Category: Statistics

JAMA Netw Open. 2025 Feb 3;8(2):e2460312. doi: 10.1001/jamanetworkopen.2024.60312.

ABSTRACT

IMPORTANCE: Metaplastic breast cancer (MpBC) is a rare, heterogeneous disease often associated with inferior outcomes. A growing body of literature describes the clinical and molecular features of MpBC, yet limited data describe the pathogenic germline variants (PGVs) in breast cancer susceptibility genes among affected individuals.

OBJECTIVE: To examine the frequency and types of PGVs in breast cancer genes among patients with MpBC.

DESIGN, SETTING, AND PARTICIPANTS: This is a descriptive retrospective cohort study of patients who received a diagnosis of MpBC at the University of Pennsylvania between January 2010 and May 2023. Electronic medical records were reviewed for demographic, clinicopathologic, and germline genetic testing information. Germline variant status was independently confirmed by a licensed genetic counselor and a physician with expertise in genetics. MpBC diagnosis and subtype were confirmed by a breast pathologist. Participants were identified via query of an institutional pathology database for reports signed between January 2010 and May 2023 including the term metaplastic. Among 320 initially obtained reports, 272 individuals had confirmed MpBC and were included in the study.

EXPOSURE: Germline genetic testing to investigate the presence of PGVs in breast cancer susceptibility genes.

MAIN OUTCOMES AND MEASURES: The primary outcome measurement was the prevalence of PGVs in breast cancer susceptibility genes among participants. The hypothesis that individuals with MpBC have an enrichment of PGVs in genes associated with inherited breast cancer risk was formulated before data collection.

RESULTS: The total sample size was 272 women, and the median age at diagnosis was 58 years (range, 20-102 years); all were biological female patients; 143 of 272 (52.6%) had documentation of germline genetic testing; and participants with testing were significantly younger than those without (median age, 53 years [range, 20-79 years] vs 63 years [range, 29-102 years]; P < .001). Of the 143 patients, 24 (16.8%) had a PGV in a breast cancer susceptibility gene (BRCA1, n = 17; BRCA2, n = 5; PALB2, n = 1; CHEK2, n = 1). Patients with PGV-associated MpBC received a diagnosis at a younger age than those with sporadic disease, but there were no significant differences in hormone receptor positivity, ERBB2 status, or metaplastic subtype.

CONCLUSIONS AND RELEVANCE: In this cohort study of patients with MpBC, a substantial proportion of clinically tested patients had a PGV in a breast cancer susceptibility gene, most commonly BRCA1. Germline testing was high yield in patients with MpBC, many of whom would be included in current germline testing eligibility criteria.

PMID:39964682 | DOI:10.1001/jamanetworkopen.2024.60312

JAMA Netw Open. 2025 Feb 3;8(2):e2460315. doi: 10.1001/jamanetworkopen.2024.60315.

ABSTRACT

IMPORTANCE: Acute pulmonary embolism (PE) is a major cause of morbidity and mortality in patients with cancer in the US and worldwide.

OBJECTIVES: To assess the trends in PE-related mortality from 2011 to 2020 among US patients with cancer across age, sex, ethnic and racial groups, urbanicity, and regionality.

DESIGN, SETTING, AND PARTICIPANTS: This cohort study used the Centers for Disease Control and Prevention Wide-Ranging Online Data for Epidemiologic Research data system to determine national trends in age-adjusted mortality rates (AAMRs) due to acute PE among US patients with cancer aged 15 years or older from January 2011 to December 2020. Concomitant trends in cancer mortality and incidence that may have contributed to PE-related mortality were obtained from US Cancer Statistics. Data were analyzed from September to November 2024.

EXPOSURE: PE-related mortality.

MAIN OUTCOMES AND MEASURES: The primary outcome was PE-related deaths among individuals with cancer. AAMRs and cancer incidence were assessed using joinpoint regression modeling, expressed as an average annual percentage change (AAPC) with relative 95% CIs.

