Category: Statistics

JAMA Netw Open. 2025 Jul 1;8(7):e2518906. doi: 10.1001/jamanetworkopen.2025.18906.

ABSTRACT

IMPORTANCE: Tumor-infiltrating lymphocytes (TILs) are a provocative biomarker in melanoma, influencing diagnosis, prognosis, and immunotherapy outcomes; however, traditional pathologist-read TIL assessment on hematoxylin and eosin-stained slides is prone to interobserver variability, leading to inconsistent clinical decisions. Therefore, development of newer TIL scoring approaches that produce more reliable and consistent readouts is important.

OBJECTIVE: To evaluate the analytical and clinical validity of a machine learning algorithm for TIL quantification in melanoma compared with traditional pathologist-read methods.

DESIGN, SETTING, AND PARTICIPANTS: This multioperator, global, multi-institutional prognostic study compared TIL scoring reproducibility between traditional pathologist-read methods and an artificial intelligence (AI)-driven approach. The study was conducted using retrospective cohorts of patients with melanoma between January 2022 and June 2023 across 45 institutions, with tissue evaluated by participants from academic, clinical, and research institutions. Participants were selected to ensure diverse expertise and professional backgrounds.

MAIN OUTCOMES AND MEASURES: Intraclass correlation coefficient (ICC) values were calculated for the manual and AI-assisted arms using log-transformed data. Kendall W values were calculated for Clark scores (brisk = 3, nonbrisk = 2, and sparse = 1). Reliabilities of ICC and W values were classified as moderate (0.40-0.60), good (0.61-0.80), or excellent (>0.80). AI TIL measurements were dichotomized using the 16.6 and median cutoffs. Univariable and multivariable Cox regression analyses assessed the prognostic value of TIL scores adjusted for clinicopathologic variables.

RESULTS: There were 111 patients with melanoma in the independent testing cohort (median [range] age at diagnosis, 61.0 [25.0-87.0] years; 56 [50.5%] male) who contributed melanoma whole tissue sections. A total of 98 participants evaluated TILs on 60 hematoxylin and eosin-stained melanoma tissue sections. All 40 participants in the manual arm were pathologists, while the AI-assisted arm included 11 pathologists and 47 nonpathologists (scientists). The AI algorithm demonstrated superior reproducibility, with ICCs higher than 0.90 for all machine learning TIL variables, significantly outperforming manual assessments (ICC, 0.61 for AI-derived stromal TILs vs Kendall W, 0.44 for manual Clark TIL scoring). AI-based TIL scores showed prognostic associations with patient outcomes (n = 111) using the median cutoff approach with a hazard ratio (HR) of 0.45 (95% CI, 0.26-0.80; P = .005), and using the cutoff of 16.6, with an HR of 0.56 (95% CI, 0.32-0.98; P = .04).

CONCLUSIONS AND RELEVANCE: In this prognostic study of TIL quantification in melanoma, the AI algorithm demonstrated superior reproducibility and prognostic associations compared with traditional methods. Although the retrospective nature of the cohorts limits demonstration of clinical utility, the publicly available dataset and open-source AI tool offer a foundation for future validation and integration into melanoma management.

PMID:40608341 | DOI:10.1001/jamanetworkopen.2025.18906

JAMA Netw Open. 2025 Jul 1;8(7):e2518960. doi: 10.1001/jamanetworkopen.2025.18960.

ABSTRACT

IMPORTANCE: Secure firearm storage is a key strategy for firearm injury prevention. Understanding trends in firearm storage behaviors over time is critical for developing tailored interventions.

OBJECTIVE: To characterize trends in household firearm storage from 2013 to 2022 in Washington State and to assess time-varying associations between demographic groups and secure storage.

DESIGN, SETTING, AND PARTICIPANTS: This survey study with repeated cross-sectional measures used data from the Behavioral Risk Factor Surveillance System, a state-based survey of adults (aged ≥18 years) living in Washington State that was conducted by telephone and included questions regarding firearm storage in the 2013, 2015, 2016, 2018, 2020, or 2022 cycles. Data were analyzed from August 2024 through April 2025.

