Bacterial meningitis can be a devastating disease in children. Overall mortality in children in the United States is 4%1 while long‐term morbidity is present in up to 25%2 of surviving children. The introduction of Haemophilus influenzae type B vaccine, heptavalent pneumococcal conjugate vaccine, and the quadrivalent meningococcal conjugate vaccine has altered the epidemiology of bacterial meningitis.24 Currently, little is known about the epidemiology of systemic complications and associated focal infections that occur during episodes of bacterial meningitis in children and how the presence of such complications affects in‐hospital healthcare resource utilization.

In a randomized controlled trial, the administration of adjuvant corticosteroids was associated with lower mortality rates in adults with bacterial meningitis due to all causes, with the greatest reduction in those with pneumococcal meningitis.5 In a post hoc analysis of data from this trial, reductions in systemic complications, such as septic shock, pneumonia, and acute respiratory distress syndrome, rather than neurologic complications were thought to be the underlying reason for the decrease in mortality associated with pneumococcal meningitis among corticosteroid recipients.6 However, children with bacterial meningitis have an overall 4‐fold lower mortality rate than adults with bacterial meningitis. An even greater difference in mortality rates exists between children and adults with pneumococcal meningitis.1, 5 Children do not benefit from adjuvant corticosteroids as adults do.1, 5, 7 Therefore, the pathogenesis of bacterial meningitis may differ in children from adults and account for the difference in response to adjuvant corticosteroids. Understanding the epidemiology of systemic complications and associated focal infections can aid in the understanding of the pathogenesis of the disease in varying age groups of children.

Previous studies in children have documented the frequency of certain bacterial meningitis‐associated conditions such as respiratory failure, pneumonia, endocarditis, and mastoiditis. Researchers have used the presence of such conditions to predict either mortality or neurologic sequelae in children.810 These studies were small and only included a few types of complications associated with bacterial meningitis. In‐hospital healthcare resource utilization, which may be an important indicator of in‐hospital morbidity, was also not considered as an outcome. In‐hospital morbidity may represent aspects of disease burden not captured by mortality rates or markers for long‐term morbidity alone. In future vaccine efficacy trails or novel therapeutics evaluations, consideration of these associated conditions is important.

The quantification of the use of in‐hospital healthcare utilization is also important for hospital planning and resource allocation in children with bacterial meningitis. A child presenting with bacterial meningitis and a systemic complication or an associated focal infection may require additional resource planning initially to expedite care to enhance recovery and decrease hospital length of stay (LOS).

Our goal was to document the frequency of bacterial meningitis‐associated conditions (systemic complications and associated focal infections) in a large cohort of children with bacterial meningitis treated at tertiary care children's hospitals in the United States, and determine how the presence of such conditions impacted in‐hospital healthcare resource utilization.

Patients and Methods

Results

Bacterial Meningitis‐Associated Conditions

Overall, 574 (25%) of children with bacterial meningitis suffered a systemic complication or an associated focal infection. Figure 1 shows the types of associated condition stratified by age category. Older children had a higher frequency of associated focal infections while younger children had a higher frequency of systemic complications (P = 0.002, chi‐square test for trend). Figure 2 shows the distribution of specific conditions among children in each age category. The frequency of sepsis decreased with age (P < 0.001, chi‐square test) while the frequency of mastoiditis (P < 0.001, Fisher's exact test) and osteomyelitis (P = 0.005, Fisher's exact test) increased with age. There did not appear to be substantial variability in the proportion of patients with SIRS or sepsis across hospitals, suggesting that hospital‐level variability in coding for these conditions was likely minimal. The median proportion of patients with SIRS by hospital was 2.4% (IQR, 1.2‐4.8%) while the median proportion of patients with sepsis by hospital was 13.4% (IQR, 10.0‐16.9%).

Figure 1

Figure 2

Of the 151 children with an associated focal infection, only 3 (2%) of children had more than 1 infection (1 child had mastoiditis and endocarditis, 1 child had pneumonia and osteomyelitis, and 1 child had pneumonia and endocarditis). However, of the 479 children with systemic complications, 116 (24%) had more than 1 systemic disease (Table 1).

Table 1.Children With Systemic Complications

Types of Systemic Complications	Systemic Complications in All Bacterial Meningitis, n (%)	Systemic Complications in Meningococcal Meningitis, n (%)	Systemic Complications in Pneumococcal Meningitis, n (%)
NOTE: Percentages may not add to 100 due to rounding. Abbreviations: SIRS, systemic inflammatory response syndrome.
Sepsis only	209 (44)	16 (21)	69 (54)
Respiratory failure only	139 (29)	38 (49)	30 (24)
SIRS only	15 (3)	9 (12)	1 (1)
Sepsis and respiratory failure	52 (11)	4 (5)	18 (14)
SIRS and sepsis	27 (6)	2 (3)	4 (3)
SIRS and respiratory failure	9 (2)	5 (6)	0 (0)
SIRS and respiratory failure and sepsis	28 (6)	3 (4)	5 (4)
Total systemic complications	479	77	127

In sub‐group analyses, 269 children had meningococcal meningitis and 470 children had pneumococcal meningitis. Of the children with meningococcal meningitis, 31.2% had a meningitis‐associated condition: 26.4% had a systemic complication, 2.6% had a focal infection, and 2.2% had both conditions. The most common associated conditions in children with meningococcal meningitis were respiratory failure (18.6%; n = 50), sepsis (9.3%; n = 25), and SIRS (7.1%; n = 19). In children with pneumococcal meningitis, 32.3% had a meningitis‐associated complication: 24.7% had a systemic complication, 5.3% had a focal infection, and 2.3% had both conditions. The most common associated conditions in children with pneumococcal meningitis were sepsis (20.4%; n = 96), respiratory failure (11.3%; n = 53), and pneumonia (4.7%; n = 22); mastoiditis was present in 2.3% (n = 11) of children with pneumococcal meningitis. Respiratory failure was more common in meningococcal meningitis (18.6%) than in pneumococcal meningitis (11.3%; P = 0.006). In contrast, sepsis was less common in meningococcal meningitis (9.3%) than in pneumococcal meningitis (20.4%; P < 0.001).

Hospital Charges

Overall, the median charges per hospital ranged from $20,158 to $53,823. In‐hospital charges for children with bacterial meningitis with and without any identified associated conditions are presented in Table 2. In multivariate analyses, the presence of systemic conditions, associated focal infections, or both conditions was independently associated with significantly higher total in‐hospital charges (Table 2). When conditions were considered individually, bone and joint infections (213% increase; 95% CI, 113‐260%), endocarditis (108% increase; 95% CI, 23‐258%), and pneumonia (107% increase; 95% CI, 58‐171%) were associated with the highest increases in total hospital charges (Figure 3). In contrast, SIRS and mastoiditis were not associated with higher hospital charges (Figure 3).

Table 2.Comparison of Total Hospital Charges and Length of Stay by Complication Type

	Charges		LOS
	Median, $ (IQR)	Adjusted Increase,* % (95% CI)	Median, days (IQR)	Adjusted Increase,* % (95% CI)
Abbreviations: CI, confidence interval; IQR, interquartile range. * Adjusted for age category, race, sex, vancomycin receipt, and adjuvant corticosteroid therapy. P value <0.001 for all comparisons.
None (n = 1,745)	$27,110 (15,823‐48,307)	Reference**	9 (6‐14)	Reference
Systemic (n = 423)	$66,690 (39,546136,756)	136 (108269)	14 (923)	72 (5196)
Focal Infection (n = 95)	$58,016 (29,056125,813)	118 (77168)	13 (928)	78 (40126)
Both (n = 56)	$130,744 (62,397299,288)	351 (237503)	21.5 (1245)	211 (142303)

Figure 3

LOS

The median LOS was 9 days (IQR, 6‐15 days); 5% of children had a LOS >42 days. Table 2 summarizes difference in LOS by the presence and absence of systemic conditions and focal infections. In multivariate analyses, the presence of systemic conditions, associated focal infections, or both conditions was independently associated with a significantly longer LOS (Table 2). When conditions were considered individually, endocarditis (152% increase; 95% CI, 60‐300%) and pneumonia (136% increase; 95% CI, 85‐201%) were associated with the greatest adjusted increases in LOS (Figure 4); only mastoiditis was not associated with an increased LOS compared with those without complications.

Figure 4

Discussion

To our knowledge, this is the first study to examine bacterial meningitis‐associated conditions in children and their impact on in‐hospital resource utilization. We found that 25% of the cohort of children with bacterial meningitis suffered from at least one focal infection or systemic complication. This represents a significant invasive disease burden among children with bacterial meningitis who do not have underlying comorbid conditions. Younger children were more likely to have systemic complications when compared with older children, specifically due to a higher frequency of sepsis in children <1 year. Older children were more likely to have an associated focal infection, specifically due to an increase in mastoiditis and osteomyelitis in children >1 year. Only 2% of children had more than 1 focal infection, while 24% of children had more than 1 systemic complication.

Importantly, the presence of a systemic complication in a child with bacterial meningitis increased their in‐hospital adjusted charges by 136%. The presence of a focal infection increased in‐hospital adjusted charges by 118%. A child with both a systemic complication and a focal infection and had a 351% increase in in‐hospital adjusted charges.

The presence of systemic complications or associated focal infections was significantly associated with higher in‐hospital charges and longer hospital LOS. Most individual meningitis‐associated conditions included in this study were associated with higher in‐hospital charges with the exception of SIRS and mastoiditis. All individual meningitis‐associated conditions were associated with a longer LOS except mastoiditis. This finding is not surprising as the LOS for children with mastoiditis is typically shorter than for children with bacterial meningitis. Glikich et al.14 reported a mean LOS of approximately 8 days for children with mastoiditis. As meningitis in the context of mastoiditis is likely caused by direct extension of infection, patients with meningitis and mastoiditis likely required extended hospitalization to treat meningitis rather than mastoiditis. In contrast, patients with meningitis occurring in the context of metastatic dissemination of infection (eg, endocarditis, pneumonia) often have hemodynamic instability requiring prolonged intensive care support.

A study of children with sepsis found that increasing severity of illness was associated with greater hospital resource utilization.15 Our study shows that this may also be true in children with bacterial meningitis. We found that in children with bacterial meningitis, having systemic complications or an associated focal infection was associated with greater in‐hospital resource utilization. This finding may therefore indicate greater in‐hospital morbidity among children with a bacterial meningitis‐associated condition. Since mortality rates for bacterial meningitis are low in children, in‐hospital morbidity may be a better indicator of disease burden.

Our data show that, in contrast to adults, bacterial meningitis in children is not typically associated with other focal infections. Some focal complications such as mastoiditis and osteomyelitis disproportionately affect older children. These complications are typically accompanied by overt clinical manifestations. Therefore, we believe that the evaluation for the presence of concomitant focal infections can be guided by clinical examination findings and that routine radiologic evaluation for focal complications may not be necessary. Additionally, focal infections tend to occur in the absence of concomitant systemic complications. Of the 151 children with at least 1 associated focal infection, only 37% had a systemic complication. Bacterial meningitis may lie on a continuum of invasive disease depending on the virulence factors of the invading pathogen as well as specific host factors. Understanding the epidemiology of these associated conditions can enhance our understanding of the pathogenesis of bacterial meningitis in children. Understanding why some children suffer from septicemia rather than bacteremia may help in developing novel therapeutics.

There are several limitations to our study. First, since we identified focal infections and systemic complications using billing charges and ICD‐9 discharge diagnosis codes, it was impossible to determine when these conditions represented true complications of bacterial meningitis and when they represented the primary source of infection. Therefore, some of our primary outcomes may represent the cause of meningitis rather than a direct complication. We attempted to minimize such misclassification by limiting the cohort to those with a primary discharge diagnosis of bacterial meningitis though such misclassification is still possible.

Second, the use of ICD‐9 codes to accurately identify systemic complications and associated focal infections is a potential limitation. For example, respiratory failure, defined as the requirement of endotracheal intubation in our study, may not capture children receiving non‐invasive mechanical ventilation (eg, bilevel positive airway pressure). If use of noninvasive ventilation strategies did not depend exclusively on illness severity, our study would underestimate the frequency of respiratory failure. Furthermore, there may be inconsistencies among pediatric physicians in coding conditions such as SIRS and sepsis. Even in the clinical setting, a uniform definition of SIRS and sepsis is problematic due to physiologic differences between adults and children of varying age groups.16 An international panel of pediatricians proposed age‐specific definitions for sepsis and SIRS, while acknowledging the paucity of evidence to support some of their recommendations.16 None of the proposed definitions could be applied using administrative data. Limitations in the use of ICD‐9 discharge diagnosis codes to identify children with bacterial meningitis were discussed previously.1

Third, only free‐standing children's hospitals were included in the analysis. It is likely that many children with uncomplicated bacterial meningitis are treated at community hospitals or smaller academic centers. Our study may overestimate the rate of bacterial meningitis‐associated focal infections and systemic complications since participating hospitals serve as regional referral centers. To address the potential for such referral bias, we repeated the analysis while restricting the cohort to those children who had a lumbar puncture performed at the treating facility. No difference in frequency of associated conditions or in‐hospital resource utilization was found between children transferred and children not transferred. Finally, the PHIS database reports billed charge data rather than cost data. Billed data may overestimate the actual economic impact of bacterial meningitis‐associated complications since payers often reimburse at lesser rates. Resource utilization may also vary widely between hospitals and geographic locations as previously shown.15

In conclusion, bacterial meningitis remains an important cause of morbidity in children. Systemic complications such as sepsis and respiratory failure are common. Respiratory failure occurred more commonly among patients with meningococcal meningitis while sepsis occurred more commonly among patients with pneumococcal meningitis. While focal complications are uncommon, children >5 years of age are more likely than younger children to have concomitant mastoiditis or osteomyelitis. The presence of both systemic and focal complications is associated with substantially greater resource utilization than either complication alone.

Dr. Shah had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis. Study concept and design: Shah, Mongelluzzo; acquisition of data: Shah, Mohamad; analysis and interpretation of data: Mongelluzzo, Mohamad, Ten Have, Shah; drafting of the manuscript: Mongelluzzo; critical revision of the manuscript for important intellectual content: Mongelluzzo, Mohamad, Ten Have, Shah; statistical analysis: Shah, Mongelluzzo, Ten Have; obtained funding: Shah, Mongelluzzo; administrative, technical, or material support: Shah; study supervision: Shah.

Bacterial meningitis can be a devastating disease in children. Overall mortality in children in the United States is 4%1 while long‐term morbidity is present in up to 25%2 of surviving children. The introduction of Haemophilus influenzae type B vaccine, heptavalent pneumococcal conjugate vaccine, and the quadrivalent meningococcal conjugate vaccine has altered the epidemiology of bacterial meningitis.24 Currently, little is known about the epidemiology of systemic complications and associated focal infections that occur during episodes of bacterial meningitis in children and how the presence of such complications affects in‐hospital healthcare resource utilization.

In a randomized controlled trial, the administration of adjuvant corticosteroids was associated with lower mortality rates in adults with bacterial meningitis due to all causes, with the greatest reduction in those with pneumococcal meningitis.5 In a post hoc analysis of data from this trial, reductions in systemic complications, such as septic shock, pneumonia, and acute respiratory distress syndrome, rather than neurologic complications were thought to be the underlying reason for the decrease in mortality associated with pneumococcal meningitis among corticosteroid recipients.6 However, children with bacterial meningitis have an overall 4‐fold lower mortality rate than adults with bacterial meningitis. An even greater difference in mortality rates exists between children and adults with pneumococcal meningitis.1, 5 Children do not benefit from adjuvant corticosteroids as adults do.1, 5, 7 Therefore, the pathogenesis of bacterial meningitis may differ in children from adults and account for the difference in response to adjuvant corticosteroids. Understanding the epidemiology of systemic complications and associated focal infections can aid in the understanding of the pathogenesis of the disease in varying age groups of children.

Previous studies in children have documented the frequency of certain bacterial meningitis‐associated conditions such as respiratory failure, pneumonia, endocarditis, and mastoiditis. Researchers have used the presence of such conditions to predict either mortality or neurologic sequelae in children.810 These studies were small and only included a few types of complications associated with bacterial meningitis. In‐hospital healthcare resource utilization, which may be an important indicator of in‐hospital morbidity, was also not considered as an outcome. In‐hospital morbidity may represent aspects of disease burden not captured by mortality rates or markers for long‐term morbidity alone. In future vaccine efficacy trails or novel therapeutics evaluations, consideration of these associated conditions is important.

The quantification of the use of in‐hospital healthcare utilization is also important for hospital planning and resource allocation in children with bacterial meningitis. A child presenting with bacterial meningitis and a systemic complication or an associated focal infection may require additional resource planning initially to expedite care to enhance recovery and decrease hospital length of stay (LOS).

Our goal was to document the frequency of bacterial meningitis‐associated conditions (systemic complications and associated focal infections) in a large cohort of children with bacterial meningitis treated at tertiary care children's hospitals in the United States, and determine how the presence of such conditions impacted in‐hospital healthcare resource utilization.

