Longer term and higher doses of opioid analgesics (OAs) have been associated with multiple adverse outcomes such as loss of work, cognitive decline, and poor function.[1, 2, 3, 4] One of the most widely reported complications of opioid therapy is drug overdose.[5, 6, 7, 8, 9] In population‐based studies, daily morphine equivalent doses >100 mg have been associated with significantly increased risk of drug overdose.[5, 6, 7, 8, 9, 10] Among health maintenance organization (HMO) enrollees filling at least 2 prescriptions for opioids, our group reported that daily opioid doses 100 mg were associated with approximately threefold greater adjusted odds of drug overdose.[10] We also observed over a twofold increase in odds of drug overdose for lower daily doses of 50 to 99 mg if the patient also received a high total opioid dose (>1830 mg) over a 6‐month period. This analysis suggests that clinicians may need to monitor not only daily dose but also total dose of opioids to reduce the risk of drug overdose.

Yet drug overdose represents only a small subset of all hospitalizations for persons receiving long‐term or higher doses of opioids for noncancer pain. These patients have significant demand for urgent care services, including hospitalization, for diverse reasons such as adverse effects of opioids, underlying cause of chronic pain, and comorbidities such as mental health disorders.[11] In a cohort of elderly primary care patients who were high hospital utilizers, Freund and colleagues reported that chronic pain and depression were the most common conditions co‐occurring with their other comorbidities.[12] However, little is known about the association of opioid dose with the risk of all‐cause hospitalization for patients with noncancer pain.

In this article we examined hospitalizations for a national cohort of HMO enrollees with noncancer pain who filled at least 2 prescriptions for schedule II or III opioids over a 3.5‐year timeframe. This retrospective cohort analysis aims to identify clinically useful opioid dose measures for clinicians, administrators, and policymakers to use in identifying patients at increased risk of future hospitalization who may warrant interventions to reduce this risk.

METHODS

RESULTS

Of 87,688 study subjects, 54.8% were women, and the mean age was 53.8 years (standard deviation [SD]=5.5). Nearly half of the cohort resided in Southern states (Table 1). In the baseline 6‐month interval, the most common chronic noncancer pain conditions were musculoskeletal involving large joint arthritis/arthralgia (38.4%) and back pain (28.2%). In regard to mental health and substance use conditions, both anxiety/PTSD and depression were diagnosed in approximately 7% of the cohort, whereas psychosis, and alcohol and other substance use disorders were each diagnosed in <2%. In the baseline interval, 12.7% of subjects were hospitalized. The majority of patients received a daily opioid dose of 20 to 49 mg, and the median total dose was 450 mg. The median percent of time exposed to opioids was 6.7% among all study subjects and 70% for those with a high total dose (>1830 mg).

In the 3 study years, an average of 12% of the cohort was hospitalized yearly (Table 2), or 1120 hospitalizations per 10,000 person‐years. Among those who were hospitalized, inpatient days averaged 6.5 (SD=8.5). The highest proportion of hospitalized subjects was 6.5%, occurring in the 6‐month interval immediately following the first opioid treatment interval. In subsequent 6‐month intervals, hospitalization rates were relatively stable, ranging from 5.2% to 6.1% (Table 2). As shown, future hospitalization rates increased monotonically, with increasing total or daily dose within each 6‐month interval.

In unadjusted analyses, a significant interaction between daily dose and total dose (P<0.001) revealed that, within each daily dose category, the odds of hospitalization differed by total dose (all P<0.05, Table 3). When the total dose was >1830 mg, the odds of future hospitalization rose monotonically with increasing daily dose (i.e., <20, 2049, 5099, 100 mg): 1.33, 1.84, 1.96, and 2.08 (P<0.05 for all comparisons vs no opioids). On the other hand, when the total dose was 450 mg or less, all daily dose categories including a very high daily dose (100 mg) were not associated with future hospitalization (all P>0.05 vs no opioids). When the total dose was 451 to 1830 mg, a nonlinear association with hospitalization appeared with higher odds for lower daily doses. For the outcome of hospital days per 6‐month interval, increasing daily dose was also associated with more hospital days per 6‐month interval when the total dose was high (>1830 mg), whereas for lower total doses, daily dose was weakly positive or even protective versus no opioids.

In the model adjusting for all covariates (Table 4), the interaction between total dose and daily dose was also significant (P=0.002). When the total dose was high (>1830 mg), the adjusted odds of future hospitalization were significantly increased by 35% to 44% for daily doses of 20 to 49 mg or greater versus no opioids (P<0.05 for all comparisons). When the total dose was <1830 mg, the majority of daily dose categories were not significantly associated with hospitalization. Similarly, in the fully adjusted analysis of hospital days, the number of inpatient days were increased by 28% to 48% when the total dose was >1830 mg and daily dose was >20 mg, but these associations were nonsignificant or protective when the total dose was lower.

In a sensitivity analysis, we examined the percentage of days covered by filled opioid prescriptions within a 6‐month interval for subjects receiving high‐dose therapy (Table 5). Compared with no opioid therapy, the adjusted odds of future hospitalization were 5% greater for low total opioid dose (11830 mg) and 21% greater for high total dose (>1830 mg) when the duration of treatment was shorter (50% of the 6‐month interval). However, the odds were increased by 41% to 51% for a high total dose (>1830 mg), with longer periods of treatment (>50% of the interval). For hospital days as the outcome, subjects with high total doses (>1830 mg) and longer periods of treatment (>50% of the interval) had 41% to 71% more hospital days per 6‐month interval than those with no opioid therapy.

DISCUSSION

In a national cohort of HMO enrollees who filled at least 2 prescriptions for OAs, 12% were hospitalized annually. Other studies of opioid users have focused on only a fraction of these hospitalizations. For example, a recent Agency for Healthcare Research and Quality study reported that the rate of hospitalization for complications from accidental or deliberate overuse of opioids more than doubled from 11.7/10,000 in 1993 to 29.5/10,000 in 2010.[20] However, in our cohort, the all‐cause hospitalization rate was 1120 per 10,000 person‐years, or over 40 times greater than the rate for complications from overuse of opioids. By comparison, hospitalization for heart failure was only 32.8/10,000 nationally in 2010.[21] Thus, our study confirms the significant demand for hospital care by patients treated with opioids. A novel finding of our study is that the total dose of prescriptions filled over 6 months is significantly associated with an increased risk of future hospitalization. When the total dose within 6 months was in the top quartile (>1830 mg in our cohort), the adjusted odds of future hospitalization ranged from 35% to 44% greater than no opioids for daily opioid doses above 20 mg/day. On the other hand, when the total dose was 1830 mg, the daily opioid dose was only weakly associated with future hospitalization. These associations were similar for hospital days per 6‐month interval as the outcome.

Edlund and colleagues examined the total dose of opioids in a national cohort of veterans with chronic noncancer pain who filled at least 1 opioid prescription.[22] In 2011, the 60th percentile for the total opioid dose for these veterans was 3610 mg within a year, which is roughly equivalent to our top quartile (1830 mg) over a 6‐month interval. These data support replicating our study in veterans to evaluate whether a similarly increased risk of hospitalization appears for those with high total opioid doses. In support of a concern among veterans, a population‐based, cross‐sectional study of hospitalized veterans reported a high rate of chronic opioid therapy (90 days) in the 6 months prior to hospitalization.[23]

Other studies have reported increased risk of hospitalization with chronic opioid therapy. Among 1045 patients followed up to 1‐year post‐transplantation, long‐term opioids were associated with up to a sixfold greater risk of at least 4 admissions within that year.[24] Among 13,127 Danish adults on opioid therapy, the odds of future hospitalization from injuries were increased by 74% for long‐term therapy and 46% for short‐term therapy versus no opioids and by threefold and 1.6‐fold, respectively, for hospitalization due to toxicity/poisoning.[9] However, none of these studies examined the dose of opioids.

In a sensitivity analysis, we found that when a subject received a high total opioid dose within 6 months, treatment for more than 50% of the interval (i.e., >3 months) was associated with a significantly increased risk of future hospitalization and significantly more hospital days. Because the strongest evidence for the benefit of opioids for chronic noncancer pain comes from trials of <3 months,[25] these data lend additional support to recommendations to minimize both dose and duration of opioid therapy.

