analysis
• urinary frequency, urgency, and/or pain on urination
• diarrhea, constipation, nausea, dyschezia, bowel cramping, abdominal distention, and early satiety.

A daunting clinical challenge is that symptoms of endometriosis overlap with other gynecologic and nongynecologic problems including pelvic infection, adhesions, ovarian cysts, fibroids, irritable bowel syndrome, inflammatory bowel disease, interstitial cystitis, myofascial pain, depression, and history of sexual abuse.

Diagnosing endometriosis relies on identifying flags on physical exam

Physical examination findings that raise the likelihood that the patient has endometriosis include:

• fixed and retroverted uterus
• adnexal mass
• lesions of the cervix or posterior fornix that visually appear to be endometriosis
• uterosacral ligament abnormalities, including tenderness, thickening, and/or nodularity^14,15
• lateral displacement of the cervix (FIGURE 2)^16,17
• severe cervical stenosis.

Illustration: Marcia Hartsock for OBG Management

In one study of 57 women with a surgical diagnosis of endometriosis, uterosacral ligament abnormalities, lateral displacement of the cervix, and cervical stenosis were observed in 47%, 28%, and 19% of the women, respectively.¹⁷ In this same study 22 women had none of these findings, but 8 had a complex ovarian mass consistent with endometriosis.

The possibility of endometriosis increases as the number of history and physical examination findings suggestive of endometriosis increase.

Related article:
Endometriosis and pain: Expert answers to 6 questions targeting your management options

When transvaginal ultrasound can aid diagnosis

Most women with endometriosis have normal transvaginal ultrasonography (TVUS) results because ultrasound cannot detect small isolated peritoneal lesions of endometriosis present in Stage I disease, the most common stage of endometriosis. However, ultrasound is useful in detecting both ovarian endometriomas and nodules of deep infiltrating endometriosis (DIE).¹⁸ TVUS has excellent sensitivity (>90%) and specificity (>90%) for the detection of ovarian endometriomas because these cysts have characteristic, homogenous, low-level internal echoes.^19,20 For the diagnosis of DIE of the uterosacral ligaments and rectovaginal septum, TVUS has fair sensitivity (>50%) and excellent specificity (>90%).²¹ In most studies, magnetic resonance imaging performs no better than TVUS for imaging ovarian endometriomas and DIE. Hence, TVUS is the preferred imaging modality for detecting endometriosis.²²

This is a practice gap we can close

Clinicians take great pride in accurately solving patient problems in a timely and efficient manner. Substantial research indicates that we can improve the timeliness of our diagnosis of endometriosis. By acknowledging patients’ pain symptoms and recognizing the myriad symptoms and physical examination and imaging findings that are associated with endometriosis, we will close the gap and make this diagnosis with greater speed.

Share your thoughts! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

Endometriosis is a common gynecologic problem in adolescents and women. It often presents with pelvic pain, an ovarian endometrioma, and/or subfertility. In a prospective study of 116,678 nurses, the incidence of a new surgical diagnosis of endometriosis was greatest among women aged 25 to 29 years and lowest among women older than age 44.¹ Using the incidence data from this study, the calculated prevalence of endometriosis in this large cohort of women of reproductive age was approximately 8%.

Although endometriosis is known to be a very common gynecologic problem, many studies report that there can be long delays between onset of pelvic pain symptoms and the diagnosis of endometriosis (Figure 1).²⁻⁶ Combining the results from 5 studies, involving 1,187 women, the mean age of onset of pelvic pain symptoms was 22.1 years, and the mean age at the diagnosis of endometriosis was 30.7 years. This is a difference of 8.6 years between the age of symptom onset and age at diagnosis.²⁻⁶

What factors contribute to the diagnosis delay?

Both patient and physician factors contribute to the reported lengthy delay between symptom onset and endometriosis diagnosis.^7,8 Differentiating dysmenorrhea due to primary and secondary causes is difficult for both patients and physicians. Women may conceal the severity of menstrual pain to avoid both the embarrassment of drawing attention to themselves and being stigmatized as unable to cope. Most disappointing is that many women with endometriosis reported that they asked their clinician if endometriosis could be the cause of their severe dysmenorrhea and were told, “No.”^7,8

Of interest, the reported delay in the diagnosis of endometriosis is much shorter for women who pre-sent with infertility than for women who present with pelvic pain. In one study from the United States, the delay to diagnosis was 3.13 years for women who presented with infertility and 6.35 years for women who presented with severe pelvic pain.³ This suggests that clinicians and patients more rapidly pursue the diagnosis of endometriosis in women with infertility, but not pelvic pain.

Related article:
Endometriosis: Expert answers to 7 crucial questions on diagnosis

Initial treatment of pelvic pain with NSAIDs and estrogen—progestin contraceptives

Many women with undiagnosed endometriosis present with pelvic pain symptoms including moderate to severe dysmenorrhea. These women are often empirically treated with nonsteroidal anti-inflammatory drugs (NSAIDs) and combination estrogen−progestin contraceptives in either a cyclic or continuous manner.^9,10 Since many women with endometriosis will have a reduction in their pelvic pain with NSAID and contraceptive treatment, diagnosis of their endometriosis may be delayed until their disease progresses years after their initial presentation. It is important to gently alert these women to the possibility that they have undiagnosed endometriosis as the cause of their pain symptoms and encourage them to report any worsening pain symptoms in a timely manner.

Sometimes women with pelvic pain are treated with NSAIDs and contraceptives but no significant reduction in pain symptoms occurs. For these women, speedy consideration should be given to offering a laparoscopy to determine the cause of their pain.

Related article:
Avoiding “shotgun” treatment: New thoughts on endometriosis-associated pelvic pain

Diagnosing endometriosis relies on identifying flags in the patient’s history

The gold standard for endometriosis diagnosis is surgical visualization of endometriosis lesions, most often with laparoscopy, plus histologic confirmation of endometriosis on a tissue biopsy.^9,10 A key to reducing the time between onset of symptoms and diagnosis of endometriosis is identifying adolescents and women who are at high risk for having the disease. These women should be offered a laparoscopy procedure. In women with moderate to severe pelvic pain of at least 6 months duration, medical history, physical examination, and imaging studies can be helpful in identifying those at increased risk for endometriosis.

Items from the patient history that might raise the likelihood of endometriosis include:

• abdominopelvic pain, dysmenorrhea, menorrhagia, subfertility, dyspareunia and/or postcoital bleeding¹¹
• symptoms of dysmenorrhea and/or dyspareunia that are not responsive to NSAIDs or estrogen−progestin contraceptives¹²
• symptoms of dysmenorrhea and/or dyspareunia associated with absenteeism from school or work¹³
• multiple visits to the emergency department for severe dysmenorrhea
• endometriosis in the patient’s mother or sister
• subfertility with regular ovulation, patent fallopian tubes, and a partner with a normal semen analysis
• urinary frequency, urgency, and/or pain on urination
• diarrhea, constipation, nausea, dyschezia, bowel cramping, abdominal distention, and early satiety.

A daunting clinical challenge is that symptoms of endometriosis overlap with other gynecologic and nongynecologic problems including pelvic infection, adhesions, ovarian cysts, fibroids, irritable bowel syndrome, inflammatory bowel disease, interstitial cystitis, myofascial pain, depression, and history of sexual abuse.

Diagnosing endometriosis relies on identifying flags on physical exam

Physical examination findings that raise the likelihood that the patient has endometriosis include:

• fixed and retroverted uterus
• adnexal mass
• lesions of the cervix or posterior fornix that visually appear to be endometriosis
• uterosacral ligament abnormalities, including tenderness, thickening, and/or nodularity^14,15
• lateral displacement of the cervix (FIGURE 2)^16,17
• severe cervical stenosis.

Illustration: Marcia Hartsock for OBG Management

In one study of 57 women with a surgical diagnosis of endometriosis, uterosacral ligament abnormalities, lateral displacement of the cervix, and cervical stenosis were observed in 47%, 28%, and 19% of the women, respectively.¹⁷ In this same study 22 women had none of these findings, but 8 had a complex ovarian mass consistent with endometriosis.

The possibility of endometriosis increases as the number of history and physical examination findings suggestive of endometriosis increase.

Related article:
Endometriosis and pain: Expert answers to 6 questions targeting your management options

When transvaginal ultrasound can aid diagnosis

Most women with endometriosis have normal transvaginal ultrasonography (TVUS) results because ultrasound cannot detect small isolated peritoneal lesions of endometriosis present in Stage I disease, the most common stage of endometriosis. However, ultrasound is useful in detecting both ovarian endometriomas and nodules of deep infiltrating endometriosis (DIE).¹⁸ TVUS has excellent sensitivity (>90%) and specificity (>90%) for the detection of ovarian endometriomas because these cysts have characteristic, homogenous, low-level internal echoes.^19,20 For the diagnosis of DIE of the uterosacral ligaments and rectovaginal septum, TVUS has fair sensitivity (>50%) and excellent specificity (>90%).²¹ In most studies, magnetic resonance imaging performs no better than TVUS for imaging ovarian endometriomas and DIE. Hence, TVUS is the preferred imaging modality for detecting endometriosis.²²

This is a practice gap we can close

Clinicians take great pride in accurately solving patient problems in a timely and efficient manner. Substantial research indicates that we can improve the timeliness of our diagnosis of endometriosis. By acknowledging patients’ pain symptoms and recognizing the myriad symptoms and physical examination and imaging findings that are associated with endometriosis, we will close the gap and make this diagnosis with greater speed.

Share your thoughts! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

She wanted to labor on hands and knees

During prenatal visits, a woman, pregnant with her fourth child, discussed undergoing labor and delivery in any position other than on her back; the ObGyn agreed. When she arrived at the hospital in labor, the patient told the nurse that she preferred to labor on her hands and knees. The nurse disagreed because of the fetal heart-rate monitor.

When the patient began hard labor, she turned herself over onto her hands and knees and again informed the nurse that she could not labor on her back. The nurse flipped the patient onto her back by taking her wrists and pulling the patient’s hands out from under her. The nurse then delayed delivery until the ObGyn arrived by putting pressure on the baby’s head. During delivery, a second nurse forcibly pressed the patient’s left knee back toward her chest, leaving her legs in an asymmetric position.

Two months later, the patient reported chronic severe pelvic pain and was found to have pudendal neuralgia. She underwent nerve blocks and takes medication for chronic pain.

PATIENT’S CLAIM:

The ObGyn did not assume responsibility when he arrived for the delivery. The nurses did not follow the standard of care. The patient’s injury was the result of tension and compression due to malpositioning of the patient’s legs during delivery.

DEFENDANTS’ DEFENSE:

There was no breach in the standard of care. The patient’s injury, if any, had not been caused by the delivery.

VERDICT:

A $16 million Alabama verdict was returned.

Related article:
10 tips for overcoming common challenges of intrapartum fetal monitoring

Late-term abortion: $1.4M award

Although genetic testing was scheduled for a 37-year-old woman’s 15-week prenatal visit, the ObGyn’s staff failed to draw blood. At 19 weeks’ gestation (April 24), blood was drawn. The ObGyn signed off on test results that showed a high risk for fetal anomaly on May 2, but the patient was not informed until May 22. The ObGyn scheduled amniocentesis for June 3. On May 30, the hospital, based in Illinois, cancelled the test, telling the ObGyn that it was because the patient was over 24 weeks’ pregnant and there was no labor and delivery unit to respond if complications arose. Instead of notifying the patient, the ObGyn arranged for amniocentesis to be performed elsewhere on June 3. The ObGyn saw the amniocentesis results on June 13, but did not tell the patient until July 3, when he advised her to terminate the pregnancy because the baby had severe cardiac defects and Down syndrome; he felt the child would not survive or have very poor quality of life. The ObGyn arranged for the patient to undergo a third-trimester abortion in Kansas and paid all expenses. On July 14, the patient began the 5-day abortion process at 30+ weeks’ gestation.

PATIENT’S CLAIM:

She was never offered additional genetic testing or expedited amniocentesis. She was not told that abortion is illegal in Illinois after 23 6/7 weeks’ gestation. The ObGyn had a motive for paying for her abortion. He never counseled her about options to keep the child. She endured extreme pain and emotional trauma during the abortion and was later found to have posttraumatic stress disorder, multidepressive disorder, and anxiety as a result of the experience. She countered the ObGyn’s contact information claim by saying that her phone number had not changed.

PHYSICIAN’S DEFENSE:

The ObGyn admitted negligence in failing to timely communicate test results but contended that the patient was more than 50% responsible for any delay by failing to update her contact information when she moved. The ObGyn denied causation of any injuries or damage.

VERDICT:

A $1,439,250 Illinois verdict was returned.

Related article:
4 Supreme Court decisions important to ObGyns from the 2015−2016 term

Did delay in delivery cause infant's death?

A woman presented to the hospital in labor. During delivery, the patient’s ObGyn encountered shoulder dystocia. The infant died shortly after birth.

PARENTS’ CLAIM:

The ObGyn and hospital nurses were negligent. The nurses failed to monitor labor and properly communicate with the ObGyn. The ObGyn failed to appreciate the baby’s large size and order a cesarean delivery. The infant’s death was due to a hypoxic event during delivery.

DEFENDANTS’ DEFENSE:

The baby gained an unexpected amount of weight between the last prenatal visit and labor. There was no reason to expect a complication to vaginal delivery. The nurses denied negligence. The child’s sudden death was caused by a genetic cardiac condition.

VERDICT:

A Tennessee defense verdict was returned.

Related article:
Shoulder dystocia: Taking the fear out of management

These cases were selected by the editors of OBG Management from Medical Malpractice Verdicts, Settlements & Experts, with permission of the editor, Lewis Laska (www.verdictslaska.com). The information available to the editors about the cases presented here is sometimes incomplete. Moreover, the cases may or may not have merit. Nevertheless, these cases represent the types of clinical situations that typically result in litigation and are meant to illustrate nationwide variation in jury verdicts and awards.

Share your thoughts! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

She wanted to labor on hands and knees

During prenatal visits, a woman, pregnant with her fourth child, discussed undergoing labor and delivery in any position other than on her back; the ObGyn agreed. When she arrived at the hospital in labor, the patient told the nurse that she preferred to labor on her hands and knees. The nurse disagreed because of the fetal heart-rate monitor.

When the patient began hard labor, she turned herself over onto her hands and knees and again informed the nurse that she could not labor on her back. The nurse flipped the patient onto her back by taking her wrists and pulling the patient’s hands out from under her. The nurse then delayed delivery until the ObGyn arrived by putting pressure on the baby’s head. During delivery, a second nurse forcibly pressed the patient’s left knee back toward her chest, leaving her legs in an asymmetric position.

Two months later, the patient reported chronic severe pelvic pain and was found to have pudendal neuralgia. She underwent nerve blocks and takes medication for chronic pain.

PATIENT’S CLAIM:

The ObGyn did not assume responsibility when he arrived for the delivery. The nurses did not follow the standard of care. The patient’s injury was the result of tension and compression due to malpositioning of the patient’s legs during delivery.

DEFENDANTS’ DEFENSE:

There was no breach in the standard of care. The patient’s injury, if any, had not been caused by the delivery.

VERDICT:

A $16 million Alabama verdict was returned.

Related article:
10 tips for overcoming common challenges of intrapartum fetal monitoring

Late-term abortion: $1.4M award

Although genetic testing was scheduled for a 37-year-old woman’s 15-week prenatal visit, the ObGyn’s staff failed to draw blood. At 19 weeks’ gestation (April 24), blood was drawn. The ObGyn signed off on test results that showed a high risk for fetal anomaly on May 2, but the patient was not informed until May 22. The ObGyn scheduled amniocentesis for June 3. On May 30, the hospital, based in Illinois, cancelled the test, telling the ObGyn that it was because the patient was over 24 weeks’ pregnant and there was no labor and delivery unit to respond if complications arose. Instead of notifying the patient, the ObGyn arranged for amniocentesis to be performed elsewhere on June 3. The ObGyn saw the amniocentesis results on June 13, but did not tell the patient until July 3, when he advised her to terminate the pregnancy because the baby had severe cardiac defects and Down syndrome; he felt the child would not survive or have very poor quality of life. The ObGyn arranged for the patient to undergo a third-trimester abortion in Kansas and paid all expenses. On July 14, the patient began the 5-day abortion process at 30+ weeks’ gestation.

PATIENT’S CLAIM:

She was never offered additional genetic testing or expedited amniocentesis. She was not told that abortion is illegal in Illinois after 23 6/7 weeks’ gestation. The ObGyn had a motive for paying for her abortion. He never counseled her about options to keep the child. She endured extreme pain and emotional trauma during the abortion and was later found to have posttraumatic stress disorder, multidepressive disorder, and anxiety as a result of the experience. She countered the ObGyn’s contact information claim by saying that her phone number had not changed.

PHYSICIAN’S DEFENSE:

The ObGyn admitted negligence in failing to timely communicate test results but contended that the patient was more than 50% responsible for any delay by failing to update her contact information when she moved. The ObGyn denied causation of any injuries or damage.

VERDICT:

A $1,439,250 Illinois verdict was returned.

Related article:
4 Supreme Court decisions important to ObGyns from the 2015−2016 term

Did delay in delivery cause infant's death?

A woman presented to the hospital in labor. During delivery, the patient’s ObGyn encountered shoulder dystocia. The infant died shortly after birth.

PARENTS’ CLAIM:

The ObGyn and hospital nurses were negligent. The nurses failed to monitor labor and properly communicate with the ObGyn. The ObGyn failed to appreciate the baby’s large size and order a cesarean delivery. The infant’s death was due to a hypoxic event during delivery.

DEFENDANTS’ DEFENSE:

The baby gained an unexpected amount of weight between the last prenatal visit and labor. There was no reason to expect a complication to vaginal delivery. The nurses denied negligence. The child’s sudden death was caused by a genetic cardiac condition.

VERDICT:

A Tennessee defense verdict was returned.

Related article:
Shoulder dystocia: Taking the fear out of management

These cases were selected by the editors of OBG Management from Medical Malpractice Verdicts, Settlements & Experts, with permission of the editor, Lewis Laska (www.verdictslaska.com). The information available to the editors about the cases presented here is sometimes incomplete. Moreover, the cases may or may not have merit. Nevertheless, these cases represent the types of clinical situations that typically result in litigation and are meant to illustrate nationwide variation in jury verdicts and awards.

Share your thoughts! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

She wanted to labor on hands and knees

During prenatal visits, a woman, pregnant with her fourth child, discussed undergoing labor and delivery in any position other than on her back; the ObGyn agreed. When she arrived at the hospital in labor, the patient told the nurse that she preferred to labor on her hands and knees. The nurse disagreed because of the fetal heart-rate monitor.

When the patient began hard labor, she turned herself over onto her hands and knees and again informed the nurse that she could not labor on her back. The nurse flipped the patient onto her back by taking her wrists and pulling the patient’s hands out from under her. The nurse then delayed delivery until the ObGyn arrived by putting pressure on the baby’s head. During delivery, a second nurse forcibly pressed the patient’s left knee back toward her chest, leaving her legs in an asymmetric position.

Two months later, the patient reported chronic severe pelvic pain and was found to have pudendal neuralgia. She underwent nerve blocks and takes medication for chronic pain.

PATIENT’S CLAIM:

The ObGyn did not assume responsibility when he arrived for the delivery. The nurses did not follow the standard of care. The patient’s injury was the result of tension and compression due to malpositioning of the patient’s legs during delivery.

DEFENDANTS’ DEFENSE:

There was no breach in the standard of care. The patient’s injury, if any, had not been caused by the delivery.

VERDICT:

A $16 million Alabama verdict was returned.

