Category: Statistics

JAMA Netw Open. 2026 Jul 1;9(7):e2622016. doi: 10.1001/jamanetworkopen.2026.22016.

ABSTRACT

IMPORTANCE: Opioid prescribing has decreased in the US for more than a decade. However, less is known about how prescribing trends differ by prescriber specialty and separately for short-term and long-term opioid prescribing for which the indications, clinical decision-making, and potential for harm vary substantially.

OBJECTIVE: To characterize opioid analgesic prescribing trends in the US between January 1, 2010, and December 31, 2024, by prescriber specialty, analyzing short-term and long-term opioid use episodes separately.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study used administrative claims data for commercially insured and Medicare Advantage patients (aged ≥65 years and <65 years with long-term disability) with at least 90 days of enrollment in medical and prescription coverage. Statistical analysis was performed from May 2025 to September 2025.

MAIN OUTCOMES AND MEASURES: Prescribing volume, daily dose, and prescription duration.

RESULTS: The study identified 110 288 218 opioid fills by 14 959 612 patients during the 15-year study period. Among opioid fills, 59.4% were attributed to female patients; the median (IQR) age at opioid fill was 61 (50-71) years. From 2010 to 2024, decreases in opioid prescribing were greater for long-term than short-term opioid use episodes (eg, commercial cohort: 81.3% reduction [from 33.2 fills per 100-person years in 2010 to 6.2 in 2024] and 74.6% reduction [from 36.2 fills per 100-person years in 2010 to 9.2 in 2024]), respectively. The proportion of all opioid fills written by primary care clinicians decreased (eg, Medicare Advantage with disability cohort: from 52.2% to 47.2%), whereas those written by pain medicine specialists increased (eg, Medicare Advantage with disability cohort: from 17.1% to 25.7%). Distinct trends for short-term and long-term opioid use episodes were observed in prescription characteristics, such as daily dose and duration.

CONCLUSIONS AND RELEVANCE: In this cross-sectional study of opioid prescribing trends in the US between 2010 and 2024, decreases in prescribing were greatest in long-term opioid use episodes, and prescribing was increasingly concentrated among pain medicine specialists. These findings suggest the need to balance reductions in opioid prescribing with provision of nonopioid pain treatments and raise concerns about access to care for patients taking opioids for chronic pain management.

PMID:42412429 | DOI:10.1001/jamanetworkopen.2026.22016

NEJM Catal Innov Care Deliv. 2026 Jan;7(1):CAT240478. doi: 10.1056/CAT.24.0478. Epub 2025 Dec 17.

ABSTRACT

Heart failure remains a leading cause of potentially avoidable hospital admissions throughout the United States and a significant driver of unnecessary, avoidable costs within value-based care programs. In a baseline population of 3233 hospital emergency department (ED) visits for heart failure, which occurred over 12 months ending in June 2023 across six WellSpan Health acute care hospitals, 2868 (88.7%) resulted in inpatient admissions. Among those admissions, 2535 (96.0%) were deemed potentially avoidable by the U.S. Centers for Medicare and Medicaid Services (CMS) Prevention Quality Indicator (PQI) 08 heart failure admission quality indicator. Collectively, these admissions totaled US$27,618,825 in potentially avoidable total costs of care. Through process improvements that modified triaging algorithms related to heart failure exacerbations where shortness of breath or lower extremity edema were present, 92.2% of triaged patients with heart failure were managed in an ambulatory setting, with 84.7% avoiding an ED visit within 24 hours. The 12% cost reduction goal was almost achieved just 1 year into the intervention, with an 11.2% reduction in potentially avoidable admissions (CMS PQI 08, a core quality measure) and a US$3,352,248 reduction in total costs of care. This novel intervention executed with Lean management principles was a pragmatic process improvement that can be readily replicated by other systems to reduce unnecessary, avoidable referrals of mild to moderate heart failure exacerbations to hospital EDs.

PMID:42412420 | DOI:10.1056/CAT.24.0478

Transbound Emerg Dis. 2026;2026(1):e5747471. doi: 10.1155/tbed/5747471.

