Preferences for Management of Pediatric Pneumonia: A Clinician Survey of Artificially Generated Patient Cases.

BACKGROUND: It is unknown which factors are associated with chest radiograph (CXR) and antibiotic use for suspected community-acquired pneumonia (CAP) in children. We evaluated factors associated with CXR and antibiotic preferences among clinicians for children with suspected CAP using case scenarios generated through artificial intelligence (AI). METHODS: We performed a survey of general pediatric, pediatric emergency medicine, and emergency medicine attending physicians employed by a private physician contractor. Respondents were given 5 unique, AI-generated case scenarios. We used generalized estimating equations to identify factors associated with CXR and antibiotic use. We evaluated the cluster-weighted correlation between clinician suspicion and clinical prediction model risk estimates for CAP using 2 predictive models. RESULTS: A total of 172 respondents provided responses to 839 scenarios. Factors associated with CXR acquisition (OR, [95% CI]) included presence of crackles (4.17 [2.19, 7.95]), prior pneumonia (2.38 [1.32, 4.20]), chest pain (1.90 [1.18, 3.05]) and fever (1.82 [1.32, 2.52]). The decision to use antibiotics before knowledge of CXR results included past hospitalization for pneumonia (4.24 [1.88, 9.57]), focal decreased breath sounds (3.86 [1.98, 7.52]), and crackles (3.45 [2.15, 5.53]). After revealing CXR results to clinicians, these results were the sole predictor associated with antibiotic decision-making. Suspicion for CAP correlated with one of 2 prediction models for CAP (Spearman's rho = 0.25). Factors associated with a greater suspicion of pneumonia included prior pneumonia, duration of illness, worsening course of illness, shortness of breath, vomiting, decreased oral intake or urinary output, respiratory distress, head nodding, focal decreased breath sounds, focal rhonchi, fever, and crackles, and lower pulse oximetry. CONCLUSIONS: Ordering preferences for CXRs demonstrated similarities and differences with evidence-based risk models for CAP. Clinicians relied heavily on CXR findings to guide antibiotic ordering. These findings can be used within decision support systems to promote evidence-based management practices for pediatric CAP.

Community-Acquired Pneumonia Diagnosis Following Emergency Department Visits for Respiratory Illness.

Community-acquired pneumonia (CAP) is often considered for children presenting to the emergency department (ED) with respiratory symptoms. It is unclear how often children are diagnosed with CAP following an ED visit for respiratory illness. We performed a retrospective case-control study to evaluate 7-day CAP diagnosis among children 3 months to 18 years discharged from the ED with respiratory illness from 2011 to 2021 and who receive care at 4 hospital-affiliated primary care clinics. Logistic regression was performed to assess for predictors of 7-day CAP diagnosis. Seventy-four (0.7%, 95% confidence interval [CI] = 0.6%, 0.9%) of 10 329 children were diagnosed with CAP within 7 days, and fever at the index visit was associated with increased odds of diagnosis (odds ratio [OR] = 3.32, 95% CI = 1.75-6.28). Community-acquired pneumonia diagnosis after discharge from the ED with respiratory illness is rare, even among children who are febrile at time of initial evaluation.

Antibiotic route and outcomes for children hospitalized with pneumonia.

BACKGROUND: Emerging evidence suggests that initial oral and intravenous (IV) antibiotics have similar efficacy in pediatric community-acquired pneumonia (CAP), but further data are needed. OBJECTIVE: We determined the association between hospital-level initial oral antibiotic rates and outcomes in pediatric CAP. DESIGNS, SETTINGS AND PARTICIPANTS: This retrospective cohort study included children hospitalized with CAP at 43 hospitals in the Pediatric Health Information System (2016-2022). Hospitals were grouped by whether initial antibiotics were given orally in a high, moderate, or low proportion of patients. MAIN OUTCOME AND MEASURES: Regression models examined associations between high versus low oral-utilizing hospitals and length of stay (LOS, primary outcome), intensive care unit (ICU) transfers, escalated respiratory care, complicated CAP, cost, readmissions, and emergency department (ED) revisits. RESULTS: Initial oral antibiotics were used in 16% (interquartile range: 10%-20%) of 30,207 encounters, ranging from 1% to 68% across hospitals. Comparing high versus low oral-utilizing hospitals (oral rate: 32% [27%-47%] and 10% [9%-11%], respectively), there were no differences in LOS, intensive care unit, complicated CAP, cost, or ED revisits. Escalated respiratory care occurred in 1.3% and 0.5% of high and low oral-utilizing hospitals, respectively (relative ratio [RR]: 2.96 [1.12, 7.81]), and readmissions occurred in 1.5% and 0.8% (RR: 1.68 [1.31, 2.17]). Initial oral antibiotics varied across hospitals without a difference in LOS. While high oral-utilizing hospitals had higher escalated respiratory care and readmission rates, these were rare, the clinical significance of these small differences is uncertain, and there were no differences in other clinically relevant outcomes. This suggests some children may benefit from initial IV antibiotics, but most would probably do well with oral antibiotics.

