Category: Statistics

JMIR Hum Factors. 2026 Apr 14;13:e81230. doi: 10.2196/81230.

ABSTRACT

BACKGROUND: Images play an important role in reducing stigma related to mental health, which often is distorted in the media. In recent years, generative artificial intelligence (AI) has been used to generate images related to mental health. However, first reports suggest that AI-generated images do not depict mental health conditions accurately. In-depth studies on the topic of mental health representations in AI-generated images are still missing.

OBJECTIVE: The main objective of this study is to analyze and compare the visual representation of depression in mass media and in AI-generated images.

METHODS: The methodologies used were discussion groups (15 participants) and a quasi-experimental online survey (792 interviewees), aimed at people with depression and young people.

RESULTS: The results showed that both the images used in the media and those generated by AI reproduced stereotypes and stigmas about depression. However, participants considered AI-generated pictures to be more stereotypical, stigmatizing, and more likely to have a negative impact on people with depression. In contrast, media images were considered more appropriate, realistic, inclusive, and that better reflected the relationship between gender and depression. Statistically significant differences were observed between the control and test groups in both people with depression and young people (P<.001), indicating that when people were aware of what images were AI-generated, they tended to reject them to a greater extent.

CONCLUSIONS: Considering the current trend toward the widespread use of AI in mental health communication, it is crucial to promote closer collaborations between science journalists, AI developers, and mental health experts, including patients’ associations, as well as a shift toward user-participatory AI design.

PMID:41980240 | DOI:10.2196/81230

Neurology. 2026 May 12;106(9):e214931. doi: 10.1212/WNL.0000000000214931. Epub 2026 Apr 14.

ABSTRACT

BACKGROUND AND OBJECTIVES: Interest in plasma biomarkers for neurodegenerative disorders is growing, but their reliance on glomerular filtration makes kidney function a key potential confounder. This study assesses the effect of kidney function (eGFR) on plasma biomarkers of neurodegeneration and on their accuracy for detecting cerebral amyloidosis.

METHODS: This observational study aims at studying the effect of kidney function on blood biomarkers through simultaneous measurements of creatinine, plasma, and CSF biomarkers (Aβ42, Aβ40, p-tau181, p-tau217, neurofilament light chain [NfL], glial fibrillary acidic protein [GFAP], and brain-derived tau [BD-tau]), as well as plasma biomarker ratios (Aβ42/Aβ40, p-tau217/Aβ42, NfL/p-tau217, and p-tau181/Aβ42), in the ALZAN cohort of patients. This prospective multicenter cohort (#NCT05427448), recruited across Montpellier, Nîmes, and Perpignan hospitals, included patients from November 2022 to July 2024 who met the following criteria: age ≥18 years, informed consent, and concomitant CSF and blood sampling. To determine the association between plasma biomarkers and kidney function, we performed univariable linear regression analyses.

RESULTS: A total of 420 patients were included (mean age: 71.1 years, %female: 53.2) and subdivided into 3 groups according to eGFR value (ml/minute/1.73 m²): <60 (n = 36), [60-90] (n = 194), and >90 (n = 190). All mean plasma biomarker levels were significantly higher in the eGFR < 60 group. Except for p-tau217, a statistically significant inverse correlation was observed between eGFR and individual plasma biomarkers. Furthermore, age-adjusted univariable linear regression analysis revealed an association between eGFR and all plasma biomarkers. Kidney dysfunction significantly impaired the specificity of several biomarkers (p-tau181, GFAP, NfL, and BD-tau; p < 0.001) for detecting cerebral amyloidosis (CSF Aβ42/Aβ40 < 7%), whereas p-tau217 was unaffected. It is important to note that ratio-based plasma biomarkers were not influenced by reduced kidney function.

DISCUSSION: Impaired kidney function was linked to increased plasma cerebral amyloidosis biomarkers, but ratio-based measures (especially p-tau217/Aβ42) showed stable sensitivity and specificity for detecting cerebral amyloidosis across all eGFR groups. Additional studies including more patients with lower eGFR values in diverse diagnostic settings are needed to clarify the influence of kidney function on these biomarkers.

CLASSIFICATION OF EVIDENCE: This Class II evidence shows that kidney function influences individual blood biomarkers used for cerebral amyloidosis detection, but not their ratios.

TRIAL REGISTRATION INFORMATION: NCT05427448.

PMID:41980230 | DOI:10.1212/WNL.0000000000214931

Age Ageing. 2026 Apr 4;55(4):afag093. doi: 10.1093/ageing/afag093.

ABSTRACT

BACKGROUND AND AIM: Falls in older adults are a major health concern. Risk factors include medications, but uncertainty remains about potential fall risk-increasing drugs (FRIDs). This nationwide cohort study examined whether non-selective beta-blocker eye drops (timolol) increase the risk of fall-related injuries compared to topical prostaglandin analogue (TPA) eye drops.

