Key Points
Question
Is pediatric volume in the emergency department associated with possible delayed diagnosis in serious conditions?
Findings
In this cohort study including 58 998 children with a wide range of conditions, possible delayed diagnosis occurred in 15.8% of children. Delay was 26.7% less common with each 2-fold increase in volume, a statistically significant association.
Meaning
Possible delayed diagnosis is common and may be less likely in emergency departments with greater pediatric volume.
This cohort study evaluates the association of annual pediatric volume in the emergency department with delayed diagnosis.
Abstract
Importance
Diagnostic delays are common in the emergency department (ED) and may predispose to worse outcomes.
Objective
To evaluate the association of annual pediatric volume in the ED with delayed diagnosis.
Design, Setting, and Participants
This retrospective cohort study included all children younger than 18 years treated at 954 EDs in 8 states with a first-time diagnosis of any of 23 acute, serious conditions: bacterial meningitis, compartment syndrome, complicated pneumonia, craniospinal abscess, deep neck infection, ectopic pregnancy, encephalitis, intussusception, Kawasaki disease, mastoiditis, myocarditis, necrotizing fasciitis, nontraumatic intracranial hemorrhage, orbital cellulitis, osteomyelitis, ovarian torsion, pulmonary embolism, pyloric stenosis, septic arthritis, sinus venous thrombosis, slipped capital femoral epiphysis, stroke, or testicular torsion. Patients were identified using the Healthcare Cost and Utilization Project State ED and Inpatient Databases. Data were collected from January 2015 to December 2019, and data were analyzed from July to December 2023.
Exposure
Annual volume of children at the first ED visited.
Main Outcomes and Measures
Possible delayed diagnosis, defined as a patient with an ED discharge within 7 days prior to diagnosis. A secondary outcome was condition-specific complications. Rates of possible delayed diagnosis and complications were determined. The association of volume with delayed diagnosis across conditions was evaluated using conditional logistic regression matching on condition, age, and medical complexity. Condition-specific volume-delay associations were tested using hierarchical logistic models with log volume as the exposure, adjusting for age, sex, payer, medical complexity, and hospital urbanicity. The association of delayed diagnosis with complications by condition was then examined using logistic regressions.
Results
Of 58 998 included children, 37 211 (63.1%) were male, and the mean (SD) age was 7.1 (5.8) years. A total of 6709 (11.4%) had a complex chronic condition. Delayed diagnosis occurred in 9296 (15.8%; 95% CI, 15.5-16.1). Each 2-fold increase in annual pediatric volume was associated with a 26.7% (95% CI, 22.5-30.7) decrease in possible delayed diagnosis. For 21 of 23 conditions (all except ectopic pregnancy and sinus venous thrombosis), there were decreased rates of possible delayed diagnosis with increasing ED volume. Condition-specific complications were 11.2% (95% CI, 3.1-20.0) more likely among patients with a possible delayed diagnosis compared with those without.
Conclusions and Relevance
EDs with fewer pediatric encounters had more possible delayed diagnoses across 23 serious conditions. Tools to support timely diagnosis in low-volume EDs are needed.
Introduction
Diagnostic delays are common in the emergency department (ED) setting.1,2 Children are at particular risk of diagnostic delays because serious illness may be more subtle, varying developmental states render children less able to communicate specific symptoms, and pediatric clinical presentations overlap with far more common, nonserious illness. Diagnostic delays put children at higher risk of illness complications.3,4 Structural differences between EDs may also play a role in diagnostic delays and illness complications: most EDs primarily provide clinical care for adults, and one-third of US EDs evaluate fewer than 5 children per day, leading to variable experience and clinician comfort caring for children.5,6
Pediatric volume is a measure of EDs’ pediatric experience. We recently identified that EDs with lower pediatric volume had higher rates of delayed diagnosis of appendicitis in children.7 These delays in diagnosis were associated with more than double the risk of serious complications. Whether this volume-delay association applies to other pediatric conditions is unclear. Understanding the types of conditions for which ED pediatric experience plays a role in delayed diagnosis would illuminate the potential impact of interventions that support high-quality diagnosis in such EDs.
