Search Results (419)

Background/Objectives: Loneliness is a relevant social determinant of health in older age and has been associated with adverse physical, psychological and social outcomes. This study aimed to estimate the prevalence of severe loneliness and to identify its epidemiological and clinical correlates among community-dwelling older adults at risk of falls in Andalusia, Spain. Methods: A cross-sectional analytical study was conducted with 237 adults aged 65 years and older living in private households. Loneliness was assessed using the 10-item version of the UCLA Loneliness Scale, which showed excellent internal consistency in this sample (Cronbach’s alpha = 0.900). After reviewing the scoring direction of the instrument, severe loneliness was operationally defined using the corrected UCLA total score, with higher scores indicating greater loneliness. Sociodemographic and clinical variables included age, sex, living arrangements, economic level, family relationships, pain, sleep quality, depression history, physical exercise, outings from home, polymedication and analgesic use. Bivariate analyses were performed to compare participants with and without severe loneliness, and logistic regression was used to examine independent correlates of severe loneliness. Results: Severe loneliness was identified in 51 participants, corresponding to 21.5% of the analytical sample. The corrected UCLA-10 loneliness burden score ranged from 0 to 30, with higher scores indicating greater loneliness. Participants with severe loneliness were more likely to report poorer sleep quality and severe pain. In the adjusted logistic regression model including age, sex, sleep quality and severe pain, better sleep quality was associated with lower odds of severe loneliness (OR = 0.46, 95% CI: 0.30–0.72, p = 0.001), while severe pain was associated with higher odds of severe loneliness (OR = 3.15, 95% CI: 1.56–6.35, p = 0.001). Age (OR = 1.03, 95% CI: 0.99–1.07, p = 0.184) and female sex (OR = 2.00, 95% CI: 0.83–4.81, p = 0.124) were not statistically significant in the fully adjusted model. Conclusions: Severe loneliness affected a clinically relevant proportion of community-dwelling older adults at risk of falls in Andalusia. The findings should be interpreted as cross-sectional associations and not as evidence of causal pathways. Sleep quality and severe pain emerged as the main independent clinical correlates of severe loneliness in the adjusted model, supporting the relevance of multidimensional assessment in frail older adults. Full article

Background/Objectives: Tooth loss reflects cumulative biological and social processes across the life course. However, population-level causal evidence on the influence of structural social vulnerability and multimorbidity on tooth-loss severity remains limited in middle-income contexts. This study evaluated the causal impacts of social vulnerability and multimorbidity on tooth-loss severity in Chilean adults under explicit potential-outcomes assumptions. Methods: We analyzed nationally representative data from the Chilean National Health Survey 2016–2017 ($N=5165$ adults aged ≥20 years with oral examination; analytic sample $n=4521$). Outcomes comprised ordinal severity ($y_1$: functioning dentition, moderate loss, severe loss, edentulism) and continuous tooth count ($y_2$). Exposures included a Social Vulnerability Index (SVI, 0–1) and Multimorbidity Score (MS, 0–1). We estimated confounder-adjusted proportional-odds and survey-weighted linear regression models. Population-averaged causal contrasts were obtained via g-computation comparing 75th and 25th exposure percentiles, with 95% confidence intervals from probability-proportional-to-size bootstrap (1000 replications). Age-dependent edentulism trajectories were generated using discrete-time Markov projections. Results: In the weighted population, 72.6% retained functional dentition, whereas 5.5% were edentulous. Increasing SVI from 0.091 to 0.345 was associated with a 0.110-point severity increase and 1.95 fewer teeth. Increasing MS from 0.00 to 0.20 was associated with a 0.062-point severity increase and 1.20 fewer teeth. SVI showed larger population-averaged effects than multimorbidity. Conclusions: Within a potential-outcomes framework and under the stated identifying assumptions, structural social vulnerability and multimorbidity each exerted independent effects on tooth-loss severity, with socioeconomic disadvantage showing the stronger distributional gradient across the life course. Because the data are cross-sectional, this causal interpretation is conditional on those assumptions rather than established by the design. Full article

