Samir S. Shah, MD, MSCE

In recent decades, hospital lengths of stay have decreased and there has been a shift toward outpatient management for many pediatric conditions. In 2003, one‐third of all children admitted to US hospitals experienced 1‐day inpatient stays, an increase from 19% in 1993.1 Some hospitals have developed dedicated observation units for the care of children, with select diagnoses, who are expected to respond to less than 24 hours of treatment.26 Expansion of observation services has been suggested as an approach to lessen emergency department (ED) crowding7 and alleviate high‐capacity conditions within hospital inpatient units.8

In contrast to care delivered in a dedicated observation unit, observation status is an administrative label applied to patients who do not meet inpatient criteria as defined by third parties such as InterQual. While the decision to admit a patient is ultimately at the discretion of the ordering physician, many hospitals use predetermined criteria to assign observation status to patients admitted to observation and inpatient units.9 Treatment provided under observation status is designated by hospitals and payers as outpatient care, even when delivered in an inpatient bed.10 As outpatient‐designated care, observation cases do not enter publicly available administrative datasets of hospital discharges that have traditionally been used to understand hospital resource utilization, including the National Hospital Discharge Survey and the Kid's Inpatient Database.11, 12

We hypothesize that there has been an increase in observation status care delivered to children in recent years, and that the majority of children under observation were discharged home without converting to inpatient status. To determine trends in pediatric observation status care, we conducted the first longitudinal, multicenter evaluation of observation status code utilization following ED treatment in a sample of US freestanding children's hospitals. In addition, we focused on the most recent year of data among top ranking diagnoses to assess the current state of observation status stay outcomes (including conversion to inpatient status and return visits).

METHODS

DATA ANALYSIS

For each of the 6 years of study, we calculated the following proportions to determine trends over time: 1) the number of Observation Status admissions from the ED as a proportion of the total number of ED visits resulting in Observation or Inpatient admission, and 2) the number of 1‐Day Stays admitted from the ED as a proportion of the total number of ED visits resulting in Observation or Inpatient admissions. Trends were analyzed using linear regression. Trends were also calculated for the total volume of admissions from the ED and the case‐mix index (CMI). CMI was assessed to evaluate for changes in the severity of illness for children admitted from the ED over the study period. Each hospital's CMI was calculated as an average of their Observation and Inpatient Status discharges' charge weights during the study period. Charge weights were calculated at the All Patient Refined Diagnosis Related Groups (APR‐DRG)/severity of illness level (3M Health Information Systems, St Paul, MN) and were normalized national average charges derived by Thomson‐Reuters from their Pediatric Projected National Database. Weights were then assigned to each discharge based on the discharge's APR‐DRG and severity level assignment.

To assess the current outcomes for observation, we analyzed stays with associated observation billing codes from the most recent year of available data (2009). Stays with Observation patient type were considered to have been discharged from observation, while those with an Inpatient Status patient type were considered to have converted to an inpatient admission during the observation period.

Using the 2009 data, we calculated descriptive statistics for patient characteristics (eg, age, gender, payer) comparing Observation Stays, 1‐Day Stays, and longer‐duration Inpatient admissions using chi‐square statistics. Age was categorized using the American Academy of Pediatrics groupings: <30 days, 30 days1 year, 12 years, 34 years, 512 years, 1317 years, >18 years. Designated payer was categorized into government, private, and other, including self‐pay and uninsured groups.

We used the Severity Classification Systems (SCS) developed for pediatric emergency care to estimate severity of illness for the visit.17 In this 5‐level system, each ICD‐9 diagnosis code is associated with a score related to the intensity of ED resources needed to care for a child with that diagnosis. In our analyses, each case was assigned the maximal SCS category based on the highest severity ICD‐9 code associated with the stay. Within the SCS, a score of 1 indicates minor illness (eg, diaper dermatitis) and 5 indicates major illness (eg, septic shock). The proportions of visits within categorical SCS scores were compared for Observation Stays, 1‐Day Stays, and longer‐duration Inpatient admissions using chi‐square statistics.

We determined the top 10 ranking diagnoses for which children were admitted from the ED in 2009 using the Diagnosis Grouping System (DGS).18 The DGS was designed specifically to categorize pediatric ED visits into clinically meaningful groups. The ICD‐9 code for the principal discharge diagnosis was used to assign records to 1 of the 77 DGS subgroups. Within each of the top ranking DGS subgroups, we determined the proportion of Observation Stays, 1‐Day Stays, and longer‐duration Inpatient admissions.

To provide clinically relevant outcomes of Observation Stays for common conditions, we selected stays with observation charges from within the top 10 ranking observation stay DGS subgroups in 2009. Outcomes for observation included: 1) immediate outcome of the observation stay (ie, discharge or conversion to inpatient status), 2) return visits to the ED in the 3 days following observation, and 3) readmissions to the hospital in the 3 and 30 days following observation. Bivariate comparisons of return visits and readmissions for Observation versus 1‐Day Stays within DGS subgroups were analyzed using chi‐square tests. Multivariate analyses of return visits and readmissions were conducted using Generalized Estimating Equations adjusting for severity of illness by SCS score and clustering by hospital. To account for local practice patterns, we also adjusted for a grouped treatment variable that included the site level proportion of children admitted to Observation Status, 1‐Day‐Stays, and longer Inpatient admissions. All statistical analyses were performed using SAS (version 9.2, SAS Institute, Inc, Cary, NC); P values <0.05 were considered statistically significant.

RESULTS

Trends in Short‐Stays

An increase in proportion of Observation Stays was mirrored by a decrease in proportion of 1‐Day Stays over the study period (Figure 1). In 2009, there were 1.4 times more Observation Stays than 1‐Day Stays (25,653 vs 18,425) compared with 14,242 and 20,747, respectively, in 2004. This shift toward more Observation Stays occurred as hospitals faced a 16% increase in the total number of admissions from the ED (91,318 to 108,217) and change in CMI from 1.48 to 1.51. Over the study period, roughly 40% of all admissions from the ED were Short‐Stays (Observation and 1‐Day Stays). Median LOS for Observation Status stays was 1 day (interquartile range [IQR]: 11).

Figure 1

DISCUSSION

In this first, multicenter longitudinal study of pediatric observation following an ED visit, we found that Observation Status code utilization has increased steadily over the past 6 years and, in 2007, the proportion of children admitted to observation status surpassed the proportion of children experiencing a 1‐day inpatient admission. Taken together, Short‐Stays made up more than 40% of the hospital‐based care delivered to children admitted from an ED. Stable trends in CMI over time suggest that observation status may be replacing inpatient status designated care for pediatric Short‐Stays in these hospitals. Our findings suggest the lines between outpatient observation and short‐stay inpatient care are becoming increasingly blurred. These trends have occurred in the setting of changing policies for hospital reimbursement, requirements for patients to meet criteria to qualify for inpatient admissions, and efforts to avoid stays deemed unnecessary or inappropriate by their brief duration.19 Therefore there is a growing need to understand the impact of children under observation on the structure, delivery, and financing of acute hospital care for children.

Our results also have implications for pediatric health services research that relies on hospital administrative databases that do not contain observation stays. Currently, observation stays are systematically excluded from many inpatient administrative datasets.11, 12 Analyses of datasets that do not account for observation stays likely result in underestimation of hospitalization rates and hospital resource utilization for children. This may be particularly important for high‐volume conditions, such as asthma and acute infections, for which children commonly require brief periods of hospital‐based care beyond an ED encounter. Data from pediatric observation status admissions should be consistently included in hospital administrative datasets to allow for more comprehensive analyses of hospital resource utilization among children.

Prior research has shown that the diagnoses commonly treated in pediatric observation units overlap with the diagnoses for which children experience 1‐Day Stays.1, 20 We found a similar pattern of conditions for which children were under Observation Status and 1‐Day Stays with comparable severity of illness between the groups in terms of SCS scores. Our findings imply a need to determine how and why hospitals differentiate Observation Status from 1‐Day‐Stay groups in order to improve the assignment of observation status. Assuming continued pressures from payers to provide more care in outpatient or observation settings, there is potential for expansion of dedicated observation services for children in the US. Without designated observation units or processes to group patients with lower severity conditions, there may be limited opportunities to realize more efficient hospital care simply through the application of the label of observation status.

For more than 30 years, observation services have been provided to children who require a period of monitoring to determine their response to therapy and the need for acute inpatient admission from the ED.21While we were not able to determine the location of care for observation status patients in this study, we know that few children's hospitals have dedicated observation units and, even when an observation unit is present, not all observation status patients are cared for in dedicated observation units.9 This, in essence, means that most children under observation status are cared for in virtual observation by inpatient teams using inpatient beds. If observation patients are treated in inpatient beds and consume the same resources as inpatients, then cost‐savings based on reimbursement contracts with payers may not reflect an actual reduction in services. Pediatric institutions will need to closely monitor the financial implications of observation status given the historical differences in payment for observation and inpatient care.

With more than 70% of children being discharged home following observation, our results are comparable to the published literature2, 5, 6, 22, 23 and guidelines for observation unit operations.24 Similar to prior studies,4, 15, 2530 our results also indicate that return visits and readmissions following observation are uncommon events. Our findings can serve as initial benchmarks for condition‐specific outcomes for pediatric observation care. Studies are needed both to identify the clinical characteristics predictive of successful discharge home from observation and to explore the hospital‐to‐hospital variability in outcomes for observation. Such studies are necessary to identify the most successful healthcare delivery models for pediatric observation stays.

LIMITATIONS

The primary limitation to our results is that data from a subset of freestanding children's hospitals may not reflect observation stays at other children's hospitals or the community hospitals that care for children across the US. Only 18 of 42 current PHIS member hospitals have provided both outpatient visit and inpatient stay data for each year of the study period and were considered eligible. In an effort to ensure the quality of observation stay data, we included the 16 hospitals that assigned observation charges to at least 90% of their observation status stays in the PHIS database. The exclusion of the 2 hospitals where <90% of observation status patients were assigned observation charges likely resulted in an underestimation of the utilization of observation status.

