Category: Statistics

Anaesthesiologie. 2026 Jan 30. doi: 10.1007/s00101-026-01648-4. Online ahead of print.

ABSTRACT

BACKGROUND: A regression model to estimate the duration from the onset of resuscitation efforts to the return of spontaneous circulation (ROSC) could help improving both resuscitation care and the quality control of registries. This study aims to evaluate the prediction accuracy and to identify challenges for future model development.

METHOD: Regression models based on M5P, random forest (RF) algorithms and a linear regression (LR) modified using M5P were retrospectively developed using a Belgian cohort of 84 individuals in whom ROSC was achieved. Model performance was assessed using quality metrics, such as the correlation coefficient (CC), coefficient of determination (R²), and root mean square error (RMSE) in a cross-validation approach.

RESULTS: In the cohort 61.9% were male with a mean age of 65.7 years. A shockable rhythm was present in 27.7% of cases and the bystander cardiopulmonary resuscitation (CPR) rate was 48.2%. The no-flow time averaged 5.13 min. The mean time from CPR onset to first defibrillation was 7.81 min and to first medication administration 11.31 min. The ROSC occurred after an average of 16.8 min, the LR showed the highest correlation (0.73, 95% confidence interval, CI 0.72-0.74) and R² (0.53 [0.52-0.55]) along with the lowest RMSE (6.76 min [6.63-6.90]). The M5P yielded similar not significantly different values (CC 0.72 [0.70-0.73], R² 0.52 [0.50-0.53], RMSE 6.84 min [6.69-6.99]). In contrast, RF performed significantly worse (CC 0.62 [0.61-0.63], R² 0.38 [0.37-0.40], RMSE 7.89 min [7.82-7.96], all p < 0.01). Only LR showed no significant difference between predicted and actual values in terms of mean (p = 0.75) and variance (p = 0.15). The proportion of potentially prematurely terminated resuscitation attempts, defined as cases with actual ROSC occurring later than predicted ROSC plus RMSE, ranged from 13% (M5P) to 18% (LR).

CONCLUSION: The duration from the start of CPR to ROSC appears to be a process that is suitable for modelling with machine learning algorithms. At this early stage of development, the individual regression models did not demonstrate sufficient validity possibly due to low sample size and simplified data structure; however, the findings indicated potential for an application as a quality assurance tool to compare actual vs. predicted time to ROSC. Therefore, to increase the robustness the results require further evaluation in a larger cohort with additional variables and improved data quality based on the Utstein criteria.

PMID:41615433 | DOI:10.1007/s00101-026-01648-4

J Gerontol A Biol Sci Med Sci. 2026 Jan 30:glag019. doi: 10.1093/gerona/glag019. Online ahead of print.

ABSTRACT

BACKGROUND: Visual impairment is a potential risk factor for depression and other outcomes in older adults. In population-based studies, vision can be measured using self-report or performance-based visual acuity, but epidemiologic associations often depend on which measure is used.

METHODS: In this Research Practice article, we illustrate the use of propensity scores to harmonize analyses of self-reported and performance-based vision in older adults. Using 2021 data from the National Health and Aging Trends Study (NHATS; n = 2,447), we measured associations between self-reported visual difficulty, distance visual impairment (logMAR >0.3), and depression. To harmonize self-reported and performance-based measures of vision, we modeled distance visual impairment as a function of self-reported vision and covariates and calculated exposure misclassification overlap weights. External validation was conducted using the Canadian Longitudinal Study on Aging (CLSA) and the Longitudinal Aging Study in India (LASI).

RESULTS: Self-reported visual difficulty was associated with depression (adjusted OR 2.32, 95% CI: 1.46-3.69), but distance visual impairment was not (OR 1.41, 95% CI: 0.99-2.01). After exposure misclassification overlap weighting, self-reported vision was no longer associated with depression, and results mirrored the association between distance visual impairment and depression (OR 1.49, 95% CI: 0.93-2.36). Similar findings were observed in CLSA and LASI.

CONCLUSIONS: Associations between vision and depression in older adults differ according to how vision is measured. In studies that measure self-reported vision but not visual acuity, propensity score methods that leverage known relationships between the two can be used to approximate associations between reduced visual acuity and health outcomes.

PMID:41615430 | DOI:10.1093/gerona/glag019

Pharmacoepidemiol Drug Saf. 2026 Feb;35(2):e70323. doi: 10.1002/pds.70323.