Background: Perioperative outcomes in pelvic organ prolapse (POP) surgery remain difficult to predict due to substantial heterogeneity in both surgical techniques and patient characteristics. Existing studies typically evaluate these factors in isolation, limiting their ability to support individualized risk stratification. This study introduces a stratified analytical framework to disentangle the relative impact of procedural and patient-related determinants across common vaginal reconstructive approaches. Methods: A retrospective cohort of 376 women undergoing POP surgery between 2020 and 2025 was analyzed. Patients were stratified into three procedure groups: sacrospinous fixation with mid-urethral sling (SFM + TOT/TVT), anterior and posterior repair with sling (A&P + TOT/TVT), and isolated anterior and posterior repair (A&P alone). Key outcomes included intraoperative blood loss, length of hospitalization, postoperative hospital stay and catheterization time. Within-group predictors were assessed using stratified odds ratios and synthesized via a random-effects model. Results: Procedure type was consistently associated with recovery-related outcomes, although it explained only a modest proportion of outcome variability. Patients undergoing A&P repair exhibited significantly prolonged hospitalization (8.00 vs. 6.29 and 6.94 days), postoperative recovery (4.99 vs. 3.48 and 4.17 days), and catheterization duration (3.31 vs. 2.33 and 2.86 days) (all p < 0.001). In contrast, intraoperative blood loss was primarily driven by patient-specific factors, including concomitant hysterectomy, prolapse severity, obesity, age, and obstetric history. Prolonged hospitalization was strongly associated with combined procedural complexity and clinical burden, while catheterization duration was influenced by postoperative complications and parity. Conclusions: This study demonstrates that perioperative outcomes in POP surgery arise from distinct and interacting domains: procedural factors predominantly shape recovery trajectories, whereas patient characteristics govern intraoperative risk. The proposed stratified random-effects framework enables integrated evaluation across heterogeneous surgical groups and provides an exploratory basis for identifying domains that may inform future individualized perioperative risk models. Full article

Background/Objectives: Strategies to improve the Social Health Authority (SHA)’s equity can be identified by analyzing the Kenya Demographic and Health Survey (KDHS) 2022. This study reports evidence of determinants of health insurance coverage in Kenya. Methods: Household- and individual-level datasets from the Kenya Demographic and Health Survey conducted between February and July 2022 were combined to form the analyzed dataset. Proportions of individuals with and without health insurance were estimated. The associations between potential determinants and health insurance status were calculated using the Rao–Scott chi-square. Logistic regression was used to analyze the determinants of health insurance coverage. Results: Most of the 14,232 participants were literate (75%), relatively poor (56%), in good health (79%), connected to electricity (55%), and radio listeners (61%). About 34% had health insurance, with 93% of the insured covered by the NHIF. Twenty predictors (Adjusted F = 4.2–434.1, p < 0.0001) were included in the complex sample logistic regression model, but only nine were statistically significant predictors of health insurance coverage. The key predictors were education level; wealth index; ownership of a solar panel, television, smartphone, and computer; age; and recent outpatient care (11–80% differences in odds). Conclusions: Health insurance coverage remains low in Kenya due to low education levels, poor economic status, and disparities in access to media. The SHA can emphasize media campaigns in the informal sector to increase premium payments. Accelerating socioeconomic advancement and adopting tax-based funding could speed up Kenya’s progress towards UHC. Full article

