Pediatrics最新文献

Background: There has been a substantial burden of hypertension in children and adolescents. Given the availability of primary prevention strategies, it is important to determine predictors for early identification of children and adolescents at risk of hypertension. This study aims to attempt and validate machine learning (ML) algorithms for accurately predicting blood pressure (BP) status (normal, prehypertension, and hypertension) over 1- and 3-year periods, identifying key predictors without compromising model performance.

Methods: We included a population-based cohort of primary 1 to secondary 6 students (typically aged 6 to 18 years) during the academic years of 1995 to 1996 and 2019 to 2020 in Hong Kong. Thirty-six easy-assessed predictors were initially model childhood BP status. Multiple ML algorithms, decision tree, random forest, k-nearest neighbor, eXtreme Gradient Boosting (XGBoost), and multinomial logistic regression (MLR), were used. Model evaluation was performed by various accuracy metrics. The Shapley Additive Explanations (SHAP) was used to identify key features for both predictions.

Results: A total of 923 301 and 602 179 visit pairs were used for the 1- and 3-year predictions, respectively. XGBoost demonstrated the highest prediction accuracies for 1-year (macro-area under the receiver operating characteristic curve [AUROC] = 0.92, micro-AUROC = 0.91) and 3-year (macro-AUROC = 0.91, micro-AUROC = 0.90) periods. The traditional MLR approach had the lowest accuracies for 1- (macro-AUROC = 0.70, micro-AUROC = 0.68) and 3-year (macro-AUROC = 0.70, micro-AUROC = 0.68) predictions. The SHAP values identified 17 key predictors without the need for direct BP measurements or laboratory tests.

Conclusion: ML prediction models can accurately predict childhood prehypertension and hypertension at 1 and 3 years, independent of BP and laboratory measurements. The identified key predictors may inform areas for personalized prevention in hypertension.

The COVID-19 pandemic exacerbated the ongoing adolescent mental health crisis in part by disrupting connections to schools, communities, and primary care. Reading has been shown to support mental health by fostering identification with narratives, situations, or characters and promoting empathy. Inspired by Reach Out and Read, we developed Turning Pages, a program designed to enhance the clinic experience for adolescents by offering books during annual physicals in a primary care clinic. The clinic serves a diverse population of 1750 adolescent and young adult patients, with approximately 40% covered by California's Medicaid program. Patients and clinicians recommended books that were meaningful during their adolescence, and we partnered with a local bookshop to curate a diverse and inclusive collection. At the start of each visit, medical assistants invited patients to select a book and offered an optional survey assessing program impact. A total of 326 books were acquired through the support of various grants and donations. Between May 2023 and January 2024, 295 books were distributed to adolescent patients. Of the 65 patients who completed the survey, 98% appreciated receiving a book, 81% felt it improved their clinic experience, and 68% reported feeling more connected to their medical team. Institutional support and positive feedback from patients, clinicians, and staff have ensured ongoing funding for the program. Turning Pages demonstrates potential as a sustainable program that enhances the clinic experience and strengthens connections to care.

Background and objectives: Ongoing systems-level changes aim to better identify and remedy the unmet health care needs of children with medical complexity (CMC). In tandem, home- and community-based services are expanding to support autistic children and their families. Despite the potential for overlap, CMC and autistic children are treated independently in services, research, and policy. We estimated the overlapping prevalence of CMC and autism among US children and health care expenditures for autistic CMC in comparison with other children.

Methods: We analyzed 2 national cross-sectional surveys: the National Survey of Children's Health (NSCH; 2017-2018, 2019-2022, and 2021-2022) and the Medical Expenditures Panel Survey (MEPS; combined 2010-2021), selecting for children aged 0 to 17 years. CMC were defined using 2 different algorithms varying in stringency.

Results: In the most recent 2021 to 2022 NSCH (n = 103 748), the prevalence of CMC among autistic children was 59.28% (95% CI, 55.61%-62.84%) using one algorithm and 17.56% (95% CI, 14.41%-21.24%) using the more stringent algorithm. Forty-one percent of CMC were autistic using either algorithm. In the MEPS data (n = 55 637), autistic CMC had significantly greater median health care expenditures compared with other CMC and other autistic children.

