This retrospective cohort study examines the incidence of return emergency department visits among patients who left without being seen at a prior index visit.

Objective: Social media has become an important tool in monitoring infectious disease outbreaks such as coronavirus disease 2019 and highly pathogenic avian influenza (HPAI). Influenced by the recent announcement of a possible human death from H5N2 avian influenza, we analyzed tweets collected from X (formerly Twitter) to describe the messaging regarding the HPAI outbreak, including mis- and dis-information, concerns, and health education. Methods: We collected tweets involving keywords relating to HPAI for 5 days (June 04 to June 08, 2024). Using topic modeling, emotion, sentiment, and user demographic analyses, we were able to describe the population and the HPAI-related topics that users discussed. Results: With an original pool of 14,796 tweets, we analyzed a final data set of 13,319 tweets from 10,421 unique X users, with 50.4% of the tweets exhibiting negative sentiments (< 0 on a scale of -4 to +4). Predominant emotions were anger and fear shown in 36.4% and 29.5% of tweets, respectively. We identified 5 distinct, descriptive topics within the tweets. The use of emotionally charged language and spread of misinformation were substantial. Conclusions: Mis- and dis-information about the causes of and ways to prevent HPAI infections were common. A large portion of the tweets contained references to a planned epidemic or "plandemic" to influence the upcoming 2024 US presidential election. These tweets were countered by a limited number of tweets discussing infection locations, case reports, and preventive measures. Our study can be used by public health officials and clinicians to influence the discourse on current and future outbreaks.

Background: At a large quaternary health system, tissue specimens were frequently sent to the microbiology laboratory with an incorrect wound culture order meant for swab specimens due to poor electronic health record menu design. Wound cultures were also requested in chronic wound cases with a low index of suspicion for acute infection. Objective: To present a case report on specific changes to the design of the electronic test menu that resulted in higher numbers of appropriate ordering practices. Methods: "Wound Culture" test was renamed to "Wound Swab Culture" to distinguish from tissue specimens and "Tissue Culture" was added as a new available quick order in the microbiology menu alongside the existing wound culture quick order. In addition, a diagnostic questionnaire was added to "Wound Swab Culture" quick orders that inquired about the presence of pus/exudate and erythema and if the wound was a surgical wound to guide and assess the appropriateness of the culture order. Results: The number of tissue specimens erroneously submitted with a wound culture order decreased from 6.6% in July 2022 (pre-intervention) to 0% in July 2023 (post-intervention). The diagnostic questionnaire was utilized in 27.5% of wound culture orders. In 6 out of 98 orders (6.1%) the wound was not surgical and there was absence of pus/exudate and erythema (p = 0.038). Conversely, 92 out of 98 orders (93.9%) had at least one “Yes” response. Total numbers of tests six months before and after the test menu design interventions showed that tissue culture orders increased from 228 prior to the intervention to 349 post-intervention. Wound culture orders decreased from 575 to 460 (p < 0.0001). Conclusions: Our case report underscores how targeted electronic health record optimization can be associated with more appropriate microbiology test ordering practices for potential wound infections.

Purpose of review This review examines the current state and future prospects of machine learning (ML) in infection prevention and control (IPC) and antimicrobial stewardship (ASP), highlighting its potential to transform healthcare practices by enhancing the precision, efficiency, and effectiveness of interventions against infections and antimicrobial resistance. Recent findings ML has shown promise in improving surveillance and detection of infections, predicting infection risk, and optimizing antimicrobial use through the development of predictive analytics, natural language processing, and personalized medicine approaches. However, challenges remain, including issues related to data quality, model interpretability, ethical considerations, and integration into clinical workflows. Summary Despite these challenges, the future of ML in IPC and ASP is promising, with interdisciplinary collaboration identified as a key factor in overcoming existing barriers. ML's role in advancing personalized medicine, real-time disease monitoring, and effective IPC and ASP strategies signifies a pivotal shift towards safer, more efficient healthcare environments and improved patient care in the face of global antimicrobial resistance challenges.