Related article:
10 tips for overcoming common challenges of intrapartum fetal monitoring

Late-term abortion: $1.4M award

Although genetic testing was scheduled for a 37-year-old woman’s 15-week prenatal visit, the ObGyn’s staff failed to draw blood. At 19 weeks’ gestation (April 24), blood was drawn. The ObGyn signed off on test results that showed a high risk for fetal anomaly on May 2, but the patient was not informed until May 22. The ObGyn scheduled amniocentesis for June 3. On May 30, the hospital, based in Illinois, cancelled the test, telling the ObGyn that it was because the patient was over 24 weeks’ pregnant and there was no labor and delivery unit to respond if complications arose. Instead of notifying the patient, the ObGyn arranged for amniocentesis to be performed elsewhere on June 3. The ObGyn saw the amniocentesis results on June 13, but did not tell the patient until July 3, when he advised her to terminate the pregnancy because the baby had severe cardiac defects and Down syndrome; he felt the child would not survive or have very poor quality of life. The ObGyn arranged for the patient to undergo a third-trimester abortion in Kansas and paid all expenses. On July 14, the patient began the 5-day abortion process at 30+ weeks’ gestation.

PATIENT’S CLAIM:

She was never offered additional genetic testing or expedited amniocentesis. She was not told that abortion is illegal in Illinois after 23 6/7 weeks’ gestation. The ObGyn had a motive for paying for her abortion. He never counseled her about options to keep the child. She endured extreme pain and emotional trauma during the abortion and was later found to have posttraumatic stress disorder, multidepressive disorder, and anxiety as a result of the experience. She countered the ObGyn’s contact information claim by saying that her phone number had not changed.

PHYSICIAN’S DEFENSE:

The ObGyn admitted negligence in failing to timely communicate test results but contended that the patient was more than 50% responsible for any delay by failing to update her contact information when she moved. The ObGyn denied causation of any injuries or damage.

VERDICT:

A $1,439,250 Illinois verdict was returned.

Related article:
4 Supreme Court decisions important to ObGyns from the 2015−2016 term

Did delay in delivery cause infant's death?

A woman presented to the hospital in labor. During delivery, the patient’s ObGyn encountered shoulder dystocia. The infant died shortly after birth.

PARENTS’ CLAIM:

The ObGyn and hospital nurses were negligent. The nurses failed to monitor labor and properly communicate with the ObGyn. The ObGyn failed to appreciate the baby’s large size and order a cesarean delivery. The infant’s death was due to a hypoxic event during delivery.

DEFENDANTS’ DEFENSE:

The baby gained an unexpected amount of weight between the last prenatal visit and labor. There was no reason to expect a complication to vaginal delivery. The nurses denied negligence. The child’s sudden death was caused by a genetic cardiac condition.

VERDICT:

A Tennessee defense verdict was returned.

Related article:
Shoulder dystocia: Taking the fear out of management

These cases were selected by the editors of OBG Management from Medical Malpractice Verdicts, Settlements & Experts, with permission of the editor, Lewis Laska (www.verdictslaska.com). The information available to the editors about the cases presented here is sometimes incomplete. Moreover, the cases may or may not have merit. Nevertheless, these cases represent the types of clinical situations that typically result in litigation and are meant to illustrate nationwide variation in jury verdicts and awards.

Share your thoughts! Send your Letter to the Editor to [email protected]. Please include your name and the city and state in which you practice.

A randomized controlled multicenter study published by Carson et al. in JAMA concluded that, for patients with “chronic critical illness” (defined as requiring 7 days of mechanical ventilation), palliative care team-led informational and emotional support meetings did not reduce anxiety or depression for families and may have increased posttraumatic stress disorder symptoms (2016:316[1]:51-62. doi: 10.1001/jama.2016.8474).

This report may surprise surgeons, as well as practitioners in other specialties, as the disconnect between palliative care and critical care services has been previously perceived as an education and access issue, not an outcome problem.

Dr. Rivet is a colon and rectal surgeon with training and board certification in hospice and palliative medicine. She is an assistant professor, departments of surgery and internal medicine, Virginia Commonwealth University, Richmond. She has no disclosures.

A randomized controlled multicenter study published by Carson et al. in JAMA concluded that, for patients with “chronic critical illness” (defined as requiring 7 days of mechanical ventilation), palliative care team-led informational and emotional support meetings did not reduce anxiety or depression for families and may have increased posttraumatic stress disorder symptoms (2016:316[1]:51-62. doi: 10.1001/jama.2016.8474).

This report may surprise surgeons, as well as practitioners in other specialties, as the disconnect between palliative care and critical care services has been previously perceived as an education and access issue, not an outcome problem.

Dr. Rivet is a colon and rectal surgeon with training and board certification in hospice and palliative medicine. She is an assistant professor, departments of surgery and internal medicine, Virginia Commonwealth University, Richmond. She has no disclosures.

A randomized controlled multicenter study published by Carson et al. in JAMA concluded that, for patients with “chronic critical illness” (defined as requiring 7 days of mechanical ventilation), palliative care team-led informational and emotional support meetings did not reduce anxiety or depression for families and may have increased posttraumatic stress disorder symptoms (2016:316[1]:51-62. doi: 10.1001/jama.2016.8474).

This report may surprise surgeons, as well as practitioners in other specialties, as the disconnect between palliative care and critical care services has been previously perceived as an education and access issue, not an outcome problem.

Dr. Rivet is a colon and rectal surgeon with training and board certification in hospice and palliative medicine. She is an assistant professor, departments of surgery and internal medicine, Virginia Commonwealth University, Richmond. She has no disclosures.

SAFER CESAREAN SCALPEL

FOR MORE INFORMATION, VISIT: www.sureglide.info

RF ENERGY TO IMPROVE SEXUAL FUNCTION

According to Viveve Medical, results of the recent Viveve I multicenter, blinded, randomized, sham controlled study showed no serious adverse effects plus improvement in arousal and/or orgasm self-reported by 9 of 10 women who noted vaginal laxity and sexual dysfunction following vaginal childbirth.

FOR MORE INFORMATION, VISIT: www.viveve.com

NEXT-GENERATION ENDOMETRIAL ABLATION

FOR MORE INFORMATION, VISIT: www.novasure.com

SAFER CESAREAN SCALPEL

FOR MORE INFORMATION, VISIT: www.sureglide.info

RF ENERGY TO IMPROVE SEXUAL FUNCTION

According to Viveve Medical, results of the recent Viveve I multicenter, blinded, randomized, sham controlled study showed no serious adverse effects plus improvement in arousal and/or orgasm self-reported by 9 of 10 women who noted vaginal laxity and sexual dysfunction following vaginal childbirth.

FOR MORE INFORMATION, VISIT: www.viveve.com

NEXT-GENERATION ENDOMETRIAL ABLATION

FOR MORE INFORMATION, VISIT: www.novasure.com

SAFER CESAREAN SCALPEL

FOR MORE INFORMATION, VISIT: www.sureglide.info

RF ENERGY TO IMPROVE SEXUAL FUNCTION

According to Viveve Medical, results of the recent Viveve I multicenter, blinded, randomized, sham controlled study showed no serious adverse effects plus improvement in arousal and/or orgasm self-reported by 9 of 10 women who noted vaginal laxity and sexual dysfunction following vaginal childbirth.

FOR MORE INFORMATION, VISIT: www.viveve.com

NEXT-GENERATION ENDOMETRIAL ABLATION

FOR MORE INFORMATION, VISIT: www.novasure.com

The care of hospitalized patients requires practitioners from multiple disciplines to assess and communicate the patient’s status in a dynamic manner during hospitalization. Although optimal teamwork is needed for patient care to be delivered reliably and efficiently, care within hospitals is typically delivered in a fragmented manner.¹ A bedside model for daily interdisciplinary rounds (IDR) has been proposed as a method to provide a structured process and engage all team members in a patient-centered, system-of-care delivery.² Specific advantages of convening rounds in the presence of the patient include the ability to directly assess care (eg, presence of a potentially unnecessary urinary catheter), patient engagement in key aspects of their care and disposition, and an increased opportunity for team members to develop a shared understanding of the patient’s views and needs.

Implementing dramatic changes to the workflow of multiple disciplines will require rigorous evidence to support a concerted effort from leadership and buy-in from stakeholders at the front line of patient care. Despite the urgency for evidence, there has been little investigation of this strategy. A systematic review³ identified 30 studies published between 1998 and 2013 addressing interdisciplinary interventions on medical wards, none of which examined a bedside IDR model. In a study performed after the period assessed by the systematic review, Stein et al⁴ described the restructuring of a medical ward as an accountable care unit (ACU), which included a bedside model for rounds by the interdisciplinary team. The change was associated with decreased mortality and length of stay (LOS), although the study did not isolate the impact of rounds or use a concurrent control group and presented aggregate rather than patient-level outcomes. The lack of convincing data may be a reason bedside rounds are not widely employed by hospitals. To provide high-quality evidence, we performed a large, prospective controlled trial comparing a structured bedside model (mobile interdisciplinary care rounds [MICRO]) with standard rounds.

This study took place at the Mount Sinai Hospital, which is a 1171-bed tertiary care academic medical center in New York City, New York. A nonteaching unit offered the ability to use a prospective controlled design. Patients were assigned to the north and the south wings of the unit in a quasi-randomized manner, rather than based on diagnosis or acuity. We transformed IDR to a bedside model on the north side of the unit (MICRO group), while the south side of the unit continued using standard conference room-based IDR (control group). The north and south sides of the unit contain 17 and 14 beds, respectively. During the study period, nurses and hospitalists cared for patients on both sides of the study unit, although on any given day were assigned only patients on 1 side of the unit. The unit uses a clinical microsystem model, which has been defined as “a group of clinicians and staff working together with a shared clinical purpose to provide care for a population of patients,” and has a defined set of characteristics associated with high performance.^5,6 Our microsystem model has incorporated features as described by Stein’s ACU model,⁴ including co-leadership by a hospitalist and a nurse manager, geographic assignment of patients to teams, and unit-level data reports. One hospitalist is assigned geographically to each area of the unit in a 2- to 4-week rotation. Coverage of the unit does not include house staff; patients are primarily assigned to hospitalists working with nurse practitioners. Patients were enrolled prospectively during their initial IDR by a research coordinator. Patient-level data and outcomes were collected prospectively by a research coordinator who attended IDR on the intervention and the control sides of the study unit daily.

Inclusion Criteria

All patients admitted to the medicine service on the study unit were eligible. Patients were greater than 18 years and admitted for an acute medical condition. Patients admitted to another unit and later transferred to the study unit were enrolled at the time of transfer. Patients could be included more than once if hospitalized on the study unit on more than 1 occasion. Most patients were covered by hospitalists, although patients covered by private physicians were included. Patients from other departments, including family medicine, are uncommonly admitted to the unit and were excluded. Patients were also excluded if they were admitted and discharged over the same weekend, because the MICRO rounds occur during weekdays and there was no opportunity to offer the intervention on Saturdays and Sundays.

MICRO Intervention

Interdisciplinary rounds occurred daily at 10:00 am for the control group and at 10:30 am for the MICRO group, and were attended by the hospitalist caring for the majority of patients on the unit, staff nurses, and the unit medical director, nurse manager, social worker, and case manager. Rounds on the control unit focused on the plan of care and disposition but did not follow any set structure and were typically 25 to 30 minutes in duration.

The MICRO rounds occurred at the bedside and followed a structured script (Appendix 1) that was designed to limit discussion of each patient to 3 minutes or less, and included speaking roles for the hospitalist, nurse, and social worker. For private physicians, the nurse practitioner assigned to the patient performed the role of the hospitalist. Rounds were expected to be approximately 50 minutes in duration. Patients were further engaged by asking for their main goal for the day. A patient safety checklist was reviewed. Initially, this task was performed by the nurse manager, who did not verbalize the items unless a deficiency was noted. After 6 months’ experience, this responsibility was given to the staff nurse, who reviewed the checklist verbally as part of the bedside script. Patients were seen daily, including those being discharged later that same day.

Staff and Clinician Education

We developed and implemented a curriculum based on a modified version of the Agency for Healthcare Research and Quality’s TeamStepps^® program to ensure that all team members were provided with the basic principles of communication within the healthcare setting. The curriculum consisted of interactive didactics on essential elements of teamwork, including team structure, communication, situation monitoring, and mutual support, as well as the purpose and structure of the MICRO model. The curriculum was delivered to nurses at 3 monthly staff meetings on the study unit and to hospitalists during 3 hospital medicine grand rounds over a 3-month period. Nurses and physicians providing care on both geographic areas of the study unit received the education program because no group of practitioners was designated to only 1 geographic area.

Outcomes

Primary and Secondary Outcomes

The primary outcomes were clinical deterioration (CD) and length of stay. Clinical deterioration was a composite outcome defined a priori as death; escalation of care (ie, transfer to an intensive care unit, intermediate care unit, or teaching unit); or a hospital-acquired complication (ie, venous thromboembolism, fall, stage III-IV pressure ulcer, catheter-associated urinary tract infection, central-line associated bloodstream infection, or Clostridium difficile-associated diarrhea). The LOS was calculated as the mean LOS with outliers excluded (outliers defined as having a LOS 100 days or longer or 2.5 or more standard deviations from the expected LOS).

Process metrics on IDR, such as the duration of rounds, attendance by members of the interdisciplinary team, the percentage of patients discussed, or the effectiveness of communication, were not collected. We assessed patient satisfaction based on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey.

Patient Safety Culture Survey

To assess the impact on the perceptions of patient safety, we administered the Agency for Healthcare Research and Quality (AHRQ) Hospital Survey on Patient Safety Culture to all staff and clinicians working on both sides of the study unit immediately before and 12 months after implementation of the MICRO model. Results are reported for the AHRQ dimensions that were most relevant to the MICRO intervention: “teamwork within units,” “overall perceptions of safety,” “communication,” “openness,” “overall patient safety grade,” and “handoffs and transitions.” The survey represents pre- and post-comparison. All nurses and hospitalists on both the MICRO and control sides of the study unit had received the TeamStepps curriculum and participated in MICRO rounds by the time of the postintervention survey. We added 3 questions specifically assessing the perception of the efficiency and effectiveness of IDR. Postintervention respondents reflected on their overall impression of IDR, which included their experiences on both sides of the unit, because no group of nurses or hospitalists was exposed only to the MICRO side or the control side of the unit. Responses to survey questions were recorded on a 5-point Likert scale (from “strongly disagree” to “strongly agree” for opinion questions; and “never,” rarely,” sometimes,” “most of the time,” and “always” for frequency questions) and given a score from 1 to 5. The question asking for an overall grade for patient safety was scored from 1 to 5 points corresponding to letter grade choices F, D, C, B, A.

Statistical Analysis

The sample size was based on the estimate of the baseline rate of the primary outcome of CD and the projected decrease by the MICRO intervention. A study using the Global Trigger Tool developed by the Institute for Healthcare Improvement provided a best estimate of 16% as the baseline rate for CD.⁷ A total of 2000 hospitalizations were planned to be included to have a power of at least 80% to detect a 25% reduction in the annual incidence of CD with a 2-tailed type I error rate of 0.05. Comparisons of dichotomous event rates were made using chi square tests at a 2-tailed level for significance of 0.05. The LOS was analyzed using the nonparametric median test and multivariable regression analysis. We used a generalized linear model with gamma distribution and log link for all analyses of LOS, where LOS was the outcome variable, and intervention vs. control unit type was the predictor variable. Age, sex, race, payer, case mix, and comorbidities defined with the Elixhauser algorithm were used as covariates.⁸ We used multivariable logistic regression for analysis of CD, where the dependent variable was CD. Predictor variables included intervention, patient age, sex, race, payer, case mix and comorbidities. Patient satisfaction data were compared using the chi square test. The Student t test for dependent means was used to analyze the patient safety culture survey data.

The study protocol was submitted to the Icahn Mount Sinai School of Medicine’s institutional review board and determined to be exempt from full review.

A total of 2005 hospitalizations were included over the 12-month study period, consisting of 1089 hospitalizations in the MICRO group and 916 in the control group. Bedside and standard IDR were completed daily, Monday through Friday without exception. The demographic characteristics and comorbidities were similar for the 2 groups (Table). Hospitalizations of patients who were initially admitted to another unit and subsequently transferred to the study unit accounted for 11.1% of hospitalizations.

Risk-adjusted LOS was similar for the groups (6.6 vs 7.0 days, P = 0.17, for the MICRO and control groups, respectively). On subgroup analysis, a reduction in LOS was noted for patients transferred to the study unit (10.4 vs 14.0 days, P = 0.02, for the MICRO and control groups, respectively). The LOS was unchanged for patients admitted directly to the study unit (6.0 vs 5.8 days, P = 0.93). There was no difference in the incidence of clinical deterioration for the MICRO or control groups (7.7% vs 9.3%, odds ratio, 0.89; 95% confidence interval, 0.61-1.22, P = 0.46).

The finding of a LOS benefit for the MICRO group limited to patients transferred to the study unit prompted a comparison of patients transferred to the study unit and patients directly admitted to the study unit from the emergency department (Appendix 2). Compared to patients admitted directly to the study unit, patients transferred to the study unit were more likely to have Medicaid or no insurance, more likely to be discharged to a facility, had longer LOS, and were more likely to experience CD.

Patient Satisfaction

There were 175 and 140 responses to the HCAHPS survey for the MICRO and the control groups, respectively. Patients in the MICRO group were more likely to report that “doctors, nurses, or other hospital staff talk with you about whether you would have the help you needed when you left the hospital” (88% vs 78%, P = 0.01). Responses for all other HCAHPS items were similar for the 2 groups.

Clinician/Staff Survey

The response rate was 96% (30 nurses and 17 hospitalists) pre-intervention and 100% (30 nurses and 22 hospitalists) postintervention. Hospitalists and nurses gave significantly higher scores for the dimensions “teamwork within units,” “overall perception of patient safety,” and “patient safety grade” on the postintervention survey compared to the pre-intervention survey (Figure 1). Hospitalists and nurses rated the efficiency of IDR and the ability of IDR to identify safety issues higher on the postintervention survey compared to the pre-intervention survey (Figure 2).

We transformed daily IDR from a standard conference room model to a structured bedside model with scripted roles, and performed a rigorous comparison using patient-level data. Our finding that transforming daily IDR from a standard conference room model to a bedside model did not significantly reduce LOS suggests either that the model is ineffective or needs to be incorporated into more comprehensive efforts to improve clinical outcomes. Studies suggest that bedside rounding can improve outcomes when implemented in the context of comprehensive restructuring of patient care.^4,9 Stein et al.⁴ have described the reorganization of a medical ward as an “accountable care unit.” The ACU model included daily IDRs at the bedside, as well as geographic-based teams, co-leadership by a hospitalist and nurse manager, and unit-level reporting. Although no definitive conclusions can be drawn based on their descriptive report, transformation of the unit was associated with reduced LOS and mortality. Similarly, Kara et al.⁹ found that the number of elements of an “accountable care team” model implemented by each unit was associated with greater reductions in LOS and cost. In contrast, our findings of a lack of an effect are consistent with a recent cluster-randomized trial by O’Leary et al,¹⁰ which found that implementation of patient-centered bedside rounds did not improve patient satisfaction or perceptions of shared decision-making compared to units using a model of structured IDRs in a conference room setting. It is notable that the control groups in both the O’Leary trial¹⁰ and this study did not represent usual care, because these groups featured localization of the clinical teams and high-quality IDR. In our trial, it is plausible that the control side of the unit was functioning at a high level, which would have decreased our ability to further improve outcomes. Whether restructuring unit processes, including implementation of bedside IDR, improves care compared to usual care without these processes is unknown.