Our study has several limitations. First, we did not assess the immediate risk of hospitalization after starting opioid therapy. Second, our outcome of hospitalization represents only 1 measure of risk. Thus, our data should not be regarded as supporting short‐term use of high‐dose opioids over 100 to 120 mg per day.[26] In an earlier study, we reported that either a high daily dose (100 mg) or a moderately high daily dose (5099 mg) plus a high total dose (>1830 mg) increased the risk of drug overdose.[10] Third, we could not examine the reason for hospitalizations in this analysis. Therefore, we cannot presume that opioid therapy caused these hospitalizations, but it likely serves as a proxy for other factors such as disability and mental health disorders that increase risk of hospitalization. However, we did adjust for pain conditions as well as mental health and substance abuse disorders that are known to increase the risk of hospitalization in other cohorts.[27, 28, 29, 30] In a national veterans study, the most common clinical conditions associated with long‐term opioid therapy were major depression and PTSD.[22] Last, we did also not consider the number of prescribers of opioids. In a Medicare study, 1 versus 4 prescribers of OAs increased patients' annual hospitalization rate from 1.6% to 4.8%, respectively.[31]

Although the total opioid dose categories observed for our study population may differ from those in other cohorts, these data offer additional evidence for clinicians to consider this measure when assessing risk for hospitalization, and among subjects on high total doses, the percentage of time on opioids offers an additional measure of risk. Because opioid users with noncancer pain are heavy consumers of healthcare services,^32,33 public health benefits and reductions in costs of care may be substantial if opportunities can be identified to reduce hospital utilization by persons treated with higher doses of OAs.

Disclosures

The work on this project was supported by an intramural grant from the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant 1UL TR001120. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The authors report no conflicts of interest.

Longer term and higher doses of opioid analgesics (OAs) have been associated with multiple adverse outcomes such as loss of work, cognitive decline, and poor function.[1, 2, 3, 4] One of the most widely reported complications of opioid therapy is drug overdose.[5, 6, 7, 8, 9] In population‐based studies, daily morphine equivalent doses >100 mg have been associated with significantly increased risk of drug overdose.[5, 6, 7, 8, 9, 10] Among health maintenance organization (HMO) enrollees filling at least 2 prescriptions for opioids, our group reported that daily opioid doses 100 mg were associated with approximately threefold greater adjusted odds of drug overdose.[10] We also observed over a twofold increase in odds of drug overdose for lower daily doses of 50 to 99 mg if the patient also received a high total opioid dose (>1830 mg) over a 6‐month period. This analysis suggests that clinicians may need to monitor not only daily dose but also total dose of opioids to reduce the risk of drug overdose.

Yet drug overdose represents only a small subset of all hospitalizations for persons receiving long‐term or higher doses of opioids for noncancer pain. These patients have significant demand for urgent care services, including hospitalization, for diverse reasons such as adverse effects of opioids, underlying cause of chronic pain, and comorbidities such as mental health disorders.[11] In a cohort of elderly primary care patients who were high hospital utilizers, Freund and colleagues reported that chronic pain and depression were the most common conditions co‐occurring with their other comorbidities.[12] However, little is known about the association of opioid dose with the risk of all‐cause hospitalization for patients with noncancer pain.

In this article we examined hospitalizations for a national cohort of HMO enrollees with noncancer pain who filled at least 2 prescriptions for schedule II or III opioids over a 3.5‐year timeframe. This retrospective cohort analysis aims to identify clinically useful opioid dose measures for clinicians, administrators, and policymakers to use in identifying patients at increased risk of future hospitalization who may warrant interventions to reduce this risk.

METHODS

RESULTS

Of 87,688 study subjects, 54.8% were women, and the mean age was 53.8 years (standard deviation [SD]=5.5). Nearly half of the cohort resided in Southern states (Table 1). In the baseline 6‐month interval, the most common chronic noncancer pain conditions were musculoskeletal involving large joint arthritis/arthralgia (38.4%) and back pain (28.2%). In regard to mental health and substance use conditions, both anxiety/PTSD and depression were diagnosed in approximately 7% of the cohort, whereas psychosis, and alcohol and other substance use disorders were each diagnosed in <2%. In the baseline interval, 12.7% of subjects were hospitalized. The majority of patients received a daily opioid dose of 20 to 49 mg, and the median total dose was 450 mg. The median percent of time exposed to opioids was 6.7% among all study subjects and 70% for those with a high total dose (>1830 mg).

In the 3 study years, an average of 12% of the cohort was hospitalized yearly (Table 2), or 1120 hospitalizations per 10,000 person‐years. Among those who were hospitalized, inpatient days averaged 6.5 (SD=8.5). The highest proportion of hospitalized subjects was 6.5%, occurring in the 6‐month interval immediately following the first opioid treatment interval. In subsequent 6‐month intervals, hospitalization rates were relatively stable, ranging from 5.2% to 6.1% (Table 2). As shown, future hospitalization rates increased monotonically, with increasing total or daily dose within each 6‐month interval.

In unadjusted analyses, a significant interaction between daily dose and total dose (P<0.001) revealed that, within each daily dose category, the odds of hospitalization differed by total dose (all P<0.05, Table 3). When the total dose was >1830 mg, the odds of future hospitalization rose monotonically with increasing daily dose (i.e., <20, 2049, 5099, 100 mg): 1.33, 1.84, 1.96, and 2.08 (P<0.05 for all comparisons vs no opioids). On the other hand, when the total dose was 450 mg or less, all daily dose categories including a very high daily dose (100 mg) were not associated with future hospitalization (all P>0.05 vs no opioids). When the total dose was 451 to 1830 mg, a nonlinear association with hospitalization appeared with higher odds for lower daily doses. For the outcome of hospital days per 6‐month interval, increasing daily dose was also associated with more hospital days per 6‐month interval when the total dose was high (>1830 mg), whereas for lower total doses, daily dose was weakly positive or even protective versus no opioids.

In the model adjusting for all covariates (Table 4), the interaction between total dose and daily dose was also significant (P=0.002). When the total dose was high (>1830 mg), the adjusted odds of future hospitalization were significantly increased by 35% to 44% for daily doses of 20 to 49 mg or greater versus no opioids (P<0.05 for all comparisons). When the total dose was <1830 mg, the majority of daily dose categories were not significantly associated with hospitalization. Similarly, in the fully adjusted analysis of hospital days, the number of inpatient days were increased by 28% to 48% when the total dose was >1830 mg and daily dose was >20 mg, but these associations were nonsignificant or protective when the total dose was lower.

In a sensitivity analysis, we examined the percentage of days covered by filled opioid prescriptions within a 6‐month interval for subjects receiving high‐dose therapy (Table 5). Compared with no opioid therapy, the adjusted odds of future hospitalization were 5% greater for low total opioid dose (11830 mg) and 21% greater for high total dose (>1830 mg) when the duration of treatment was shorter (50% of the 6‐month interval). However, the odds were increased by 41% to 51% for a high total dose (>1830 mg), with longer periods of treatment (>50% of the interval). For hospital days as the outcome, subjects with high total doses (>1830 mg) and longer periods of treatment (>50% of the interval) had 41% to 71% more hospital days per 6‐month interval than those with no opioid therapy.

DISCUSSION

In a national cohort of HMO enrollees who filled at least 2 prescriptions for OAs, 12% were hospitalized annually. Other studies of opioid users have focused on only a fraction of these hospitalizations. For example, a recent Agency for Healthcare Research and Quality study reported that the rate of hospitalization for complications from accidental or deliberate overuse of opioids more than doubled from 11.7/10,000 in 1993 to 29.5/10,000 in 2010.[20] However, in our cohort, the all‐cause hospitalization rate was 1120 per 10,000 person‐years, or over 40 times greater than the rate for complications from overuse of opioids. By comparison, hospitalization for heart failure was only 32.8/10,000 nationally in 2010.[21] Thus, our study confirms the significant demand for hospital care by patients treated with opioids. A novel finding of our study is that the total dose of prescriptions filled over 6 months is significantly associated with an increased risk of future hospitalization. When the total dose within 6 months was in the top quartile (>1830 mg in our cohort), the adjusted odds of future hospitalization ranged from 35% to 44% greater than no opioids for daily opioid doses above 20 mg/day. On the other hand, when the total dose was 1830 mg, the daily opioid dose was only weakly associated with future hospitalization. These associations were similar for hospital days per 6‐month interval as the outcome.

Edlund and colleagues examined the total dose of opioids in a national cohort of veterans with chronic noncancer pain who filled at least 1 opioid prescription.[22] In 2011, the 60th percentile for the total opioid dose for these veterans was 3610 mg within a year, which is roughly equivalent to our top quartile (1830 mg) over a 6‐month interval. These data support replicating our study in veterans to evaluate whether a similarly increased risk of hospitalization appears for those with high total opioid doses. In support of a concern among veterans, a population‐based, cross‐sectional study of hospitalized veterans reported a high rate of chronic opioid therapy (90 days) in the 6 months prior to hospitalization.[23]

Other studies have reported increased risk of hospitalization with chronic opioid therapy. Among 1045 patients followed up to 1‐year post‐transplantation, long‐term opioids were associated with up to a sixfold greater risk of at least 4 admissions within that year.[24] Among 13,127 Danish adults on opioid therapy, the odds of future hospitalization from injuries were increased by 74% for long‐term therapy and 46% for short‐term therapy versus no opioids and by threefold and 1.6‐fold, respectively, for hospitalization due to toxicity/poisoning.[9] However, none of these studies examined the dose of opioids.