Background: Health accessibility is a key determinant of equitable cancer care. In many countries, specialized oncology services are concentrated in urban and socioeconomically advantaged regions, forcing many patients to travel long distances for treatment. Consequently, geographic and social characteristics may be impactful in determining cancer healthcare outcomes. Objective: The aim of this study was to evaluate the association between the municipal-level Social Progress Index (SPI) and geographic travel burden, stage at diagnosis, treatment, and survival in patients with pancreatic cancer in São Paulo state, Brazil. Methods: We conducted a population-based study using data from “Fundação Oncocentro” on adults with pancreatic adenocarcinoma (2005–2025). The SPI (0–100 scale), a composite measure of municipal social and environmental development, was the primary exposure. It is structured into 3 dimensions and 12 components: Basic Human Needs (nutrition, medical care, water and sanitation, housing, safety); Foundations of Well-being (education, information access, health, environmental quality); and Opportunity (rights, freedom of choice, social inclusion, higher education). Municipal residence and cancer center locations were geocoded, and travel distance (km) was estimated. Multivariable Cox, logistic, and linear regression models assessed associations between SPI and overall survival, stage IV at diagnosis, surgery, and travel distance. Results: A total of 13,478 patients were included (mean follow-up 15.1 ± 27.2 months; mean age 62.3 years; 50.4% male). Stage IV disease was frequent (46.3%), and surgery was performed in 33% of cases. Over half of patients (53.2%) traveled more than 10 km for treatment. Increasing SPI was strongly associated with shorter travel distance (β −62.6 km per SPI unit; p < 0.001) and higher odds of surgery (OR 1.04; p < 0.001) and remained independently associated with a higher likelihood of undergoing surgical treatment (adjusted OR 1.04; p < 0.001). The proportion of stage IV disease did not decrease with increasing SPI and was slightly higher in the highest quartile (49.3%). In survival analysis, SPI demonstrated a protective effect in univariate modeling (HR 0.987; p < 0.001), but lost significance in multivariable analysis (p = 0.125). Travel burden was not retained as an independent predictor of survival after adjustment. Conclusions: Municipal-level SPI was a strong determinant of healthcare access and the likelihood of receiving surgical treatment for pancreatic cancer. Social and geographic vulnerability directly influence care pathways, revealing structural inequities in access to treatment. SPI-based stratification may serve as a practical tool to identify priority regions for transport support and equitable allocation of oncology services. Full article

Introduction: Cerebrovascular disorders are major contributors to maternal morbidity and mortality during pregnancy. In this nationwide study in Hungary, we evaluated the frequency of cerebrovascular diagnoses during first recorded pregnancies in a 17-year period. Material and Methods: In the framework of the NEUROHUN project utilising nationwide administrative healthcare data, we included women with at least one delivery and with at least one cerebrovascular diagnosis during their first pregnancies recorded between 2004 and 2020. To minimise the number of misclassified first pregnancies due to database limitations appearing towards the beginning of the database, trend analyses using linear regression models were restricted to the 2011–2020 period. Results: During first recorded pregnancies in the 17-year study time frame (n = 952,451), the frequency of ICD-10 cerebrovascular diagnoses was 0.17% (n = 1614), with an estimated overall prevalence rate of 169.4 per 100,000 women (95% CI: 161.4–177.9). Transient ischaemic attack (TIA) was the most prevalent specific diagnosis, with a rate of 72.7 per 100,000 (95% CI: 67.4–78.3). In a multiple linear regression model on the mean age at first recorded birth within 2004–2020, women diagnosed with a cerebrovascular disorder were, on average, 1.935 years older at the time of their first birth compared to those without a diagnosis (mean difference: 1.935 years; 95% CI [1.188–2.682], p < 0.001). This analysis, adjusted for calendar year trends between 2011 and 2020, suggests that higher maternal age is an important factor associated with these events. In a sensitivity analysis of the linear regression using the diagnoses of G45, I60, I61, I63, and I67 we found that the relationship between the presence of diagnosis and mean age remained significant in the case of G45, I63 and I67, but not for I60 and I61. In the logistic regression model, compared to the reference group of women < 25 years, the prevalence for all evaluable cerebrovascular diagnoses was significantly higher in the >34 age group, and was also significantly higher for TIA (G45) and cerebral infarction (I63) diagnoses in the 25–34 age group. The rate of cerebral infarction among cerebrovascular disorders showed an increasing trend towards higher maternal age (<25 years age group: 12%; 25–34 years age group: 16.5%; >34 years age group: 20.0%). Also, when compared to the reference category of diagnosed women < 25 years, the increase in the odds of cerebral infarction was significant at the 5% level among women > 34 years. In contrast, there was no increment in the proportion of intracranial bleedings at older age. Discussion and Conclusion: The prevalence of most cerebrovascular diagnoses increases significantly with higher maternal age. Allowing for the limitations of our study, we found that in a Central-Eastern European population, the prevalence of cerebrovascular diagnoses during first recorded pregnancies between 2004 and 2020 was 169.4 per 100,000 (0.17%), with TIA being the most common diagnosis in approximately one-third of cases. The rate of cerebral infarctions among cerebrovascular diagnoses was almost twice as high in those over 34 years of age compared to those below 25. The frequency of pregnancy-related ischemic strokes and cerebral haemorrhages in the Central-Eastern European population corresponds to published values. Full article