We found that the MICRO intervention significantly decreased LOS compared to the control group for patients transferred to the study unit. This analysis was exploratory and the finding was unexpected. Patients were transferred to the study unit from units of higher acuity, and were more likely to have Medicaid or no insurance and be discharged to facilities rather than home, suggesting that these patients had substantial disposition challenges. It is plausible that this is the population for which bedside IDRs may have the greatest impact. This was a secondary analysis, however, and should be considered as hypothesis-generating for future investigations.

Although the impact on outcomes of bedside IDRs is uncertain, potential benefits and practical barriers have been examined. Gonzalo et al.¹¹ surveyed inpatient physicians and nurses at a hospital employing bedside IDRs and found that the benefits ranked the highest were communication, coordination, and teamwork, and the lowest-ranked benefits were related to efficiency and outcomes. The 6 greatest barriers concerned the time required to complete bedside IDR. These results indicate that the time investiture by staff may be a barrier to widespread adoption. More modest changes, such as increasing the structure of standard conference room rounds, may improve care, although the data are mixed. O’Leary et al.¹² assessed the value of a structured approach in a conference room setting, which primarily entailed implementing a checklist for newly admitted patients, and found no difference in LOS. Follow-up studies by these investigators found mixed results on the ability of structured IDR to decrease the incidence of adverse events.^13,14

The results of our AHRQ survey of patient safety culture found that several important aspects of teamwork and safety were perceived as improved by the intervention, including the “overall grade on patient safety.” Other studies have similarly shown increases in teamwork and safety ratings through redesign of IDR. O’Leary et al.¹² surveyed residents and nurses on a unit that implemented a structured, conference room-based IDR and found that providers on the intervention unit rated the teamwork climate higher than providers on the control unit. Our finding that hospitalists and nurses gave higher ratings for IDR being “efficient” and “a good use of my time” on the postintervention survey than the pre-intervention survey suggests that initial concerns about the additional time commitment may be offset by gains in overall efficiency and in development of an environment of enhanced communication, teamwork, and safety.

This study has several limitations. First, the trial may have been underpowered to find small differences between the groups. The trends for decreased LOS and clinical deterioration in the MICRO group may suggest that bedside IDR can provide a small but clinically significant benefit that would be statistically significant only in a larger trial. Second, patients were not randomized to the 2 groups. The impact is diminished, however, because the routine hospital process for assigning patients to the 2 areas in which the groups were located is random and based solely on bed availability. Third, nurses and hospitalists caring for patients in the control group likely experienced improved communication practices from the unit-wide TeamStepps education and from participating in the MICRO protocol when caring for patients on the intervention side of the unit. Fourth, we did not collect data on the effectiveness of communication and are unable to assess the fidelity with which the structured protocol was followed or whether interprofessional communication was fostered or hindered. Lastly, the study was implemented on a nonteaching unit at a single academic medical center. The protocol and the results may not be generalizable to other hospitals or units that include house staff.

In conclusion, transforming IDR from a conference room model to a bedside model did not reduce overall LOS or clinical deterioration on a unit using features of an ACU structure. Although several beneficial effects were noted, including a reduction in LOS for patients transferred to the study unit and higher ratings of the patient safety climate and efficiency of IDR, implementing bedside IDR in this setting has marginal benefit. Future studies should assess whether a comprehensive transformation of the inpatient model of care, including patient-centered bedside IDR, geographic cohorting of teams, and co-leadership, improves outcomes compared to models without these features.

Disclosures

This trial was funded by Medline’s Prevention Above All Discoveries Grant Program. The authors report no financial conflicts of interest.

The care of hospitalized patients requires practitioners from multiple disciplines to assess and communicate the patient’s status in a dynamic manner during hospitalization. Although optimal teamwork is needed for patient care to be delivered reliably and efficiently, care within hospitals is typically delivered in a fragmented manner.¹ A bedside model for daily interdisciplinary rounds (IDR) has been proposed as a method to provide a structured process and engage all team members in a patient-centered, system-of-care delivery.² Specific advantages of convening rounds in the presence of the patient include the ability to directly assess care (eg, presence of a potentially unnecessary urinary catheter), patient engagement in key aspects of their care and disposition, and an increased opportunity for team members to develop a shared understanding of the patient’s views and needs.

Implementing dramatic changes to the workflow of multiple disciplines will require rigorous evidence to support a concerted effort from leadership and buy-in from stakeholders at the front line of patient care. Despite the urgency for evidence, there has been little investigation of this strategy. A systematic review³ identified 30 studies published between 1998 and 2013 addressing interdisciplinary interventions on medical wards, none of which examined a bedside IDR model. In a study performed after the period assessed by the systematic review, Stein et al⁴ described the restructuring of a medical ward as an accountable care unit (ACU), which included a bedside model for rounds by the interdisciplinary team. The change was associated with decreased mortality and length of stay (LOS), although the study did not isolate the impact of rounds or use a concurrent control group and presented aggregate rather than patient-level outcomes. The lack of convincing data may be a reason bedside rounds are not widely employed by hospitals. To provide high-quality evidence, we performed a large, prospective controlled trial comparing a structured bedside model (mobile interdisciplinary care rounds [MICRO]) with standard rounds.

This study took place at the Mount Sinai Hospital, which is a 1171-bed tertiary care academic medical center in New York City, New York. A nonteaching unit offered the ability to use a prospective controlled design. Patients were assigned to the north and the south wings of the unit in a quasi-randomized manner, rather than based on diagnosis or acuity. We transformed IDR to a bedside model on the north side of the unit (MICRO group), while the south side of the unit continued using standard conference room-based IDR (control group). The north and south sides of the unit contain 17 and 14 beds, respectively. During the study period, nurses and hospitalists cared for patients on both sides of the study unit, although on any given day were assigned only patients on 1 side of the unit. The unit uses a clinical microsystem model, which has been defined as “a group of clinicians and staff working together with a shared clinical purpose to provide care for a population of patients,” and has a defined set of characteristics associated with high performance.^5,6 Our microsystem model has incorporated features as described by Stein’s ACU model,⁴ including co-leadership by a hospitalist and a nurse manager, geographic assignment of patients to teams, and unit-level data reports. One hospitalist is assigned geographically to each area of the unit in a 2- to 4-week rotation. Coverage of the unit does not include house staff; patients are primarily assigned to hospitalists working with nurse practitioners. Patients were enrolled prospectively during their initial IDR by a research coordinator. Patient-level data and outcomes were collected prospectively by a research coordinator who attended IDR on the intervention and the control sides of the study unit daily.

Inclusion Criteria

All patients admitted to the medicine service on the study unit were eligible. Patients were greater than 18 years and admitted for an acute medical condition. Patients admitted to another unit and later transferred to the study unit were enrolled at the time of transfer. Patients could be included more than once if hospitalized on the study unit on more than 1 occasion. Most patients were covered by hospitalists, although patients covered by private physicians were included. Patients from other departments, including family medicine, are uncommonly admitted to the unit and were excluded. Patients were also excluded if they were admitted and discharged over the same weekend, because the MICRO rounds occur during weekdays and there was no opportunity to offer the intervention on Saturdays and Sundays.

MICRO Intervention

Interdisciplinary rounds occurred daily at 10:00 am for the control group and at 10:30 am for the MICRO group, and were attended by the hospitalist caring for the majority of patients on the unit, staff nurses, and the unit medical director, nurse manager, social worker, and case manager. Rounds on the control unit focused on the plan of care and disposition but did not follow any set structure and were typically 25 to 30 minutes in duration.

The MICRO rounds occurred at the bedside and followed a structured script (Appendix 1) that was designed to limit discussion of each patient to 3 minutes or less, and included speaking roles for the hospitalist, nurse, and social worker. For private physicians, the nurse practitioner assigned to the patient performed the role of the hospitalist. Rounds were expected to be approximately 50 minutes in duration. Patients were further engaged by asking for their main goal for the day. A patient safety checklist was reviewed. Initially, this task was performed by the nurse manager, who did not verbalize the items unless a deficiency was noted. After 6 months’ experience, this responsibility was given to the staff nurse, who reviewed the checklist verbally as part of the bedside script. Patients were seen daily, including those being discharged later that same day.

Staff and Clinician Education

We developed and implemented a curriculum based on a modified version of the Agency for Healthcare Research and Quality’s TeamStepps^® program to ensure that all team members were provided with the basic principles of communication within the healthcare setting. The curriculum consisted of interactive didactics on essential elements of teamwork, including team structure, communication, situation monitoring, and mutual support, as well as the purpose and structure of the MICRO model. The curriculum was delivered to nurses at 3 monthly staff meetings on the study unit and to hospitalists during 3 hospital medicine grand rounds over a 3-month period. Nurses and physicians providing care on both geographic areas of the study unit received the education program because no group of practitioners was designated to only 1 geographic area.

Outcomes

Primary and Secondary Outcomes

The primary outcomes were clinical deterioration (CD) and length of stay. Clinical deterioration was a composite outcome defined a priori as death; escalation of care (ie, transfer to an intensive care unit, intermediate care unit, or teaching unit); or a hospital-acquired complication (ie, venous thromboembolism, fall, stage III-IV pressure ulcer, catheter-associated urinary tract infection, central-line associated bloodstream infection, or Clostridium difficile-associated diarrhea). The LOS was calculated as the mean LOS with outliers excluded (outliers defined as having a LOS 100 days or longer or 2.5 or more standard deviations from the expected LOS).

Process metrics on IDR, such as the duration of rounds, attendance by members of the interdisciplinary team, the percentage of patients discussed, or the effectiveness of communication, were not collected. We assessed patient satisfaction based on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey.

Patient Safety Culture Survey

To assess the impact on the perceptions of patient safety, we administered the Agency for Healthcare Research and Quality (AHRQ) Hospital Survey on Patient Safety Culture to all staff and clinicians working on both sides of the study unit immediately before and 12 months after implementation of the MICRO model. Results are reported for the AHRQ dimensions that were most relevant to the MICRO intervention: “teamwork within units,” “overall perceptions of safety,” “communication,” “openness,” “overall patient safety grade,” and “handoffs and transitions.” The survey represents pre- and post-comparison. All nurses and hospitalists on both the MICRO and control sides of the study unit had received the TeamStepps curriculum and participated in MICRO rounds by the time of the postintervention survey. We added 3 questions specifically assessing the perception of the efficiency and effectiveness of IDR. Postintervention respondents reflected on their overall impression of IDR, which included their experiences on both sides of the unit, because no group of nurses or hospitalists was exposed only to the MICRO side or the control side of the unit. Responses to survey questions were recorded on a 5-point Likert scale (from “strongly disagree” to “strongly agree” for opinion questions; and “never,” rarely,” sometimes,” “most of the time,” and “always” for frequency questions) and given a score from 1 to 5. The question asking for an overall grade for patient safety was scored from 1 to 5 points corresponding to letter grade choices F, D, C, B, A.

Statistical Analysis

The sample size was based on the estimate of the baseline rate of the primary outcome of CD and the projected decrease by the MICRO intervention. A study using the Global Trigger Tool developed by the Institute for Healthcare Improvement provided a best estimate of 16% as the baseline rate for CD.⁷ A total of 2000 hospitalizations were planned to be included to have a power of at least 80% to detect a 25% reduction in the annual incidence of CD with a 2-tailed type I error rate of 0.05. Comparisons of dichotomous event rates were made using chi square tests at a 2-tailed level for significance of 0.05. The LOS was analyzed using the nonparametric median test and multivariable regression analysis. We used a generalized linear model with gamma distribution and log link for all analyses of LOS, where LOS was the outcome variable, and intervention vs. control unit type was the predictor variable. Age, sex, race, payer, case mix, and comorbidities defined with the Elixhauser algorithm were used as covariates.⁸ We used multivariable logistic regression for analysis of CD, where the dependent variable was CD. Predictor variables included intervention, patient age, sex, race, payer, case mix and comorbidities. Patient satisfaction data were compared using the chi square test. The Student t test for dependent means was used to analyze the patient safety culture survey data.

The study protocol was submitted to the Icahn Mount Sinai School of Medicine’s institutional review board and determined to be exempt from full review.

A total of 2005 hospitalizations were included over the 12-month study period, consisting of 1089 hospitalizations in the MICRO group and 916 in the control group. Bedside and standard IDR were completed daily, Monday through Friday without exception. The demographic characteristics and comorbidities were similar for the 2 groups (Table). Hospitalizations of patients who were initially admitted to another unit and subsequently transferred to the study unit accounted for 11.1% of hospitalizations.

Risk-adjusted LOS was similar for the groups (6.6 vs 7.0 days, P = 0.17, for the MICRO and control groups, respectively). On subgroup analysis, a reduction in LOS was noted for patients transferred to the study unit (10.4 vs 14.0 days, P = 0.02, for the MICRO and control groups, respectively). The LOS was unchanged for patients admitted directly to the study unit (6.0 vs 5.8 days, P = 0.93). There was no difference in the incidence of clinical deterioration for the MICRO or control groups (7.7% vs 9.3%, odds ratio, 0.89; 95% confidence interval, 0.61-1.22, P = 0.46).

The finding of a LOS benefit for the MICRO group limited to patients transferred to the study unit prompted a comparison of patients transferred to the study unit and patients directly admitted to the study unit from the emergency department (Appendix 2). Compared to patients admitted directly to the study unit, patients transferred to the study unit were more likely to have Medicaid or no insurance, more likely to be discharged to a facility, had longer LOS, and were more likely to experience CD.

Patient Satisfaction

There were 175 and 140 responses to the HCAHPS survey for the MICRO and the control groups, respectively. Patients in the MICRO group were more likely to report that “doctors, nurses, or other hospital staff talk with you about whether you would have the help you needed when you left the hospital” (88% vs 78%, P = 0.01). Responses for all other HCAHPS items were similar for the 2 groups.

Clinician/Staff Survey

The response rate was 96% (30 nurses and 17 hospitalists) pre-intervention and 100% (30 nurses and 22 hospitalists) postintervention. Hospitalists and nurses gave significantly higher scores for the dimensions “teamwork within units,” “overall perception of patient safety,” and “patient safety grade” on the postintervention survey compared to the pre-intervention survey (Figure 1). Hospitalists and nurses rated the efficiency of IDR and the ability of IDR to identify safety issues higher on the postintervention survey compared to the pre-intervention survey (Figure 2).

We transformed daily IDR from a standard conference room model to a structured bedside model with scripted roles, and performed a rigorous comparison using patient-level data. Our finding that transforming daily IDR from a standard conference room model to a bedside model did not significantly reduce LOS suggests either that the model is ineffective or needs to be incorporated into more comprehensive efforts to improve clinical outcomes. Studies suggest that bedside rounding can improve outcomes when implemented in the context of comprehensive restructuring of patient care.^4,9 Stein et al.⁴ have described the reorganization of a medical ward as an “accountable care unit.” The ACU model included daily IDRs at the bedside, as well as geographic-based teams, co-leadership by a hospitalist and nurse manager, and unit-level reporting. Although no definitive conclusions can be drawn based on their descriptive report, transformation of the unit was associated with reduced LOS and mortality. Similarly, Kara et al.⁹ found that the number of elements of an “accountable care team” model implemented by each unit was associated with greater reductions in LOS and cost. In contrast, our findings of a lack of an effect are consistent with a recent cluster-randomized trial by O’Leary et al,¹⁰ which found that implementation of patient-centered bedside rounds did not improve patient satisfaction or perceptions of shared decision-making compared to units using a model of structured IDRs in a conference room setting. It is notable that the control groups in both the O’Leary trial¹⁰ and this study did not represent usual care, because these groups featured localization of the clinical teams and high-quality IDR. In our trial, it is plausible that the control side of the unit was functioning at a high level, which would have decreased our ability to further improve outcomes. Whether restructuring unit processes, including implementation of bedside IDR, improves care compared to usual care without these processes is unknown.

We found that the MICRO intervention significantly decreased LOS compared to the control group for patients transferred to the study unit. This analysis was exploratory and the finding was unexpected. Patients were transferred to the study unit from units of higher acuity, and were more likely to have Medicaid or no insurance and be discharged to facilities rather than home, suggesting that these patients had substantial disposition challenges. It is plausible that this is the population for which bedside IDRs may have the greatest impact. This was a secondary analysis, however, and should be considered as hypothesis-generating for future investigations.

Although the impact on outcomes of bedside IDRs is uncertain, potential benefits and practical barriers have been examined. Gonzalo et al.¹¹ surveyed inpatient physicians and nurses at a hospital employing bedside IDRs and found that the benefits ranked the highest were communication, coordination, and teamwork, and the lowest-ranked benefits were related to efficiency and outcomes. The 6 greatest barriers concerned the time required to complete bedside IDR. These results indicate that the time investiture by staff may be a barrier to widespread adoption. More modest changes, such as increasing the structure of standard conference room rounds, may improve care, although the data are mixed. O’Leary et al.¹² assessed the value of a structured approach in a conference room setting, which primarily entailed implementing a checklist for newly admitted patients, and found no difference in LOS. Follow-up studies by these investigators found mixed results on the ability of structured IDR to decrease the incidence of adverse events.^13,14

The results of our AHRQ survey of patient safety culture found that several important aspects of teamwork and safety were perceived as improved by the intervention, including the “overall grade on patient safety.” Other studies have similarly shown increases in teamwork and safety ratings through redesign of IDR. O’Leary et al.¹² surveyed residents and nurses on a unit that implemented a structured, conference room-based IDR and found that providers on the intervention unit rated the teamwork climate higher than providers on the control unit. Our finding that hospitalists and nurses gave higher ratings for IDR being “efficient” and “a good use of my time” on the postintervention survey than the pre-intervention survey suggests that initial concerns about the additional time commitment may be offset by gains in overall efficiency and in development of an environment of enhanced communication, teamwork, and safety.

This study has several limitations. First, the trial may have been underpowered to find small differences between the groups. The trends for decreased LOS and clinical deterioration in the MICRO group may suggest that bedside IDR can provide a small but clinically significant benefit that would be statistically significant only in a larger trial. Second, patients were not randomized to the 2 groups. The impact is diminished, however, because the routine hospital process for assigning patients to the 2 areas in which the groups were located is random and based solely on bed availability. Third, nurses and hospitalists caring for patients in the control group likely experienced improved communication practices from the unit-wide TeamStepps education and from participating in the MICRO protocol when caring for patients on the intervention side of the unit. Fourth, we did not collect data on the effectiveness of communication and are unable to assess the fidelity with which the structured protocol was followed or whether interprofessional communication was fostered or hindered. Lastly, the study was implemented on a nonteaching unit at a single academic medical center. The protocol and the results may not be generalizable to other hospitals or units that include house staff.

In conclusion, transforming IDR from a conference room model to a bedside model did not reduce overall LOS or clinical deterioration on a unit using features of an ACU structure. Although several beneficial effects were noted, including a reduction in LOS for patients transferred to the study unit and higher ratings of the patient safety climate and efficiency of IDR, implementing bedside IDR in this setting has marginal benefit. Future studies should assess whether a comprehensive transformation of the inpatient model of care, including patient-centered bedside IDR, geographic cohorting of teams, and co-leadership, improves outcomes compared to models without these features.

Disclosures

This trial was funded by Medline’s Prevention Above All Discoveries Grant Program. The authors report no financial conflicts of interest.

The care of hospitalized patients requires practitioners from multiple disciplines to assess and communicate the patient’s status in a dynamic manner during hospitalization. Although optimal teamwork is needed for patient care to be delivered reliably and efficiently, care within hospitals is typically delivered in a fragmented manner.¹ A bedside model for daily interdisciplinary rounds (IDR) has been proposed as a method to provide a structured process and engage all team members in a patient-centered, system-of-care delivery.² Specific advantages of convening rounds in the presence of the patient include the ability to directly assess care (eg, presence of a potentially unnecessary urinary catheter), patient engagement in key aspects of their care and disposition, and an increased opportunity for team members to develop a shared understanding of the patient’s views and needs.