In a sensitivity analysis, we found that when a subject received a high total opioid dose within 6 months, treatment for more than 50% of the interval (i.e., >3 months) was associated with a significantly increased risk of future hospitalization and significantly more hospital days. Because the strongest evidence for the benefit of opioids for chronic noncancer pain comes from trials of <3 months,[25] these data lend additional support to recommendations to minimize both dose and duration of opioid therapy.

Our study has several limitations. First, we did not assess the immediate risk of hospitalization after starting opioid therapy. Second, our outcome of hospitalization represents only 1 measure of risk. Thus, our data should not be regarded as supporting short‐term use of high‐dose opioids over 100 to 120 mg per day.[26] In an earlier study, we reported that either a high daily dose (100 mg) or a moderately high daily dose (5099 mg) plus a high total dose (>1830 mg) increased the risk of drug overdose.[10] Third, we could not examine the reason for hospitalizations in this analysis. Therefore, we cannot presume that opioid therapy caused these hospitalizations, but it likely serves as a proxy for other factors such as disability and mental health disorders that increase risk of hospitalization. However, we did adjust for pain conditions as well as mental health and substance abuse disorders that are known to increase the risk of hospitalization in other cohorts.[27, 28, 29, 30] In a national veterans study, the most common clinical conditions associated with long‐term opioid therapy were major depression and PTSD.[22] Last, we did also not consider the number of prescribers of opioids. In a Medicare study, 1 versus 4 prescribers of OAs increased patients' annual hospitalization rate from 1.6% to 4.8%, respectively.[31]

Although the total opioid dose categories observed for our study population may differ from those in other cohorts, these data offer additional evidence for clinicians to consider this measure when assessing risk for hospitalization, and among subjects on high total doses, the percentage of time on opioids offers an additional measure of risk. Because opioid users with noncancer pain are heavy consumers of healthcare services,^32,33 public health benefits and reductions in costs of care may be substantial if opportunities can be identified to reduce hospital utilization by persons treated with higher doses of OAs.

Disclosures

The work on this project was supported by an intramural grant from the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant 1UL TR001120. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The authors report no conflicts of interest.

Longer term and higher doses of opioid analgesics (OAs) have been associated with multiple adverse outcomes such as loss of work, cognitive decline, and poor function.[1, 2, 3, 4] One of the most widely reported complications of opioid therapy is drug overdose.[5, 6, 7, 8, 9] In population‐based studies, daily morphine equivalent doses >100 mg have been associated with significantly increased risk of drug overdose.[5, 6, 7, 8, 9, 10] Among health maintenance organization (HMO) enrollees filling at least 2 prescriptions for opioids, our group reported that daily opioid doses 100 mg were associated with approximately threefold greater adjusted odds of drug overdose.[10] We also observed over a twofold increase in odds of drug overdose for lower daily doses of 50 to 99 mg if the patient also received a high total opioid dose (>1830 mg) over a 6‐month period. This analysis suggests that clinicians may need to monitor not only daily dose but also total dose of opioids to reduce the risk of drug overdose.

Yet drug overdose represents only a small subset of all hospitalizations for persons receiving long‐term or higher doses of opioids for noncancer pain. These patients have significant demand for urgent care services, including hospitalization, for diverse reasons such as adverse effects of opioids, underlying cause of chronic pain, and comorbidities such as mental health disorders.[11] In a cohort of elderly primary care patients who were high hospital utilizers, Freund and colleagues reported that chronic pain and depression were the most common conditions co‐occurring with their other comorbidities.[12] However, little is known about the association of opioid dose with the risk of all‐cause hospitalization for patients with noncancer pain.

In this article we examined hospitalizations for a national cohort of HMO enrollees with noncancer pain who filled at least 2 prescriptions for schedule II or III opioids over a 3.5‐year timeframe. This retrospective cohort analysis aims to identify clinically useful opioid dose measures for clinicians, administrators, and policymakers to use in identifying patients at increased risk of future hospitalization who may warrant interventions to reduce this risk.

METHODS

RESULTS

Of 87,688 study subjects, 54.8% were women, and the mean age was 53.8 years (standard deviation [SD]=5.5). Nearly half of the cohort resided in Southern states (Table 1). In the baseline 6‐month interval, the most common chronic noncancer pain conditions were musculoskeletal involving large joint arthritis/arthralgia (38.4%) and back pain (28.2%). In regard to mental health and substance use conditions, both anxiety/PTSD and depression were diagnosed in approximately 7% of the cohort, whereas psychosis, and alcohol and other substance use disorders were each diagnosed in <2%. In the baseline interval, 12.7% of subjects were hospitalized. The majority of patients received a daily opioid dose of 20 to 49 mg, and the median total dose was 450 mg. The median percent of time exposed to opioids was 6.7% among all study subjects and 70% for those with a high total dose (>1830 mg).

In the 3 study years, an average of 12% of the cohort was hospitalized yearly (Table 2), or 1120 hospitalizations per 10,000 person‐years. Among those who were hospitalized, inpatient days averaged 6.5 (SD=8.5). The highest proportion of hospitalized subjects was 6.5%, occurring in the 6‐month interval immediately following the first opioid treatment interval. In subsequent 6‐month intervals, hospitalization rates were relatively stable, ranging from 5.2% to 6.1% (Table 2). As shown, future hospitalization rates increased monotonically, with increasing total or daily dose within each 6‐month interval.

In unadjusted analyses, a significant interaction between daily dose and total dose (P<0.001) revealed that, within each daily dose category, the odds of hospitalization differed by total dose (all P<0.05, Table 3). When the total dose was >1830 mg, the odds of future hospitalization rose monotonically with increasing daily dose (i.e., <20, 2049, 5099, 100 mg): 1.33, 1.84, 1.96, and 2.08 (P<0.05 for all comparisons vs no opioids). On the other hand, when the total dose was 450 mg or less, all daily dose categories including a very high daily dose (100 mg) were not associated with future hospitalization (all P>0.05 vs no opioids). When the total dose was 451 to 1830 mg, a nonlinear association with hospitalization appeared with higher odds for lower daily doses. For the outcome of hospital days per 6‐month interval, increasing daily dose was also associated with more hospital days per 6‐month interval when the total dose was high (>1830 mg), whereas for lower total doses, daily dose was weakly positive or even protective versus no opioids.

In the model adjusting for all covariates (Table 4), the interaction between total dose and daily dose was also significant (P=0.002). When the total dose was high (>1830 mg), the adjusted odds of future hospitalization were significantly increased by 35% to 44% for daily doses of 20 to 49 mg or greater versus no opioids (P<0.05 for all comparisons). When the total dose was <1830 mg, the majority of daily dose categories were not significantly associated with hospitalization. Similarly, in the fully adjusted analysis of hospital days, the number of inpatient days were increased by 28% to 48% when the total dose was >1830 mg and daily dose was >20 mg, but these associations were nonsignificant or protective when the total dose was lower.

In a sensitivity analysis, we examined the percentage of days covered by filled opioid prescriptions within a 6‐month interval for subjects receiving high‐dose therapy (Table 5). Compared with no opioid therapy, the adjusted odds of future hospitalization were 5% greater for low total opioid dose (11830 mg) and 21% greater for high total dose (>1830 mg) when the duration of treatment was shorter (50% of the 6‐month interval). However, the odds were increased by 41% to 51% for a high total dose (>1830 mg), with longer periods of treatment (>50% of the interval). For hospital days as the outcome, subjects with high total doses (>1830 mg) and longer periods of treatment (>50% of the interval) had 41% to 71% more hospital days per 6‐month interval than those with no opioid therapy.

DISCUSSION

In a national cohort of HMO enrollees who filled at least 2 prescriptions for OAs, 12% were hospitalized annually. Other studies of opioid users have focused on only a fraction of these hospitalizations. For example, a recent Agency for Healthcare Research and Quality study reported that the rate of hospitalization for complications from accidental or deliberate overuse of opioids more than doubled from 11.7/10,000 in 1993 to 29.5/10,000 in 2010.[20] However, in our cohort, the all‐cause hospitalization rate was 1120 per 10,000 person‐years, or over 40 times greater than the rate for complications from overuse of opioids. By comparison, hospitalization for heart failure was only 32.8/10,000 nationally in 2010.[21] Thus, our study confirms the significant demand for hospital care by patients treated with opioids. A novel finding of our study is that the total dose of prescriptions filled over 6 months is significantly associated with an increased risk of future hospitalization. When the total dose within 6 months was in the top quartile (>1830 mg in our cohort), the adjusted odds of future hospitalization ranged from 35% to 44% greater than no opioids for daily opioid doses above 20 mg/day. On the other hand, when the total dose was 1830 mg, the daily opioid dose was only weakly associated with future hospitalization. These associations were similar for hospital days per 6‐month interval as the outcome.