We sought to evaluate the association of EDs’ pediatric volume with possible delayed diagnosis across a set of serious pediatric conditions and to determine the association of delay with complications and hospital utilization.
Methods
Study Design, Setting, and Participants
We conducted a retrospective cohort study of all children younger than 18 years who visited a hospital (either through the ED or via direct admission) with an included condition between 2015 and 2019 in Arkansas, Florida, Georgia, Iowa, Maryland, Nebraska, New York, and Wisconsin. Together, these states include 20.9% of US children and represent geographically diverse regions. Patients were drawn from the Healthcare Cost and Utilization Project State Emergency Department Database (SEDD) and State Inpatient Database (SID).8 The SEDD and SID include all ED and inpatient clinical care and allow for patients to be followed up longitudinally between encounters and facilities.
The 23 included conditions were bacterial meningitis, compartment syndrome, complicated pneumonia, craniospinal abscess, deep neck infection (ie, retropharyngeal or parapharyngeal abscess), ectopic pregnancy, encephalitis, intussusception, Kawasaki disease, mastoiditis, myocarditis, necrotizing fasciitis, nontraumatic intracranial hemorrhage, orbital cellulitis, osteomyelitis, ovarian torsion, pulmonary embolism, pyloric stenosis, septic arthritis, sinus venous thrombosis, slipped capital femoral epiphysis, stroke, and testicular torsion. We selected these conditions a priori because we thought they were (1) frequently missed, (2) nonchronic (eg, unlike new cancer diagnoses), (3) challenging to diagnose in children, and (4) specifically identifiable using diagnosis codes, based on prior evidence when available or consensus opinion when prior evidence was not available (eTable 1 in Supplement 1).9 We included only the first time each patient presented with the condition and had a 1-year wash-in period (2014) wherein if a patient presented for a given condition, they were not included in the study. We excluded patients who had no longitudinal identifier in the SEDD or SID, who had a prior hospitalization within 30 days, who visited a hospital with no recorded ED visits among children, who had a missing hospital identifier, or who were transferred at the time of diagnosis but without a record from the receiving hospital (ie, when a patient was transferred across state lines).
We adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.10 The Ann & Robert Lurie Children’s Hospital Institutional Review Board approved this study as exempt from review with a waiver of informed consent.
Outcomes
The main outcome was possible delayed diagnosis, defined as having an ED discharge within the 7 days prior to the diagnosis of an included condition (ie, a revisit to diagnosis). The index encounter was defined as the earliest opportunity for diagnosis during an ED visit. For patients with a possible delayed diagnosis, the index encounter was defined as the prior ED discharge. For patients without possible delayed diagnosis, the index encounter was defined as the diagnosis encounter.
Secondary outcomes included condition-specific complications (eTable 1 in Supplement 1) and measures of hospital utilization. Condition-specific complications were identified using diagnosis codes, procedure codes, and disposition (eg, transfer to a rehabilitation facility). Condition-specific complications for 14 conditions were previously defined and validated.9 For the other 9, we defined specific outcomes by team consensus. Utilization measures included critical procedures (eg, cardiopulmonary resuscitation, dialysis, mechanical ventilation, or extracorporeal membrane oxygenation), intensive care unit (ICU) hospitalization, length of stay at the time of diagnosis, and number of subsequent ED/inpatient encounters in the 90 days following the diagnosis encounter.
