PEDIATRICS/ORIGINAL RESEARCH
Repeated Emergency Department Visits Among Children Admitted With Meningitis or Septicemia: A Population-Based Study Samuel Vaillancourt, MDCM, MPH*; Astrid Guttmann, MDCM, MSc; Qi Li, MSc; Ian Y. M. Chan, BHSc, MPH; Marian J. Vermeulen, MHSc; Michael J. Schull, MD, MSc *Corresponding Author. E-mail: [email protected], Twitter: @VaillancourtSam.
Study objective: Early diagnosis of children with meningitis or septicemia remains a significant challenge in emergency medicine. We seek to describe the frequency of repeated emergency department (ED) visits among children admitted with meningitis or septicemia in Ontario, Canada. Methods: In this retrospective cohort study, using health administrative data, we included all children aged 30 days to 5 years who were hospitalized with a final diagnosis of meningitis or septicemia in Ontario between 2005 and 2010. ED visits at any hospital in the preceding 5 days were identified as potential repeated ED visits. We used generalized estimating equations to model the association of sex, age, triage score, immunocompromised state, visit timing, type of ED, and annual patient volume on the risk of repeated ED visits. Results: Of 521 children, 114 (21.9%) had repeated ED visits before admission. Children admitted on initial visit and those with repeated visits had similar median lengths of stay (13 versus 12 days), critical care use (21.1% versus 16.7%), and mortality (mean 2.9%). One in 3 children repeating visits returned to a different hospital. Repeated visits were associated with older age, a less acute triage score, and initial visit to a community hospital without available pediatric consultation. Conclusion: In this cohort, repeated ED visits among children with meningitis or septicemia were common, yet they had health outcomes similar to those of children admitted on initial visit. One in 3 returned to a different ED, making it unlikely that EDs and clinicians can learn from these critical events without a regionalized reporting system. [Ann Emerg Med. 2015;65:625-632.] Please see page 626 for the Editor’s Capsule Summary of this article. A feedback survey is available with each research article published on the Web at www.annemergmed.com. A podcast for this article is available at www.annemergmed.com. 0196-0644/$-see front matter Copyright © 2014 by the American College of Emergency Physicians. http://dx.doi.org/10.1016/j.annemergmed.2014.10.022
SEE EDITORIAL, P. 633. INTRODUCTION Background Despite advances in vaccinations and antibiotics, meningitis and septicemia remain significant causes of death of children in developed countries.1,2 Early diagnosis and treatment are important determinants of mortality and morbidity, but signs and symptoms are often difficult to differentiate from benign self-limited febrile illnesses.3,4 Since the introduction in Ontario, Canada, of the Haemophilus influenza (Hib) vaccine in the early 1990s and pneumococcal vaccination in 2005, the incidence of invasive infections by these 2 pathogens is estimated to have decreased by 95% and 60%, respectively, resulting in less clinical experience with their often subtle presentations for emergency physicians.5-9 Volume 65, no. 6 : June 2015
Importance Several studies and guidelines have emphasized the adverse effect of delays in care, but most have focused on treatment delays after diagnosis and little is known about ED contacts before diagnosis with meningitis or septicemia.10-12 One retrospective regional study of 21 children who died of severe bacterial infection (including meningitis) identified through a death registry in France found that 6 cases (29%) were inappropriately discharged home after initial presentation.13 However, a prospective but single-center study of children admitted with serious bacterial infection found that only 2 of 74 children (2.7%) had been missed on initial presentation.14 Both studies focused on children after the introduction of the Hib vaccine and before the pneumococcal vaccine. The early recognition of children with serious infection has been the focus of studies trying to devise decision rules and identify Annals of Emergency Medicine 625
Repeated Emergency Department Visits Among Children With Meningitis or Septicemia
Editor’s Capsule Summary
What is already known on this topic Emergency physicians fear missing early presentations of bacterial meningitis and septicemia, particularly in healthy children. What question this study addressed How often do children visit the emergency department (ED) in the 5 days before diagnosis of meningitis or septicemia? What this study adds to our knowledge In this database of all Ontario ED visits during a 5-year period, 1 in 5 children with meningitis or septicemia had a previous ED visit. Children who return to the ED and receive a diagnosis of meningitis and septicemia have clinical outcomes similar to those who receive a diagnosis at the initial visit. How this is relevant to clinical practice This study supports current clinical practice for the timely identification of serious bacterial infections in children.