Edlund and colleagues examined the total dose of opioids in a national cohort of veterans with chronic noncancer pain who filled at least 1 opioid prescription.[22] In 2011, the 60th percentile for the total opioid dose for these veterans was 3610 mg within a year, which is roughly equivalent to our top quartile (1830 mg) over a 6‐month interval. These data support replicating our study in veterans to evaluate whether a similarly increased risk of hospitalization appears for those with high total opioid doses. In support of a concern among veterans, a population‐based, cross‐sectional study of hospitalized veterans reported a high rate of chronic opioid therapy (90 days) in the 6 months prior to hospitalization.[23]

Other studies have reported increased risk of hospitalization with chronic opioid therapy. Among 1045 patients followed up to 1‐year post‐transplantation, long‐term opioids were associated with up to a sixfold greater risk of at least 4 admissions within that year.[24] Among 13,127 Danish adults on opioid therapy, the odds of future hospitalization from injuries were increased by 74% for long‐term therapy and 46% for short‐term therapy versus no opioids and by threefold and 1.6‐fold, respectively, for hospitalization due to toxicity/poisoning.[9] However, none of these studies examined the dose of opioids.

In a sensitivity analysis, we found that when a subject received a high total opioid dose within 6 months, treatment for more than 50% of the interval (i.e., >3 months) was associated with a significantly increased risk of future hospitalization and significantly more hospital days. Because the strongest evidence for the benefit of opioids for chronic noncancer pain comes from trials of <3 months,[25] these data lend additional support to recommendations to minimize both dose and duration of opioid therapy.

Our study has several limitations. First, we did not assess the immediate risk of hospitalization after starting opioid therapy. Second, our outcome of hospitalization represents only 1 measure of risk. Thus, our data should not be regarded as supporting short‐term use of high‐dose opioids over 100 to 120 mg per day.[26] In an earlier study, we reported that either a high daily dose (100 mg) or a moderately high daily dose (5099 mg) plus a high total dose (>1830 mg) increased the risk of drug overdose.[10] Third, we could not examine the reason for hospitalizations in this analysis. Therefore, we cannot presume that opioid therapy caused these hospitalizations, but it likely serves as a proxy for other factors such as disability and mental health disorders that increase risk of hospitalization. However, we did adjust for pain conditions as well as mental health and substance abuse disorders that are known to increase the risk of hospitalization in other cohorts.[27, 28, 29, 30] In a national veterans study, the most common clinical conditions associated with long‐term opioid therapy were major depression and PTSD.[22] Last, we did also not consider the number of prescribers of opioids. In a Medicare study, 1 versus 4 prescribers of OAs increased patients' annual hospitalization rate from 1.6% to 4.8%, respectively.[31]

Although the total opioid dose categories observed for our study population may differ from those in other cohorts, these data offer additional evidence for clinicians to consider this measure when assessing risk for hospitalization, and among subjects on high total doses, the percentage of time on opioids offers an additional measure of risk. Because opioid users with noncancer pain are heavy consumers of healthcare services,^32,33 public health benefits and reductions in costs of care may be substantial if opportunities can be identified to reduce hospital utilization by persons treated with higher doses of OAs.

Disclosures

The work on this project was supported by an intramural grant from the National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health, through Grant 1UL TR001120. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The authors report no conflicts of interest.

Cardiopulmonary arrest (CPA) is a major cause of morbidity and mortality, both in the out‐of‐hospital environment as well as the inpatient setting.[1, 2] Unlike the out‐of‐hospital environment, inpatient CPA is unique in that healthcare providers are present during the prearrest period. In theory, this allows the opportunity to intervene and potentially prevent arrest. However, multiple investigators have demonstrated that the vast majority of inpatient CPA victims demonstrate abnormal vital signs prior to arrest without antecedent therapeutic intervention.[3, 4, 5]

Rapid response teams (RRTs) were created to institutionalize the response to at‐risk patients based on chief complaint or vital sign abnormalities. Early evaluation by a critical care team and initiation of appropriate therapy based on defined activation criteria should prevent deterioration in a substantial portion of patients at risk for CPA. Unfortunately, RRTs have not consistently demonstrated improved outcomes on the incidence of CPA and hospital mortality.[6, 7, 8, 9, 10, 11, 12, 13, 14] Although some investigators have reported a decrease in nonintensive care unit (ICU) arrests, it has been posited that this finding appears to be highly associated with either an increase in ICU arrests or more aggressive do not attempt resuscitation (DNAR) orders.[8]

Another potential explanation is an absence of training models that focus on the primary inpatient healthcare providers who are directly responsible for the afferent portion of an RRT program. Here we describe our experience with a novel RRT curriculum, in which unit managers (ie, charge nurse) play an essential role, and substantial education is directed toward primary responders such as bedside nurses and respiratory therapists. This approach is implemented through our novel resuscitation curriculum, which represents a comprehensive approach to inpatient resuscitation management built around critical links between continuous quality improvement (CQI) data, training, and special initiatives.

METHODS

RESULTS

Starting preimplementation year 2006 through postimplementation years 2007 to 2011, the incidence of non‐ICU CPA decreased from 2.7 to 1.1 arrests per 1000 discharges (P<0.0001). The incidence of ICU CPA remained unchanged following program implementation (P=0.532) (Figure 1). Overall hospital mortality also decreased over the study period 2006 to 2011 (2.12%1.74%, P<0.001) (Figure 2). Overall hospital CMI for fiscal years 2005/2006 through 2011/2012 were significantly different (1.47 vs 1.67, P<0.0001). No difference was observed in the pre‐ and postimplementation period likelihood of DNAR status among nonsurvivors with initial return of spontaneous circulation at the time of CPA resuscitation (76% vs 75%, P=0.841).

Figure 1

Figure 2

Starting fiscal year 2005to 2011, there were a total of 546 total CPAs with 247 non‐ICU CPAs observed between both hospital systems. Since its implementation starting at Q3 of 2007, a total of 1729 RRT activations throughout all inpatient areas were observed through 2011. The overall relationship between Code Blue activations starting July 2005 and the RRT since its implementation is displayed in Figure 3. No relationship was detected between the number of Code Blue and RRT activations on each unit (r=0.17, P=0.242). However, the year‐over‐year (fiscal years 2007/20082008/2009) change in RRT activations for each unit was inversely related to the change in Code Blue activations; the individual units with an increase in RRT activations experienced a decrease in Code Blue activations, and units with a decrease in RRT activations experienced an increase in Code Blue activations (r=0.68, P<0.001) (Figure 4).

Figure 3

Figure 4

The time per RRT activation appeared to stabilize after the first program year, whereas the number of RRT activations per month has increased over time. The RRT activation etiologies based on chief complaint reported to the RRT nurse are displayed in Figure 5. Only 3% of activations resulted in no intervention; most of these represented seizures that resolved prior to RRT arrival or syncope episodes for noninpatients. The most common interventions were airway management (27%), fluid therapy (18%), and respiratory treatments (15%). The vast majority of patients (99%) survived the RRT response. A total of 56% of RRT activations resulted in disposition to a higher level of care (43% upgraded to ICU status, 13% upgraded to intermediate care unit status), whereas the remaining 44% of patients stayed on their original unit.

Figure 5

DISCUSSION

One of the primary rationales for hospital admission is the ability to observe and monitor patients to identify deterioration and prevent CPA. Thus, the failure of RRT programs to consistently demonstrate improvements in overall hospital mortality is somewhat perplexing. Here we present 4 years of data starting at the initiation of our RRT program. During our period of observation, we noted a significant inverse relationship between the activations of the RRT and incidence of CPA. Additionally, we noticed a significant decrease in non‐ICU arrests and in overall hospital mortality that appears to be associated with initiation of our novel RRT (Figures 1 and 2, respectively). In postintervention year 1, the unadjusted implementation of our RRT was associated with approximately 57% of the improvement of in‐hospital mortality. We noted a decrease of non‐ICU mortality from 45 deaths in fiscal year 2006 to 2007 to 21 deaths in fiscal year 2007 to 2008. Additionally, the number of non‐ICU Code Blue activations also declined during this period, from 56 to 34, with a survival to hospital discharge rate of 38% (fiscal year 2007 to 2008). In the second full year postimplementation, the activation of the RRT appears to be associated with a similar amount of reduction in in‐hospital mortality (52%) (see Supporting Table 2 in the online version of this article). We do not believe that initiation of our novel RRT accounts for all the variance in reduction of overall mortality. However, we posit that much of our success likely reflects our unique combined approach to a multifaceted life‐support training curriculum. Finally, though not statistically investigated, RRT activation etiology appears to be relatively uniform with respiratory, suspected cardiovascular, and clinical intuition accounting for a large majority of activations through the study period (Figure 5).