Variables
The main exposure variable was annual pediatric volume in the ED, which was the mean annual number of pediatric visits to each ED across the ED’s years of inclusion.11 Patient-level covariates included age (categorized as younger than 2 years [preverbal], aged 2 to 5 years [preschool aged], aged 6 to 10 years [school aged], and 11 years or older [adolescent], except where noted as continuous), sex, payer (ie, public, private, or other), and history of a prior complex chronic condition (CCC) in any previous encounter in the SEDD or SID.12 The hospital-level covariate was urbanicity, defined as urban, micropolitan, or rural based on the modal urbanicity among patients visiting the hospital.13
Statistical Analysis
To evaluate the face validity of our possible delayed diagnosis construct, we reported the top 5 most common initial diagnoses from the encounter preceding diagnosis (in which the diagnosis was not made) among children with a possible delayed diagnosis. For this list, we grouped conditions using Clinical Classifications Software version 2019.1 (Agency for Healthcare Research and Quality) based on the first-listed diagnosis.14 The overall delayed diagnosis rate was reported as a proportion and binomial exact confidence interval. We determined whether there was a time trend in delayed diagnosis rates by constructing a logistic model with the outcome of delayed diagnosis and year-quarter as the covariate, including random intercepts for hospital and condition.
We next assessed the association of volume and possible delayed diagnosis using 2 types of models: an all-conditions model and condition specific models. In the all-conditions model, we matched each case patient (those with possible delayed diagnoses) to 4 control patients by condition (exact match), CCC (exact match), and age (nearest neighbor matched on continuous age). We then performed conditional logistic regression with the outcome of possible delayed diagnosis and sole covariate of log2 pediatric ED volume, conditioned on the matched set.3 The goal of this matched analysis was to control for condition, CCC, and age, which we believed to be the most important risks for delayed diagnosis. We used log volume to report the relative odds of delayed diagnosis with each 2-fold increase in volume. 95% CIs accounted for hospital-level clustering.15 We evaluated how model fit changed with inclusion of the volume variable using the change in Akaike information criterion (AIC), with a difference in AIC from the null model less than 2 representing an improvement in fit.16
In condition-specific models, we assessed the association between pediatric ED volume and delayed diagnosis by condition. Each condition-specific unadjusted model was constructed as a hierarchical logistic regression model with patients nested within hospitals. The outcome was delayed diagnosis; the main covariate was log2 pediatric volume with a random intercept for ED. The adjusted model added all covariates. For ectopic pregnancy, ovarian torsion, and testicular torsion, we did not include the sex covariate because each affects only a single sex; for ectopic pregnancy, we further omitted age since all patients were 11 years or older. All models conformed to previously recommended principles of volume association studies (volume should be modeled continuously, model fit should be evaluated, and hierarchical modeling should be used).17
Patients who experience a possible delayed diagnosis may change hospitals between encounters. Thus, we reported the proportion of patients who changed hospitals and the ratio of volume between the diagnosis and index EDs.
We evaluated the association of possible delayed diagnosis with outcomes. We first used the all-conditions model with condition-specific complications as the outcome and possible delayed diagnosis as the exposure. We then repeated the condition-specific adjusted models using the outcome of complications with the exposure of possible delayed diagnosis. Risk of ICU hospitalization and critical procedures after a possible delayed diagnosis was evaluated using χ2 tests. Rank sum tests assessed differences in length of stay and subsequent encounters. Throughout the analysis, we used an α less than .05 for statistical significance, and all P values were 2-tailed. All analyses were conducted using R version 4.3.0 (The R Foundation).
Sensitivity Analyses
We repeated the all-conditions model using a 3-day revisit window instead of 7 days to assess the impact of a higher-specificity time interval on the volume-delay association. To assess for selection bias, we also repeated the main model excluding patients from Arkansas and Florida, the states with the most children missing a longitudinal identifier. Finally, we repeated the matching process and all-conditions model excluding children transferred during their index encounter, as hospitals receiving higher numbers of transfers may be at lower risk of delayed diagnosis.
Results
Across 954 EDs, there were 87 220 patients with an included condition, of whom we excluded 24 694 (28.3%), with 22 245 of these excluded children (90.1%) from Arkansas or Florida for missing a longitudinal identifier, 3008 (3.4%) for a previous hospitalization within 30 days,18 310 (0.4%) for having an index hospital with no ED visits for children, 126 (0.1%) for missing hospital information, and 84 (0.1%) for missing receiving hospital information after a transfer. Of 58 998 included children, 37 211 (63.1%) were male, and the mean (SD) age was 7.1 (5.8) years (Table). The median (IQR) hospital volume was 3813 (1495-8241) encounters per year.