Patients and Methods

Results

Bacterial Meningitis‐Associated Conditions

Overall, 574 (25%) of children with bacterial meningitis suffered a systemic complication or an associated focal infection. Figure 1 shows the types of associated condition stratified by age category. Older children had a higher frequency of associated focal infections while younger children had a higher frequency of systemic complications (P = 0.002, chi‐square test for trend). Figure 2 shows the distribution of specific conditions among children in each age category. The frequency of sepsis decreased with age (P < 0.001, chi‐square test) while the frequency of mastoiditis (P < 0.001, Fisher's exact test) and osteomyelitis (P = 0.005, Fisher's exact test) increased with age. There did not appear to be substantial variability in the proportion of patients with SIRS or sepsis across hospitals, suggesting that hospital‐level variability in coding for these conditions was likely minimal. The median proportion of patients with SIRS by hospital was 2.4% (IQR, 1.2‐4.8%) while the median proportion of patients with sepsis by hospital was 13.4% (IQR, 10.0‐16.9%).

Figure 1

Figure 2

Of the 151 children with an associated focal infection, only 3 (2%) of children had more than 1 infection (1 child had mastoiditis and endocarditis, 1 child had pneumonia and osteomyelitis, and 1 child had pneumonia and endocarditis). However, of the 479 children with systemic complications, 116 (24%) had more than 1 systemic disease (Table 1).

Table 1.Children With Systemic Complications

Types of Systemic Complications	Systemic Complications in All Bacterial Meningitis, n (%)	Systemic Complications in Meningococcal Meningitis, n (%)	Systemic Complications in Pneumococcal Meningitis, n (%)
NOTE: Percentages may not add to 100 due to rounding. Abbreviations: SIRS, systemic inflammatory response syndrome.
Sepsis only	209 (44)	16 (21)	69 (54)
Respiratory failure only	139 (29)	38 (49)	30 (24)
SIRS only	15 (3)	9 (12)	1 (1)
Sepsis and respiratory failure	52 (11)	4 (5)	18 (14)
SIRS and sepsis	27 (6)	2 (3)	4 (3)
SIRS and respiratory failure	9 (2)	5 (6)	0 (0)
SIRS and respiratory failure and sepsis	28 (6)	3 (4)	5 (4)
Total systemic complications	479	77	127

In sub‐group analyses, 269 children had meningococcal meningitis and 470 children had pneumococcal meningitis. Of the children with meningococcal meningitis, 31.2% had a meningitis‐associated condition: 26.4% had a systemic complication, 2.6% had a focal infection, and 2.2% had both conditions. The most common associated conditions in children with meningococcal meningitis were respiratory failure (18.6%; n = 50), sepsis (9.3%; n = 25), and SIRS (7.1%; n = 19). In children with pneumococcal meningitis, 32.3% had a meningitis‐associated complication: 24.7% had a systemic complication, 5.3% had a focal infection, and 2.3% had both conditions. The most common associated conditions in children with pneumococcal meningitis were sepsis (20.4%; n = 96), respiratory failure (11.3%; n = 53), and pneumonia (4.7%; n = 22); mastoiditis was present in 2.3% (n = 11) of children with pneumococcal meningitis. Respiratory failure was more common in meningococcal meningitis (18.6%) than in pneumococcal meningitis (11.3%; P = 0.006). In contrast, sepsis was less common in meningococcal meningitis (9.3%) than in pneumococcal meningitis (20.4%; P < 0.001).

Hospital Charges

Overall, the median charges per hospital ranged from $20,158 to $53,823. In‐hospital charges for children with bacterial meningitis with and without any identified associated conditions are presented in Table 2. In multivariate analyses, the presence of systemic conditions, associated focal infections, or both conditions was independently associated with significantly higher total in‐hospital charges (Table 2). When conditions were considered individually, bone and joint infections (213% increase; 95% CI, 113‐260%), endocarditis (108% increase; 95% CI, 23‐258%), and pneumonia (107% increase; 95% CI, 58‐171%) were associated with the highest increases in total hospital charges (Figure 3). In contrast, SIRS and mastoiditis were not associated with higher hospital charges (Figure 3).

Table 2.Comparison of Total Hospital Charges and Length of Stay by Complication Type

	Charges		LOS
	Median, $ (IQR)	Adjusted Increase,* % (95% CI)	Median, days (IQR)	Adjusted Increase,* % (95% CI)
Abbreviations: CI, confidence interval; IQR, interquartile range. * Adjusted for age category, race, sex, vancomycin receipt, and adjuvant corticosteroid therapy. P value <0.001 for all comparisons.
None (n = 1,745)	$27,110 (15,823‐48,307)	Reference**	9 (6‐14)	Reference
Systemic (n = 423)	$66,690 (39,546136,756)	136 (108269)	14 (923)	72 (5196)
Focal Infection (n = 95)	$58,016 (29,056125,813)	118 (77168)	13 (928)	78 (40126)
Both (n = 56)	$130,744 (62,397299,288)	351 (237503)	21.5 (1245)	211 (142303)

Figure 3

LOS

The median LOS was 9 days (IQR, 6‐15 days); 5% of children had a LOS >42 days. Table 2 summarizes difference in LOS by the presence and absence of systemic conditions and focal infections. In multivariate analyses, the presence of systemic conditions, associated focal infections, or both conditions was independently associated with a significantly longer LOS (Table 2). When conditions were considered individually, endocarditis (152% increase; 95% CI, 60‐300%) and pneumonia (136% increase; 95% CI, 85‐201%) were associated with the greatest adjusted increases in LOS (Figure 4); only mastoiditis was not associated with an increased LOS compared with those without complications.

Figure 4

Discussion

To our knowledge, this is the first study to examine bacterial meningitis‐associated conditions in children and their impact on in‐hospital resource utilization. We found that 25% of the cohort of children with bacterial meningitis suffered from at least one focal infection or systemic complication. This represents a significant invasive disease burden among children with bacterial meningitis who do not have underlying comorbid conditions. Younger children were more likely to have systemic complications when compared with older children, specifically due to a higher frequency of sepsis in children <1 year. Older children were more likely to have an associated focal infection, specifically due to an increase in mastoiditis and osteomyelitis in children >1 year. Only 2% of children had more than 1 focal infection, while 24% of children had more than 1 systemic complication.

Importantly, the presence of a systemic complication in a child with bacterial meningitis increased their in‐hospital adjusted charges by 136%. The presence of a focal infection increased in‐hospital adjusted charges by 118%. A child with both a systemic complication and a focal infection and had a 351% increase in in‐hospital adjusted charges.

The presence of systemic complications or associated focal infections was significantly associated with higher in‐hospital charges and longer hospital LOS. Most individual meningitis‐associated conditions included in this study were associated with higher in‐hospital charges with the exception of SIRS and mastoiditis. All individual meningitis‐associated conditions were associated with a longer LOS except mastoiditis. This finding is not surprising as the LOS for children with mastoiditis is typically shorter than for children with bacterial meningitis. Glikich et al.14 reported a mean LOS of approximately 8 days for children with mastoiditis. As meningitis in the context of mastoiditis is likely caused by direct extension of infection, patients with meningitis and mastoiditis likely required extended hospitalization to treat meningitis rather than mastoiditis. In contrast, patients with meningitis occurring in the context of metastatic dissemination of infection (eg, endocarditis, pneumonia) often have hemodynamic instability requiring prolonged intensive care support.

A study of children with sepsis found that increasing severity of illness was associated with greater hospital resource utilization.15 Our study shows that this may also be true in children with bacterial meningitis. We found that in children with bacterial meningitis, having systemic complications or an associated focal infection was associated with greater in‐hospital resource utilization. This finding may therefore indicate greater in‐hospital morbidity among children with a bacterial meningitis‐associated condition. Since mortality rates for bacterial meningitis are low in children, in‐hospital morbidity may be a better indicator of disease burden.

Our data show that, in contrast to adults, bacterial meningitis in children is not typically associated with other focal infections. Some focal complications such as mastoiditis and osteomyelitis disproportionately affect older children. These complications are typically accompanied by overt clinical manifestations. Therefore, we believe that the evaluation for the presence of concomitant focal infections can be guided by clinical examination findings and that routine radiologic evaluation for focal complications may not be necessary. Additionally, focal infections tend to occur in the absence of concomitant systemic complications. Of the 151 children with at least 1 associated focal infection, only 37% had a systemic complication. Bacterial meningitis may lie on a continuum of invasive disease depending on the virulence factors of the invading pathogen as well as specific host factors. Understanding the epidemiology of these associated conditions can enhance our understanding of the pathogenesis of bacterial meningitis in children. Understanding why some children suffer from septicemia rather than bacteremia may help in developing novel therapeutics.

There are several limitations to our study. First, since we identified focal infections and systemic complications using billing charges and ICD‐9 discharge diagnosis codes, it was impossible to determine when these conditions represented true complications of bacterial meningitis and when they represented the primary source of infection. Therefore, some of our primary outcomes may represent the cause of meningitis rather than a direct complication. We attempted to minimize such misclassification by limiting the cohort to those with a primary discharge diagnosis of bacterial meningitis though such misclassification is still possible.

Second, the use of ICD‐9 codes to accurately identify systemic complications and associated focal infections is a potential limitation. For example, respiratory failure, defined as the requirement of endotracheal intubation in our study, may not capture children receiving non‐invasive mechanical ventilation (eg, bilevel positive airway pressure). If use of noninvasive ventilation strategies did not depend exclusively on illness severity, our study would underestimate the frequency of respiratory failure. Furthermore, there may be inconsistencies among pediatric physicians in coding conditions such as SIRS and sepsis. Even in the clinical setting, a uniform definition of SIRS and sepsis is problematic due to physiologic differences between adults and children of varying age groups.16 An international panel of pediatricians proposed age‐specific definitions for sepsis and SIRS, while acknowledging the paucity of evidence to support some of their recommendations.16 None of the proposed definitions could be applied using administrative data. Limitations in the use of ICD‐9 discharge diagnosis codes to identify children with bacterial meningitis were discussed previously.1

Third, only free‐standing children's hospitals were included in the analysis. It is likely that many children with uncomplicated bacterial meningitis are treated at community hospitals or smaller academic centers. Our study may overestimate the rate of bacterial meningitis‐associated focal infections and systemic complications since participating hospitals serve as regional referral centers. To address the potential for such referral bias, we repeated the analysis while restricting the cohort to those children who had a lumbar puncture performed at the treating facility. No difference in frequency of associated conditions or in‐hospital resource utilization was found between children transferred and children not transferred. Finally, the PHIS database reports billed charge data rather than cost data. Billed data may overestimate the actual economic impact of bacterial meningitis‐associated complications since payers often reimburse at lesser rates. Resource utilization may also vary widely between hospitals and geographic locations as previously shown.15

In conclusion, bacterial meningitis remains an important cause of morbidity in children. Systemic complications such as sepsis and respiratory failure are common. Respiratory failure occurred more commonly among patients with meningococcal meningitis while sepsis occurred more commonly among patients with pneumococcal meningitis. While focal complications are uncommon, children >5 years of age are more likely than younger children to have concomitant mastoiditis or osteomyelitis. The presence of both systemic and focal complications is associated with substantially greater resource utilization than either complication alone.

Dr. Shah had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis. Study concept and design: Shah, Mongelluzzo; acquisition of data: Shah, Mohamad; analysis and interpretation of data: Mongelluzzo, Mohamad, Ten Have, Shah; drafting of the manuscript: Mongelluzzo; critical revision of the manuscript for important intellectual content: Mongelluzzo, Mohamad, Ten Have, Shah; statistical analysis: Shah, Mongelluzzo, Ten Have; obtained funding: Shah, Mongelluzzo; administrative, technical, or material support: Shah; study supervision: Shah.

Bacterial meningitis can be a devastating disease in children. Overall mortality in children in the United States is 4%1 while long‐term morbidity is present in up to 25%2 of surviving children. The introduction of Haemophilus influenzae type B vaccine, heptavalent pneumococcal conjugate vaccine, and the quadrivalent meningococcal conjugate vaccine has altered the epidemiology of bacterial meningitis.24 Currently, little is known about the epidemiology of systemic complications and associated focal infections that occur during episodes of bacterial meningitis in children and how the presence of such complications affects in‐hospital healthcare resource utilization.

In a randomized controlled trial, the administration of adjuvant corticosteroids was associated with lower mortality rates in adults with bacterial meningitis due to all causes, with the greatest reduction in those with pneumococcal meningitis.5 In a post hoc analysis of data from this trial, reductions in systemic complications, such as septic shock, pneumonia, and acute respiratory distress syndrome, rather than neurologic complications were thought to be the underlying reason for the decrease in mortality associated with pneumococcal meningitis among corticosteroid recipients.6 However, children with bacterial meningitis have an overall 4‐fold lower mortality rate than adults with bacterial meningitis. An even greater difference in mortality rates exists between children and adults with pneumococcal meningitis.1, 5 Children do not benefit from adjuvant corticosteroids as adults do.1, 5, 7 Therefore, the pathogenesis of bacterial meningitis may differ in children from adults and account for the difference in response to adjuvant corticosteroids. Understanding the epidemiology of systemic complications and associated focal infections can aid in the understanding of the pathogenesis of the disease in varying age groups of children.

Previous studies in children have documented the frequency of certain bacterial meningitis‐associated conditions such as respiratory failure, pneumonia, endocarditis, and mastoiditis. Researchers have used the presence of such conditions to predict either mortality or neurologic sequelae in children.810 These studies were small and only included a few types of complications associated with bacterial meningitis. In‐hospital healthcare resource utilization, which may be an important indicator of in‐hospital morbidity, was also not considered as an outcome. In‐hospital morbidity may represent aspects of disease burden not captured by mortality rates or markers for long‐term morbidity alone. In future vaccine efficacy trails or novel therapeutics evaluations, consideration of these associated conditions is important.

The quantification of the use of in‐hospital healthcare utilization is also important for hospital planning and resource allocation in children with bacterial meningitis. A child presenting with bacterial meningitis and a systemic complication or an associated focal infection may require additional resource planning initially to expedite care to enhance recovery and decrease hospital length of stay (LOS).

Our goal was to document the frequency of bacterial meningitis‐associated conditions (systemic complications and associated focal infections) in a large cohort of children with bacterial meningitis treated at tertiary care children's hospitals in the United States, and determine how the presence of such conditions impacted in‐hospital healthcare resource utilization.

Patients and Methods

Results

Bacterial Meningitis‐Associated Conditions

Overall, 574 (25%) of children with bacterial meningitis suffered a systemic complication or an associated focal infection. Figure 1 shows the types of associated condition stratified by age category. Older children had a higher frequency of associated focal infections while younger children had a higher frequency of systemic complications (P = 0.002, chi‐square test for trend). Figure 2 shows the distribution of specific conditions among children in each age category. The frequency of sepsis decreased with age (P < 0.001, chi‐square test) while the frequency of mastoiditis (P < 0.001, Fisher's exact test) and osteomyelitis (P = 0.005, Fisher's exact test) increased with age. There did not appear to be substantial variability in the proportion of patients with SIRS or sepsis across hospitals, suggesting that hospital‐level variability in coding for these conditions was likely minimal. The median proportion of patients with SIRS by hospital was 2.4% (IQR, 1.2‐4.8%) while the median proportion of patients with sepsis by hospital was 13.4% (IQR, 10.0‐16.9%).

Figure 1

Figure 2

Of the 151 children with an associated focal infection, only 3 (2%) of children had more than 1 infection (1 child had mastoiditis and endocarditis, 1 child had pneumonia and osteomyelitis, and 1 child had pneumonia and endocarditis). However, of the 479 children with systemic complications, 116 (24%) had more than 1 systemic disease (Table 1).

Table 1.Children With Systemic Complications

Types of Systemic Complications	Systemic Complications in All Bacterial Meningitis, n (%)	Systemic Complications in Meningococcal Meningitis, n (%)	Systemic Complications in Pneumococcal Meningitis, n (%)
NOTE: Percentages may not add to 100 due to rounding. Abbreviations: SIRS, systemic inflammatory response syndrome.
Sepsis only	209 (44)	16 (21)	69 (54)
Respiratory failure only	139 (29)	38 (49)	30 (24)
SIRS only	15 (3)	9 (12)	1 (1)
Sepsis and respiratory failure	52 (11)	4 (5)	18 (14)
SIRS and sepsis	27 (6)	2 (3)	4 (3)
SIRS and respiratory failure	9 (2)	5 (6)	0 (0)
SIRS and respiratory failure and sepsis	28 (6)	3 (4)	5 (4)
Total systemic complications	479	77	127

In sub‐group analyses, 269 children had meningococcal meningitis and 470 children had pneumococcal meningitis. Of the children with meningococcal meningitis, 31.2% had a meningitis‐associated condition: 26.4% had a systemic complication, 2.6% had a focal infection, and 2.2% had both conditions. The most common associated conditions in children with meningococcal meningitis were respiratory failure (18.6%; n = 50), sepsis (9.3%; n = 25), and SIRS (7.1%; n = 19). In children with pneumococcal meningitis, 32.3% had a meningitis‐associated complication: 24.7% had a systemic complication, 5.3% had a focal infection, and 2.3% had both conditions. The most common associated conditions in children with pneumococcal meningitis were sepsis (20.4%; n = 96), respiratory failure (11.3%; n = 53), and pneumonia (4.7%; n = 22); mastoiditis was present in 2.3% (n = 11) of children with pneumococcal meningitis. Respiratory failure was more common in meningococcal meningitis (18.6%) than in pneumococcal meningitis (11.3%; P = 0.006). In contrast, sepsis was less common in meningococcal meningitis (9.3%) than in pneumococcal meningitis (20.4%; P < 0.001).