Abandoned, lost or otherwise discarded fishing gear (ALDFG) is a global challenge that negatively affects marine ecosystems and fishers’ livelihoods. Survey results from 1864 capture fishers in Vietnam show that the total mass of ALDFG is 82,720 kg/year, with an average loss rate of 0.24% of the total plastic used. The average ALDFG mass from gillnet fisheries is 26.0 ± 1.2 kg/vessel/year, trap fisheries 16.1 ± 2.6 kg/vessel/year, hook and line fisheries 13.8 ± 1.4 kg/vessel/year, trawl fisheries 8.8 ± 1.1 kg/vessel/year, stick-held falling net fisheries 5.1 ± 2.1 kg/vessel/year, and purse seine fisheries 4.2 ± 1.4 kg/vessel/year. Polyethylene (PE) accounts for 47.4%, polyamide (PA) 29.5% and polypropylene (PP) 17.2% of the total ALDFG mass. Results from the selected partial proportional odds model show that fishers’ concern about ALDFG increases with education, fishing experience, vessel length, and stronger awareness of environmental and economic impacts, whereas age, days at sea, perceived navigation hazards, and perceived impacts on fish stocks are negatively associated with concern; perceived tourism impacts show a statistically significant threshold-varying effect across concern levels. These findings provide a scientific basis for developing sustainable fishing-gear management policies, aiming to reduce marine plastic pollution in Vietnam. These findings provide evidence for conservation-oriented fishing-gear management in Vietnam by identifying the gear types and fisher perceptions most relevant to reducing ghost fishing, marine plastic leakage, and risks to marine biodiversity and coastal ecosystems. Full article

Widespread exposure to per- and polyfluoroalkyl substances (PFAS) is a growing public health concern, but its link to muscle damage remains largely unexplored. As PFAS exposure is associated with liver dysfunction, which is an established risk factor for muscle damage, we examined their associations and potential mediating pathways. A total of 1261 participants were recruited from Guangdong province, China, from November 2018 to August 2019 and examined for muscle mass, strength, serum PFAS levels, and biomarkers of liver function. The key results demonstrated significant positive associations between serum PFAS exposure and sarcopenia risk. Specifically, a per ln ng/mL increase in linear perfluorooctane sulfonate (PFOS), branch PFOS, and perfluorooctanoic acid (PFOA) was associated with adjusted odds ratios of 2.32 (95% CI: 1.77 to 3.00), 2.18 (95% CI: 1.67 to 2.90) and 3.01 (95% CI: 1.96 to 4.70), respectively. Analysis of PFAS mixtures via the BKMR model revealed a linear dose–response relationship of sarcopenia, with PFOS and PFOA being the primary contributor. Importantly, mediation analyses showed that liver function biomarkers served as significant mediators of the PFAS–sarcopenia association. Notably, liver synthesis function markers (albumin and globin) mediated a substantial proportion of the association, ranging from 3.48% to 82.42%, whereas liver injury markers (aspartate aminotransferase and gamma-glutamyl transferase) accounted for only 1.54% to 15.44%. This study underscores the need to be aware of the increased risk of muscle damage associated with PFAS exposure, which may primarily operate through liver function abnormalities. Full article