Second, there is potential for misclassification of patient type given institutional variations in the assignment of patient status. The PHIS database does not contain information about the factors that were considered in the assignment of observation status. At the time of admission from the ED, observation or inpatient status is assigned. While this decision is clearly reserved for the admitting physician, the process is not standardized across hospitals.9 Some institutions have Utilization Managers on site to help guide decision‐making, while others allow the assignment to be made by physicians without specific guidance. As a result, some patients may be assigned to observation status at admission and reassigned to inpatient status following Utilization Review, which may bias our results toward overestimation of the number of observation stays that converted to inpatient status.

The third limitation to our results relates to return visits. An accurate assessment of return visits is subject to the patient returning to the same hospital. If children do not return to the same hospital, our results would underestimate return visits and readmissions. In addition, we did not assess the reason for return visit as there was no way to verify if the return visit was truly related to the index visit without detailed chart review. Assuming children return to the same hospital for different reasons, our results would overestimate return visits associated with observation stays. We suspect that many 3‐day return visits result from the progression of acute illness or failure to respond to initial treatment, and 30‐day readmissions reflect recurrent hospital care needs related to chronic illnesses.

Lastly, severity classification is difficult when analyzing administrative datasets without physiologic patient data, and the SCS may not provide enough detail to reveal clinically important differences between patient groups.

CONCLUSIONS

Short‐stay hospitalizations following ED visits are common among children, and the majority of pediatric short‐stays are under observation status. Analyses of inpatient administrative databases that exclude observation stays likely result in an underestimation of hospital resource utilization for children. Efforts are needed to ensure that patients under observation status are accounted for in hospital administrative datasets used for pediatric health services research, and healthcare resource allocation, as it relates to hospital‐based care. While the clinical outcomes for observation patients appear favorable in terms of conversion to inpatient admissions and return visits, the financial implications of observation status care within children's hospitals are currently unknown.

In recent decades, hospital lengths of stay have decreased and there has been a shift toward outpatient management for many pediatric conditions. In 2003, one‐third of all children admitted to US hospitals experienced 1‐day inpatient stays, an increase from 19% in 1993.1 Some hospitals have developed dedicated observation units for the care of children, with select diagnoses, who are expected to respond to less than 24 hours of treatment.26 Expansion of observation services has been suggested as an approach to lessen emergency department (ED) crowding7 and alleviate high‐capacity conditions within hospital inpatient units.8

In contrast to care delivered in a dedicated observation unit, observation status is an administrative label applied to patients who do not meet inpatient criteria as defined by third parties such as InterQual. While the decision to admit a patient is ultimately at the discretion of the ordering physician, many hospitals use predetermined criteria to assign observation status to patients admitted to observation and inpatient units.9 Treatment provided under observation status is designated by hospitals and payers as outpatient care, even when delivered in an inpatient bed.10 As outpatient‐designated care, observation cases do not enter publicly available administrative datasets of hospital discharges that have traditionally been used to understand hospital resource utilization, including the National Hospital Discharge Survey and the Kid's Inpatient Database.11, 12

We hypothesize that there has been an increase in observation status care delivered to children in recent years, and that the majority of children under observation were discharged home without converting to inpatient status. To determine trends in pediatric observation status care, we conducted the first longitudinal, multicenter evaluation of observation status code utilization following ED treatment in a sample of US freestanding children's hospitals. In addition, we focused on the most recent year of data among top ranking diagnoses to assess the current state of observation status stay outcomes (including conversion to inpatient status and return visits).

METHODS

DATA ANALYSIS

For each of the 6 years of study, we calculated the following proportions to determine trends over time: 1) the number of Observation Status admissions from the ED as a proportion of the total number of ED visits resulting in Observation or Inpatient admission, and 2) the number of 1‐Day Stays admitted from the ED as a proportion of the total number of ED visits resulting in Observation or Inpatient admissions. Trends were analyzed using linear regression. Trends were also calculated for the total volume of admissions from the ED and the case‐mix index (CMI). CMI was assessed to evaluate for changes in the severity of illness for children admitted from the ED over the study period. Each hospital's CMI was calculated as an average of their Observation and Inpatient Status discharges' charge weights during the study period. Charge weights were calculated at the All Patient Refined Diagnosis Related Groups (APR‐DRG)/severity of illness level (3M Health Information Systems, St Paul, MN) and were normalized national average charges derived by Thomson‐Reuters from their Pediatric Projected National Database. Weights were then assigned to each discharge based on the discharge's APR‐DRG and severity level assignment.

To assess the current outcomes for observation, we analyzed stays with associated observation billing codes from the most recent year of available data (2009). Stays with Observation patient type were considered to have been discharged from observation, while those with an Inpatient Status patient type were considered to have converted to an inpatient admission during the observation period.

Using the 2009 data, we calculated descriptive statistics for patient characteristics (eg, age, gender, payer) comparing Observation Stays, 1‐Day Stays, and longer‐duration Inpatient admissions using chi‐square statistics. Age was categorized using the American Academy of Pediatrics groupings: <30 days, 30 days1 year, 12 years, 34 years, 512 years, 1317 years, >18 years. Designated payer was categorized into government, private, and other, including self‐pay and uninsured groups.

We used the Severity Classification Systems (SCS) developed for pediatric emergency care to estimate severity of illness for the visit.17 In this 5‐level system, each ICD‐9 diagnosis code is associated with a score related to the intensity of ED resources needed to care for a child with that diagnosis. In our analyses, each case was assigned the maximal SCS category based on the highest severity ICD‐9 code associated with the stay. Within the SCS, a score of 1 indicates minor illness (eg, diaper dermatitis) and 5 indicates major illness (eg, septic shock). The proportions of visits within categorical SCS scores were compared for Observation Stays, 1‐Day Stays, and longer‐duration Inpatient admissions using chi‐square statistics.

We determined the top 10 ranking diagnoses for which children were admitted from the ED in 2009 using the Diagnosis Grouping System (DGS).18 The DGS was designed specifically to categorize pediatric ED visits into clinically meaningful groups. The ICD‐9 code for the principal discharge diagnosis was used to assign records to 1 of the 77 DGS subgroups. Within each of the top ranking DGS subgroups, we determined the proportion of Observation Stays, 1‐Day Stays, and longer‐duration Inpatient admissions.

To provide clinically relevant outcomes of Observation Stays for common conditions, we selected stays with observation charges from within the top 10 ranking observation stay DGS subgroups in 2009. Outcomes for observation included: 1) immediate outcome of the observation stay (ie, discharge or conversion to inpatient status), 2) return visits to the ED in the 3 days following observation, and 3) readmissions to the hospital in the 3 and 30 days following observation. Bivariate comparisons of return visits and readmissions for Observation versus 1‐Day Stays within DGS subgroups were analyzed using chi‐square tests. Multivariate analyses of return visits and readmissions were conducted using Generalized Estimating Equations adjusting for severity of illness by SCS score and clustering by hospital. To account for local practice patterns, we also adjusted for a grouped treatment variable that included the site level proportion of children admitted to Observation Status, 1‐Day‐Stays, and longer Inpatient admissions. All statistical analyses were performed using SAS (version 9.2, SAS Institute, Inc, Cary, NC); P values <0.05 were considered statistically significant.

RESULTS

Trends in Short‐Stays

An increase in proportion of Observation Stays was mirrored by a decrease in proportion of 1‐Day Stays over the study period (Figure 1). In 2009, there were 1.4 times more Observation Stays than 1‐Day Stays (25,653 vs 18,425) compared with 14,242 and 20,747, respectively, in 2004. This shift toward more Observation Stays occurred as hospitals faced a 16% increase in the total number of admissions from the ED (91,318 to 108,217) and change in CMI from 1.48 to 1.51. Over the study period, roughly 40% of all admissions from the ED were Short‐Stays (Observation and 1‐Day Stays). Median LOS for Observation Status stays was 1 day (interquartile range [IQR]: 11).

Figure 1

DISCUSSION

In this first, multicenter longitudinal study of pediatric observation following an ED visit, we found that Observation Status code utilization has increased steadily over the past 6 years and, in 2007, the proportion of children admitted to observation status surpassed the proportion of children experiencing a 1‐day inpatient admission. Taken together, Short‐Stays made up more than 40% of the hospital‐based care delivered to children admitted from an ED. Stable trends in CMI over time suggest that observation status may be replacing inpatient status designated care for pediatric Short‐Stays in these hospitals. Our findings suggest the lines between outpatient observation and short‐stay inpatient care are becoming increasingly blurred. These trends have occurred in the setting of changing policies for hospital reimbursement, requirements for patients to meet criteria to qualify for inpatient admissions, and efforts to avoid stays deemed unnecessary or inappropriate by their brief duration.19 Therefore there is a growing need to understand the impact of children under observation on the structure, delivery, and financing of acute hospital care for children.

Our results also have implications for pediatric health services research that relies on hospital administrative databases that do not contain observation stays. Currently, observation stays are systematically excluded from many inpatient administrative datasets.11, 12 Analyses of datasets that do not account for observation stays likely result in underestimation of hospitalization rates and hospital resource utilization for children. This may be particularly important for high‐volume conditions, such as asthma and acute infections, for which children commonly require brief periods of hospital‐based care beyond an ED encounter. Data from pediatric observation status admissions should be consistently included in hospital administrative datasets to allow for more comprehensive analyses of hospital resource utilization among children.

Prior research has shown that the diagnoses commonly treated in pediatric observation units overlap with the diagnoses for which children experience 1‐Day Stays.1, 20 We found a similar pattern of conditions for which children were under Observation Status and 1‐Day Stays with comparable severity of illness between the groups in terms of SCS scores. Our findings imply a need to determine how and why hospitals differentiate Observation Status from 1‐Day‐Stay groups in order to improve the assignment of observation status. Assuming continued pressures from payers to provide more care in outpatient or observation settings, there is potential for expansion of dedicated observation services for children in the US. Without designated observation units or processes to group patients with lower severity conditions, there may be limited opportunities to realize more efficient hospital care simply through the application of the label of observation status.

For more than 30 years, observation services have been provided to children who require a period of monitoring to determine their response to therapy and the need for acute inpatient admission from the ED.21While we were not able to determine the location of care for observation status patients in this study, we know that few children's hospitals have dedicated observation units and, even when an observation unit is present, not all observation status patients are cared for in dedicated observation units.9 This, in essence, means that most children under observation status are cared for in virtual observation by inpatient teams using inpatient beds. If observation patients are treated in inpatient beds and consume the same resources as inpatients, then cost‐savings based on reimbursement contracts with payers may not reflect an actual reduction in services. Pediatric institutions will need to closely monitor the financial implications of observation status given the historical differences in payment for observation and inpatient care.