Hospital Charges

Overall, the median charges per hospital ranged from $20,158 to $53,823. In‐hospital charges for children with bacterial meningitis with and without any identified associated conditions are presented in Table 2. In multivariate analyses, the presence of systemic conditions, associated focal infections, or both conditions was independently associated with significantly higher total in‐hospital charges (Table 2). When conditions were considered individually, bone and joint infections (213% increase; 95% CI, 113‐260%), endocarditis (108% increase; 95% CI, 23‐258%), and pneumonia (107% increase; 95% CI, 58‐171%) were associated with the highest increases in total hospital charges (Figure 3). In contrast, SIRS and mastoiditis were not associated with higher hospital charges (Figure 3).

Table 2.Comparison of Total Hospital Charges and Length of Stay by Complication Type

	Charges		LOS
	Median, $ (IQR)	Adjusted Increase,* % (95% CI)	Median, days (IQR)	Adjusted Increase,* % (95% CI)
Abbreviations: CI, confidence interval; IQR, interquartile range. * Adjusted for age category, race, sex, vancomycin receipt, and adjuvant corticosteroid therapy. P value <0.001 for all comparisons.
None (n = 1,745)	$27,110 (15,823‐48,307)	Reference**	9 (6‐14)	Reference
Systemic (n = 423)	$66,690 (39,546136,756)	136 (108269)	14 (923)	72 (5196)
Focal Infection (n = 95)	$58,016 (29,056125,813)	118 (77168)	13 (928)	78 (40126)
Both (n = 56)	$130,744 (62,397299,288)	351 (237503)	21.5 (1245)	211 (142303)

Figure 3

LOS

The median LOS was 9 days (IQR, 6‐15 days); 5% of children had a LOS >42 days. Table 2 summarizes difference in LOS by the presence and absence of systemic conditions and focal infections. In multivariate analyses, the presence of systemic conditions, associated focal infections, or both conditions was independently associated with a significantly longer LOS (Table 2). When conditions were considered individually, endocarditis (152% increase; 95% CI, 60‐300%) and pneumonia (136% increase; 95% CI, 85‐201%) were associated with the greatest adjusted increases in LOS (Figure 4); only mastoiditis was not associated with an increased LOS compared with those without complications.

Figure 4

Discussion

To our knowledge, this is the first study to examine bacterial meningitis‐associated conditions in children and their impact on in‐hospital resource utilization. We found that 25% of the cohort of children with bacterial meningitis suffered from at least one focal infection or systemic complication. This represents a significant invasive disease burden among children with bacterial meningitis who do not have underlying comorbid conditions. Younger children were more likely to have systemic complications when compared with older children, specifically due to a higher frequency of sepsis in children <1 year. Older children were more likely to have an associated focal infection, specifically due to an increase in mastoiditis and osteomyelitis in children >1 year. Only 2% of children had more than 1 focal infection, while 24% of children had more than 1 systemic complication.

Importantly, the presence of a systemic complication in a child with bacterial meningitis increased their in‐hospital adjusted charges by 136%. The presence of a focal infection increased in‐hospital adjusted charges by 118%. A child with both a systemic complication and a focal infection and had a 351% increase in in‐hospital adjusted charges.

The presence of systemic complications or associated focal infections was significantly associated with higher in‐hospital charges and longer hospital LOS. Most individual meningitis‐associated conditions included in this study were associated with higher in‐hospital charges with the exception of SIRS and mastoiditis. All individual meningitis‐associated conditions were associated with a longer LOS except mastoiditis. This finding is not surprising as the LOS for children with mastoiditis is typically shorter than for children with bacterial meningitis. Glikich et al.14 reported a mean LOS of approximately 8 days for children with mastoiditis. As meningitis in the context of mastoiditis is likely caused by direct extension of infection, patients with meningitis and mastoiditis likely required extended hospitalization to treat meningitis rather than mastoiditis. In contrast, patients with meningitis occurring in the context of metastatic dissemination of infection (eg, endocarditis, pneumonia) often have hemodynamic instability requiring prolonged intensive care support.

A study of children with sepsis found that increasing severity of illness was associated with greater hospital resource utilization.15 Our study shows that this may also be true in children with bacterial meningitis. We found that in children with bacterial meningitis, having systemic complications or an associated focal infection was associated with greater in‐hospital resource utilization. This finding may therefore indicate greater in‐hospital morbidity among children with a bacterial meningitis‐associated condition. Since mortality rates for bacterial meningitis are low in children, in‐hospital morbidity may be a better indicator of disease burden.

Our data show that, in contrast to adults, bacterial meningitis in children is not typically associated with other focal infections. Some focal complications such as mastoiditis and osteomyelitis disproportionately affect older children. These complications are typically accompanied by overt clinical manifestations. Therefore, we believe that the evaluation for the presence of concomitant focal infections can be guided by clinical examination findings and that routine radiologic evaluation for focal complications may not be necessary. Additionally, focal infections tend to occur in the absence of concomitant systemic complications. Of the 151 children with at least 1 associated focal infection, only 37% had a systemic complication. Bacterial meningitis may lie on a continuum of invasive disease depending on the virulence factors of the invading pathogen as well as specific host factors. Understanding the epidemiology of these associated conditions can enhance our understanding of the pathogenesis of bacterial meningitis in children. Understanding why some children suffer from septicemia rather than bacteremia may help in developing novel therapeutics.

There are several limitations to our study. First, since we identified focal infections and systemic complications using billing charges and ICD‐9 discharge diagnosis codes, it was impossible to determine when these conditions represented true complications of bacterial meningitis and when they represented the primary source of infection. Therefore, some of our primary outcomes may represent the cause of meningitis rather than a direct complication. We attempted to minimize such misclassification by limiting the cohort to those with a primary discharge diagnosis of bacterial meningitis though such misclassification is still possible.

Second, the use of ICD‐9 codes to accurately identify systemic complications and associated focal infections is a potential limitation. For example, respiratory failure, defined as the requirement of endotracheal intubation in our study, may not capture children receiving non‐invasive mechanical ventilation (eg, bilevel positive airway pressure). If use of noninvasive ventilation strategies did not depend exclusively on illness severity, our study would underestimate the frequency of respiratory failure. Furthermore, there may be inconsistencies among pediatric physicians in coding conditions such as SIRS and sepsis. Even in the clinical setting, a uniform definition of SIRS and sepsis is problematic due to physiologic differences between adults and children of varying age groups.16 An international panel of pediatricians proposed age‐specific definitions for sepsis and SIRS, while acknowledging the paucity of evidence to support some of their recommendations.16 None of the proposed definitions could be applied using administrative data. Limitations in the use of ICD‐9 discharge diagnosis codes to identify children with bacterial meningitis were discussed previously.1

Third, only free‐standing children's hospitals were included in the analysis. It is likely that many children with uncomplicated bacterial meningitis are treated at community hospitals or smaller academic centers. Our study may overestimate the rate of bacterial meningitis‐associated focal infections and systemic complications since participating hospitals serve as regional referral centers. To address the potential for such referral bias, we repeated the analysis while restricting the cohort to those children who had a lumbar puncture performed at the treating facility. No difference in frequency of associated conditions or in‐hospital resource utilization was found between children transferred and children not transferred. Finally, the PHIS database reports billed charge data rather than cost data. Billed data may overestimate the actual economic impact of bacterial meningitis‐associated complications since payers often reimburse at lesser rates. Resource utilization may also vary widely between hospitals and geographic locations as previously shown.15

In conclusion, bacterial meningitis remains an important cause of morbidity in children. Systemic complications such as sepsis and respiratory failure are common. Respiratory failure occurred more commonly among patients with meningococcal meningitis while sepsis occurred more commonly among patients with pneumococcal meningitis. While focal complications are uncommon, children >5 years of age are more likely than younger children to have concomitant mastoiditis or osteomyelitis. The presence of both systemic and focal complications is associated with substantially greater resource utilization than either complication alone.

Dr. Shah had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the analysis. Study concept and design: Shah, Mongelluzzo; acquisition of data: Shah, Mohamad; analysis and interpretation of data: Mongelluzzo, Mohamad, Ten Have, Shah; drafting of the manuscript: Mongelluzzo; critical revision of the manuscript for important intellectual content: Mongelluzzo, Mohamad, Ten Have, Shah; statistical analysis: Shah, Mongelluzzo, Ten Have; obtained funding: Shah, Mongelluzzo; administrative, technical, or material support: Shah; study supervision: Shah.

A 57‐year‐old man with Hepatitis B and C was scheduled for an outpatient computed tomography (CT)‐guided biopsy of the left lobe of the liver for grading and staging of his liver disease at an outside hospital. Complete metabolic profile (CMP), complete blood count (CBC) and international normalized ratio (INR) were performed and were normal. Following his liver biopsy, the patient became hypotensive and developed shock. He received fluid resuscitation. Clinical exam and an abdominal x‐ray failed to identify a hepatic or intraperitoneal source of bleeding to explain the cause of hypotension. A chest radiograph showed an enlarged cardiac silhouette with clear lung fields. An electrocardiogram (ECG) demonstrated (Figure 1) 1‐mm ST segment elevation in the anterior leads. An emergent echocardiogram showed a significant pericardial effusion with echocardiographic evidence of cardiac tamponade. The patient underwent emergent pericardiocentesis with removal of 400 mL of hemorrhagic fluid. The etiology of the hemorrhagic fluid was thought to be due to myocardial injury secondary to the liver biopsy. Subsequently, the histopathology of the liver biopsy reported (Figure 2) features suggestive of cirrhosis of the liver. Additional tissues from the biopsy included pericardium, myocardium and coronary artery. The patient was transferred to our institution for a cardiac catheterization and coronary angiogram to evaluate his coronary anatomy. The coronary angiogram was normal without any evidence of dye extravasation. A follow‐up echocardiogram and CT scan of the chest showed residual pericardial and pleural fluid. The remainder of the hospital stay was uneventful and he was discharged on day 7 in good condition. He was doing well a month later at the time of his follow‐up visit.0, 0

Figure 1

Figure 2

Discussion

Serious complications of liver biopsy occur in less than 1% of biopsied patients and include intraperitoneal1 or intrahepatic hemorrhage,2, 3 pneumothorax, hemothorax, hemobilia,1 and injury to the gall bladder, colon, kidney and lung. Other rare complications of percutaneous liver biopsy include biliary ascites, bile pleuritis, bile peritonitis, subcutaneous emphysema, pneumoperitoneum, subphrenic abscess, carcinoid crisis, anaphylaxis after biopsy of an echinococcal cyst, pancreatitis due to hemobilia, and breakage of the biopsy needle.1, 4, 5 Bleeding is more common in the elderly and in patients with cirrhosis and liver cancer.1, 6 The most common cause of shock in a patient after liver biopsy is hypovolemia from intrahepatic or intraperitoneal bleeding. The incidence of such complications may be higher than commonly appreciated as these complications are likely underreported.

Ultrasound (US)‐guided and CT‐guided techniques are 2 common methods for performing liver biopsy. US‐guided and CT‐guided liver biopsy each have their own benefits and limitations. The CT‐guided biopsy provides an excellent resolution of the liver architecture and is done from the epigastric region, but uses a static view. The US‐guided liver biopsy, on the other hand, is done from the mid‐axillary intercostal line and is performed with real‐time images.

Our patient's complication was presumably caused by accidental passage of the biopsy needle from the liver into the pericardium during respiratory movements. The coronary artery tissue was most likely not an epicardial vessel as supported by the normal angiogram without evidence of coronary artery perforation.

Our patient had an inadvertent biopsy of cardiac tissue during a liver biopsy resulting in a hemorrhagic pericardial effusion and tamponade. Cardiac tamponade as a complication of liver biopsy and intrathoracic percutaneous procedures has been described, although this has very rarely been reported.7, 8 Fortunately, the patient's coronary artery anatomy was normal and he did well without surgical intervention.

Unusual procedural complications are reported in the literature with the intention of increasing awareness and improving patient safety. Percutaneous liver biopsy is performed to stage and grade liver disease for assessment and treatment. However, as with any invasive procedure, there are potential risks and complications. In patients who present with hypotension and shock following liver biopsy, myocardial injury with cardiac tamponade should be considered in the differential diagnosis.

A 57‐year‐old man with Hepatitis B and C was scheduled for an outpatient computed tomography (CT)‐guided biopsy of the left lobe of the liver for grading and staging of his liver disease at an outside hospital. Complete metabolic profile (CMP), complete blood count (CBC) and international normalized ratio (INR) were performed and were normal. Following his liver biopsy, the patient became hypotensive and developed shock. He received fluid resuscitation. Clinical exam and an abdominal x‐ray failed to identify a hepatic or intraperitoneal source of bleeding to explain the cause of hypotension. A chest radiograph showed an enlarged cardiac silhouette with clear lung fields. An electrocardiogram (ECG) demonstrated (Figure 1) 1‐mm ST segment elevation in the anterior leads. An emergent echocardiogram showed a significant pericardial effusion with echocardiographic evidence of cardiac tamponade. The patient underwent emergent pericardiocentesis with removal of 400 mL of hemorrhagic fluid. The etiology of the hemorrhagic fluid was thought to be due to myocardial injury secondary to the liver biopsy. Subsequently, the histopathology of the liver biopsy reported (Figure 2) features suggestive of cirrhosis of the liver. Additional tissues from the biopsy included pericardium, myocardium and coronary artery. The patient was transferred to our institution for a cardiac catheterization and coronary angiogram to evaluate his coronary anatomy. The coronary angiogram was normal without any evidence of dye extravasation. A follow‐up echocardiogram and CT scan of the chest showed residual pericardial and pleural fluid. The remainder of the hospital stay was uneventful and he was discharged on day 7 in good condition. He was doing well a month later at the time of his follow‐up visit.0, 0

Figure 1

Figure 2

Discussion

Serious complications of liver biopsy occur in less than 1% of biopsied patients and include intraperitoneal1 or intrahepatic hemorrhage,2, 3 pneumothorax, hemothorax, hemobilia,1 and injury to the gall bladder, colon, kidney and lung. Other rare complications of percutaneous liver biopsy include biliary ascites, bile pleuritis, bile peritonitis, subcutaneous emphysema, pneumoperitoneum, subphrenic abscess, carcinoid crisis, anaphylaxis after biopsy of an echinococcal cyst, pancreatitis due to hemobilia, and breakage of the biopsy needle.1, 4, 5 Bleeding is more common in the elderly and in patients with cirrhosis and liver cancer.1, 6 The most common cause of shock in a patient after liver biopsy is hypovolemia from intrahepatic or intraperitoneal bleeding. The incidence of such complications may be higher than commonly appreciated as these complications are likely underreported.

Ultrasound (US)‐guided and CT‐guided techniques are 2 common methods for performing liver biopsy. US‐guided and CT‐guided liver biopsy each have their own benefits and limitations. The CT‐guided biopsy provides an excellent resolution of the liver architecture and is done from the epigastric region, but uses a static view. The US‐guided liver biopsy, on the other hand, is done from the mid‐axillary intercostal line and is performed with real‐time images.

Our patient's complication was presumably caused by accidental passage of the biopsy needle from the liver into the pericardium during respiratory movements. The coronary artery tissue was most likely not an epicardial vessel as supported by the normal angiogram without evidence of coronary artery perforation.

Our patient had an inadvertent biopsy of cardiac tissue during a liver biopsy resulting in a hemorrhagic pericardial effusion and tamponade. Cardiac tamponade as a complication of liver biopsy and intrathoracic percutaneous procedures has been described, although this has very rarely been reported.7, 8 Fortunately, the patient's coronary artery anatomy was normal and he did well without surgical intervention.

Unusual procedural complications are reported in the literature with the intention of increasing awareness and improving patient safety. Percutaneous liver biopsy is performed to stage and grade liver disease for assessment and treatment. However, as with any invasive procedure, there are potential risks and complications. In patients who present with hypotension and shock following liver biopsy, myocardial injury with cardiac tamponade should be considered in the differential diagnosis.

A 57‐year‐old man with Hepatitis B and C was scheduled for an outpatient computed tomography (CT)‐guided biopsy of the left lobe of the liver for grading and staging of his liver disease at an outside hospital. Complete metabolic profile (CMP), complete blood count (CBC) and international normalized ratio (INR) were performed and were normal. Following his liver biopsy, the patient became hypotensive and developed shock. He received fluid resuscitation. Clinical exam and an abdominal x‐ray failed to identify a hepatic or intraperitoneal source of bleeding to explain the cause of hypotension. A chest radiograph showed an enlarged cardiac silhouette with clear lung fields. An electrocardiogram (ECG) demonstrated (Figure 1) 1‐mm ST segment elevation in the anterior leads. An emergent echocardiogram showed a significant pericardial effusion with echocardiographic evidence of cardiac tamponade. The patient underwent emergent pericardiocentesis with removal of 400 mL of hemorrhagic fluid. The etiology of the hemorrhagic fluid was thought to be due to myocardial injury secondary to the liver biopsy. Subsequently, the histopathology of the liver biopsy reported (Figure 2) features suggestive of cirrhosis of the liver. Additional tissues from the biopsy included pericardium, myocardium and coronary artery. The patient was transferred to our institution for a cardiac catheterization and coronary angiogram to evaluate his coronary anatomy. The coronary angiogram was normal without any evidence of dye extravasation. A follow‐up echocardiogram and CT scan of the chest showed residual pericardial and pleural fluid. The remainder of the hospital stay was uneventful and he was discharged on day 7 in good condition. He was doing well a month later at the time of his follow‐up visit.0, 0

Figure 1

Figure 2

Discussion

Serious complications of liver biopsy occur in less than 1% of biopsied patients and include intraperitoneal1 or intrahepatic hemorrhage,2, 3 pneumothorax, hemothorax, hemobilia,1 and injury to the gall bladder, colon, kidney and lung. Other rare complications of percutaneous liver biopsy include biliary ascites, bile pleuritis, bile peritonitis, subcutaneous emphysema, pneumoperitoneum, subphrenic abscess, carcinoid crisis, anaphylaxis after biopsy of an echinococcal cyst, pancreatitis due to hemobilia, and breakage of the biopsy needle.1, 4, 5 Bleeding is more common in the elderly and in patients with cirrhosis and liver cancer.1, 6 The most common cause of shock in a patient after liver biopsy is hypovolemia from intrahepatic or intraperitoneal bleeding. The incidence of such complications may be higher than commonly appreciated as these complications are likely underreported.