ABSTRACT

PURPOSE: This nationwide cohort study examined the effects of discontinuation versus continuation of renin-angiotensin system inhibitors (RASis) on major renal and cardiovascular outcomes after the estimated glomerular filtration rate (eGFR) decreased to below 45 mL/min/1.73 m² in patients with type 2 diabetes and treated with RASis.

METHODS: Using linked Taiwanese databases with claims and clinical data, we identified patients with type 2 diabetes who used RASis during 2016-2020, and either discontinued or continued RASis within 180 days when their eGFR fell below 45 mL/min/1.73 m². The outcomes of interest included end-stage renal disease (ESRD), myocardial infarction, stroke, heart failure, and all-cause mortality. We estimated the hazard ratios (HRs) and 95% confidence intervals (CIs) for RASi discontinuation versus RASi continuation using on-treatment and intention-to-treat analyses and inverse probability weighting to adjust for baseline and time-varying covariates.

RESULTS: We identified 251 853 eligible patients, of whom 37 108 (15%) discontinued RASis and 214 745 (85%) continued RASis. The on-treatment HR associated with RASi discontinuation was 2.52 (95% CI, 2.33-2.73) for ESRD, 1.18 (1.08-1.30) for myocardial infarction, 1.28 (1.19-1.37) for stroke, 1.18 (1.13-1.24) for heart failure, and 1.77 (1.70-1.84) for all-cause mortality. Results from the intention-to-treat analysis were similar, albeit more conservative. Findings remained consistent across eGFR strata (≥ 30 to < 45 and < 30 mL/min/1.73 m²), urine albumin-creatinine ratio categories (≥ 300 and < 300 mg/g), and patient subgroups with various baseline characteristics.

CONCLUSION: Our results support continuing RASi treatment even when the eGFR declines to below 45 mL/min/1.73 m² based on potential renal, cardiovascular, and survival benefits.

PMID:41614370 | DOI:10.1002/pds.70323

Bioinformatics. 2026 Jan 30:btag044. doi: 10.1093/bioinformatics/btag044. Online ahead of print.

ABSTRACT

MOTIVATION: Phylogenetic trees describe the evolutionary history among biological species based on their genomic data. Maximum Likelihood (ML) based phylogenetic inference tools search for the tree and evolutionary model that best explain the observed genomic data. Given the independence of likelihood score calculations between different genomic sites, parallel computation is commonly deployed. This is followed by a parallel summation over the per-site scores to obtain the overall likelihood score of the tree. However, basic arithmetic operations on IEEE 754 floating-point numbers, such as addition and multiplication, inherently introduce rounding errors. Consequently, the order by which floating-point operations are executed affects the exact resulting likelihood value since these operations are not associative. Moreover, parallel reduction algorithms in numerical codes re-associate operations as a function of the core count and cluster network topology, inducing different round-off errors. These low-level deviations can cause heuristic searches to diverge and induce high-level result discrepancies (e.g., yield topologically distinct phylogenies). This effect has also been observed in multiple scientific fields beyond phylogenetics.

RESULTS: We observe that varying the degree of parallelism results in diverging phylogenetic tree searches (high level results) for over 31% out of 10 179 empirical datasets. More importantly, 8% of these diverging datasets yield trees that are statistically significantly worse than the best known ML tree for the dataset (AU-test, p<0.05). To alleviate this, we develop a variant of the widely used phylogenetic inference tool RAxML-NG, which does yield bit-reproducible results under varying core-counts, with a slowdown of only 0 to 12.7% (median 0.8%) on up to 768 cores. For this, we introduce the ReproRed reduction algorithm, which yields bit-identical results under varying core-counts, by maintaining a fixed operation order that is independent of the communication pattern. ReproRed is thus applicable to all associative reduction operations-in contrast to competitors, which are confined to summation. Our ReproRed reduction algorithm only exchanges the theoretical minimum number of messages, overlaps communication with computation, and utilizes fast base-cases for local reductions. ReproRed is able to all-reduce (via a subsequent broadcast) 4.1×106 operands across 48 to 768 cores in 19.7 to 48.61 μs, thereby exhibiting a slowdown of 13 to 93% over a non-reproducible all-reduce algorithm. ReproRed outperforms the state-of-the-art reproducible all-reduction algorithm ReproBLAS (offers summation only) beyond 10 000 elements per core. In summary, we re-assess non-reproducibility in parallel phylogenetic inference, present the first bit-reproducible parallel phylogenetic inference tool, as well as introduce a general algorithm and open-source code for conducting reproducible associative parallel reduction operations.