Analyzing data from a national deidentified electronic health record-based data set using a matched case-control study design, we found that antibiotic use and severity of illness were independent risk factors for healthcare-associated candidemia in adult patients hospitalized with SARS-CoV-2 infection. Interleukin-6 inhibitor and corticosteroid use were not independent risk factors.

Background The increased prevalence of antimicrobial-resistant (AMR) infections is a significant global health threat, resulting in increased disease, deaths, and costs. The drivers of AMR are complex and potentially impacted by socioeconomic factors. We investigated the relationships between geographic and socioeconomic factors and AMR.Methods We collected select patient bacterial culture results from 2015 to 2020 from electronic health records of 2 expansive healthcare systems within the Dallas-Fort Worth, Texas, metropolitan area. Among individuals with electronic health records who resided in the 4 most populous counties in Dallas-Fort Worth, culture data were aggregated. Case counts for each organism studied were standardized per 1000 persons per area population. Using residential addresses, the cultures were geocoded and linked to socioeconomic index values. Spatial autocorrelation tests identified geographic clusters of high and low AMR organism prevalence and correlations with established socioeconomic indices.Results We found significant clusters of AMR organisms in areas with high levels of deprivation, as measured by the area deprivation index (ADI). We found a significant spatial autocorrelation between ADI and the prevalence of AMR organisms, particularly for AmpC beta-lactamase and methicillin-resistant Staphylococcus aureus, with 14% and 13%, respectively, of the variability in prevalence rates being attributable to their relationship with the ADI values of the neighboring locations.Conclusions We found that areas with a high ADI are more likely to have higher rates of AMR organisms. Interventions that improve socioeconomic factors such as poverty, unemployment, decreased access to healthcare, crowding, and sanitation in these areas of high prevalence may reduce the spread of AMR. Antimicrobial resistance (AMR) and its spread poses a significant threat to health worldwide. By identifying locations with high AMR prevalence through geospatial analyses, we discovered patterns of co-occurrence between socioeconomic factors, indicated by the area deprivation index, and AMR prevalence.

Introduction: During the COVID-19 pandemic, social media became increasingly relied upon for health information in Canada. By analyzing georeferenced tweets using natural language processing, we aimed to understand regional discussions and concerns about school closures, masking, vaccines, and lockdowns during the pandemic's first two years. Methods: Using Twitter's application programming interface, we collected English-language tweets with keywords related to COVID-19 posted between January 1, 2020 and February 22, 2022 from Canadian users. Results: Out of all retained tweets, 2,851,951 (47.9%) were about vaccines, 1,344,008 (22.6%) about lockdowns, 1,011,909 (17%) about schooling, and 752,014 (12.6%) about masking. Tweets on schooling received the most engagement, with the highest rates of likes (17.3%), retweets (18.7%), replies (10%), and quotes (6.8%). The most common emotions expressed were trust, fear, and anticipation, with lockdown tweets showing greater fear and sadness. Overall, sentiment was negative, particularly regarding lockdowns in the Northwest Territories and Alberta. Discussion: During the COVID-19 pandemic, Twitter became an essential tool for analyzing public sentiment regarding government actions. Users showed the most interest in vaccines, followed by lockdowns, schooling, and masking, with the highest engagement on schooling tweets. Our analysis of sentiment, emotion, and content revealed valuable insights into public beliefs about COVID-19 in Canada, highlighting regional differences and shifts in sentiment, particularly negative reactions to school closures as government recommendations evolved. Our study adds to the growing evidence supporting the use of natural language processing for real-time analysis of social media content to early identify public health concerns.