Ultrasound (US)‐guided and CT‐guided techniques are 2 common methods for performing liver biopsy. US‐guided and CT‐guided liver biopsy each have their own benefits and limitations. The CT‐guided biopsy provides an excellent resolution of the liver architecture and is done from the epigastric region, but uses a static view. The US‐guided liver biopsy, on the other hand, is done from the mid‐axillary intercostal line and is performed with real‐time images.

Our patient's complication was presumably caused by accidental passage of the biopsy needle from the liver into the pericardium during respiratory movements. The coronary artery tissue was most likely not an epicardial vessel as supported by the normal angiogram without evidence of coronary artery perforation.

Our patient had an inadvertent biopsy of cardiac tissue during a liver biopsy resulting in a hemorrhagic pericardial effusion and tamponade. Cardiac tamponade as a complication of liver biopsy and intrathoracic percutaneous procedures has been described, although this has very rarely been reported.7, 8 Fortunately, the patient's coronary artery anatomy was normal and he did well without surgical intervention.

Unusual procedural complications are reported in the literature with the intention of increasing awareness and improving patient safety. Percutaneous liver biopsy is performed to stage and grade liver disease for assessment and treatment. However, as with any invasive procedure, there are potential risks and complications. In patients who present with hypotension and shock following liver biopsy, myocardial injury with cardiac tamponade should be considered in the differential diagnosis.

Classically, infections have been categorized as either community‐acquired (CAI) or nosocomial in origin. Until recently, this scheme was thought adequate to capture the differences in the microbiology and outcomes in the corresponding scenarios. However, recent evidence suggests that this distinction may no longer be valid. For example, with the spread and diffusion of healthcare delivery beyond the confines of the hospital along with the increasing use of broad spectrum antibiotics both in and out of the hospital, pathogens such as methicillin‐resistant Staphylococcus aureus (MRSA) and Pseudomonas aeruginosa (PA), traditionally thought to be confined to the hospital, are now seen in patients presenting from the community to the emergency department (ED).1, 2 Reflecting this shift in epidemiology, some national guidelines now recognize healthcare‐associated infection (HCAI) as a distinct entity.3 The concept of HCAI allows the clinician to identify patients who, despite suffering a community onset infection, still may be at risk for a resistant bacterial pathogen. Recent studies in both bloodstream infection and pneumonia have clearly demonstrated that those with HCAI have distinct microbiology and outcomes relative to those with pure CAI.47

Most work focusing on establishing HCAI has not addressed skin and soft tissue infections. These infections, although not often fatal, account for an increasing number of admissions to the hospital.8, 9 In addition, they may be associated with substantial morbidity and cost.8 Given that many pathogens such as S. aureus, which may be resistant to typical antimicrobials used in the ED, are also major culprits in complicated skin and skin structure infections (cSSSI), the HCAI paradigm may apply in cSSSI. Furthermore, because of these patterns of increased resistance, HCA‐cSSSI patients, similar to other HCAI groups, may be at an increased risk of being treated with initially inappropriate antibiotic therapy.7, 10

Since in the setting of other types of infection inappropriate empiric treatment has been shown to be associated with increased mortality and costs,7, 1015 and since indirect evidence suggests a similar impact on healthcare utilization among cSSSI patients,8 we hypothesized that among a cohort of patients hospitalized with a cSSSI, the initial empiric choice of therapy is independently associated with hospital length of stay (LOS). We performed a retrospective cohort study to address this question.

Methods

Results

Of the 717 patients with culture‐positive cSSSI admitted during the study period, 527 (73.5%) were classified as HCAI. The most common reason for classification as an HCAI was recent hospitalization. Among those with an HCA‐cSSSI, 405 (76.9%) received appropriate empiric treatment, with nearly one‐quarter receiving inappropriate initial coverage. Those receiving inappropriate antibiotic were more likely to be African American, and had a higher likelihood of having end‐stage renal disease (ESRD) than those with appropriate coverage (Table 1). While those patients treated appropriately had higher rates of both cellulitis and abscess as the presenting infection, a substantially higher proportion of those receiving inappropriate initial treatment had a decubitus ulcer (29.5% vs. 10.9%, P <0.001), a device‐associated infection (42.6% vs. 28.6%, P = 0.004), and had evidence of bacteremia (68.9% vs. 57.8%, P = 0.028) than those receiving appropriate empiric coverage (Table 2).

The pathogens recovered from the appropriately and inappropriately treated groups are listed in Figure 1. While S. aureus overall was more common among those treated appropriately, the frequency of MRSA did not differ between the groups. Both E. faecalis and E. faecium were recovered more frequently in the inappropriate group, resulting in a similar pattern among the vancomycin‐resistant enterococcal species. Likewise, P. aeruginosa, P. mirabilis, and A. baumannii were all more frequently seen in the group treated inappropriately than in the group getting appropriate empiric coverage. A mixed infection was also more likely to be present among those not exposed (16.5%) than among those exposed (7.5%) to appropriate early therapy (P = 0.001) (Figure 1).

Figure 1

In terms of processes of care and outcomes (Table 3), commensurate with the higher prevalence of abscess in the appropriately treated group, the rate of I&D was significantly higher in this cohort (36.8%) than in the inappropriately treated (23.0%) group (P = 0.005). Need for initial ICU care did not differ as a function of appropriateness of therapy (P = 0.635).

The unadjusted mortality rate was low overall and did not vary based on initial treatment (Table 3). In a generalized linear model with the log‐transformed LOS as the dependent variable, adjusting for multiple potential confounders, initial inappropriate antibiotic therapy had an attributable incremental increase in the hospital LOS of 1.8 days (95% CI, 1.42.3) (Table 4).

Because bacteremia is known to be an effect modifier of the relationship between the empiric choice of antibiotic and infection outcomes, we further explored its role in the HCAI cSSSI on the outcomes of interest (Table 5). Similar to the effect detected in the overall cohort, treatment with inappropriate therapy was associated with an increase in the hospital LOS, but not hospital mortality in those with bacteremia, though this phenomenon was observed only among patients with secondary bacteremia, and not among those without (Table 5).

Discussion

This retrospective analysis provides evidence that inappropriate empiric antibiotic therapy for HCA‐cSSSI independently prolongs hospital LOS. The impact of inappropriate initial treatment on LOS is independent of many important confounders. In addition, we observed that this effect, while present among patients with secondary bacteremia, is absent among those without a blood stream infection.

To the best of our knowledge, ours is the first cohort study to examine the outcomes associated with inappropriate treatment of a HCAI cSSSI within the context of available microbiology data. Edelsberg et al.8 examined clinical and economic outcomes associated with the failure of the initial treatment of cSSSI. While not specifically focusing on HCAI patients, these authors noted an overall 23% initial therapy failure rate. Among those patients who failed initial therapy, the risk of hospital death was nearly 3‐fold higher (adjusted odds ratio [OR], 2.91; 95% CI, 2.343.62), and they incurred the mean of 5.4 additional hospital days, compared to patients treated successfully with the initial regimen.8 Our study confirms Edelsberg et al.'s8 observation of prolonged hospital LOS in association with treatment failure, and builds upon it by defining the actual LOS increment attributable to inappropriate empiric therapy. It is worth noting that the study by Edelsberg et al.,8 however, lacked explicit definition of the HCAI population and microbiology data, and used treatment failure as a surrogate marker for inappropriate treatment. It is likely these differences between our two studies in the underlying population and exposure definitions that account for the differences in the mortality data between that study and ours.

It is not fundamentally surprising that early exposure to inappropriate empiric therapy alters healthcare resource utilization outcomes for the worse. Others have demonstrated that infection with a resistant organism results in prolongation of hospital LOS and costs. For example, in a large cohort of over 600 surgical hospitalizations requiring treatment for a gram‐negative infection, antibiotic resistance was an independent predictor of increased LOS and costs.15 These authors quantified the incremental burden of early gram‐negative resistance at over $11,000 in hospital costs.15 Unfortunately, the treatment differences for resistant and sensitive organisms were not examined.15 Similarly, Shorr et al. examined risk factors for prolonged hospital LOS and increased costs in a cohort of 291 patients with MRSA sterile site infection.17 Because in this study 23% of the patients received inappropriate empiric therapy, the authors were able to examine the impact of this exposure on utilization outcomes.17 In an adjusted analysis, inappropriate initial treatment was associated with an incremental increase in the LOS of 2.5 days, corresponding to the unadjusted cost differential of nearly $6,000.17 Although focusing on a different population, our results are consistent with these previous observations that antibiotic resistance and early inappropriate therapy affect hospital utilization parameters, in our case by adding nearly 2 days to the hospital LOS.

Our study has a number of limitations. First, as a retrospective cohort study it is prone to various forms of bias, most notably selection bias. To minimize the possibility of such, we established a priori case definitions and enrolled consecutive patients over a specific period of time. Second, as in any observational study, confounding is an issue. We dealt with this statistically by constructing a multivariable regression model; however, the possibility of residual confounding remains. Third, because some of the wound and ulcer cultures likely were obtained with a swab and thus represented colonization, rather than infection, we may have over‐estimated the rate of inappropriate therapy, and this needs to be followed up in future prospective studies. Similarly, we may have over‐estimated the likelihood of inappropriate therapy among polymicrobial and mixed infections as well, given that, for example, a gram‐negative organism may carry a different clinical significance when cultured from blood (infection) than when it is detected in a decubitus ulcer (potential colonization). Fourth, because we limited our cohort to patients without deep‐seated infections, such as necrotizing fasciitis, other procedures were not collected. This omission may have led to either over‐estimation or under‐estimation of the impact of inappropriate therapy on the outcomes of interest.

The fact that our cohort represents a single large urban academic tertiary care medical center may limit the generalizability of our results only to centers that share similar characteristics. Finally, similar to most other studies of this type, ours lacks data on posthospitalization outcomes and for this reason limits itself to hospital outcomes only.

In summary, we have shown that, similar to other populations with HCAI, a substantial proportion (nearly 1/4) of cSSSI patients with HCAI receive inappropriate empiric therapy for their infection, and this early exposure, though not affecting hospital mortality, is associated with a significant prolongation of the hospitalization by as much as 2 days. Studies are needed to refine decision rules for risk‐stratifying patients with cSSSI HCAI in order to determine the probability of infection with a resistant organism. In turn, such instruments at the bedside may assure improved utilization of appropriately targeted empiric therapy that will both optimize individual patient outcomes and reduce the risk of emergence of antimicrobial resistance.

Classically, infections have been categorized as either community‐acquired (CAI) or nosocomial in origin. Until recently, this scheme was thought adequate to capture the differences in the microbiology and outcomes in the corresponding scenarios. However, recent evidence suggests that this distinction may no longer be valid. For example, with the spread and diffusion of healthcare delivery beyond the confines of the hospital along with the increasing use of broad spectrum antibiotics both in and out of the hospital, pathogens such as methicillin‐resistant Staphylococcus aureus (MRSA) and Pseudomonas aeruginosa (PA), traditionally thought to be confined to the hospital, are now seen in patients presenting from the community to the emergency department (ED).1, 2 Reflecting this shift in epidemiology, some national guidelines now recognize healthcare‐associated infection (HCAI) as a distinct entity.3 The concept of HCAI allows the clinician to identify patients who, despite suffering a community onset infection, still may be at risk for a resistant bacterial pathogen. Recent studies in both bloodstream infection and pneumonia have clearly demonstrated that those with HCAI have distinct microbiology and outcomes relative to those with pure CAI.47

Most work focusing on establishing HCAI has not addressed skin and soft tissue infections. These infections, although not often fatal, account for an increasing number of admissions to the hospital.8, 9 In addition, they may be associated with substantial morbidity and cost.8 Given that many pathogens such as S. aureus, which may be resistant to typical antimicrobials used in the ED, are also major culprits in complicated skin and skin structure infections (cSSSI), the HCAI paradigm may apply in cSSSI. Furthermore, because of these patterns of increased resistance, HCA‐cSSSI patients, similar to other HCAI groups, may be at an increased risk of being treated with initially inappropriate antibiotic therapy.7, 10

Since in the setting of other types of infection inappropriate empiric treatment has been shown to be associated with increased mortality and costs,7, 1015 and since indirect evidence suggests a similar impact on healthcare utilization among cSSSI patients,8 we hypothesized that among a cohort of patients hospitalized with a cSSSI, the initial empiric choice of therapy is independently associated with hospital length of stay (LOS). We performed a retrospective cohort study to address this question.

Methods

Results

Of the 717 patients with culture‐positive cSSSI admitted during the study period, 527 (73.5%) were classified as HCAI. The most common reason for classification as an HCAI was recent hospitalization. Among those with an HCA‐cSSSI, 405 (76.9%) received appropriate empiric treatment, with nearly one‐quarter receiving inappropriate initial coverage. Those receiving inappropriate antibiotic were more likely to be African American, and had a higher likelihood of having end‐stage renal disease (ESRD) than those with appropriate coverage (Table 1). While those patients treated appropriately had higher rates of both cellulitis and abscess as the presenting infection, a substantially higher proportion of those receiving inappropriate initial treatment had a decubitus ulcer (29.5% vs. 10.9%, P <0.001), a device‐associated infection (42.6% vs. 28.6%, P = 0.004), and had evidence of bacteremia (68.9% vs. 57.8%, P = 0.028) than those receiving appropriate empiric coverage (Table 2).

The pathogens recovered from the appropriately and inappropriately treated groups are listed in Figure 1. While S. aureus overall was more common among those treated appropriately, the frequency of MRSA did not differ between the groups. Both E. faecalis and E. faecium were recovered more frequently in the inappropriate group, resulting in a similar pattern among the vancomycin‐resistant enterococcal species. Likewise, P. aeruginosa, P. mirabilis, and A. baumannii were all more frequently seen in the group treated inappropriately than in the group getting appropriate empiric coverage. A mixed infection was also more likely to be present among those not exposed (16.5%) than among those exposed (7.5%) to appropriate early therapy (P = 0.001) (Figure 1).

Figure 1

In terms of processes of care and outcomes (Table 3), commensurate with the higher prevalence of abscess in the appropriately treated group, the rate of I&D was significantly higher in this cohort (36.8%) than in the inappropriately treated (23.0%) group (P = 0.005). Need for initial ICU care did not differ as a function of appropriateness of therapy (P = 0.635).

The unadjusted mortality rate was low overall and did not vary based on initial treatment (Table 3). In a generalized linear model with the log‐transformed LOS as the dependent variable, adjusting for multiple potential confounders, initial inappropriate antibiotic therapy had an attributable incremental increase in the hospital LOS of 1.8 days (95% CI, 1.42.3) (Table 4).

Because bacteremia is known to be an effect modifier of the relationship between the empiric choice of antibiotic and infection outcomes, we further explored its role in the HCAI cSSSI on the outcomes of interest (Table 5). Similar to the effect detected in the overall cohort, treatment with inappropriate therapy was associated with an increase in the hospital LOS, but not hospital mortality in those with bacteremia, though this phenomenon was observed only among patients with secondary bacteremia, and not among those without (Table 5).

Discussion

This retrospective analysis provides evidence that inappropriate empiric antibiotic therapy for HCA‐cSSSI independently prolongs hospital LOS. The impact of inappropriate initial treatment on LOS is independent of many important confounders. In addition, we observed that this effect, while present among patients with secondary bacteremia, is absent among those without a blood stream infection.

To the best of our knowledge, ours is the first cohort study to examine the outcomes associated with inappropriate treatment of a HCAI cSSSI within the context of available microbiology data. Edelsberg et al.8 examined clinical and economic outcomes associated with the failure of the initial treatment of cSSSI. While not specifically focusing on HCAI patients, these authors noted an overall 23% initial therapy failure rate. Among those patients who failed initial therapy, the risk of hospital death was nearly 3‐fold higher (adjusted odds ratio [OR], 2.91; 95% CI, 2.343.62), and they incurred the mean of 5.4 additional hospital days, compared to patients treated successfully with the initial regimen.8 Our study confirms Edelsberg et al.'s8 observation of prolonged hospital LOS in association with treatment failure, and builds upon it by defining the actual LOS increment attributable to inappropriate empiric therapy. It is worth noting that the study by Edelsberg et al.,8 however, lacked explicit definition of the HCAI population and microbiology data, and used treatment failure as a surrogate marker for inappropriate treatment. It is likely these differences between our two studies in the underlying population and exposure definitions that account for the differences in the mortality data between that study and ours.