ABSTRACT

Highly pathogenic avian influenza (HPAI) is a devastating viral disease causing substantial economic losses in the poultry industry and posing potential zoonotic risks. Located along the East Asian-Australasian Flyway (EAAF), South Korea has experienced recurrent outbreaks of HPAI since 2003. Following the severe 2016-2017 epidemic, the government implemented strengthened control measures, including restrictions on duck farming and organizational restructuring. This study quantitatively evaluated structural changes in the spatiotemporal patterns and transmission dynamics of HPAI before and after the 2017 policy reinforcement, utilizing a complete dataset covering 12 epidemic waves between 2003 and 2025. Our analysis suggests a distinct shift in HPAI occurrence patterns from large-scale, clustered epidemics to more sporadic occurrences in the post-2017 period. Pre-2017 epidemics, particularly the 5th and 6th waves, exhibited intense spatiotemporal clustering and high transmission potential. Conversely, post-2017 epidemics showed a significant reduction in outbreak density and the disappearance of large-scale clusters. Notably, the 12th wave displayed a more circular diffusion pattern with outbreaks confined to specific regions, suggesting relatively more geographically contained spread. However, despite the overall reduction in scale, high spatiotemporal interaction intensity was intermittently observed, such as in the 11th wave, indicating that residual risks of explosive local transmission persist even during smaller epidemics. These findings suggest that the post-2017 pattern was temporally consistent with strengthened control policies aimed at reducing mechanical connectivity between farms, although this ecological analysis cannot separate policy effects from other time-varying epidemiological and surveillance-related factors. Nevertheless, a decline in case numbers does not necessarily imply the elimination of local transmission risk, highlighting the need to advance precise and risk-based surveillance and response strategies to effectively manage residual risks.

PMID:42412408 | DOI:10.1155/tbed/5747471

JMIR Form Res. 2026 Jul 7;10:e96755. doi: 10.2196/96755.

ABSTRACT

BACKGROUND: Manual chart abstraction from electronic health records is a critical step in clinical outcomes research but is time-intensive and prone to human error. Advances in artificial intelligence (AI), particularly large language models, offer the potential to automate the extraction of structured data from unstructured clinical documentation with improved efficiency and consistency.

OBJECTIVE: This study aimed to evaluate the accuracy and efficiency of an AI-assisted approach for extracting patient-reported outcomes from clinical notes compared with traditional human abstraction.

METHODS: We conducted a retrospective study of 26 patients treated with low-dose radiation therapy for osteoarthritis. Human reviewers abstracted numeric rating scale (NRS; 0-10) pain scores at baseline, the end of treatment, and the first follow-up, and von Pannewitz score (VPS; 0-4) improvement scores at posttreatment time points. A HIPAA (Health Insurance Portability and Accountability Act)-compliant generative pretrained transformer-based AI system was prompted to extract the same end points from clinical notes. Concordance was assessed using exact match rates, the intraclass correlation coefficient for the NRS, and weighted Cohen κ for the VPS. The time required for AI vs manual abstraction was recorded. The AI system was not trained or fine-tuned on study data, and performance was evaluated directly against human abstraction to reflect real-world deployment.

RESULTS: The AI system demonstrated high concordance with human abstraction, achieving an exact match rate of 92% for the NRS (95% CI 84-96; intraclass correlation coefficient=0.96) and 94% for the VPS (95% CI 84-98; κ=0.91). All discrepancies were minor, and no spurious values were generated. The AI system identified 1 clinically relevant data point missed during manual review. Average abstraction time per patient decreased from approximately 30 minutes to 2 minutes, representing time savings of >90%. The system also captured trends in analgesic use, but these results were not statistically significant, including reductions without escalation.

CONCLUSIONS: AI-assisted data abstraction demonstrated high concordance with human review in this single-institution cohort while substantially reducing the time requirements. These findings support the feasibility of AI-assisted abstraction workflows, although further validation across larger and more diverse datasets is needed.

PMID:42412398 | DOI:10.2196/96755

Neurol Ther. 2026 Jul 7. doi: 10.1007/s40120-026-00991-3. Online ahead of print.

ABSTRACT

INTRODUCTION: In the course of their disease, most patients with Parkinson’s disease (PD) will experience OFF episodes, during which symptoms worsen despite symptomatic treatment. Current on-demand treatments for OFF episodes include inhaled levodopa and subcutaneous and sublingual apomorphine, however no head-to-head comparison of these treatments is available. We performed a network meta-analysis (NMA) to provide robust comparative evidence for on-demand OFF episode treatments.

METHODS: Randomized controlled trials assessing OFF episode treatments in patients with PD were identified in a systematic literature review. A feasibility assessment was conducted considering OFF time, Unified Parkinson’s Disease Rating Scale (UPDRS) Part III scores, Patient Global Impression of Change score (PGI-C), and safety outcomes. An NMA was carried out using a random effects model.