Background/Objectives: Intra-articular hyaluronic acid (HA) injections are widely used to treat pain and functional limitations in knee osteoarthritis (KOA). However, evidence on their real-world effectiveness in routine clinical practice and their impact on analgesic medication use remains limited. This study assessed changes in analgesic prescribing following HA injections in a large real-world outpatient population. Methods: This retrospective cohort study used data from the IQVIA™ Disease Analyzer database in Germany. Patients with knee osteoarthritis who received a first 20 mg hyaluronic acid injection between 2011 and 2024 and had at least three months of observation before and after treatment were included. Outcomes were (1) the proportion of patients with any analgesic prescription (EPHMRA ATC codes M01A, N02A, N02B) and (2) changes in the number of prescribed analgesic pills within three months before versus after injection. Multivariable logistic regression models evaluated factors associated with post-treatment analgesic prescriptions and reductions in pill counts. Results: A total of 4696 patients were included (mean age 64.5 years; 53.9% women), including 524 treated with Recosyn and 4172 treated with other HA products. The proportion of patients with analgesic prescriptions decreased from 28.6% before injection to 26.2% after injection (p = 0.004). Among Recosyn-treated patients, the proportion declined from 25.0% to 20.0% (p = 0.028). Overall, 70.8% of patients with baseline analgesic use achieved a ≥10% reduction in pill counts and 70.0% achieved a ≥20% reduction. In multivariable analyses, treatment with Recosyn was associated with lower odds of receiving an analgesic prescription after injection (OR 0.67; 95% CI 0.53–0.85) and higher odds of achieving a ≥10% reduction in analgesic pill count (OR 1.64; 95% CI 1.05–2.58). Conclusions: In routine outpatient practice, HA injections were accompanied by modest reductions in analgesic prescribing among patients with KOA. Numerically greater reductions were observed among patients treated with Recosyn compared with other HA products; however, these findings should be interpreted with caution given the observational design, potential residual confounding, and the absence of a non-treated comparator group. These results should be considered hypothesis-generating. Full article

Background: Iron deficiency (ID) progresses through three stages: iron deficiency stores (IDS), iron deficiency erythropoiesis (IDE), and iron deficiency anemia (IDA). Neglecting subclinical ID may be harmful to school-aged children and increase the public health burden. Although diet is a key modifiable factor, most studies only focus on overall ID or merely the clinical IDA stage. This study combines a dietary index with pattern analysis to take advantage of their complementary strengths and explore their associations with ID progression. Methods: This cross-sectional study included 2493 participants from rural Guangzhou between June 2022 and May 2023. Demographic, lifestyle, anthropometric, and dietary data were collected via structured questionnaires. Blood samples were analyzed for iron status. Factor analysis identified dietary patterns, and the Chinese Dietary Guidelines Index for Children and Adolescents [CDGI(2021)-C] assessed dietary quality. We used ordinal logistic regression, multivariable logistic regression, and restricted cubic spline (RCS) models to examine dietary associations with ID stages. Results: IDS, IDE, and IDA proportions were 68.22%, 17.45%, and 14.33%, respectively. All four dietary patterns correlated positively with CDGI(2021)-C, most strongly for the fruit–vegetable (r_s = 0.552) and cereal–tuber–legume patterns (r_s = 0.386). Higher CDGI(2021)-C (OR = 0.852, 95% CI: 0.751–0.966, p-trend = 0.012), fruit–vegetable (OR = 0.866, 95%CI: 0.748–0.993, p-trend = 0.047), and meat–offal patterns (OR = 0.733, 95%CI: 0.611–0.868, p-trend < 0.001) were inversely associated with advancing ID stages, while the snack–fast food pattern was positively associated (OR = 1.233, 95% CI: 1.094–1.381, p-trend < 0.001). In IDS, higher adherence to CDGI(2021)-C, fruit–vegetable, and meat–offal patterns was associated with lower odds (all p-trend < 0.05). RCS showed nonlinear associations for the snack–fast food and cereal–tuber–legume patterns, with risk peaking at moderate-to-high adherence to these patterns (both p-nonlinear < 0.05). In IDE and IDA, the snack–fast food pattern risk rose steeply at moderate-to-high adherence (p-nonlinear = 0.036), whereas the cereal–tuber–legume pattern’s ORs fluctuated near 1 (p-nonlinear = 0.020). Conclusions: Dietary pattern and index analyses showed consistent associations across ID stages. Adherence to dietary guidelines slows ID progression, especially in early subclinical stages. More fruits, vegetables, and heme-iron-rich foods, alongside less fast food and snacks, slow ID progression. Though dietary intervention effects weaken in later stages, reducing fast food and snacks confers long-term benefits. These findings inform targeted nutrition policies to prevent ID progression in children. Full article