RESULTS: From 2011 to 2020, a total of 27 280 194 individuals aged 15 years or older (13 897 519 male [50.9%]; 13 382 675 female [49.1%]) died in the US. The AAMR for PE-related mortality in patients with cancer increased during this time period (AAPC, 2.5%; 95% CI, 1.4% to 3.6%; P = .001), without differences between sexes (P for parallelism = .38). The AAMR increased among those aged 15 to 64 years (AAPC, 3.2%; 95% CI, 1.9% to 4.6%; P = .001), non-Hispanic and non-Latinx White individuals (AAPC, 2.7%; 95% CI, 1.52% to 3.94%; P = .001), non-Hispanic and non-Latinx Black or African American individuals (AAPC, 2.2%; 95% CI, 0.7% to 3.7%; P = .001), Hispanic and Latinx individuals (AAPC, 2.6%; 95% CI, 0.7% to 4.5%; P = .006), and among individuals residing in the Southern US (AAPC, 3.7%; 95% CI, 1.3% to 6.2%; P = .003). During the same period, age-adjusted cancer incidence and cancer-related mortality decreased while the absolute number of new cancer diagnoses and cancer-related deaths increased.

CONCLUSIONS AND RELEVANCE: This cohort study found that despite decreases in cancer-related mortality rates, age-adjusted PE-related mortality in US patients with cancer increased over the last decade; concerning trends included rising PE-related mortality in younger individuals aged 15 to 64 years, particular ethnic and racial groups, and the Southern region of the US. Recognition of such patterns may inform further research into thromboprophylaxis and treatment of PE as a complication of cancer and cancer-directed therapy.

PMID:39964681 | DOI:10.1001/jamanetworkopen.2024.60315

JAMA Netw Open. 2025 Feb 3;8(2):e2460465. doi: 10.1001/jamanetworkopen.2024.60465.

ABSTRACT

IMPORTANCE: Despite the importance of patient trust in health care, there are no patient-reported outcome measures (PROMs) for trust in their clinician that have been developed empirically in Spanish, which is the second most common language in the US.

OBJECTIVE: To develop and validate a Spanish-language PROM for trust in pregnancy care clinician.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study used a national online panel of patients who reported a Spanish language preference and had limited English proficiency and were currently pregnant or had given birth within the 12 months before the survey. Participants resided in the United States, and data were collected from January to May 2024.

EXPOSURES: Participants had clinical interactions during pregnancy and/or postpartum care. Data collected included demographics, Confianza (Trust) Scale candidate items, and 4 measures for concurrent validity evidence: the Trust in Physician Scale (TPS), the Mothers on Respect Index, the Edinburgh Postpartum Depression Scale (EPDS), and the Patient-Reported Outcomes Measurement Information System Global 10.

MAIN OUTCOMES AND MEASURES: The main outcomes were psychometric properties of the Confianza scale and its association with validated scales (validity coefficients). Item response theory (IRT) analyses were conducted to evaluate the psychometric properties of the candidate items, select the best item subset for the Confianza scale, examine its correlation with other measures, and compare scores according to demographic characteristics.

RESULTS: Of the included 204 participants (mean [SD] age, 26 [7] years; 62 participants from South America [30%]; 32 participants from Mexico [16%]), 117 participants were pregnant (57%), and 87 were within 1-year post partum (43%) at the time of survey completion. Four items were removed based on exploratory factor analysis. Using results from IRT analysis on the remaining 12 items, 5 items were selected to represent communication, caring, competency, accompaniment, and overall trust for the final measure. The 5-item Confianza scale had high measurement precision, with reliability above 0.90 across a wide range of the trust continuum. The Confianza scale (mean [SD] score, 21.5 [4.6] out of 25) was positively correlated with the TPS (r = 0.47; 95% CI, 0.36 to 0.57; P < .001) and negatively correlated with the EPDS (r = -0.41; 95% CI, -0.52 to -0.29; P <.001). Higher trust scores were obtained when there was language concordance with clinicians (mean [SD], 23.6 [2.3] vs 20.0 [5.3]; P < .001) and care continuity (mean [SD], 22.3 [3.8] vs 20.9 [5.3]; P = .001).

CONCLUSIONS AND RELEVANCE: In this cross-sectional study of pregnant and postpartum Spanish-speaking individuals, a Spanish-language PROM for trust in pregnancy care clinician had initial validity.

PMID:39964680 | DOI:10.1001/jamanetworkopen.2024.60465

JAMA Netw Open. 2025 Feb 3;8(2):e2460471. doi: 10.1001/jamanetworkopen.2024.60471.

ABSTRACT

IMPORTANCE: Research on cumulative disabilities and tobacco use in the current tobacco landscape is lacking.

OBJECTIVE: To examine prevalence of tobacco use across cumulative disabilities within and across years among US adults.