EXPOSURES: Demographic variables including sex, age, rurality (urban vs small town or rural), veteran status, and the presence of children in the home.

MAIN OUTCOMES AND MEASURES: The primary outcome was the percentage of adults who reported secure firearm storage (all household firearms stored unloaded and locked). Logistic regression models that included each of the selected demographic variables, year, and interactions between them were used to estimate time-varying associations, with weighted percentages used to account for complex survey design and to be representative of the adult population of Washington State.

RESULTS: A total of 77 275 survey respondents (mean [SD] age, 46.7 [18.9] years) in Washington State were included in the analysis. Of these, 51.7% (95% CI, 51.2%-52.1%) were female. The percentage of adults reporting household firearm presence was 35.7% (95% CI, 34.4%-37.0%) in 2013 and 33.3% (95% CI, 32.5%-34.1%) in 2022. Among households with firearms (n = 27 721), secure storage was reported among 34.9% (95% CI, 32.8%-37.1%) in 2013 and 48.8% (95% CI, 47.2%-50.3%) in 2022. The odds of secure storage significantly increased per year overall (odds ratio [OR], 1.07; 95% CI, 1.06-1.08), and among nearly all subgroups: males (OR, 1.08; 95% CI, 1.07-1.10), females (OR, 1.06; 95% CI, 1.04-1.08), veterans (OR, 1.09; 95% CI, 1.07-1.10), nonveterans (OR, 1.07; 95% CI, 1.05-1.08), those with children (1.08; 95% CI, 1.06-1.10), and those without (OR, 1.07; 95% CI, 1.06-1.08). The odds increased in urban areas (OR, 1.07; 95% CI, 1.05-1.08) and suburban areas or large towns (OR, 1.08; 95% CI, 1.05-1.10) but not in small towns or rural areas (OR, 1.01; 95% CI, 0.97-1.06).

CONCLUSIONS AND RELEVANCE: In this survey study of Washington State adults, the percentage of adults reporting secure storage of household firearms increased from 2013 to 2022, but differences persisted across demographic groups. These findings underscore the need for more tailored interventions to address these disparities.

PMID:40608340 | DOI:10.1001/jamanetworkopen.2025.18960

JAMA Netw Open. 2025 Jul 1;8(7):e2518967. doi: 10.1001/jamanetworkopen.2025.18967.

ABSTRACT

IMPORTANCE: Goal-concordant care (GCC) is recognized as the highest quality of care and most important outcome measure for serious illness research, yet practical methods for measuring it are lacking.

OBJECTIVE: To measure GCC using clinical notes in patients’ medical records.

DESIGN, SETTING, AND PARTICIPANTS: This longitudinal cohort study involved a retrospective medical record review in 3 urban hospitals in a single health system. Participants included adults with a hospital encounter of 3 or more days between April 1 and July 31, 2019, and 50% or higher predicted 6-month mortality risk. Data abstraction occurred from July 2021 through June 2022.

EXPOSURE: Acute care hospitalization and a 50% or higher predicted 6-month mortality risk.

MAIN OUTCOMES AND MEASURES: Pairs of clinicians independently reviewed clinical notes from admission through 6 months or death to classify the care received during each epoch between patients’ documented goals of care (GOC) discussions, into 1 of 4 categories: (1) comfort focused, (2) maintain or improve function, (3) life extension, or (4) unclear. The GOC discussions had been previously classified using the same 4 categories. The primary study outcome was GCC, defined as the alignment of classification of care received and GOC. Secondary outcomes included goal-discordant care, if GOC and care-received classifications were misaligned, and uncertain concordance, if either care received or GOC was classified as unclear or GOC were not documented. Interrater reliability for classification of care received was assessed using Cohen κ statistics.

