Category: Statistics

JAMA Netw Open. 2026 Mar 2;9(3):e262103. doi: 10.1001/jamanetworkopen.2026.2103.

ABSTRACT

IMPORTANCE: Patients with cancer face an increased risk of Clostridioides difficile infection (CDI) due to several factors, including chemotherapy and repeated health care exposures. However, recent national-level data characterizing the prevalence of CDI, associated clinical outcomes, and resource utilization in this population remain limited.

OBJECTIVE: To characterize the hospitalization-level prevalence, clinical outcomes, and health care resource utilization associated with CDI among hospitalized patients with cancer in the US.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study used data from the National Inpatient Sample, a nationally representative database, to identify hospitalizations among adults with a diagnosis of cancer between January 1, 2016, and December 31, 2022. Hospitalizations with CDI were identified using International Statistical Classification of Diseases and Related Health Problems, Tenth Revision codes. Data were analyzed from May to June 2025.

EXPOSURES: Presence of CDI as a primary or secondary diagnosis during the index hospitalization.

MAIN OUTCOMES AND MEASURES: The primary outcome was in-hospital, all-cause mortality. Secondary outcomes included need for kidney replacement therapy, mechanical ventilation, vasopressor support, and colonoscopy. Multivariable logistic regression was used to adjust for demographics, comorbidities, cancer type, and hospital characteristics.

RESULTS: Of 32 083 671 cancer-related hospitalizations (overall study population mean [SD] patient age, 69.4 [13.9] years; 16 050 025 [50.0%] male), 450 360 (1.4%) involved a diagnosis of CDI. Compared with hospitalizations without CDI, those with CDI involved older patients, a higher proportion of women, and higher prevalence of hematologic cancers, cirrhosis, solid organ transplant, bone marrow transplant, chronic kidney disease, and inflammatory bowel disease. Patients with CDI infection had higher in-hospital mortality (7.3% vs 4.5%; adjusted odds ratio [aOR], 1.62; 95% CI, 1.58-1.67) and greater use of critical care interventions. CDI was associated with higher rates of critical care interventions, including kidey replacement therapy (44.3 vs 20.4 per 1000 hospitalizations; aOR, 2.00; 95% CI, 1.92-2.08; P < .001), mechanical ventilation (68.1 vs 35.6 per 1000 hospitalizations; aOR, 1.89; 95% CI, 1.84-1.95; P < .001), and vasopressor use (25.6 vs 11.5 per 1000 hospitalizations; aOR, 2.11; 95% CI, 2.00-2.24; P < .001). Mortality among patients with CDI varied geographically, ranging from 6.4% (95% CI, 6.2%-6.5%) in the Midwest to 8.5% (95% CI, 8.3%-8.7%) in the Northeast.

CONCLUSIONS AND RELEVANCE: In this cross-sectional study of US oncologic hospitalizations, CDI occurred in 1 in 70 cancer-related hospitalizations and was associated with significantly increased mortality and resource utilization. These findings underscore the need for targeted prevention and early intervention strategies in this vulnerable population.

PMID:41879784 | DOI:10.1001/jamanetworkopen.2026.2103

JAMA Netw Open. 2026 Mar 2;9(3):e262443. doi: 10.1001/jamanetworkopen.2026.2443.

ABSTRACT

IMPORTANCE: Large language models (LLMs) are increasingly being applied to analyze clinical data, primarily clinical text, with an increasing emphasis on integration in health care. However, the use of LLMs in pediatric care remains underexplored.

OBJECTIVE: To map the emerging literature on LLM use in pediatrics involving clinical text and identify evidence gaps and future directions for implementation and evaluation.

EVIDENCE REVIEW: PubMed/MEDLINE, Embase, Web of Science, Scopus, and preprint servers were searched for English-language original research published from January 1, 2020, to July 1, 2025. Included studies used modern transformer-based LLMs with pediatric clinical text as input. Two reviewers independently screened studies using predefined criteria. Data were extracted by one reviewer and verified by another. Findings were descriptively synthesized, and adherence to the Minimum Information for Medical AI Reporting (MINIMAR) standards was assessed.