With more than 70% of children being discharged home following observation, our results are comparable to the published literature2, 5, 6, 22, 23 and guidelines for observation unit operations.24 Similar to prior studies,4, 15, 2530 our results also indicate that return visits and readmissions following observation are uncommon events. Our findings can serve as initial benchmarks for condition‐specific outcomes for pediatric observation care. Studies are needed both to identify the clinical characteristics predictive of successful discharge home from observation and to explore the hospital‐to‐hospital variability in outcomes for observation. Such studies are necessary to identify the most successful healthcare delivery models for pediatric observation stays.

LIMITATIONS

The primary limitation to our results is that data from a subset of freestanding children's hospitals may not reflect observation stays at other children's hospitals or the community hospitals that care for children across the US. Only 18 of 42 current PHIS member hospitals have provided both outpatient visit and inpatient stay data for each year of the study period and were considered eligible. In an effort to ensure the quality of observation stay data, we included the 16 hospitals that assigned observation charges to at least 90% of their observation status stays in the PHIS database. The exclusion of the 2 hospitals where <90% of observation status patients were assigned observation charges likely resulted in an underestimation of the utilization of observation status.

Second, there is potential for misclassification of patient type given institutional variations in the assignment of patient status. The PHIS database does not contain information about the factors that were considered in the assignment of observation status. At the time of admission from the ED, observation or inpatient status is assigned. While this decision is clearly reserved for the admitting physician, the process is not standardized across hospitals.9 Some institutions have Utilization Managers on site to help guide decision‐making, while others allow the assignment to be made by physicians without specific guidance. As a result, some patients may be assigned to observation status at admission and reassigned to inpatient status following Utilization Review, which may bias our results toward overestimation of the number of observation stays that converted to inpatient status.

The third limitation to our results relates to return visits. An accurate assessment of return visits is subject to the patient returning to the same hospital. If children do not return to the same hospital, our results would underestimate return visits and readmissions. In addition, we did not assess the reason for return visit as there was no way to verify if the return visit was truly related to the index visit without detailed chart review. Assuming children return to the same hospital for different reasons, our results would overestimate return visits associated with observation stays. We suspect that many 3‐day return visits result from the progression of acute illness or failure to respond to initial treatment, and 30‐day readmissions reflect recurrent hospital care needs related to chronic illnesses.

Lastly, severity classification is difficult when analyzing administrative datasets without physiologic patient data, and the SCS may not provide enough detail to reveal clinically important differences between patient groups.

CONCLUSIONS

Short‐stay hospitalizations following ED visits are common among children, and the majority of pediatric short‐stays are under observation status. Analyses of inpatient administrative databases that exclude observation stays likely result in an underestimation of hospital resource utilization for children. Efforts are needed to ensure that patients under observation status are accounted for in hospital administrative datasets used for pediatric health services research, and healthcare resource allocation, as it relates to hospital‐based care. While the clinical outcomes for observation patients appear favorable in terms of conversion to inpatient admissions and return visits, the financial implications of observation status care within children's hospitals are currently unknown.

In recent decades, hospital lengths of stay have decreased and there has been a shift toward outpatient management for many pediatric conditions. In 2003, one‐third of all children admitted to US hospitals experienced 1‐day inpatient stays, an increase from 19% in 1993.1 Some hospitals have developed dedicated observation units for the care of children, with select diagnoses, who are expected to respond to less than 24 hours of treatment.26 Expansion of observation services has been suggested as an approach to lessen emergency department (ED) crowding7 and alleviate high‐capacity conditions within hospital inpatient units.8

In contrast to care delivered in a dedicated observation unit, observation status is an administrative label applied to patients who do not meet inpatient criteria as defined by third parties such as InterQual. While the decision to admit a patient is ultimately at the discretion of the ordering physician, many hospitals use predetermined criteria to assign observation status to patients admitted to observation and inpatient units.9 Treatment provided under observation status is designated by hospitals and payers as outpatient care, even when delivered in an inpatient bed.10 As outpatient‐designated care, observation cases do not enter publicly available administrative datasets of hospital discharges that have traditionally been used to understand hospital resource utilization, including the National Hospital Discharge Survey and the Kid's Inpatient Database.11, 12

We hypothesize that there has been an increase in observation status care delivered to children in recent years, and that the majority of children under observation were discharged home without converting to inpatient status. To determine trends in pediatric observation status care, we conducted the first longitudinal, multicenter evaluation of observation status code utilization following ED treatment in a sample of US freestanding children's hospitals. In addition, we focused on the most recent year of data among top ranking diagnoses to assess the current state of observation status stay outcomes (including conversion to inpatient status and return visits).

METHODS

DATA ANALYSIS

For each of the 6 years of study, we calculated the following proportions to determine trends over time: 1) the number of Observation Status admissions from the ED as a proportion of the total number of ED visits resulting in Observation or Inpatient admission, and 2) the number of 1‐Day Stays admitted from the ED as a proportion of the total number of ED visits resulting in Observation or Inpatient admissions. Trends were analyzed using linear regression. Trends were also calculated for the total volume of admissions from the ED and the case‐mix index (CMI). CMI was assessed to evaluate for changes in the severity of illness for children admitted from the ED over the study period. Each hospital's CMI was calculated as an average of their Observation and Inpatient Status discharges' charge weights during the study period. Charge weights were calculated at the All Patient Refined Diagnosis Related Groups (APR‐DRG)/severity of illness level (3M Health Information Systems, St Paul, MN) and were normalized national average charges derived by Thomson‐Reuters from their Pediatric Projected National Database. Weights were then assigned to each discharge based on the discharge's APR‐DRG and severity level assignment.

To assess the current outcomes for observation, we analyzed stays with associated observation billing codes from the most recent year of available data (2009). Stays with Observation patient type were considered to have been discharged from observation, while those with an Inpatient Status patient type were considered to have converted to an inpatient admission during the observation period.

Using the 2009 data, we calculated descriptive statistics for patient characteristics (eg, age, gender, payer) comparing Observation Stays, 1‐Day Stays, and longer‐duration Inpatient admissions using chi‐square statistics. Age was categorized using the American Academy of Pediatrics groupings: <30 days, 30 days1 year, 12 years, 34 years, 512 years, 1317 years, >18 years. Designated payer was categorized into government, private, and other, including self‐pay and uninsured groups.

We used the Severity Classification Systems (SCS) developed for pediatric emergency care to estimate severity of illness for the visit.17 In this 5‐level system, each ICD‐9 diagnosis code is associated with a score related to the intensity of ED resources needed to care for a child with that diagnosis. In our analyses, each case was assigned the maximal SCS category based on the highest severity ICD‐9 code associated with the stay. Within the SCS, a score of 1 indicates minor illness (eg, diaper dermatitis) and 5 indicates major illness (eg, septic shock). The proportions of visits within categorical SCS scores were compared for Observation Stays, 1‐Day Stays, and longer‐duration Inpatient admissions using chi‐square statistics.

We determined the top 10 ranking diagnoses for which children were admitted from the ED in 2009 using the Diagnosis Grouping System (DGS).18 The DGS was designed specifically to categorize pediatric ED visits into clinically meaningful groups. The ICD‐9 code for the principal discharge diagnosis was used to assign records to 1 of the 77 DGS subgroups. Within each of the top ranking DGS subgroups, we determined the proportion of Observation Stays, 1‐Day Stays, and longer‐duration Inpatient admissions.

To provide clinically relevant outcomes of Observation Stays for common conditions, we selected stays with observation charges from within the top 10 ranking observation stay DGS subgroups in 2009. Outcomes for observation included: 1) immediate outcome of the observation stay (ie, discharge or conversion to inpatient status), 2) return visits to the ED in the 3 days following observation, and 3) readmissions to the hospital in the 3 and 30 days following observation. Bivariate comparisons of return visits and readmissions for Observation versus 1‐Day Stays within DGS subgroups were analyzed using chi‐square tests. Multivariate analyses of return visits and readmissions were conducted using Generalized Estimating Equations adjusting for severity of illness by SCS score and clustering by hospital. To account for local practice patterns, we also adjusted for a grouped treatment variable that included the site level proportion of children admitted to Observation Status, 1‐Day‐Stays, and longer Inpatient admissions. All statistical analyses were performed using SAS (version 9.2, SAS Institute, Inc, Cary, NC); P values <0.05 were considered statistically significant.

RESULTS

Trends in Short‐Stays

An increase in proportion of Observation Stays was mirrored by a decrease in proportion of 1‐Day Stays over the study period (Figure 1). In 2009, there were 1.4 times more Observation Stays than 1‐Day Stays (25,653 vs 18,425) compared with 14,242 and 20,747, respectively, in 2004. This shift toward more Observation Stays occurred as hospitals faced a 16% increase in the total number of admissions from the ED (91,318 to 108,217) and change in CMI from 1.48 to 1.51. Over the study period, roughly 40% of all admissions from the ED were Short‐Stays (Observation and 1‐Day Stays). Median LOS for Observation Status stays was 1 day (interquartile range [IQR]: 11).

Figure 1

DISCUSSION

In this first, multicenter longitudinal study of pediatric observation following an ED visit, we found that Observation Status code utilization has increased steadily over the past 6 years and, in 2007, the proportion of children admitted to observation status surpassed the proportion of children experiencing a 1‐day inpatient admission. Taken together, Short‐Stays made up more than 40% of the hospital‐based care delivered to children admitted from an ED. Stable trends in CMI over time suggest that observation status may be replacing inpatient status designated care for pediatric Short‐Stays in these hospitals. Our findings suggest the lines between outpatient observation and short‐stay inpatient care are becoming increasingly blurred. These trends have occurred in the setting of changing policies for hospital reimbursement, requirements for patients to meet criteria to qualify for inpatient admissions, and efforts to avoid stays deemed unnecessary or inappropriate by their brief duration.19 Therefore there is a growing need to understand the impact of children under observation on the structure, delivery, and financing of acute hospital care for children.