prognostic factors.3,14,15 Yet to our knowledge, no study has examined the frequency of repeated emergency department (ED) visits before admission with meningitis or septicemia among children at a population level, and whether ED characteristics such as access to consultations or patient volume are related to these events. Goals of This Investigation Our objectives were, first, to determine the frequency of repeated visits before admission for meningitis or septicemia among children presenting to the EDs of the province of Ontario, Canada. Second, we wanted to identify individual and ED visit characteristics associated with repeated visits before admission. In particular, we wanted to test for association with type of ED and access to pediatric consultants, annual ED patient volumes, and time of visit. MATERIALS AND METHODS Study Design and Setting We conducted a retrospective cohort study using linked, anonymized, health administrative data from Ontario, Canada, with a total population of 13.2 million and 670,765 children younger than 5 years in 2006.16 We evaluated all nonelective hospital admissions from April 6, 2005, to March 1, 2010. We chose to focus on the period after 2005, the year 626 Annals of Emergency Medicine
Vaillancourt et al
of the introduction of the pneumococcal vaccination program, to examine a population of children who were covered by vaccination programs against both pneumococcus and Hib. Physician training varies across EDs in the province of Ontario, Canada. The majority of physicians working in pediatric EDs have focused pediatric training through completing a pediatric residency or emergency medicine specialty training. In 2009 to 2010, 1,350 physicians worked principally as emergency physicians, of whom 41% had family medicine training, 43% had family medicine training with an additional year of training in emergency medicine, and 16% were specialist emergency physicians.17 Specialist physicians tended to work in academic or pediatric EDs; family physicians with an additional year of emergency medicine training tended to work in highvolume community hospitals, which are more likely to have pediatric consultants; and family physicians without additional training were more likely to provide coverage in smaller community EDs that tend to have less consultant coverage. Data Collection and Processing Patient records were linked with a unique encrypted identifier at the Institute for Clinical Evaluative Sciences. We used data provided by the Canadian Institute for Health Information: the National Ambulatory Care Reporting System, which contains data from ED visits, and the Discharge Abstract Database, which contains hospital discharge data. A data quality study on the National Ambulatory Care Reporting System database showed high levels of agreement in regard to the fields used in this study.18 We used Statistics Canada 2001 census data to determine neighborhood income quintiles and patient postal code from the ED record to determine rural residence. We used data from a previous published survey, supplemented by additional investigation for hospitals with missing data, to determine the availability of pediatricians for consultation.19 Selection of Participants We included children aged 30 days to 5 years at hospital admission, with meningitis or septicemia as a most responsible diagnosis (see Appendix E1, available online at http://www.annemergmed.com, for International Classification of Diseases, 10th Revision codes used) and with a minimum length of stay of 4 days or who had died in the hospital. Given the difficulty on initial presentation of differentiating between bacterial and viral causes of meningitis, we included both.20-23 Because administrative data do not include detailed clinical information, we used mortality, critical care unit use, and inhospital length of Volume 65, no. 6 : June 2015
Vaillancourt et al
Repeated Emergency Department Visits Among Children With Meningitis or Septicemia
stay as markers of disease severity. We excluded children with a length of stay of fewer than 4 days because many of them are admitted for 3 days or fewer for observation pending blood or cerebrospinal fluid culture results and may still be coded as having sepsis or meningitis as a “ruleout” diagnosis. This decision was based on clinical experience and supported by data on hospitalizations for meningitis in the United States in 2006, which showed a mean length of stay of 9.1 days overall and 16.6 days for the bacterial meningitis subgroup. We excluded infants younger than 30 days because they are subject to heightened risk for disease and it is common practice in this population to initiate parenteral antibiotics and admit to the hospital for any suspicion of infection. We also excluded patients who had been discharged from the hospital for any reason within the previous 14 days of the index hospital admission for meningitis or septicemia because this readmission could be the result of progression of disease or nosocomial infections and would significantly change the medical decisionmaking process in the ED. To identify repeated visits, we looked for any previous ED visit