Table. Demographic Features of the Cohort of 58 998 Children.
| Characteristic | Total, No. (%) |
|---|---|
| Condition | |
| Bacterial meningitis | 2007 (3.4) |
| Compartment syndrome | 774 (1.3) |
| Complicated pneumonia | 6409 (10.9) |
| Craniospinal abscess | 770 (1.3) |
| Deep neck infection | 2527 (4.3) |
| Ectopic pregnancy | 375 (0.6) |
| Encephalitis | 2055 (3.5) |
| Intussusception | 5841 (9.9) |
| Kawasaki disease | 4611 (7.8) |
| Mastoiditis | 3103 (5.3) |
| Myocarditis | 958 (1.6) |
| Necrotizing fasciitis | 161 (0.3) |
| Nontraumatic intracranial hemorrhage | 3252 (5.5) |
| Orbital cellulitis | 4121 (7.0) |
| Osteomyelitis | 1778 (3.0) |
| Ovarian torsion | 1460 (2.5) |
| Pulmonary embolism | 1029 (1.7) |
| Pyloric stenosis | 5396 (9.1) |
| Slipped capital femoral epiphysis | 1280 (2.2) |
| Septic arthritis | 2969 (5.0) |
| Sinus venous thrombosis | 116 (0.2) |
| Stroke | 1913 (3.2) |
| Testicular torsion | 6093 (10.3) |
| Complex chronic condition | 6709 (11.4) |
| Age, y | |
| <2 | 17 438 (29.6) |
| 2-5 | 11 982 (20.3) |
| 6-10 | 9857 (16.7) |
| ≥11 | 19 721 (33.4) |
| Sex | |
| Female | 21 787 (36.9) |
| Male | 37 211 (63.1) |
| Primary payer | |
| Private | 23 473 (39.8) |
| Public | 31 927 (54.1) |
| Other | 3506 (5.9) |
| Missing | 92 (0.2) |
| Index hospital urbanicity | |
| Urban | 55 664 (94.3) |
| Micropolitan | 1979 (3.4) |
| Rural | 1355 (2.3) |
Possible delayed diagnosis occurred in 9296 patients (15.8%; 95% CI, 15.5-16.1), of whom 7058 (75.9%) had their index encounter within the 3 days preceding diagnosis. Diagnoses from the index encounter of patients with possible delayed diagnosis are shown in eTable 2 in Supplement 1. Possible delayed diagnoses increased 2.2% (95% CI, 0.5-4.0) per year. Delayed diagnosis rates by condition are shown in Figure 1.
Figure 1. Proportions of Patients With a Possible Delayed Diagnosis by Condition.
Possible delay was defined as a prior emergency department discharge within 7 days.
Risk of Possible Delayed Diagnosis
In the all-conditions model, the risk of delayed diagnosis decreased by 26.7% (95% CI, 22.5-30.7) per 2-fold increase in annual pediatric volume. The inclusion of ED volume improved the model fit substantially, with a change in AIC between the volume model and null model of −1755.
In the condition-specific models for the 23 assessed conditions, 20 conditions in unadjusted models and 21 in adjusted models had a significantly lower risk of delayed diagnosis with higher pediatric ED volume (Figure 2). The median association, stroke, was associated with a 26.5% (95% CI, 11.1-39.3) decreased adjusted odds of possible delayed diagnosis per 2-fold increase in volume. The strongest association between volume and delay was in pyloric stenosis, with a 50.5% (95% CI, 42.0-57.7) decreased adjusted odds of possible delayed diagnosis per 2-fold increase in of pediatric volume. Volume did not have an association with delayed diagnosis for sinus venous thrombosis or ectopic pregnancy in unadjusted and adjusted models; orbital cellulitis did not have an association in the unadjusted model only.
Figure 2. Association of Pediatric Volume and Delayed Diagnosis by Condition.