Background and Objectives: There is limited evidence supporting the incorporation of dietary patterns into heart failure (HF) management. The Dietary Approaches to Stop Hypertension (DASH) diet is linked to cardiovascular disease prevention, but evidence correlating DASH adherence to HF risk is sparse. This study is the first prospective investigation into the relationship between the DASH diet, incident HF and its associated risk factors—hypertension and diabetes mellitus (DM)—in Australian women. Materials and Methods: Survey data (2001–2022) from the Australian Longitudinal Study on Women’s Health (ALSWH) was analysed, where DASH diet scores were calculated from food frequency questionnaire (FFQ) responses and categorised into quintiles. Multivariable logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for the association between DASH adherence and incident HF. Cox proportional hazards models were used to calculate hazard ratios (HRs) and 95% CIs for secondary endpoints, hypertension and DM, and dietary exposure was modelled as a time-varying covariate. Results: 10 594 women (mean age 52.5 ± 1.45 years) participated and, at 21-year follow-up, there were 136 (1.3%) cases of HF, 2182 (20.6%) and 994 (5.7%) cases of hypertension and DM, respectively. After adjustment for covariates (including age and socioeconomic factors), no association was found between the highest DASH quintile and incident HF [OR 0.73, 95% CI: 0.37–1.43; p = 0.20]. However, adjusted HRs for hypertension and DM—0.73 (95% CI: 0.63–0.84; p < 0.001) and 0.65 (95% CI: 0.53–0.81; p < 0.001), respectively—indicated significant associations. Conclusions: In Australian women, DASH diet adherence was associated with a significantly lower risk of hypertension and DM, both of which are HF risk factors. The finding of no direct statistically significant association between the DASH diet and incident HF might reflect the small incidence of HF in our cohort. Full article

Conventional anthropometric measures used to predict dyslipidemia, such as body mass index and waist circumference, vary over time and may not fully capture early-life influences on metabolic risk. Fixed skeletal proportions, including limb length ratios, remain stable after physical maturity and may reflect developmental exposures relevant to lipid metabolism. This study examined the association between the upper arm–to–upper leg length ratio (UA/UL), a fixed skeletal proportion metric with established links to diabetes risk and dyslipidemia; this represents an application not previously reported in a nationally representative U.S. population. We conducted a cross-sectional analysis of adults aged ≥20 years using data from the National Health and Nutrition Examination Survey (NHANES) 2017–March 2020 (n = 7569). The UA/UL ratio was calculated from standardized upper arm and upper leg length measurements and categorized into quartiles based on the weighted sample distribution. Dyslipidemia was defined according to National Cholesterol Education Program Adult Treatment Panel III criteria or current lipid-lowering medication use. Survey-weighted logistic regression models were used to estimate odds ratios (ORs) and 95% confidence intervals (CIs) across progressively adjusted models. Dyslipidemia prevalence increased across UA/UL quartiles (58.4% in Q1 to 81.3% in Q4; p < 0.001). In unadjusted analyses, individuals in the highest UA/UL quartile had greater odds of dyslipidemia compared with the lowest quartile (OR 3.10, 95% CI 2.49–3.86). Associations remained significant after adjustment for demographic factors and for anthropometric measures considered separately. However, the association was attenuated and no longer statistically significant in fully adjusted models that included demographics, adiposity measures, hypertension, and diabetes. In sex-stratified analyses, the association was attenuated and no longer statistically significant in either sex after full adjustment; formal interaction testing confirmed no significant effect modification by sex (p-for-interaction = 0.943). Full article

Background: Thyroid nodules are common, yet only a small proportion are malignant. The independent role of nodule size in malignancy risk remains debated, particularly after adjustment for clinical, biochemical, and sonographic features. Methods: A retrospective cohort study was conducted on adult patients with thyroid nodules evaluated between 2018 and 2025 at a tertiary care center. Clinical, laboratory, ultrasound, cytology, and histopathology data were extracted. Thyroid-stimulating hormone (TSH), free thyroxine (free T4), and sonographic characteristics were analyzed. Univariable and multivariable logistic regression were performed. Missing ultrasound data were addressed using multiple imputation (m = 20), with pooled estimates derived using Rubin’s rules. The final multivariable model included 446 patients. Results: A total of 446 patients were included, of whom 91 (20.4%) had thyroid malignancy. Malignant nodules were significantly larger than benign nodules (2.30 [1.80] cm vs. 1.80 [1.13] cm; p = 0.015). In univariable analysis, TSH, free T4, and multiple ultrasound features were associated with malignancy. In multivariable analysis, nodule size remained the strongest independent predictor of malignancy (adjusted odds ratio [aOR] 1.51; p < 0.001). Hypoechogenicity (aOR 2.07; p = 0.020) and microcalcifications (aOR 1.86; p = 0.047) also remained independently significant, whereas thyroid function parameters were not associated with malignancy after adjustment. Conclusions: Thyroid nodule size is the strongest independent predictor of malignancy, with select ultrasound features retaining additional predictive value. These findings support incorporating nodule size more prominently into thyroid cancer risk stratification while maintaining key sonographic features. Full article