that did not result in hospital admission and in which 120 hours (5 days) or fewer had elapsed between the registration times of the initial ED visit and the admitting ED visit (Figure 1). In cases in which the previous visit resulted in a transfer, only the first visit was considered. If they had no other ED visit, these cases were not considered to have repeated ED visits. Coded reason for the initial ED visit for all cases was manually reviewed by one of the investigators (S.V.) to exclude presentations that were unlikely to be related to an infectious process. The final list was reviewed by 2 other investigators independently (A.G. and M.J.S.) and no disagreement was identified. The International Classification of Diseases, 10th Revision, Canada codes of the visit reason used to define repeated visits are shown in Appendix E1, available online at http:// www.annemergmed.com. Patients were characterized according to age, sex, rural or urban residence, neighborhood income quintile, and immunocompromised state (as defined by the Agency for Healthcare Research and Quality for the pediatric patient safety measures).24 Initial ED visits were described with regard to acuity (using the Canadian Triage Acuity Scale),25 hospital type (pediatric, nonpediatric academic, Repeat ED Visits Admitted on Initial ED Visit
First ED visit
Admitting (Repeat) ED visit
Admitting ED visit
Hospitalization
Hospitalization
Initial ED visit
Figure 1. Classification of ED visits for meningitis or septicemia. Volume 65, no. 6 : June 2015
community without pediatric consultation availability, or community with pediatric consultation), annual ED patient volume, ED registration time (8 AM to 3:59 PM as day, 4 PM to 11:59 PM as evening, and midnight to 7:59 AM as night), and weekday versus weekend or holiday. In a regression model, age was categorized in clinically relevant groups of 30 to 90 days, a period during which heightened suspicion has been traditionally taught, greater than 90 days to 2 years, in which children have little ability to communicate verbally, and greater than 2 to less than or equal to 5 years. Type of ED was divided both in terms of type of hospital and in regard to availability of pediatric consultation service. The 5 levels of the Canadian Triage Acuity Scale scores were grouped as emergency (levels 1 and 2), urgent (level 3), and less urgent (levels 4 and 5).25 We analyzed the effect of annual ED patient volume, divided in 3 equal groups (low, medium, and high), which may affect the clinical proficiency of the department, including triage and ED nurses, physicians, and access to resources and tests. The effect of timing of the ED visit was analyzed by looking at nighttime visits between midnight and 8 AM against evening and daytime visits, hypothesizing that decisionmaking may be affected by fatigue and reduced access to consultation services. We also compared weekend and statutory holidays against weekdays, which may also affect ease of access to consultants. Outcome Measures We tracked the clinical outcomes of all the children admitted during the study period according to whether the diagnosis was made after repeated visits or on initial ED visit. We used hospital length of stay, critical care unit stay, and mortality. Primary Data Analysis In our primary descriptive analysis, we looked at baseline characteristics of patients and of the initial ED visit (referred to as “admitting ED visit” for children who received a diagnosis on initial visit and “first ED visit” for those who were admitted on a repeated visit) (Figure 1). In our secondary analysis, we conducted a multivariable logistic regression to examine predictors of repeated ED visits before admission for meningitis or septicemia. We used generalized estimating equations to account for the potential clustering of outcomes within EDs.26 More details on the modelbuilding process can be found in Appendix E2, available online at http://www.annemergmed.com. All statistical analysis was carried out with SAS (version 9.2 for UNIX; SAS Institute, Inc., Cary, NC). This study was approved by the Research Ethics Board at Sunnybrook Health Sciences Centre. Annals of Emergency Medicine 627
Repeated Emergency Department Visits Among Children With Meningitis or Septicemia
Vaillancourt et al
Table 1. Characteristics of patients and ED at initial ED visit.* Characteristics Patient characteristics Age Mean mo (SD) 30–90 days 91 days–2 y >2–5 y Sex Female Income quintile 5 (lowest) 4 3 2 1 (highest) Missing Rural Immunocompromised Diagnosis Meningitis Septicemia ED visit characteristics Triage acuity Emergency (CTAS 1–2) Urgent (CTAS 3) Less urgent (CTAS 4–5) ED type Pediatric ED Academic ED† Community ED without pediatric consultation Community ED with pediatric consultation Average annual ED patient volume Low
Repeated Emergency Department Visits Among Children Admitted With Meningitis or Septicemia: A Population-Based Study Samuel Vaillancourt, MDCM, MPH*; Astrid Guttmann, MDCM, MSc; Qi Li, MSc; Ian Y. M. Chan, BHSc, MPH; Marian J. Vermeulen, MHSc; Michael J. Schull, MD, MSc *Corresponding Author. E-mail: [email protected], Twitter: @VaillancourtSam.