RESULTS: Among 109 patients (53 female [49%]), the median (IQR) age was 70 (63-79) years. The most common serious illnesses were cardiac disease (76 patients [70%]), metastatic cancer (50 patients [45%]), and chronic kidney disease (42 patients [39%]). Interrater reliability for care-received classification was almost perfect (95% interrater agreement, Cohen κ = 0.92; 95% CI, 0.86-0.99). A total of 398 epochs of care were identified, 198 (50%) of which were classified as goal concordant. Of the remaining 200 epochs, 74 (19%) were classified as goal discordant and 126 (32%) of uncertain concordance. During at least 1 epoch of care over the 6-month follow-up, 85 patients (78%) received care of uncertain concordance and 43 (39%) received goal-discordant care.

CONCLUSIONS AND RELEVANCE: In this cohort study of seriously ill adults, GCC was measured using clinical notes alone. These findings can inform automated text-based classification methods to improve the efficiency and scalability of this method and facilitate pragmatic and reliable measurement of GCC in serious illness research and quality improvement efforts.

PMID:40608339 | DOI:10.1001/jamanetworkopen.2025.18967

JAMA Netw Open. 2025 Jul 1;8(7):e2518977. doi: 10.1001/jamanetworkopen.2025.18977.

ABSTRACT

IMPORTANCE: Primary care practitioners (PCPs) and staff in Veterans Health Administration (VHA) clinics with staffing shortages have reported higher rates of burnout that may be associated with higher workloads. Introducing PCPs through the Clinical Research Hub (CRH) virtual contingency staffing program into these clinics may help reduce burnout but may also disrupt workflows and increase burnout.

OBJECTIVE: To understand how rates of burnout among VHA PCPs and staff vary by staffing level and CRH program use.

DESIGN, SETTING, AND PARTICIPANTS: This survey study used annual, repeated, cross-sectional VHA employee survey data from fiscal years 2018 to 2022 to examine associations between staffing and burnout before and after implementation of the CRH program.

EXPOSURE: Clinical Research Hub virtual contingency PCP visits.

MAIN OUTCOME AND MEASURES: The main outcome was burnout as measured using multilevel, mixed-effects logistic regression to estimate the association between health care system-level PCP staffing and individual-level PCP and staff burnout before and after implementation of the CRH program. An interaction term was used to test the association between program use and burnout in health care systems with full and less-than-full PCP staffing, controlling for PCP, staff, and health care system characteristics. Estimated marginal means of burnout were calculated from model results.

RESULTS: Survey responses from 134 640 PCPs and staff (53% younger than 49 years; 70% female) in 139 VHA health care systems were analyzed. From fiscal years 2018 to 2022, 38% of PCPs and staff experienced burnout, and CRH visits ranged from a median of 0 to 127.6 (IQR, 76.7-237.4) visits per 1000 patients at the health care system level. In estimations derived from the full model, the probability of burnout was higher in clinics without full PCP staffing before program implementation (34.3% [95% CI, 33.4%-35.2%] without full staffing vs 36.5% [95% CI, 35.3%-37.8%] with full staffing) and in the lowest tertile of CRH visits (37.4% [95% CI, 36.4%-38.4%] without full staffing vs 40.2% [95% CI, 38.3%-42.1%] with full staffing). However, burnout did not differ by staffing at higher levels of CRH visits.

CONCLUSIONS AND RELEVANCE: In this survey study of VHA PCPs and staff, the association between low staffing and burnout was mitigated at higher levels of CRH program use, suggesting that contingency PCPs may alleviate high workload in short-staffed clinics.

PMID:40608338 | DOI:10.1001/jamanetworkopen.2025.18977

JAMA Netw Open. 2025 Jul 1;8(7):e2519029. doi: 10.1001/jamanetworkopen.2025.19029.