Implementing dramatic changes to the workflow of multiple disciplines will require rigorous evidence to support a concerted effort from leadership and buy-in from stakeholders at the front line of patient care. Despite the urgency for evidence, there has been little investigation of this strategy. A systematic review³ identified 30 studies published between 1998 and 2013 addressing interdisciplinary interventions on medical wards, none of which examined a bedside IDR model. In a study performed after the period assessed by the systematic review, Stein et al⁴ described the restructuring of a medical ward as an accountable care unit (ACU), which included a bedside model for rounds by the interdisciplinary team. The change was associated with decreased mortality and length of stay (LOS), although the study did not isolate the impact of rounds or use a concurrent control group and presented aggregate rather than patient-level outcomes. The lack of convincing data may be a reason bedside rounds are not widely employed by hospitals. To provide high-quality evidence, we performed a large, prospective controlled trial comparing a structured bedside model (mobile interdisciplinary care rounds [MICRO]) with standard rounds.

This study took place at the Mount Sinai Hospital, which is a 1171-bed tertiary care academic medical center in New York City, New York. A nonteaching unit offered the ability to use a prospective controlled design. Patients were assigned to the north and the south wings of the unit in a quasi-randomized manner, rather than based on diagnosis or acuity. We transformed IDR to a bedside model on the north side of the unit (MICRO group), while the south side of the unit continued using standard conference room-based IDR (control group). The north and south sides of the unit contain 17 and 14 beds, respectively. During the study period, nurses and hospitalists cared for patients on both sides of the study unit, although on any given day were assigned only patients on 1 side of the unit. The unit uses a clinical microsystem model, which has been defined as “a group of clinicians and staff working together with a shared clinical purpose to provide care for a population of patients,” and has a defined set of characteristics associated with high performance.^5,6 Our microsystem model has incorporated features as described by Stein’s ACU model,⁴ including co-leadership by a hospitalist and a nurse manager, geographic assignment of patients to teams, and unit-level data reports. One hospitalist is assigned geographically to each area of the unit in a 2- to 4-week rotation. Coverage of the unit does not include house staff; patients are primarily assigned to hospitalists working with nurse practitioners. Patients were enrolled prospectively during their initial IDR by a research coordinator. Patient-level data and outcomes were collected prospectively by a research coordinator who attended IDR on the intervention and the control sides of the study unit daily.

Inclusion Criteria

All patients admitted to the medicine service on the study unit were eligible. Patients were greater than 18 years and admitted for an acute medical condition. Patients admitted to another unit and later transferred to the study unit were enrolled at the time of transfer. Patients could be included more than once if hospitalized on the study unit on more than 1 occasion. Most patients were covered by hospitalists, although patients covered by private physicians were included. Patients from other departments, including family medicine, are uncommonly admitted to the unit and were excluded. Patients were also excluded if they were admitted and discharged over the same weekend, because the MICRO rounds occur during weekdays and there was no opportunity to offer the intervention on Saturdays and Sundays.

MICRO Intervention

Interdisciplinary rounds occurred daily at 10:00 am for the control group and at 10:30 am for the MICRO group, and were attended by the hospitalist caring for the majority of patients on the unit, staff nurses, and the unit medical director, nurse manager, social worker, and case manager. Rounds on the control unit focused on the plan of care and disposition but did not follow any set structure and were typically 25 to 30 minutes in duration.

The MICRO rounds occurred at the bedside and followed a structured script (Appendix 1) that was designed to limit discussion of each patient to 3 minutes or less, and included speaking roles for the hospitalist, nurse, and social worker. For private physicians, the nurse practitioner assigned to the patient performed the role of the hospitalist. Rounds were expected to be approximately 50 minutes in duration. Patients were further engaged by asking for their main goal for the day. A patient safety checklist was reviewed. Initially, this task was performed by the nurse manager, who did not verbalize the items unless a deficiency was noted. After 6 months’ experience, this responsibility was given to the staff nurse, who reviewed the checklist verbally as part of the bedside script. Patients were seen daily, including those being discharged later that same day.

Staff and Clinician Education

We developed and implemented a curriculum based on a modified version of the Agency for Healthcare Research and Quality’s TeamStepps^® program to ensure that all team members were provided with the basic principles of communication within the healthcare setting. The curriculum consisted of interactive didactics on essential elements of teamwork, including team structure, communication, situation monitoring, and mutual support, as well as the purpose and structure of the MICRO model. The curriculum was delivered to nurses at 3 monthly staff meetings on the study unit and to hospitalists during 3 hospital medicine grand rounds over a 3-month period. Nurses and physicians providing care on both geographic areas of the study unit received the education program because no group of practitioners was designated to only 1 geographic area.

Outcomes

Primary and Secondary Outcomes

The primary outcomes were clinical deterioration (CD) and length of stay. Clinical deterioration was a composite outcome defined a priori as death; escalation of care (ie, transfer to an intensive care unit, intermediate care unit, or teaching unit); or a hospital-acquired complication (ie, venous thromboembolism, fall, stage III-IV pressure ulcer, catheter-associated urinary tract infection, central-line associated bloodstream infection, or Clostridium difficile-associated diarrhea). The LOS was calculated as the mean LOS with outliers excluded (outliers defined as having a LOS 100 days or longer or 2.5 or more standard deviations from the expected LOS).

Process metrics on IDR, such as the duration of rounds, attendance by members of the interdisciplinary team, the percentage of patients discussed, or the effectiveness of communication, were not collected. We assessed patient satisfaction based on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey.

Patient Safety Culture Survey

To assess the impact on the perceptions of patient safety, we administered the Agency for Healthcare Research and Quality (AHRQ) Hospital Survey on Patient Safety Culture to all staff and clinicians working on both sides of the study unit immediately before and 12 months after implementation of the MICRO model. Results are reported for the AHRQ dimensions that were most relevant to the MICRO intervention: “teamwork within units,” “overall perceptions of safety,” “communication,” “openness,” “overall patient safety grade,” and “handoffs and transitions.” The survey represents pre- and post-comparison. All nurses and hospitalists on both the MICRO and control sides of the study unit had received the TeamStepps curriculum and participated in MICRO rounds by the time of the postintervention survey. We added 3 questions specifically assessing the perception of the efficiency and effectiveness of IDR. Postintervention respondents reflected on their overall impression of IDR, which included their experiences on both sides of the unit, because no group of nurses or hospitalists was exposed only to the MICRO side or the control side of the unit. Responses to survey questions were recorded on a 5-point Likert scale (from “strongly disagree” to “strongly agree” for opinion questions; and “never,” rarely,” sometimes,” “most of the time,” and “always” for frequency questions) and given a score from 1 to 5. The question asking for an overall grade for patient safety was scored from 1 to 5 points corresponding to letter grade choices F, D, C, B, A.

Statistical Analysis

The sample size was based on the estimate of the baseline rate of the primary outcome of CD and the projected decrease by the MICRO intervention. A study using the Global Trigger Tool developed by the Institute for Healthcare Improvement provided a best estimate of 16% as the baseline rate for CD.⁷ A total of 2000 hospitalizations were planned to be included to have a power of at least 80% to detect a 25% reduction in the annual incidence of CD with a 2-tailed type I error rate of 0.05. Comparisons of dichotomous event rates were made using chi square tests at a 2-tailed level for significance of 0.05. The LOS was analyzed using the nonparametric median test and multivariable regression analysis. We used a generalized linear model with gamma distribution and log link for all analyses of LOS, where LOS was the outcome variable, and intervention vs. control unit type was the predictor variable. Age, sex, race, payer, case mix, and comorbidities defined with the Elixhauser algorithm were used as covariates.⁸ We used multivariable logistic regression for analysis of CD, where the dependent variable was CD. Predictor variables included intervention, patient age, sex, race, payer, case mix and comorbidities. Patient satisfaction data were compared using the chi square test. The Student t test for dependent means was used to analyze the patient safety culture survey data.

The study protocol was submitted to the Icahn Mount Sinai School of Medicine’s institutional review board and determined to be exempt from full review.

A total of 2005 hospitalizations were included over the 12-month study period, consisting of 1089 hospitalizations in the MICRO group and 916 in the control group. Bedside and standard IDR were completed daily, Monday through Friday without exception. The demographic characteristics and comorbidities were similar for the 2 groups (Table). Hospitalizations of patients who were initially admitted to another unit and subsequently transferred to the study unit accounted for 11.1% of hospitalizations.

Risk-adjusted LOS was similar for the groups (6.6 vs 7.0 days, P = 0.17, for the MICRO and control groups, respectively). On subgroup analysis, a reduction in LOS was noted for patients transferred to the study unit (10.4 vs 14.0 days, P = 0.02, for the MICRO and control groups, respectively). The LOS was unchanged for patients admitted directly to the study unit (6.0 vs 5.8 days, P = 0.93). There was no difference in the incidence of clinical deterioration for the MICRO or control groups (7.7% vs 9.3%, odds ratio, 0.89; 95% confidence interval, 0.61-1.22, P = 0.46).

The finding of a LOS benefit for the MICRO group limited to patients transferred to the study unit prompted a comparison of patients transferred to the study unit and patients directly admitted to the study unit from the emergency department (Appendix 2). Compared to patients admitted directly to the study unit, patients transferred to the study unit were more likely to have Medicaid or no insurance, more likely to be discharged to a facility, had longer LOS, and were more likely to experience CD.

Patient Satisfaction

There were 175 and 140 responses to the HCAHPS survey for the MICRO and the control groups, respectively. Patients in the MICRO group were more likely to report that “doctors, nurses, or other hospital staff talk with you about whether you would have the help you needed when you left the hospital” (88% vs 78%, P = 0.01). Responses for all other HCAHPS items were similar for the 2 groups.

Clinician/Staff Survey

The response rate was 96% (30 nurses and 17 hospitalists) pre-intervention and 100% (30 nurses and 22 hospitalists) postintervention. Hospitalists and nurses gave significantly higher scores for the dimensions “teamwork within units,” “overall perception of patient safety,” and “patient safety grade” on the postintervention survey compared to the pre-intervention survey (Figure 1). Hospitalists and nurses rated the efficiency of IDR and the ability of IDR to identify safety issues higher on the postintervention survey compared to the pre-intervention survey (Figure 2).

We transformed daily IDR from a standard conference room model to a structured bedside model with scripted roles, and performed a rigorous comparison using patient-level data. Our finding that transforming daily IDR from a standard conference room model to a bedside model did not significantly reduce LOS suggests either that the model is ineffective or needs to be incorporated into more comprehensive efforts to improve clinical outcomes. Studies suggest that bedside rounding can improve outcomes when implemented in the context of comprehensive restructuring of patient care.^4,9 Stein et al.⁴ have described the reorganization of a medical ward as an “accountable care unit.” The ACU model included daily IDRs at the bedside, as well as geographic-based teams, co-leadership by a hospitalist and nurse manager, and unit-level reporting. Although no definitive conclusions can be drawn based on their descriptive report, transformation of the unit was associated with reduced LOS and mortality. Similarly, Kara et al.⁹ found that the number of elements of an “accountable care team” model implemented by each unit was associated with greater reductions in LOS and cost. In contrast, our findings of a lack of an effect are consistent with a recent cluster-randomized trial by O’Leary et al,¹⁰ which found that implementation of patient-centered bedside rounds did not improve patient satisfaction or perceptions of shared decision-making compared to units using a model of structured IDRs in a conference room setting. It is notable that the control groups in both the O’Leary trial¹⁰ and this study did not represent usual care, because these groups featured localization of the clinical teams and high-quality IDR. In our trial, it is plausible that the control side of the unit was functioning at a high level, which would have decreased our ability to further improve outcomes. Whether restructuring unit processes, including implementation of bedside IDR, improves care compared to usual care without these processes is unknown.

We found that the MICRO intervention significantly decreased LOS compared to the control group for patients transferred to the study unit. This analysis was exploratory and the finding was unexpected. Patients were transferred to the study unit from units of higher acuity, and were more likely to have Medicaid or no insurance and be discharged to facilities rather than home, suggesting that these patients had substantial disposition challenges. It is plausible that this is the population for which bedside IDRs may have the greatest impact. This was a secondary analysis, however, and should be considered as hypothesis-generating for future investigations.

Although the impact on outcomes of bedside IDRs is uncertain, potential benefits and practical barriers have been examined. Gonzalo et al.¹¹ surveyed inpatient physicians and nurses at a hospital employing bedside IDRs and found that the benefits ranked the highest were communication, coordination, and teamwork, and the lowest-ranked benefits were related to efficiency and outcomes. The 6 greatest barriers concerned the time required to complete bedside IDR. These results indicate that the time investiture by staff may be a barrier to widespread adoption. More modest changes, such as increasing the structure of standard conference room rounds, may improve care, although the data are mixed. O’Leary et al.¹² assessed the value of a structured approach in a conference room setting, which primarily entailed implementing a checklist for newly admitted patients, and found no difference in LOS. Follow-up studies by these investigators found mixed results on the ability of structured IDR to decrease the incidence of adverse events.^13,14

The results of our AHRQ survey of patient safety culture found that several important aspects of teamwork and safety were perceived as improved by the intervention, including the “overall grade on patient safety.” Other studies have similarly shown increases in teamwork and safety ratings through redesign of IDR. O’Leary et al.¹² surveyed residents and nurses on a unit that implemented a structured, conference room-based IDR and found that providers on the intervention unit rated the teamwork climate higher than providers on the control unit. Our finding that hospitalists and nurses gave higher ratings for IDR being “efficient” and “a good use of my time” on the postintervention survey than the pre-intervention survey suggests that initial concerns about the additional time commitment may be offset by gains in overall efficiency and in development of an environment of enhanced communication, teamwork, and safety.

This study has several limitations. First, the trial may have been underpowered to find small differences between the groups. The trends for decreased LOS and clinical deterioration in the MICRO group may suggest that bedside IDR can provide a small but clinically significant benefit that would be statistically significant only in a larger trial. Second, patients were not randomized to the 2 groups. The impact is diminished, however, because the routine hospital process for assigning patients to the 2 areas in which the groups were located is random and based solely on bed availability. Third, nurses and hospitalists caring for patients in the control group likely experienced improved communication practices from the unit-wide TeamStepps education and from participating in the MICRO protocol when caring for patients on the intervention side of the unit. Fourth, we did not collect data on the effectiveness of communication and are unable to assess the fidelity with which the structured protocol was followed or whether interprofessional communication was fostered or hindered. Lastly, the study was implemented on a nonteaching unit at a single academic medical center. The protocol and the results may not be generalizable to other hospitals or units that include house staff.

In conclusion, transforming IDR from a conference room model to a bedside model did not reduce overall LOS or clinical deterioration on a unit using features of an ACU structure. Although several beneficial effects were noted, including a reduction in LOS for patients transferred to the study unit and higher ratings of the patient safety climate and efficiency of IDR, implementing bedside IDR in this setting has marginal benefit. Future studies should assess whether a comprehensive transformation of the inpatient model of care, including patient-centered bedside IDR, geographic cohorting of teams, and co-leadership, improves outcomes compared to models without these features.

Disclosures

This trial was funded by Medline’s Prevention Above All Discoveries Grant Program. The authors report no financial conflicts of interest.

Attending rounds at academic medical centers are often disconnected from patients and non-physician care team members. Time spent bedside is consistently less than one third of total rounding time, with observational studies reporting a range of 9% to 33% over the past several decades.^1-8 Rounds are often conducted outside patient rooms, denying patients, families, and nurses the opportunity to participate and offer valuable insights. Lack of bedside rounds thus limits patient and family engagement, patient input into the care plan, teaching of the physical examination, and communication and collaboration with nurses. In one study, physicians and nurses on rounds engaged in interprofessional communication in only 12% of patient cases.¹ Studies have found interdisciplinary bedside rounds have several benefits, including subjectively improved communication and teamwork between physicians and nurses; increased patient satisfaction, including feeling more cared for by the medical team; and decreased length of stay and costs of care.^2-10

However, there are many barriers to conducting interdisciplinary bedside rounds at large academic medical centers. Patients cared for by a single medical team are often geographically dispersed to several nursing units, and nurses are unable to predict when physicians will round on their patients. This situation limits nursing involvement on rounds and keeps doctors and nurses isolated from each other.² Regionalization of care teams reduces this fragmentation by facilitating more interaction among doctors, patients, families, and nursing staff.

There are few data on how regionalized patients and interdisciplinary bedside rounds affect rounding time and the nature of rounds. This information is needed to understand how these structural changes mediate their effects, whether other steps are required to optimize outcomes, and how to maximize efficiency. We used time-motion analysis (TMA) to investigate how regionalization of medical teams, encouragement of bedside rounding, and systematic inclusion of nurses on ward rounds affect amount of time spent with patients, nursing presence on rounds, and total rounding time.

Setting

This prospective interventional study, approved by the Institutional Review Board of Partners HealthCare, was conducted on the general medical wards at Brigham and Women’s Hospital, an academic 793-bed tertiary-care center in Boston, Massachusetts. Housestaff teams consist of 1 attending, 1 resident, and 2 interns with or without a medical student. Before June 20, 2013, daily rounds on medical inpatients were conducted largely on the patient unit but outside patient rooms. After completing most of a rounding discussion outside a patient’s room, the team might walk in to examine or speak with the patient. A typical medical team had patients dispersed over 7 medical units on average, and over as many as 13. As nurses were unit based, they did not consistently participate in rounds.

Intervention

In June 2013, as part of a general medical service care redesign initiative, the general medical teams were regionalized to specific inpatient units. The goal was to have teams admit patients predominantly to the team’s designated unit and to have all patients on a unit be cared for by the unit’s assigned team as often as possible, with an 85% goal for both. Toward those ends, the admitting structure was changed from a traditional 4-day call cycle to daily admitting for all teams, based on each unit’s bed availability.¹¹

Teams were also expected to conduct rounds with nurses, and a system for facilitating these rounds was established. As physician and nurse care teams were now geographically co-located, it became possible for residents and nurses to check a rounding sheet for the planned patient rounding order, which had been set by the resident and nurse-in-charge before rounds. No more than about 5 minutes was needed to prepare each day’s order. The rounding sheet prioritized sick patients, newly admitted patients, and planned morning discharges, but patients were also always grouped by nurse. For example, the physician team rounded with the first nurse on all 3 of a nurse’s patients, and then proceeded to the next group of 3 patients with the next nurse, until all patients were seen.

Teams were encouraged to conduct patient- and family-centered rounds exclusively at bedside, except when bedside rounding was thought to be detrimental to a patient (eg, one with delirium). After an intern’s bedside presentation, which included a brief summary and details about overnight events and vital signs, the concerns of the patient, family, and nurse were shared, a focused physical examination performed, relevant data (eg, laboratory test results and imaging studies) reviewed, and the day’s plan formulated. The entire team, including the attending, was expected to have read new patients’ admission notes before rounds. Bedside rounds could thus be focused more on patient assessment and patient/family engagement and less on data transfer.

Several actions were taken to facilitate these changes. Residents, attendings, nurses, and other interdisciplinary team members participated in a series of focus groups and conferences to define workflows and share best practices for patient- and family-centered bedside rounds. Tips on bedside rounding were included in a general medicine rotation guidebook made available to residents and attendings. At the beginning of each post-intervention general medicine rotation, attendings and residents attended brief orientation sessions to review the new daily schedule, have interdisciplinary huddles, and share expectations for patient- and family-centered bedside rounds. On the general medicine units, new medical directors were hired to partner with existing nursing directors to support adoption of the workflows. Last, an interdisciplinary leadership team was formed to support the care redesign efforts. This team started meeting every 2 weeks.