Background and Objective Clinical documentation is essential for conveying medical decision-making, communication between providers and patients, and capturing quality, billing, and regulatory measures during emergency department (ED) visits. Growing evidence suggests the benefits of note template standardization; however, variations in documentation practices are common. The primary objective of this study is to measure the utilization and coding performance of a standardized ED note template implemented across a nine-hospital health system. Methods This was a retrospective study before and after the implementation of a standardized ED note template. A multi-disciplinary group consensus was built around standardized note elements, provider note workflows within the electronic health record (EHR), and how to incorporate newly required medical decision-making elements. The primary outcomes measured included the proportion of ED visits using standardized note templates, and the distribution of billing codes in the 6 months before and after implementation. Results In the preimplementation period, a total of six legacy ED note templates were being used across nine EDs, with the most used template accounting for approximately 36% of ED visits. Marked variations in documentation elements were noted across six legacy templates. After the implementation, 82% of ED visits system-wide used a single standardized note template. Following implementation, we observed a 1% increase in the proportion of ED visits coded as highest acuity and an unchanged proportion coded as second highest acuity. Conclusion We observed a greater than twofold increase in the use of a standardized ED note template across a nine-hospital health system in anticipation of the new 2023 coding guidelines. The development and utilization of a standardized note template format relied heavily on multi-disciplinary stakeholder engagement to inform design that worked for varied documentation practices within the EHR. After the implementation of a standardized note template, we observed better-than-anticipated coding performance.

BACKGROUND:In 2011, the American Board of Medical Specialties established clinical informatics (CI) as a subspecialty in medicine, jointly administered by the American Board of Pathology and the American Board of Preventive Medicine. Subsequently, many institutions created CI fellowship training programs to meet the growing need for informaticists. Although many programs share similar features, there is considerable variation in program funding and administrative structures.OBJECTIVES:The aim of our study was to characterize CI fellowship program features, including governance structures, funding sources, and expenses.METHODS:We created a cross-sectional online REDCap survey with 44 items requesting information on program administration, fellows, administrative support, funding sources, and expenses. We surveyed program directors of programs accredited by the Accreditation Council for Graduate Medical Education between 2014 and 2021.RESULTS:We invited 54 program directors, of which 41 (76%) completed the survey. The average administrative support received was $27,732/year. Most programs (85.4%) were accredited to have two or more fellows per year. Programs were administratively housed under six departments: Internal Medicine (17; 41.5%), Pediatrics (7; 17.1%), Pathology (6; 14.6%), Family Medicine (6; 14.6%), Emergency Medicine (4; 9.8%), and Anesthesiology (1; 2.4%). Funding sources for CI fellowship program directors included: hospital or health systems (28.3%), clinical departments (28.3%), graduate medical education office (13.2%), biomedical informatics department (9.4%), hospital information technology (9.4%), research and grants (7.5%), and other sources (3.8%) that included philanthropy and external entities.CONCLUSION:CI fellowships have been established in leading academic and community health care systems across the country. Due to their unique training requirements, these programs require significant resources for education, administration, and recruitment. There continues to be considerable heterogeneity in funding models between programs. Our survey findings reinforce the need for reformed federal funding models for informatics practice and training.

BACKGROUND:Lack of consensus on the appropriate look-back period for multi-drug resistance (MDR) complicates antimicrobial clinical decision support. We compared the predictive performance of different MDR look-back periods for five common MDR mechanisms (MRSA, VRE, ESBL, AmpC, CRE).METHODS:We mapped microbiological cultures to MDR mechanisms and labeled them at different look-back periods. We compared predictive performance for each look-back period-MDR combination using precision, recall, F1 scores, and odds ratios.RESULTS:Longer look-back periods resulted in lower odds ratios, lower precisions, higher recalls, and lower delta changes in precision and recall compared to shorter periods. We observed higher precision with more information available to clinicians.CONCLUSION:A previously positive MDR culture may have significant enough precision depending on the mechanism of resistance and varying information available. One year is a clinically relevant and statistically sound look-back period for empiric antimicrobial decision-making at varying points of care for the studied population.