ABSTRACT

IMPORTANCE: Hospital ratings including the US News & World Report’s Best Hospitals rankings and the Centers for Medicare & Medicaid Services’ (CMS’) Overall Hospital Quality Star Rating (Overall Star Rating) measure different outcomes and are weakly correlated. Therefore, methods for defining and measuring reliable excellence, defined as consistently great performance across all quality measures, are needed.

OBJECTIVE: To assess a measure of reliable excellence using the 45 quality measures reported in the Overall Star Ratings.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study used hospital-level data from the 2023 and 2024 CMS Overall Star Ratings at all US hospitals with a 2023 and 2024 CMS Overall Star Rating.

EXPOSURES: The exposure was the CMS Overall Star Rating summary score, a continuous variable calculated from the weighted z scores of 45 quality measures used in the Overall Star Rating. A total of 100 000 simulations were run in which all US hospitals’ CMS Overall Star Rating summary scores were calculated through summation of z scores from randomly generated measure weights, as opposed to the existing weights used in the Overall Star Rating method.

MAIN OUTCOMES AND MEASURES: Reliable excellence, defined as achieving a 90th percentile (or better) CMS Overall Star Rating summary score on at least 50 000 of 100 000 simulations. The percentage of hospitals achieving reliable excellence was calculated both overall and stratified by CMS Ovearll Star Ratings.

RESULTS: There were 2700 hospitals in the analysis, with 335 5-star hospitals (12.4%), 727 4-star hospitals (26.9%), 799 3-star hospitals (29.6%), 572 2-star hospitals (21.2%), and 267 1-star hospitals (9.9%) in the 2024 CMS Overall Star Rating. A total of 244 of 2700 hospitals (9.0%) met the study definition of reliable excellence, whereas 1287 of 2700 hospitals (47.7%) achieved excellence in at least 1 simulation.

CONCLUSIONS AND RELEVANCE: This cross-sectional study of 2700 US hospitals found that only 244 hospitals (9.0%), including less than two-thirds of the CMS 5-star rated hospitals, were reliably excellent across 100 000 CMS Overall Star Rating scoring simulations using random measure weightings. These findings lend credence to the ubiquity of inconsistent greatness in health care quality and illuminate the need for methods to distinguish hospitals that provide reliably excellent care.

PMID:40608337 | DOI:10.1001/jamanetworkopen.2025.19029

JAMA Netw Open. 2025 Jul 1;8(7):e2519038. doi: 10.1001/jamanetworkopen.2025.19038.

ABSTRACT

IMPORTANCE: Real-time prescription benefit (RTPB) tools provide point-of-care information for clinicians at the time of prescribing and may reduce prescription costs for patients and payers.

OBJECTIVE: To assess trends in prescription use and spending among Medicare Advantage beneficiaries at a national health insurer during the first year of clinician access to an RTPB tool.

DESIGN, SETTING, AND PARTICIPANTS: This cohort study used 2018 to 2020 administrative data from a national insurer to compare prescription fills for beneficiaries receiving prescriptions from clinicians at practices with an RTPB tool with fills prescribed by clinicians without access to the tool. Trends in prescription spending and fills in the year after practices adopted an RTPB tool (in March 2019) were measured using a difference-in-differences design. Data were analyzed from November 2022 to June 2024.

EXPOSURE: Access to an RTPB tool within a national electronic health record software vendor.

MAIN OUTCOMES AND MEASURES: The main outcomes were total prescription spending, beneficiary out-of-pocket spending, and number of prescription fills. Secondary outcomes included percentage of fills with the insurer-owned mail-order pharmacy, percentage of fills with a 90-day supply, and subgroup analyses in drug classes appearing most frequently in the RTPB tool and high-cost prescription drug classes.