Study Design

We used a pre–post analysis to study the effects of care redesign. Analysis was performed at the same time of year for 2 consecutive years to control for the stage of training and experience of the housestaff. TMA was performed by trained medical students using computer tablets linked to a customized Microsoft Access database form (Redmond, Washington). The form and the database were designed with specific buttons that, when pressed, recorded the time of particular events, such as the coming and going of each participant, the location of rounds, and the beginning and the end of rounding encounters with a patient. One research assistant using an Access entry form was able to dynamically track all events in real time, as they occurred. We collected data on 4 teams at baseline and 5 teams after the intervention. Each of the 4 baseline teams was followed for 4 consecutive weekdays—16 rounds total, April-June 2013—to capture the 4-day call cycle. Each of the 5 post-intervention teams was followed for 5 consecutive weekdays—25 rounds total, April–June 2014—to capture the 5-day cycle. (Because of technical difficulties, data from 1 rounding session were not captured.) For inclusion in the statistical analyses, TMA captured 166 on-service patients before the intervention and 304 afterward. Off-service patients, those with an attending other than the team attending, were excluded because their rounds were conducted separately.

We examined 2 primary outcomes, the proportion of time each clinical team member was present on rounds and the proportion of bedside rounding time. Secondary outcomes were round duration, rounding time per patient, and total non-patient time per rounding session (total rounding time minus total patient time).

Statistical Analysis

TMA data were organized in an Access database and analyzed with SAS Version 9.3 (SAS Institute, Cary, North Carolina). We analyzed the data by round session as well as by patient.

Data are presented as means with standard deviations, medians with interquartile ranges, and proportions, as appropriate. For analyses by round session, we used unadjusted linear regression; for patient-level analyses, we used general estimating equations to adjust for clustering of patients within each session; for nurse presence during any part of a round by patient, we used a χ² test. Total non-patient time per round session was compared with use of patient-clustered general estimating equations using a γ distribution to account for the non-normality of the data.

Patient and Care Team Characteristics

Over the first year of the initiative, 85% of a team’s patients were on their assigned unit, and 87% of a unit’s patients were with the assigned team. Census numbers were 10.4 patients per general medicine team in April-June 2013 and 12.7 patients per team in April-June 2014, a 22% increase after care redesign. There were no statistically significant differences in patient characteristics, including age, sex, race, language, admission source, and comorbidity measure (Elixhauser score), between the pre-intervention and post-intervention study periods, except for a slightly higher proportion of patients admitted from home and fewer patients admitted directly from clinic (Table 1).

Primary Outcomes

Mean proportion of time the nurse was present on rounds per round session increased significantly (P < 0.001), from 24.1% to 67.8% (Figure 1A, Table 2). For individual patient encounters, the increased overall nursing presence was attributable to having more nurses on rounds and having nurses present for a larger proportion of individual rounding encounters (Figure 1B, Table 2). Nurses were present for at least some part of rounds for 53% of patients before the intervention and 93% afterward (P < 0.001). Mean proportion of round time by each of the 2 interns on each team decreased from 59.6% to 49.6% (P = 0.007).

Total bedside rounding time increased significantly ( P < 0.001), from 39.9% before the intervention to 55.8% afterward (Table 2). Meanwhile, percentage of rounding time spent on the unit but outside patient rooms decreased significantly ( P = 0.004), from 55.2% to 42.2%, as did rounding time on a unit completely different from the patient’s (4.9% before intervention, 2.0% afterward; P = 0.03). Again, patient-level results were similar (Figure 2, Table 2), but the decreased time spent on the unit, outside the patient rooms, was not significant.

Secondary Outcomes

Total rounding time decreased significantly, from a mean of 182 minutes (3.0 hours) at baseline to a mean of 146 minutes (2.4 hours) after the intervention, despite the higher post-intervention census. (When adjusted for patient census, the difference increased from 35.5 to 53.8 minutes; Table 2.) Mean rounding time per patient decreased significantly, from 14.7 minutes at baseline to 10.5 minutes after the intervention. For newly admitted patients, mean rounding time per patient decreased from 30.0 minutes before implementation to 16.3 minutes afterward. Mean rounding time also decreased, though much less, for subsequent-day patients (Table 2). For both new and existing patients, the decrease in rounding time largely was a reduction in time spent rounding outside patient rooms, with minimal impact on bedside time (Table 2). Mean time nurses were present during a patient’s rounds increased significantly, from 4.5 to 8.0 minutes (Table 2). Total nurse rounding time increased from 45.1 minutes per session to 98.8 minutes. Rounding time not related to patient discussion or evaluation decreased from 22.7 minutes per session to 13.3 minutes ( P = 0.003).

TMA of our care redesign initiative showed that this multipronged intervention, which included team regionalization, encouragement of bedside rounding with nurses, call structure changes, and attendings’ reading of admission notes before rounds, resulted in an increased proportion of rounding time spent with patients and an increased proportion of time nurses were present on rounds. Secondarily, round duration decreased even as patient census increased.

Regionalized teams have been found to improve interdisciplinary communication.¹ The present study elaborates on that finding by demonstrating a dramatic increase in nursing presence on rounds, likely resulting from the unit’s use of rounding schedules and nurses’ prioritization of rounding orders, both of which were made possible by geographic co-localization. Other research has noted that one of the most significant barriers to interdisciplinary rounds is difficulty coordinating the start times of physician/nurse bedside rounding encounters. The system we have studied directly addresses this difficulty.⁹ Of note, nursing presence on rounds is necessary but not sufficient for true physician–nurse collaboration and effective communication,¹ as reflected in a separate study of the intervention showing no significant difference in the concordance of the patient care plan between nurses and physicians before and after regionalization.¹² Additional interventions may be needed to ensure that communication during bedside rounds is effective.

Our regionalized teams spent a significantly higher proportion of rounding time bedside, likely because of a cultural shift in expectations and the increased convenience of seeing patients on the team’s unit. Nevertheless, bedside time was not 100%. Structural barriers (eg, patients off-unit for dialysis) and cultural barriers likely contributed to the less than full adoption of bedside rounding. As described previously, cultural barriers to bedside rounding include trainees’ anxiety about being questioned in front of patients, the desire to freely exchange academic ideas in a conference room, and attendings’ doubts about their bedside teaching ability.^1,9,13 Bedside rounds provide an important opportunity to apply the principles of patient- and family-centered care, including promotion of dignity and respect, information sharing, and collaboration. Thus, overcoming the concerns of housestaff and attendings and helping them feel prepared for bedside rounds can benefit the patient experience. More attention should be given to these practices as these types of interventions are implemented at Brigham and Women’s Hospital and elsewhere.^1,13-15

Another primary concern about interdisciplinary bedside rounding is the perception that it takes more time.⁹ Therefore, it was important for us to measure round duration as a balancing measure to be considered for our intervention. Fortunately, we found round duration decreased with regionalization and encouragement of bedside rounding. This decrease was driven largely by a significant decrease in mean rounding time per new patient, which may be attributable at least in part to setting expectations that attendings and residents will read admission notes before rounds and that interns will summarize rather than recount information from admission notes. However, we also found rounding time decreases for subsequent-day patients, suggesting an underlying time savings. Spending a larger proportion of time bedside may therefore result in more efficient rounds. Bedside presentations can reduce redundancies, such as discussing a patient’s case outside his or her room and subsequently walking in and going over much of the same information with the patient. Our model de-emphasizes data transfer in favor of discussion of care plans. There was also a decrease in non-patient time, likely reflecting reduced transit time for regionalized teams. This decrease aligns with a recent finding that bedside rounding was at least as efficient as rounding outside the room.¹⁶

Of note, though a larger percentage of time was spent bedside after implementation of the care redesign, the absolute amount of bedside time did not change significantly. Our data showed that, even with shorter rounds, the same amount of absolute time can be spent bedside, face to face with the patient, by increasing the proportion of bedside rounding time. In other words, teams on average did not spend more time with patients, though the content and the structure of those encounters may have changed. This finding may be attributable to eliminating redundancy, forgoing the outside-the-room discussion, and thus the largest time reductions were realized there. In addition, teams incompletely adopted beside rounds, as reflected in the data. We expect that, with more complete adoption, an even larger proportion of time will be spent bedside, and absolute time bedside might increase as a result.

An unexpected result of the care redesign was that interns’ proportion of rounding time decreased after the intervention. This decrease most likely is attributable to interns’ being less likely to participate in rounds for a co-intern’s patient, and to their staying outside that patient’s room to give themselves more time to advance the care of their own patients. Before the intervention, when more rounding time was spent outside patient rooms, interns were more likely to join rounds for their co-intern’s patients because they could easily break away, as needed, to continue care of their own patients. The resident is now encouraged to use the morning huddle to identify which patients likely have the most educational value, and both interns are expected to join the bedside rounds for these patients.

This study had a few limitations. First, the pre–post design made it difficult to exclude the possibility that other temporal changes may have affected outcomes, though we did account for time-of-year effects by aligning our data-collection phases. In addition, the authors, including the director of the general medical service, are unaware of any co-interventions during the study period. Second, the multipronged intervention included care team regionalization, encouragement of bedside rounding with nurses, call structure changes (from 4 days to daily admitting), and attendings’ reading of admission notes before rounds. Thus, parsing which component(s) contributed to the results was difficult, though all the changes instituted likely were necessary for system redesign. For example, regionalization of clinicians to unit-based teams was made possible by switching to a daily admitting system.

Time that team members spent preparing for rounds was not recorded before or after the intervention. Thus, the decrease in total rounding time could have been accompanied by an increase in time spent preparing for rounds. However, admission notes were available in our electronic medical record before and after the intervention, and most residents and attendings were already reading them pre-intervention. After the intervention, pre-round note reading was more clearly defined as an expectation, and we were able to set the expectation that interns should use their presentations to summarize rather than recount information. In addition, in the post-intervention period, we did not include time spent preparing rounding orders; as already noted, however, preparation took only 5 minutes per day. Also, we did not analyze the content or the quality of the discussion on rounds, but simply recorded who was present where and when. Regarding the effect of the intervention on patient care, results were mixed. As reported in 2016, we saw no difference in frequency of adverse events with this intervention.¹² However, a more sensitive measure of adverse events—used in a study on handoffs—showed our regionalization efforts had an additive effect on reducing overnight adverse events.¹⁷Researchers should now focus on the effects of care redesign on clinical outcomes, interdisciplinary care team communication, patient engagement and satisfaction, provider opinions of communication, workflow, patient care, and housestaff education. Our methodology can be used as a model to link structure, process, and outcome related to rounds and thereby better understand how best to optimize patient care and efficiency. Additional studies are needed to analyze the content of rounds and their association with patient and educational outcomes. Last, it will be important to conduct a study to see if the effects we have identified can be sustained. Such a study is already under way.

In conclusion, creating regionalized care teams and encouraging focused bedside rounds increased the proportion of bedside time and the presence of nurses on rounds. Rounds were shorter despite higher patient census. TMA revealed that regionalized care teams and bedside rounding at a large academic hospital are feasible, and are useful in establishing the necessary structures for increasing physician–nurse and provider–patient interactions.

Acknowledgments

The authors acknowledge Dr. Stan Ashley, Dr. Jacqueline Somerville, and Sheila Harris for their support of the regionalization initiative.

Disclosures

Dr. Schnipper received funding from Sanofi-aventis to conduct an investigator-initiated study to implement and evaluate a multi-faceted intervention to improve transitions of care in patients discharged home on insulin. The study was also supported by funding from the Marshall A. Wolf Medical Education Fund, Brigham and Women’s Hospital, and Dr. Stan Ashley, Chief Medical Officer, Brigham and Women’s Hospital. Some of the content of this article was orally presented at the annual meeting of the Society of Hospital Medicine; March 29-April 1, 2015; National Harbor, MD.

Attending rounds at academic medical centers are often disconnected from patients and non-physician care team members. Time spent bedside is consistently less than one third of total rounding time, with observational studies reporting a range of 9% to 33% over the past several decades.^1-8 Rounds are often conducted outside patient rooms, denying patients, families, and nurses the opportunity to participate and offer valuable insights. Lack of bedside rounds thus limits patient and family engagement, patient input into the care plan, teaching of the physical examination, and communication and collaboration with nurses. In one study, physicians and nurses on rounds engaged in interprofessional communication in only 12% of patient cases.¹ Studies have found interdisciplinary bedside rounds have several benefits, including subjectively improved communication and teamwork between physicians and nurses; increased patient satisfaction, including feeling more cared for by the medical team; and decreased length of stay and costs of care.^2-10

However, there are many barriers to conducting interdisciplinary bedside rounds at large academic medical centers. Patients cared for by a single medical team are often geographically dispersed to several nursing units, and nurses are unable to predict when physicians will round on their patients. This situation limits nursing involvement on rounds and keeps doctors and nurses isolated from each other.² Regionalization of care teams reduces this fragmentation by facilitating more interaction among doctors, patients, families, and nursing staff.

There are few data on how regionalized patients and interdisciplinary bedside rounds affect rounding time and the nature of rounds. This information is needed to understand how these structural changes mediate their effects, whether other steps are required to optimize outcomes, and how to maximize efficiency. We used time-motion analysis (TMA) to investigate how regionalization of medical teams, encouragement of bedside rounding, and systematic inclusion of nurses on ward rounds affect amount of time spent with patients, nursing presence on rounds, and total rounding time.

Setting

This prospective interventional study, approved by the Institutional Review Board of Partners HealthCare, was conducted on the general medical wards at Brigham and Women’s Hospital, an academic 793-bed tertiary-care center in Boston, Massachusetts. Housestaff teams consist of 1 attending, 1 resident, and 2 interns with or without a medical student. Before June 20, 2013, daily rounds on medical inpatients were conducted largely on the patient unit but outside patient rooms. After completing most of a rounding discussion outside a patient’s room, the team might walk in to examine or speak with the patient. A typical medical team had patients dispersed over 7 medical units on average, and over as many as 13. As nurses were unit based, they did not consistently participate in rounds.

Intervention

In June 2013, as part of a general medical service care redesign initiative, the general medical teams were regionalized to specific inpatient units. The goal was to have teams admit patients predominantly to the team’s designated unit and to have all patients on a unit be cared for by the unit’s assigned team as often as possible, with an 85% goal for both. Toward those ends, the admitting structure was changed from a traditional 4-day call cycle to daily admitting for all teams, based on each unit’s bed availability.¹¹

Teams were also expected to conduct rounds with nurses, and a system for facilitating these rounds was established. As physician and nurse care teams were now geographically co-located, it became possible for residents and nurses to check a rounding sheet for the planned patient rounding order, which had been set by the resident and nurse-in-charge before rounds. No more than about 5 minutes was needed to prepare each day’s order. The rounding sheet prioritized sick patients, newly admitted patients, and planned morning discharges, but patients were also always grouped by nurse. For example, the physician team rounded with the first nurse on all 3 of a nurse’s patients, and then proceeded to the next group of 3 patients with the next nurse, until all patients were seen.

Teams were encouraged to conduct patient- and family-centered rounds exclusively at bedside, except when bedside rounding was thought to be detrimental to a patient (eg, one with delirium). After an intern’s bedside presentation, which included a brief summary and details about overnight events and vital signs, the concerns of the patient, family, and nurse were shared, a focused physical examination performed, relevant data (eg, laboratory test results and imaging studies) reviewed, and the day’s plan formulated. The entire team, including the attending, was expected to have read new patients’ admission notes before rounds. Bedside rounds could thus be focused more on patient assessment and patient/family engagement and less on data transfer.

Several actions were taken to facilitate these changes. Residents, attendings, nurses, and other interdisciplinary team members participated in a series of focus groups and conferences to define workflows and share best practices for patient- and family-centered bedside rounds. Tips on bedside rounding were included in a general medicine rotation guidebook made available to residents and attendings. At the beginning of each post-intervention general medicine rotation, attendings and residents attended brief orientation sessions to review the new daily schedule, have interdisciplinary huddles, and share expectations for patient- and family-centered bedside rounds. On the general medicine units, new medical directors were hired to partner with existing nursing directors to support adoption of the workflows. Last, an interdisciplinary leadership team was formed to support the care redesign efforts. This team started meeting every 2 weeks.

Study Design

We used a pre–post analysis to study the effects of care redesign. Analysis was performed at the same time of year for 2 consecutive years to control for the stage of training and experience of the housestaff. TMA was performed by trained medical students using computer tablets linked to a customized Microsoft Access database form (Redmond, Washington). The form and the database were designed with specific buttons that, when pressed, recorded the time of particular events, such as the coming and going of each participant, the location of rounds, and the beginning and the end of rounding encounters with a patient. One research assistant using an Access entry form was able to dynamically track all events in real time, as they occurred. We collected data on 4 teams at baseline and 5 teams after the intervention. Each of the 4 baseline teams was followed for 4 consecutive weekdays—16 rounds total, April-June 2013—to capture the 4-day call cycle. Each of the 5 post-intervention teams was followed for 5 consecutive weekdays—25 rounds total, April–June 2014—to capture the 5-day cycle. (Because of technical difficulties, data from 1 rounding session were not captured.) For inclusion in the statistical analyses, TMA captured 166 on-service patients before the intervention and 304 afterward. Off-service patients, those with an attending other than the team attending, were excluded because their rounds were conducted separately.

We examined 2 primary outcomes, the proportion of time each clinical team member was present on rounds and the proportion of bedside rounding time. Secondary outcomes were round duration, rounding time per patient, and total non-patient time per rounding session (total rounding time minus total patient time).

Statistical Analysis

TMA data were organized in an Access database and analyzed with SAS Version 9.3 (SAS Institute, Cary, North Carolina). We analyzed the data by round session as well as by patient.

Data are presented as means with standard deviations, medians with interquartile ranges, and proportions, as appropriate. For analyses by round session, we used unadjusted linear regression; for patient-level analyses, we used general estimating equations to adjust for clustering of patients within each session; for nurse presence during any part of a round by patient, we used a χ² test. Total non-patient time per round session was compared with use of patient-clustered general estimating equations using a γ distribution to account for the non-normality of the data.

Patient and Care Team Characteristics

Over the first year of the initiative, 85% of a team’s patients were on their assigned unit, and 87% of a unit’s patients were with the assigned team. Census numbers were 10.4 patients per general medicine team in April-June 2013 and 12.7 patients per team in April-June 2014, a 22% increase after care redesign. There were no statistically significant differences in patient characteristics, including age, sex, race, language, admission source, and comorbidity measure (Elixhauser score), between the pre-intervention and post-intervention study periods, except for a slightly higher proportion of patients admitted from home and fewer patients admitted directly from clinic (Table 1).

Primary Outcomes

Mean proportion of time the nurse was present on rounds per round session increased significantly (P < 0.001), from 24.1% to 67.8% (Figure 1A, Table 2). For individual patient encounters, the increased overall nursing presence was attributable to having more nurses on rounds and having nurses present for a larger proportion of individual rounding encounters (Figure 1B, Table 2). Nurses were present for at least some part of rounds for 53% of patients before the intervention and 93% afterward (P < 0.001). Mean proportion of round time by each of the 2 interns on each team decreased from 59.6% to 49.6% (P = 0.007).