Model estimates show the change in odds of delayed diagnosis for each 2-fold increase in hospital pediatric volume. Unadjusted models included a random intercept for emergency department. Adjusted models added age, sex, complex chronic condition, and hospital urbanicity. SCFE indicates slipped capital femoral epiphysis.
Among patients with a possible delayed diagnosis, 4644 (50.0%) changed hospitals between encounters. Among patients who changed hospitals, the median (IQR) annual pediatric volume at the diagnosis ED was 4.4-fold (1.8-11.1) greater than the index ED (eFigure in Supplement 1).
Association of Possible Delayed Diagnosis With Condition-Specific Complications
Condition-specific complications were 11.2% (95% CI, 3.1-20.0) more likely among patients with a possible delayed diagnosis compared with those without. Complications were significantly more likely after a possible delayed diagnosis for patients with craniospinal abscess, deep neck infection, intussusception, myocarditis, orbital cellulitis, pyloric stenosis, septic arthritis, stroke, and testicular torsion (Figure 3). They were significantly less likely after possible delayed diagnosis in bacterial meningitis and ectopic pregnancy.
Figure 3. Adjusted Odds Ratios (aORs) of Complications for a Possible Delayed Diagnosis of 23 Conditions, With Adjustment for Age, Sex, Complex Chronic Condition, and Hospital Urbanicity.
Conditions with a significantly lower likelihood of complications or higher likelihood are highlighted. ICH indicates intracranial hemorrhage; SCFE, slipped capital femoral epiphysis.
ICU hospitalization was more likely after possible delayed diagnosis (2083 of 9296 [22.4%] vs 10 455 of 49 702 [21.0%]; P = .003). Critical procedures were not significantly more likely after possible delayed diagnosis (783 of 9296 [8.4%] vs 4498 of 49 702 [9.0%]; P = .05). The 50th, 75th, 90th, and 95th percentile number of encounters in the 90 days following diagnosis were 1, 2, 2, and 3, respectively, for those without possible delay and 1, 2, 3, and 4 for those with possible delay (P < .001). The median (IQR) length of stay was 2 (0-5) days and 3 (1-7) days for those without and with delayed diagnosis, respectively (P < .001).
Sensitivity Analyses
Using a stricter threshold of revisits within 3 days instead of 7 days to define possible delayed diagnosis, the all-conditions model result was unchanged: the risk of possible delayed diagnosis was 27.3% (95% CI, 23.8-30.7) lower per 2-fold increase in annual pediatric volume. Excluding Arkansas and Florida children, the risk decreased to 26.0% (95% CI, 21.4-30.4) per 2-fold increase in annual pediatric volume. Index interhospital transfers occurred in 4984 patients (8.4%). Excluding those patients, the all-conditions model revealed that each 2-fold increase in volume was associated with a 24.9% (95% CI, 20.2-29.3) decrease in delayed diagnosis.
Discussion
Across 8 states representing more than one-fifth of US children, possible delayed diagnosis of serious conditions occurred in 15% of children and was more likely to occur in EDs with lower pediatric volume. This finding was consistent across nearly all the 23 conditions studied. For 9 of the conditions, complications were significantly more likely after a possible delayed diagnosis. Long hospital length of stay, ICU hospitalization, and subsequent ED and inpatient encounters were each more common after a possible delayed diagnosis. Taken together, our findings indicate that EDs with more pediatric encounters had lower rates of possible delayed diagnosis, which in turn was associated with fewer complications across a wide array of conditions. Importantly, most US EDs have low pediatric volume, meaning that the problem is widespread.