Harnessing the Power of Generative AI for Clinical Summaries: Perspectives From Emergency Physicians.

STUDY OBJECTIVE: The workload of clinical documentation contributes to health care costs and professional burnout. The advent of generative artificial intelligence language models presents a promising solution. The perspective of clinicians may contribute to effective and responsible implementation of such tools. This study sought to evaluate 3 uses for generative artificial intelligence for clinical documentation in pediatric emergency medicine, measuring time savings, effort reduction, and physician attitudes and identifying potential risks and barriers. METHODS: This mixed-methods study was performed with 10 pediatric emergency medicine attending physicians from a single pediatric emergency department. Participants were asked to write a supervisory note for 4 clinical scenarios, with varying levels of complexity, twice without any assistance and twice with the assistance of ChatGPT Version 4.0. Participants evaluated 2 additional ChatGPT-generated clinical summaries: a structured handoff and a visit summary for a family written at an 8th grade reading level. Finally, a semistructured interview was performed to assess physicians' perspective on the use of ChatGPT in pediatric emergency medicine. Main outcomes and measures included between subjects' comparisons of the effort and time taken to complete the supervisory note with and without ChatGPT assistance. Effort was measured using a self-reported Likert scale of 0 to 10. Physicians' scoring of and attitude toward the ChatGPT-generated summaries were measured using a 0 to 10 Likert scale and open-ended questions. Summaries were scored for completeness, accuracy, efficiency, readability, and overall satisfaction. A thematic analysis was performed to analyze the content of the open-ended questions and to identify key themes. RESULTS: ChatGPT yielded a 40% reduction in time and a 33% decrease in effort for supervisory notes in intricate cases, with no discernible effect on simpler notes. ChatGPT-generated summaries for structured handoffs and family letters were highly rated, ranging from 7.0 to 9.0 out of 10, and most participants favored their inclusion in clinical practice. However, there were several critical reservations, out of which a set of general recommendations for applying ChatGPT to clinical summaries was formulated. CONCLUSION: Pediatric emergency medicine attendings in our study perceived that ChatGPT can deliver high-quality summaries while saving time and effort in many scenarios, but not all.

Interpretation of Antibiotic Trials in Pediatric Pneumonia.

This cohort study assesses radiographic evidence of pneumonia and antibiotic use in children with clinically suspected community-acquired pneumonia.

Diagnosis and Management of Pneumonia in Infants Less Than 90 Days of Age.

BACKGROUND: Current guidelines and recent studies on pediatric pneumonia pertain to children older than 3 months of age. Little information exists regarding the diagnostic evaluation, management, and outcomes of infants less than 90 days with pneumonia. METHODS: We compared infants <90 days of age diagnosed with pneumonia across 38 US children's hospitals from 2016 to 2021 to children 90 days to 5 years of age. We evaluated whether differences exist in patient characteristics, diagnostic testing, antibiotic treatment, and outcomes between young infants and older children. Additionally, we assessed seasonal variability and trends over time in pneumonia diagnoses by age group. RESULTS: Among 109 796 children diagnosed with pneumonia, 3128 (2.8%) were <90 days of age. Compared with older children, infants <90 days had more laboratory testing performed (88.6% vs 48.8%, P < .001; median number of laboratory tests 4 [interquartile range: 2-5] vs 0 [interquartile range: 0-3] respectively), with wide variation in testing across hospitals. Chest radiograph utilization did not differ by age group. Infants <90 days were more likely to be hospitalized and require respiratory support than older children. Seasonal variation was observed for pneumonia encounters in both age groups. CONCLUSIONS: Infants <90 days with pneumonia were more likely to undergo laboratory testing, be hospitalized, and require respiratory support than children 90 days to 5 years of age. This may reflect inherent differences in the pathophysiology of pneumonia by age, the manner in which pneumonia is diagnosed, or possible overuse of testing in infants.