Multiple potential explanations exist to account for the lack of outcomes data to fully support RRT programs, with a failure to follow published RRT activation guidelines listed as a key factor.[7, 8, 19] It is unclear whether this reflects inconsistent activation protocols, as many hospitals lack specific published guidelines, whereas others may have more have rigid protocols. Another consideration is possible reluctance to activate a specialized team by primary in‐hospital caregivers due in part to intimidation or a lack of specific training. Interestingly, the routine presence of a physician with RRT activations has been postulated to be inhibitory in this regard and result in critical delays to resuscitative care.[20, 21, 22]

The vast majority of our RRT activations survived the initial response. This is an important metric not only for determining the competency of RRT providers in initiating therapies but also reflects the willingness of inpatient staff to activate RRT early in the course of a patient's deterioration, as delayed activation has been previously associated with increased mortality.[23, 24, 25] The relatively high RRT survival rate could be interpreted as reflecting some degree of overactivation. Of note, based upon hospital CMI, overall patient acuity appeared to continuously increase during and after the observation period. However, the incidence of RRT activations in which no therapies were initiated was extremely low, and more than half of patients were transferred to a higher level of care. We posit that benchmarking such metrics in the future may help institutions guide their resuscitation programs.

Our current RRT configuration of a dedicated critical care trained nurse and respiratory therapist plus unit manager (ie, charge nurse) is a departure from the traditional RRT historically consisting of a dedicated critical care trained nurse and/or respiratory therapist, and physician(s).[26] Perhaps more important than team configuration is our novel concept of individual unit managers regularly rounding on their own most at‐risk patients, which may add a layer of familiarity and increased likelihood of identifying even subtle changes associated with eventual decompensation. Unfortunately, we did not assess the charge nurse decision‐making process regarding RRT activation. This approach differs from Gerdik et al., who have demonstrated a positive association with the ability of the patient and/or family to activate the RRT. [27] Though similar, our approach also differs from the strategy of a dedicated RRT nurse rounding on high risk patients identified through physician and nurse surveys that have also shown a significant reduction in admission deaths.[26, 28, 29] Although the strategies may differ on the specific team members initiating the deployment of the RRT, what they appear to have in common is the proactive component of the identification of at‐risk patients. Additionally, we employ an annual RRT educational seminar for potential primary responders including bedside nurses, respiratory therapists, and physical therapists. Our novel resuscitation program and RRT education allows modulation of the life‐support curriculum to emphasize the importance of early recognition and response to at‐risk patients based upon our activation criteria and evaluation of activation trends (see Supporting Table 1 in the online version of this article). Finally, the presence of the charge nurse as part of the RRT is important, not only as part of the afferent arm of the program but also to enhance unit responsibility for detecting deterioration. It is our belief that an aggressive resuscitation CQI program with efferent links to unit managers amplifies the perception of ownership by primary providers and enhances the culture of resuscitation.

A lack of understanding as to the etiology of CPA in the inpatient environment may also limit the effectiveness of protocol and monitoring strategies. Our current resuscitation program places great emphasis on the taxonomy of our in‐hospital cardiac arrests and classifying inpatient events to help guide CQI efforts. These classifications provide a scaffolding for life‐support education and can result in changes to treatment algorithms or initiate new programs to target particular patient populations. An example includes the implementation of respiratory monitoring strategies in perioperative patients at high risk for obstructive sleep apnea.

Several limitations to this analysis must be considered. The study was not a randomized prospective trial and lacks internal validation. The before‐and‐after study was limited by the inclusion of only 1 complete preimplementation year (2006), which may have introduced a bias related to the inherent inability to properly evaluate secular trends. As such, this study cannot compare the relative effectiveness of our novel RRT participants and curriculum versus the traditional RRTs previously described.[26] Additionally, we also excluded data from both the emergency department and operating arena. Both are reported to have overall higher survival rates due to differences in arrest etiology, monitoring, anticipation, and available personal.[30, 31] We used CMI coefficients to explore the possibility of a decrease in patient acuity during the study. However, we noticed an increasing case‐mix coefficient value suggesting higher patient acuity, which would predict increased mortality rather than the decrease we observed.

One must also consider how the introduction of an RRT may increase DNAR orders and subsequently affect overall mortality by artificially lowering it.[32] Trends in DNAR during our period of observation were not significantly different. However, if an increase in DNAR orders did artificially improve non‐ICU CPA outcomes, one would expect unchanged or increased overall hospital mortality. In contrast, we found improvement in all outcomes including overall hospital mortality (Figure 2).

CONCLUSIONS

Our novel RRT program, with an emphasis on inclusion of non‐ICU charge nurses as part of the team and universal RRT education integrated within life support training, appears to be effective at decreasing the incidence of non‐ICU CPA and overall hospital mortality.

Disclosure: Nothing to report.

Cardiopulmonary arrest (CPA) is a major cause of morbidity and mortality, both in the out‐of‐hospital environment as well as the inpatient setting.[1, 2] Unlike the out‐of‐hospital environment, inpatient CPA is unique in that healthcare providers are present during the prearrest period. In theory, this allows the opportunity to intervene and potentially prevent arrest. However, multiple investigators have demonstrated that the vast majority of inpatient CPA victims demonstrate abnormal vital signs prior to arrest without antecedent therapeutic intervention.[3, 4, 5]

Rapid response teams (RRTs) were created to institutionalize the response to at‐risk patients based on chief complaint or vital sign abnormalities. Early evaluation by a critical care team and initiation of appropriate therapy based on defined activation criteria should prevent deterioration in a substantial portion of patients at risk for CPA. Unfortunately, RRTs have not consistently demonstrated improved outcomes on the incidence of CPA and hospital mortality.[6, 7, 8, 9, 10, 11, 12, 13, 14] Although some investigators have reported a decrease in nonintensive care unit (ICU) arrests, it has been posited that this finding appears to be highly associated with either an increase in ICU arrests or more aggressive do not attempt resuscitation (DNAR) orders.[8]

Another potential explanation is an absence of training models that focus on the primary inpatient healthcare providers who are directly responsible for the afferent portion of an RRT program. Here we describe our experience with a novel RRT curriculum, in which unit managers (ie, charge nurse) play an essential role, and substantial education is directed toward primary responders such as bedside nurses and respiratory therapists. This approach is implemented through our novel resuscitation curriculum, which represents a comprehensive approach to inpatient resuscitation management built around critical links between continuous quality improvement (CQI) data, training, and special initiatives.

METHODS

RESULTS

Starting preimplementation year 2006 through postimplementation years 2007 to 2011, the incidence of non‐ICU CPA decreased from 2.7 to 1.1 arrests per 1000 discharges (P<0.0001). The incidence of ICU CPA remained unchanged following program implementation (P=0.532) (Figure 1). Overall hospital mortality also decreased over the study period 2006 to 2011 (2.12%1.74%, P<0.001) (Figure 2). Overall hospital CMI for fiscal years 2005/2006 through 2011/2012 were significantly different (1.47 vs 1.67, P<0.0001). No difference was observed in the pre‐ and postimplementation period likelihood of DNAR status among nonsurvivors with initial return of spontaneous circulation at the time of CPA resuscitation (76% vs 75%, P=0.841).

Figure 1

Figure 2

Starting fiscal year 2005to 2011, there were a total of 546 total CPAs with 247 non‐ICU CPAs observed between both hospital systems. Since its implementation starting at Q3 of 2007, a total of 1729 RRT activations throughout all inpatient areas were observed through 2011. The overall relationship between Code Blue activations starting July 2005 and the RRT since its implementation is displayed in Figure 3. No relationship was detected between the number of Code Blue and RRT activations on each unit (r=0.17, P=0.242). However, the year‐over‐year (fiscal years 2007/20082008/2009) change in RRT activations for each unit was inversely related to the change in Code Blue activations; the individual units with an increase in RRT activations experienced a decrease in Code Blue activations, and units with a decrease in RRT activations experienced an increase in Code Blue activations (r=0.68, P<0.001) (Figure 4).

Figure 3

Figure 4

The time per RRT activation appeared to stabilize after the first program year, whereas the number of RRT activations per month has increased over time. The RRT activation etiologies based on chief complaint reported to the RRT nurse are displayed in Figure 5. Only 3% of activations resulted in no intervention; most of these represented seizures that resolved prior to RRT arrival or syncope episodes for noninpatients. The most common interventions were airway management (27%), fluid therapy (18%), and respiratory treatments (15%). The vast majority of patients (99%) survived the RRT response. A total of 56% of RRT activations resulted in disposition to a higher level of care (43% upgraded to ICU status, 13% upgraded to intermediate care unit status), whereas the remaining 44% of patients stayed on their original unit.