The association of volume with possible delay has several potential explanations. First, experience in providing clinical care for children may matter. Clinicians that spend more time evaluating children may develop more diagnostic accuracy, may be more aware of the range of presentations for serious conditions, and could be less subject to cognitive biases.19 Second, EDs with higher pediatric volume are more common in facilities with other pediatric resources, such as specialists available for consultation. Specialist consultation can provide both expertise and can add a second set of eyes to the facts of a case. Imaging and diagnostic resources are also likely to be more available at high-volume centers, with such resources supporting timely diagnosis. Third, higher pediatric readiness in higher-volume EDs is associated with the presence of clinical practice guidelines and local pathways that can guide care to avoid cognitive biases and missed diagnoses.5 Fourth, outpatient clinicians may refer children they suspect of having serious diagnoses to high-volume EDs, which may prime receiving clinicians to consider serious conditions. Finally, the patients evaluated at high-volume EDs tend to be sicker, which increases the base rate of serious conditions and could prompt a lower threshold to initiate workups.
Not all revisits leading to a serious diagnosis are true delays in diagnosis. Many represent progression of disease or the later development of a serious condition that was not present during the index visit. For example, in Kawasaki disease, the diagnosis should generally not be entertained with fewer than 5 days of fever. Thus, a child who visited on the fourth day of fever who eventually revisits and is diagnosed with Kawasaki disease does not necessarily have a delay in diagnosis. Similarly, community-acquired pneumonia may progress to complicated pneumonia, even with appropriate outpatient treatment. Thus, true rates of delayed diagnosis are not calculable from all revisits. Instead, the rates we report should be regarded as a ceiling. Based on the conditions most commonly seen in preceding encounters among those with possible delayed diagnosis, many are plausibly true missed diagnoses. We do believe that the association of volume with possible delayed diagnosis is more likely to be true than it is to be explainable by confounding by progression of disease, as the relative likelihood of progression of disease vs true delayed diagnosis is unlikely to vary by ED volume.
For 9 conditions, complications were unsurprisingly more common after possible delayed diagnosis. For the other conditions, there are several reasons why complications might either be unchanged or less common after possible delayed diagnosis. We had variable power to measure complications; for some conditions, the confidence intervals around complication rates were wide, obscuring potential associations. Second, progression of disease would tend to bias the complication association toward the null. Third, for some conditions, delays in diagnosis may not actually lead to a worse outcome. For example, in many cases, myocarditis may progress slowly enough that a delayed diagnosis might not be harmful.20 Finally, for some conditions, patients with real delays in diagnosis may be more likely than those with timely diagnosis to have an indolent form of the disease that is predestined to be less complicated.21 For example, patients who are well enough that bacterial meningitis is missed during an ED visit may have less pathogenic etiologies, whereas patients with meningococcal meningitis are easier to identify in the first encounter.22
What can be done to reduce delays in diagnosis related to low pediatric volume? First, making diagnostic expertise available to EDs with relatively low pediatric volume could help, for example, through teleconsultation with pediatric emergency medicine–trained clinicians or pediatric specialists.23,24,25 Given our findings, trials of the effectiveness of expanded access to teleconsultation are warranted. Second, better integration of clinical practice guidelines and clinical decision support into commercially available electronic health records could provide guidance to clinicians that would reduce delays.26,27,28 Third, deepening relationships between referring and receiving EDs could lower barriers to communication between sites. Finally, stronger education of clinicians at low-volume EDs could also help, but the mechanism for achieving this is not certain, particularly since clinicians’ focus may reasonably be on their largest constituency: adults.29
Limitations
This study had several limitations. First, some patients with possible delayed diagnosis had progression of disease and not a true delay; this would be unlikely to affect the main outcome. Second, certain conditions, such as sinus venous thrombosis, were rare, which limited power. Third, we did not include several important conditions, such as chronic conditions or those without accurate diagnosis coding.
Conclusions
In conclusion, EDs with fewer pediatric encounters were more likely to have possible diagnostic delays across many serious conditions, with many associated with excess complications. Studies of cost-effective strategies to reduce delayed diagnosis are pressing in pediatrics, since most EDs across the US have low pediatric volume.