Conclusions: There is extensive overlap of CMC and autism among children in the United States. When medical complexity and autism are both evident, expenditures are significantly higher than for either category alone. Despite this overlap and the associated high need, CMC and autism are generally treated as separate groups in services, research, and policy. These findings underscore the importance of cohesively understanding service needs across CMC, autistic children, and their caretakers.

Objectives: Homelessness is a public health crisis in the United States, yet homelessness prevalence, especially among children and youth, is not well understood. In this study, we use an indirect estimation method known as multiple systems estimation to further evaluate prevalence of youth experiencing homelessness in Denver, Colorado.

Methods: We performed a multiple systems estimation ("capture-recapture") analysis to estimate annual homelessness among youth aged 14 to 17 years in the city and county of Denver, Colorado from 2017 to 2021. We used 3 citywide datasets linked at the individual level and stratified by year, as follows: (1) Denver-Metro Homeless Management Information System, (2) Denver Public Schools McKinney-Vento data, and (3) Colorado's Statewide Automated Child Welfare Information System TRAILS. We identified "known" youth experiencing homelessness from the 3 datasets and developed log-linear models with calculated 95% CIs to estimate the unknown population. We combined these counts to estimate the total youth homeless population and stratified counts by race, ethnicity, and gender.

Results: The known count of youth experiencing homelessness increased from 840 in 2017 to 1040 in 2021. We estimated the total number of youth experiencing homelessness increased from 2880 (95% CI, 2199-3290) in 2017 to 7084 (95% CI, 2826-9046) in 2021. This corresponds with an increase from 10.4% to 25.1% in the total prevalence of youth experiencing homelessness in Denver. We found 74.9% to 83.1% of youth experiencing homelessness were Black/African American or Hispanic.

Conclusions: The scope of youth homelessness increased notably and may be far greater than previously understood. Using multiple systems estimation may improve homelessness prevalence estimates and facilitate more effective resource allocation and service delivery.

Background: Traditional classification of children's and nonchildren's hospitals is based on physical structure and branding. We grouped hospitals with the most similar types of pediatric patients.

Methods: Retrospective analysis of 2.8 million hospitalizations in 3993 hospitals for patients 0 to 20 years in the 2019 Kids' Inpatient Database. After stratifying low-volume hospitals (greater than 100 annual admissions), we grouped the remaining hospitals using K-means clustering by case-mix of neonatal services and pediatric diagnosis diversity (DD).

Results: Clustering distinguished 6 hospital groups. Group 1 (n = 1665 [1.6% of hospitalizations]) represented low pediatric volume hospitals (13 annual pediatric hospitalizations [IQR 3-82]). Group 2 (n = 118 hospitals [1.1% of hospitalizations]) provided no neonatal care, had low DD (12 [IQR 4-34]), and had a median age of 17 years. Group 3 (n = 1156 [19.7% of hospitalizations]) hospitals provided low-severity neonatal care with low DD (13 [IQR 7-19). Group 4 (n = 674 hospitals, [24.0% of hospitalizations]) provided moderate-severity neonatal care (2.2 [ IQR 2.1-2.4]) and increased DD (24 [ IQR 6-34]). Group 5 (n = 238 hospitals [20.5% of hospitalizations]) had a similar severity of neonatal care as group 4 (2.3 [IQR 2.1-2.5]), but 2.7 times greater DD (64 [IQR 55-77]). Group 6 (n = 142 hospitals [33.0% of hospitalizations]) had the highest-severity neonatal care (2.6 [IQR 2.3-3.1]) and the greatest DD (127 [113-140]).

Conclusion: Children receive inpatient care across 6 groups of hospitals, distinguished by neonatal case-mix and DD. Future studies should investigate the utility of these groups for peer comparisons with health care use and outcomes.

Background: Febrile young infants are at risk of invasive bacterial infections (IBIs; bacteremia or bacterial meningitis). American Academy of Pediatrics (AAP) guidelines recommend that when procalcitonin testing is unavailable, C-reactive protein (CRP), absolute neutrophil count (ANC) and temperature should be used to identify low-risk infants. We sought to determine the optimal combination of these inflammatory markers to predict IBI when procalcitonin is unavailable.