Association of Chest Radiography With Outcomes in Pediatric Pneumonia: A Population-Based Study.

OBJECTIVE: Chest radiograph (CXR) is often performed for the evaluation of community-acquired pneumonia (CAP) in the ED setting. We sought to evaluate the association of undergoing CXR with 7-day hospitalization after emergency department (ED) discharge among patients with CAP. METHODS: This was a retrospective cohort study including children 3 months to 17 years discharged from any ED within 8 states from 2014 to 2019. We evaluated the association of CXR performance with 7-day hospitalization at both the patient and ED levels using mixed-effects logistic regression models accounting for markers of illness severity. Secondary outcomes included 7-day ED revisits and 7-day hospitalization with severe CAP. RESULTS: Among 206 694 children with CAP, rates of 7-day ED revisit, hospitalization, and severe CAP were 8.9%, 1.6%, and 0.4%, respectively. After adjusting for illness severity, CXR was associated with fewer 7-day hospitalizations (1.6% vs. 1.7%, adjusted odds ratio: [aOR] 0.82, 95% confidence interval [CI]: 0.73-0.92). CXR performance varied somewhat between EDs (median 91.5%, IQR: 85.3%-95.0%). EDs in the highest quartile had fewer 7-day hospitalizations (1.4% vs 1.9%, aOR: 0.78, 95% CI: 0.65-0.94), ED revisits (8.5% vs 9.4%, aOR: 0.88, 95% CI: 0.80-0.96) and hospitalizations for severe CAP (0.3% vs. 0.5%, aOR: 0.70, 95% CI: 0.51-0.97) as compared to EDs with the lowest quartile of CXR utilization. CONCLUSIONS: Among children discharged from the ED with CAP, performance of CXR was associated with a small but significant reduction in hospitalization within 7 days. CXR may be helpful in the prognostic evaluation of children with CAP discharged from the ED.

Management of Pediatric Pneumonia: A Decade After the Pediatric Infectious Diseases Society and Infectious Diseases Society of America Guideline.

BACKGROUND: Incomplete uptake of guidelines can lead to nonstandardized care, increased expenditures, and adverse clinical outcomes. The objective of this study was to evaluate the impact of the 2011 Pediatric Infectious Diseases Society and Infectious Diseases Society of America (PIDS/IDSA) pediatric community-acquired pneumonia (CAP) guideline that emphasized aminopenicillin use and de-emphasized the use of chest radiographs (CXRs) in certain populations. METHODS: This quasi-experimental study queried a national administrative database of children's hospitals to identify children aged 3 months-18 years with CAP who visited 1 of 28 participating hospitals from 2009 to 2021. PIDS/IDSA pediatric CAP guideline recommendations regarding antibiotic therapy, diagnostic testing, and imaging were evaluated. Segmented regression interrupted time series was used to measure guideline-concordant practices with interruptions for guideline publication and the Coronavirus Disease 2019 (COVID-19) pandemic. RESULTS: Of 315 384 children with CAP, 71 804 (22.8%) were hospitalized. Among hospitalized children, there was a decrease in blood culture performance (0.5% per quarter) and increase in aminopenicillin prescribing (1.1% per quarter). Among children discharged from the emergency department (ED), there was an increase in aminopenicillin prescription (0.45% per quarter), whereas the rate of obtaining CXRs declined (0.12% per quarter). However, use of CXRs rebounded during the COVID-19 pandemic (increase of 1.56% per quarter). Hospital length of stay, ED revisit rates, and hospital readmission rates remained stable. CONCLUSIONS: Guideline publication was associated with an increase of aminopenicillin prescribing. However, rates of diagnostic testing did not materially change, suggesting the need to consider implementation strategies to meaningfully change clinical practice for children with CAP.

Trends in the Use of Procalcitonin at US Children's Hospital Emergency Departments.