Study objective: Early diagnosis of children with meningitis or septicemia remains a significant challenge in emergency medicine. We seek to describe the frequency of repeated emergency department (ED) visits among children admitted with meningitis or septicemia in Ontario, Canada. Methods: In this retrospective cohort study, using health administrative data, we included all children aged 30 days to 5 years who were hospitalized with a final diagnosis of meningitis or septicemia in Ontario between 2005 and 2010. ED visits at any hospital in the preceding 5 days were identified as potential repeated ED visits. We used generalized estimating equations to model the association of sex, age, triage score, immunocompromised state, visit timing, type of ED, and annual patient volume on the risk of repeated ED visits. Results: Of 521 children, 114 (21.9%) had repeated ED visits before admission. Children admitted on initial visit and those with repeated visits had similar median lengths of stay (13 versus 12 days), critical care use (21.1% versus 16.7%), and mortality (mean 2.9%). One in 3 children repeating visits returned to a different hospital. Repeated visits were associated with older age, a less acute triage score, and initial visit to a community hospital without available pediatric consultation. Conclusion: In this cohort, repeated ED visits among children with meningitis or septicemia were common, yet they had health outcomes similar to those of children admitted on initial visit. One in 3 returned to a different ED, making it unlikely that EDs and clinicians can learn from these critical events without a regionalized reporting system. [Ann Emerg Med. 2015;65:625-632.] Please see page 626 for the Editor’s Capsule Summary of this article. A feedback survey is available with each research article published on the Web at www.annemergmed.com. A podcast for this article is available at www.annemergmed.com. 0196-0644/$-see front matter Copyright © 2014 by the American College of Emergency Physicians. http://dx.doi.org/10.1016/j.annemergmed.2014.10.022
SEE EDITORIAL, P. 633. INTRODUCTION Background Despite advances in vaccinations and antibiotics, meningitis and septicemia remain significant causes of death of children in developed countries.1,2 Early diagnosis and treatment are important determinants of mortality and morbidity, but signs and symptoms are often difficult to differentiate from benign self-limited febrile illnesses.3,4 Since the introduction in Ontario, Canada, of the Haemophilus influenza (Hib) vaccine in the early 1990s and pneumococcal vaccination in 2005, the incidence of invasive infections by these 2 pathogens is estimated to have decreased by 95% and 60%, respectively, resulting in less clinical experience with their often subtle presentations for emergency physicians.5-9 Volume 65, no. 6 : June 2015
Importance Several studies and guidelines have emphasized the adverse effect of delays in care, but most have focused on treatment delays after diagnosis and little is known about ED contacts before diagnosis with meningitis or septicemia.10-12 One retrospective regional study of 21 children who died of severe bacterial infection (including meningitis) identified through a death registry in France found that 6 cases (29%) were inappropriately discharged home after initial presentation.13 However, a prospective but single-center study of children admitted with serious bacterial infection found that only 2 of 74 children (2.7%) had been missed on initial presentation.14 Both studies focused on children after the introduction of the Hib vaccine and before the pneumococcal vaccine. The early recognition of children with serious infection has been the focus of studies trying to devise decision rules and identify Annals of Emergency Medicine 625
Repeated Emergency Department Visits Among Children With Meningitis or Septicemia
Editor’s Capsule Summary
What is already known on this topic Emergency physicians fear missing early presentations of bacterial meningitis and septicemia, particularly in healthy children. What question this study addressed How often do children visit the emergency department (ED) in the 5 days before diagnosis of meningitis or septicemia? What this study adds to our knowledge In this database of all Ontario ED visits during a 5-year period, 1 in 5 children with meningitis or septicemia had a previous ED visit. Children who return to the ED and receive a diagnosis of meningitis and septicemia have clinical outcomes similar to those who receive a diagnosis at the initial visit. How this is relevant to clinical practice This study supports current clinical practice for the timely identification of serious bacterial infections in children.