DESIGN, SETTING, AND PARTICIPANTS: Representative cross-sectional surveys of the US adult (aged ≥18 years) civilian, noninstitutionalized population from the Population Assessment of Tobacco and Health (PATH) Study were compared across 2013 to 2014 (wave 1), 2016 to 2018 (wave 4), and 2022 to 2023 (wave 7). Age-adjusted estimates and change scores were compared using design-based significance tests. Data were analyzed from November 2023 and April 2024.

MAIN OUTCOMES AND MEASURES: Past 30-day use of tobacco (cigarette, electronic nicotine [e-product], traditional cigar, filtered cigar, cigarillo, pipe, hookah, and smokeless products); disability types (mobility, self-care, independent living, cognitive, hearing, and vision) summed to generate a 4-level cumulative disability measure (0, 1, 2, or ≥3).

RESULTS: In 2013 to 2014, among 32 314 participants at wave 1, 3110 participants were 65 years or older (18.2%), 15 993 (51.9%) were female, 5536 (15.2%) were Hispanic, 4496 (11.2%) were non-Hispanic Black, 19 295 (66.0%) were non-Hispanic White, and 2428 (7.5%) were another non-Hispanic race. A total of 32 314 adults were included in 2013 to 2014, 33 638 in 2016 to 2018, and 30 681 in 2022 to 2023. In 2013 to 2014, any tobacco use prevalence for those with no disabilities was 25.4% (95% CI, 24.7% to 26.1%), while prevalence was 40.2% (95% CI, 38.6% to 41.8%), 48.9% (95% CI, 45.4% to 52.5%), and 51.8% (95% CI, 48.6% to 55.0%) for those with 1, 2, and 3 or more disabilities, respectively. Any tobacco, cigarette, e-product, cigarillo, filtered cigar, and pipe use were consistently higher among those with multiple disabilities compared with those with 0 or 1 disability. Prevalence was highest among those with 3 or more disabilities for all products. Tobacco use declined over time, except e-product use increased. However, not all declines in prevalence between 2013 to 2014 and 2022 to 2023 were significant among those with 3 or more disabilities. The declines for those who used any tobacco (-4.6%; 95% CI, -8.7% to -0.6%) and cigarettes (-8.9%; 95% CI, -12.7% to -5.1%) were small, and increases in e-product use were large (4.6%; 95% CI, 1.0% to 8.3%) for those with 3 or more disabilities compared with other groups.

CONCLUSIONS AND RELEVANCE: In this survey study, disparities in tobacco use existed across cumulative disabilities. Despite declines in tobacco use over time, differences in use across cumulative disabilities persisted or worsened, especially among those with 3 or more disabilities. Targeted prevention, screening, and cessation efforts inclusive of multiple products are needed.

PMID:39964679 | DOI:10.1001/jamanetworkopen.2024.60471

Stat Med. 2025 Feb 28;44(5):e10304. doi: 10.1002/sim.10304.

ABSTRACT

Stepped wedge cluster randomized trials (SWCRTs) often face challenges related to potential confounding by time. Traditional frequentist methods may not provide adequate coverage of an intervention’s true effect using confidence intervals, whereas Bayesian approaches show potential for better coverage of intervention effects. However, Bayesian methods remain underexplored in the context of SWCRTs. To bridge this gap, we propose two innovative Bayesian hierarchical penalized spline models. Our first model accommodates large numbers of clusters and time periods, focusing on immediate intervention effects. To evaluate this approach, we compared this model to traditional frequentist methods. We then extend our approach to account for time-varying intervention effects, conducting a comprehensive comparison with an existing Bayesian monotone effect curve model and alternative frequentist methods. The proposed models were applied in the Primary Palliative Care for Emergency Medicine stepped wedge trial to evaluate the effectiveness of the intervention. Through extensive simulations and real-world application, we demonstrate the robustness of our proposed Bayesian models. Notably, the Bayesian immediate effect model consistently achieves the nominal coverage probability, providing more reliable interval estimations while maintaining high estimation accuracy. Furthermore, our proposed Bayesian time-varying effect model represents a significant advancement over the existing Bayesian monotone effect curve model, offering improved accuracy and reliability in estimation while also achieving higher coverage probability than alternative frequentist methods. To the best of our knowledge, this marks the first development of Bayesian hierarchical spline modeling for SWCRTs. Our proposed models offer promising tools for researchers and practitioners, enabling more precise evaluation of intervention impacts.

PMID:39964677 | DOI:10.1002/sim.10304

Stat Med. 2025 Feb 28;44(5):e70004. doi: 10.1002/sim.70004.