FINDINGS: The review included 40 studies published between 2023 and 2025. Twenty-three studies were conducted in the US, and all were retrospective observational studies using clinical data from sources such as electronic health records. Participant sample sizes ranged from 10 to 172 683. Although all pediatric age subgroups were represented, early childhood populations (aged 0-5 years) were underrepresented. The most common LLM clinical applications were diagnostic decision support in 24 studies (60.0%) and treatment planning in 7 studies (17.5%). Although all 40 studies conducted clinical evaluation of LLMs and 30 included discussions of ethics or data privacy, 39 studies (97.5%) did not meet full MINIMAR standards, 34 (85.0%) did not report use of Health Insurance Portability and Accountability Act-compliant models, and 30 (75.0%) lacked fine-tuning for pediatric-specific data. Among 33 studies assessing model performance against human annotations, 10 (30.3%) did not include clinicians as annotators; among 26 studies with multiple annotators, only 9 (34.6%) reported interannotator agreement statistics.

CONCLUSIONS AND RELEVANCE: This scoping review found that diagnostic decision support and treatment planning were commonly proposed applications of LLMs in pediatrics. However, gaps in scientific rigor and limited use of pediatric-specific data may hinder their safe and effective implementation in pediatrics. Future studies should use standardized evaluation and reporting methods, increase clinician involvement, and expand research to underrepresented ages and clinical applications.

PMID:41879783 | DOI:10.1001/jamanetworkopen.2026.2443

JAMA Netw Open. 2026 Mar 2;9(3):e263050. doi: 10.1001/jamanetworkopen.2026.3050.

ABSTRACT

IMPORTANCE: Although buprenorphine is highly effective at preventing overdose, access is limited in rural areas. Rural primary care is a critical site to expand access; prescribing support tailored for this setting is urgently needed.

OBJECTIVE: To test the feasibility and preliminary effectiveness of a brief buprenorphine prescribing support program (BPSP) tailored for rural primary care professionals (PCPs).

DESIGN, SETTING, AND PARTICIPANTS: This cluster randomized pilot clinical trial compared 27 Ohio community health centers with 63 PCPs (10 physicians, 50 nurse practitioners, and 3 physician assistants) receiving the support program at baseline and immediately after the intervention with those receiving only copies of the American Society of Addiction Medicine’s buprenorphine prescribing guidelines. Data were collected from July 25, 2024, to February 28, 2025.

INTERVENTIONS: The BPSP is a brief, hour-long, asynchronous and online training program, followed by an optional, live booster session, that provides clinical skills on buprenorphine prescribing, addresses misinformation on the safety of buprenorphine, and addresses stigma toward buprenorphine. Health centers were randomized with 2 centers allocated to the intervention for every 1 allocated to the control condition.

MAIN OUTCOMES AND MEASURES: Primary implementation outcomes were feasibility (measured via the 4-item Feasibility of Intervention Measure [scale of 1 to 5, with higher scores indicating greater feasibility]), acceptability (measured via the 4-item Acceptability of Intervention Measure [scale of 1 to 5, with lhigher scores indicating greater acceptability]), and appropriateness (measured via the 4-item Intervention Appropriateness Measure [scale of 1 to 5, with higher scores indicating greater appropriateness]) of the BPSP; willingness to treat opioid use disorder (OUD) in primary care (using a measure developed and tested with a different sample of PCPs); and likelihood of prescribing buprenorphine in the next 6 months (measured via a 5-point Likert scale ranging from 1.00 [extremely unlikely] to 5.00 [extremely likely]).