It is not fundamentally surprising that early exposure to inappropriate empiric therapy alters healthcare resource utilization outcomes for the worse. Others have demonstrated that infection with a resistant organism results in prolongation of hospital LOS and costs. For example, in a large cohort of over 600 surgical hospitalizations requiring treatment for a gram‐negative infection, antibiotic resistance was an independent predictor of increased LOS and costs.15 These authors quantified the incremental burden of early gram‐negative resistance at over $11,000 in hospital costs.15 Unfortunately, the treatment differences for resistant and sensitive organisms were not examined.15 Similarly, Shorr et al. examined risk factors for prolonged hospital LOS and increased costs in a cohort of 291 patients with MRSA sterile site infection.17 Because in this study 23% of the patients received inappropriate empiric therapy, the authors were able to examine the impact of this exposure on utilization outcomes.17 In an adjusted analysis, inappropriate initial treatment was associated with an incremental increase in the LOS of 2.5 days, corresponding to the unadjusted cost differential of nearly $6,000.17 Although focusing on a different population, our results are consistent with these previous observations that antibiotic resistance and early inappropriate therapy affect hospital utilization parameters, in our case by adding nearly 2 days to the hospital LOS.

Our study has a number of limitations. First, as a retrospective cohort study it is prone to various forms of bias, most notably selection bias. To minimize the possibility of such, we established a priori case definitions and enrolled consecutive patients over a specific period of time. Second, as in any observational study, confounding is an issue. We dealt with this statistically by constructing a multivariable regression model; however, the possibility of residual confounding remains. Third, because some of the wound and ulcer cultures likely were obtained with a swab and thus represented colonization, rather than infection, we may have over‐estimated the rate of inappropriate therapy, and this needs to be followed up in future prospective studies. Similarly, we may have over‐estimated the likelihood of inappropriate therapy among polymicrobial and mixed infections as well, given that, for example, a gram‐negative organism may carry a different clinical significance when cultured from blood (infection) than when it is detected in a decubitus ulcer (potential colonization). Fourth, because we limited our cohort to patients without deep‐seated infections, such as necrotizing fasciitis, other procedures were not collected. This omission may have led to either over‐estimation or under‐estimation of the impact of inappropriate therapy on the outcomes of interest.

The fact that our cohort represents a single large urban academic tertiary care medical center may limit the generalizability of our results only to centers that share similar characteristics. Finally, similar to most other studies of this type, ours lacks data on posthospitalization outcomes and for this reason limits itself to hospital outcomes only.

In summary, we have shown that, similar to other populations with HCAI, a substantial proportion (nearly 1/4) of cSSSI patients with HCAI receive inappropriate empiric therapy for their infection, and this early exposure, though not affecting hospital mortality, is associated with a significant prolongation of the hospitalization by as much as 2 days. Studies are needed to refine decision rules for risk‐stratifying patients with cSSSI HCAI in order to determine the probability of infection with a resistant organism. In turn, such instruments at the bedside may assure improved utilization of appropriately targeted empiric therapy that will both optimize individual patient outcomes and reduce the risk of emergence of antimicrobial resistance.

Classically, infections have been categorized as either community‐acquired (CAI) or nosocomial in origin. Until recently, this scheme was thought adequate to capture the differences in the microbiology and outcomes in the corresponding scenarios. However, recent evidence suggests that this distinction may no longer be valid. For example, with the spread and diffusion of healthcare delivery beyond the confines of the hospital along with the increasing use of broad spectrum antibiotics both in and out of the hospital, pathogens such as methicillin‐resistant Staphylococcus aureus (MRSA) and Pseudomonas aeruginosa (PA), traditionally thought to be confined to the hospital, are now seen in patients presenting from the community to the emergency department (ED).1, 2 Reflecting this shift in epidemiology, some national guidelines now recognize healthcare‐associated infection (HCAI) as a distinct entity.3 The concept of HCAI allows the clinician to identify patients who, despite suffering a community onset infection, still may be at risk for a resistant bacterial pathogen. Recent studies in both bloodstream infection and pneumonia have clearly demonstrated that those with HCAI have distinct microbiology and outcomes relative to those with pure CAI.47

Most work focusing on establishing HCAI has not addressed skin and soft tissue infections. These infections, although not often fatal, account for an increasing number of admissions to the hospital.8, 9 In addition, they may be associated with substantial morbidity and cost.8 Given that many pathogens such as S. aureus, which may be resistant to typical antimicrobials used in the ED, are also major culprits in complicated skin and skin structure infections (cSSSI), the HCAI paradigm may apply in cSSSI. Furthermore, because of these patterns of increased resistance, HCA‐cSSSI patients, similar to other HCAI groups, may be at an increased risk of being treated with initially inappropriate antibiotic therapy.7, 10

Since in the setting of other types of infection inappropriate empiric treatment has been shown to be associated with increased mortality and costs,7, 1015 and since indirect evidence suggests a similar impact on healthcare utilization among cSSSI patients,8 we hypothesized that among a cohort of patients hospitalized with a cSSSI, the initial empiric choice of therapy is independently associated with hospital length of stay (LOS). We performed a retrospective cohort study to address this question.

Methods

Results

Of the 717 patients with culture‐positive cSSSI admitted during the study period, 527 (73.5%) were classified as HCAI. The most common reason for classification as an HCAI was recent hospitalization. Among those with an HCA‐cSSSI, 405 (76.9%) received appropriate empiric treatment, with nearly one‐quarter receiving inappropriate initial coverage. Those receiving inappropriate antibiotic were more likely to be African American, and had a higher likelihood of having end‐stage renal disease (ESRD) than those with appropriate coverage (Table 1). While those patients treated appropriately had higher rates of both cellulitis and abscess as the presenting infection, a substantially higher proportion of those receiving inappropriate initial treatment had a decubitus ulcer (29.5% vs. 10.9%, P <0.001), a device‐associated infection (42.6% vs. 28.6%, P = 0.004), and had evidence of bacteremia (68.9% vs. 57.8%, P = 0.028) than those receiving appropriate empiric coverage (Table 2).

The pathogens recovered from the appropriately and inappropriately treated groups are listed in Figure 1. While S. aureus overall was more common among those treated appropriately, the frequency of MRSA did not differ between the groups. Both E. faecalis and E. faecium were recovered more frequently in the inappropriate group, resulting in a similar pattern among the vancomycin‐resistant enterococcal species. Likewise, P. aeruginosa, P. mirabilis, and A. baumannii were all more frequently seen in the group treated inappropriately than in the group getting appropriate empiric coverage. A mixed infection was also more likely to be present among those not exposed (16.5%) than among those exposed (7.5%) to appropriate early therapy (P = 0.001) (Figure 1).

Figure 1

In terms of processes of care and outcomes (Table 3), commensurate with the higher prevalence of abscess in the appropriately treated group, the rate of I&D was significantly higher in this cohort (36.8%) than in the inappropriately treated (23.0%) group (P = 0.005). Need for initial ICU care did not differ as a function of appropriateness of therapy (P = 0.635).

The unadjusted mortality rate was low overall and did not vary based on initial treatment (Table 3). In a generalized linear model with the log‐transformed LOS as the dependent variable, adjusting for multiple potential confounders, initial inappropriate antibiotic therapy had an attributable incremental increase in the hospital LOS of 1.8 days (95% CI, 1.42.3) (Table 4).

Because bacteremia is known to be an effect modifier of the relationship between the empiric choice of antibiotic and infection outcomes, we further explored its role in the HCAI cSSSI on the outcomes of interest (Table 5). Similar to the effect detected in the overall cohort, treatment with inappropriate therapy was associated with an increase in the hospital LOS, but not hospital mortality in those with bacteremia, though this phenomenon was observed only among patients with secondary bacteremia, and not among those without (Table 5).

Discussion

This retrospective analysis provides evidence that inappropriate empiric antibiotic therapy for HCA‐cSSSI independently prolongs hospital LOS. The impact of inappropriate initial treatment on LOS is independent of many important confounders. In addition, we observed that this effect, while present among patients with secondary bacteremia, is absent among those without a blood stream infection.

To the best of our knowledge, ours is the first cohort study to examine the outcomes associated with inappropriate treatment of a HCAI cSSSI within the context of available microbiology data. Edelsberg et al.8 examined clinical and economic outcomes associated with the failure of the initial treatment of cSSSI. While not specifically focusing on HCAI patients, these authors noted an overall 23% initial therapy failure rate. Among those patients who failed initial therapy, the risk of hospital death was nearly 3‐fold higher (adjusted odds ratio [OR], 2.91; 95% CI, 2.343.62), and they incurred the mean of 5.4 additional hospital days, compared to patients treated successfully with the initial regimen.8 Our study confirms Edelsberg et al.'s8 observation of prolonged hospital LOS in association with treatment failure, and builds upon it by defining the actual LOS increment attributable to inappropriate empiric therapy. It is worth noting that the study by Edelsberg et al.,8 however, lacked explicit definition of the HCAI population and microbiology data, and used treatment failure as a surrogate marker for inappropriate treatment. It is likely these differences between our two studies in the underlying population and exposure definitions that account for the differences in the mortality data between that study and ours.

It is not fundamentally surprising that early exposure to inappropriate empiric therapy alters healthcare resource utilization outcomes for the worse. Others have demonstrated that infection with a resistant organism results in prolongation of hospital LOS and costs. For example, in a large cohort of over 600 surgical hospitalizations requiring treatment for a gram‐negative infection, antibiotic resistance was an independent predictor of increased LOS and costs.15 These authors quantified the incremental burden of early gram‐negative resistance at over $11,000 in hospital costs.15 Unfortunately, the treatment differences for resistant and sensitive organisms were not examined.15 Similarly, Shorr et al. examined risk factors for prolonged hospital LOS and increased costs in a cohort of 291 patients with MRSA sterile site infection.17 Because in this study 23% of the patients received inappropriate empiric therapy, the authors were able to examine the impact of this exposure on utilization outcomes.17 In an adjusted analysis, inappropriate initial treatment was associated with an incremental increase in the LOS of 2.5 days, corresponding to the unadjusted cost differential of nearly $6,000.17 Although focusing on a different population, our results are consistent with these previous observations that antibiotic resistance and early inappropriate therapy affect hospital utilization parameters, in our case by adding nearly 2 days to the hospital LOS.

Our study has a number of limitations. First, as a retrospective cohort study it is prone to various forms of bias, most notably selection bias. To minimize the possibility of such, we established a priori case definitions and enrolled consecutive patients over a specific period of time. Second, as in any observational study, confounding is an issue. We dealt with this statistically by constructing a multivariable regression model; however, the possibility of residual confounding remains. Third, because some of the wound and ulcer cultures likely were obtained with a swab and thus represented colonization, rather than infection, we may have over‐estimated the rate of inappropriate therapy, and this needs to be followed up in future prospective studies. Similarly, we may have over‐estimated the likelihood of inappropriate therapy among polymicrobial and mixed infections as well, given that, for example, a gram‐negative organism may carry a different clinical significance when cultured from blood (infection) than when it is detected in a decubitus ulcer (potential colonization). Fourth, because we limited our cohort to patients without deep‐seated infections, such as necrotizing fasciitis, other procedures were not collected. This omission may have led to either over‐estimation or under‐estimation of the impact of inappropriate therapy on the outcomes of interest.

The fact that our cohort represents a single large urban academic tertiary care medical center may limit the generalizability of our results only to centers that share similar characteristics. Finally, similar to most other studies of this type, ours lacks data on posthospitalization outcomes and for this reason limits itself to hospital outcomes only.

In summary, we have shown that, similar to other populations with HCAI, a substantial proportion (nearly 1/4) of cSSSI patients with HCAI receive inappropriate empiric therapy for their infection, and this early exposure, though not affecting hospital mortality, is associated with a significant prolongation of the hospitalization by as much as 2 days. Studies are needed to refine decision rules for risk‐stratifying patients with cSSSI HCAI in order to determine the probability of infection with a resistant organism. In turn, such instruments at the bedside may assure improved utilization of appropriately targeted empiric therapy that will both optimize individual patient outcomes and reduce the risk of emergence of antimicrobial resistance.

Staphylococcus aureus (S. aureus) is well recognized as a major cause of nosocomial and healthcare‐associated pneumonia (HCAP). These infections are associated with substantial morbidity and mortality, and generate high medical costs.14 The significance of S. aureus in community‐acquired pneumonia (CAP) is less clear. Historically, S. aureus has not been considered a common pathogen in CAP, usually arising in association with or following influenza or an influenza‐like syndrome.58 However, with the increasing prevalence of methicillin‐resistant S. aureus (MRSA) as a cause of HCAPand most recently, serious community‐acquired non‐pulmonary infectionsinterest in this pathogen and its impact beyond the hospital have been expanding. During the 2003 to 2004 influenza season, 15 cases of influenza‐associated MRSA CAP were identified.6 In January 2007, the US Centers for Disease Control and Prevention (CDC) received reports of an additional 10 cases of severe MRSA CAP, resulting in six deaths, among previously healthy children and adults in Louisiana and Georgia.7 Kallen et al.9 recently reported 51 cases from 19 states between November 2006 and April 2007. To the best of our knowledge, these reports represent the largest case series describing MRSA in CAP.

Earlier studies have suggested a relationship between the outcome of S. aureus infection and the presence of various patient and strain characteristics.1013 Most observations regarding the adverse impact of MRSA on patient outcomes have arisen from analyses of cohorts of patients with either mixed infections, bacteremias, or nosocomial infection. There is little information on outcomes in patients with S. aureus CAP. To address this issue, we conducted a retrospective study of patients with culture‐proven S. aureus pneumonia admitted to a large urban hospital.

Materials and Methods

Sample Selection

The source population for the study consisted of all admissions to Henry Ford Hospital between January 2005 and May 2008 (study period) Patients were included in the study sample if they had: (1) a discharge diagnosis (principal or secondary) of pneumonia (International Classification of Diseases 9th Edition, Clinical Modification [ICD‐9‐CM] diagnosis codes 481.X 486.X) on their discharge summary or in their medical record; (2) a positive chest x‐ray (ie, infiltrate, consolidation, pleural effusion) within 48 hours of hospital admission; (3) an abnormal temperature (>38.3C [101.0F] or <36C [96.8F]), an abnormal white blood count (WBC) (>12,000/mm³ or <4,000/mm³), or increased sputum production on their day of hospital admission; and (4) a positive blood or respiratory culture for S. aureus within 48 hours of hospital admission.

Among these identified patients, those with probable HCAP were excluded from the study sample, based on evidence of: (1) hospitalization for 2 days during the 90‐day period preceding the index hospitalization; (2) admission to hospital from a nursing home or long‐term care facility; (3) hemodialysis 30 days prior to hospital admission; (4) receipt of cancer chemotherapy, IV antibiotic therapy, or wound care 30 days prior to hospital admission; or (5) receipt of an immunosuppressant at the time of hospital admission. All remaining patients were assumed to have CAP.

Results

There were 282 admissions to Henry Ford Hospital between January 2005 and May 2008 of patients with pneumonia who had positive blood or respiratory cultures for S. aureus within 48 hours of admission. (The total number of admissions over this period of patients with a principal diagnosis of pneumonia was 3894). Twelve patients had a negative chest x‐ray or negative clinical findings for pneumonia on the day of hospital admission and were excluded from the analysis. An additional 142 (53%) patients had evidence of HCAP and were excluded from the analysis. The final study sample therefore consisted of 128 patients with S. aureus CAP; their demographic and clinical characteristics are presented in Table 1.

Mean (SD) age of study subjects was 60 (17.0) years; 72% were nonwhite, and 58% were males. Prevalence of selected comorbidities, including acute renal failure, coronary artery disease, congestive heart failure, chronic obstructive pulmonary disease, diabetes, myocardial infarction, and peripheral vascular disease was high (>20%). Mean (SD) CURB score at admission was 1.8 (1.1); scores did not differ significantly between MRSA vs. MSSA patients. Eight patients (6%) had evidence of positive S. aureus cultures in the year prior to admission.

A total of 55 (43%) patients had positive MRSA cultures within the first 48 hours of admission, while the remainder (n = 73 [57%]) had cultures positive for MSSA. Among the 55 patients with MRSA, 23 (42%) had USA300 MRSA strains that produced PVL toxin. Only 18% of MRSA patients had MICs to vancomycin of 1 g/mL; most remaining patients (73%) had 1.5 g/mL MIC values. Thirty‐one patients (24%) died prior to hospital discharge (Table 2). Eighty‐eight (69%) patients received mechanical ventilation, and 79% were admitted to ICU; 25% of patients underwent thoracic surgery for pneumonia. Mean (SD) length of stay in hospital was 17.0 (15.7) days, and mean (SD) total charges per hospitalization were $127,922 ($154,605). (Mean length of stay in hospital was 5.8 days for all 3894 patients with a principal discharge diagnosis of pneumonia, and mean total charges were $29,448).