AVAILABILITY AND IMPLEMENTATION: ReproRed: https://doi.org/10.5281/zenodo.15004918 (LGPL)-Reproducible RAxML-NG version https://doi.org/10.5281/zenodo.15017407 (GPL).

SUPPLEMENTARY INFORMATION: https://doi.org/10.5281/zenodo.15524754.

FUNDING: This project received funding from the Klaus Tschira Foundation, the European Union via European Research Council (ERC) Horizon 2020 research and innovation grant No. 882500, and via the EU ERA Chair (HORIZON-WIDERA-2022-TALENTS-01: 2023-2028) program grant No. 101087081 (Comp-Biodiv-GR). The authors gratefully acknowledge the Gauss Centre for Supercomputing e. V. (www.gauss-centre.eu) for funding this project by providing computing time on the GCS Supercomputer SuperMUC-NG at Leibniz Supercomputing Centre (www.lrz.de).

PMID:41614360 | DOI:10.1093/bioinformatics/btag044

Indian J Dent Res. 2026 Jan 29. doi: 10.4103/ijdr.ijdr_138_25. Online ahead of print.

ABSTRACT

PURPOSE: To evaluate and compare chairside clinical time and acceptance of prefabricated and conventional band and loop space maintainers in children with premature loss of primary molars.

METHODS: Sixty children aged 4-8 years with premature loss of primary first or second molars requiring a space maintainer were randomized into two groups: group 1, conventional band and loop space maintainer (CBLSM), and group 2, prefabricated band and loop space maintainer (PBLSM). The clinical evaluation for patient acceptance was carried out at six months follow-up using a five-point Likert scale. Patient comfort during clinical procedure, chairside clinical time, space loss, gingival index and plaque index were also recorded and compared between the two groups.

RESULTS: After six months, both space maintainers were well accepted by the children, showing no statistically significant difference between the groups (P > 0.05). In group 1, 40% of children, and in group 2, 46.6% of them showed mild discomfort measured using the Sound, Eye, Motor (SEM) scale, while the others were comfortable during the clinical procedure. The mean chairside clinical time required was 14.12 ± 3.90 and 19.96 ± 5.55 min for group 1 and group 2, respectively, and there was a statistically significant difference between the two groups (P < 0.001). The plaque index, the gingival index and space loss after six months had no statistically significant differences between groups.

CONCLUSION: The CBLSM and PBLSM were comparable in terms of patient acceptance, comfort during clinical procedure and space loss after six months. However, the mean chairside clinical time required was significantly higher in PBLSM group.

PMID:41614351 | DOI:10.4103/ijdr.ijdr_138_25

Otolaryngol Head Neck Surg. 2026 Jan 30. doi: 10.1002/ohn.70132. Online ahead of print.

ABSTRACT

OBJECTIVE: To investigate the diagnostic utility of asymmetrical tonsils in detecting tonsillar malignancy.

DATA SOURCES: PubMed, Embase, Scopus, and Cochrane Library; from inception until December 17, 2024.

REVIEW METHODS: We included observational studies of adult/pediatric patients undergoing excisional tonsillectomy or incisional tonsillar biopsy that reported at least one diagnostic accuracy outcome for tonsillar asymmetry in predicting malignancy. We pooled estimates using frequentist univariate random-effects generalized linear mixed models, examined and adjusted for publication bias via visual inspection, Egger’s test, and trim-and-fill, performed influence and cumulative meta-analyses, and used a Bayesian bivariate model as a sensitivity analysis. Outcome measures included the following: sensitivity, specificity, positive/negative likelihood ratio (LR+/LR-), and positive/negative predictive value (NPV/PPV) with 95% confidence interval (95% CI).

RESULTS: Twenty-nine studies (5178 participants) from 422 records were included. The risk of bias was low-moderate. The sensitivity and specificity of tonsillar asymmetry as a diagnostic marker for malignancy were 77.2% (95% CI: 68.6%-84.0%) and 96.4% (95% CI: 91.6%-98.6%), respectively. The LR- was 0.24 (0.17-0.34) and LR+ was 21.44 (8.05-57.0). The NPV and PPV were 99.8% (95% CI: 99.1%-99.9%) and 4.31% (95% CI: 1.83%-9.80%), without considering clinical risks. With concomitant high-risk clinical features such as lymphadenopathy, the PPV (probability of malignancy given asymmetrical tonsils) was 38.5% (30.3%-47.4%). Without other high-risk features, the PPV was 0.16% (0.15%-0.18%). The overall quality of evidence was high.