RESULTS: The sample included 2 805 060 beneficiaries (mean [SD] age 70.9 [9.2] years; 56.7% female; 14.7% Black individuals; 80.5% White individuals), with mean (SD) monthly out-of-pocket costs of $29.1 ($90.4), total prescription costs of $213.2 ($1066.3), and 2.6 (2.1) prescription fills per month. After introduction of the RTPB tool, there was no change in prescription spending (estimated out-of-pocket spending change, 1.2% [95% CI, -0.7% to 3.0%]; estimated total prescription spending change, 0.5% [95% CI: -0.2% to 1.2%]) or number of prescription fills (estimated change, 0.01 [95% CI, -0.01 to 0.02]) among beneficiaries prescribed medication by clinicians at practices with the RTPB tool.

CONCLUSIONS AND RELEVANCE: In this cohort study of 2.8 million patients, simply providing clinicians access to a RTPB tool was not associated with the anticipated benefits to patients and payers in the first year the tool was released. Further research on how to design and deploy RTPB tools to maximize potential benefits is needed.

PMID:40608336 | DOI:10.1001/jamanetworkopen.2025.19038

JAMA Netw Open. 2025 Jul 1;8(7):e2521158. doi: 10.1001/jamanetworkopen.2025.21158.

ABSTRACT

IMPORTANCE: Umbilical cord blood-derived cells (UCBCs) are increasingly being evaluated for neuroprotective properties in perinatal brain injury.

OBJECTIVE: To report early neurodevelopmental outcomes of extremely preterm infants who received autologous UCBCs in the CORD-SaFe study.

DESIGN, SETTING, AND PARTICIPANTS: This study reports early follow-up on the preplanned secondary aims of a phase 1 safety and feasibility nonrandomized clinical trial conducted between May 2021 and November 2023, with early follow-up completed in August 2024. Participants were infants born at less than 28 weeks’ completed gestation who received autologous UCBCs in the CORD-SaFe study at Monash Children’s Hospital, Australia. A contemporaneous cohort of noninfused infants born during the study period was included for comparison. Data were analyzed from October to December 2024.

INTERVENTION: Autologous UCBC administered intravenously in the second postnatal week of life.

MAIN OUTCOMES AND MEASURES: Infants underwent brain magnetic resonance imaging to assess structure and injury (Kidokoro score) at term-equivalent age. Assessments at 52 to 54 weeks postmenstrual age included General Movements Assessment, Hammersmith Infant Neurological Examination score, and clinical examination to diagnose risk of cerebral palsy.

RESULTS: A total of 23 infants (median [IQR] gestation, 26 [25-27] weeks; median [IQR] birth weight, 748 [645-981] grams; 17 [73.9%] male) were administered UCBCs at a median (IQR) dose of 42.3 (31.1-63.2) million cells/kg. The contemporaneous cohort included 93 infants (median [IQR] gestation, 26 (24-27) weeks; median [IQR] birth weight, 769 [660-1017] grams; 39 [41.9%] male). Median (IQR) Kidokoro score was 2 (1-3) for the UCBCs group and 3 (2-5) for the contemporaneous cohort, with no statistically significant difference observed between the groups (adjusted median difference, 0 [95% CI, -1.78 to 1.78]). No infants in the UCBC group were assessed as high risk for cerebral palsy compared with 6 of 87 assessed infants (6.8%) in the contemporaneous group; however, the difference was not statistically significant (adjusted log odds, 0.31 [95% CI, -0.76 to 1.38]). No differences in Hammersmith Infant Neurological Examination score (adjusted log odds, -1.50 [95% CI, -5.78 to 2.78]) and absent fidgety movements (adjusted odds ratio, 0.24 [95% CI, 0.20 to 3.04]) were observed between groups.

CONCLUSIONS AND RELEVANCE: This phase 1 nonrandomized clinical trial assessing the safety and feasibility of autologous UCBCs in extremely preterm infants did not find significant differences in brain imaging parameters and early neurodevelopmental outcomes between the cell therapy and contemporaneous untreated groups. It was encouraging to note no infants who received UCBCs were assessed as high risk for cerebral palsy at 52 to 54 weeks postmenstrual age, and the absence of high risk for CP merits further study.

TRIAL REGISTRATION: ANZCTR.org.au Identifier: ACTRN12619001637134.