Our results also have implications for pediatric health services research that relies on hospital administrative databases that do not contain observation stays. Currently, observation stays are systematically excluded from many inpatient administrative datasets.11, 12 Analyses of datasets that do not account for observation stays likely result in underestimation of hospitalization rates and hospital resource utilization for children. This may be particularly important for high‐volume conditions, such as asthma and acute infections, for which children commonly require brief periods of hospital‐based care beyond an ED encounter. Data from pediatric observation status admissions should be consistently included in hospital administrative datasets to allow for more comprehensive analyses of hospital resource utilization among children.

Prior research has shown that the diagnoses commonly treated in pediatric observation units overlap with the diagnoses for which children experience 1‐Day Stays.1, 20 We found a similar pattern of conditions for which children were under Observation Status and 1‐Day Stays with comparable severity of illness between the groups in terms of SCS scores. Our findings imply a need to determine how and why hospitals differentiate Observation Status from 1‐Day‐Stay groups in order to improve the assignment of observation status. Assuming continued pressures from payers to provide more care in outpatient or observation settings, there is potential for expansion of dedicated observation services for children in the US. Without designated observation units or processes to group patients with lower severity conditions, there may be limited opportunities to realize more efficient hospital care simply through the application of the label of observation status.

For more than 30 years, observation services have been provided to children who require a period of monitoring to determine their response to therapy and the need for acute inpatient admission from the ED.21While we were not able to determine the location of care for observation status patients in this study, we know that few children's hospitals have dedicated observation units and, even when an observation unit is present, not all observation status patients are cared for in dedicated observation units.9 This, in essence, means that most children under observation status are cared for in virtual observation by inpatient teams using inpatient beds. If observation patients are treated in inpatient beds and consume the same resources as inpatients, then cost‐savings based on reimbursement contracts with payers may not reflect an actual reduction in services. Pediatric institutions will need to closely monitor the financial implications of observation status given the historical differences in payment for observation and inpatient care.

With more than 70% of children being discharged home following observation, our results are comparable to the published literature2, 5, 6, 22, 23 and guidelines for observation unit operations.24 Similar to prior studies,4, 15, 2530 our results also indicate that return visits and readmissions following observation are uncommon events. Our findings can serve as initial benchmarks for condition‐specific outcomes for pediatric observation care. Studies are needed both to identify the clinical characteristics predictive of successful discharge home from observation and to explore the hospital‐to‐hospital variability in outcomes for observation. Such studies are necessary to identify the most successful healthcare delivery models for pediatric observation stays.

LIMITATIONS

The primary limitation to our results is that data from a subset of freestanding children's hospitals may not reflect observation stays at other children's hospitals or the community hospitals that care for children across the US. Only 18 of 42 current PHIS member hospitals have provided both outpatient visit and inpatient stay data for each year of the study period and were considered eligible. In an effort to ensure the quality of observation stay data, we included the 16 hospitals that assigned observation charges to at least 90% of their observation status stays in the PHIS database. The exclusion of the 2 hospitals where <90% of observation status patients were assigned observation charges likely resulted in an underestimation of the utilization of observation status.

Second, there is potential for misclassification of patient type given institutional variations in the assignment of patient status. The PHIS database does not contain information about the factors that were considered in the assignment of observation status. At the time of admission from the ED, observation or inpatient status is assigned. While this decision is clearly reserved for the admitting physician, the process is not standardized across hospitals.9 Some institutions have Utilization Managers on site to help guide decision‐making, while others allow the assignment to be made by physicians without specific guidance. As a result, some patients may be assigned to observation status at admission and reassigned to inpatient status following Utilization Review, which may bias our results toward overestimation of the number of observation stays that converted to inpatient status.

The third limitation to our results relates to return visits. An accurate assessment of return visits is subject to the patient returning to the same hospital. If children do not return to the same hospital, our results would underestimate return visits and readmissions. In addition, we did not assess the reason for return visit as there was no way to verify if the return visit was truly related to the index visit without detailed chart review. Assuming children return to the same hospital for different reasons, our results would overestimate return visits associated with observation stays. We suspect that many 3‐day return visits result from the progression of acute illness or failure to respond to initial treatment, and 30‐day readmissions reflect recurrent hospital care needs related to chronic illnesses.

Lastly, severity classification is difficult when analyzing administrative datasets without physiologic patient data, and the SCS may not provide enough detail to reveal clinically important differences between patient groups.

CONCLUSIONS

Short‐stay hospitalizations following ED visits are common among children, and the majority of pediatric short‐stays are under observation status. Analyses of inpatient administrative databases that exclude observation stays likely result in an underestimation of hospital resource utilization for children. Efforts are needed to ensure that patients under observation status are accounted for in hospital administrative datasets used for pediatric health services research, and healthcare resource allocation, as it relates to hospital‐based care. While the clinical outcomes for observation patients appear favorable in terms of conversion to inpatient admissions and return visits, the financial implications of observation status care within children's hospitals are currently unknown.

Traumatic lumbar puncture (LP) occurs when peripheral blood is introduced into the cerebrospinal fluid (CSF) as a result of needle trauma, which causes bleeding into the subarachnoid space. Traumatic LPs occur in up to 30% of LPs performed in children.1, 2 In addition to affecting the CSF white blood cell count, the presence of CSF red blood cells (RBCs) is associated with higher CSF protein concentrations due to the higher protein concentration in plasma compared with CSF and to the release of protein from lysed red blood cells. CSF protein concentration has been used in clinical decision rules for the prediction of bacterial meningitis in children.3 Elevated protein levels are difficult to interpret in cases of traumatic LP, and a diagnosis of bacterial meningitis may be more difficult to exclude on the basis of CSF test results.4

The interpretation of CSF protein levels is further complicated in the youngest infants due to both the changing composition of the CSF as well as the higher rates of traumatic LPs.5 Therefore, studies establishing a correction factor, adjusting observed CSF protein levels for the presence of CSF RBCs, that included predominantly older children may not be generalizable to neonates and young infants.6 We sought to determine the relationship between CSF RBC count and CSF protein in infants 56 days of age who underwent LP in the emergency department (ED).

METHODS

RESULTS

There were 1986 infants, 56 days of age or younger, who underwent LP in the ED during the study period. Patients were excluded for the following reasons: missing medical record number (n = 16); missing CSF WBC, CSF RBC, or CSF protein values (n = 290); conditions known to elevate CSF protein concentrations (n = 426, as follows: presence of a ventricular shunt device [n = 48], serious bacterial infection [n = 149], congenital infection [n = 2], positive CSF polymerase chain reaction [PCR] test for either enterovirus or herpes simplex virus [n = 97], seizure prior to presentation [n = 98], or elevated serum bilirubin [n = 32]). An additional 13 patients with a CSF RBC count >150,000 cells/mm³ were also excluded.

For the remaining 1241 study infants, the median age was 34 days (interquartile range: 19 days‐46 days) and 554 patients (45%) were male. The median CSF RBC count was 40 cells/mm³ (interquartile range: 2‐1080 cells/mm³); 11.8% of patients had a CSF RBC count >10,000 cells/mm³.

CSF protein increased linearly with increasing CSF RBCs (Figure 1). The increase in the CSF protein concentration of 1.9 mg/dL per 1000 CSF RBCs for all patients was similar between different age groups and delivery types (Table 1). Restricting analysis to those patients without pleocytosis also yielded comparable results; applying 2 other definitions of pleocytosis did not change the magnitude of the association (Table 1).

Figure 1

In a subanalysis, we then included subjects with a CSF RBC count >150,000/mm³; one extreme outlier with a CSF RBC equal to 3,160,000/mm³ remained excluded. Inclusion of more traumatic samples lessened the overall correction factor. The CSF protein increased by 1.22 mg/dL (95% confidence interval: 1.14‐1.29 mg/dL) per 1000 RBC/mm³ increase in the CSF. In the subset without CSF pleocytosis, the CSF protein increased by 1.44 mg/dL (95% confidence interval: 1.33‐1.57 mg/dL) per 1000 RBC/mm³.

Three children had high CSF protein values (>500 mg/dL) despite the relative paucity of CSF RBCs. Two of these infants had respiratory syncytial virus bronchiolitis; neither infant had signs or symptoms of neurological illness. While details of the labor and delivery were not available, the CSF sample for one of these infants was reported to have xanthochromia, and the other infant was reported to have had a traumatic LP with a CSF sample that subsequently cleared. The third infant had fever without a specific source identified, but had a birth history of vaginal delivery and prolonged labor. The CSF sample from LP for this patient was reported as grossly bloody by the performing clinicians and by the Clinical Microbiology Laboratory, despite a CSF red blood cell count of only 5500 cells/mm³.

DISCUSSION

In a large cohort of infants 56 days of age, CSF protein increased by approximately 2 mg/dL for every 1000 cell/mm³ increase in CSF RBCs. This correction factor is higher than previously reported correction factors from studies including older infants and children.6, 18 Some of this difference may be explained by the presence of old blood related to the trauma of labor and delivery. Previous work has demonstrated that the presence of xanthochromia, another RBC breakdown product, in the CSF of young infants was associated with maternal labor and elevated CSF protein.19 Consistent with this hypothesis, the correction factor was nominally higher in those infants born by vaginal delivery compared with those born by cesarean section.

Several infants in our study had high CSF protein levels despite a paucity of CSF RBCs. By convention at our institution, the protein and glucose values are determined from the second tube, and the WBCs and RBCs are determined from the third tube. However, we could not determine the order in which the specimens for protein and RBCs were collected for individual specimens. Additionally, it is possible that delayed clearance of blood from a traumatic LP would cause the CSF protein level to be high, as measured in the second tube, but lead to few RBCs in the third tube. These circumstances could explain the discrepancy between CSF protein and CSF RBCs counts for some patients.

The CSF protein adjustment factor for infants 56 days of age in our study was almost twice the correction of 1.1 mg/dL for every 1000 RBC increase reported by Nigrovic et al among infants 90 days of age.6 There are differences in the design of the 2 studies. We excluded subjects with exceedingly large numbers of CSF RBCs and restricted inclusion to those 56 days of age or younger. When subjects with >150,000 RBCs/mm³ were included, the correction decreased to a value comparable to that reported by Nigrovic et al.6 Therefore, it is possible that inclusion of subjects with grossly bloody specimens in prior studies skewed the association between CSF protein and CSF RBCs. The number of subjects in our cohort with >150,000 CSF RBCs was too small to calculate a relevant correction factor for infants with exceedingly high CSF RBC counts.