ABSTRACT

For clinical prediction applications, we are often faced with small sample size data compared to the number of covariates. Such data pose problems for variable selection and prediction, especially when the covariate-response relationship is complicated. To address these challenges, we propose to incorporate external information on the covariates into Bayesian additive regression trees (BART), a sum-of-trees prediction model that utilizes priors on the tree parameters to prevent overfitting. To incorporate external information, an empirical Bayes (EB) framework is developed that estimates, assisted by a model, prior covariate weights in the BART model. The proposed EB framework enables the estimation of the other prior parameters of BART as well, rendering an appealing and computationally efficient alternative to cross-validation. We show that the method finds relevant covariates and that it improves prediction compared to default BART in simulations. If the covariate-response relationship is non-linear, the method benefits from the flexibility of BART to outperform regression-based learners. Finally, the benefit of incorporating external information is shown in an application to diffuse large B-cell lymphoma prognosis based on clinical covariates, gene mutations, DNA translocations, and DNA copy number data.

PMID:39964672 | DOI:10.1002/sim.70004

Obes Surg. 2025 Feb 18. doi: 10.1007/s11695-025-07743-6. Online ahead of print.

ABSTRACT

BACKGROUND: Laparoscopic Roux-En-Y Gastric Bypass (LRYGB) contributes significantly to higher weight loss at 6 to 12 months when compared to Laparoscopic Sleeve Gastrectomy (LSG) in patients with severe obesity (SO-body mass index (BMI) ≥ 50 kg/m²). However, there is still no consensus regarding the best procedure in terms of mortality and complication rates. We performed a systematic review and meta-analysis to compare the complication rates between these two surgical procedures.

METHODS: PubMed, EMBASE, and Cochrane Central were searched for studies that compared LRYGB and LSG in SO patients. We pooled outcomes for mortality and complications, defined as bleeding, cardiovascular events, conversion to open procedure, and a composite endpoint of leak, abscess, fistulas, and reoperation. Length of stay and operative time were also pooled. A random-effects model was used, and statistical analyses were performed using R version 4.4.0.

RESULTS: A total of 156,767 patients from 28 observational studies were included, of whom 79,324 (50.6%) underwent LRYGB and 77,443 (49.4%) LSG. Length of stay (MD 0.45; 95% CI 0.42-0.48; P < 0.01) and operative time (MD 58.88; 95% CI 37.88-79.87; P < 0.01) were lower in the LSG group. Overall, there was no difference in mortality (OR 1.28; 95% CI 0.80-2.04; P = 0.311) and in complication rates (OR 1.22; 95% CI 0.85-1.76; P = 0.287). A subgroup analysis showed lower conversion to open procedure for patients who underwent LSG (OR 2.75; 95% CI 1.90-3.98; P < 0.001), and no difference was noted in bleeding (OR 0.98; 95% CI 0.47-2.07; P = 0.965), cardiovascular events (OR 0.99; 95% CI 0.43-2.29; P = 0.983), and a composite endpoint of leak, abscess, and fistulas (OR 0.82; 95% CI 0.67-1.01; P = 0.066).

CONCLUSION: Our meta-analysis suggests that there is no difference in mortality and complication rates between the two groups. However, length of stay and operative time were lower in SO patients who underwent LSG.

PMID:39964665 | DOI:10.1007/s11695-025-07743-6

Jpn J Ophthalmol. 2025 Feb 18. doi: 10.1007/s10384-025-01166-7. Online ahead of print.

ABSTRACT

PURPOSE: This study investigated the long-term efficacy and safety of Autologous Platelet Concentrate (APC) injection as an adjunct to pars plana vitrectomy (PPV) in the treatment of macular holes (MHs) with a high risk of surgical failure, including large, highly myopic, or recurrent MHs.

STUDY DESIGN: Retrospective study.

METHODS: This study included patients who underwent PPV with or without APC injection for MHs at Seoul St. Mary’s Hospital between March 2017 and February 2020, with a minimum postoperative follow-up period of 3 years. Anatomical success rates, alterations in foveal microstructure, and improvements in best-corrected visual acuity (BCVA) were evaluated.