RESULTS: Of the 63 participating PCPs, 48 participants were allocated to the intervention condition and 15 to the control condition (49 female [78%]; mean [SD] age, 45.5 [11.4] years). Forty-nine participants (78%) practiced in rural areas and 10 (16%) had ever prescribed buprenorphine. Participants receiving the BPSP rated it as highly feasible (median score, 4.25 [IQR, 4.00-5.00]), acceptable (median score, 4.88 of 5.00 [IQR, 4.00-5.00]), and appropriate (median score, 5.00 of 5.00 [IQR, 4.00-5.00]). Participants receiving the BPSP had significantly higher willingness to treat OUD; 86% of rank comparisons improved post intervention. Intention to prescribe buprenorphine in the next 6 months also significantly increased; 98% of rank comparisons improved post intervention. Participants who completed the BPSP showed significant improvements in having correct information about buprenorphine, had greater confidence treating addiction and prescribing buprenorphine, and had lower stigma and greater empathy toward patients with OUD.

CONCLUSIONS AND RELEVANCE: In this cluster randomized clinical trial of the BPSF, brief prescribing support was feasible, acceptable, and appropriate for implementation in rural primary care, and intentions to prescribe buprenorphine increased. A larger trial is needed to confirm preliminary findings.

TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT05505227.

PMID:41879782 | DOI:10.1001/jamanetworkopen.2026.3050

JAMA Netw Open. 2026 Mar 2;9(3):e263410. doi: 10.1001/jamanetworkopen.2026.3410.

ABSTRACT

IMPORTANCE: Despite safety warnings, antipsychotics remain commonly prescribed to older adults, particularly those with dementia, and for off-label indications. However, little is known about how prescribing patterns vary by clinician type over time.

OBJECTIVE: To examine trends in antipsychotic prescribing by clinician type among Medicare beneficiaries from 2013 to 2023.

DESIGN, SETTING, AND PARTICIPANTS: This repeated cross-sectional study used the Medicare Part D Prescribers by Provider and Drug dataset from 2013 to 2023. Participants included Medicare beneficiaries enrolled in Medicare Part D. Analyses were performed between November 2024 and June 2025.

EXPOSURE: Prescribing clinician type, categorized as psychiatrist, primary care physician (PCP), advanced practice registered nurse (APRN) or physician assistant (PA), or other physician.

MAIN OUTCOMES AND MEASURES: Annual antipsychotic prescription claims, the proportion of claims by clinician type, the number of prescribing clinicians, and the mean number of claims per clinician.

RESULTS: From 2013 to 2023, annual antipsychotic prescription claims decreased from 10.8 million to 7.8 million for psychiatrists (average annual percentage change [AAPC], -3.2% [95% CI, -3.7% to -2.7%]; P < .001) and from 7.4 million to 5.7 million for PCPs (AAPC, -2.6% [95% CI, -3.2% to -2.3%]; P < .001). In contrast, claims prescribed by APRNs or PAs more than tripled from 2013 to 2023, increasing from 3.1 million to 9.5 million (AAPC, 11.8% [95% CI, 10.9%-12.7%]; P < .001). The proportion of prescriptions decreased for psychiatrists (48.4% to 32.4%) and PCPs (33.0% to 23.8%) but increased for APRNs and PAs (13.8% to 39.6%) from 2013 to 2023. These shifts were associated primarily with the expanding number of prescribing APRNs and PAs rather than an increase in the mean number of claims per clinician. Although trends were observed in both rural and urban areas, rural areas experienced larger shifts from PCPs to APRNs and PAs, whereas urban areas experienced larger shifts from psychiatrists to APRNs and PAs.

CONCLUSIONS AND RELEVANCE: In this cross-sectional study of Medicare Part D data from 2013 to 2023, antipsychotic prescribing shifted from psychiatrists and PCPs to APRNs and PAs, reflecting growth in nonphysician prescribers. These findings suggest evolving roles in antipsychotic medication management and highlight the need for appropriate training for the full range of clinician types who prescribe these medications to Medicare enrollees.

PMID:41879781 | DOI:10.1001/jamanetworkopen.2026.3410

JAMA Netw Open. 2026 Mar 2;9(3):e263439. doi: 10.1001/jamanetworkopen.2026.3439.