PMID:40608334 | DOI:10.1001/jamanetworkopen.2025.21158

JAMA Oncol. 2025 Jul 3. doi: 10.1001/jamaoncol.2025.1891. Online ahead of print.

ABSTRACT

IMPORTANCE: Non-clear cell renal cell carcinomas (nccRCCs) present considerable challenges owing to their heterogeneity and limited clinical trial representation. Understanding the benefits of combining immunotherapy and targeted therapy for these subtypes is crucial for improving patient outcomes.

OBJECTIVE: To evaluate the efficacy of various first-line immunotherapy combinations and targeted therapy in treating metastatic nccRCC.

DATA SOURCES: A systematic literature search was conducted across PubMed, Embase, and Cochrane Library databases from inception until December 31, 2024, using relevant keywords and medical subject headings terms.

STUDY SELECTION: Studies were included if they involved patients with nccRCC, reported on immune checkpoint inhibitor (ICI)-based therapies, and provided data on objective response rate (ORR), progression-free survival (PFS), overall survival (OS), and disease control rate (DCR).

DATA EXTRACTION AND SYNTHESIS: Two independent reviewers extracted data, with discrepancies resolved by a third expert. Observational study quality was assessed using the Newcastle-Ottawa Scale. A random-effects meta-analysis was performed, and heterogeneity was evaluated using the I2 statistic.

MAIN OUTCOME AND MEASURES: The primary outcomes of interest were ORR, PFS, OS, and DCR.

RESULTS: The analysis included 23 studies encompassing various subtypes of nccRCC. Pooled results indicated an ORR of 26.6% and a DCR of 57.8% for nccRCC treatments. Median PFS was 6.59 months, and the median OS was 21.11 months. ICIs demonstrated significant efficacy in nccRCC, exhibiting marked clinical activity across different subtypes. Although monotherapy with ICIs showed effectiveness, combination therapies yielded superior clinical outcomes.

CONCLUSIONS AND RELEVANCE: This systematic review and meta-analysis found that ICIs, particularly when combined with targeted therapies, showed promising efficacy in treating metastatic nccRCC. These findings support their integration into treatment guidelines and emphasize the importance of personalized treatment strategies. Future research should focus on long-term outcomes, safety profiles, and the identification of biomarkers to optimize patient selection and improve outcomes.

PMID:40608318 | DOI:10.1001/jamaoncol.2025.1891

JAMA Health Forum. 2025 Jul 3;6(7):e251640. doi: 10.1001/jamahealthforum.2025.1640.

ABSTRACT

IMPORTANCE: There is a growing consensus that commercial prices vary in ways that do not reflect quality of care and are a key factor in high health care spending in the US.

OBJECTIVE: To assess the geographic variation in commercial prices relative to Medicare rates for both hospital and professional services at the state and substate levels, estimate the change in these prices and determine which characteristics are associated with higher hospital prices.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study analyzed deidentified aggregated health care claims data for 2 time frames of service, from January 1, 2020, through December 31, 2020, and from June 1, 2022, through May 31, 2023, to construct commercial-to-Medicare price ratios for hospital and professional services at the state and geozip levels (491 geozips correspond to combinations of zip codes in 50 states and the District of Columbia). Multivariable regression models were estimated to assess the association between commercial-to-Medicare relative hospital prices and various market characteristics at the geozip level. Data analysis was conducted from July through November 2024.

EXPOSURES: Exposures defined at the geozip level included hospital and insurer market concentrations, the share of hospitals beds associated with nonprofit hospitals, the share of beds associated with health systems, the presence of a major teaching hospital, mean household income, the share of the population who had public health insurance, and the share who were uninsured.

MAIN OUTCOMES AND MEASURES: Commercial prices relative to Medicare rates for inpatient, outpatient, combined hospital, and professional services.