prognostic factors.3,14,15 Yet to our knowledge, no study has examined the frequency of repeated emergency department (ED) visits before admission with meningitis or septicemia among children at a population level, and whether ED characteristics such as access to consultations or patient volume are related to these events. Goals of This Investigation Our objectives were, first, to determine the frequency of repeated visits before admission for meningitis or septicemia among children presenting to the EDs of the province of Ontario, Canada. Second, we wanted to identify individual and ED visit characteristics associated with repeated visits before admission. In particular, we wanted to test for association with type of ED and access to pediatric consultants, annual ED patient volumes, and time of visit. MATERIALS AND METHODS Study Design and Setting We conducted a retrospective cohort study using linked, anonymized, health administrative data from Ontario, Canada, with a total population of 13.2 million and 670,765 children younger than 5 years in 2006.16 We evaluated all nonelective hospital admissions from April 6, 2005, to March 1, 2010. We chose to focus on the period after 2005, the year 626 Annals of Emergency Medicine
Vaillancourt et al
of the introduction of the pneumococcal vaccination program, to examine a population of children who were covered by vaccination programs against both pneumococcus and Hib. Physician training varies across EDs in the province of Ontario, Canada. The majority of physicians working in pediatric EDs have focused pediatric training through completing a pediatric residency or emergency medicine specialty training. In 2009 to 2010, 1,350 physicians worked principally as emergency physicians, of whom 41% had family medicine training, 43% had family medicine training with an additional year of training in emergency medicine, and 16% were specialist emergency physicians.17 Specialist physicians tended to work in academic or pediatric EDs; family physicians with an additional year of emergency medicine training tended to work in highvolume community hospitals, which are more likely to have pediatric consultants; and family physicians without additional training were more likely to provide coverage in smaller community EDs that tend to have less consultant coverage. Data Collection and Processing Patient records were linked with a unique encrypted identifier at the Institute for Clinical Evaluative Sciences. We used data provided by the Canadian Institute for Health Information: the National Ambulatory Care Reporting System, which contains data from ED visits, and the Discharge Abstract Database, which contains hospital discharge data. A data quality study on the National Ambulatory Care Reporting System database showed high levels of agreement in regard to the fields used in this study.18 We used Statistics Canada 2001 census data to determine neighborhood income quintiles and patient postal code from the ED record to determine rural residence. We used data from a previous published survey, supplemented by additional investigation for hospitals with missing data, to determine the availability of pediatricians for consultation.19 Selection of Participants We included children aged 30 days to 5 years at hospital admission, with meningitis or septicemia as a most responsible diagnosis (see Appendix E1, available online at http://www.annemergmed.com, for International Classification of Diseases, 10th Revision codes used) and with a minimum length of stay of 4 days or who had died in the hospital. Given the difficulty on initial presentation of differentiating between bacterial and viral causes of meningitis, we included both.20-23 Because administrative data do not include detailed clinical information, we used mortality, critical care unit use, and inhospital length of Volume 65, no. 6 : June 2015
Vaillancourt et al
Repeated Emergency Department Visits Among Children With Meningitis or Septicemia
stay as markers of disease severity. We excluded children with a length of stay of fewer than 4 days because many of them are admitted for 3 days or fewer for observation pending blood or cerebrospinal fluid culture results and may still be coded as having sepsis or meningitis as a “ruleout” diagnosis. This decision was based on clinical experience and supported by data on hospitalizations for meningitis in the United States in 2006, which showed a mean length of stay of 9.1 days overall and 16.6 days for the bacterial meningitis subgroup. We excluded infants younger than 30 days because they are subject to heightened risk for disease and it is common practice in this population to initiate parenteral antibiotics and admit to the hospital for any suspicion of infection. We also excluded patients who had been discharged from the hospital for any reason within the previous 14 days of the index hospital admission for meningitis or septicemia because this readmission could be the result of progression of