The results of this study should be considered in the context of several limitations. Details regarding labor and delivery were not available. We suspect that old blood related to the trauma of birth provides partial explanation for the higher correction factor in neonates and young infants compared with older children. However, differences in CSF blood‐brain barrier permeability may also contribute to these differences, independent of the CSF RBC count. Additionally, though the study population included a large number of neonates and young infants, a relatively small proportion of subjects had high CSF RBC counts. Therefore, our results may not be generalizable to those with exceedingly high CSF RBCs. Finally, available clinical prediction rules to identify patients with CSF pleocytosis, who are at very low risk for bacterial meningitis, include CSF protein as a predictor.3, 20, 21 Although CSF protein in children with traumatic LPs may need adjustment prior to application of the clinical prediction rule, further study is needed before implementing this approach.

In conclusion, we found that CSF protein concentrations increased by approximately 2 mg/dL for every 1000 CSF RBCs. Correction of CSF protein for those with extremely high CSF RBCs may not be appropriate, as conventional linear models do not apply. These data may assist clinicians in interpreting CSF protein concentrations in infants 56 days of age and younger in the context of traumatic LPs.

Traumatic lumbar puncture (LP) occurs when peripheral blood is introduced into the cerebrospinal fluid (CSF) as a result of needle trauma, which causes bleeding into the subarachnoid space. Traumatic LPs occur in up to 30% of LPs performed in children.1, 2 In addition to affecting the CSF white blood cell count, the presence of CSF red blood cells (RBCs) is associated with higher CSF protein concentrations due to the higher protein concentration in plasma compared with CSF and to the release of protein from lysed red blood cells. CSF protein concentration has been used in clinical decision rules for the prediction of bacterial meningitis in children.3 Elevated protein levels are difficult to interpret in cases of traumatic LP, and a diagnosis of bacterial meningitis may be more difficult to exclude on the basis of CSF test results.4

The interpretation of CSF protein levels is further complicated in the youngest infants due to both the changing composition of the CSF as well as the higher rates of traumatic LPs.5 Therefore, studies establishing a correction factor, adjusting observed CSF protein levels for the presence of CSF RBCs, that included predominantly older children may not be generalizable to neonates and young infants.6 We sought to determine the relationship between CSF RBC count and CSF protein in infants 56 days of age who underwent LP in the emergency department (ED).

METHODS

RESULTS

There were 1986 infants, 56 days of age or younger, who underwent LP in the ED during the study period. Patients were excluded for the following reasons: missing medical record number (n = 16); missing CSF WBC, CSF RBC, or CSF protein values (n = 290); conditions known to elevate CSF protein concentrations (n = 426, as follows: presence of a ventricular shunt device [n = 48], serious bacterial infection [n = 149], congenital infection [n = 2], positive CSF polymerase chain reaction [PCR] test for either enterovirus or herpes simplex virus [n = 97], seizure prior to presentation [n = 98], or elevated serum bilirubin [n = 32]). An additional 13 patients with a CSF RBC count >150,000 cells/mm³ were also excluded.

For the remaining 1241 study infants, the median age was 34 days (interquartile range: 19 days‐46 days) and 554 patients (45%) were male. The median CSF RBC count was 40 cells/mm³ (interquartile range: 2‐1080 cells/mm³); 11.8% of patients had a CSF RBC count >10,000 cells/mm³.

CSF protein increased linearly with increasing CSF RBCs (Figure 1). The increase in the CSF protein concentration of 1.9 mg/dL per 1000 CSF RBCs for all patients was similar between different age groups and delivery types (Table 1). Restricting analysis to those patients without pleocytosis also yielded comparable results; applying 2 other definitions of pleocytosis did not change the magnitude of the association (Table 1).

Figure 1

In a subanalysis, we then included subjects with a CSF RBC count >150,000/mm³; one extreme outlier with a CSF RBC equal to 3,160,000/mm³ remained excluded. Inclusion of more traumatic samples lessened the overall correction factor. The CSF protein increased by 1.22 mg/dL (95% confidence interval: 1.14‐1.29 mg/dL) per 1000 RBC/mm³ increase in the CSF. In the subset without CSF pleocytosis, the CSF protein increased by 1.44 mg/dL (95% confidence interval: 1.33‐1.57 mg/dL) per 1000 RBC/mm³.

Three children had high CSF protein values (>500 mg/dL) despite the relative paucity of CSF RBCs. Two of these infants had respiratory syncytial virus bronchiolitis; neither infant had signs or symptoms of neurological illness. While details of the labor and delivery were not available, the CSF sample for one of these infants was reported to have xanthochromia, and the other infant was reported to have had a traumatic LP with a CSF sample that subsequently cleared. The third infant had fever without a specific source identified, but had a birth history of vaginal delivery and prolonged labor. The CSF sample from LP for this patient was reported as grossly bloody by the performing clinicians and by the Clinical Microbiology Laboratory, despite a CSF red blood cell count of only 5500 cells/mm³.

DISCUSSION

In a large cohort of infants 56 days of age, CSF protein increased by approximately 2 mg/dL for every 1000 cell/mm³ increase in CSF RBCs. This correction factor is higher than previously reported correction factors from studies including older infants and children.6, 18 Some of this difference may be explained by the presence of old blood related to the trauma of labor and delivery. Previous work has demonstrated that the presence of xanthochromia, another RBC breakdown product, in the CSF of young infants was associated with maternal labor and elevated CSF protein.19 Consistent with this hypothesis, the correction factor was nominally higher in those infants born by vaginal delivery compared with those born by cesarean section.

Several infants in our study had high CSF protein levels despite a paucity of CSF RBCs. By convention at our institution, the protein and glucose values are determined from the second tube, and the WBCs and RBCs are determined from the third tube. However, we could not determine the order in which the specimens for protein and RBCs were collected for individual specimens. Additionally, it is possible that delayed clearance of blood from a traumatic LP would cause the CSF protein level to be high, as measured in the second tube, but lead to few RBCs in the third tube. These circumstances could explain the discrepancy between CSF protein and CSF RBCs counts for some patients.

The CSF protein adjustment factor for infants 56 days of age in our study was almost twice the correction of 1.1 mg/dL for every 1000 RBC increase reported by Nigrovic et al among infants 90 days of age.6 There are differences in the design of the 2 studies. We excluded subjects with exceedingly large numbers of CSF RBCs and restricted inclusion to those 56 days of age or younger. When subjects with >150,000 RBCs/mm³ were included, the correction decreased to a value comparable to that reported by Nigrovic et al.6 Therefore, it is possible that inclusion of subjects with grossly bloody specimens in prior studies skewed the association between CSF protein and CSF RBCs. The number of subjects in our cohort with >150,000 CSF RBCs was too small to calculate a relevant correction factor for infants with exceedingly high CSF RBC counts.

The results of this study should be considered in the context of several limitations. Details regarding labor and delivery were not available. We suspect that old blood related to the trauma of birth provides partial explanation for the higher correction factor in neonates and young infants compared with older children. However, differences in CSF blood‐brain barrier permeability may also contribute to these differences, independent of the CSF RBC count. Additionally, though the study population included a large number of neonates and young infants, a relatively small proportion of subjects had high CSF RBC counts. Therefore, our results may not be generalizable to those with exceedingly high CSF RBCs. Finally, available clinical prediction rules to identify patients with CSF pleocytosis, who are at very low risk for bacterial meningitis, include CSF protein as a predictor.3, 20, 21 Although CSF protein in children with traumatic LPs may need adjustment prior to application of the clinical prediction rule, further study is needed before implementing this approach.

In conclusion, we found that CSF protein concentrations increased by approximately 2 mg/dL for every 1000 CSF RBCs. Correction of CSF protein for those with extremely high CSF RBCs may not be appropriate, as conventional linear models do not apply. These data may assist clinicians in interpreting CSF protein concentrations in infants 56 days of age and younger in the context of traumatic LPs.

Traumatic lumbar puncture (LP) occurs when peripheral blood is introduced into the cerebrospinal fluid (CSF) as a result of needle trauma, which causes bleeding into the subarachnoid space. Traumatic LPs occur in up to 30% of LPs performed in children.1, 2 In addition to affecting the CSF white blood cell count, the presence of CSF red blood cells (RBCs) is associated with higher CSF protein concentrations due to the higher protein concentration in plasma compared with CSF and to the release of protein from lysed red blood cells. CSF protein concentration has been used in clinical decision rules for the prediction of bacterial meningitis in children.3 Elevated protein levels are difficult to interpret in cases of traumatic LP, and a diagnosis of bacterial meningitis may be more difficult to exclude on the basis of CSF test results.4

The interpretation of CSF protein levels is further complicated in the youngest infants due to both the changing composition of the CSF as well as the higher rates of traumatic LPs.5 Therefore, studies establishing a correction factor, adjusting observed CSF protein levels for the presence of CSF RBCs, that included predominantly older children may not be generalizable to neonates and young infants.6 We sought to determine the relationship between CSF RBC count and CSF protein in infants 56 days of age who underwent LP in the emergency department (ED).

METHODS

RESULTS

There were 1986 infants, 56 days of age or younger, who underwent LP in the ED during the study period. Patients were excluded for the following reasons: missing medical record number (n = 16); missing CSF WBC, CSF RBC, or CSF protein values (n = 290); conditions known to elevate CSF protein concentrations (n = 426, as follows: presence of a ventricular shunt device [n = 48], serious bacterial infection [n = 149], congenital infection [n = 2], positive CSF polymerase chain reaction [PCR] test for either enterovirus or herpes simplex virus [n = 97], seizure prior to presentation [n = 98], or elevated serum bilirubin [n = 32]). An additional 13 patients with a CSF RBC count >150,000 cells/mm³ were also excluded.

For the remaining 1241 study infants, the median age was 34 days (interquartile range: 19 days‐46 days) and 554 patients (45%) were male. The median CSF RBC count was 40 cells/mm³ (interquartile range: 2‐1080 cells/mm³); 11.8% of patients had a CSF RBC count >10,000 cells/mm³.