RESULTS: This cross-sectional study of 1.2 billion claim lines in 2020 and 1.5 billion claim lines from June 2022 through May 2023 found that private insurers’ in-network allowed amounts were 246% (ratio [SD], 2.46 [0.6]) of the Medicare rates for hospital services and 124% (ratio [SD], 1.24 [0.3]) of the Medicare rates for professional services. The mean commercial-to-Medicare price ratio for professional services slightly declined from 2020 to 2022-2023, while the mean (SD) price ratio for hospital services increased by 5.5%, from 2.34 (0.5) in 2020 to 2.46 (0.6) in 2022-2023. There was substantial variation in the commercial-to-Medicare price ratios across states and geozips. Geozips with very high hospital market concentration levels (Herfindahl-Hirschman Index [HHI]>3500) were associated with a commercial-to-Medicare price ratio higher by 0.21 (95% CI, 0.02-0.39; P = .03) relative to geozips with HHI levels lower than 1500, which represents an 8.4% increase above the 2022-2023 mean. High insurer concentration was negatively associated with the commercial-to-Medicare hospital price ratios (-0.13; 95% CI, -0.26 to 0.01; P = .04), whereas having a major teaching hospital in the geozip (0.20; 95% CI, 0.06-0.34; P = .01), being in the highest household income quartile (0.35; 95% CI, 0.13-0.57; P = .002), and the share of the population who were uninsured (0.03; 95% CI, 0.01-0.05; P < .001) were positively associated with price ratios.

CONCLUSIONS AND RELEVANCE: Examination of a major claims database revealed substantial geographic variation in commercial-to-Medicare price ratios and increases in the price ratio for hospital services over time. Substate market and hospital characteristics were also associated with higher commercial-to-Medicare relative prices. These factors, including high hospital market concentration, could be used to identify and target specific areas more amenable to policies aimed at curbing hospital price growth.

PMID:40608307 | DOI:10.1001/jamahealthforum.2025.1640

JAMA Health Forum. 2025 Jul 3;6(7):e251748. doi: 10.1001/jamahealthforum.2025.1748.

ABSTRACT

IMPORTANCE: Medicare beneficiaries dually enrolled in Medicare and Medicaid have some of the highest care needs. Finding ways to support dually eligible beneficiaries in the Medicare Advantage (MA) program has become a policy goal.

OBJECTIVE: To determine if enrollment in different MA plan types is associated with differences in disenrollment.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study included Medicare enrollment data from dually eligible Medicare beneficiaries in 2021. Analyses were conducted between March 2024 and February 2025. Data were analyzed from January through March 2025.

EXPOSURE: Enrollment in different MA plan types, including those that exclusively serve dual-eligible beneficiaries (coordination-only, dual-eligible special needs plans [D-SNPs] and fully integrated D-SNPs [FIDE-SNPs]), standard MA plans that serve dual-eligible and non-dual-eligible beneficiaries, and D-SNP look-alike plans, defined as standard MA plans that primarily enroll dual-eligible beneficiaries.

MAIN OUTCOMES AND MEASURES: One-year disenrollment from one plan to another or to traditional Medicare.

RESULTS: Among 2 698 434 dually eligible beneficiaries in 2021, the mean (SD) age was 66.9 (14.1) years, and 62.5% were female individuals. Of dual-eligible beneficiaries enrolled in FIDE-SNPs in 2021, 19 001 (8.1%) disenrolled by 2022. Of those enrolled in coordination-only D-SNPs, D-SNP look-alikes, and standard MA plans in 2021, disenrollment rates were 18.3%, 30.5%, and 28.2%, respectively. Disenrollment rates were higher for Black beneficiaries and those who used more health services, including inpatient stays and more days of nursing home care.

CONCLUSIONS AND RELEVANCE: The results of this cross-sectional study suggest that FIDE-SNPs retained their members at higher rates, which could be a sign of improved care experiences. Understanding how FIDE-SNPs may be affecting patient care will be important moving forward.

PMID:40608306 | DOI:10.1001/jamahealthforum.2025.1748