OBJECTIVES: Procalcitonin (PCT) was approved by the Food and Drug Administration in 2016. We assessed changes in PCT utilization over time in emergency departments (EDs) at US Children's Hospitals and identified the most common conditions associated with PCT testing. METHODS: We performed a cross-sectional study of children <18 years of age presenting to 1 of 33 EDs contributing data to the Pediatric Health Information System between 2016 and 2020. We examined trends in PCT utilization during an ED encounter between institutions and over the study period. Using All Patients Refined Diagnosis Related Groups, we identified the most common conditions for which PCT was obtained (overall, and relative to the performance of a complete blood count). RESULTS: The overall rate of PCT testing increased from 0.2% of all ED visits in 2016 to 1.8% in 2020. Across hospitals, the proportion of ED encounters with PCT obtained ranged from 0.0005% to 4.3% with marked variability in overall use. Among children who had PCT testing performed, the most common diagnoses were fever (10.7%), infections of the upper respiratory tract (9.2%), and pneumonia (5.9%). Relative to the performance of a complete blood count, rates of PCT testing were highest among children with sepsis (28.7%), fever (21.4%), pulmonary edema/respiratory failure (17.3%), and bronchiolitis/respiratory syncytial virus pneumonia (15.6%). CONCLUSIONS: PCT utilization in the ED has increased over the past 5 years with variation between hospitals. PCT is most frequently obtained for children with respiratory infections and febrile illnesses.

Antibiotics and outcomes of CF pulmonary exacerbations in children infected with MRSA and Pseudomonas aeruginosa.

BACKGROUND: Limited data exist to inform antibiotic selection among people with cystic fibrosis (CF) with airway infection by multiple CF-related microorganisms. This study aimed to determine among children with CF co-infected with methicillin-resistant Staphylococcus aureus (MRSA) and Pseudomonas aeruginosa (Pa) if the addition of anti-MRSA antibiotics to antipseudomonal antibiotic treatment for pulmonary exacerbations (PEx) would be associated with improved clinical outcomes compared with antipseudomonal antibiotics alone. METHODS: Retrospective cohort study using data from the CF Foundation Patient Registry-Pediatric Health Information System linked dataset. The odds of returning to baseline lung function and having a subsequent PEx requiring intravenous antibiotics were compared between PEx treated with anti-MRSA and antipseudomonal antibiotics and those treated with antipseudomonal antibiotics alone, adjusting for confounding by indication using inverse probability of treatment weighting. RESULTS: 943 children with CF co-infected with MRSA and Pa contributed 2,989 PEx for analysis. Of these, 2,331 (78%) PEx were treated with both anti-MRSA and antipseudomonal antibiotics and 658 (22%) PEx were treated with antipseudomonal antibiotics alone. Compared with PEx treated with antipseudomonal antibiotics alone, the addition of anti-MRSA antibiotics to antipseudomonal antibiotic therapy was not associated with a higher odds of returning to ≥90% or ≥100% of baseline lung function or a lower odds of future PEx requiring intravenous antibiotics. CONCLUSIONS: Children with CF co-infected with MRSA and Pa may not benefit from the addition of anti-MRSA antibiotics for PEx treatment. Prospective studies evaluating optimal antibiotic selection strategies for PEx treatment are needed to optimize clinical outcomes following PEx treatment.

Respiratory virus testing and clinical outcomes among children hospitalized with pneumonia.

BACKGROUND: Despite the increased availability of diagnostic tests for respiratory viruses, their clinical utility for children with community-acquired pneumonia (CAP) remains uncertain. OBJECTIVE: To identify patterns of respiratory virus testing across children's hospitals prior to the COVID-19 pandemic and to determine whether hospital-level rates of viral testing were associated with clinical outcomes. DESIGN, SETTING, AND PARTICIPANTS: Multicenter retrospective cohort study of children hospitalized for CAP at 19 children's hospitals in the United States from 2010-2019. MAIN OUTCOMES AND MEASURES: Using a novel method to identify the performance of viral testing, we assessed time trends in the use of viral tests, both overall and stratified by testing method. Adjusted proportions of encounters with viral testing were compared across hospitals and were correlated with length of stay, antibiotic and oseltamivir use, and performance of ancillary laboratory testing. RESULTS: There were 46,038 hospitalizations for non-severe CAP among children without complex chronic conditions. The proportion with viral testing increased from 38.8% to 44.2% during the study period (p < .001). Molecular testing increased (27.2% to 40.0%, p < .001) and antigen testing decreased (33.2% to 7.8%, p < .001). Hospital-specific adjusted proportions of testing ranged from 10.0% to 83.5% and were not associated with length of stay, antibiotic use, or antiviral use. Hospitals that performed more viral testing did not have lower rates of ancillary laboratory testing. CONCLUSIONS: Viral testing practices varied widely across children's hospitals and were not associated with clinically important process or outcome measures. Viral testing may not influence clinical management for many children hospitalized with CAP.