Notwithstanding the fact that patients with MRSA were more than twice as likely to receive inappropriate initial therapy (44% for MRSA vs. 18% for MSSA; odds ratio [OR], 3.57; 95% confidence interval [CI], 1.607.97; P = 0.0015), there were no statistically significant differences in in‐hospital mortality (26% for MSSA vs. 22% for MRSA; P = 0.678), surgery for pneumonia (26% vs. 24%; P = 0.838), or admission to ICU (82% vs. 74%; P = 0.382). Patients with MSSA CAP were more likely to receive mechanical ventilation (77% vs. 58%; P = 0.034). Mean (SD) total charges per admission did not differ between patients with MSSA vs. MRSA CAP ($135,784 [$170,046] vs $117,489 [$132,164], respectively [P = 0.510]). (30‐day mortality following hospital discharge was 24% for MRSA CAP patients and 30% for MSSA CAP patients.)

Among patients with MRSA CAP, there were no notable differences in outcomes between those with strains with the PVL toxin vs. those without it (Table 2). Mean (SD) length of stay in hospital, however, was significantly longer among the former patients (25 [22.6] days vs. 13 [7.7] days for those without PVL toxin; P = 0.020). Reflecting this difference, mean (SD) total charges per admission were $76,909 higher among MRSA patients with PVL‐positive strains vs. those with PVL‐negative strains ($162,124 [$186,923] vs. $85,215 [$57,957] respectively; P = 0.066).

Among MRSA patients with MICs to vancomycin of 1.5 g/mL, 17% died in hospital, 20% underwent surgery for pneumonia, 60% received mechanical ventilation, and 75% were admitted to ICU. Mean (SD) total charges per admission were $104,514 ($112,606) (Table 2).

Discussion

To the best of our knowledge, this is the largest observational study to date of outcomes in patients with S. aureus CAP. Our results indicate that MRSA represents almost one‐half of all such infections, with USA300 strains accounting for a substantial proportion of these cases. In addition, for most outcome measures, there were no significant differences between patients with MSSA vs. MRSA CAP. In patients with MRSA CAP, however, those with PVL‐positive strains had longer stays in the hospital and had higher total costs of hospitalization.

Our study highlights the significant clinical and economic burden of S. aureus CAP. In the US, the mortality rate in patients hospitalized with CAP is about 8%, while mean length of stay is approximately 5 days (it was 5.8 days in our institution among all patients with CAP). In contrast, we found that most patients with S. aureus CAP were admitted to ICU, and that nearly one in four died during hospitalization. One possible explanation for these findings is that patients with S. aureus CAP may often receive inappropriate initial antibiotic therapy, a major predictor of adverse outcome in infection. The overall rate of inappropriate treatment, however, was low in our study (although we note that optimal antibiotic treatment is not clearly defined for the USA300 strain, and as a result, we assumed that vancomycin would be appropriate for all MRSA infections). A more likely explanation for our findings is the severity of CAP associated with S. aureus, whether MRSA or MSSA. Given the nature of S. aureus infections, the need for prevention (and potentially, development of a vaccine) remains crucial.

While MRSA has been studied extensively as a cause of bacteremia, as well as HCAP and VAP, it has not been well studied in CAP. Patients with MRSA bacteremia have been shown to have a higher mortality risk and higher healthcare costs than those with MSSA infections, and a meta‐analysis of 31 studies of patients with S. aureus bloodstream infections demonstrated a significant increase in mortality among patients with MRSA vs. MSSA bacteremia.1820 Outcomes when comparing MRSA and MSSA in VAP are slightly more variable. A higher rate of mortality has been reported in patients with pneumonia caused by MRSA vs. MSSA, but others found no difference after controlling for potential confounders.21, 22 We found that case fatality, surgery for pneumonia, ICU admission, length of stay, and total hospital charges do not differ significantly between patients with MRSA vs. MSSA CAP. This suggests that appreciation of the burden of MRSA can only be done in the context of the syndrome in question, and that conclusions from analyses of bacteremia and VAP may not be generalizable to CAP.

In the US, an increasing number of community‐associated infections are due to MRSA. Skin and skin structure infections comprise the majority of community‐associated MRSA infections and are caused by a single pulsed field type, termed USA300. These strains are believed to have distinctive virulence and epidemiologic characteristics. USA300 isolates typically are resistant only to beta‐lactam and macrolide antimicrobial agents and contain genes for the PVL toxin, which typically are not present in strains of healthcare‐associated MRSA.23 Recent studies of acute pneumonia with animal models and in humans have suggested that the PVL toxinalone or in combination with other virulence factorsis associated with the development of necrotizing pneumonia.2430 In our study, 23 patients with MRSA CAP had strains that contained PVL toxin, and these patients had longer stays in hospital and higher total hospital charges than those with MRSA CAP not containing the PVL toxin. This is an important finding, as length of stay is an important proxy for morbidity and case severity. This also extends the finding of previous case studies reporting that MRSA CAP with PVL toxin is associated with worse outcomes.2430 Moreover, we clearly document that PVL‐positive strains are emerging as a cause of pulmonary infection in broader clinical scenarios. As such, physicians must remain vigilant for this toxin‐producing strain. The full extent of the impact of PVL‐positive strains in our study is unknown, as we cannot ascertain whether morbidity was worse because of the PVL toxin itself, or because most patients were not treated with clindamycin or linezolid, which inhibit toxin production.

We failed to observe a correlation between MICs to vancomycin and the outcomes we studied. The clinical significance of reduced susceptibility of S. aureus to vancomycin remains a controversial issue. In our study, there was no significant association between MICs to vancomycin and mortality, need for surgery, ICU admission, length of stay, or total hospital charges. Sakoulas et al.11 reported that as vancomycin MICs for MRSA isolates rose within the susceptible range from 0.5 mcg/mL to 2.0 mcg/mL, so too did the number of clinical failures among bacteremic patients receiving vancomycin. Moise‐Broder et al.31 reported a similar finding in a study of 63 patients with MRSA bacteremia, of whom 45 failed or were intolerant of vancomycin therapy. In a more recent report, Soriano et al.12 prospectively evaluated 414 episodes of bacteremia. The authors concluded that mortality in patients with MRSA bacteremia is significantly higher when the empiric antimicrobial agent is inappropriate and when vancomycin is used to treat infections involving strains with MICs >1.0 mcg/mL. In pneumonia, Hidayat et al.13 reported that patients with infections due to pathogens with higher MICs to vancomycin had a higher rate of mortality than those with lower MICs. Our findings likely differ from this report because of the proportion of subjects with PVL‐positive strains, and because of the skew in the distribution of vancomycin MICs. More specifically, the majority of patients had higher MIC strains, clustered around 1.5 mg/L, leaving us with few lower MIC strains and fewer strains with MICs of 2.0 mcg/mL for comparison, and therefore limited statistical power to assess the relationship between MICs and outcomes.

There were no clinical features that clearly distinguished patients with MRSA vs. MSSA CAP. This suggests that if physicians hope to ensure that patients with MRSA CAP receive appropriate initial antibiotic therapy, they cannot base therapeutic decision‐making (ie, use of an anti‐MRSA treatment) on clinical criteria. Presently, national guidelines recommend diagnostic testing only if the results might affect clinical decisions (eg, antimicrobial management).32 Furthermore, the IDSA/ATS guidelines recommend sputum cultures along with blood cultures and other diagnostic tests only in select cases (eg, those with severe disease). Since beginning optimal therapy quickly can reduce mortality in pneumonia,33 our results indicate a need for both rapid diagnostic tests to identify patients with MRSA, and reevaluation of current recommendations for diagnostic testing in CAP.

Our study has several important limitations. First, its retrospective design exposes it to many forms of bias. Second, because the diagnosis of pneumonia can be challenging, we required that all patients have both clinical and radiologic evidence of the disease. However, we may have thereby excluded some admissions for S. aureus pneumonia among the over 3000 pneumonia admissions to our hospital during the study period.

Third, we only included patients with culture evidence of infection. This was necessary by design, but as a result we may have failed to identify some patients with S. aureus infections because of the limitations of respiratory culture technology. Patients on general practice units often do not get cultured, and accordingly we may have missed those with milder disease caused by S. aureus.

Fourth, we used only one method of MIC testing to determine vancomycin susceptibility. It would have been of interest to have used automated dilution testing also and to have compared both methods. Comparison of laboratory methods was not among our study objectives, and most of the earlier studies examining epidemiology and outcomes used a single laboratory testing method only.

Finally, since the study was conducted at a large urban teaching hospital in a city with a long history of issues with MRSA and resistance, the generalizability of our results may be limited.

In summary, our study provides further evidence that S. aureus is an important pathogen in CAP. MRSA CAP has few obvious characteristics that differentiate it from MSSA CAP. MRSA and MSSA CAP are both very serious infections that should be treated aggressively to avoid poor outcomes.

Staphylococcus aureus (S. aureus) is well recognized as a major cause of nosocomial and healthcare‐associated pneumonia (HCAP). These infections are associated with substantial morbidity and mortality, and generate high medical costs.14 The significance of S. aureus in community‐acquired pneumonia (CAP) is less clear. Historically, S. aureus has not been considered a common pathogen in CAP, usually arising in association with or following influenza or an influenza‐like syndrome.58 However, with the increasing prevalence of methicillin‐resistant S. aureus (MRSA) as a cause of HCAPand most recently, serious community‐acquired non‐pulmonary infectionsinterest in this pathogen and its impact beyond the hospital have been expanding. During the 2003 to 2004 influenza season, 15 cases of influenza‐associated MRSA CAP were identified.6 In January 2007, the US Centers for Disease Control and Prevention (CDC) received reports of an additional 10 cases of severe MRSA CAP, resulting in six deaths, among previously healthy children and adults in Louisiana and Georgia.7 Kallen et al.9 recently reported 51 cases from 19 states between November 2006 and April 2007. To the best of our knowledge, these reports represent the largest case series describing MRSA in CAP.

Earlier studies have suggested a relationship between the outcome of S. aureus infection and the presence of various patient and strain characteristics.1013 Most observations regarding the adverse impact of MRSA on patient outcomes have arisen from analyses of cohorts of patients with either mixed infections, bacteremias, or nosocomial infection. There is little information on outcomes in patients with S. aureus CAP. To address this issue, we conducted a retrospective study of patients with culture‐proven S. aureus pneumonia admitted to a large urban hospital.

Materials and Methods

Sample Selection

The source population for the study consisted of all admissions to Henry Ford Hospital between January 2005 and May 2008 (study period) Patients were included in the study sample if they had: (1) a discharge diagnosis (principal or secondary) of pneumonia (International Classification of Diseases 9th Edition, Clinical Modification [ICD‐9‐CM] diagnosis codes 481.X 486.X) on their discharge summary or in their medical record; (2) a positive chest x‐ray (ie, infiltrate, consolidation, pleural effusion) within 48 hours of hospital admission; (3) an abnormal temperature (>38.3C [101.0F] or <36C [96.8F]), an abnormal white blood count (WBC) (>12,000/mm³ or <4,000/mm³), or increased sputum production on their day of hospital admission; and (4) a positive blood or respiratory culture for S. aureus within 48 hours of hospital admission.

Among these identified patients, those with probable HCAP were excluded from the study sample, based on evidence of: (1) hospitalization for 2 days during the 90‐day period preceding the index hospitalization; (2) admission to hospital from a nursing home or long‐term care facility; (3) hemodialysis 30 days prior to hospital admission; (4) receipt of cancer chemotherapy, IV antibiotic therapy, or wound care 30 days prior to hospital admission; or (5) receipt of an immunosuppressant at the time of hospital admission. All remaining patients were assumed to have CAP.

Results

There were 282 admissions to Henry Ford Hospital between January 2005 and May 2008 of patients with pneumonia who had positive blood or respiratory cultures for S. aureus within 48 hours of admission. (The total number of admissions over this period of patients with a principal diagnosis of pneumonia was 3894). Twelve patients had a negative chest x‐ray or negative clinical findings for pneumonia on the day of hospital admission and were excluded from the analysis. An additional 142 (53%) patients had evidence of HCAP and were excluded from the analysis. The final study sample therefore consisted of 128 patients with S. aureus CAP; their demographic and clinical characteristics are presented in Table 1.

Mean (SD) age of study subjects was 60 (17.0) years; 72% were nonwhite, and 58% were males. Prevalence of selected comorbidities, including acute renal failure, coronary artery disease, congestive heart failure, chronic obstructive pulmonary disease, diabetes, myocardial infarction, and peripheral vascular disease was high (>20%). Mean (SD) CURB score at admission was 1.8 (1.1); scores did not differ significantly between MRSA vs. MSSA patients. Eight patients (6%) had evidence of positive S. aureus cultures in the year prior to admission.

A total of 55 (43%) patients had positive MRSA cultures within the first 48 hours of admission, while the remainder (n = 73 [57%]) had cultures positive for MSSA. Among the 55 patients with MRSA, 23 (42%) had USA300 MRSA strains that produced PVL toxin. Only 18% of MRSA patients had MICs to vancomycin of 1 g/mL; most remaining patients (73%) had 1.5 g/mL MIC values. Thirty‐one patients (24%) died prior to hospital discharge (Table 2). Eighty‐eight (69%) patients received mechanical ventilation, and 79% were admitted to ICU; 25% of patients underwent thoracic surgery for pneumonia. Mean (SD) length of stay in hospital was 17.0 (15.7) days, and mean (SD) total charges per hospitalization were $127,922 ($154,605). (Mean length of stay in hospital was 5.8 days for all 3894 patients with a principal discharge diagnosis of pneumonia, and mean total charges were $29,448).

Notwithstanding the fact that patients with MRSA were more than twice as likely to receive inappropriate initial therapy (44% for MRSA vs. 18% for MSSA; odds ratio [OR], 3.57; 95% confidence interval [CI], 1.607.97; P = 0.0015), there were no statistically significant differences in in‐hospital mortality (26% for MSSA vs. 22% for MRSA; P = 0.678), surgery for pneumonia (26% vs. 24%; P = 0.838), or admission to ICU (82% vs. 74%; P = 0.382). Patients with MSSA CAP were more likely to receive mechanical ventilation (77% vs. 58%; P = 0.034). Mean (SD) total charges per admission did not differ between patients with MSSA vs. MRSA CAP ($135,784 [$170,046] vs $117,489 [$132,164], respectively [P = 0.510]). (30‐day mortality following hospital discharge was 24% for MRSA CAP patients and 30% for MSSA CAP patients.)

Among patients with MRSA CAP, there were no notable differences in outcomes between those with strains with the PVL toxin vs. those without it (Table 2). Mean (SD) length of stay in hospital, however, was significantly longer among the former patients (25 [22.6] days vs. 13 [7.7] days for those without PVL toxin; P = 0.020). Reflecting this difference, mean (SD) total charges per admission were $76,909 higher among MRSA patients with PVL‐positive strains vs. those with PVL‐negative strains ($162,124 [$186,923] vs. $85,215 [$57,957] respectively; P = 0.066).

Among MRSA patients with MICs to vancomycin of 1.5 g/mL, 17% died in hospital, 20% underwent surgery for pneumonia, 60% received mechanical ventilation, and 75% were admitted to ICU. Mean (SD) total charges per admission were $104,514 ($112,606) (Table 2).

Discussion

To the best of our knowledge, this is the largest observational study to date of outcomes in patients with S. aureus CAP. Our results indicate that MRSA represents almost one‐half of all such infections, with USA300 strains accounting for a substantial proportion of these cases. In addition, for most outcome measures, there were no significant differences between patients with MSSA vs. MRSA CAP. In patients with MRSA CAP, however, those with PVL‐positive strains had longer stays in the hospital and had higher total costs of hospitalization.

Our study highlights the significant clinical and economic burden of S. aureus CAP. In the US, the mortality rate in patients hospitalized with CAP is about 8%, while mean length of stay is approximately 5 days (it was 5.8 days in our institution among all patients with CAP). In contrast, we found that most patients with S. aureus CAP were admitted to ICU, and that nearly one in four died during hospitalization. One possible explanation for these findings is that patients with S. aureus CAP may often receive inappropriate initial antibiotic therapy, a major predictor of adverse outcome in infection. The overall rate of inappropriate treatment, however, was low in our study (although we note that optimal antibiotic treatment is not clearly defined for the USA300 strain, and as a result, we assumed that vancomycin would be appropriate for all MRSA infections). A more likely explanation for our findings is the severity of CAP associated with S. aureus, whether MRSA or MSSA. Given the nature of S. aureus infections, the need for prevention (and potentially, development of a vaccine) remains crucial.

While MRSA has been studied extensively as a cause of bacteremia, as well as HCAP and VAP, it has not been well studied in CAP. Patients with MRSA bacteremia have been shown to have a higher mortality risk and higher healthcare costs than those with MSSA infections, and a meta‐analysis of 31 studies of patients with S. aureus bloodstream infections demonstrated a significant increase in mortality among patients with MRSA vs. MSSA bacteremia.1820 Outcomes when comparing MRSA and MSSA in VAP are slightly more variable. A higher rate of mortality has been reported in patients with pneumonia caused by MRSA vs. MSSA, but others found no difference after controlling for potential confounders.21, 22 We found that case fatality, surgery for pneumonia, ICU admission, length of stay, and total hospital charges do not differ significantly between patients with MRSA vs. MSSA CAP. This suggests that appreciation of the burden of MRSA can only be done in the context of the syndrome in question, and that conclusions from analyses of bacteremia and VAP may not be generalizable to CAP.