CONCLUSION: Tonsillar asymmetry has a high specificity and moderate sensitivity for tonsillar malignancy. Due to the low prevalence of malignancy, the probability of malignancy is less than 1% if no other suspicious clinical features are present.

PMID:41614350 | DOI:10.1002/ohn.70132

Nurs Crit Care. 2026 Mar;31(2):e70316. doi: 10.1111/nicc.70316.

ABSTRACT

BACKGROUND: Late-onset sepsis is a major cause of morbidity and mortality in NICUs. This study reports the attributable mortality of late-onset sepsis across different gestational age groups.

AIM: The objective of this study is to assess the attributable mortality of neonatal late-onset sepsis according to gestational age.

STUDY DESIGN: A matched cohort study using a 10-year, single-centre historical cohort with propensity score matching to reduce confounders. Attributable mortality was determined in the total matched cohort and by gestational age: extremely preterm (< 28 weeks), very preterm (28-32 weeks + 6 days), late preterm (33-36 weeks +6 days) and full-term (≥ 37 weeks).

RESULTS: The matched cohort included 4210 neonates. Attributable mortality of late-onset sepsis in the total cohort was 6.1% (95% CI, 4.7-7.8). It was highest in neonates born between 28 and 36 weeks: 7.1% (95% CI, 4.5-9.8) for very preterm, 8.1% (95% CI, 5.1-11.1) for late preterm and 2.8% (95% CI, 0.01-5.8) for term neonates. For extremely preterm newborns, it was 6.6% (95% CI, -1.1 to 14.5).

CONCLUSIONS: Late-onset sepsis significantly impacts neonates, particularly those born between 28 and 36 weeks, even after adjusting for confounders. Larger-scale studies are needed to further understand the impact across gestational ages. Integrating gestational age into risk stratification tools may improve prevention, early detection and management of neonatal late-onset sepsis.

RELEVANCE TO CLINICAL PRACTICE: Gestational age-specific risk profiles enable targeted monitoring and timely intervention for neonates at greatest risk of sepsis. Personalised care strengthens prevention and management, while nursing expertise in this context is vital for early detection and reducing mortality.

PMID:41614337 | DOI:10.1111/nicc.70316

Nurs Crit Care. 2026 Mar;31(2):e70351. doi: 10.1111/nicc.70351.

ABSTRACT

BACKGROUND: Maternal mental health issues such as depression, anxiety and stress are prevalent among mothers of neonates in the neonatal intensive care unit (NICU). These psychological challenges can impact both maternal well-being and the mother-neonate bond.

AIM: This meta-analysis aims to evaluate the effectiveness of nursing interventions in improving maternal mental health outcomes, including depression, anxiety, stress, mother-neonate interaction and sleep quality in NICU settings.

STUDY DESIGN: Systematic review and meta-analysis were conducted in databases including PubMed, Scopus, Cochrane Library, CINAHL and PsycINFO. Data were pooled using standardised mean differences (SMDs) and analysed using a random effects model. Heterogeneity was assessed using I² statistics, and publication bias was evaluated with funnel plots and Egger’s test.

RESULTS: Final number of included studies was 36. Nursing interventions significantly reduced maternal depression (SMD = -0.296, p = 0.002), stress (SMD = -1.326, p < 0.001) and anxiety (SMD = -0.694, p = 0.001). However, no significant effects were observed for mother-neonate interaction (SMD = 0.071, p = 0.903). Heterogeneity was high for stress (I² = 95.0%) and anxiety (I² = 89.2%), while there was no publication bias detected for depression, anxiety or stress.

CONCLUSION: Nursing interventions significantly reduce maternal depression, anxiety and stress in the NICU. However, their impact on mother-neonate interaction and sleep quality remains unclear.

RELEVANCE TO CLINICAL PRACTICE: Given the substantial impact of stress and anxiety on maternal well-being, implementing structured interventions such as psychoeducation, stress management and mindfulness could significantly reduce the psychological burden on mothers.

PMID:41614336 | DOI:10.1111/nicc.70351

Cancer Med. 2026 Feb;15(2):e71555. doi: 10.1002/cam4.71555.