Objective: Social media's arrival eased the sharing of mis- and disinformation. False information proved challenging throughout the coronavirus disease 2019 (COVID-19) pandemic with many clinicians and researchers analyzing the "infodemic." We systemically reviewed and synthesized COVID-19 mis- and disinformation literature, identifying the prevalence and content of false information and exploring mitigation and prevention strategies.Design: We identified and analyzed publications on COVID-19-related mis- and disinformation published from March 1, 2020, to December 31, 2022, in PubMed. We performed a manual topic review of the abstracts along with automated topic modeling to organize and compare the different themes. We also conducted sentiment (ranked -3 to +3) and emotion analysis (rated as predominately happy, sad, angry, surprised, or fearful) of the abstracts.Results: We reviewed 868 peer-reviewed scientific publications of which 639 (74%) had abstracts available for automatic topic modeling and sentiment analysis. More than a third of publications described mitigation and prevention-related issues. The mean sentiment score for the publications was 0.685, and 56% of studies had a negative sentiment (fear and sadness as the most common emotions).Conclusions: Our comprehensive analysis reveals a significant proliferation of dis- and misinformation research during the COVID-19 pandemic. Our study illustrates the pivotal role of social media in amplifying false information. Research into the infodemic was characterized by negative sentiments. Combining manual and automated topic modeling provided a nuanced understanding of the complexities of COVID-19-related misinformation, highlighting themes such as the source and effect of misinformation, and strategies for mitigation and prevention.

Racial and ethnic bias in Large Language Models (LLMs) used for healthcare tasks is a growing concern, as it may contribute to health disparities. In response, LLM operators implemented safeguards against prompts that are overtly seeking certain bias. Our study investigates potential racial and ethnic bias in GPT-3.5-turbo, a popular LLM, in generating healthcare consumer-directed text in absence of overtly biased queries. In this cross-sectional study, GPT-3.5-turbo was prompted to generate discharge instructions for patients with Human Immunodeficiency Virus (HIV). Each patient’s encounter de-identified metadata including race/ethnicity as a variable were passed over in a table format through a prompt four times, altering only the race/ethnicity information (African American, Asian, Hispanic White, Non-Hispanic White) each time, while keeping all other information constant. The prompt requested the model to write discharge instructions for each encounter without explicitly mentioning race, ethnicity, or insurance type. The LLM-generated instructions were analyzed for sentiment, subjectivity, reading ease, and word usage by race/ethnicity and insurance type. The average polarity of GPT-3.5-turbo generated patient instructions across the different racial/ethnic groups was comparable, ranging from 0.14 to 0.15, with an average subjectivity of 0.46 for all groups. Differences in polarity and subjectivity across racial/ethnic groups were not statistically significant. However, word frequency varied across racial/ethnic groups, and subjectivity differed across insurance types, with commercial insurance eliciting the most subjective responses. GPT-3.5-turbo was relatively invariant to race/ethnicity and insurance type in terms of linguistic and readability measures. Further studies are needed to validate these results and assess their implications.

Background Social connectedness decreases human mortality, improves cancer survival, cardiovascular health, and body mass, results in better-controlled glucose levels, and strengthens mental health. However, few public health studies have leveraged large social media data sets to classify user network structure and geographic reach rather than the sole use of social media platforms. Objective The objective of this study was to determine the association between population-level digital social connectedness and reach and depression in the population across geographies of the United States. Methods Our study used an ecological assessment of aggregated, cross-sectional population measures of social connectedness, and self-reported depression across all counties in the United States. This study included all 3142 counties in the contiguous United States. We used measures obtained between 2018 and 2020 for adult residents in the study area. The study’s main exposure of interest is the Social Connectedness Index (SCI), a pair-wise composite index describing the “strength of connectedness between 2 geographic areas as represented by Facebook friendship ties.” This measure describes the density and geographical reach of average county residents’ social network using Facebook friendships and can differentiate between local and long-distance Facebook connections. The study’s outcome of interest is self-reported depressive disorder as published by the Centers for Disease Control and Prevention. Results On average, 21% (21/100) of all adult residents in the United States reported a depressive disorder. Depression frequency was the lowest for counties in the Northeast (18.6%) and was highest for southern counties (22.4%). Social networks in northeastern counties involved moderately local connections (SCI 5-10 the 20th percentile for n=70, 36% of counties), whereas social networks in Midwest, southern, and western counties contained mostly local connections (SCI 1-2 the 20th percentile for n=598, 56.7%, n=401, 28.2%, and n=159, 38.4%, respectively). As the quantity and distance that social connections span (ie, SCI) increased, the prevalence of depressive disorders decreased by 0.3% (SE 0.1%) per rank. Conclusions Social connectedness and depression showed, after adjusting for confounding factors such as income, education, cohabitation, natural resources, employment categories, accessibility, and urbanicity, that a greater social connectedness score is associated with a decreased prevalence of depression.