Figure 5

DISCUSSION

One of the primary rationales for hospital admission is the ability to observe and monitor patients to identify deterioration and prevent CPA. Thus, the failure of RRT programs to consistently demonstrate improvements in overall hospital mortality is somewhat perplexing. Here we present 4 years of data starting at the initiation of our RRT program. During our period of observation, we noted a significant inverse relationship between the activations of the RRT and incidence of CPA. Additionally, we noticed a significant decrease in non‐ICU arrests and in overall hospital mortality that appears to be associated with initiation of our novel RRT (Figures 1 and 2, respectively). In postintervention year 1, the unadjusted implementation of our RRT was associated with approximately 57% of the improvement of in‐hospital mortality. We noted a decrease of non‐ICU mortality from 45 deaths in fiscal year 2006 to 2007 to 21 deaths in fiscal year 2007 to 2008. Additionally, the number of non‐ICU Code Blue activations also declined during this period, from 56 to 34, with a survival to hospital discharge rate of 38% (fiscal year 2007 to 2008). In the second full year postimplementation, the activation of the RRT appears to be associated with a similar amount of reduction in in‐hospital mortality (52%) (see Supporting Table 2 in the online version of this article). We do not believe that initiation of our novel RRT accounts for all the variance in reduction of overall mortality. However, we posit that much of our success likely reflects our unique combined approach to a multifaceted life‐support training curriculum. Finally, though not statistically investigated, RRT activation etiology appears to be relatively uniform with respiratory, suspected cardiovascular, and clinical intuition accounting for a large majority of activations through the study period (Figure 5).

Multiple potential explanations exist to account for the lack of outcomes data to fully support RRT programs, with a failure to follow published RRT activation guidelines listed as a key factor.[7, 8, 19] It is unclear whether this reflects inconsistent activation protocols, as many hospitals lack specific published guidelines, whereas others may have more have rigid protocols. Another consideration is possible reluctance to activate a specialized team by primary in‐hospital caregivers due in part to intimidation or a lack of specific training. Interestingly, the routine presence of a physician with RRT activations has been postulated to be inhibitory in this regard and result in critical delays to resuscitative care.[20, 21, 22]

The vast majority of our RRT activations survived the initial response. This is an important metric not only for determining the competency of RRT providers in initiating therapies but also reflects the willingness of inpatient staff to activate RRT early in the course of a patient's deterioration, as delayed activation has been previously associated with increased mortality.[23, 24, 25] The relatively high RRT survival rate could be interpreted as reflecting some degree of overactivation. Of note, based upon hospital CMI, overall patient acuity appeared to continuously increase during and after the observation period. However, the incidence of RRT activations in which no therapies were initiated was extremely low, and more than half of patients were transferred to a higher level of care. We posit that benchmarking such metrics in the future may help institutions guide their resuscitation programs.

Our current RRT configuration of a dedicated critical care trained nurse and respiratory therapist plus unit manager (ie, charge nurse) is a departure from the traditional RRT historically consisting of a dedicated critical care trained nurse and/or respiratory therapist, and physician(s).[26] Perhaps more important than team configuration is our novel concept of individual unit managers regularly rounding on their own most at‐risk patients, which may add a layer of familiarity and increased likelihood of identifying even subtle changes associated with eventual decompensation. Unfortunately, we did not assess the charge nurse decision‐making process regarding RRT activation. This approach differs from Gerdik et al., who have demonstrated a positive association with the ability of the patient and/or family to activate the RRT. [27] Though similar, our approach also differs from the strategy of a dedicated RRT nurse rounding on high risk patients identified through physician and nurse surveys that have also shown a significant reduction in admission deaths.[26, 28, 29] Although the strategies may differ on the specific team members initiating the deployment of the RRT, what they appear to have in common is the proactive component of the identification of at‐risk patients. Additionally, we employ an annual RRT educational seminar for potential primary responders including bedside nurses, respiratory therapists, and physical therapists. Our novel resuscitation program and RRT education allows modulation of the life‐support curriculum to emphasize the importance of early recognition and response to at‐risk patients based upon our activation criteria and evaluation of activation trends (see Supporting Table 1 in the online version of this article). Finally, the presence of the charge nurse as part of the RRT is important, not only as part of the afferent arm of the program but also to enhance unit responsibility for detecting deterioration. It is our belief that an aggressive resuscitation CQI program with efferent links to unit managers amplifies the perception of ownership by primary providers and enhances the culture of resuscitation.

A lack of understanding as to the etiology of CPA in the inpatient environment may also limit the effectiveness of protocol and monitoring strategies. Our current resuscitation program places great emphasis on the taxonomy of our in‐hospital cardiac arrests and classifying inpatient events to help guide CQI efforts. These classifications provide a scaffolding for life‐support education and can result in changes to treatment algorithms or initiate new programs to target particular patient populations. An example includes the implementation of respiratory monitoring strategies in perioperative patients at high risk for obstructive sleep apnea.

Several limitations to this analysis must be considered. The study was not a randomized prospective trial and lacks internal validation. The before‐and‐after study was limited by the inclusion of only 1 complete preimplementation year (2006), which may have introduced a bias related to the inherent inability to properly evaluate secular trends. As such, this study cannot compare the relative effectiveness of our novel RRT participants and curriculum versus the traditional RRTs previously described.[26] Additionally, we also excluded data from both the emergency department and operating arena. Both are reported to have overall higher survival rates due to differences in arrest etiology, monitoring, anticipation, and available personal.[30, 31] We used CMI coefficients to explore the possibility of a decrease in patient acuity during the study. However, we noticed an increasing case‐mix coefficient value suggesting higher patient acuity, which would predict increased mortality rather than the decrease we observed.

One must also consider how the introduction of an RRT may increase DNAR orders and subsequently affect overall mortality by artificially lowering it.[32] Trends in DNAR during our period of observation were not significantly different. However, if an increase in DNAR orders did artificially improve non‐ICU CPA outcomes, one would expect unchanged or increased overall hospital mortality. In contrast, we found improvement in all outcomes including overall hospital mortality (Figure 2).

CONCLUSIONS

Our novel RRT program, with an emphasis on inclusion of non‐ICU charge nurses as part of the team and universal RRT education integrated within life support training, appears to be effective at decreasing the incidence of non‐ICU CPA and overall hospital mortality.

Disclosure: Nothing to report.

Cardiopulmonary arrest (CPA) is a major cause of morbidity and mortality, both in the out‐of‐hospital environment as well as the inpatient setting.[1, 2] Unlike the out‐of‐hospital environment, inpatient CPA is unique in that healthcare providers are present during the prearrest period. In theory, this allows the opportunity to intervene and potentially prevent arrest. However, multiple investigators have demonstrated that the vast majority of inpatient CPA victims demonstrate abnormal vital signs prior to arrest without antecedent therapeutic intervention.[3, 4, 5]

Rapid response teams (RRTs) were created to institutionalize the response to at‐risk patients based on chief complaint or vital sign abnormalities. Early evaluation by a critical care team and initiation of appropriate therapy based on defined activation criteria should prevent deterioration in a substantial portion of patients at risk for CPA. Unfortunately, RRTs have not consistently demonstrated improved outcomes on the incidence of CPA and hospital mortality.[6, 7, 8, 9, 10, 11, 12, 13, 14] Although some investigators have reported a decrease in nonintensive care unit (ICU) arrests, it has been posited that this finding appears to be highly associated with either an increase in ICU arrests or more aggressive do not attempt resuscitation (DNAR) orders.[8]

Another potential explanation is an absence of training models that focus on the primary inpatient healthcare providers who are directly responsible for the afferent portion of an RRT program. Here we describe our experience with a novel RRT curriculum, in which unit managers (ie, charge nurse) play an essential role, and substantial education is directed toward primary responders such as bedside nurses and respiratory therapists. This approach is implemented through our novel resuscitation curriculum, which represents a comprehensive approach to inpatient resuscitation management built around critical links between continuous quality improvement (CQI) data, training, and special initiatives.

METHODS

RESULTS

Starting preimplementation year 2006 through postimplementation years 2007 to 2011, the incidence of non‐ICU CPA decreased from 2.7 to 1.1 arrests per 1000 discharges (P<0.0001). The incidence of ICU CPA remained unchanged following program implementation (P=0.532) (Figure 1). Overall hospital mortality also decreased over the study period 2006 to 2011 (2.12%1.74%, P<0.001) (Figure 2). Overall hospital CMI for fiscal years 2005/2006 through 2011/2012 were significantly different (1.47 vs 1.67, P<0.0001). No difference was observed in the pre‐ and postimplementation period likelihood of DNAR status among nonsurvivors with initial return of spontaneous circulation at the time of CPA resuscitation (76% vs 75%, P=0.841).

Figure 1

Figure 2

Starting fiscal year 2005to 2011, there were a total of 546 total CPAs with 247 non‐ICU CPAs observed between both hospital systems. Since its implementation starting at Q3 of 2007, a total of 1729 RRT activations throughout all inpatient areas were observed through 2011. The overall relationship between Code Blue activations starting July 2005 and the RRT since its implementation is displayed in Figure 3. No relationship was detected between the number of Code Blue and RRT activations on each unit (r=0.17, P=0.242). However, the year‐over‐year (fiscal years 2007/20082008/2009) change in RRT activations for each unit was inversely related to the change in Code Blue activations; the individual units with an increase in RRT activations experienced a decrease in Code Blue activations, and units with a decrease in RRT activations experienced an increase in Code Blue activations (r=0.68, P<0.001) (Figure 4).