ABSTRACT

IMPORTANCE: Patient-directed discharge (PDD), when patients leave the hospital prior to completing recommended medical treatment, is associated with increased morbidity and mortality and occurs in 10% to 20% of hospitalizations for patients with opioid use disorder (OUD). Understanding risk factors associated with PDD is essential to improving outcomes for this population.

OBJECTIVE: To investigate whether hospitalized patients with OUD who received higher doses of methadone during the first 24, 48, and 72 hours after first contact with the emergency department had decreased odds of PDD.

DESIGN, SETTING, AND PARTICIPANTS: This retrospective observational cohort study was conducted at a single academic health center in the northeastern US during the fentanyl era (July 1, 2019, to June 30, 2022). Hospitalized adults with OUD and without methadone listed in their medication history who received methadone during the first 72 hours were included. Data were analyzed from April 2025 through February 2026.

EXPOSURES: Cumulative dose of methadone received for patients 24, 48, and 72 hours after initial evaluation in the emergency department.

MAIN OUTCOMES AND MEASURES: PDD by 48, 72, or 96 hours or ever, as indicated by discharge disposition in the patient electronic health record.

RESULTS: A total of 554 patients were included in the study. For analysis, participants were separated into cohorts based on cumulative dose by 24 hours (325 patients), 48 hours (488 patients), and 72 hours (454 patients) after presentation to the emergency department, with the main analysis among patients in the 24-hour cohort. Among 325 patients (184 male [56.6%]; median [IQR] age, 49.0 [36.0-59.0] years) receiving methadone within 24 hours of presentation to the emergency department, the incidence of PDD was 45 patients (13.8%). In an adjusted logistic regression model, each additional 10 mg of methadone in the first 24 hours was associated with lower odds of PDD (adjusted odds ratio [aOR], 0.71; 95% CI, 0.44-0.98) at 48 hours. Results were similar for PDD at 72 hours (aOR, 0.68; 95% CI, 0.50-0.85), 96 hours (aOR, 0.72; 95% CI, 0.56-0.88), or ever (aOR, 0.79; 95% CI, 0.67-0.91) in the 24-hour cohort and qualitatively similar but with smaller decreases in odds or nonsignificant outcomes for cumulative methadone dose in the 48-hour cohort (eg, PDD at 96 hours: aOR, 0.91; 95% CI, 0.82-0.99) and nonsignificant outcomes in the 72-hour cohort (eg, PDD at 96 hours: aOR, 0.98; 95% CI, 0.89-1.06).

CONCLUSIONS AND RELEVANCE: In this study, higher cumulative doses of methadone during the first 48 hours of care were associated with substantial reductions in the incidence of PDD. These findings suggest that early and adequate treatment of withdrawal with methadone may be associated with reduced PDD among hospitalized patients with OUD in the fentanyl era.

PMID:41879780 | DOI:10.1001/jamanetworkopen.2026.3439

J Diabetes Investig. 2026 Mar 25. doi: 10.1111/jdi.70268. Online ahead of print.

ABSTRACT

AIMS: To quantify the risk of cardiovascular disease (CVD) and all-cause mortality associated with diabetic foot ulcer (DFU)-related amputation using a large-scale, nationally representative cohort.

MATERIALS AND METHODS: This longitudinal study included individuals with diabetes who underwent standardized national health checkups between 2009 and 2012, with follow-up data obtained from the Korean National Health Insurance Service claims database. CVD was defined as an incident myocardial infarction or stroke occurring after the index health checkup. Cox proportional hazards models were used to estimate the hazard ratios (HRs) for CVD and mortality in individuals with DFU-related amputation compared with those with diabetes but without amputation.