Nonoperative Management of Uncomplicated Appendicitis.

BACKGROUND AND OBJECTIVES: Several studies have revealed the success of nonoperative management (NOM) of uncomplicated appendicitis in children. Large studies of current NOM utilization and its outcomes in children are lacking. METHODS: We queried the Pediatric Health Information System database to identify children <19 years of age with a diagnosis code for appendicitis. We used linear trend analysis to assess the subsequent utilization and outcomes of NOM in children with nonperforated appendicitis over time. We calculated the proportion of children experiencing treatment failure, defined as either a subsequent appendectomy or hospitalization with a diagnosis code of perforated appendicitis. RESULTS: We identified 117 705 children with appendicitis over the 9-year study period. Of the 73 544 children with nonperforated appendicitis, 10 394 (14.1%) underwent NOM. The odds of NOM significantly increased (odds ratio 1.10 per study quarter, 95% confidence interval [CI] 1.05-1.15). The 1-year and 5-year failure rates were 18.6% and 23.3%, respectively. Children who experienced failure of NOM had higher rates of perforation at the time of failure than did the general cohort at the time of initial presentation (45.7% vs 37.5%, P < .001). Patients undergoing NOM had higher rates of subsequent related emergency department visits (8.0% vs 5.1%, P < .001) and hospitalizations (4.2% vs 1.4%, P < .001) over a 12-month follow-up period. CONCLUSIONS: NOM of nonperforated appendicitis in children is increasing. Although the majority of children who undergo NOM remain recurrence-free years later, they carry a substantial risk of perforation at the time of recurrence and may experience a higher rate of postoperative complications than children undergoing an immediate appendectomy.

Case of an Infant with Intermittent Eye Swelling.

CASE PRESENTATION: An eight-week-old infant presented to the emergency department with two weeks of fluctuating swelling and erythema of her right upper eyelid. On examination, she had swelling of the right upper eyelid with ptosis and proptosis as well as a nevus simplex on the upper eyelid. Orbital magnetic resonance imaging demonstrated a proliferating orbital hemangioma. DISCUSSION: Periorbital erythema and swelling are often infectious or allergic, but in infants with a fluctuating course, underlying vascular malformation must be considered. Without early provider recognition, periocular hemangiomas have the potential to cause vision-related complications.

Impact of Viral Radiographic Features on Antibiotic Treatment for Pediatric Pneumonia.

BACKGROUND/OBJECTIVES: Chest radiograph (CXR) is routinely performed among children with suspected pneumonia, though it is not clear how specific radiographic findings impact antibiotic treatment for pneumonia. We evaluated the impact of viral radiographic features on antibiotic treatment among children undergoing pneumonia evaluation in the emergency department (ED). METHODS: Children presenting to a pediatric ED who underwent a CXR for pneumonia evaluation were prospectively enrolled. Prior to CXR performance, physicians indicated their level of suspicion for pneumonia. The CXR report was reviewed to assess for the presence of viral features (peribronchial cuffing, perihilar markings, and interstitial infiltrate) as well as radiographic features suggestive of pneumonia (consolidation, infiltrate, and opacity). The relationship between viral radiographic features and antibiotic treatment was assessed based on the level of clinical suspicion for pneumonia prior to CXR. RESULTS: Patients with normal CXRs (n = 400) and viral features alone (n = 370) were managed similarly, with 8.0% and 8.6% of patients receiving antibiotic treatment, respectively (P = .75). Compared with children with radiographic pneumonia (n = 174), patients with concurrent viral features and radiographic pneumonia (n = 177) were treated with antibiotics less frequently (86.2% vs 54.3%, P < .001). Among children with isolated viral features on CXR, antibiotic treatment rates were correlated with pre-CXR level of suspicion for pneumonia. CONCLUSIONS: Among children with suspected pneumonia, the presence of viral features alone on CXR is not associated with increased rates of antibiotic use. Among children with radiographic pneumonia, the addition of viral features on CXR is associated with lower rates of antibiotic use, as compared to children with radiographic pneumonia alone.

Development of the Novel Pneumonia Risk Score to Predict Radiographic Pneumonia in Children.