Figure 3

Figure 4

The time per RRT activation appeared to stabilize after the first program year, whereas the number of RRT activations per month has increased over time. The RRT activation etiologies based on chief complaint reported to the RRT nurse are displayed in Figure 5. Only 3% of activations resulted in no intervention; most of these represented seizures that resolved prior to RRT arrival or syncope episodes for noninpatients. The most common interventions were airway management (27%), fluid therapy (18%), and respiratory treatments (15%). The vast majority of patients (99%) survived the RRT response. A total of 56% of RRT activations resulted in disposition to a higher level of care (43% upgraded to ICU status, 13% upgraded to intermediate care unit status), whereas the remaining 44% of patients stayed on their original unit.

Figure 5

DISCUSSION

One of the primary rationales for hospital admission is the ability to observe and monitor patients to identify deterioration and prevent CPA. Thus, the failure of RRT programs to consistently demonstrate improvements in overall hospital mortality is somewhat perplexing. Here we present 4 years of data starting at the initiation of our RRT program. During our period of observation, we noted a significant inverse relationship between the activations of the RRT and incidence of CPA. Additionally, we noticed a significant decrease in non‐ICU arrests and in overall hospital mortality that appears to be associated with initiation of our novel RRT (Figures 1 and 2, respectively). In postintervention year 1, the unadjusted implementation of our RRT was associated with approximately 57% of the improvement of in‐hospital mortality. We noted a decrease of non‐ICU mortality from 45 deaths in fiscal year 2006 to 2007 to 21 deaths in fiscal year 2007 to 2008. Additionally, the number of non‐ICU Code Blue activations also declined during this period, from 56 to 34, with a survival to hospital discharge rate of 38% (fiscal year 2007 to 2008). In the second full year postimplementation, the activation of the RRT appears to be associated with a similar amount of reduction in in‐hospital mortality (52%) (see Supporting Table 2 in the online version of this article). We do not believe that initiation of our novel RRT accounts for all the variance in reduction of overall mortality. However, we posit that much of our success likely reflects our unique combined approach to a multifaceted life‐support training curriculum. Finally, though not statistically investigated, RRT activation etiology appears to be relatively uniform with respiratory, suspected cardiovascular, and clinical intuition accounting for a large majority of activations through the study period (Figure 5).

Multiple potential explanations exist to account for the lack of outcomes data to fully support RRT programs, with a failure to follow published RRT activation guidelines listed as a key factor.[7, 8, 19] It is unclear whether this reflects inconsistent activation protocols, as many hospitals lack specific published guidelines, whereas others may have more have rigid protocols. Another consideration is possible reluctance to activate a specialized team by primary in‐hospital caregivers due in part to intimidation or a lack of specific training. Interestingly, the routine presence of a physician with RRT activations has been postulated to be inhibitory in this regard and result in critical delays to resuscitative care.[20, 21, 22]

The vast majority of our RRT activations survived the initial response. This is an important metric not only for determining the competency of RRT providers in initiating therapies but also reflects the willingness of inpatient staff to activate RRT early in the course of a patient's deterioration, as delayed activation has been previously associated with increased mortality.[23, 24, 25] The relatively high RRT survival rate could be interpreted as reflecting some degree of overactivation. Of note, based upon hospital CMI, overall patient acuity appeared to continuously increase during and after the observation period. However, the incidence of RRT activations in which no therapies were initiated was extremely low, and more than half of patients were transferred to a higher level of care. We posit that benchmarking such metrics in the future may help institutions guide their resuscitation programs.

Our current RRT configuration of a dedicated critical care trained nurse and respiratory therapist plus unit manager (ie, charge nurse) is a departure from the traditional RRT historically consisting of a dedicated critical care trained nurse and/or respiratory therapist, and physician(s).[26] Perhaps more important than team configuration is our novel concept of individual unit managers regularly rounding on their own most at‐risk patients, which may add a layer of familiarity and increased likelihood of identifying even subtle changes associated with eventual decompensation. Unfortunately, we did not assess the charge nurse decision‐making process regarding RRT activation. This approach differs from Gerdik et al., who have demonstrated a positive association with the ability of the patient and/or family to activate the RRT. [27] Though similar, our approach also differs from the strategy of a dedicated RRT nurse rounding on high risk patients identified through physician and nurse surveys that have also shown a significant reduction in admission deaths.[26, 28, 29] Although the strategies may differ on the specific team members initiating the deployment of the RRT, what they appear to have in common is the proactive component of the identification of at‐risk patients. Additionally, we employ an annual RRT educational seminar for potential primary responders including bedside nurses, respiratory therapists, and physical therapists. Our novel resuscitation program and RRT education allows modulation of the life‐support curriculum to emphasize the importance of early recognition and response to at‐risk patients based upon our activation criteria and evaluation of activation trends (see Supporting Table 1 in the online version of this article). Finally, the presence of the charge nurse as part of the RRT is important, not only as part of the afferent arm of the program but also to enhance unit responsibility for detecting deterioration. It is our belief that an aggressive resuscitation CQI program with efferent links to unit managers amplifies the perception of ownership by primary providers and enhances the culture of resuscitation.

A lack of understanding as to the etiology of CPA in the inpatient environment may also limit the effectiveness of protocol and monitoring strategies. Our current resuscitation program places great emphasis on the taxonomy of our in‐hospital cardiac arrests and classifying inpatient events to help guide CQI efforts. These classifications provide a scaffolding for life‐support education and can result in changes to treatment algorithms or initiate new programs to target particular patient populations. An example includes the implementation of respiratory monitoring strategies in perioperative patients at high risk for obstructive sleep apnea.

Several limitations to this analysis must be considered. The study was not a randomized prospective trial and lacks internal validation. The before‐and‐after study was limited by the inclusion of only 1 complete preimplementation year (2006), which may have introduced a bias related to the inherent inability to properly evaluate secular trends. As such, this study cannot compare the relative effectiveness of our novel RRT participants and curriculum versus the traditional RRTs previously described.[26] Additionally, we also excluded data from both the emergency department and operating arena. Both are reported to have overall higher survival rates due to differences in arrest etiology, monitoring, anticipation, and available personal.[30, 31] We used CMI coefficients to explore the possibility of a decrease in patient acuity during the study. However, we noticed an increasing case‐mix coefficient value suggesting higher patient acuity, which would predict increased mortality rather than the decrease we observed.

One must also consider how the introduction of an RRT may increase DNAR orders and subsequently affect overall mortality by artificially lowering it.[32] Trends in DNAR during our period of observation were not significantly different. However, if an increase in DNAR orders did artificially improve non‐ICU CPA outcomes, one would expect unchanged or increased overall hospital mortality. In contrast, we found improvement in all outcomes including overall hospital mortality (Figure 2).

CONCLUSIONS

Our novel RRT program, with an emphasis on inclusion of non‐ICU charge nurses as part of the team and universal RRT education integrated within life support training, appears to be effective at decreasing the incidence of non‐ICU CPA and overall hospital mortality.

Disclosure: Nothing to report.

The theory behind rapid response teams (RRTs), namely to provide critical care resources to patients with clinical deterioration on the wards, is such common sense that failure to do so seems unethical. This idea, combined with evidence that many cardiac arrests on the wards are predictable and potentially preventable events, led to the proliferation of RRTs across the country and a Joint Commission mandate.[1] However, data from clinical trials have failed to consistently confirm the value of these teams, likely a product of the wide variability in implementation practices across institutions.[2]

In this issue of the Journal of Hospital Medicine, Davis and colleagues demonstrate improvements in both mortality and cardiac arrest rates outside the intensive care unit (ICU) following implementation of their rapid response system in 2 hospitals.[3] Although several other studies have shown similar results, what makes this implementation unique is the bundle approach that included proactive rounding by the charge nurse from each unit, annual focused training of team members and staff, and an integrated, continuous, quality‐improvement feedback loop. Bundles are common in successful quality‐improvement work, but can be challenging for deciphering which of the individual components are driving the results, leaving readers to venture an educated guess. In the current bundle, the novel use of the charge nurse has some significant appeal as a candidate primary driver of the impact, because it likely had 2 distinct actions: (1) proactive rounding and (2) promoting a culture change, both of which are well supported in the literature.^4,5

Several studies, including this one, have demonstrated a dose‐response association between the number of RRT activations and patient outcomes, with a low number of RRT activations deemed a major contributor to the neutral results of the large multicenter, randomized, controlled MERIT trial.[6, 7] Additionally, delays in treatment and transfer to the ICU for unstable patients are known to increase mortality.[8] One way to increase the number of patients seen by the RRT and decrease activation delays is by instituting proactive rounding by the team on high‐risk patients. This was the strategy employed in a landmark ward‐randomized trial by Priestley and colleagues, which demonstrated a significant improvement in mortality from proactive rounding on patients deemed to be at high risk of clinical deterioration as calculated by an early warning score or due to caregiver concern.⁴

Identification of at‐risk patients for proactive rounding can be accomplished with gestalt, as was done by the charge nurse in the current study, or using specific individual criteria such as recent discharge from an ICU. Alternatively, this can be accomplished using composite vital signbased risk scores, such as the Modified Early Warning Score (MEWS).[9] Recently, several newer algorithms that integrate vital signs, laboratory data, and demographics have been shown to outperform the MEWS.[10, 11] Such systems promise an exciting age of real‐time computer‐generated risk stratification, with the ability to automate and standardize the selection of patients for proactive rounding across institutions.