disease or nosocomial infections and would significantly change the medical decisionmaking process in the ED. To identify repeated visits, we looked for any previous ED visit that did not result in hospital admission and in which 120 hours (5 days) or fewer had elapsed between the registration times of the initial ED visit and the admitting ED visit (Figure 1). In cases in which the previous visit resulted in a transfer, only the first visit was considered. If they had no other ED visit, these cases were not considered to have repeated ED visits. Coded reason for the initial ED visit for all cases was manually reviewed by one of the investigators (S.V.) to exclude presentations that were unlikely to be related to an infectious process. The final list was reviewed by 2 other investigators independently (A.G. and M.J.S.) and no disagreement was identified. The International Classification of Diseases, 10th Revision, Canada codes of the visit reason used to define repeated visits are shown in Appendix E1, available online at http:// www.annemergmed.com. Patients were characterized according to age, sex, rural or urban residence, neighborhood income quintile, and immunocompromised state (as defined by the Agency for Healthcare Research and Quality for the pediatric patient safety measures).24 Initial ED visits were described with regard to acuity (using the Canadian Triage Acuity Scale),25 hospital type (pediatric, nonpediatric academic, Repeat ED Visits Admitted on Initial ED Visit
First ED visit
Admitting (Repeat) ED visit
Admitting ED visit
Hospitalization
Hospitalization
Initial ED visit
Figure 1. Classification of ED visits for meningitis or septicemia. Volume 65, no. 6 : June 2015
community without pediatric consultation availability, or community with pediatric consultation), annual ED patient volume, ED registration time (8 AM to 3:59 PM as day, 4 PM to 11:59 PM as evening, and midnight to 7:59 AM as night), and weekday versus weekend or holiday. In a regression model, age was categorized in clinically relevant groups of 30 to 90 days, a period during which heightened suspicion has been traditionally taught, greater than 90 days to 2 years, in which children have little ability to communicate verbally, and greater than 2 to less than or equal to 5 years. Type of ED was divided both in terms of type of hospital and in regard to availability of pediatric consultation service. The 5 levels of the Canadian Triage Acuity Scale scores were grouped as emergency (levels 1 and 2), urgent (level 3), and less urgent (levels 4 and 5).25 We analyzed the effect of annual ED patient volume, divided in 3 equal groups (low, medium, and high), which may affect the clinical proficiency of the department, including triage and ED nurses, physicians, and access to resources and tests. The effect of timing of the ED visit was analyzed by looking at nighttime visits between midnight and 8 AM against evening and daytime visits, hypothesizing that decisionmaking may be affected by fatigue and reduced access to consultation services. We also compared weekend and statutory holidays against weekdays, which may also affect ease of access to consultants. Outcome Measures We tracked the clinical outcomes of all the children admitted during the study period according to whether the diagnosis was made after repeated visits or on initial ED visit. We used hospital length of stay, critical care unit stay, and mortality. Primary Data Analysis In our primary descriptive analysis, we looked at baseline characteristics of patients and of the initial ED visit (referred to as “admitting ED visit” for children who received a diagnosis on initial visit and “first ED visit” for those who were admitted on a repeated visit) (Figure 1). In our secondary analysis, we conducted a multivariable logistic regression to examine predictors of repeated ED visits before admission for meningitis or septicemia. We used generalized estimating equations to account for the potential clustering of outcomes within EDs.26 More details on the modelbuilding process can be found in Appendix E2, available online at http://www.annemergmed.com. All statistical analysis was carried out with SAS (version 9.2 for UNIX; SAS Institute, Inc., Cary, NC). This study was approved by the Research Ethics Board at Sunnybrook Health Sciences Centre. Annals of Emergency Medicine 627
Repeated Emergency Department Visits Among Children With Meningitis or Septicemia
Vaillancourt et al
Table 1. Characteristics of patients and ED at initial ED visit.* Characteristics Patient characteristics Age Mean mo (SD) 30–90 days 91 days–2 y >2–5 y Sex Female Income quintile 5 (lowest) 4 3 2 1 (highest) Missing Rural Immunocompromised Diagnosis Meningitis Septicemia ED visit characteristics Triage acuity Emergency (CTAS 1–2) Urgent (CTAS 3) Less urgent (CTAS 4–5) ED type Pediatric ED Academic ED† Community ED without pediatric consultation Community ED with pediatric consultation Average annual ED patient volume Low