Large Language Models (LLM) are AI tools that can respond human-like to voice or free-text commands without training on specific tasks. However, concerns have been raised about their potential racial bias in healthcare tasks. In this study, ChatGPT was used to generate healthcare-related text for patients with HIV, analyzing data from 100 deidentified electronic health record encounters. Each patient's data were fed four times with all information remaining the same except for race/ethnicity (African American, Asian, Hispanic White, Non-Hispanic White). The text output was analyzed for sentiment, subjectivity, reading ease, and most used words by race/ethnicity and insurance type. Results showed that instructions for African American, Asian, Hispanic White, and Non-Hispanic White patients had an average polarity of 0.14, 0.14, 0.15, and 0.14, respectively, with an average subjectivity of 0.46 for all races/ethnicities. The differences in polarity and subjectivity across races/ethnicities were not statistically significant. However, there was a statistically significant difference in word frequency across races/ethnicities and a statistically significant difference in subjectivity across insurance types with commercial insurance eliciting the most subjective responses and Medicare and other payer types the lowest. The study suggests that ChatGPT is relatively invariant to race/ethnicity and insurance type in terms of linguistic and readability measures. Further studies are needed to validate these results and assess their implications.

Received: 22 September 2022 Accepted after revision: 02 December 2022 Accepted Manuscript online:19 December 2022

Background An undiagnosed HIV infection remains a public health challenge. In the digital era, social media and digital health communication have been widely used to accelerate research, improve consumer health, and facilitate public health interventions including HIV prevention. Objective We aimed to evaluate and compare the projected cost and efficacy of different simulated Facebook (FB) advertisement (ad) approaches targeting at-risk populations for HIV based on new HIV diagnosis rates by age group and geographic region in the United States. Methods We used the FB ad platform to simulate (without actually launching) an automatically placed video ad for a 10-day duration targeting at-risk populations for HIV. We compared the estimated total ad audience, daily reach, daily clicks, and cost. We tested ads for the age group of 13 to 24 years (in which undiagnosed HIV is most prevalent), other age groups, US geographic regions and states, and different campaign budgets. We then estimated the ad cost per new HIV diagnosis based on HIV positivity rates and the average health care industry conversion rate. Results On April 20, 2021, the potential reach of targeted ads to at-risk populations for HIV in the United States was approximately 16 million for all age groups and 3.3 million for age group 13 to 24 years, with the highest potential reach in California, Texas, Florida, and New York. When using different FB ad budgets, the daily reach and daily clicks per US dollar followed a cumulative distribution curve of an exponential function. Using multiple US $10 ten-day ads, the cost per every new HIV diagnosis ranged from US $13.09 to US $37.82, with an average cost of US $19.45. In contrast, a 1-time national ad had a cost of US $72.76 to US $452.25 per new HIV diagnosis (mean US $166.79). The estimated cost per new HIV diagnosis ranged from US $13.96 to US $55.10 for all age groups (highest potential reach and lowest cost in the age groups 20-29 and 30-39 years) and from US $12.55 to US $24.67 for all US regions (with the highest potential reach of 6.2 million and the lowest cost per new HIV diagnosis at US $12.55 in the US South). Conclusions Targeted personalized FB ads are a potential means to encourage at-risk populations for HIV to be tested, especially those aged 20 to 39 years in the US South, where the disease burden and potential reach on FB are high and the ad cost per new HIV diagnosis is low. Considering the cost efficiency of ads, the combined cost of multiple low-cost ads may be more economical than a single high-cost ad, suggesting that local FB ads could be more cost-effective than a single large-budget national FB ad.