eTable 1. Inclusion and Exclusion Criteria
eTable 2. Index Encounter Diagnoses
eFigure. Volumes of Diagnosis vs Index Encounter EDs
Data Sharing Statement
References
- 1.Plint AC, Stang A, Newton AS, et al. Adverse events in the paediatric emergency department: a prospective cohort study. BMJ Qual Saf. 2021;30(3):216-227. doi: 10.1136/bmjqs-2019-010055 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Lam D, Dominguez F, Leonard J, Wiersma A, Grubenhoff JA. Use of e-triggers to identify diagnostic errors in the paediatric ED. BMJ Qual Saf. 2022;31(10):735-743. doi: 10.1136/bmjqs-2021-013683 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Michelson KA, McGarghan FLE, Patterson EE, Samuels-Kalow ME, Waltzman ML, Greco KF. Delayed diagnosis of serious paediatric conditions in 13 regional emergency departments. BMJ Qual Saf. Published online September 30, 2022. doi: 10.1136/bmjqs-2022-015314 [DOI] [PubMed] [Google Scholar]
- 4.Naiditch JA, Lautz TB, Daley S, Pierce MC, Reynolds M. The implications of missed opportunities to diagnose appendicitis in children. Acad Emerg Med. 2013;20(6):592-596. doi: 10.1111/acem.12144 [DOI] [PubMed] [Google Scholar]
- 5.Remick KE, Hewes HA, Ely M, et al. National assessment of pediatric readiness of US emergency departments during the COVID-19 pandemic. JAMA Netw Open. 2023;6(7):e2321707. doi: 10.1001/jamanetworkopen.2023.21707 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Michelson KA, Hudgins JD, Lyons TW, Monuteaux MC, Bachur RG, Finkelstein JA. Trends in capability of hospitals to provide definitive acute care for children: 2008 to 2016. Pediatrics. 2020;145(1):e20192203. doi: 10.1542/peds.2019-2203 [DOI] [PubMed] [Google Scholar]
- 7.Michelson KA, Bachur RG, Rangel SJ, Monuteaux MC, Mahajan P, Finkelstein JA. Emergency department volume and delayed diagnosis of pediatric appendicitis: a retrospective cohort study. Ann Surg. 2023;278(6):833-838. doi: 10.1097/SLA.0000000000005972 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Healthcare Cost and Utilization Project . SEDD overview. Accessed January 9, 2024. https://www.hcup-us.ahrq.gov/seddoverview.jsp
- 9.Michelson KA, Dart AH, Finkelstein JA, Bachur RG. Validation of an automated system for identifying complications of serious pediatric emergencies. Hosp Pediatr. 2021;11(8):864-878. doi: 10.1542/hpeds.2020-005792 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative . The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. J Clin Epidemiol. 2008;61(4):344-349. doi: 10.1016/j.jclinepi.2007.11.008 [DOI] [PubMed] [Google Scholar]
- 11.French B, Farjah F, Flum DR, Heagerty PJ. A general framework for estimating volume-outcome associations from longitudinal data. Stat Med. 2012;31(4):366-382. doi: 10.1002/sim.4410 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Feudtner C, Feinstein JA, Zhong W, Hall M, Dai D. Pediatric complex chronic conditions classification system version 2: updated for ICD-10 and complex medical technology dependence and transplantation. BMC Pediatr. 2014;14(1):199. doi: 10.1186/1471-2431-14-199 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.US Centers for Disease Control and Prevention . NCHS Urban-Rural Classification Scheme for Counties. Accessed January 9, 2024. https://www.cdc.gov/nchs/data_access/urban_rural.htm
- 14.Healthcare Cost and Utilization Project . Tools Archive for Clinical Classifications Software Refined. Accessed January 9, 2024. https://hcup-us.ahrq.gov/toolssoftware/ccsr/ccsr_archive.jsp
- 15.Fay MP, Graubard BI. Small-sample adjustments for Wald-type tests using sandwich estimators. Biometrics. 2001;57(4):1198-1206. doi: 10.1111/j.0006-341X.2001.01198.x [DOI] [PubMed] [Google Scholar]