ABSTRACT

BACKGROUND: The optimal therapeutic strategy for patients with T2-3N0-3 M0 or T1N1-3 M0 hypopharyngeal squamous cell carcinoma (HPSCC) and the use of postoperative radiotherapy with or without systemic therapy for patients with T1-2N1M0 HPSCC remain controversial. We aimed to determine whether these additional treatments improve the prognosis in HPSCC.

METHODS: We retrospectively analyzed the databases held by the SEER (surveillance, epidemiology, and end results) program and a tertiary referral center in China to evaluate the survival outcomes of surgical intervention for T2-3N0-3 M0 and T1N1-3 M0 HPSCC and of postoperative radiotherapy for T1-2N1M0 disease.

RESULTS: The SEER contained data for 1235 patients with T2-3N0-3 M0 or T1N1-3 M0 HPSCC, of whom 220 underwent surgery as their first treatment and 737 received non-surgical treatment. There was no statistically significant difference in overall survival (OS) between these two groups. Data were also available for 30 patients in the SEER who were treated by surgery alone (n = 11), surgery plus postoperative radiotherapy (n = 7), or surgery plus postoperative radiotherapy with systemic therapy (n = 12). Similarly, 23 patients at our hospital were identified to have been treated by surgery alone (n = 7), surgery plus postoperative radiotherapy (n = 10), or surgery plus postoperative radiotherapy with systemic therapy (n = 6). The SEER data indicated that postoperative radiotherapy improved OS (hazard ratio 0.281, 95% confidence interval 0.079-0.998; p = 0.036). This finding was supported by the data from our hospital, although the improvement in OS was not statistically significant (hazard ratio 0.360, 95% confidence interval 0.057-2.261; p = 0.224). Postoperative radiotherapy with systemic therapy seemed not to improve OS beyond that achieved by postoperative radiotherapy alone.

CONCLUSIONS: There was no significant difference in OS in patients with T2-3N0-3 M0 or T1N1-3 M0 HPSCC according to whether or not they underwent surgery as first-line treatment. Surgery plus postoperative radiotherapy was associated with a more favorable prognosis than surgery alone in patients with T1-2N1M0 HPSCC.

PMID:41614332 | DOI:10.1002/cam4.71555

J Am Heart Assoc. 2026 Jan 30:e046142. doi: 10.1161/JAHA.125.046142. Online ahead of print.

ABSTRACT

BACKGROUND: The DCP (Diuretic Comparison Project), a pragmatic trial, evaluated whether chlorthalidone compared with hydrochlorothiazide would reduce the risk of nonfatal cardiovascular disease or noncancer-related death. The intent-to-treat analysis found no difference in such comparison (hazard ratio, 1.04 [95% CI, 0.94-1.16]). The objective of the current study is to estimate the per-protocol effect of chlorthalidone (12.5/25 mg daily) compared with hydrochlorothiazide (25/50 mg daily) in preventing major adverse cardiovascular events among older patients with hypertension.

METHODS: The effect of adhering to treatment strategies was assessed by censoring at first instance of nonadherence, defined as a gap (>90-day gap in drug coverage), switch (switching between study medications), and discontinuation (stop taking chlorthalidone/hydrochlorothiazide altogether for >90 days before the end of the study). The primary outcome was a composite of nonfatal myocardial infarction, stroke, heart failure resulting in hospitalization, urgent coronary revascularization for unstable angina, and noncancer-related death. Using inverse probability weighting, we evaluated the per-protocol effect of chlorthalidone and hydrochlorothiazide using DCP trial data.

RESULTS: Nonadherence was found in 5476 (40%) participants; among 8047 (60%) adherents, 3905 (49%) were randomized to chlorthalidone and 4142 (51%) to hydrochlorothiazide. After censoring time when participants deviated from the assigned treatments, the estimated 5-year risk ratio of the composite primary outcome of nonfatal cardiovascular disease and noncancer-related death was 1.36 (95% CI, 0.96-2.12) in chlorthalidone compared with hydrochlorothiazide.

CONCLUSIONS: The per-protocol analysis indicated a lower risk with hydrochlorothiazide compared with chlorthalidone in preventing nonfatal cardiovascular disease and noncancer-related death; however, this difference was not statistically significant using dispensation data to identify adherence.

REGISTRATION: URL: https://www.clinicaltrials.gov; Unique identifier: NCT02185417.

PMID:41614323 | DOI:10.1161/JAHA.125.046142