ABSTRACT Objectives Outpatient parenteral antimicrobial therapy (OPAT) use has increased significantly as it provides safe and reliable administration of long-term antimicrobials for severe infections. Benefits of OPAT include fewer antibiotic or line-related complications, increased patient satisfaction, shorter hospitalizations, and lower costs. Although OPAT programs carefully screen patients for eligibility and safety prior to enrollment, complications can occur. There is a paucity of studies identifying predictors of clinical outcomes in OPAT patients. Here, we seek to identify baseline predictors of OPAT outcomes utilizing machine learning methodologies. Methods We used electronic health record data from patients treated with OPAT between February 2019 and June 2022 at a large academic tertiary care hospital in Dallas, Texas. Three primary outcomes were examined: 1) clinical improvement at 30 days without evidence of reinfection; 2) patient actively being followed at 30 days; and 3) occurrence of any adverse event while on OPAT. Potential predictors were determined a priori , including demographic and clinical characteristics, OPAT setting, intravenous line type, and antimicrobials administered. Three classifiers were used to predict each outcome: logistic regression, random forest, and extreme gradient boosting (XGBoost). Model performance was measured using AUC, F1, and accuracy scores. Results We included 664 unique patients in the study, of whom 57% were male. At 30 days, clinical improvement was present in 78% of patients. Two-thirds of patients (67%) were actively followed at 30 days, and 30% experienced an adverse event while on OPAT. The XGBoost model performed best for predicting treatment success (average AUC = 0.873), with significant predictors including ID consultation and the use of vancomycin. The logistic regression model was best for predicting adverse outcomes (average AUC = 0.710). Risk factors for adverse outcomes included management in the home setting and the use of vancomycin, daptomycin, or piperacillin-tazobactam. Conclusion Outcomes of patients undergoing OPAT can be predicted with the use of easily-obtainable clinical and demographic factors. Patients requiring certain antimicrobial therapies, such as vancomycin or daptomycin, may derive less benefit from early hospital discharge and OPAT.

AbstractBackgroundSocial media platforms like Twitter provide important insights into the public's perceptions of global outbreaks like monkeypox. By analyzing tweets, we aimed to identify public knowledge and opinions on the monkeypox virus and related public health issues.MethodsWe analyzed English-language tweets using the keyword “monkeypox” from 1 May to 23 July 2022. We reported gender, ethnicity, and race of Twitter users and analyzed tweets to identify predominant sentiment and emotions. We performed topic modeling and compared cohorts of users who self-identify as LGBTQ+ (an abreviation for lesbian, gay, bisexual, transgender, queer, and/or questioning) allies versus users who do not, and cohorts identified as “bots” versus humans.ResultsA total of 48 330 tweets were written by LGBTQ+ self-identified advocates or allies. The mean sentiment score for all tweets was −0.413 on a −4 to +4 scale. Negative tweets comprised 39% of tweets. The most common emotions expressed were fear and sadness. Topic modeling identified unique topics among the 4 cohorts analyzed.ConclusionsThe spread of mis- and disinformation about monkeypox was common in our tweet library. Various conspiracy theories about the origins of monkeypox, its relationship to global economic concerns, and homophobic and racial comments were common. Conversely, many other tweets helped to provide information about monkeypox vaccines, disease symptoms, and prevention methods. Discussion of rising monkeypox case numbers globally was also a large aspect of the conversation.ConclusionsWe demonstrated that Twitter is an effective means of tracking sentiment about public healthcare issues. We gained insight into a subset of people, self-identified LGBTQ+ allies, who were more affected by monkeypox.