Total bedside rounding time increased significantly ( P < 0.001), from 39.9% before the intervention to 55.8% afterward (Table 2). Meanwhile, percentage of rounding time spent on the unit but outside patient rooms decreased significantly ( P = 0.004), from 55.2% to 42.2%, as did rounding time on a unit completely different from the patient’s (4.9% before intervention, 2.0% afterward; P = 0.03). Again, patient-level results were similar (Figure 2, Table 2), but the decreased time spent on the unit, outside the patient rooms, was not significant.

Secondary Outcomes

Total rounding time decreased significantly, from a mean of 182 minutes (3.0 hours) at baseline to a mean of 146 minutes (2.4 hours) after the intervention, despite the higher post-intervention census. (When adjusted for patient census, the difference increased from 35.5 to 53.8 minutes; Table 2.) Mean rounding time per patient decreased significantly, from 14.7 minutes at baseline to 10.5 minutes after the intervention. For newly admitted patients, mean rounding time per patient decreased from 30.0 minutes before implementation to 16.3 minutes afterward. Mean rounding time also decreased, though much less, for subsequent-day patients (Table 2). For both new and existing patients, the decrease in rounding time largely was a reduction in time spent rounding outside patient rooms, with minimal impact on bedside time (Table 2). Mean time nurses were present during a patient’s rounds increased significantly, from 4.5 to 8.0 minutes (Table 2). Total nurse rounding time increased from 45.1 minutes per session to 98.8 minutes. Rounding time not related to patient discussion or evaluation decreased from 22.7 minutes per session to 13.3 minutes ( P = 0.003).

TMA of our care redesign initiative showed that this multipronged intervention, which included team regionalization, encouragement of bedside rounding with nurses, call structure changes, and attendings’ reading of admission notes before rounds, resulted in an increased proportion of rounding time spent with patients and an increased proportion of time nurses were present on rounds. Secondarily, round duration decreased even as patient census increased.

Regionalized teams have been found to improve interdisciplinary communication.¹ The present study elaborates on that finding by demonstrating a dramatic increase in nursing presence on rounds, likely resulting from the unit’s use of rounding schedules and nurses’ prioritization of rounding orders, both of which were made possible by geographic co-localization. Other research has noted that one of the most significant barriers to interdisciplinary rounds is difficulty coordinating the start times of physician/nurse bedside rounding encounters. The system we have studied directly addresses this difficulty.⁹ Of note, nursing presence on rounds is necessary but not sufficient for true physician–nurse collaboration and effective communication,¹ as reflected in a separate study of the intervention showing no significant difference in the concordance of the patient care plan between nurses and physicians before and after regionalization.¹² Additional interventions may be needed to ensure that communication during bedside rounds is effective.

Our regionalized teams spent a significantly higher proportion of rounding time bedside, likely because of a cultural shift in expectations and the increased convenience of seeing patients on the team’s unit. Nevertheless, bedside time was not 100%. Structural barriers (eg, patients off-unit for dialysis) and cultural barriers likely contributed to the less than full adoption of bedside rounding. As described previously, cultural barriers to bedside rounding include trainees’ anxiety about being questioned in front of patients, the desire to freely exchange academic ideas in a conference room, and attendings’ doubts about their bedside teaching ability.^1,9,13 Bedside rounds provide an important opportunity to apply the principles of patient- and family-centered care, including promotion of dignity and respect, information sharing, and collaboration. Thus, overcoming the concerns of housestaff and attendings and helping them feel prepared for bedside rounds can benefit the patient experience. More attention should be given to these practices as these types of interventions are implemented at Brigham and Women’s Hospital and elsewhere.^1,13-15

Another primary concern about interdisciplinary bedside rounding is the perception that it takes more time.⁹ Therefore, it was important for us to measure round duration as a balancing measure to be considered for our intervention. Fortunately, we found round duration decreased with regionalization and encouragement of bedside rounding. This decrease was driven largely by a significant decrease in mean rounding time per new patient, which may be attributable at least in part to setting expectations that attendings and residents will read admission notes before rounds and that interns will summarize rather than recount information from admission notes. However, we also found rounding time decreases for subsequent-day patients, suggesting an underlying time savings. Spending a larger proportion of time bedside may therefore result in more efficient rounds. Bedside presentations can reduce redundancies, such as discussing a patient’s case outside his or her room and subsequently walking in and going over much of the same information with the patient. Our model de-emphasizes data transfer in favor of discussion of care plans. There was also a decrease in non-patient time, likely reflecting reduced transit time for regionalized teams. This decrease aligns with a recent finding that bedside rounding was at least as efficient as rounding outside the room.¹⁶

Of note, though a larger percentage of time was spent bedside after implementation of the care redesign, the absolute amount of bedside time did not change significantly. Our data showed that, even with shorter rounds, the same amount of absolute time can be spent bedside, face to face with the patient, by increasing the proportion of bedside rounding time. In other words, teams on average did not spend more time with patients, though the content and the structure of those encounters may have changed. This finding may be attributable to eliminating redundancy, forgoing the outside-the-room discussion, and thus the largest time reductions were realized there. In addition, teams incompletely adopted beside rounds, as reflected in the data. We expect that, with more complete adoption, an even larger proportion of time will be spent bedside, and absolute time bedside might increase as a result.

An unexpected result of the care redesign was that interns’ proportion of rounding time decreased after the intervention. This decrease most likely is attributable to interns’ being less likely to participate in rounds for a co-intern’s patient, and to their staying outside that patient’s room to give themselves more time to advance the care of their own patients. Before the intervention, when more rounding time was spent outside patient rooms, interns were more likely to join rounds for their co-intern’s patients because they could easily break away, as needed, to continue care of their own patients. The resident is now encouraged to use the morning huddle to identify which patients likely have the most educational value, and both interns are expected to join the bedside rounds for these patients.

This study had a few limitations. First, the pre–post design made it difficult to exclude the possibility that other temporal changes may have affected outcomes, though we did account for time-of-year effects by aligning our data-collection phases. In addition, the authors, including the director of the general medical service, are unaware of any co-interventions during the study period. Second, the multipronged intervention included care team regionalization, encouragement of bedside rounding with nurses, call structure changes (from 4 days to daily admitting), and attendings’ reading of admission notes before rounds. Thus, parsing which component(s) contributed to the results was difficult, though all the changes instituted likely were necessary for system redesign. For example, regionalization of clinicians to unit-based teams was made possible by switching to a daily admitting system.

Time that team members spent preparing for rounds was not recorded before or after the intervention. Thus, the decrease in total rounding time could have been accompanied by an increase in time spent preparing for rounds. However, admission notes were available in our electronic medical record before and after the intervention, and most residents and attendings were already reading them pre-intervention. After the intervention, pre-round note reading was more clearly defined as an expectation, and we were able to set the expectation that interns should use their presentations to summarize rather than recount information. In addition, in the post-intervention period, we did not include time spent preparing rounding orders; as already noted, however, preparation took only 5 minutes per day. Also, we did not analyze the content or the quality of the discussion on rounds, but simply recorded who was present where and when. Regarding the effect of the intervention on patient care, results were mixed. As reported in 2016, we saw no difference in frequency of adverse events with this intervention.¹² However, a more sensitive measure of adverse events—used in a study on handoffs—showed our regionalization efforts had an additive effect on reducing overnight adverse events.¹⁷Researchers should now focus on the effects of care redesign on clinical outcomes, interdisciplinary care team communication, patient engagement and satisfaction, provider opinions of communication, workflow, patient care, and housestaff education. Our methodology can be used as a model to link structure, process, and outcome related to rounds and thereby better understand how best to optimize patient care and efficiency. Additional studies are needed to analyze the content of rounds and their association with patient and educational outcomes. Last, it will be important to conduct a study to see if the effects we have identified can be sustained. Such a study is already under way.

In conclusion, creating regionalized care teams and encouraging focused bedside rounds increased the proportion of bedside time and the presence of nurses on rounds. Rounds were shorter despite higher patient census. TMA revealed that regionalized care teams and bedside rounding at a large academic hospital are feasible, and are useful in establishing the necessary structures for increasing physician–nurse and provider–patient interactions.

Acknowledgments

The authors acknowledge Dr. Stan Ashley, Dr. Jacqueline Somerville, and Sheila Harris for their support of the regionalization initiative.

Disclosures

Dr. Schnipper received funding from Sanofi-aventis to conduct an investigator-initiated study to implement and evaluate a multi-faceted intervention to improve transitions of care in patients discharged home on insulin. The study was also supported by funding from the Marshall A. Wolf Medical Education Fund, Brigham and Women’s Hospital, and Dr. Stan Ashley, Chief Medical Officer, Brigham and Women’s Hospital. Some of the content of this article was orally presented at the annual meeting of the Society of Hospital Medicine; March 29-April 1, 2015; National Harbor, MD.

Attending rounds at academic medical centers are often disconnected from patients and non-physician care team members. Time spent bedside is consistently less than one third of total rounding time, with observational studies reporting a range of 9% to 33% over the past several decades.^1-8 Rounds are often conducted outside patient rooms, denying patients, families, and nurses the opportunity to participate and offer valuable insights. Lack of bedside rounds thus limits patient and family engagement, patient input into the care plan, teaching of the physical examination, and communication and collaboration with nurses. In one study, physicians and nurses on rounds engaged in interprofessional communication in only 12% of patient cases.¹ Studies have found interdisciplinary bedside rounds have several benefits, including subjectively improved communication and teamwork between physicians and nurses; increased patient satisfaction, including feeling more cared for by the medical team; and decreased length of stay and costs of care.^2-10

However, there are many barriers to conducting interdisciplinary bedside rounds at large academic medical centers. Patients cared for by a single medical team are often geographically dispersed to several nursing units, and nurses are unable to predict when physicians will round on their patients. This situation limits nursing involvement on rounds and keeps doctors and nurses isolated from each other.² Regionalization of care teams reduces this fragmentation by facilitating more interaction among doctors, patients, families, and nursing staff.

There are few data on how regionalized patients and interdisciplinary bedside rounds affect rounding time and the nature of rounds. This information is needed to understand how these structural changes mediate their effects, whether other steps are required to optimize outcomes, and how to maximize efficiency. We used time-motion analysis (TMA) to investigate how regionalization of medical teams, encouragement of bedside rounding, and systematic inclusion of nurses on ward rounds affect amount of time spent with patients, nursing presence on rounds, and total rounding time.

Setting

This prospective interventional study, approved by the Institutional Review Board of Partners HealthCare, was conducted on the general medical wards at Brigham and Women’s Hospital, an academic 793-bed tertiary-care center in Boston, Massachusetts. Housestaff teams consist of 1 attending, 1 resident, and 2 interns with or without a medical student. Before June 20, 2013, daily rounds on medical inpatients were conducted largely on the patient unit but outside patient rooms. After completing most of a rounding discussion outside a patient’s room, the team might walk in to examine or speak with the patient. A typical medical team had patients dispersed over 7 medical units on average, and over as many as 13. As nurses were unit based, they did not consistently participate in rounds.

Intervention

In June 2013, as part of a general medical service care redesign initiative, the general medical teams were regionalized to specific inpatient units. The goal was to have teams admit patients predominantly to the team’s designated unit and to have all patients on a unit be cared for by the unit’s assigned team as often as possible, with an 85% goal for both. Toward those ends, the admitting structure was changed from a traditional 4-day call cycle to daily admitting for all teams, based on each unit’s bed availability.¹¹

Teams were also expected to conduct rounds with nurses, and a system for facilitating these rounds was established. As physician and nurse care teams were now geographically co-located, it became possible for residents and nurses to check a rounding sheet for the planned patient rounding order, which had been set by the resident and nurse-in-charge before rounds. No more than about 5 minutes was needed to prepare each day’s order. The rounding sheet prioritized sick patients, newly admitted patients, and planned morning discharges, but patients were also always grouped by nurse. For example, the physician team rounded with the first nurse on all 3 of a nurse’s patients, and then proceeded to the next group of 3 patients with the next nurse, until all patients were seen.

Teams were encouraged to conduct patient- and family-centered rounds exclusively at bedside, except when bedside rounding was thought to be detrimental to a patient (eg, one with delirium). After an intern’s bedside presentation, which included a brief summary and details about overnight events and vital signs, the concerns of the patient, family, and nurse were shared, a focused physical examination performed, relevant data (eg, laboratory test results and imaging studies) reviewed, and the day’s plan formulated. The entire team, including the attending, was expected to have read new patients’ admission notes before rounds. Bedside rounds could thus be focused more on patient assessment and patient/family engagement and less on data transfer.

Several actions were taken to facilitate these changes. Residents, attendings, nurses, and other interdisciplinary team members participated in a series of focus groups and conferences to define workflows and share best practices for patient- and family-centered bedside rounds. Tips on bedside rounding were included in a general medicine rotation guidebook made available to residents and attendings. At the beginning of each post-intervention general medicine rotation, attendings and residents attended brief orientation sessions to review the new daily schedule, have interdisciplinary huddles, and share expectations for patient- and family-centered bedside rounds. On the general medicine units, new medical directors were hired to partner with existing nursing directors to support adoption of the workflows. Last, an interdisciplinary leadership team was formed to support the care redesign efforts. This team started meeting every 2 weeks.

Study Design

We used a pre–post analysis to study the effects of care redesign. Analysis was performed at the same time of year for 2 consecutive years to control for the stage of training and experience of the housestaff. TMA was performed by trained medical students using computer tablets linked to a customized Microsoft Access database form (Redmond, Washington). The form and the database were designed with specific buttons that, when pressed, recorded the time of particular events, such as the coming and going of each participant, the location of rounds, and the beginning and the end of rounding encounters with a patient. One research assistant using an Access entry form was able to dynamically track all events in real time, as they occurred. We collected data on 4 teams at baseline and 5 teams after the intervention. Each of the 4 baseline teams was followed for 4 consecutive weekdays—16 rounds total, April-June 2013—to capture the 4-day call cycle. Each of the 5 post-intervention teams was followed for 5 consecutive weekdays—25 rounds total, April–June 2014—to capture the 5-day cycle. (Because of technical difficulties, data from 1 rounding session were not captured.) For inclusion in the statistical analyses, TMA captured 166 on-service patients before the intervention and 304 afterward. Off-service patients, those with an attending other than the team attending, were excluded because their rounds were conducted separately.

We examined 2 primary outcomes, the proportion of time each clinical team member was present on rounds and the proportion of bedside rounding time. Secondary outcomes were round duration, rounding time per patient, and total non-patient time per rounding session (total rounding time minus total patient time).

Statistical Analysis

TMA data were organized in an Access database and analyzed with SAS Version 9.3 (SAS Institute, Cary, North Carolina). We analyzed the data by round session as well as by patient.

Data are presented as means with standard deviations, medians with interquartile ranges, and proportions, as appropriate. For analyses by round session, we used unadjusted linear regression; for patient-level analyses, we used general estimating equations to adjust for clustering of patients within each session; for nurse presence during any part of a round by patient, we used a χ² test. Total non-patient time per round session was compared with use of patient-clustered general estimating equations using a γ distribution to account for the non-normality of the data.

Patient and Care Team Characteristics

Over the first year of the initiative, 85% of a team’s patients were on their assigned unit, and 87% of a unit’s patients were with the assigned team. Census numbers were 10.4 patients per general medicine team in April-June 2013 and 12.7 patients per team in April-June 2014, a 22% increase after care redesign. There were no statistically significant differences in patient characteristics, including age, sex, race, language, admission source, and comorbidity measure (Elixhauser score), between the pre-intervention and post-intervention study periods, except for a slightly higher proportion of patients admitted from home and fewer patients admitted directly from clinic (Table 1).

Primary Outcomes

Mean proportion of time the nurse was present on rounds per round session increased significantly (P < 0.001), from 24.1% to 67.8% (Figure 1A, Table 2). For individual patient encounters, the increased overall nursing presence was attributable to having more nurses on rounds and having nurses present for a larger proportion of individual rounding encounters (Figure 1B, Table 2). Nurses were present for at least some part of rounds for 53% of patients before the intervention and 93% afterward (P < 0.001). Mean proportion of round time by each of the 2 interns on each team decreased from 59.6% to 49.6% (P = 0.007).

Total bedside rounding time increased significantly ( P < 0.001), from 39.9% before the intervention to 55.8% afterward (Table 2). Meanwhile, percentage of rounding time spent on the unit but outside patient rooms decreased significantly ( P = 0.004), from 55.2% to 42.2%, as did rounding time on a unit completely different from the patient’s (4.9% before intervention, 2.0% afterward; P = 0.03). Again, patient-level results were similar (Figure 2, Table 2), but the decreased time spent on the unit, outside the patient rooms, was not significant.

Secondary Outcomes

Total rounding time decreased significantly, from a mean of 182 minutes (3.0 hours) at baseline to a mean of 146 minutes (2.4 hours) after the intervention, despite the higher post-intervention census. (When adjusted for patient census, the difference increased from 35.5 to 53.8 minutes; Table 2.) Mean rounding time per patient decreased significantly, from 14.7 minutes at baseline to 10.5 minutes after the intervention. For newly admitted patients, mean rounding time per patient decreased from 30.0 minutes before implementation to 16.3 minutes afterward. Mean rounding time also decreased, though much less, for subsequent-day patients (Table 2). For both new and existing patients, the decrease in rounding time largely was a reduction in time spent rounding outside patient rooms, with minimal impact on bedside time (Table 2). Mean time nurses were present during a patient’s rounds increased significantly, from 4.5 to 8.0 minutes (Table 2). Total nurse rounding time increased from 45.1 minutes per session to 98.8 minutes. Rounding time not related to patient discussion or evaluation decreased from 22.7 minutes per session to 13.3 minutes ( P = 0.003).

TMA of our care redesign initiative showed that this multipronged intervention, which included team regionalization, encouragement of bedside rounding with nurses, call structure changes, and attendings’ reading of admission notes before rounds, resulted in an increased proportion of rounding time spent with patients and an increased proportion of time nurses were present on rounds. Secondarily, round duration decreased even as patient census increased.

Regionalized teams have been found to improve interdisciplinary communication.¹ The present study elaborates on that finding by demonstrating a dramatic increase in nursing presence on rounds, likely resulting from the unit’s use of rounding schedules and nurses’ prioritization of rounding orders, both of which were made possible by geographic co-localization. Other research has noted that one of the most significant barriers to interdisciplinary rounds is difficulty coordinating the start times of physician/nurse bedside rounding encounters. The system we have studied directly addresses this difficulty.⁹ Of note, nursing presence on rounds is necessary but not sufficient for true physician–nurse collaboration and effective communication,¹ as reflected in a separate study of the intervention showing no significant difference in the concordance of the patient care plan between nurses and physicians before and after regionalization.¹² Additional interventions may be needed to ensure that communication during bedside rounds is effective.