BACKGROUND: The diagnosis of pneumonia in children is challenging, given the wide overlap of many of the symptoms and physical examination findings with other common respiratory illnesses. We sought to derive and validate the novel Pneumonia Risk Score (PRS), a clinical tool utilizing signs and symptoms available to clinicians to determine a child's risk of radiographic pneumonia. METHODS: We prospectively enrolled children 3 months to 18 years in whom a chest radiograph (CXR) was obtained in the emergency department to evaluate for pneumonia. Before CXR, we collected information regarding symptoms, physical examination findings, and the physician-estimated probability of radiographic pneumonia. Logistic regression was used to predict the presence of radiographic pneumonia, and the PRS was validated in a distinct cohort of children with suspected pneumonia. RESULTS: Among 1181 children included in the study, 206 (17%) had radiographic pneumonia. The PRS included age in years, triage oxygen saturation, presence of fever, presence of rales, and presence of wheeze. The area under the curve (AUC) of the PRS was 0.71 (95% confidence interval [CI]: 0.68-0.75), while the AUC of clinician judgment was 0.61 (95% CI: 0.56-0.66) (P < 0.001). Among 2132 children included in the validation cohort, the PRS demonstrated an AUC of 0.69 (95% CI: 0.65-0.73). CONCLUSIONS: In children with suspected pneumonia, the PRS is superior to clinician judgment in predicting the presence of radiographic pneumonia. Use of the PRS may help efforts to support the judicious use of antibiotics and chest radiography among children with suspected pneumonia.

Mycoplasma Pneumoniae Testing and Treatment Among Children With Community-Acquired Pneumonia.

OBJECTIVES: To describe testing and treatment practices for Mycoplasma pneumoniae (Mp) among children hospitalized with community-acquired pneumonia (CAP). METHODS: We conducted a retrospective cohort study using the Pediatric Health Information Systems database. We included children 3 months to 18 years old hospitalized with CAP between 2012 and 2018 and excluded children who were transferred from another hospital and those with complex chronic conditions. We examined the proportion of patients receiving Mp testing and macrolide therapy at the hospital level and trends in Mp testing and macrolide prescription over time. At the patient level, we examined differences in demographics, illness severity (eg, blood gas, chest tube placement), and outcomes (eg, ICU admission, length of stay, readmission) among patients with and without Mp testing. RESULTS: Among 103 977 children hospitalized with CAP, 17.3% underwent Mp testing and 31.1% received macrolides. We found no correlation between Mp testing and macrolide treatment at the hospital level (R 2 = 0.05; P = .11). Patients tested for Mp were more likely to have blood gas analysis (15.8% vs 12.8%; P < .1), chest tube placement (1.4% vs 0.8%; P < .1), and ICU admission (3.1% vs 1.4%; P < .1). Mp testing increased (from 15.8% to 18.6%; P < .001), and macrolide prescription decreased (from 40.9% to 20.6%; P < .001) between 2012 and 2018. CONCLUSIONS: Nearly one-third of hospitalized children with CAP received macrolide antibiotics, although macrolide prescription decreased over time. Clinicians were more likely to perform Mp testing in children with severe illness, and Mp testing and macrolide treatment were not correlated at the hospital level.

Prediction of patient disposition: comparison of computer and human approaches and a proposed synthesis.

OBJECTIVE: To compare the accuracy of computer versus physician predictions of hospitalization and to explore the potential synergies of hybrid physician-computer models. MATERIALS AND METHODS: A single-center prospective observational study in a tertiary pediatric hospital in Boston, Massachusetts, United States. Nine emergency department (ED) attending physicians participated in the study. Physicians predicted the likelihood of admission for patients in the ED whose hospitalization disposition had not yet been decided. In parallel, a random-forest computer model was developed to predict hospitalizations from the ED, based on data available within the first hour of the ED encounter. The model was tested on the same cohort of patients evaluated by the participating physicians. RESULTS: 198 pediatric patients were considered for inclusion. Six patients were excluded due to incomplete or erroneous physician forms. Of the 192 included patients, 54 (28%) were admitted and 138 (72%) were discharged. The positive predictive value for the prediction of admission was 66% for the clinicians, 73% for the computer model, and 86% for a hybrid model combining the two. To predict admission, physicians relied more heavily on the clinical appearance of the patient, while the computer model relied more heavily on technical data-driven features, such as the rate of prior admissions or distance traveled to hospital. DISCUSSION: Computer-generated predictions of patient disposition were more accurate than clinician-generated predictions. A hybrid prediction model improved accuracy over both individual predictions, highlighting the complementary and synergistic effects of both approaches. CONCLUSION: The integration of computer and clinician predictions can yield improved predictive performance.