BACKGROUND:Breastfeeding is a critical health intervention in infants. Recent literature reported that the COVID-19 pandemic resulted in significant mental health issues in pregnant and breastfeeding women due to social isolation and lack of direct professional support. These maternal mental health issues affected infant nutrition and decreased breastfeeding rates during COVID-19. Twitter, a popular social media platform, can provide insight into public perceptions and sentiment about various health-related topics. With evidence of significant mental health issues among women during the COVID-19 pandemic, the perception of infant nutrition, specifically breastfeeding, remains unknown.METHODS:We aimed to understand public perceptions and sentiment regarding breastfeeding during the COVID-19 pandemic through Twitter analysis using natural language processing techniques. We collected and analyzed tweets related to breastfeeding and COVID-19 during the pandemic from January 2020 to May 2022. We used Python software (v3.9.0) for all data processing and analyses. We performed sentiment and emotion analysis of the tweets using natural language processing libraries and topic modeling using an unsupervised machine-learning algorithm.RESULTS:We analyzed 40,628 tweets related to breastfeeding and COVID-19 generated by 28,216 users. Emotion analysis revealed predominantly "Positive emotions" regarding breastfeeding, comprising 72% of tweets. The overall tweet sentiment was positive, with a mean weekly sentiment of 0.25 throughout, and was affected by external events. Topic modeling revealed six significant themes related to breastfeeding and COVID-19. Passive immunity through breastfeeding after maternal vaccination had the highest mean positive sentiment score of 0.32.CONCLUSIONS:Our study provides insight into public perceptions and sentiment regarding breastfeeding during the COVID-19 pandemic. Contrary to other topics we explored in the context of COVID (e.g., ivermectin, disinformation), we found that breastfeeding had an overall positive sentiment during the pandemic despite the documented rise in mental health challenges in pregnant and breastfeeding mothers. The wide range of topics on Twitter related to breastfeeding provides an opportunity for active engagement by the medical community and timely dissemination of advice, support, and guidance. Future studies should leverage social media analysis to gain real-time insight into public health topics of importance in child health and apply targeted interventions.

OBJECTIVES Throughout the pandemic, children with COVID-19 have experienced hospitalization, ICU admission, invasive respiratory support, and death. Using a multisite, national dataset, we investigate risk factors associated with these outcomes in children with COVID-19. METHODS Our data source (Optum deidentified COVID-19 Electronic Health Record Dataset) included children aged 0 to 18 years testing positive for COVID-19 between January 1, 2020, and January 20, 2022. Using ordinal logistic regression, we identified factors associated with an ordinal outcome scale: nonhospitalization, hospitalization, or a severe composite outcome (ICU, intensive respiratory support, death). To contrast hospitalization for COVID-19 and incidental positivity on hospitalization, we secondarily identified patient factors associated with hospitalizations with a primary diagnosis of COVID-19. RESULTS In 165 437 children with COVID-19, 3087 (1.8%) were hospitalized without complication, 2954 (1.8%) experienced ICU admission and/or intensive respiratory support, and 31 (0.02%) died. We grouped patients by age: 0 to 4 years old (35 088), and 5 to 11 years old (75 574), 12 to 18 years old (54 775). Factors positively associated with worse outcomes were preexisting comorbidities and residency in the Southern United States. In 0- to 4-year-old children, there was a nonlinear association between age and worse outcomes, with worse outcomes in 0- to 2-year-old children. In 5- to 18-year-old patients, vaccination was protective. Findings were similar in our secondary analysis of hospitalizations with a primary diagnosis of COVID-19, though region effects were no longer observed. CONCLUSIONS Among children with COVID-19, preexisting comorbidities and residency in the Southern United States were positively associated with worse outcomes, whereas vaccination was negatively associated. Our study population was highly insured; future studies should evaluate underinsured populations to confirm generalizability.