CSF protein increased linearly with increasing CSF RBCs (Figure 1). The increase in the CSF protein concentration of 1.9 mg/dL per 1000 CSF RBCs for all patients was similar between different age groups and delivery types (Table 1). Restricting analysis to those patients without pleocytosis also yielded comparable results; applying 2 other definitions of pleocytosis did not change the magnitude of the association (Table 1).

Figure 1

In a subanalysis, we then included subjects with a CSF RBC count >150,000/mm³; one extreme outlier with a CSF RBC equal to 3,160,000/mm³ remained excluded. Inclusion of more traumatic samples lessened the overall correction factor. The CSF protein increased by 1.22 mg/dL (95% confidence interval: 1.14‐1.29 mg/dL) per 1000 RBC/mm³ increase in the CSF. In the subset without CSF pleocytosis, the CSF protein increased by 1.44 mg/dL (95% confidence interval: 1.33‐1.57 mg/dL) per 1000 RBC/mm³.

Three children had high CSF protein values (>500 mg/dL) despite the relative paucity of CSF RBCs. Two of these infants had respiratory syncytial virus bronchiolitis; neither infant had signs or symptoms of neurological illness. While details of the labor and delivery were not available, the CSF sample for one of these infants was reported to have xanthochromia, and the other infant was reported to have had a traumatic LP with a CSF sample that subsequently cleared. The third infant had fever without a specific source identified, but had a birth history of vaginal delivery and prolonged labor. The CSF sample from LP for this patient was reported as grossly bloody by the performing clinicians and by the Clinical Microbiology Laboratory, despite a CSF red blood cell count of only 5500 cells/mm³.

DISCUSSION

In a large cohort of infants 56 days of age, CSF protein increased by approximately 2 mg/dL for every 1000 cell/mm³ increase in CSF RBCs. This correction factor is higher than previously reported correction factors from studies including older infants and children.6, 18 Some of this difference may be explained by the presence of old blood related to the trauma of labor and delivery. Previous work has demonstrated that the presence of xanthochromia, another RBC breakdown product, in the CSF of young infants was associated with maternal labor and elevated CSF protein.19 Consistent with this hypothesis, the correction factor was nominally higher in those infants born by vaginal delivery compared with those born by cesarean section.

Several infants in our study had high CSF protein levels despite a paucity of CSF RBCs. By convention at our institution, the protein and glucose values are determined from the second tube, and the WBCs and RBCs are determined from the third tube. However, we could not determine the order in which the specimens for protein and RBCs were collected for individual specimens. Additionally, it is possible that delayed clearance of blood from a traumatic LP would cause the CSF protein level to be high, as measured in the second tube, but lead to few RBCs in the third tube. These circumstances could explain the discrepancy between CSF protein and CSF RBCs counts for some patients.

The CSF protein adjustment factor for infants 56 days of age in our study was almost twice the correction of 1.1 mg/dL for every 1000 RBC increase reported by Nigrovic et al among infants 90 days of age.6 There are differences in the design of the 2 studies. We excluded subjects with exceedingly large numbers of CSF RBCs and restricted inclusion to those 56 days of age or younger. When subjects with >150,000 RBCs/mm³ were included, the correction decreased to a value comparable to that reported by Nigrovic et al.6 Therefore, it is possible that inclusion of subjects with grossly bloody specimens in prior studies skewed the association between CSF protein and CSF RBCs. The number of subjects in our cohort with >150,000 CSF RBCs was too small to calculate a relevant correction factor for infants with exceedingly high CSF RBC counts.

The results of this study should be considered in the context of several limitations. Details regarding labor and delivery were not available. We suspect that old blood related to the trauma of birth provides partial explanation for the higher correction factor in neonates and young infants compared with older children. However, differences in CSF blood‐brain barrier permeability may also contribute to these differences, independent of the CSF RBC count. Additionally, though the study population included a large number of neonates and young infants, a relatively small proportion of subjects had high CSF RBC counts. Therefore, our results may not be generalizable to those with exceedingly high CSF RBCs. Finally, available clinical prediction rules to identify patients with CSF pleocytosis, who are at very low risk for bacterial meningitis, include CSF protein as a predictor.3, 20, 21 Although CSF protein in children with traumatic LPs may need adjustment prior to application of the clinical prediction rule, further study is needed before implementing this approach.

In conclusion, we found that CSF protein concentrations increased by approximately 2 mg/dL for every 1000 CSF RBCs. Correction of CSF protein for those with extremely high CSF RBCs may not be appropriate, as conventional linear models do not apply. These data may assist clinicians in interpreting CSF protein concentrations in infants 56 days of age and younger in the context of traumatic LPs.

Mycoplasma pneumoniae is a common cause of community‐acquired pneumonia (CAP), among school‐age children and adolescents.14 Though pneumonia caused by M. pneumoniae is typically self‐limited, severe illness may occur.5 M. pneumoniae has also been implicated in airway inflammation, which may lead to the onset and development of chronic pulmonary disease.610 Few studies have directly addressed appropriate treatment strategies for M. pneumoniae pneumonia,11 and, despite its high prevalence and potential for causing severe complications, treatment recommendations remain inconsistent.

The efficacy of macrolide therapy in particular for M. pneumoniae remains unclear. In vitro susceptibility studies have shown bacteriostatic activity of erythromycin, clarithromycin, and azithromycin against M. pneumoniae.1218 Additionally, several small retrospective studies have shown that among children with atypical CAP (including M. pneumoniae pneumonia), those treated with macrolides were less likely to have persistence or progression of signs and symptoms after 3 days of therapy.19, 20 Lu et al21 found a shorter duration of fever among macrolide recipients compared with non‐recipients. In adults, Shames et al22 found a shorter duration of fever and hospitalization among erythromycin recipients compared with controls. Other randomized controlled trials have also addressed the use of macrolides in treatment of M. pneumoniae, but the ability to draw meaningful conclusions is limited by small samples sizes and by lack of details about the number of patients with M. pneumoniae.11

In addition to their antimicrobial effect, macrolides also have anti‐inflammatory properties.2327 The importance of these anti‐inflammatory properties is supported by studies showing clinical cure in patients treated with macrolides despite persistence of M. pneumoniae organisms,2831 clinical improvement despite the administration of doses that provide tissue levels below the minimum inhibitory concentration of the organism,3234 and clinical cure in patients with macrolide‐resistant M. pneumoniae.18, 35

The objectives of the current study were to examine the impact of macrolide therapy on the length of stay (LOS) and short‐ and longer‐term readmissions, including longer‐term asthma‐related readmissions, in children hospitalized with M. pneumoniae pneumonia.

METHODS

RESULTS

DISCUSSION

This multicenter study examined the role of macrolide therapy in children hospitalized with M. pneumoniae pneumonia. Empiric macrolide therapy was associated with an approximately 30% shorter hospital LOS and, in stratified analysis, remained associated with a significantly shorter hospital LOS in both systemic corticosteroid recipients and non‐recipients. Empiric macrolide therapy was not associated with short‐ or longer‐term hospital readmission.

Previous small randomized trials have been inconclusive regarding the potential benefit of macrolide therapy in M. pneumoniae pneumonia.11 Our study, which demonstrated a shorter LOS among macrolides recipients compared with non‐recipients, has several advantages over prior studies including a substantively larger sample size and multicenter design. Animal models support our observations regarding the potential beneficial antimicrobial role of macrolides. M. pneumoniae concentrations in bronchoalveolar lavage specimens were significantly lower among experimentally infected mice treated with clarithromycin, a macolide‐class antibiotic, compared with either placebo or dexamethasone.45 Combination therapy with clarithromycin and dexamethasone reduced histopathologic inflammation to a greater degree than dexamethasone alone.45

While the relative importance of the antimicrobial and anti‐inflammatory properties of macrolides is not known, observational studies of children infected with macrolide‐resistant M. pneumoniae suggest that the antimicrobial properties of macrolides may provide disproportionate clinical benefit. The duration of fever in macrolide recipients with macrolide‐resistant M. pneumoniae (median duration, 9 days) reported by Suzuki et al46 was significantly longer than those with macrolide‐susceptible infections (median duration, 5 days), and similar to the duration of fever in patients with M. pneumoniae infection treated with placebo (median duration, 8 days) reported by Kingston et al.47 Additionally, macrolide therapy was associated with significant improvements in lung function in patients with asthma and concomitant M. pneumoniae infection, but not in patients with asthma without documented M. pneumoniae infection.9 As corticosteroids also have anti‐inflammatory properties, we expect that any anti‐inflammatory benefit of macrolide therapy would be mitigated by the concomitant administration of corticosteroids. The shorter LOS associated with empiric macrolide therapy in our study was comparable among corticosteroid recipients and non‐recipients.

Atypical bacterial pathogens, including M. pneumoniae, are associated with diffuse lower airway inflammation6, 48 and airway hyperresponsiveness,6 and have been implicated as a cause of acute asthma exacerbations.7, 4954 Among patients with previously diagnosed asthma, acute M. pneumoniae infection was identified in up to 20% of those having acute exacerbations.7, 54 Macrolide therapy has a beneficial effect on lung function and airway hyperresponsiveness in adults with asthma.9, 55 Among mice infected with M. pneumoniae, 3 days of macrolide therapy resulted in a significant reduction in airway hyperresponsiveness compared with placebo or dexamethasone; however, after 6 days of therapy, there was no significant difference in airway hyperresponsiveness between those receiving macrolides, dexamethasone, or placebo, suggesting that the benefit of macrolides on airway hyperresponsiveness may be brief. Our findings of a shorter LOS but no difference in readmissions at 28 days or longer, for macrolide recipients compared with non‐recipients, support the limited benefit of macrolide therapy beyond the initial reduction in bacterial load seen in the first few days of therapy.

M. pneumoniae infection has also been implicated as a cause of chronic pulmonary disease, including asthma.610 In the mouse model, peribronchial and perivascular mononuclear infiltrates, increased airway methacholine reactivity, and increased airway obstruction were observed 530 days after M. pneumoniae inoculation.6 M. pneumoniae has been identified in 26 (50%) of 51 children experiencing their first asthma attack,7 and 23 (42%) of 55 adults with chronic, stable asthma.9 Nevertheless, results of other studies addressing the issue are inconsistent, and the role of M. pneumoniae in the development of asthma remains unclear.56 In order to investigate the impact of macrolide therapy on the development of chronic pulmonary disease requiring hospitalization, we examined the readmission rates in the 15 months following index discharge. The proportion of children hospitalized with asthma following the hospitalization for M. pneumoniae pneumonia was higher for both macrolide recipients and non‐recipients compared with the 24‐months prior to infection. These results support a possible role for M. pneumoniae in chronic pulmonary disease. However, macrolide therapy was not associated with long‐term overall hospital readmission or long‐term asthma readmission, either in the entire cohort or in the subset of patients without prior asthma hospitalizations or medications.