Large language models (LLMs) are increasingly used in professional examinations, but their relative performance on dental board-style questions remains unclear. This study compared two reasoning-optimized models, GPT-o3 and GPT-5T, with a general-purpose multimodal model, GPT-4o, using 399 Japanese dental board-style multiple-choice questions from 2018 to 2022. All questions were presented in Japanese, and items originally accompanied by charts, photographs, or other figures were analyzed separately from items without visual materials. Accuracy and item-level agreement were assessed using pairwise McNemar tests, stratified analyses according to the original presence of visual materials, the Breslow–Day test for homogeneity of odds ratios, and two-proportion z-tests. GPT-5T achieved the highest overall accuracy (294/399, 73.7%), followed by GPT-o3 (257/399, 64.4%) and GPT-4o (255/399, 63.9%). Pairwise McNemar tests showed that GPT-5T outperformed both GPT-4o (Holm-adjusted p = 0.00098) and GPT-o3 (Holm-adjusted p = 0.00072), whereas GPT-o3 and GPT-4o did not differ significantly (Holm-adjusted p = 0.920). Accuracy was lower for questions originally containing visual materials than for questions without such materials across all three models (GPT-4o: 49.7% vs. 72.2%; GPT-o3: 55.1% vs. 69.8%; GPT-5T: 59.9% vs. 81.8%). The advantage of GPT-5T was more evident in questions without visual materials, and heterogeneity across question formats was observed for GPT-5T versus GPT-o3. GPT-5T showed the strongest performance in this dataset. Questions originally containing visual materials were associated with lower accuracy across all models. Because the comparison was based on distinct item groups rather than experimentally manipulated visual conditions, this result should be interpreted as a difference across question formats and may also reflect differences in item composition and difficulty between the two groups. Full article

Background/Objectives: Diabetic retinopathy (DR) is a major microvascular complication of type 1 diabetes mellitus (T1DM) and remains an important cause of preventable visual impairment. Region-specific data on the incidence and clinical predictors of DR among patients with T1DM in Saudi Arabia remain limited. This study aimed to determine the incidence of DR and identify associated demographic and systemic risk factors among patients with T1DM at a tertiary care center in Riyadh, Saudi Arabia. Methods: This retrospective cohort study included 449 patients with T1DM aged ≥9 years who were followed at King Abdulaziz Medical City, Riyadh, between 2015 and 2025. Patients were selected using a consecutive non-probability sampling technique. Data were extracted from the BESTCare 2.0A electronic medical record system and supplemented, when required, by phone-based interviews to verify selected clinical and demographic variables. Patients were classified as controls without DR or cases with DR, including non-vision-threatening DR and vision-threatening DR (VTDR), according to the International Clinical Diabetic Retinopathy Severity Scale. Multivariable logistic regression, Cox proportional hazards models, and temporal trend analysis were performed, with statistical significance set at p < 0.05. Results: The overall incidence rate of DR was 92.66 per 1000 person-years, with similar rates among males and females. In multivariable logistic regression, older age at T1DM diagnosis, longer diabetes duration, hypertension, hyperlipidemia, and albuminuria were independently associated with DR. Mean HbA1c and HbA1c variability were not independently associated with DR after adjustment. In Cox regression, older age at T1DM diagnosis was associated with higher hazards of both DR and VTDR, while hypertension was associated with VTDR. Among patients with DR, younger age at T1DM diagnosis was associated with higher odds of proliferative disease in exploratory severity analysis. Conclusions: DR was common among patients with T1DM in this tertiary-care cohort and was mainly associated with disease duration, age at diagnosis, and systemic vascular comorbidities. These findings support the importance of routine ophthalmologic screening and integrated management of systemic risk factors in patients with T1DM. Full article