Methods: This was a secondary analysis of prospectively collected data for all febrile infants aged 60 days or younger evaluated at a tertiary pediatric emergency department (January 2018 to July 2023). Previously healthy term infants aged 8 to 60 days with rectal temperatures of 38.0°C or greater meeting AAP inclusion/exclusion criteria were analyzed. A decision rule was derived by classification and regression tree analysis with 10-fold cross-validation then compared to AAP-recommended thresholds of ANC ≤ 5200/mm3, CRP ≤ 20 mg/L, and temperature ≤ 38.5°C.

Results: Among 1987 infants, 38 (1.9%) had IBIs. The AAP-recommended thresholds missed no IBIs (sensitivity: 100.0% [95% CI, 88.6%-100.0%]; negative predictive value (NPV): 100.0% [95% CI, 99.5%-100.0%]; specificity: 50.7% [95% CI, 48.5%-53.0%]). Optimal derived thresholds were CRP ≤ 22.2mg/L, temperature ≤ 39.0°C, and ANC ≤ 4500/mm3; urinalysis and age were not selected. The derived rule also missed no IBIs (sensitivity: 100.0% [95% CI, 88.6%-100.0%]; NPV: 100.0% [95% CI, 99.7%-100.0%]); however, specificity improved to 83.8% (95% CI, 82.1%-85.4%). Area under the receiver operating curve for the cross-validated rule (91.9% [95% CI, 91.1%-92.7%]) was higher than at AAP-recommended thresholds (75.4% (95% CI, 74.3%-76.5%]).

Conclusions: The combination of ANC, CRP, and temperature at statistically derived thresholds improved diagnostic accuracy for identifying infants at low risk of IBIs compared to AAP-recommended thresholds.

In response to a record number of immigrant families arriving in the United States through the southern border, a multidisciplinary team at a tertiary care children's hospital developed an inpatient asylum protocol (IAP) whose goals were to identify, screen, and support hospitalized asylum-seeking patients and their families. Identified patients were provided with specialized social work, case management, and legal support and were longitudinally followed after hospital discharge to ensure successful engagement with community resources. A total of 47 patients were enrolled over 2.5 years. Our patient population demonstrated significant food, housing, and legal needs. Medical complexity was also substantial, with half of our patients requiring both inpatient subspecialty consultations and outpatient follow-up and 13% qualifying for our institution's complex care program. As a result of the IAP, all families received social work support that continued after discharge, 93% of families without legal aid on admission were connected with legal resources, and 96% of our patients without a primary care physician on arrival had established one on discharge. At the time of this review, half of our patients and families had graduated from the program, whereas the rest continued to require monthly follow-up because of ongoing needs. This novel program was successful in evaluating and addressing the needs of newly arrived asylum-seeking patients and their families in the inpatient setting. We hope that our protocol will aid other institutions in adopting similar interventions.

Objectives: To assess whether exposure to an infant safe sleep initiative was associated with maternal report of infant safe sleep practice at home and to identify other predictive factors.

Methods: After linking Pennsylvania data on infant safe sleep initiative implementation at 27 hospitals to birth certificate and Pregnancy Risk Assessment Monitoring System (PRAMS) data from 2017 to 2021, we generated descriptive statistics to compare infant safe sleep practice and other characteristics between respondents exposed to the initiative and all other PRAMS respondents with a hospital birth. Using multivariable logistic regression, we modeled the association between exposure to the initiative and maternal self-report of placing their infant to sleep on their back, on a separate surface, without soft objects, or room sharing without bed sharing.

Results: PRAMS respondents who gave birth in a hospital that had implemented the infant safe sleep initiative were more likely to report placing their infant on their back to sleep, on a separate sleep surface, or without soft objects compared with those who were not exposed to the initiative after adjusting for maternal characteristics and birth parameters. No significant effect was observed on room sharing without bed sharing. When the overall number of reported infant safe sleep practices reported by respondents was considered as an outcome, the association with exposure to the safe sleep initiative persisted.

Conclusion: This study demonstrates the association between exposure to a hospital-based infant safe sleep initiative and subsequent safe sleep practice at home. Additional study on the equity of such initiatives may be warranted.