This study had several limitations. First, because the identification of children with M. pneumoniae pneumonia relied on ICD‐9 discharge diagnosis codes, it is possible that there was misclassification of disease. We minimized the inclusion of children without M. pneumoniae by including only children who received antibiotic therapy on the first day of hospitalization and by excluding patients younger than 6 years of age, a group at relatively low‐risk for M. pneumoniae infection. Further, our algorithm for identification of M. pneumoniae pneumonia was validated through review of the medical records at 1 institution and was found to have a high positive predictive value. However, the positive predictive value of these ICD‐9 codes may vary across institutions. Additionally, the sensitivity of ICD‐9 codes for identifying children with M. pneumoniae pneumonia is not known. Also, not all children with pneumonia undergo testing for M. pneumoniae, and different tests have varying sensitivity and specificity.57, 58 Thus, some children with M. pneumoniae pneumonia were not diagnosed and so were not included in our study. It is not known how inclusion of these children would affect our results.

Second, the antibiotic information used in this study was limited to empiric antibiotic therapy. It is possible that some patients received macrolide therapy before admission. It is also likely that identification of M. pneumoniae during the hospitalization prompted the addition or substitution of macrolide therapy for some patients. If this therapy was initiated beyond the first day of hospitalization, these children would be classified as macrolide non‐recipients. Since macrolide administration was associated with a shorter hospital LOS, such misclassification would bias our results towards finding no difference in LOS between macrolide recipients and non‐recipients. It is therefore possible that the benefit of macrolide therapy is even greater than found in our study.

Third, there may be unmeasured confounding or residual confounding by indication for adjunct corticosteroid therapy related to clinical presentation. We expect that corticosteroid recipients would be sicker than non‐recipients. We included variables associated with a greater severity of illness (such as intensive care unit admission) in the multivariable analysis. Additionally, the shorter LOS among macrolide recipients remained when the analysis was stratified by receipt or non‐receipt of systemic corticosteroid therapy.

Fourth, we were only able to record readmissions occurring at the same hospital as the index admission; any readmission presenting to a different hospital following their index admission did not appear in our records, and was therefore not counted. It is thus possible that the true number of readmissions is higher than that represented here. Finally, despite the large number of patients included in this study, the number of short‐term readmissions was relatively small. Thus, we may have been underpowered to detect small but significant differences in short‐term readmission rates.

In conclusion, macrolide therapy was associated with shorter hospital LOS, but not with short‐term or longer‐term readmission in children presenting with M. pneumoniae pneumonia.

Mycoplasma pneumoniae is a common cause of community‐acquired pneumonia (CAP), among school‐age children and adolescents.14 Though pneumonia caused by M. pneumoniae is typically self‐limited, severe illness may occur.5 M. pneumoniae has also been implicated in airway inflammation, which may lead to the onset and development of chronic pulmonary disease.610 Few studies have directly addressed appropriate treatment strategies for M. pneumoniae pneumonia,11 and, despite its high prevalence and potential for causing severe complications, treatment recommendations remain inconsistent.

The efficacy of macrolide therapy in particular for M. pneumoniae remains unclear. In vitro susceptibility studies have shown bacteriostatic activity of erythromycin, clarithromycin, and azithromycin against M. pneumoniae.1218 Additionally, several small retrospective studies have shown that among children with atypical CAP (including M. pneumoniae pneumonia), those treated with macrolides were less likely to have persistence or progression of signs and symptoms after 3 days of therapy.19, 20 Lu et al21 found a shorter duration of fever among macrolide recipients compared with non‐recipients. In adults, Shames et al22 found a shorter duration of fever and hospitalization among erythromycin recipients compared with controls. Other randomized controlled trials have also addressed the use of macrolides in treatment of M. pneumoniae, but the ability to draw meaningful conclusions is limited by small samples sizes and by lack of details about the number of patients with M. pneumoniae.11

In addition to their antimicrobial effect, macrolides also have anti‐inflammatory properties.2327 The importance of these anti‐inflammatory properties is supported by studies showing clinical cure in patients treated with macrolides despite persistence of M. pneumoniae organisms,2831 clinical improvement despite the administration of doses that provide tissue levels below the minimum inhibitory concentration of the organism,3234 and clinical cure in patients with macrolide‐resistant M. pneumoniae.18, 35

The objectives of the current study were to examine the impact of macrolide therapy on the length of stay (LOS) and short‐ and longer‐term readmissions, including longer‐term asthma‐related readmissions, in children hospitalized with M. pneumoniae pneumonia.

METHODS

RESULTS

DISCUSSION

This multicenter study examined the role of macrolide therapy in children hospitalized with M. pneumoniae pneumonia. Empiric macrolide therapy was associated with an approximately 30% shorter hospital LOS and, in stratified analysis, remained associated with a significantly shorter hospital LOS in both systemic corticosteroid recipients and non‐recipients. Empiric macrolide therapy was not associated with short‐ or longer‐term hospital readmission.

Previous small randomized trials have been inconclusive regarding the potential benefit of macrolide therapy in M. pneumoniae pneumonia.11 Our study, which demonstrated a shorter LOS among macrolides recipients compared with non‐recipients, has several advantages over prior studies including a substantively larger sample size and multicenter design. Animal models support our observations regarding the potential beneficial antimicrobial role of macrolides. M. pneumoniae concentrations in bronchoalveolar lavage specimens were significantly lower among experimentally infected mice treated with clarithromycin, a macolide‐class antibiotic, compared with either placebo or dexamethasone.45 Combination therapy with clarithromycin and dexamethasone reduced histopathologic inflammation to a greater degree than dexamethasone alone.45

While the relative importance of the antimicrobial and anti‐inflammatory properties of macrolides is not known, observational studies of children infected with macrolide‐resistant M. pneumoniae suggest that the antimicrobial properties of macrolides may provide disproportionate clinical benefit. The duration of fever in macrolide recipients with macrolide‐resistant M. pneumoniae (median duration, 9 days) reported by Suzuki et al46 was significantly longer than those with macrolide‐susceptible infections (median duration, 5 days), and similar to the duration of fever in patients with M. pneumoniae infection treated with placebo (median duration, 8 days) reported by Kingston et al.47 Additionally, macrolide therapy was associated with significant improvements in lung function in patients with asthma and concomitant M. pneumoniae infection, but not in patients with asthma without documented M. pneumoniae infection.9 As corticosteroids also have anti‐inflammatory properties, we expect that any anti‐inflammatory benefit of macrolide therapy would be mitigated by the concomitant administration of corticosteroids. The shorter LOS associated with empiric macrolide therapy in our study was comparable among corticosteroid recipients and non‐recipients.

Atypical bacterial pathogens, including M. pneumoniae, are associated with diffuse lower airway inflammation6, 48 and airway hyperresponsiveness,6 and have been implicated as a cause of acute asthma exacerbations.7, 4954 Among patients with previously diagnosed asthma, acute M. pneumoniae infection was identified in up to 20% of those having acute exacerbations.7, 54 Macrolide therapy has a beneficial effect on lung function and airway hyperresponsiveness in adults with asthma.9, 55 Among mice infected with M. pneumoniae, 3 days of macrolide therapy resulted in a significant reduction in airway hyperresponsiveness compared with placebo or dexamethasone; however, after 6 days of therapy, there was no significant difference in airway hyperresponsiveness between those receiving macrolides, dexamethasone, or placebo, suggesting that the benefit of macrolides on airway hyperresponsiveness may be brief. Our findings of a shorter LOS but no difference in readmissions at 28 days or longer, for macrolide recipients compared with non‐recipients, support the limited benefit of macrolide therapy beyond the initial reduction in bacterial load seen in the first few days of therapy.

M. pneumoniae infection has also been implicated as a cause of chronic pulmonary disease, including asthma.610 In the mouse model, peribronchial and perivascular mononuclear infiltrates, increased airway methacholine reactivity, and increased airway obstruction were observed 530 days after M. pneumoniae inoculation.6 M. pneumoniae has been identified in 26 (50%) of 51 children experiencing their first asthma attack,7 and 23 (42%) of 55 adults with chronic, stable asthma.9 Nevertheless, results of other studies addressing the issue are inconsistent, and the role of M. pneumoniae in the development of asthma remains unclear.56 In order to investigate the impact of macrolide therapy on the development of chronic pulmonary disease requiring hospitalization, we examined the readmission rates in the 15 months following index discharge. The proportion of children hospitalized with asthma following the hospitalization for M. pneumoniae pneumonia was higher for both macrolide recipients and non‐recipients compared with the 24‐months prior to infection. These results support a possible role for M. pneumoniae in chronic pulmonary disease. However, macrolide therapy was not associated with long‐term overall hospital readmission or long‐term asthma readmission, either in the entire cohort or in the subset of patients without prior asthma hospitalizations or medications.

This study had several limitations. First, because the identification of children with M. pneumoniae pneumonia relied on ICD‐9 discharge diagnosis codes, it is possible that there was misclassification of disease. We minimized the inclusion of children without M. pneumoniae by including only children who received antibiotic therapy on the first day of hospitalization and by excluding patients younger than 6 years of age, a group at relatively low‐risk for M. pneumoniae infection. Further, our algorithm for identification of M. pneumoniae pneumonia was validated through review of the medical records at 1 institution and was found to have a high positive predictive value. However, the positive predictive value of these ICD‐9 codes may vary across institutions. Additionally, the sensitivity of ICD‐9 codes for identifying children with M. pneumoniae pneumonia is not known. Also, not all children with pneumonia undergo testing for M. pneumoniae, and different tests have varying sensitivity and specificity.57, 58 Thus, some children with M. pneumoniae pneumonia were not diagnosed and so were not included in our study. It is not known how inclusion of these children would affect our results.

