Tag: statistics

JMIR Public Health Surveill. 2025 Sep 30;11:e72591. doi: 10.2196/72591.

ABSTRACT

BACKGROUND: Scalable and accurate screening tools are critical for public mental health strategies, especially in low- and middle-income countries (LMICs). While the Generalized Anxiety Disorder Scale (GAD-7) and Patient Health Questionnaire (PHQ-9) are widely used, their full application in large-scale programs can pose feasibility challenges. By contrast, shorter versions like GAD-2 and PHQ-2 reduce burdens but fail to capture symptom diversity.

OBJECTIVE: This study aimed to optimize screening for anxiety and depression severity using classification and regression tree (CART) models, identifying concise and high-performing decision rules based on the GAD-7 and PHQ-9 items, and to test their reproducibility in 5 independent datasets.

METHODS: A cross-sectional, nonprobabilistic study was conducted with 20,585 Brazilian adults from all 27 states and more than 3,000 cities, collected using digital outreach. Anxiety and depression symptoms were assessed using the GAD-7 and PHQ-9. CART models were trained and tested on bootstrapped samples (70% training, 30% testing), totaling 45,000 trees per scale. Each model used combinations of scale items and sociodemographic predictors. Robustness was evaluated via 10-fold cross-validation and evaluation across 3 hyperparameter configurations (minsplit and minbucket=500, 1000, 2000). Performance metrics included accuracy, sensitivity, specificity, precision, F1-score, and area under the curve (AUC).

RESULTS: The CART models produced concise, high-performing decision rules-using only 2 items for the GAD-7 and 3 for the PHQ-9. No sociodemographic variable appeared in the final classification paths. For GAD-7, the models achieved an accuracy of 86.1% for minimal or mild severity and 85.1% for severe cases, with both categories showing AUC values above 0.900. By contrast, the moderate severity class had lower performance, with accuracy around 51% and an AUC of 0.728. For PHQ-9, the models achieved 81.7% accuracy for minimal or mild cases and 78.8% for severe cases, with AUCs again exceeding 0.900 for the extreme classes; the moderate or moderately severe class showed 66.9% accuracy and an AUC of 0.776. The most frequently repeated rules included the following: “GAD2<2 and GAD4<2” for identifying minimal or mild anxiety and “GAD2≥2 and GAD4=3” for severe anxiety; for depression, “PHQ2<2and PHQ4<2” for minimal or mild cases and “PHQ2≥2 and PHQ8≥2” for severe cases. These rule-based models demonstrated stable performance across thousands of bootstrapped replications and showed reproducibility in 5 independent datasets through external validation.

CONCLUSIONS: CART models enabled simplified, symptom-specific pathways for stratifying anxiety and depression severity with high precision and minimal item burden. These rule-based shortcuts offer an efficient alternative to fixed short forms (eg, GAD-2, PHQ-2) by preserving symptom diversity and severity discrimination. The findings support and lay the groundwork for adaptive, cost-effective screening and intervention models, especially in resource-limited settings and LMICs.

PMID:41027019 | DOI:10.2196/72591

JMIR Mhealth Uhealth. 2025 Sep 30;13:e68824. doi: 10.2196/68824.

ABSTRACT

BACKGROUND: Engagement with digital mental health interventions is often measured as a summary-level variable and remains underresearched despite its importance for meaningful symptom change. This study deepens understanding of engagement in a digital eating disorder intervention, recovery record, by measuring engagement with unique components of the app, on 2 different devices (phone and watch), and at a summary level.

OBJECTIVE: This study described and modeled how individuals engaged with the app across a variety of measures of engagement and identified baseline predictors of engagement.

METHODS: Participants with current binge-eating behavior were recruited as part of the Binge Eating Genetics Initiative study to use a digital eating disorder intervention for 4 weeks. Demographic and severity of illness variables were captured in the baseline survey at enrollment, and engagement data were captured through both an iPhone and Apple Watch version of the intervention. Engagement was characterized by log type (urge, behavior, mood, or meal), device type (logs on phone or watch), and overall usage (total logs) and averaged each week for 4 weeks. Descriptives were tabulated for demographic and engagement variables, and multilevel growth models were conducted for each measure of engagement with baseline characteristics and time as predictors.

RESULTS: Participants (N=893) self-reported as primarily White (743/871, 85%), non-Hispanic (801/893, 90%), females (772/893, 87%) with a mean age of 29.6 (SD 7.4) years and mean current BMI of 32.5 (SD 9.8) kg/m2 and used the app for a mean of 24 days. Most logs were captured on phones (217,143/225,927; 96%), and mood logs were the most used app component (174,818/282,136; 62% of logs). All measures of engagement declined over time, as illustrated by the visualizations, but each measure of engagement illustrated unique participant trajectories over time. Time was a significant negative predictor in every multilevel model. Sex and ethnicity were also significant predictors across several measures of engagement, with female and Hispanic participants demonstrating greater engagement than male and non-Hispanic counterparts. Other baseline characteristics (age, current BMI, and binge episodes in the past 28 days) were significant predictors of 1 measure of engagement each.