Our regionalized teams spent a significantly higher proportion of rounding time bedside, likely because of a cultural shift in expectations and the increased convenience of seeing patients on the team’s unit. Nevertheless, bedside time was not 100%. Structural barriers (eg, patients off-unit for dialysis) and cultural barriers likely contributed to the less than full adoption of bedside rounding. As described previously, cultural barriers to bedside rounding include trainees’ anxiety about being questioned in front of patients, the desire to freely exchange academic ideas in a conference room, and attendings’ doubts about their bedside teaching ability.^1,9,13 Bedside rounds provide an important opportunity to apply the principles of patient- and family-centered care, including promotion of dignity and respect, information sharing, and collaboration. Thus, overcoming the concerns of housestaff and attendings and helping them feel prepared for bedside rounds can benefit the patient experience. More attention should be given to these practices as these types of interventions are implemented at Brigham and Women’s Hospital and elsewhere.^1,13-15

Another primary concern about interdisciplinary bedside rounding is the perception that it takes more time.⁹ Therefore, it was important for us to measure round duration as a balancing measure to be considered for our intervention. Fortunately, we found round duration decreased with regionalization and encouragement of bedside rounding. This decrease was driven largely by a significant decrease in mean rounding time per new patient, which may be attributable at least in part to setting expectations that attendings and residents will read admission notes before rounds and that interns will summarize rather than recount information from admission notes. However, we also found rounding time decreases for subsequent-day patients, suggesting an underlying time savings. Spending a larger proportion of time bedside may therefore result in more efficient rounds. Bedside presentations can reduce redundancies, such as discussing a patient’s case outside his or her room and subsequently walking in and going over much of the same information with the patient. Our model de-emphasizes data transfer in favor of discussion of care plans. There was also a decrease in non-patient time, likely reflecting reduced transit time for regionalized teams. This decrease aligns with a recent finding that bedside rounding was at least as efficient as rounding outside the room.¹⁶

Of note, though a larger percentage of time was spent bedside after implementation of the care redesign, the absolute amount of bedside time did not change significantly. Our data showed that, even with shorter rounds, the same amount of absolute time can be spent bedside, face to face with the patient, by increasing the proportion of bedside rounding time. In other words, teams on average did not spend more time with patients, though the content and the structure of those encounters may have changed. This finding may be attributable to eliminating redundancy, forgoing the outside-the-room discussion, and thus the largest time reductions were realized there. In addition, teams incompletely adopted beside rounds, as reflected in the data. We expect that, with more complete adoption, an even larger proportion of time will be spent bedside, and absolute time bedside might increase as a result.

An unexpected result of the care redesign was that interns’ proportion of rounding time decreased after the intervention. This decrease most likely is attributable to interns’ being less likely to participate in rounds for a co-intern’s patient, and to their staying outside that patient’s room to give themselves more time to advance the care of their own patients. Before the intervention, when more rounding time was spent outside patient rooms, interns were more likely to join rounds for their co-intern’s patients because they could easily break away, as needed, to continue care of their own patients. The resident is now encouraged to use the morning huddle to identify which patients likely have the most educational value, and both interns are expected to join the bedside rounds for these patients.

This study had a few limitations. First, the pre–post design made it difficult to exclude the possibility that other temporal changes may have affected outcomes, though we did account for time-of-year effects by aligning our data-collection phases. In addition, the authors, including the director of the general medical service, are unaware of any co-interventions during the study period. Second, the multipronged intervention included care team regionalization, encouragement of bedside rounding with nurses, call structure changes (from 4 days to daily admitting), and attendings’ reading of admission notes before rounds. Thus, parsing which component(s) contributed to the results was difficult, though all the changes instituted likely were necessary for system redesign. For example, regionalization of clinicians to unit-based teams was made possible by switching to a daily admitting system.

Time that team members spent preparing for rounds was not recorded before or after the intervention. Thus, the decrease in total rounding time could have been accompanied by an increase in time spent preparing for rounds. However, admission notes were available in our electronic medical record before and after the intervention, and most residents and attendings were already reading them pre-intervention. After the intervention, pre-round note reading was more clearly defined as an expectation, and we were able to set the expectation that interns should use their presentations to summarize rather than recount information. In addition, in the post-intervention period, we did not include time spent preparing rounding orders; as already noted, however, preparation took only 5 minutes per day. Also, we did not analyze the content or the quality of the discussion on rounds, but simply recorded who was present where and when. Regarding the effect of the intervention on patient care, results were mixed. As reported in 2016, we saw no difference in frequency of adverse events with this intervention.¹² However, a more sensitive measure of adverse events—used in a study on handoffs—showed our regionalization efforts had an additive effect on reducing overnight adverse events.¹⁷Researchers should now focus on the effects of care redesign on clinical outcomes, interdisciplinary care team communication, patient engagement and satisfaction, provider opinions of communication, workflow, patient care, and housestaff education. Our methodology can be used as a model to link structure, process, and outcome related to rounds and thereby better understand how best to optimize patient care and efficiency. Additional studies are needed to analyze the content of rounds and their association with patient and educational outcomes. Last, it will be important to conduct a study to see if the effects we have identified can be sustained. Such a study is already under way.

In conclusion, creating regionalized care teams and encouraging focused bedside rounds increased the proportion of bedside time and the presence of nurses on rounds. Rounds were shorter despite higher patient census. TMA revealed that regionalized care teams and bedside rounding at a large academic hospital are feasible, and are useful in establishing the necessary structures for increasing physician–nurse and provider–patient interactions.

Acknowledgments

The authors acknowledge Dr. Stan Ashley, Dr. Jacqueline Somerville, and Sheila Harris for their support of the regionalization initiative.

Disclosures

Dr. Schnipper received funding from Sanofi-aventis to conduct an investigator-initiated study to implement and evaluate a multi-faceted intervention to improve transitions of care in patients discharged home on insulin. The study was also supported by funding from the Marshall A. Wolf Medical Education Fund, Brigham and Women’s Hospital, and Dr. Stan Ashley, Chief Medical Officer, Brigham and Women’s Hospital. Some of the content of this article was orally presented at the annual meeting of the Society of Hospital Medicine; March 29-April 1, 2015; National Harbor, MD.

In recent years, rapid response teams (RRTs) have been widely implemented to improve patient safety and quality of care. RRTs traditionally are activated by providers to address a clinically deteriorating patient; trained nurses, respiratory care specialists, and physicians are brought bedside to assist in triage and management. After the Joint Commission¹ endorsed patient engagement as a strategy for enhancing patient safety, new initiatives were developed to meet the challenge. Programs designed to enhance patient engagement have taken a variety of forms, including educational campaigns encouraging patients to report adverse events, requests for handwashing by providers, and the institution of patient- and family-activated RRTs.² Patient involvement is viewed favorably and has been shown to increase patients’ perception of health care quality.³ Although these initiatives are presumed helpful in encouraging communication, there is limited evidence that more communication leads to safety improvements. Despite the increasing prevalence of patient-activated RRTs in the United States, they have gone largely unevaluated in the adult population, and their efficacy remains unclear.

Condition Help (CH) is a patient- and family-initiated RRT designed to prevent medical errors and communication problems and improve patient safety. Patients and families are encouraged to call the CH hotline if they believe that there has been a breakdown in care or that their health is in imminent danger. This RRT was inspired by the case of Josie King, an 18-month-old girl who died of preventable causes at a large children’s hospital.⁴ After her daughter’s death, Sorrel King started the Josie King Foundation, an organization committed to preventing medical errors and creating a culture of patient safety. With the support of this foundation, CH was launched in 2005 at the Children’s Hospital of Pittsburgh at the University of Pittsburgh Medical Center (UPMC). Later it was implemented at the UPMC adult tertiary-care center, and now it is available in all UPMC facilities.

On admission, patients receive a brochure that details the purpose of CH and provides examples of when and how to call the CH hotline. In this brochure, patients are instructed to call CH in 3 situations: “1) There is an emergency and you cannot get the attention of hospital staff, 2) You see a change in the patient’s condition and the health care team is not recognizing the concern, or 3) There is breakdown in how care is given or uncertainty over what needs to be done.” These instructions are printed on bulletins placed in elevators and hallways throughout the hospital. Patients and families may activate the system at any time and can even do so from home.

When a patient or family member calls the hotline, an operator notifies the CH team. This team, which consists of a patient care liaison (or an on-duty administrator) and the unit charge nurse, convenes bedside to address the patient’s concern. The team was designed without a physician to ensure that the primary team remains in charge of the care plan. CH is kept separate from our traditional RRT and does not compete for resources (personnel, equipment, time) with the RRT, which is designed to address a clinically deteriorating patient.

In this article, we describe the characteristics of patients for whom CH was activated at our adult hospital. We also describe reasons for calls, whether changes in care were implemented, and outcomes, including traditional RRT activation, transfer to intensive care unit (ICU), and inpatient mortality. As CH was designed with patient safety as a goal, we tracked 2 types of calls, those involving safety issues and those involving nonsafety issues.

This study was approved by the quality improvement committee at the University of Pittsburgh and was considered exempt from review by the university’s Institutional Review Board.

Our integrated health system consists of more than 20 hospitals serving a tristate region. UPMC Presbyterian and UPMC Montefiore are adult tertiary-care referral hospitals with more than 750 medical/surgical beds and 150 critical care beds and more than 30,000 annual inpatient admissions. These hospitals are physically connected and function as a single large medical center. We reviewed all CH events that occurred at this combined hospital during the period January 2012 through June 2015. The dates coincided with CH data acquisition.

CH was available 24 hours a day 7 days a week. A patient care liaison (or an on-duty administrator) and the unit charge nurse responded to CH calls. Data from all calls included date and time of call, day of week, primary service, patient location, unique patient identifiers, call initiator (patient or family), whether a call led to changes in care, and primary reason for call. Each call reason was sorted into 1 of 10 categories: pain control, staff problem, lack of communication between patient/family and care team, questions about patient management, care delays, delays in a particular service, questions about discharge, administrative issues, acute psychiatric needs, and unknown/other. In addition, after a call, we reviewed all charts to determine if a safety issue was involved; Dr. Eden and Dr. Bump independently reviewed calls for safety issues and discussed any differences until they reached consensus. We also recorded outcomes, including activation of a traditional RRT or transfer to ICU within 24 hours of CH call, inpatient mortality, and against medical advice (AMA) discharges. Given that many calls were made by patients who called more than once (during a single admission or over multiple admissions), we also sorted patients into one-time callers and repeat callers for comparison. Patient satisfaction data were unavailable for review.

Patient demographic data are presented as means, standard deviations, and percentages, and call characteristics as percentages. Chi-square tests and t tests were used for analyses except for comparisons having few observations. For those, Fisher exact test was used. All analyses were performed with SAS Version 9.4 (SAS Institute, Cary, North Carolina).

From January 2012 through June 2015, 367 CH calls were made, about 105 annually. During this period, there were about 33,000 admissions, 800 combined grievances and complaints, 170 AMA discharges, 155 cardiac arrests, 2300 traditional RRT activations, and 1200 inpatient deaths per year. The 367 CH calls were made by 240 patients (Table 1). Of these 240 patients, 43 (18%) activated the CH team with multiple calls; their calls accounted for (46.3%) of all calls (170/367). The majority of calls were made by patients (76.8%) rather than family members (21.8%). Mean (SD) patient age was 45.8 (17.4) years. Mean (SD) number of admissions per patient per year was 2.7 (3.5). More events were activated for patients admitted to medical services (66%) than surgical services (34%). Calls were evenly distributed between time of day and day of week.

The most common reason for CH calls was inadequate pain control (48.2%), followed by dissatisfaction with staff (12.5%); the remaining calls were evenly distributed among the other categories. The majority of calls involved nonsafety issues (83.4%) rather than safety issues (11.4%); in 5.2% of calls, the distinction could not be made because of lack of information (Table 2). In 152 (41.4%) of the 367 total calls, a change in care or alteration in management was made. Of these 152 calls, 99 (65.1%) involved distinct changes in the care plan, such as medication changes, imaging or additional testing, or consultation with other physicians; the other 53 calls (34.9%) involved additional patient counseling or nonmedical changes. Our traditional RRT was activated within 24 hours of CH in 19 cases (5.2%); of the 19 patients, 6 were transferred to ICU. Seven patients (2.9%) died during admission. Twelve (3.3%) were discharged AMA. We compared outcomes of patients who made safety-issue calls with those of patients who made nonsafety-issue calls. The composite outcome of RRT activation, ICU transfer, and mortality was found for 6 (14.3%) of the 42 safety-issue calls and 15 (4.9%) of the 306 nonsafety-issue calls (P = 0.0291).

Patient- and family-activated RRTs provide unique opportunities for patient and family engagement during inpatient hospital stays. Our study described the results obtained with use of a well-established patient-activated RRT over several years, one of the longer observation periods reported in the literature. We found that, with use of patient-activated RRTs, patient safety issues were identified, though these were far outnumbered by nonsafety issues.

Almost half of all CH events were related to pain. Pain as the primary driver for RRT activation may be attributable to several factors, including degree of illness, poor communication about pain management expectations, positive reinforcement of narcotic-seeking behavior as a result of CH activation, and high rate of opiate use in the catchment area. A striking finding of our analysis was repeat calling; only 43 (18%) of the 240 callers were repeat callers, but they made almost half of all the calls. In some cases, during a single admission, multiple calls were made because the first had no effect on care or management; more typically, though, multiple calls were made over several admissions. Repeat callers were admitted more often per year, and they used hospital services more. They should be further studied with a goal of designing programs that better meet their needs and that prospectively address expectations of pain control.

Our study was unique in describing several outcomes related to CH events. We found that traditional RRTs were seldom activated, level of care was seldom escalated, and mortality was rare, though these outcomes occurred more often for safety-issue calls than nonsafety-issue calls. We also found that activation of CH teams often led to changes in medical management, though we could not determine whether these changes in care led to different patient outcomes.

Patient-initiated RRTs are described in a limited number of pediatric and adult studies, all with findings differing from ours. In the pediatric models, most calls were initiated by family members, were less frequent, and tended to signal higher patient acuity.^5,6 For example, in a pediatric RRT model,⁵ family members activated the RRT only twice within the study year, but both calls resulted in ICU transfer. Most descriptions of patient-activated RRTs in adult hospitals are from pilot studies, which similarly identified infrequent RRT calls but often did not identify call reasons or specific outcomes.⁷ A single-center study concluded that, after implementation of a mixed-model RRT⁸—a traditional practitioner-activated RRT later enhanced with a patient/family activation mechanism—non-ICU codes decreased, and there was a statistically significant drop in hospital-wide mortality rates. However, this RRT was patient-activated only 25 times over 2 years, and the specific outcomes of those events were not described.

Other initiatives have been designed to enhance patient care and communication. Purposeful rounding systems⁹ involve hourly rounding by bedside nurses and daily rounding by nurse leaders to improve timely patient care and provide proactive service. Such systems ideally preempt calls involving dissatisfaction and nonsafety issues. Although they would reduce the number of patient-dissatisfaction calls made in the CH system, they may not be any better than the CH system is in its main purpose, identifying safety issues. In addition, whether patient-activated RRTs or purposeful rounding systems are better at addressing patient dissatisfaction is unclear.

This study had its limitations. First, like other studies, it was a single-center observational study without a concurrent control group. Second, because CH was first implemented 10 years ago, we could not compare patient outcomes or traditional RRT use before and after program initiation. Third, our study cohort consisted of patients hospitalized at one academic tertiary-care center in one region, and the hospital is a training site for multiple residencies and fellowships. These factors likely affect the generalizability of our data to smaller or community-based centers. Fourth, some determinations were subjective (eg, whether calls involved safety or nonsafety issues). We tried to minimize bias by having 2 authors independently review cases, but the process did not reflect patient experience or perspective. Fifth, our hospital adopted its traditional RRT years before its CH system. The criteria used by hospital personnel for traditional RRT activation are designed to encourage staff to call for help at early signs of patient deterioration. Consequently, traditional RRT activations substantially outnumber CH calls. Whether this resulted in fewer CH safety calls is unclear. Sixth, we did not capture the financial implications of using CH teams.

Although patient-activated RRTs identified patient safety issues, questions about the utility or necessity of these RRTs remain. In our era of limited hospital resources, the case has not been definitively made that these teams are practical, based on patient outcomes, though other studies have found improved patient satisfaction.⁷ Most of the RRT calls in our study involved patient dissatisfaction and communication issues. CH may not be the ideal approach for managing these issues, but it represents the last line of patient advocacy once other systems have failed.

We think patient-activated RRTs have the potential to effect patient engagement in safe care. Given the importance of establishing a culture of patient safety and engagement, and increased detection of safety-related events, CH remains active throughout our hospital system. Newer iterations of CH may benefit from stricter language in defining appropriate occasions for calling RRTs, and from descriptions of other resources for patient advocacy within the hospital. These modifications could end up restricting RRT activations to patient complaints and preserving CH resources for patients with safety concerns. Our study lays the groundwork for other institutions that are considering similar interventions. Studies should now start evaluating how well patient- and family-activated RRTs improve patient satisfaction, staff satisfaction, and patient outcomes.

Patient- and family-activated RRTs were designed to engage patients and families in safe care. Although CH detects patient safety issues, these are far outnumbered by nonsafety issues. CH demonstrates a commitment to patient engagement and a culture that emphasizes patient safety.

Acknowledgements

This work was presented as a poster at the annual meeting of the Society of Hospital Medicine; March 6-9, 2016; San Diego, CA.

Disclosure

Nothing to report.

In recent years, rapid response teams (RRTs) have been widely implemented to improve patient safety and quality of care. RRTs traditionally are activated by providers to address a clinically deteriorating patient; trained nurses, respiratory care specialists, and physicians are brought bedside to assist in triage and management. After the Joint Commission¹ endorsed patient engagement as a strategy for enhancing patient safety, new initiatives were developed to meet the challenge. Programs designed to enhance patient engagement have taken a variety of forms, including educational campaigns encouraging patients to report adverse events, requests for handwashing by providers, and the institution of patient- and family-activated RRTs.² Patient involvement is viewed favorably and has been shown to increase patients’ perception of health care quality.³ Although these initiatives are presumed helpful in encouraging communication, there is limited evidence that more communication leads to safety improvements. Despite the increasing prevalence of patient-activated RRTs in the United States, they have gone largely unevaluated in the adult population, and their efficacy remains unclear.

Condition Help (CH) is a patient- and family-initiated RRT designed to prevent medical errors and communication problems and improve patient safety. Patients and families are encouraged to call the CH hotline if they believe that there has been a breakdown in care or that their health is in imminent danger. This RRT was inspired by the case of Josie King, an 18-month-old girl who died of preventable causes at a large children’s hospital.⁴ After her daughter’s death, Sorrel King started the Josie King Foundation, an organization committed to preventing medical errors and creating a culture of patient safety. With the support of this foundation, CH was launched in 2005 at the Children’s Hospital of Pittsburgh at the University of Pittsburgh Medical Center (UPMC). Later it was implemented at the UPMC adult tertiary-care center, and now it is available in all UPMC facilities.

On admission, patients receive a brochure that details the purpose of CH and provides examples of when and how to call the CH hotline. In this brochure, patients are instructed to call CH in 3 situations: “1) There is an emergency and you cannot get the attention of hospital staff, 2) You see a change in the patient’s condition and the health care team is not recognizing the concern, or 3) There is breakdown in how care is given or uncertainty over what needs to be done.” These instructions are printed on bulletins placed in elevators and hallways throughout the hospital. Patients and families may activate the system at any time and can even do so from home.

When a patient or family member calls the hotline, an operator notifies the CH team. This team, which consists of a patient care liaison (or an on-duty administrator) and the unit charge nurse, convenes bedside to address the patient’s concern. The team was designed without a physician to ensure that the primary team remains in charge of the care plan. CH is kept separate from our traditional RRT and does not compete for resources (personnel, equipment, time) with the RRT, which is designed to address a clinically deteriorating patient.

In this article, we describe the characteristics of patients for whom CH was activated at our adult hospital. We also describe reasons for calls, whether changes in care were implemented, and outcomes, including traditional RRT activation, transfer to intensive care unit (ICU), and inpatient mortality. As CH was designed with patient safety as a goal, we tracked 2 types of calls, those involving safety issues and those involving nonsafety issues.

This study was approved by the quality improvement committee at the University of Pittsburgh and was considered exempt from review by the university’s Institutional Review Board.

Our integrated health system consists of more than 20 hospitals serving a tristate region. UPMC Presbyterian and UPMC Montefiore are adult tertiary-care referral hospitals with more than 750 medical/surgical beds and 150 critical care beds and more than 30,000 annual inpatient admissions. These hospitals are physically connected and function as a single large medical center. We reviewed all CH events that occurred at this combined hospital during the period January 2012 through June 2015. The dates coincided with CH data acquisition.