Second, the antibiotic information used in this study was limited to empiric antibiotic therapy. It is possible that some patients received macrolide therapy before admission. It is also likely that identification of M. pneumoniae during the hospitalization prompted the addition or substitution of macrolide therapy for some patients. If this therapy was initiated beyond the first day of hospitalization, these children would be classified as macrolide non‐recipients. Since macrolide administration was associated with a shorter hospital LOS, such misclassification would bias our results towards finding no difference in LOS between macrolide recipients and non‐recipients. It is therefore possible that the benefit of macrolide therapy is even greater than found in our study.

Third, there may be unmeasured confounding or residual confounding by indication for adjunct corticosteroid therapy related to clinical presentation. We expect that corticosteroid recipients would be sicker than non‐recipients. We included variables associated with a greater severity of illness (such as intensive care unit admission) in the multivariable analysis. Additionally, the shorter LOS among macrolide recipients remained when the analysis was stratified by receipt or non‐receipt of systemic corticosteroid therapy.

Fourth, we were only able to record readmissions occurring at the same hospital as the index admission; any readmission presenting to a different hospital following their index admission did not appear in our records, and was therefore not counted. It is thus possible that the true number of readmissions is higher than that represented here. Finally, despite the large number of patients included in this study, the number of short‐term readmissions was relatively small. Thus, we may have been underpowered to detect small but significant differences in short‐term readmission rates.

In conclusion, macrolide therapy was associated with shorter hospital LOS, but not with short‐term or longer‐term readmission in children presenting with M. pneumoniae pneumonia.

Mycoplasma pneumoniae is a common cause of community‐acquired pneumonia (CAP), among school‐age children and adolescents.14 Though pneumonia caused by M. pneumoniae is typically self‐limited, severe illness may occur.5 M. pneumoniae has also been implicated in airway inflammation, which may lead to the onset and development of chronic pulmonary disease.610 Few studies have directly addressed appropriate treatment strategies for M. pneumoniae pneumonia,11 and, despite its high prevalence and potential for causing severe complications, treatment recommendations remain inconsistent.

The efficacy of macrolide therapy in particular for M. pneumoniae remains unclear. In vitro susceptibility studies have shown bacteriostatic activity of erythromycin, clarithromycin, and azithromycin against M. pneumoniae.1218 Additionally, several small retrospective studies have shown that among children with atypical CAP (including M. pneumoniae pneumonia), those treated with macrolides were less likely to have persistence or progression of signs and symptoms after 3 days of therapy.19, 20 Lu et al21 found a shorter duration of fever among macrolide recipients compared with non‐recipients. In adults, Shames et al22 found a shorter duration of fever and hospitalization among erythromycin recipients compared with controls. Other randomized controlled trials have also addressed the use of macrolides in treatment of M. pneumoniae, but the ability to draw meaningful conclusions is limited by small samples sizes and by lack of details about the number of patients with M. pneumoniae.11

In addition to their antimicrobial effect, macrolides also have anti‐inflammatory properties.2327 The importance of these anti‐inflammatory properties is supported by studies showing clinical cure in patients treated with macrolides despite persistence of M. pneumoniae organisms,2831 clinical improvement despite the administration of doses that provide tissue levels below the minimum inhibitory concentration of the organism,3234 and clinical cure in patients with macrolide‐resistant M. pneumoniae.18, 35

The objectives of the current study were to examine the impact of macrolide therapy on the length of stay (LOS) and short‐ and longer‐term readmissions, including longer‐term asthma‐related readmissions, in children hospitalized with M. pneumoniae pneumonia.

METHODS

RESULTS

DISCUSSION

This multicenter study examined the role of macrolide therapy in children hospitalized with M. pneumoniae pneumonia. Empiric macrolide therapy was associated with an approximately 30% shorter hospital LOS and, in stratified analysis, remained associated with a significantly shorter hospital LOS in both systemic corticosteroid recipients and non‐recipients. Empiric macrolide therapy was not associated with short‐ or longer‐term hospital readmission.

Previous small randomized trials have been inconclusive regarding the potential benefit of macrolide therapy in M. pneumoniae pneumonia.11 Our study, which demonstrated a shorter LOS among macrolides recipients compared with non‐recipients, has several advantages over prior studies including a substantively larger sample size and multicenter design. Animal models support our observations regarding the potential beneficial antimicrobial role of macrolides. M. pneumoniae concentrations in bronchoalveolar lavage specimens were significantly lower among experimentally infected mice treated with clarithromycin, a macolide‐class antibiotic, compared with either placebo or dexamethasone.45 Combination therapy with clarithromycin and dexamethasone reduced histopathologic inflammation to a greater degree than dexamethasone alone.45

While the relative importance of the antimicrobial and anti‐inflammatory properties of macrolides is not known, observational studies of children infected with macrolide‐resistant M. pneumoniae suggest that the antimicrobial properties of macrolides may provide disproportionate clinical benefit. The duration of fever in macrolide recipients with macrolide‐resistant M. pneumoniae (median duration, 9 days) reported by Suzuki et al46 was significantly longer than those with macrolide‐susceptible infections (median duration, 5 days), and similar to the duration of fever in patients with M. pneumoniae infection treated with placebo (median duration, 8 days) reported by Kingston et al.47 Additionally, macrolide therapy was associated with significant improvements in lung function in patients with asthma and concomitant M. pneumoniae infection, but not in patients with asthma without documented M. pneumoniae infection.9 As corticosteroids also have anti‐inflammatory properties, we expect that any anti‐inflammatory benefit of macrolide therapy would be mitigated by the concomitant administration of corticosteroids. The shorter LOS associated with empiric macrolide therapy in our study was comparable among corticosteroid recipients and non‐recipients.

Atypical bacterial pathogens, including M. pneumoniae, are associated with diffuse lower airway inflammation6, 48 and airway hyperresponsiveness,6 and have been implicated as a cause of acute asthma exacerbations.7, 4954 Among patients with previously diagnosed asthma, acute M. pneumoniae infection was identified in up to 20% of those having acute exacerbations.7, 54 Macrolide therapy has a beneficial effect on lung function and airway hyperresponsiveness in adults with asthma.9, 55 Among mice infected with M. pneumoniae, 3 days of macrolide therapy resulted in a significant reduction in airway hyperresponsiveness compared with placebo or dexamethasone; however, after 6 days of therapy, there was no significant difference in airway hyperresponsiveness between those receiving macrolides, dexamethasone, or placebo, suggesting that the benefit of macrolides on airway hyperresponsiveness may be brief. Our findings of a shorter LOS but no difference in readmissions at 28 days or longer, for macrolide recipients compared with non‐recipients, support the limited benefit of macrolide therapy beyond the initial reduction in bacterial load seen in the first few days of therapy.

M. pneumoniae infection has also been implicated as a cause of chronic pulmonary disease, including asthma.610 In the mouse model, peribronchial and perivascular mononuclear infiltrates, increased airway methacholine reactivity, and increased airway obstruction were observed 530 days after M. pneumoniae inoculation.6 M. pneumoniae has been identified in 26 (50%) of 51 children experiencing their first asthma attack,7 and 23 (42%) of 55 adults with chronic, stable asthma.9 Nevertheless, results of other studies addressing the issue are inconsistent, and the role of M. pneumoniae in the development of asthma remains unclear.56 In order to investigate the impact of macrolide therapy on the development of chronic pulmonary disease requiring hospitalization, we examined the readmission rates in the 15 months following index discharge. The proportion of children hospitalized with asthma following the hospitalization for M. pneumoniae pneumonia was higher for both macrolide recipients and non‐recipients compared with the 24‐months prior to infection. These results support a possible role for M. pneumoniae in chronic pulmonary disease. However, macrolide therapy was not associated with long‐term overall hospital readmission or long‐term asthma readmission, either in the entire cohort or in the subset of patients without prior asthma hospitalizations or medications.

This study had several limitations. First, because the identification of children with M. pneumoniae pneumonia relied on ICD‐9 discharge diagnosis codes, it is possible that there was misclassification of disease. We minimized the inclusion of children without M. pneumoniae by including only children who received antibiotic therapy on the first day of hospitalization and by excluding patients younger than 6 years of age, a group at relatively low‐risk for M. pneumoniae infection. Further, our algorithm for identification of M. pneumoniae pneumonia was validated through review of the medical records at 1 institution and was found to have a high positive predictive value. However, the positive predictive value of these ICD‐9 codes may vary across institutions. Additionally, the sensitivity of ICD‐9 codes for identifying children with M. pneumoniae pneumonia is not known. Also, not all children with pneumonia undergo testing for M. pneumoniae, and different tests have varying sensitivity and specificity.57, 58 Thus, some children with M. pneumoniae pneumonia were not diagnosed and so were not included in our study. It is not known how inclusion of these children would affect our results.

Second, the antibiotic information used in this study was limited to empiric antibiotic therapy. It is possible that some patients received macrolide therapy before admission. It is also likely that identification of M. pneumoniae during the hospitalization prompted the addition or substitution of macrolide therapy for some patients. If this therapy was initiated beyond the first day of hospitalization, these children would be classified as macrolide non‐recipients. Since macrolide administration was associated with a shorter hospital LOS, such misclassification would bias our results towards finding no difference in LOS between macrolide recipients and non‐recipients. It is therefore possible that the benefit of macrolide therapy is even greater than found in our study.

Third, there may be unmeasured confounding or residual confounding by indication for adjunct corticosteroid therapy related to clinical presentation. We expect that corticosteroid recipients would be sicker than non‐recipients. We included variables associated with a greater severity of illness (such as intensive care unit admission) in the multivariable analysis. Additionally, the shorter LOS among macrolide recipients remained when the analysis was stratified by receipt or non‐receipt of systemic corticosteroid therapy.

Fourth, we were only able to record readmissions occurring at the same hospital as the index admission; any readmission presenting to a different hospital following their index admission did not appear in our records, and was therefore not counted. It is thus possible that the true number of readmissions is higher than that represented here. Finally, despite the large number of patients included in this study, the number of short‐term readmissions was relatively small. Thus, we may have been underpowered to detect small but significant differences in short‐term readmission rates.

In conclusion, macrolide therapy was associated with shorter hospital LOS, but not with short‐term or longer‐term readmission in children presenting with M. pneumoniae pneumonia.