CONCLUSIONS: This study highlighted that engagement is far more complex and nuanced than is typically described in research, and that specific components and mode of delivery may have unique engagement profiles and predictors. Future work would benefit from developing early engagement models informed by baseline characteristics to predict intervention outcomes, thereby tailoring digital eating disorder interventions at the individual level.

PMID:41027004 | DOI:10.2196/68824

J Holist Nurs. 2025 Sep 30:8980101251380316. doi: 10.1177/08980101251380316. Online ahead of print.

ABSTRACT

A trauma cluster occurs when two or more natural disasters are experienced simultaneously or consecutively within a short timeframe. With the predicted increase in the frequency and severity of such events, the risk of experiencing trauma clusters is expected to rise, posing significant threats to public health and safety. While the focus of disaster research traditionally centers on negative psychological outcomes, such experiences can also lead to positive effects such as post-traumatic growth and ease. As such, this study aimed to assess the levels of post-traumatic growth and ease, and their relationship, among Bahamians who experienced the trauma cluster of Hurricane Dorian and the coronavirus disease pandemic. Nearly 4 years after the trauma cluster event began, 208 participants completed an online survey including a sociodemographic form, the Post-traumatic Growth Inventory, and Ease Measure. Data were analyzed using descriptive statistics and Pearson’s correlation. Results revealed a broad range of individual post-traumatic growth and ease scores, with a significant positive correlation between overall post-traumatic growth and ease levels. The findings suggest that individuals can respond to adversity in positive ways across various aspects of life. The knowledge gained can inform holistic nursing interventions to support disaster survivors.

PMID:41026997 | DOI:10.1177/08980101251380316

Neurology. 2025 Oct 21;105(8):e214279. doi: 10.1212/WNL.0000000000214279. Epub 2025 Sep 30.

NO ABSTRACT

PMID:41026992 | DOI:10.1212/WNL.0000000000214279

Neurology. 2025 Oct 21;105(8):e214139. doi: 10.1212/WNL.0000000000214139. Epub 2025 Sep 30.

ABSTRACT

BACKGROUND AND OBJECTIVES: Muscle weakness, autoantibodies, and medication use in myasthenia gravis (MG) can complicate pregnancy and delivery. Vaginal delivery is encouraged despite increased risk of acute operative delivery. Our aim was to assess childbirth outcomes related to MG in a current, unselected, nationwide birth cohort.

METHODS: This cohort study included all singleton births in the Medical Birth Registry of Norway from 1999 to 2022. Maternal MG, defined as MG diagnosis or pyridostigmine use, was identified from current and previous pregnancies. We compared adverse pregnancy, delivery, and neonatal outcomes between MG and non-MG births, with mode of delivery as the main outcome. Odds ratios (ORs) with 95% CIs, adjusted for year of birth, maternal age, parity, civil status, and autoimmune comorbidities, were estimated with logistic regression. In subanalyses, we investigated cesarean section (C-section) rates within Robson-10 subgroups and C-section indications.

RESULTS: There were 134 MG births and 1,351,032 non-MG births. Elective C-section was twice as likely if the mother had MG (adjusted OR [aOR] 1.8, 95% CI 1.1-3.1), but emergency C-section and operative vaginal delivery were not more common. MG increased the risk of induction (aOR 1.5, 1.0-2.3), neuraxial anesthesia (aOR 1.5, 1.1-2.1), episiotomy (aOR 1.8, 1.1-3.0), preterm prelabor rupture of membranes (aOR 2.7, 1.1-6.6), prolonged hospitalization after delivery (aOR 1.8, 1.3-2.7), low birthweight (aOR 2.4, 1.3-4.4), feeding problems (aOR 4.9, 2.5-9.5), and transfer to a neonatal unit (aOR 5.1, 3.6-7.2). Transient neonatal MG (TNMG) was diagnosed in 5 of 134 children (4%). In subanalyses, the crude C-section rate was increased in preterm MG births (50% in MG vs 19% in non-MG) and in MG births after previous C-section (75% in MG vs 47% in non-MG).

DISCUSSION: In this registry-based study, the higher rate of elective interventions suggests a proactive management of MG deliveries. The lower C-section threshold for MG births with relative obstetric indications may be justified, but most vaginal deliveries were uncomplicated. Adverse outcomes were generally not increased in the MG group compared with the general population, although episiotomy and prolonged hospitalizations were more frequent. Notably, 40% of MG-exposed infants had feeding difficulties and hypotonia or required neonatal care, suggesting underdiagnosis of TNMG.