In the US, an increasing number of community‐associated infections are due to MRSA. Skin and skin structure infections comprise the majority of community‐associated MRSA infections and are caused by a single pulsed field type, termed USA300. These strains are believed to have distinctive virulence and epidemiologic characteristics. USA300 isolates typically are resistant only to beta‐lactam and macrolide antimicrobial agents and contain genes for the PVL toxin, which typically are not present in strains of healthcare‐associated MRSA.23 Recent studies of acute pneumonia with animal models and in humans have suggested that the PVL toxinalone or in combination with other virulence factorsis associated with the development of necrotizing pneumonia.2430 In our study, 23 patients with MRSA CAP had strains that contained PVL toxin, and these patients had longer stays in hospital and higher total hospital charges than those with MRSA CAP not containing the PVL toxin. This is an important finding, as length of stay is an important proxy for morbidity and case severity. This also extends the finding of previous case studies reporting that MRSA CAP with PVL toxin is associated with worse outcomes.2430 Moreover, we clearly document that PVL‐positive strains are emerging as a cause of pulmonary infection in broader clinical scenarios. As such, physicians must remain vigilant for this toxin‐producing strain. The full extent of the impact of PVL‐positive strains in our study is unknown, as we cannot ascertain whether morbidity was worse because of the PVL toxin itself, or because most patients were not treated with clindamycin or linezolid, which inhibit toxin production.

We failed to observe a correlation between MICs to vancomycin and the outcomes we studied. The clinical significance of reduced susceptibility of S. aureus to vancomycin remains a controversial issue. In our study, there was no significant association between MICs to vancomycin and mortality, need for surgery, ICU admission, length of stay, or total hospital charges. Sakoulas et al.11 reported that as vancomycin MICs for MRSA isolates rose within the susceptible range from 0.5 mcg/mL to 2.0 mcg/mL, so too did the number of clinical failures among bacteremic patients receiving vancomycin. Moise‐Broder et al.31 reported a similar finding in a study of 63 patients with MRSA bacteremia, of whom 45 failed or were intolerant of vancomycin therapy. In a more recent report, Soriano et al.12 prospectively evaluated 414 episodes of bacteremia. The authors concluded that mortality in patients with MRSA bacteremia is significantly higher when the empiric antimicrobial agent is inappropriate and when vancomycin is used to treat infections involving strains with MICs >1.0 mcg/mL. In pneumonia, Hidayat et al.13 reported that patients with infections due to pathogens with higher MICs to vancomycin had a higher rate of mortality than those with lower MICs. Our findings likely differ from this report because of the proportion of subjects with PVL‐positive strains, and because of the skew in the distribution of vancomycin MICs. More specifically, the majority of patients had higher MIC strains, clustered around 1.5 mg/L, leaving us with few lower MIC strains and fewer strains with MICs of 2.0 mcg/mL for comparison, and therefore limited statistical power to assess the relationship between MICs and outcomes.

There were no clinical features that clearly distinguished patients with MRSA vs. MSSA CAP. This suggests that if physicians hope to ensure that patients with MRSA CAP receive appropriate initial antibiotic therapy, they cannot base therapeutic decision‐making (ie, use of an anti‐MRSA treatment) on clinical criteria. Presently, national guidelines recommend diagnostic testing only if the results might affect clinical decisions (eg, antimicrobial management).32 Furthermore, the IDSA/ATS guidelines recommend sputum cultures along with blood cultures and other diagnostic tests only in select cases (eg, those with severe disease). Since beginning optimal therapy quickly can reduce mortality in pneumonia,33 our results indicate a need for both rapid diagnostic tests to identify patients with MRSA, and reevaluation of current recommendations for diagnostic testing in CAP.

Our study has several important limitations. First, its retrospective design exposes it to many forms of bias. Second, because the diagnosis of pneumonia can be challenging, we required that all patients have both clinical and radiologic evidence of the disease. However, we may have thereby excluded some admissions for S. aureus pneumonia among the over 3000 pneumonia admissions to our hospital during the study period.

Third, we only included patients with culture evidence of infection. This was necessary by design, but as a result we may have failed to identify some patients with S. aureus infections because of the limitations of respiratory culture technology. Patients on general practice units often do not get cultured, and accordingly we may have missed those with milder disease caused by S. aureus.

Fourth, we used only one method of MIC testing to determine vancomycin susceptibility. It would have been of interest to have used automated dilution testing also and to have compared both methods. Comparison of laboratory methods was not among our study objectives, and most of the earlier studies examining epidemiology and outcomes used a single laboratory testing method only.

Finally, since the study was conducted at a large urban teaching hospital in a city with a long history of issues with MRSA and resistance, the generalizability of our results may be limited.

In summary, our study provides further evidence that S. aureus is an important pathogen in CAP. MRSA CAP has few obvious characteristics that differentiate it from MSSA CAP. MRSA and MSSA CAP are both very serious infections that should be treated aggressively to avoid poor outcomes.

Staphylococcus aureus (S. aureus) is well recognized as a major cause of nosocomial and healthcare‐associated pneumonia (HCAP). These infections are associated with substantial morbidity and mortality, and generate high medical costs.14 The significance of S. aureus in community‐acquired pneumonia (CAP) is less clear. Historically, S. aureus has not been considered a common pathogen in CAP, usually arising in association with or following influenza or an influenza‐like syndrome.58 However, with the increasing prevalence of methicillin‐resistant S. aureus (MRSA) as a cause of HCAPand most recently, serious community‐acquired non‐pulmonary infectionsinterest in this pathogen and its impact beyond the hospital have been expanding. During the 2003 to 2004 influenza season, 15 cases of influenza‐associated MRSA CAP were identified.6 In January 2007, the US Centers for Disease Control and Prevention (CDC) received reports of an additional 10 cases of severe MRSA CAP, resulting in six deaths, among previously healthy children and adults in Louisiana and Georgia.7 Kallen et al.9 recently reported 51 cases from 19 states between November 2006 and April 2007. To the best of our knowledge, these reports represent the largest case series describing MRSA in CAP.

Earlier studies have suggested a relationship between the outcome of S. aureus infection and the presence of various patient and strain characteristics.1013 Most observations regarding the adverse impact of MRSA on patient outcomes have arisen from analyses of cohorts of patients with either mixed infections, bacteremias, or nosocomial infection. There is little information on outcomes in patients with S. aureus CAP. To address this issue, we conducted a retrospective study of patients with culture‐proven S. aureus pneumonia admitted to a large urban hospital.

Materials and Methods

Sample Selection

The source population for the study consisted of all admissions to Henry Ford Hospital between January 2005 and May 2008 (study period) Patients were included in the study sample if they had: (1) a discharge diagnosis (principal or secondary) of pneumonia (International Classification of Diseases 9th Edition, Clinical Modification [ICD‐9‐CM] diagnosis codes 481.X 486.X) on their discharge summary or in their medical record; (2) a positive chest x‐ray (ie, infiltrate, consolidation, pleural effusion) within 48 hours of hospital admission; (3) an abnormal temperature (>38.3C [101.0F] or <36C [96.8F]), an abnormal white blood count (WBC) (>12,000/mm³ or <4,000/mm³), or increased sputum production on their day of hospital admission; and (4) a positive blood or respiratory culture for S. aureus within 48 hours of hospital admission.

Among these identified patients, those with probable HCAP were excluded from the study sample, based on evidence of: (1) hospitalization for 2 days during the 90‐day period preceding the index hospitalization; (2) admission to hospital from a nursing home or long‐term care facility; (3) hemodialysis 30 days prior to hospital admission; (4) receipt of cancer chemotherapy, IV antibiotic therapy, or wound care 30 days prior to hospital admission; or (5) receipt of an immunosuppressant at the time of hospital admission. All remaining patients were assumed to have CAP.

Results

There were 282 admissions to Henry Ford Hospital between January 2005 and May 2008 of patients with pneumonia who had positive blood or respiratory cultures for S. aureus within 48 hours of admission. (The total number of admissions over this period of patients with a principal diagnosis of pneumonia was 3894). Twelve patients had a negative chest x‐ray or negative clinical findings for pneumonia on the day of hospital admission and were excluded from the analysis. An additional 142 (53%) patients had evidence of HCAP and were excluded from the analysis. The final study sample therefore consisted of 128 patients with S. aureus CAP; their demographic and clinical characteristics are presented in Table 1.

Mean (SD) age of study subjects was 60 (17.0) years; 72% were nonwhite, and 58% were males. Prevalence of selected comorbidities, including acute renal failure, coronary artery disease, congestive heart failure, chronic obstructive pulmonary disease, diabetes, myocardial infarction, and peripheral vascular disease was high (>20%). Mean (SD) CURB score at admission was 1.8 (1.1); scores did not differ significantly between MRSA vs. MSSA patients. Eight patients (6%) had evidence of positive S. aureus cultures in the year prior to admission.

A total of 55 (43%) patients had positive MRSA cultures within the first 48 hours of admission, while the remainder (n = 73 [57%]) had cultures positive for MSSA. Among the 55 patients with MRSA, 23 (42%) had USA300 MRSA strains that produced PVL toxin. Only 18% of MRSA patients had MICs to vancomycin of 1 g/mL; most remaining patients (73%) had 1.5 g/mL MIC values. Thirty‐one patients (24%) died prior to hospital discharge (Table 2). Eighty‐eight (69%) patients received mechanical ventilation, and 79% were admitted to ICU; 25% of patients underwent thoracic surgery for pneumonia. Mean (SD) length of stay in hospital was 17.0 (15.7) days, and mean (SD) total charges per hospitalization were $127,922 ($154,605). (Mean length of stay in hospital was 5.8 days for all 3894 patients with a principal discharge diagnosis of pneumonia, and mean total charges were $29,448).

Notwithstanding the fact that patients with MRSA were more than twice as likely to receive inappropriate initial therapy (44% for MRSA vs. 18% for MSSA; odds ratio [OR], 3.57; 95% confidence interval [CI], 1.607.97; P = 0.0015), there were no statistically significant differences in in‐hospital mortality (26% for MSSA vs. 22% for MRSA; P = 0.678), surgery for pneumonia (26% vs. 24%; P = 0.838), or admission to ICU (82% vs. 74%; P = 0.382). Patients with MSSA CAP were more likely to receive mechanical ventilation (77% vs. 58%; P = 0.034). Mean (SD) total charges per admission did not differ between patients with MSSA vs. MRSA CAP ($135,784 [$170,046] vs $117,489 [$132,164], respectively [P = 0.510]). (30‐day mortality following hospital discharge was 24% for MRSA CAP patients and 30% for MSSA CAP patients.)

Among patients with MRSA CAP, there were no notable differences in outcomes between those with strains with the PVL toxin vs. those without it (Table 2). Mean (SD) length of stay in hospital, however, was significantly longer among the former patients (25 [22.6] days vs. 13 [7.7] days for those without PVL toxin; P = 0.020). Reflecting this difference, mean (SD) total charges per admission were $76,909 higher among MRSA patients with PVL‐positive strains vs. those with PVL‐negative strains ($162,124 [$186,923] vs. $85,215 [$57,957] respectively; P = 0.066).

Among MRSA patients with MICs to vancomycin of 1.5 g/mL, 17% died in hospital, 20% underwent surgery for pneumonia, 60% received mechanical ventilation, and 75% were admitted to ICU. Mean (SD) total charges per admission were $104,514 ($112,606) (Table 2).

Discussion

To the best of our knowledge, this is the largest observational study to date of outcomes in patients with S. aureus CAP. Our results indicate that MRSA represents almost one‐half of all such infections, with USA300 strains accounting for a substantial proportion of these cases. In addition, for most outcome measures, there were no significant differences between patients with MSSA vs. MRSA CAP. In patients with MRSA CAP, however, those with PVL‐positive strains had longer stays in the hospital and had higher total costs of hospitalization.

Our study highlights the significant clinical and economic burden of S. aureus CAP. In the US, the mortality rate in patients hospitalized with CAP is about 8%, while mean length of stay is approximately 5 days (it was 5.8 days in our institution among all patients with CAP). In contrast, we found that most patients with S. aureus CAP were admitted to ICU, and that nearly one in four died during hospitalization. One possible explanation for these findings is that patients with S. aureus CAP may often receive inappropriate initial antibiotic therapy, a major predictor of adverse outcome in infection. The overall rate of inappropriate treatment, however, was low in our study (although we note that optimal antibiotic treatment is not clearly defined for the USA300 strain, and as a result, we assumed that vancomycin would be appropriate for all MRSA infections). A more likely explanation for our findings is the severity of CAP associated with S. aureus, whether MRSA or MSSA. Given the nature of S. aureus infections, the need for prevention (and potentially, development of a vaccine) remains crucial.

While MRSA has been studied extensively as a cause of bacteremia, as well as HCAP and VAP, it has not been well studied in CAP. Patients with MRSA bacteremia have been shown to have a higher mortality risk and higher healthcare costs than those with MSSA infections, and a meta‐analysis of 31 studies of patients with S. aureus bloodstream infections demonstrated a significant increase in mortality among patients with MRSA vs. MSSA bacteremia.1820 Outcomes when comparing MRSA and MSSA in VAP are slightly more variable. A higher rate of mortality has been reported in patients with pneumonia caused by MRSA vs. MSSA, but others found no difference after controlling for potential confounders.21, 22 We found that case fatality, surgery for pneumonia, ICU admission, length of stay, and total hospital charges do not differ significantly between patients with MRSA vs. MSSA CAP. This suggests that appreciation of the burden of MRSA can only be done in the context of the syndrome in question, and that conclusions from analyses of bacteremia and VAP may not be generalizable to CAP.

In the US, an increasing number of community‐associated infections are due to MRSA. Skin and skin structure infections comprise the majority of community‐associated MRSA infections and are caused by a single pulsed field type, termed USA300. These strains are believed to have distinctive virulence and epidemiologic characteristics. USA300 isolates typically are resistant only to beta‐lactam and macrolide antimicrobial agents and contain genes for the PVL toxin, which typically are not present in strains of healthcare‐associated MRSA.23 Recent studies of acute pneumonia with animal models and in humans have suggested that the PVL toxinalone or in combination with other virulence factorsis associated with the development of necrotizing pneumonia.2430 In our study, 23 patients with MRSA CAP had strains that contained PVL toxin, and these patients had longer stays in hospital and higher total hospital charges than those with MRSA CAP not containing the PVL toxin. This is an important finding, as length of stay is an important proxy for morbidity and case severity. This also extends the finding of previous case studies reporting that MRSA CAP with PVL toxin is associated with worse outcomes.2430 Moreover, we clearly document that PVL‐positive strains are emerging as a cause of pulmonary infection in broader clinical scenarios. As such, physicians must remain vigilant for this toxin‐producing strain. The full extent of the impact of PVL‐positive strains in our study is unknown, as we cannot ascertain whether morbidity was worse because of the PVL toxin itself, or because most patients were not treated with clindamycin or linezolid, which inhibit toxin production.

We failed to observe a correlation between MICs to vancomycin and the outcomes we studied. The clinical significance of reduced susceptibility of S. aureus to vancomycin remains a controversial issue. In our study, there was no significant association between MICs to vancomycin and mortality, need for surgery, ICU admission, length of stay, or total hospital charges. Sakoulas et al.11 reported that as vancomycin MICs for MRSA isolates rose within the susceptible range from 0.5 mcg/mL to 2.0 mcg/mL, so too did the number of clinical failures among bacteremic patients receiving vancomycin. Moise‐Broder et al.31 reported a similar finding in a study of 63 patients with MRSA bacteremia, of whom 45 failed or were intolerant of vancomycin therapy. In a more recent report, Soriano et al.12 prospectively evaluated 414 episodes of bacteremia. The authors concluded that mortality in patients with MRSA bacteremia is significantly higher when the empiric antimicrobial agent is inappropriate and when vancomycin is used to treat infections involving strains with MICs >1.0 mcg/mL. In pneumonia, Hidayat et al.13 reported that patients with infections due to pathogens with higher MICs to vancomycin had a higher rate of mortality than those with lower MICs. Our findings likely differ from this report because of the proportion of subjects with PVL‐positive strains, and because of the skew in the distribution of vancomycin MICs. More specifically, the majority of patients had higher MIC strains, clustered around 1.5 mg/L, leaving us with few lower MIC strains and fewer strains with MICs of 2.0 mcg/mL for comparison, and therefore limited statistical power to assess the relationship between MICs and outcomes.

There were no clinical features that clearly distinguished patients with MRSA vs. MSSA CAP. This suggests that if physicians hope to ensure that patients with MRSA CAP receive appropriate initial antibiotic therapy, they cannot base therapeutic decision‐making (ie, use of an anti‐MRSA treatment) on clinical criteria. Presently, national guidelines recommend diagnostic testing only if the results might affect clinical decisions (eg, antimicrobial management).32 Furthermore, the IDSA/ATS guidelines recommend sputum cultures along with blood cultures and other diagnostic tests only in select cases (eg, those with severe disease). Since beginning optimal therapy quickly can reduce mortality in pneumonia,33 our results indicate a need for both rapid diagnostic tests to identify patients with MRSA, and reevaluation of current recommendations for diagnostic testing in CAP.

Our study has several important limitations. First, its retrospective design exposes it to many forms of bias. Second, because the diagnosis of pneumonia can be challenging, we required that all patients have both clinical and radiologic evidence of the disease. However, we may have thereby excluded some admissions for S. aureus pneumonia among the over 3000 pneumonia admissions to our hospital during the study period.