Interestingly, the selection of the charge nurse, rather than someone who did not reside on the unit, to conduct the surveillance rounds likely had another benefit: expediting and facilitating the culture change necessary for a successful implementation. The integration of the charge nurse into the RRT likely led to a local reinforcement of important cultural changes that were already happening at the institutional level. It is clear that culture change is essential in any quality improvement endeavor, and previous literature on RRTs supports this notion.[5]

Rapid response systems are complex and include the activation criteria, team composition and training, and an administrative component. A multifaceted, bundled approach is likely to be required for success. Furthermore, regardless of what risk stratification criteria are used, proactive rounding on high‐risk patients is likely to increase the yield. Utilizing the charge nurse in that effort is a creative use of a preexisting local resource and is worthy of future study.

Disclosures: Dr. Churpek is supported by a career development award from the National Heart, Lung, and Blood Institute (K08 HL121080) and has received honoraria from CHEST for invited speaking engagements. Drs. Churpek and Edelson have a patent pending (ARCD.P0535US.P2) for risk stratification algorithms for hospitalized patients, and Dr. Edelson has an ownership interest in Quant HC (Chicago, IL), which seeks to commercialize those algorithms.

The theory behind rapid response teams (RRTs), namely to provide critical care resources to patients with clinical deterioration on the wards, is such common sense that failure to do so seems unethical. This idea, combined with evidence that many cardiac arrests on the wards are predictable and potentially preventable events, led to the proliferation of RRTs across the country and a Joint Commission mandate.[1] However, data from clinical trials have failed to consistently confirm the value of these teams, likely a product of the wide variability in implementation practices across institutions.[2]

In this issue of the Journal of Hospital Medicine, Davis and colleagues demonstrate improvements in both mortality and cardiac arrest rates outside the intensive care unit (ICU) following implementation of their rapid response system in 2 hospitals.[3] Although several other studies have shown similar results, what makes this implementation unique is the bundle approach that included proactive rounding by the charge nurse from each unit, annual focused training of team members and staff, and an integrated, continuous, quality‐improvement feedback loop. Bundles are common in successful quality‐improvement work, but can be challenging for deciphering which of the individual components are driving the results, leaving readers to venture an educated guess. In the current bundle, the novel use of the charge nurse has some significant appeal as a candidate primary driver of the impact, because it likely had 2 distinct actions: (1) proactive rounding and (2) promoting a culture change, both of which are well supported in the literature.^4,5

Several studies, including this one, have demonstrated a dose‐response association between the number of RRT activations and patient outcomes, with a low number of RRT activations deemed a major contributor to the neutral results of the large multicenter, randomized, controlled MERIT trial.[6, 7] Additionally, delays in treatment and transfer to the ICU for unstable patients are known to increase mortality.[8] One way to increase the number of patients seen by the RRT and decrease activation delays is by instituting proactive rounding by the team on high‐risk patients. This was the strategy employed in a landmark ward‐randomized trial by Priestley and colleagues, which demonstrated a significant improvement in mortality from proactive rounding on patients deemed to be at high risk of clinical deterioration as calculated by an early warning score or due to caregiver concern.⁴

Identification of at‐risk patients for proactive rounding can be accomplished with gestalt, as was done by the charge nurse in the current study, or using specific individual criteria such as recent discharge from an ICU. Alternatively, this can be accomplished using composite vital signbased risk scores, such as the Modified Early Warning Score (MEWS).[9] Recently, several newer algorithms that integrate vital signs, laboratory data, and demographics have been shown to outperform the MEWS.[10, 11] Such systems promise an exciting age of real‐time computer‐generated risk stratification, with the ability to automate and standardize the selection of patients for proactive rounding across institutions.

Interestingly, the selection of the charge nurse, rather than someone who did not reside on the unit, to conduct the surveillance rounds likely had another benefit: expediting and facilitating the culture change necessary for a successful implementation. The integration of the charge nurse into the RRT likely led to a local reinforcement of important cultural changes that were already happening at the institutional level. It is clear that culture change is essential in any quality improvement endeavor, and previous literature on RRTs supports this notion.[5]

Rapid response systems are complex and include the activation criteria, team composition and training, and an administrative component. A multifaceted, bundled approach is likely to be required for success. Furthermore, regardless of what risk stratification criteria are used, proactive rounding on high‐risk patients is likely to increase the yield. Utilizing the charge nurse in that effort is a creative use of a preexisting local resource and is worthy of future study.

Disclosures: Dr. Churpek is supported by a career development award from the National Heart, Lung, and Blood Institute (K08 HL121080) and has received honoraria from CHEST for invited speaking engagements. Drs. Churpek and Edelson have a patent pending (ARCD.P0535US.P2) for risk stratification algorithms for hospitalized patients, and Dr. Edelson has an ownership interest in Quant HC (Chicago, IL), which seeks to commercialize those algorithms.

The theory behind rapid response teams (RRTs), namely to provide critical care resources to patients with clinical deterioration on the wards, is such common sense that failure to do so seems unethical. This idea, combined with evidence that many cardiac arrests on the wards are predictable and potentially preventable events, led to the proliferation of RRTs across the country and a Joint Commission mandate.[1] However, data from clinical trials have failed to consistently confirm the value of these teams, likely a product of the wide variability in implementation practices across institutions.[2]

In this issue of the Journal of Hospital Medicine, Davis and colleagues demonstrate improvements in both mortality and cardiac arrest rates outside the intensive care unit (ICU) following implementation of their rapid response system in 2 hospitals.[3] Although several other studies have shown similar results, what makes this implementation unique is the bundle approach that included proactive rounding by the charge nurse from each unit, annual focused training of team members and staff, and an integrated, continuous, quality‐improvement feedback loop. Bundles are common in successful quality‐improvement work, but can be challenging for deciphering which of the individual components are driving the results, leaving readers to venture an educated guess. In the current bundle, the novel use of the charge nurse has some significant appeal as a candidate primary driver of the impact, because it likely had 2 distinct actions: (1) proactive rounding and (2) promoting a culture change, both of which are well supported in the literature.^4,5

Several studies, including this one, have demonstrated a dose‐response association between the number of RRT activations and patient outcomes, with a low number of RRT activations deemed a major contributor to the neutral results of the large multicenter, randomized, controlled MERIT trial.[6, 7] Additionally, delays in treatment and transfer to the ICU for unstable patients are known to increase mortality.[8] One way to increase the number of patients seen by the RRT and decrease activation delays is by instituting proactive rounding by the team on high‐risk patients. This was the strategy employed in a landmark ward‐randomized trial by Priestley and colleagues, which demonstrated a significant improvement in mortality from proactive rounding on patients deemed to be at high risk of clinical deterioration as calculated by an early warning score or due to caregiver concern.⁴

Identification of at‐risk patients for proactive rounding can be accomplished with gestalt, as was done by the charge nurse in the current study, or using specific individual criteria such as recent discharge from an ICU. Alternatively, this can be accomplished using composite vital signbased risk scores, such as the Modified Early Warning Score (MEWS).[9] Recently, several newer algorithms that integrate vital signs, laboratory data, and demographics have been shown to outperform the MEWS.[10, 11] Such systems promise an exciting age of real‐time computer‐generated risk stratification, with the ability to automate and standardize the selection of patients for proactive rounding across institutions.

Interestingly, the selection of the charge nurse, rather than someone who did not reside on the unit, to conduct the surveillance rounds likely had another benefit: expediting and facilitating the culture change necessary for a successful implementation. The integration of the charge nurse into the RRT likely led to a local reinforcement of important cultural changes that were already happening at the institutional level. It is clear that culture change is essential in any quality improvement endeavor, and previous literature on RRTs supports this notion.[5]

Rapid response systems are complex and include the activation criteria, team composition and training, and an administrative component. A multifaceted, bundled approach is likely to be required for success. Furthermore, regardless of what risk stratification criteria are used, proactive rounding on high‐risk patients is likely to increase the yield. Utilizing the charge nurse in that effort is a creative use of a preexisting local resource and is worthy of future study.

Disclosures: Dr. Churpek is supported by a career development award from the National Heart, Lung, and Blood Institute (K08 HL121080) and has received honoraria from CHEST for invited speaking engagements. Drs. Churpek and Edelson have a patent pending (ARCD.P0535US.P2) for risk stratification algorithms for hospitalized patients, and Dr. Edelson has an ownership interest in Quant HC (Chicago, IL), which seeks to commercialize those algorithms.