PMID:41026991 | DOI:10.1212/WNL.0000000000214139

Neurology. 2025 Oct 21;105(8):e214219. doi: 10.1212/WNL.0000000000214219. Epub 2025 Sep 30.

ABSTRACT

OBJECTIVES: To assess the recent prescribing practices for antiseizure medications (ASMs) to pregnant people with and without epilepsy and prescribers’ preferences beyond the first-line ASMs for patients with refractory generalized epilepsy.

METHODS: A retrospective survey extracted electronic medical data from 20 US tertiary centers (January 1, 2021-December 31, 2022) on ASM prescriptions in pregnant people with and without epilepsy. A second survey asked epileptologists for next-line ASM choices after lamotrigine and levetiracetam failure in women with generalized epilepsy.

RESULTS: The final analyses included 17 centers, 378,808 pregnant people who were prescribed ASMs-12,908 with and 365,900 without epilepsy. Key findings include the following: (1) benzodiazepines represented the most prescribed ASM group (12.13%); (2) gabapentin was the second most prescribed ASM (4.16%); (3) among pregnant people with epilepsy, levetiracetam (28.28%), lamotrigine (15.61%), and gabapentin (11.12%) were most frequent exposures; (4) other common ASMs in this group included topiramate (5.86%), oxcarbazepine (3.32%), lacosamide (2.93%), and zonisamide (2.90%); and (5) valproate use was high, with 492 (3.81%) epilepsy exposures and 1,419 overall. Thirty-two US epileptologists survey responders selected zonisamide and lacosamide as second-line ASM choices for female patients with generalized epilepsy.

DISCUSSION: Valproate use in pregnancy remains high despite its known teratogenicity. Many commonly used ASMs in pregnancy have inadequate safety information.

PMID:41026990 | DOI:10.1212/WNL.0000000000214219

Stat Methods Med Res. 2025 Sep 30:9622802251374838. doi: 10.1177/09622802251374838. Online ahead of print.

ABSTRACT

Randomized controlled trials in oncology often allow control group participants to switch to experimental treatments, a practice that, while often ethically necessary, complicates the accurate estimation of long-term treatment effects. When switching rates are high or sample sizes are limited, commonly used methods for treatment switching adjustment (such as the rank-preserving structural failure time model, inverse probability of censoring weights, and two-stage estimation) may produce imprecise estimates. Real-world data can be used to develop an external control arm for the randomized controlled trial, although this approach ignores evidence from trial subjects who did not switch and ignores evidence from the data obtained prior to switching for those subjects who did. This article introduces “augmented two-stage estimation” (ATSE), a method that combines data from non-switching participants in a randomized controlled trial with an external dataset, forming a “hybrid non-switching arm”. While aiming for more precise estimation, the augmented two-stage estimation requires strong assumptions. Namely, conditional on all the observed covariates: (1) a participant’s decision to switch treatments must be independent of their post-progression survival, and (2) individuals from the randomized controlled trial and the external cohort must be exchangeable. With a simulation study, we evaluate the augmented two-stage estimation method’s performance compared to two-stage estimation adjustment and an external control arm approach. Results indicate that performance is dependent on scenario characteristics, but when unconfounded external data are available, augmented two-stage estimation may result in less bias and improved precision compared to two-stage estimation and external control arm approaches. When external data are affected by unmeasured confounding, augmented two-stage estimation becomes prone to bias, but to a lesser extent compared to an external control arm approach.

PMID:41026983 | DOI:10.1177/09622802251374838

J Med Internet Res. 2025 Sep 30;27:e74177. doi: 10.2196/74177.

ABSTRACT

BACKGROUND: In the era of data and intelligence, artificial intelligence has been widely applied in the medical field. As the most cutting-edge technology, the large language model (LLM) has gained popularity due to its extraordinary ability to handle complex tasks and interactive features.

OBJECTIVE: This study aimed to systematically review current applications of LLMs in lung cancer (LC) care and evaluate their potential across the full-cycle management spectrum.

METHODS: Following PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, we conducted a comprehensive literature search across 6 databases up to January 1, 2025. Studies were included if they satisfied the following criteria: (1) journal articles, conference papers, and preprints; (2) studies that reported the content of LLMs in LC; (3) including original data and LC-related data presented separately; and (4) studies published in English. The exclusion criteria were as follows: (1) books and book chapters, letters, reviews, conference proceedings; (2) studies that did not report the content of LLMs in LC; and (3) no original data, and LC-related data that are not presented separately. Studies were screened independently by 2 authors (SC and ZL) and assessed for quality using Quality Assessment of Diagnostic Accuracy Studies-2, Prediction Model Risk of Bias Assessment Tool, and Risk Of Bias in Non-randomized Studies – of Interventions tools, selected based on study type. Key data items extracted included model type, application scenario, prompt method, input and output format, outcome measures, and safety considerations. Data analysis was conducted using descriptive statistics.