Third, we only included patients with culture evidence of infection. This was necessary by design, but as a result we may have failed to identify some patients with S. aureus infections because of the limitations of respiratory culture technology. Patients on general practice units often do not get cultured, and accordingly we may have missed those with milder disease caused by S. aureus.

Fourth, we used only one method of MIC testing to determine vancomycin susceptibility. It would have been of interest to have used automated dilution testing also and to have compared both methods. Comparison of laboratory methods was not among our study objectives, and most of the earlier studies examining epidemiology and outcomes used a single laboratory testing method only.

Finally, since the study was conducted at a large urban teaching hospital in a city with a long history of issues with MRSA and resistance, the generalizability of our results may be limited.

In summary, our study provides further evidence that S. aureus is an important pathogen in CAP. MRSA CAP has few obvious characteristics that differentiate it from MSSA CAP. MRSA and MSSA CAP are both very serious infections that should be treated aggressively to avoid poor outcomes.

There is a potential for discontinuity of care in the management of inpatients by hospitalists since many patients are cared for by more than one physician during their hospitalization. Previous studies have explored the impact of this type of discontinuity of care in residency programs. With the restrictions put in place by the Accreditation Council on Graduate Medical Education (ACGME) in 2003, there has been an increase in the fragmentation of care (FOC) that patients receive in the hospital. Studies have explored the impact of these changes on length of stay (LOS) and quality of care. The results have been mixed, with some studies showing that the increased FOC was associated with prolonged LOS,1, 2 as well as having a negative influence on quality.2, 3 Other studies have shown no change or a reduction in LOS,4 and an improvement in quality measures.46

There have been no prior studies on the impact of hospitalist care as the source of FOC on LOS. We therefore undertook a study to explore the impact of fragmented hospitalist care on LOS. Additionally, there has been some discussion on the impact of admission day of week on both FOC and LOS. Prior studies have mainly looked at day of the week in intensive care unit patients. One study found a modest increase in ICU LOS in patients admitted during the weekend.7 We therefore also looked at the impact of day of the week of admission on FOC in order to determine if fragmentation was just a proxy for admission day as the true indicator for increased LOS.

Methods

Results

Table 1 reviews the demographics of patients with DRG 89 (PNA) and DRG 127 (HF). A large proportion of patients with both DRGs experienced no fragmentation in their physician care during hospitalization; 4152 patients with HF (46.3%) and 685 patients with PNA (35.8%) had visits from only 1 hospitalist throughout their hospital stay, for a FOC Index of 0%. The mean fragmentation level was 21.9% for pneumonia patients, with the mean number of hospitalists seen during the stay at 2.05. For heart failure patients, the mean fragmentation level was 18.3 % and a mean of 1.78 hospitalists seen (Table 1).

Table 2 presents the results of the regression analysis on LOS. The association between fragmentation and LOS, adjusted for demographics, case mix and day of admission, was similar for PNA and HF patients. We found an increase in LOS of 0.39 days for each 10% increase in fragmentation level for pneumonia. Other variables that significantly increased LOS for PNA were larger ROM score, Sunday admission and more discharge diagnoses. Similarly, for heart failure patients, there was an increase in LOS of 0.30 days for each 10% increase in fragmentation level. Other variables associated with a significantly increase in LOS for PNA were larger ROM score, larger SOI category, more discharge diagnoses, and age. Figures 1 and 2 demonstrate the results graphically for LOS in each condition. They show the mean adjusted LOS by fragmentation level. Adjusted LOS was calculated for each patient based on the final model. For example, the average adjusted LOS for PNA patients with a fragmentation level of 20% to 30% is 8.06 days and 9.16 days for patients with a fragmentation level of 30% to 40%.

Figure 1

Figure 2

Discussion

This study demonstrated that there was a statistically significant association between FOC by hospitalist physicians and LOS for patients with DRG 89 (PNA) and DRG 127 (HF). As the percentage of fragmentation increased, the LOS increased significantly.

There are many ways in which to define FOC in the context of hospitalist care of inpatients. We chose to use an index similar to the Usual Provider of Care Index (UPC) which is a standard way to measure continuity of care in the primary care setting.9 The UPC measures the proportion of time spent with the primary provider. In the inpatient setting, we defined continuity of care as the proportion of visits by the primary hospitalist (the physician assuming the greatest number of days of the patient's care). The FOC Index is simply 1 minus the continuity of care. There are other potential measures of FOC, such as the number of handoffs or the number of physicians. We selected our measure based on its following strengths: (1) it does not have the simultaneity problem of measures such as the number of doctors or the number of handoffs that leads to endogeneitya correlation between the error term and the independent variable that biases the coefficient. Endogeneity can occur when the outcome variable and the independent variable are jointly determined, in this case, that an additional day in the hospital increases the likelihood of having an additional handoff as well as an additional handoff increasing the likelihood of an additional day; (2) fragmentation patterns that lead to handoffs on the weekend that then return to the primary hospitalist may be very different from fragmentation patterns that lead to new physicians at each transfer; (3) It provides a good comparison to other models of care where community physicians provide care to their own patients and have an effective FOC index of 0. Further research should be performed looking at different measures that may capture other aspects of care fragmentation.

This is one of the first studies to directly explore the relationship between FOC and LOS. Previous studies in the hospitalist literature have explored the impact of variables on LOS in which FOC may have been an indirect influence. For example, one study found a 13% shorter LOS among academic hospitalists at a teaching hospital who worked on a block rotation compared to a group of community hospitalists with a schedule involving more patient handoffs.10 Although there is little literature on FOC among hospitalists, the literature on medical residency programs is informative. The amount of discontinuity and the number of housestaff‐to‐housestaff transfers of responsibility has increased dramatically since the institution of more stringent work‐hour restrictions by the Accreditation Council for Graduate Medical Education (ACGME) in 2003.11 Horwitz et al.12 found that after the institution of work‐hour regulations, there was an 11% increase in number of transfers of care for a hypothetical patient admitted from Monday until Friday. They noted that programs with a night‐float system had a statistically significant increase in transitions of care compared to residencies without night float. Studies on the impact of the ACGME regulations on LOS have been mixed. One study found that LOS was reduced, and that there was improved adherence to quality indicators.4 A recent study found that there was a 44% increase in the median LOS when Short Call admitting teams were involved.2

There are several limitations to the findings of this study. The largest potential limitation is interpreting the direction of causality between FOC and LOS. This was a concurrent control study. Because the study was not randomized, there is a risk of confounding variables. Admission day of the week was 1 variable of concern for confounding due to the existence of a significant relationship between admission day LOS and between admission day and FOC. We included admission day in our modeling process. The coefficient estimates on FOC were stable, changing less than 1% when day of admission was added to the model. The risk of temporal trends was minimized by studying 1‐year's worth of data. Although we looked at variables that were potentially relevant in all hospitalized patients, every hospital and hospitalist practice has its own unique features that may impact both LOS and FOC. For example, for some hospitals, bed capacity may impact these measures. In many hospitals, hospitalist work force shortages may impact these measures by affecting the extent to which patients are cared for by full‐time hospitalists vs. locums tenens physicians. The large number of facilities in the study mitigates the influence of any individual hospital's LOS and FOC tendencies.

If increased FOC does result in prolonged LOS, the question arises as to how this fragmentation can be decreased. Although no physician can guarantee presence continually throughout a patient's inpatient stay, there are scheduling methods that reduce fragmentation and maximize the odds of a patient being followed by a single clinician. For example, the longer the block of days that a hospitalist is scheduled, the fewer hospitalists will need to care for a patient. Many of the top DRGs for patients cared for by hospitalists have lengths of stay between 4 and 5 days.13 Therefore, any schedule that has a hospitalist on for at least four days at a time will increase the likelihood that the same hospitalist will care for many of the patients throughout their stay. In making schedules, the goals of clinical efficiency and physician satisfaction must be weighed against the potential risks to the quality of patient care. For instance, the use of one hospitalist as the admitting physician for all patients may increase efficiency, but will also increase fragmentation.

A factor that may influence the impact of FOC on LOS is the quality of patient handoffs. The Joint Commission instituted this as a national patient safety goal in 2006.14 This goal was based on a Joint Commission analysis that 70% of sentinel events were caused by communication breakdowns, half of these occurring during handoffs.15

In conclusion, this study explored the relationship between FOC in patients cared for by hospitalists and LOS, using an FOC index. For every 10% increase in fragmentation, the LOS went up by 0.39 days for pneumonia and 0.30 days for heart failure. By adjusting for many variables that may impact LOS due to higher severity or complexity of illness, there is an increased likelihood that the FOC may have a causative relationship with the prolonged LOS.

There is a potential for discontinuity of care in the management of inpatients by hospitalists since many patients are cared for by more than one physician during their hospitalization. Previous studies have explored the impact of this type of discontinuity of care in residency programs. With the restrictions put in place by the Accreditation Council on Graduate Medical Education (ACGME) in 2003, there has been an increase in the fragmentation of care (FOC) that patients receive in the hospital. Studies have explored the impact of these changes on length of stay (LOS) and quality of care. The results have been mixed, with some studies showing that the increased FOC was associated with prolonged LOS,1, 2 as well as having a negative influence on quality.2, 3 Other studies have shown no change or a reduction in LOS,4 and an improvement in quality measures.46

There have been no prior studies on the impact of hospitalist care as the source of FOC on LOS. We therefore undertook a study to explore the impact of fragmented hospitalist care on LOS. Additionally, there has been some discussion on the impact of admission day of week on both FOC and LOS. Prior studies have mainly looked at day of the week in intensive care unit patients. One study found a modest increase in ICU LOS in patients admitted during the weekend.7 We therefore also looked at the impact of day of the week of admission on FOC in order to determine if fragmentation was just a proxy for admission day as the true indicator for increased LOS.

Methods

Results

Table 1 reviews the demographics of patients with DRG 89 (PNA) and DRG 127 (HF). A large proportion of patients with both DRGs experienced no fragmentation in their physician care during hospitalization; 4152 patients with HF (46.3%) and 685 patients with PNA (35.8%) had visits from only 1 hospitalist throughout their hospital stay, for a FOC Index of 0%. The mean fragmentation level was 21.9% for pneumonia patients, with the mean number of hospitalists seen during the stay at 2.05. For heart failure patients, the mean fragmentation level was 18.3 % and a mean of 1.78 hospitalists seen (Table 1).

Table 2 presents the results of the regression analysis on LOS. The association between fragmentation and LOS, adjusted for demographics, case mix and day of admission, was similar for PNA and HF patients. We found an increase in LOS of 0.39 days for each 10% increase in fragmentation level for pneumonia. Other variables that significantly increased LOS for PNA were larger ROM score, Sunday admission and more discharge diagnoses. Similarly, for heart failure patients, there was an increase in LOS of 0.30 days for each 10% increase in fragmentation level. Other variables associated with a significantly increase in LOS for PNA were larger ROM score, larger SOI category, more discharge diagnoses, and age. Figures 1 and 2 demonstrate the results graphically for LOS in each condition. They show the mean adjusted LOS by fragmentation level. Adjusted LOS was calculated for each patient based on the final model. For example, the average adjusted LOS for PNA patients with a fragmentation level of 20% to 30% is 8.06 days and 9.16 days for patients with a fragmentation level of 30% to 40%.

Figure 1

Figure 2

Discussion

This study demonstrated that there was a statistically significant association between FOC by hospitalist physicians and LOS for patients with DRG 89 (PNA) and DRG 127 (HF). As the percentage of fragmentation increased, the LOS increased significantly.

There are many ways in which to define FOC in the context of hospitalist care of inpatients. We chose to use an index similar to the Usual Provider of Care Index (UPC) which is a standard way to measure continuity of care in the primary care setting.9 The UPC measures the proportion of time spent with the primary provider. In the inpatient setting, we defined continuity of care as the proportion of visits by the primary hospitalist (the physician assuming the greatest number of days of the patient's care). The FOC Index is simply 1 minus the continuity of care. There are other potential measures of FOC, such as the number of handoffs or the number of physicians. We selected our measure based on its following strengths: (1) it does not have the simultaneity problem of measures such as the number of doctors or the number of handoffs that leads to endogeneitya correlation between the error term and the independent variable that biases the coefficient. Endogeneity can occur when the outcome variable and the independent variable are jointly determined, in this case, that an additional day in the hospital increases the likelihood of having an additional handoff as well as an additional handoff increasing the likelihood of an additional day; (2) fragmentation patterns that lead to handoffs on the weekend that then return to the primary hospitalist may be very different from fragmentation patterns that lead to new physicians at each transfer; (3) It provides a good comparison to other models of care where community physicians provide care to their own patients and have an effective FOC index of 0. Further research should be performed looking at different measures that may capture other aspects of care fragmentation.

This is one of the first studies to directly explore the relationship between FOC and LOS. Previous studies in the hospitalist literature have explored the impact of variables on LOS in which FOC may have been an indirect influence. For example, one study found a 13% shorter LOS among academic hospitalists at a teaching hospital who worked on a block rotation compared to a group of community hospitalists with a schedule involving more patient handoffs.10 Although there is little literature on FOC among hospitalists, the literature on medical residency programs is informative. The amount of discontinuity and the number of housestaff‐to‐housestaff transfers of responsibility has increased dramatically since the institution of more stringent work‐hour restrictions by the Accreditation Council for Graduate Medical Education (ACGME) in 2003.11 Horwitz et al.12 found that after the institution of work‐hour regulations, there was an 11% increase in number of transfers of care for a hypothetical patient admitted from Monday until Friday. They noted that programs with a night‐float system had a statistically significant increase in transitions of care compared to residencies without night float. Studies on the impact of the ACGME regulations on LOS have been mixed. One study found that LOS was reduced, and that there was improved adherence to quality indicators.4 A recent study found that there was a 44% increase in the median LOS when Short Call admitting teams were involved.2

There are several limitations to the findings of this study. The largest potential limitation is interpreting the direction of causality between FOC and LOS. This was a concurrent control study. Because the study was not randomized, there is a risk of confounding variables. Admission day of the week was 1 variable of concern for confounding due to the existence of a significant relationship between admission day LOS and between admission day and FOC. We included admission day in our modeling process. The coefficient estimates on FOC were stable, changing less than 1% when day of admission was added to the model. The risk of temporal trends was minimized by studying 1‐year's worth of data. Although we looked at variables that were potentially relevant in all hospitalized patients, every hospital and hospitalist practice has its own unique features that may impact both LOS and FOC. For example, for some hospitals, bed capacity may impact these measures. In many hospitals, hospitalist work force shortages may impact these measures by affecting the extent to which patients are cared for by full‐time hospitalists vs. locums tenens physicians. The large number of facilities in the study mitigates the influence of any individual hospital's LOS and FOC tendencies.

If increased FOC does result in prolonged LOS, the question arises as to how this fragmentation can be decreased. Although no physician can guarantee presence continually throughout a patient's inpatient stay, there are scheduling methods that reduce fragmentation and maximize the odds of a patient being followed by a single clinician. For example, the longer the block of days that a hospitalist is scheduled, the fewer hospitalists will need to care for a patient. Many of the top DRGs for patients cared for by hospitalists have lengths of stay between 4 and 5 days.13 Therefore, any schedule that has a hospitalist on for at least four days at a time will increase the likelihood that the same hospitalist will care for many of the patients throughout their stay. In making schedules, the goals of clinical efficiency and physician satisfaction must be weighed against the potential risks to the quality of patient care. For instance, the use of one hospitalist as the admitting physician for all patients may increase efficiency, but will also increase fragmentation.

A factor that may influence the impact of FOC on LOS is the quality of patient handoffs. The Joint Commission instituted this as a national patient safety goal in 2006.14 This goal was based on a Joint Commission analysis that 70% of sentinel events were caused by communication breakdowns, half of these occurring during handoffs.15

In conclusion, this study explored the relationship between FOC in patients cared for by hospitalists and LOS, using an FOC index. For every 10% increase in fragmentation, the LOS went up by 0.39 days for pneumonia and 0.30 days for heart failure. By adjusting for many variables that may impact LOS due to higher severity or complexity of illness, there is an increased likelihood that the FOC may have a causative relationship with the prolonged LOS.

There is a potential for discontinuity of care in the management of inpatients by hospitalists since many patients are cared for by more than one physician during their hospitalization. Previous studies have explored the impact of this type of discontinuity of care in residency programs. With the restrictions put in place by the Accreditation Council on Graduate Medical Education (ACGME) in 2003, there has been an increase in the fragmentation of care (FOC) that patients receive in the hospital. Studies have explored the impact of these changes on length of stay (LOS) and quality of care. The results have been mixed, with some studies showing that the increased FOC was associated with prolonged LOS,1, 2 as well as having a negative influence on quality.2, 3 Other studies have shown no change or a reduction in LOS,4 and an improvement in quality measures.46

There have been no prior studies on the impact of hospitalist care as the source of FOC on LOS. We therefore undertook a study to explore the impact of fragmented hospitalist care on LOS. Additionally, there has been some discussion on the impact of admission day of week on both FOC and LOS. Prior studies have mainly looked at day of the week in intensive care unit patients. One study found a modest increase in ICU LOS in patients admitted during the weekend.7 We therefore also looked at the impact of day of the week of admission on FOC in order to determine if fragmentation was just a proxy for admission day as the true indicator for increased LOS.

Methods

Results