RESULTS: Among 1,229,689 individuals with diabetes, 1,486 (0.12%) had a DFU-related amputation at baseline. The median follow-up was 8.2 years in the non-amputation group and 6.7 years in the DFU-related amputation group. People with DFU-related amputation had a significantly higher incidence rate of CVD compared with those without DFU-related amputation (32.6 vs. 11.0 per 1000 person-years) with a HR of 1.9 (95% CI 1.6-2.1). The HRs for myocardial infarction and stroke were 2.2 (95% CI 1.9-2.6) and 1.6 (95% CI 1.4-1.8), respectively. All-cause mortality significantly increased in people with DFU-related amputation compared with those without, with an HR of 2.1 (95% CI 1.9-2.3).

CONCLUSIONS: In this nationwide population-based cohort, DFU-related amputation was associated with nearly a twofold increased risk of cardiovascular events and mortality. These findings underscore the importance of comprehensive cardiovascular risk assessment and management in patients with diabetic foot complications.

PMID:41879777 | DOI:10.1111/jdi.70268

Psychol Rep. 2026 Mar 25:332941261438074. doi: 10.1177/00332941261438074. Online ahead of print.

ABSTRACT

Long-term reactivated memories undergo a time-dependent reconsolidation process during which they are susceptible to modulation. We report the results of an experiment examining the effect of multiple reminder treatments on reconsolidation of a long-term memory. Participants were trained on a procedural finger tapping task and were either reminded of that training 5 days later or not, immediately prior to learning a new tapping sequence. An additional group was given a reminder treatment of original learning on each of the interim days. Contrary to our hypothesis, results indicate that multiple reminder treatments rendered the training memory significantly more susceptible to disruption in comparison to controls that did not receive any reminders of original learning prior to learning the new motor sequence.

PMID:41879773 | DOI:10.1177/00332941261438074

JAMA. 2026 Mar 25. doi: 10.1001/jama.2026.2075. Online ahead of print.

ABSTRACT

IMPORTANCE: The Oncology Care Model (OCM) was the Centers for Medicare & Medicaid Services’ first cancer-focused alternative payment model, running from 2016 to 2022. The OCM aimed to reduce Medicare spending and improve quality of care for patients receiving chemotherapy.

OBJECTIVE: To evaluate the association of the OCM with changes in Medicare spending, utilization, and quality of care.

DESIGN, SETTING, AND PARTICIPANTS: Difference-in-differences (DID) regression analysis of 6-month chemotherapy episodes attributed to practices voluntarily participating in the OCM or propensity-matched comparison practices, adjusted for beneficiary, episode, practice, and regional characteristics. Episodes for fee-for-service Medicare beneficiaries were grouped into baseline (initiated January 2014-January 2016) and intervention (initiated July 2016-June 2022) periods.

MAIN OUTCOMES AND MEASURES: Total episode payments (Medicare spending for Parts A, B, and D, excluding OCM Monthly Enhanced Oncology Services [MEOS] payments); episode payments for Medicare Parts A, B, and D, hospitalizations, emergency department visits, and measures of quality.

RESULTS: The study population included 739 735 Medicare beneficiaries (mean age, 73.2 [SD, 8.6] years; 59.4% female; 1 746 368 episodes) undergoing chemotherapy (ie, traditional cytotoxic therapy, targeted therapy, immunotherapy, and hormonal therapy) at 202 OCM practices and 830 165 beneficiaries (mean age, 73.1 [SD, 8.8] years; 56.6% female; 1 919 516 episodes) at 534 comparison practices. Total episode payments increased from $29 206 (baseline period) to $36 190 (intervention period) for OCM episodes and from $28 788 to $36 388 for comparison episodes, for an OCM-associated spending change of -$616 [90% CI, -$912 to -$321]). Reductions in total episode payments increased over time (-$1282 in the final 6-month performance period). Statistically significant spending reductions were observed for Part A (DID, -$176 [90% CI, -$288 to -$63]) and Part B (DID, -$340 [90% CI, -$529 to -$149]) but not for Part D (DID, -$53 [90% CI, -$216 to $111]). The OCM was not associated with significant differences in hospitalizations, emergency department visits, or quality. Accounting for MEOS payments and performance-based incentive payments, the OCM resulted in an estimated net loss to Medicare of $639 million over 6 years.