RESULTS: Twenty-one trials were identified in the systematic literature review, 11 of which were included in the feasibility assessment and deemed suitable to be included in the NMA. No statistically significant difference in OFF time was observed between patients receiving inhaled levodopa and those receiving subcutaneous apomorphine, and no statistically significant difference in UPDRS Part III scores or probability of PGI-C score improvements, all-cause discontinuation, or adverse events (AEs) were observed between patients receiving inhaled levodopa, subcutaneous apomorphine, or sublingual apomorphine. Subcutaneous apomorphine had significantly higher probability of treatment discontinuation due to adverse events compared to inhaled levodopa (Log odds ratio 20.712; 95% credible interval 1.855, 55.991).

CONCLUSION: Inhaled levodopa demonstrated no statistically significant difference in efficacy with subcutaneous or sublingual apomorphine, however inhaled levodopa had a lower probability of treatment discontinuation due to AEs than subcutaneous apomorphine. These data highlight that inhaled levodopa is a suitable non-invasive and well-tolerated treatment for OFF episodes.

PMID:42412388 | DOI:10.1007/s40120-026-00991-3

Odontology. 2026 Jul 7. doi: 10.1007/s10266-026-01498-x. Online ahead of print.

ABSTRACT

This study aimed to evaluate the clinical safety, informational completeness, and patient-centered language of responses generated by four large language model (LLM)-based systems to standardized periodontal complaint-based queries. Seven standardized symptom-based periodontal queries were developed based on commonly reported patient complaints and submitted in Turkish to Copilot, Gemini, Claude, and ChatGPT. Responses were evaluated using a structured rule-based framework consisting of a Content Coverage Score (CCS), Risk of Harm Score (RHS), and Patient Language Score (PLS). Assessments were performed by two blinded human reviewers and one blinded AI-based evaluator. Human consensus, AI consensus, and combined evaluator scores were calculated. Agreement between evaluators was assessed using intraclass correlation coefficients (ICC), while human-AI differences and inter-model comparisons were analyzed using non-parametric statistical tests. Excellent agreement was observed between human reviewers, repeated AI evaluations, and human-AI consensus scores (ICC range: 0.911-0.925; p < 0.001). No significant difference was found between overall human and AI consensus scores (p = 0.985). Across the four AI systems, no statistically significant differences were observed in CCS or PLS scores in the human, AI, or combined evaluator analyses (all p > 0.05). PLS scores were generally high across models, indicating good linguistic accessibility for patients. No clearly harmful guidance was identified by the human reviewers. Overall, LLM-based systems generated clinically safe, reasonably comprehensive, and generally patient-accessible responses to common periodontal complaint-based queries. Although these systems may serve as supplementary sources of periodontal health information, they cannot replace individualized clinical evaluation and professional dental consultation.

PMID:42412385 | DOI:10.1007/s10266-026-01498-x

Ther Innov Regul Sci. 2026 Jul 7. doi: 10.1007/s43441-026-01012-z. Online ahead of print.

ABSTRACT

The ICH E9(R1) estimands framework requires precise specification of the treatment effects (estimands) a trial is designed to estimate. A recent article by Troxel et al. [1] has advanced a narrow view that only estimands using the treatment-policy strategy are scientifically defensible. In particular, the article recommends that journals adopt a new policy with regard to reporting results from clinical trials, advocating that only results based on the treatment policy strategy should appear in the main body of the paper with estimates based on other strategies relegated to supplementary materials. Treatment-policy estimands target the effect of assignment to treatment and are defined to include outcomes after non-terminal post-randomization events, such as treatment discontinuation or initiation of alternative or rescue medications. Use of this estimand requires that outcome collection continues following these intercurrent events. In the presence of missing data, estimation of effects using the treatment policy strategy typically relies on strong, unverifiable assumptions [2]. While using treatment policy strategies to address all intercurrent events is appropriate for certain scientific objectives, these estimands do not address all clinically relevant questions. Results based on estimands using alternative strategies for primary and key secondary objectives should therefore also be presented in the main body of the published paper when they address important clinical questions that are relevant to patient care or decision-making.

PMID:42412376 | DOI:10.1007/s43441-026-01012-z

Anal Sci. 2026 Jul 7. doi: 10.1007/s44211-026-00940-y. Online ahead of print.

ABSTRACT

Lysine methylation is an important epigenetic modification that regulates chromatin structure and gene expression. However, it is still difficult to distinguish its methylated states without labels at the single-molecule level. In this study, we investigate the discrimination of lysine methylation states using single-molecule tunneling measurements with gold nano-gap electrodes. The conductance decreases stepwise as the number of methyl groups increases, even though density functional theory (DFT) shows that all molecules have almost the same HOMO energy levels. This result suggests that conductance is not determined only by the electronic structure, but also by how the molecule is arranged between the electrodes. Statistical analysis of current signals shows that high-conductance events become less frequent after methylation, indicating fewer strongly coupled configurations. The relationship between current and molecular length also supports that transport depends on variations in molecular configurations. Machine learning analysis achieved an F-score of 0.76 for distinguishing methylated from unmethylated lysine. In contrast, distinguishing between mono-, di-, and trimethylated forms gave a lower F-score of 0.49, reflecting overlap in the signals. These results suggest that single-molecule tunneling currents are sensitive to stepwise lysine methylation states through differences in transient molecular configurations. This work demonstrates the potential of single-molecule tunneling measurements for label-free analysis of epigenetic modifications.