RESULTS: Fifty-four patients were included, exhibiting a trend toward enhanced anatomical success in the PPV with APC group (95.0%) compared to the PPV with flap techniques (85.7%) and PPV without the adjunct of flap techniques or APC injection (84.6%), though not statistically significant. Significant findings included increased glial proliferation within the PPV with APC group, sustained for up to one year (P < 0.01), and a significant correlation between glial proliferation and MH closure (P = 0.032). Improvements in BCVA were substantial and found to be significantly associated with preoperative BCVA, the reconstruction of the external limiting membrane (ELM) and ellipsoid zone (EZ) (P = 0.008, 0.010, and 0.047, respectively).

CONCLUSION: The use of APC as adjuvant to PPV can be advantageous in cases of challenging MHs, reflecting comparable anatomical success rates, potential for glial proliferation critical for hole closure, and established safety profile.

PMID:39964664 | DOI:10.1007/s10384-025-01166-7

Brain Imaging Behav. 2025 Feb 18. doi: 10.1007/s11682-025-00982-2. Online ahead of print.

ABSTRACT

Lesions in the basal ganglia present different neuroimaging manifestations compared to other regions. The functional connectivity and white matter (WM) microstructural alterations in patients with left basal ganglia acute ischemic stroke (AIS) remain unknown. This study aimed to explore the alterations of functional connectivity and WM microstructure, as well as their relationship with cognitive performance in patients with left basal ganglia AIS. We acquired resting-state functional MRI (rs-fMRI) and diffusion kurtosis imaging (DKI) data from 41 individuals with left basal ganglia AIS and 41 healthy controls (HC). The degree centrality (DC) method was applied to calculate the functional connectivity and Tract-Based Spatial Statistics was employed to evaluate the voxel-based group differences of diffusion metrics for the values of fractional anisotropy (FA), mean diffusivity, axial diffusivity (AD), radial diffusivity, mean kurtosis (MK), axial kurtosis, and radial kurtosis (RK). AIS showed attenuated DC in the bilateral precuneus and enhanced DC in the left caudate nucleus, compared with HC. In AIS, DC in the left caudate nucleus correlated positively with the Montreal Cognitive Assessment (MoCA) score (r = 0.681, p < 0.05). AIS had significantly decreased FA, AD, MK, and RK in WM tracts, including the internal capsule (IC), genu of corpus callosum (CC), body of CC, left superior longitudinal fasciculus (SLF), left cerebral peduncle, left corticospinal tract, anterior corona radiata (ACR), and left cingulum gyrus (CG). The MK in a cluster including the body of CC, right IC, left cingulate, SLF, ACR, and left CG was also significantly negatively correlated with MoCA scores (r = -0.508, p < 0.05). This study revealed that left basal ganglia AIS not only disrupted the functional connectivity of the whole brain but also had a pervasive impact on the WM microstructure of the whole brain. These findings provide novel insights into the underlying neural mechanisms of early cognitive decline in patients after AIS.

PMID:39964657 | DOI:10.1007/s11682-025-00982-2

J Med Syst. 2025 Feb 18;49(1):26. doi: 10.1007/s10916-025-02161-8.

ABSTRACT

Patients admitted to neonatal intensive care units are up to eight times more likely to experience medication errors than patients admitted to adult intensive care units. Prescribing errors account for up to 74% of medication errors. Electronic prescribing has been postulated as a tool to reduce errors. The objective was to analyse prescribing errors with the e-prescribing system and risk factors. All patients who were admitted for at least 24 h and who received active pharmacological treatment during the study period were included. Prescriptions were made using electronic assisted prescription software integrated into the medical record system. Treatment was reviewed daily by a pharmacist, and errors were graded according to taxonomic criteria. A total of 240 patients were included, 13,876 prescriptions were reviewed and 455 errors were found (3.3% of prescriptions were wrong). Prescribing errors were concentrated in 40 drugs/nutritional products. The most frequent error was a discrepancy between the prescription and the associated text-free field (n = 196). The drugs with the most errors were Lactobacillus acidophilus, caffeine citrate, acetaminophen, gentamycin and cholecalciferol. Patients with a birth weight from 1000 to 1500 g were 82% more likely to experience an error than those with an extremely low birth weight (< 1000 g) (OR = 1.81, 95% CI = 1.42-2.89, p < 0.05). Patients at the highest risk were those with gestational ages from 28 to 32 weeks, with a 29.80% greater risk of prescribing errors than those with gestational ages less than 28 weeks (OR = 1.29, 95% CI = 1.02-1.65, p < 0.05). Prescribing errors occur due to complex dosing rules based on patient characteristics and free-text use, highlighting process issues rather than specific medication risks.

PMID:39964641 | DOI:10.1007/s10916-025-02161-8