CONCLUSIONS AND RELEVANCE: The OCM was associated with modest reductions in Medicare payments during cancer treatment episodes without significant changes in care quality; payment reductions increased during the program’s last 3 years. However, the OCM incurred a net loss because these estimated savings were exceeded by enhanced services payments and performance-based payments to practices.

PMID:41879763 | DOI:10.1001/jama.2026.2075

JAMA Psychiatry. 2026 Mar 25. doi: 10.1001/jamapsychiatry.2026.0152. Online ahead of print.

ABSTRACT

IMPORTANCE: Methylphenidate is the leading pharmacological treatment for attention-deficit/hyperactivity disorder (ADHD) in childhood and adolescence. Individuals with ADHD have a higher risk of psychosis, but the long-term relationship between methylphenidate and risk of developing psychotic disorders is unknown.

OBJECTIVE: To estimate the relationship between methylphenidate treatment and the risk of nonaffective psychosis in children and adolescents diagnosed with ADHD.

DESIGN, SETTING, AND PARTICIPANTS: This cohort study included instrumental variable analysis of data linkage from multiple national Finnish registries for all individuals born from 1987 to 1997 (n = 697 289). These registries were used to identify childhood and adolescent ADHD diagnoses (age <18 years) from 2003 onwards. Data were analyzed from June 2023 to December 2025.

EXPOSURE: Cumulative amount of treatment with methylphenidate used in 4 intervention windows: within 1, 2, 3, and 4 years after ADHD diagnosis. Hospital district prescribing propensities (average prescribing within each hospital district, within each intervention window) were used as instruments.

MAIN OUTCOME AND MEASURES: Diagnosis of nonaffective psychotic disorder (by code from International Statistical Classification of Diseases and Related Health Problems, Tenth Revision) by the end of follow-up (December 31, 2016). Instrumental variable analyses were conducted using 2-stage least squares modeling and the Anderson-Rubin test. Risk differences (RDs) were estimated for each intervention window.

RESULTS: Among 3956 individuals diagnosed with ADHD (3181 male [80.4%], 775 female [19.6%]; median [IQR] age, 14.16 [11.78-15.93] years), 2728 (69.0%) received methylphenidate at least once. A total of 222 individuals (5.7%) were diagnosed with nonaffective psychosis by mean (SD) age 22.16 (2.39) years (range, 19.00-29.81 years). There was substantial variation in hospital district prescribing propensity (for example, first-year range, 0.07 to 0.30). Instrumental variable analysis indicated that sustained treatment with methylphenidate (30 mg/d) was not associated with the risk of nonaffective psychosis in the overall ADHD sample (1-year RD, -0.14; 95% CI, -0.85 to 0.42; and 4-year RD, -0.15; 95% CI, -0.49 to 0.11). Secondary analyses indicated a reduced risk of nonaffective psychosis among individuals diagnosed in childhood (age <13 years: 3-year RD, -0.24; 95% CI, -0.45 to -0.03; P = .03; 4-year RD, -0.21; 95% CI, -0.48 to -0.07; P = .02). An insufficiently strong instrument precluded the same secondary analyses in those diagnosed in adolescence.

CONCLUSION AND RELEVANCE: This study of national Finnish registry data for individuals with ADHD found no overall relationship between sustained treatment with methylphenidate risk of nonaffective psychosis; in secondary analyses, a potentially protective effect of methylphenidate treatment against later psychosis in children diagnosed with ADHD was found. Further research is needed to evaluate potential effects of treatment in individuals diagnosed in adolescence and adulthood.

PMID:41879751 | DOI:10.1001/jamapsychiatry.2026.0152

JAMA Surg. 2026 Mar 25. doi: 10.1001/jamasurg.2026.0471. Online ahead of print.

NO ABSTRACT

PMID:41879750 | DOI:10.1001/jamasurg.2026.0471