PMID:42412374 | DOI:10.1007/s44211-026-00940-y

Nano Converg. 2026 Jul 7;13(1):32. doi: 10.1186/s40580-026-00562-0.

ABSTRACT

This study investigates the impact of high-energy proton irradiation (33 and 100 MeV) on the radiation response and statistical uniformity of the breakdown field of Hf_xZr_1-xO₂ based metal-insulator-metal capacitors. While macroscopic electrical characteristics-such as polarization hysteresis, dielectric constant, and leakage current-exhibit remarkable radiation stability across various crystalline phases, a distinct improvement is observed in the statistical distribution of the breakdown field (E_BD). Weibull distribution analysis reveals a consistent increase in the shape factor (β) following irradiation, indicating a “healing effect” that effectively narrows the variance of dielectric breakdown. This enhancement leads to a normalized yield improvement ranging from 3.0% to 14.8%. Our findings suggest that optimized high-energy proton treatment can effectively mitigate localized defects and suppress early-stage failures. These results provide a strategic pathway for enhancing the reliability and operational lifetime of high-k dielectrics in space-qualified and radiation-hardened electronics.

PMID:42412370 | DOI:10.1186/s40580-026-00562-0

J Endocrinol Invest. 2026 Jul 7. doi: 10.1007/s40618-026-02980-7. Online ahead of print.

ABSTRACT

INTRODUCTION: The C-reactive protein to albumin ratio (CAR), an integrative biomarker of inflammation and malnutrition, has shown prognostic value in various diseases, but its role in diabetic chronic kidney disease (CKD) remains inadequately defined. This study aimed to evaluate the independent association of CAR with progression of chronic kidney disease (CKD) and all-cause mortality.

DESIGN AND METHODS: We retrospectively retrieved 231 CKD patients with diabetes who had not received erythropoietin stimulants or iron therapy. The primary outcomes were mortality and progression of CKD (composite), specifically, progression to end-stage renal disease (ESRD) and initiation of renal replacement therapy (RRT), or a doubling of serum creatinine (SCr) levels in patients not receiving RRT. We used Kaplan-Meier survival curves and constructed multivariate Cox proportional hazards models adjusted for potential confounding factors, to estimate the association of CAR with progression of CKD and all-cause mortality; in addition, we employed restricted cubic spline analysis to explore nonlinear relationships. We used the SHAP machine learning algorithm to evaluate the predictive performance of CAR and to analyze the predictive increment of CAR for clinical outcomes. Subgroup analyses were conducted to assess the robustness of the results across different subgroups and modeling choices.

RESULTS: The analysis cohort included a total of 231 adults. Kaplan-Meier curves showed a progressive and significant increase in cumulative CKD progression and mortality across CAR quartiles. In the fully adjusted model of the Cox multivariate regression analysis, a 1-unit increase in CAR (log-transformed) was associated with a 59% increase in the risk of CKD progression (HR = 1.59, 95% CI 1.20-2.09; P = 0.001) and a 32% increase in the risk of mortality (HR = 1.32, 95% CI 1.03-1.68, P = 0.029); participants in the highest quartile had a significantly higher mortality risk compared to those in the lowest quartile (Q4 vs. Q1, HR = 3.8, 95% CI 1.24-11.67; P = 0.02). Restricted cubic spline analysis revealed a significant linear relationship (nonlinear P > 0.05). Subgroup analysis indicated that CAR was consistently associated with outcomes across different age, sex, and BMI groups, with no significant interactions observed, confirming the robustness of these results. In machine learning models, SHAP analysis identified CAR as a key predictor. Compared with the baseline risk model (UTP, eGFR), adding CAR improved predictive performance for CKD progression and mortality, with enhanced C-statistic, improved discriminatory index (IDI), and improved net reclassification index (NRI).

CONCLUSIONS: The CAR serves as a robust, independent predictor of CKD progression and all-cause mortality in patients with diabetic CKD. As a readily accessible biomarker, it holds significant potential to enhance risk stratification and identify candidates warranting intensified clinical management.

PMID:42412363 | DOI:10.1007/s40618-026-02980-7