CH was available 24 hours a day 7 days a week. A patient care liaison (or an on-duty administrator) and the unit charge nurse responded to CH calls. Data from all calls included date and time of call, day of week, primary service, patient location, unique patient identifiers, call initiator (patient or family), whether a call led to changes in care, and primary reason for call. Each call reason was sorted into 1 of 10 categories: pain control, staff problem, lack of communication between patient/family and care team, questions about patient management, care delays, delays in a particular service, questions about discharge, administrative issues, acute psychiatric needs, and unknown/other. In addition, after a call, we reviewed all charts to determine if a safety issue was involved; Dr. Eden and Dr. Bump independently reviewed calls for safety issues and discussed any differences until they reached consensus. We also recorded outcomes, including activation of a traditional RRT or transfer to ICU within 24 hours of CH call, inpatient mortality, and against medical advice (AMA) discharges. Given that many calls were made by patients who called more than once (during a single admission or over multiple admissions), we also sorted patients into one-time callers and repeat callers for comparison. Patient satisfaction data were unavailable for review.

Patient demographic data are presented as means, standard deviations, and percentages, and call characteristics as percentages. Chi-square tests and t tests were used for analyses except for comparisons having few observations. For those, Fisher exact test was used. All analyses were performed with SAS Version 9.4 (SAS Institute, Cary, North Carolina).

From January 2012 through June 2015, 367 CH calls were made, about 105 annually. During this period, there were about 33,000 admissions, 800 combined grievances and complaints, 170 AMA discharges, 155 cardiac arrests, 2300 traditional RRT activations, and 1200 inpatient deaths per year. The 367 CH calls were made by 240 patients (Table 1). Of these 240 patients, 43 (18%) activated the CH team with multiple calls; their calls accounted for (46.3%) of all calls (170/367). The majority of calls were made by patients (76.8%) rather than family members (21.8%). Mean (SD) patient age was 45.8 (17.4) years. Mean (SD) number of admissions per patient per year was 2.7 (3.5). More events were activated for patients admitted to medical services (66%) than surgical services (34%). Calls were evenly distributed between time of day and day of week.

The most common reason for CH calls was inadequate pain control (48.2%), followed by dissatisfaction with staff (12.5%); the remaining calls were evenly distributed among the other categories. The majority of calls involved nonsafety issues (83.4%) rather than safety issues (11.4%); in 5.2% of calls, the distinction could not be made because of lack of information (Table 2). In 152 (41.4%) of the 367 total calls, a change in care or alteration in management was made. Of these 152 calls, 99 (65.1%) involved distinct changes in the care plan, such as medication changes, imaging or additional testing, or consultation with other physicians; the other 53 calls (34.9%) involved additional patient counseling or nonmedical changes. Our traditional RRT was activated within 24 hours of CH in 19 cases (5.2%); of the 19 patients, 6 were transferred to ICU. Seven patients (2.9%) died during admission. Twelve (3.3%) were discharged AMA. We compared outcomes of patients who made safety-issue calls with those of patients who made nonsafety-issue calls. The composite outcome of RRT activation, ICU transfer, and mortality was found for 6 (14.3%) of the 42 safety-issue calls and 15 (4.9%) of the 306 nonsafety-issue calls (P = 0.0291).

Patient- and family-activated RRTs provide unique opportunities for patient and family engagement during inpatient hospital stays. Our study described the results obtained with use of a well-established patient-activated RRT over several years, one of the longer observation periods reported in the literature. We found that, with use of patient-activated RRTs, patient safety issues were identified, though these were far outnumbered by nonsafety issues.

Almost half of all CH events were related to pain. Pain as the primary driver for RRT activation may be attributable to several factors, including degree of illness, poor communication about pain management expectations, positive reinforcement of narcotic-seeking behavior as a result of CH activation, and high rate of opiate use in the catchment area. A striking finding of our analysis was repeat calling; only 43 (18%) of the 240 callers were repeat callers, but they made almost half of all the calls. In some cases, during a single admission, multiple calls were made because the first had no effect on care or management; more typically, though, multiple calls were made over several admissions. Repeat callers were admitted more often per year, and they used hospital services more. They should be further studied with a goal of designing programs that better meet their needs and that prospectively address expectations of pain control.

Our study was unique in describing several outcomes related to CH events. We found that traditional RRTs were seldom activated, level of care was seldom escalated, and mortality was rare, though these outcomes occurred more often for safety-issue calls than nonsafety-issue calls. We also found that activation of CH teams often led to changes in medical management, though we could not determine whether these changes in care led to different patient outcomes.

Patient-initiated RRTs are described in a limited number of pediatric and adult studies, all with findings differing from ours. In the pediatric models, most calls were initiated by family members, were less frequent, and tended to signal higher patient acuity.^5,6 For example, in a pediatric RRT model,⁵ family members activated the RRT only twice within the study year, but both calls resulted in ICU transfer. Most descriptions of patient-activated RRTs in adult hospitals are from pilot studies, which similarly identified infrequent RRT calls but often did not identify call reasons or specific outcomes.⁷ A single-center study concluded that, after implementation of a mixed-model RRT⁸—a traditional practitioner-activated RRT later enhanced with a patient/family activation mechanism—non-ICU codes decreased, and there was a statistically significant drop in hospital-wide mortality rates. However, this RRT was patient-activated only 25 times over 2 years, and the specific outcomes of those events were not described.

Other initiatives have been designed to enhance patient care and communication. Purposeful rounding systems⁹ involve hourly rounding by bedside nurses and daily rounding by nurse leaders to improve timely patient care and provide proactive service. Such systems ideally preempt calls involving dissatisfaction and nonsafety issues. Although they would reduce the number of patient-dissatisfaction calls made in the CH system, they may not be any better than the CH system is in its main purpose, identifying safety issues. In addition, whether patient-activated RRTs or purposeful rounding systems are better at addressing patient dissatisfaction is unclear.

This study had its limitations. First, like other studies, it was a single-center observational study without a concurrent control group. Second, because CH was first implemented 10 years ago, we could not compare patient outcomes or traditional RRT use before and after program initiation. Third, our study cohort consisted of patients hospitalized at one academic tertiary-care center in one region, and the hospital is a training site for multiple residencies and fellowships. These factors likely affect the generalizability of our data to smaller or community-based centers. Fourth, some determinations were subjective (eg, whether calls involved safety or nonsafety issues). We tried to minimize bias by having 2 authors independently review cases, but the process did not reflect patient experience or perspective. Fifth, our hospital adopted its traditional RRT years before its CH system. The criteria used by hospital personnel for traditional RRT activation are designed to encourage staff to call for help at early signs of patient deterioration. Consequently, traditional RRT activations substantially outnumber CH calls. Whether this resulted in fewer CH safety calls is unclear. Sixth, we did not capture the financial implications of using CH teams.

Although patient-activated RRTs identified patient safety issues, questions about the utility or necessity of these RRTs remain. In our era of limited hospital resources, the case has not been definitively made that these teams are practical, based on patient outcomes, though other studies have found improved patient satisfaction.⁷ Most of the RRT calls in our study involved patient dissatisfaction and communication issues. CH may not be the ideal approach for managing these issues, but it represents the last line of patient advocacy once other systems have failed.

We think patient-activated RRTs have the potential to effect patient engagement in safe care. Given the importance of establishing a culture of patient safety and engagement, and increased detection of safety-related events, CH remains active throughout our hospital system. Newer iterations of CH may benefit from stricter language in defining appropriate occasions for calling RRTs, and from descriptions of other resources for patient advocacy within the hospital. These modifications could end up restricting RRT activations to patient complaints and preserving CH resources for patients with safety concerns. Our study lays the groundwork for other institutions that are considering similar interventions. Studies should now start evaluating how well patient- and family-activated RRTs improve patient satisfaction, staff satisfaction, and patient outcomes.

Patient- and family-activated RRTs were designed to engage patients and families in safe care. Although CH detects patient safety issues, these are far outnumbered by nonsafety issues. CH demonstrates a commitment to patient engagement and a culture that emphasizes patient safety.

Acknowledgements

This work was presented as a poster at the annual meeting of the Society of Hospital Medicine; March 6-9, 2016; San Diego, CA.

Disclosure

Nothing to report.

In recent years, rapid response teams (RRTs) have been widely implemented to improve patient safety and quality of care. RRTs traditionally are activated by providers to address a clinically deteriorating patient; trained nurses, respiratory care specialists, and physicians are brought bedside to assist in triage and management. After the Joint Commission¹ endorsed patient engagement as a strategy for enhancing patient safety, new initiatives were developed to meet the challenge. Programs designed to enhance patient engagement have taken a variety of forms, including educational campaigns encouraging patients to report adverse events, requests for handwashing by providers, and the institution of patient- and family-activated RRTs.² Patient involvement is viewed favorably and has been shown to increase patients’ perception of health care quality.³ Although these initiatives are presumed helpful in encouraging communication, there is limited evidence that more communication leads to safety improvements. Despite the increasing prevalence of patient-activated RRTs in the United States, they have gone largely unevaluated in the adult population, and their efficacy remains unclear.

Condition Help (CH) is a patient- and family-initiated RRT designed to prevent medical errors and communication problems and improve patient safety. Patients and families are encouraged to call the CH hotline if they believe that there has been a breakdown in care or that their health is in imminent danger. This RRT was inspired by the case of Josie King, an 18-month-old girl who died of preventable causes at a large children’s hospital.⁴ After her daughter’s death, Sorrel King started the Josie King Foundation, an organization committed to preventing medical errors and creating a culture of patient safety. With the support of this foundation, CH was launched in 2005 at the Children’s Hospital of Pittsburgh at the University of Pittsburgh Medical Center (UPMC). Later it was implemented at the UPMC adult tertiary-care center, and now it is available in all UPMC facilities.

On admission, patients receive a brochure that details the purpose of CH and provides examples of when and how to call the CH hotline. In this brochure, patients are instructed to call CH in 3 situations: “1) There is an emergency and you cannot get the attention of hospital staff, 2) You see a change in the patient’s condition and the health care team is not recognizing the concern, or 3) There is breakdown in how care is given or uncertainty over what needs to be done.” These instructions are printed on bulletins placed in elevators and hallways throughout the hospital. Patients and families may activate the system at any time and can even do so from home.

When a patient or family member calls the hotline, an operator notifies the CH team. This team, which consists of a patient care liaison (or an on-duty administrator) and the unit charge nurse, convenes bedside to address the patient’s concern. The team was designed without a physician to ensure that the primary team remains in charge of the care plan. CH is kept separate from our traditional RRT and does not compete for resources (personnel, equipment, time) with the RRT, which is designed to address a clinically deteriorating patient.

In this article, we describe the characteristics of patients for whom CH was activated at our adult hospital. We also describe reasons for calls, whether changes in care were implemented, and outcomes, including traditional RRT activation, transfer to intensive care unit (ICU), and inpatient mortality. As CH was designed with patient safety as a goal, we tracked 2 types of calls, those involving safety issues and those involving nonsafety issues.

This study was approved by the quality improvement committee at the University of Pittsburgh and was considered exempt from review by the university’s Institutional Review Board.

Our integrated health system consists of more than 20 hospitals serving a tristate region. UPMC Presbyterian and UPMC Montefiore are adult tertiary-care referral hospitals with more than 750 medical/surgical beds and 150 critical care beds and more than 30,000 annual inpatient admissions. These hospitals are physically connected and function as a single large medical center. We reviewed all CH events that occurred at this combined hospital during the period January 2012 through June 2015. The dates coincided with CH data acquisition.

CH was available 24 hours a day 7 days a week. A patient care liaison (or an on-duty administrator) and the unit charge nurse responded to CH calls. Data from all calls included date and time of call, day of week, primary service, patient location, unique patient identifiers, call initiator (patient or family), whether a call led to changes in care, and primary reason for call. Each call reason was sorted into 1 of 10 categories: pain control, staff problem, lack of communication between patient/family and care team, questions about patient management, care delays, delays in a particular service, questions about discharge, administrative issues, acute psychiatric needs, and unknown/other. In addition, after a call, we reviewed all charts to determine if a safety issue was involved; Dr. Eden and Dr. Bump independently reviewed calls for safety issues and discussed any differences until they reached consensus. We also recorded outcomes, including activation of a traditional RRT or transfer to ICU within 24 hours of CH call, inpatient mortality, and against medical advice (AMA) discharges. Given that many calls were made by patients who called more than once (during a single admission or over multiple admissions), we also sorted patients into one-time callers and repeat callers for comparison. Patient satisfaction data were unavailable for review.

Patient demographic data are presented as means, standard deviations, and percentages, and call characteristics as percentages. Chi-square tests and t tests were used for analyses except for comparisons having few observations. For those, Fisher exact test was used. All analyses were performed with SAS Version 9.4 (SAS Institute, Cary, North Carolina).

From January 2012 through June 2015, 367 CH calls were made, about 105 annually. During this period, there were about 33,000 admissions, 800 combined grievances and complaints, 170 AMA discharges, 155 cardiac arrests, 2300 traditional RRT activations, and 1200 inpatient deaths per year. The 367 CH calls were made by 240 patients (Table 1). Of these 240 patients, 43 (18%) activated the CH team with multiple calls; their calls accounted for (46.3%) of all calls (170/367). The majority of calls were made by patients (76.8%) rather than family members (21.8%). Mean (SD) patient age was 45.8 (17.4) years. Mean (SD) number of admissions per patient per year was 2.7 (3.5). More events were activated for patients admitted to medical services (66%) than surgical services (34%). Calls were evenly distributed between time of day and day of week.

The most common reason for CH calls was inadequate pain control (48.2%), followed by dissatisfaction with staff (12.5%); the remaining calls were evenly distributed among the other categories. The majority of calls involved nonsafety issues (83.4%) rather than safety issues (11.4%); in 5.2% of calls, the distinction could not be made because of lack of information (Table 2). In 152 (41.4%) of the 367 total calls, a change in care or alteration in management was made. Of these 152 calls, 99 (65.1%) involved distinct changes in the care plan, such as medication changes, imaging or additional testing, or consultation with other physicians; the other 53 calls (34.9%) involved additional patient counseling or nonmedical changes. Our traditional RRT was activated within 24 hours of CH in 19 cases (5.2%); of the 19 patients, 6 were transferred to ICU. Seven patients (2.9%) died during admission. Twelve (3.3%) were discharged AMA. We compared outcomes of patients who made safety-issue calls with those of patients who made nonsafety-issue calls. The composite outcome of RRT activation, ICU transfer, and mortality was found for 6 (14.3%) of the 42 safety-issue calls and 15 (4.9%) of the 306 nonsafety-issue calls (P = 0.0291).

Patient- and family-activated RRTs provide unique opportunities for patient and family engagement during inpatient hospital stays. Our study described the results obtained with use of a well-established patient-activated RRT over several years, one of the longer observation periods reported in the literature. We found that, with use of patient-activated RRTs, patient safety issues were identified, though these were far outnumbered by nonsafety issues.

Almost half of all CH events were related to pain. Pain as the primary driver for RRT activation may be attributable to several factors, including degree of illness, poor communication about pain management expectations, positive reinforcement of narcotic-seeking behavior as a result of CH activation, and high rate of opiate use in the catchment area. A striking finding of our analysis was repeat calling; only 43 (18%) of the 240 callers were repeat callers, but they made almost half of all the calls. In some cases, during a single admission, multiple calls were made because the first had no effect on care or management; more typically, though, multiple calls were made over several admissions. Repeat callers were admitted more often per year, and they used hospital services more. They should be further studied with a goal of designing programs that better meet their needs and that prospectively address expectations of pain control.

Our study was unique in describing several outcomes related to CH events. We found that traditional RRTs were seldom activated, level of care was seldom escalated, and mortality was rare, though these outcomes occurred more often for safety-issue calls than nonsafety-issue calls. We also found that activation of CH teams often led to changes in medical management, though we could not determine whether these changes in care led to different patient outcomes.

Patient-initiated RRTs are described in a limited number of pediatric and adult studies, all with findings differing from ours. In the pediatric models, most calls were initiated by family members, were less frequent, and tended to signal higher patient acuity.^5,6 For example, in a pediatric RRT model,⁵ family members activated the RRT only twice within the study year, but both calls resulted in ICU transfer. Most descriptions of patient-activated RRTs in adult hospitals are from pilot studies, which similarly identified infrequent RRT calls but often did not identify call reasons or specific outcomes.⁷ A single-center study concluded that, after implementation of a mixed-model RRT⁸—a traditional practitioner-activated RRT later enhanced with a patient/family activation mechanism—non-ICU codes decreased, and there was a statistically significant drop in hospital-wide mortality rates. However, this RRT was patient-activated only 25 times over 2 years, and the specific outcomes of those events were not described.

Other initiatives have been designed to enhance patient care and communication. Purposeful rounding systems⁹ involve hourly rounding by bedside nurses and daily rounding by nurse leaders to improve timely patient care and provide proactive service. Such systems ideally preempt calls involving dissatisfaction and nonsafety issues. Although they would reduce the number of patient-dissatisfaction calls made in the CH system, they may not be any better than the CH system is in its main purpose, identifying safety issues. In addition, whether patient-activated RRTs or purposeful rounding systems are better at addressing patient dissatisfaction is unclear.

This study had its limitations. First, like other studies, it was a single-center observational study without a concurrent control group. Second, because CH was first implemented 10 years ago, we could not compare patient outcomes or traditional RRT use before and after program initiation. Third, our study cohort consisted of patients hospitalized at one academic tertiary-care center in one region, and the hospital is a training site for multiple residencies and fellowships. These factors likely affect the generalizability of our data to smaller or community-based centers. Fourth, some determinations were subjective (eg, whether calls involved safety or nonsafety issues). We tried to minimize bias by having 2 authors independently review cases, but the process did not reflect patient experience or perspective. Fifth, our hospital adopted its traditional RRT years before its CH system. The criteria used by hospital personnel for traditional RRT activation are designed to encourage staff to call for help at early signs of patient deterioration. Consequently, traditional RRT activations substantially outnumber CH calls. Whether this resulted in fewer CH safety calls is unclear. Sixth, we did not capture the financial implications of using CH teams.

Although patient-activated RRTs identified patient safety issues, questions about the utility or necessity of these RRTs remain. In our era of limited hospital resources, the case has not been definitively made that these teams are practical, based on patient outcomes, though other studies have found improved patient satisfaction.⁷ Most of the RRT calls in our study involved patient dissatisfaction and communication issues. CH may not be the ideal approach for managing these issues, but it represents the last line of patient advocacy once other systems have failed.

We think patient-activated RRTs have the potential to effect patient engagement in safe care. Given the importance of establishing a culture of patient safety and engagement, and increased detection of safety-related events, CH remains active throughout our hospital system. Newer iterations of CH may benefit from stricter language in defining appropriate occasions for calling RRTs, and from descriptions of other resources for patient advocacy within the hospital. These modifications could end up restricting RRT activations to patient complaints and preserving CH resources for patients with safety concerns. Our study lays the groundwork for other institutions that are considering similar interventions. Studies should now start evaluating how well patient- and family-activated RRTs improve patient satisfaction, staff satisfaction, and patient outcomes.

Patient- and family-activated RRTs were designed to engage patients and families in safe care. Although CH detects patient safety issues, these are far outnumbered by nonsafety issues. CH demonstrates a commitment to patient engagement and a culture that emphasizes patient safety.

Acknowledgements

This work was presented as a poster at the annual meeting of the Society of Hospital Medicine; March 6-9, 2016; San Diego, CA.

Disclosure

Nothing to report.