RESULTS: Out of 706 studies screened, 28 were included (published between 2023 and 2024). The ability of LLMs to automatically extract medical records, popularize general knowledge about LC, and assist clinical diagnosis and treatment has been demonstrated through the systematic review, emerging visual ability, and multimodal potential. Prompt engineering was a critical component, with varying degrees of sophistication from zero-shot to fine-tuned approaches. Quality assessments revealed overall acceptable methodological rigor but noted limitations in bias control and data security reporting.

CONCLUSIONS: LLMs show considerable potential in improving LC diagnosis, communication, and decision-making. However, their responsible use requires attention to privacy, interpretability, and human oversight.

PMID:41026980 | DOI:10.2196/74177

Blood Adv. 2025 Sep 30:bloodadvances.2025017254. doi: 10.1182/bloodadvances.2025017254. Online ahead of print.

ABSTRACT

Hydroxyurea provides effective disease-modifying treatment for people with sickle cell anemia (SCA), especially when escalated to maximum tolerated dose (MTD), but has wide interpatient dosing variability due to pharmacokinetic (PK) differences. Whether hydroxyurea PK parameters differ among children with SCA in different global regions is unknown. We compared hydroxyurea PK parameters among children with SCA from five clinical trials: HUSTLE (USA, NCT00305175), TREAT (USA, NCT02286154), NOHARM (Uganda, NCT01976416), REACH (Uganda and Kenya, NCT01966731), and EXTEND (Jamaica, NCT02556099). Key hydroxyurea PK parameters were determined using HdxSim™, a validated hydroxyurea PK-software program. The results were compared across regions by one way analysis of variance. The influence of laboratory and clinical variables on PK-guided doses were evaluated by linear regression. PK profiles from 451 children with SCA were included: 146 from the USA, 265 from Africa, and 40 from the Caribbean. Children from Africa had slightly lower volumes of distribution (p<0.001), but absorption rate (p=0.07) and clearance (p=0.2) were similar across regions. The PK-recommended doses to achieve MTD were statistically different but clinically similar: 26.6 ± 5.9, 27.6 ± 6.5, and 25.2 ± 4.7 mg/kg/day, respectively (p=0.04). In multivariable regression, younger age and increased reticulocyte counts were associated with higher PK-recommended doses. Hydroxyurea PK parameters in children with SCA differ minimally across global populations, predicting clinically similar doses to achieve MTD. Individualized hydroxyurea dosing based on a PK-population model derived from American children with SCA can be used broadly to maximize the benefits of this critical medication in other global populations.

PMID:41026975 | DOI:10.1182/bloodadvances.2025017254

JCO Oncol Pract. 2025 Sep 30:OP2500266. doi: 10.1200/OP-25-00266. Online ahead of print.

ABSTRACT

PURPOSE: Germline testing is recommended for patients with high-risk localized, locally advanced, and metastatic prostate cancer; however, implementation remains suboptimal. Novel strategies are needed to engage oncology clinicians and patients in germline testing.

METHODS: A single-arm investigator-initiated study, conducted from October 1, 2022, to December 31, 2023, used video education as pretest counseling for patients meeting National Cancer Comprehensive Network germline testing criteria. Patients attended a one-on-one in-person session with an educational video, followed by prevideo and postvideo questionnaires assessing knowledge and satisfaction. The primary end point was the proportion who underwent germline testing after intervention, compared with a contemporaneous standard-of-care (SOC) group. Secondary end points included changes in knowledge and perceptions.

RESULTS: All 50 enrolled patients completed the intervention. Patients were predominantly White (78.0%), non-Hispanic (92.0%), English-speaking (98.0%), and college-educated (70%). Most had high-risk localized (46.0%) or metastatic hormone-sensitive prostate cancer (26.0%). Germline testing uptake was significantly higher in the intervention group (82.0%, 41/50) compared with the SOC group (37.94%, 107/282, P < .001). In the intervention group, urologists ordered 22.5% of tests and medical oncologists ordered 77.5%. Pathogenic/likely pathogenic alterations were identified in 10.0% of patients. Most participants (64%) indicated that the most important factor in their decision was whether results could guide treatment. Those who declined testing cited lack of clinical value. Most scored high on the prevideo genetic knowledge test (mean 9.1/11.0) with no statistical difference postvideo (P = .88). Most found the video satisfactory and useful (94%).

CONCLUSION: Germline testing uptake was high after video education. Most patients had high baseline genetic knowledge but were more likely to pursue testing after intervention if it influenced their treatment. Virtual educational aids should be integrated into clinical practice to increase testing rates.

PMID:41026959 | DOI:10.1200/OP-25-00266