- 16.Burnham KP, Anderson DR. Multimodel inference. Sociol Methods Res. 2004;33(2):261-304. doi: 10.1177/0049124104268644 [DOI] [Google Scholar]
- 17.Livingston EH, Cao J. Procedure volume as a predictor of surgical outcomes. JAMA. 2010;304(1):95-97. doi: 10.1001/jama.2010.905 [DOI] [PubMed] [Google Scholar]
- 18.Bardach NS, Vittinghoff E, Asteria-Peñaloza R, et al. Measuring hospital quality using pediatric readmission and revisit rates. Pediatrics. 2013;132(3):429-436. doi: 10.1542/peds.2012-3527 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Schiff GD, Volodarskaya M, Ruan E, et al. Characteristics of disease-specific and generic diagnostic pitfalls: a qualitative study. JAMA Netw Open. 2022;5(1):e2144531. doi: 10.1001/jamanetworkopen.2021.44531 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Freedman SB, Haladyn JK, Floh A, Kirsh JA, Taylor G, Thull-Freedman J. Pediatric myocarditis: emergency department clinical findings and diagnostic evaluation. Pediatrics. 2007;120(6):1278-1285. doi: 10.1542/peds.2007-1073 [DOI] [PubMed] [Google Scholar]
- 21.Michelson KA, Bachur RG, Grubenhoff JA, et al. Outcomes of missed diagnosis of pediatric appendicitis, new-onset diabetic ketoacidosis, and sepsis in five pediatric hospitals. J Emerg Med. 2023;65(1):e9-e18. doi: 10.1016/j.jemermed.2023.04.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kilpi T, Anttila M, Kallio MJT, Peltola H. Severity of childhood bacterial meningitis and duration of illness before diagnosis. Lancet. 1991;338(8764):406-409. doi: 10.1016/0140-6736(91)91032-P [DOI] [PubMed] [Google Scholar]
- 23.Fiks AG, Jenssen BP, Ray KN. A defining moment for pediatric primary care telehealth. JAMA Pediatr. 2021;175(1):9-10. doi: 10.1001/jamapediatrics.2020.1881 [DOI] [PubMed] [Google Scholar]
- 24.Callahan CW, Malone F, Estroff D, Person DA. Effectiveness of an Internet-based store-and-forward telemedicine system for pediatric subspecialty consultation. Arch Pediatr Adolesc Med. 2005;159(4):389-393. doi: 10.1001/archpedi.159.4.389 [DOI] [PubMed] [Google Scholar]
- 25.Foster CC, Macy ML, Simon NJ, et al. Emergency care connect: extending pediatric emergency care expertise to general emergency departments through telemedicine. Acad Pediatr. 2020;20(5):577-584. doi: 10.1016/j.acap.2020.02.028 [DOI] [PubMed] [Google Scholar]
- 26.Kharbanda AB, Vazquez-Benitez G, Ballard DW, et al. ; Clinical Research on Emergency Services and Treatments Network (CREST) and the Critical Care Research Center, HealthPartners Institute . Effect of clinical decision support on diagnostic imaging for pediatric appendicitis: a cluster randomized trial. JAMA Netw Open. 2021;4(2):e2036344. doi: 10.1001/jamanetworkopen.2020.36344 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Castaneda C, Nalley K, Mannion C, et al. Clinical decision support systems for improving diagnostic accuracy and achieving precision medicine. J Clin Bioinforma. 2015;5(1):4. doi: 10.1186/s13336-015-0019-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Sibbald M, Monteiro S, Sherbino J, LoGiudice A, Friedman C, Norman G. Should electronic differential diagnosis support be used early or late in the diagnostic process? a multicentre experimental study of Isabel. BMJ Qual Saf. 2022;31(6):426-433. doi: 10.1136/bmjqs-2021-013493 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Dave N, Bui S, Morgan C, Hickey S, Paul CL. Interventions targeted at reducing diagnostic error: systematic review. BMJ Qual Saf. 2022;31(4):297-307. doi: 10.1136/bmjqs-2020-012704 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Inclusion and Exclusion Criteria
eTable 2. Index Encounter Diagnoses
eFigure. Volumes of Diagnosis vs Index Encounter EDs
Data Sharing Statement