METHODS: Using registry data from all Danish residents, we included individuals aged ≥65 years who initiated timolol (n = 52 019) or TPA (n = 72 885) between 1996 and 2023. Propensity score matching was applied to balance baseline characteristics. The primary outcome covered hospital-treated fall-related injuries while the secondary outcome was fall-related fractures specifically.

RESULTS: No statistically significant differences were observed in the risk of fall-related injuries or fall-related fractures for timolol vs. TPA users at 14, 90 or 365 days of follow-up, before or after propensity score matching. Predefined subgroup analyses showed an increased risk of fall-related injuries among individuals aged ≥80 years [incidence rate ratio (IRR) 1.23, 95% CI 1.01-1.50] and among users of ≥3 FRIDs (IRR 1.20, 95% CI 1.01-1.44).

CONCLUSION: Our findings suggest that timolol eye drops are not associated with a significantly increased risk of fall-related injuries in the overall population of older adults. However, a marginally significant increased risk of fall-related injuries was seen among the oldest individuals and those concurrently using multiple FRIDs. These findings underscore the importance of considering a patient-centred approach when prescribing medications that could be potentially harmful.

PMID:41980196 | DOI:10.1093/ageing/afag093

JMIR Mhealth Uhealth. 2026 Apr 14;14:e78637. doi: 10.2196/78637.

ABSTRACT

BACKGROUND: Respiratory dysfunction frequently occurs during the acute phase of stroke and is associated with reduced ventilatory capacity, respiratory muscle weakness, and increased pulmonary complications. However, delivering standardized respiratory training during hospitalization is often constrained by staffing and service continuity.

OBJECTIVE: This study aimed to evaluate the efficacy, safety, and feasibility of a hospital-based comprehensive mobile-based respiratory training program (CMRTP) added to conventional rehabilitation in people with acute stroke who are inpatients.

METHODS: This single-center, assessor-blinded randomized controlled trial enrolled 40 patients within 2 weeks after stroke onset with respiratory dysfunction (forced vital capacity <80% predicted). Participants were randomized (1:1) to CMRTP plus conventional rehabilitation or conventional rehabilitation alone. The CMRTP was delivered via the WeChat-based AIRHUB platform and performed 20 minutes twice daily, 5 days per week for 2 weeks, either independently or with caregiver assistance as needed. The primary outcome was change in forced vital capacity from baseline to week 2. Secondary outcomes included forced expiratory volume in 1 second (FEV₁), peak expiratory flow, maximal inspiratory pressure, maximal expiratory pressure, and modified Barthel index. All outcomes were assessed face-to-face by a blinded senior physician, and all analyses followed an intention-to-treat principle.

RESULTS: Of 56 screened patients, 40 were randomized, and 39 completed the study. Adherence to the CMRTP reached 96%, and no serious adverse events occurred; mild, transient events (fatigue, dizziness, and hyperventilation) were recorded. Compared with the control group, the CMRTP group demonstrated greater improvement in forced vital capacity at week 2 (mean difference 0.77 L; 95% CI 0.39-1.16; P<.001; η²=0.32), with additional between-group differences in maximal inspiratory pressure (P=.001; η²=.25), maximal expiratory pressure (P<.001; η²=.08), and modified Barthel index (P=.001; η²=.26). No significant group differences were found for forced expiratory volume in 1 second or peak expiratory flow.

CONCLUSIONS: A 2-week hospital-based mobile respiratory training program is feasible and safe in people with acute stroke who are inpatients and yields clinically meaningful improvements in respiratory function and daily functional performance when added to conventional rehabilitation.

PMID:41980186 | DOI:10.2196/78637

JMIR Hum Factors. 2026 Apr 14;13:e81343. doi: 10.2196/81343.

ABSTRACT

BACKGROUND: Previous studies indicate that 37% to 92% of patients presenting to hospital emergency eye clinics (EECs) could be seen in commissioned community optometrist enhanced service schemes (ESSs), reducing pressure on hospital services and moving eye care into the community. Digital triage tools may have the potential to support effective triage and use of ESSs.

OBJECTIVE: This study sought to evaluate the effectiveness of a recently developed online symptom triage tool in real-world EEC practice and assess its usability and acceptance by patients.

METHODS: This was a prospective, real-world study comparing automated dispositions suggested by the eye+dot online triage tool with nursing triage, using ophthalmologist dispositions and subsequent clinical findings as the reference standard. Patients aged 13 years and older were sent an SMS text message invitation to use eye+dot prior to their scheduled EEC appointment. Age, time required to complete the symptom questionnaire, and acceptability (using an in-application visual Likert scale) were also recorded. The accuracy of the different triage methods at assigning patients to eye assessments within 24 hours, 48 to 72 hours, or a week based on symptom urgency was compared. Eye+dot’s accuracy in identifying patient suitability for ESSs was analyzed.

RESULTS: A total of 282 eligible patients with a mean age of 53.14 (SD 19.8; range 13-92) years were included. The mean eye+dot test duration was 5.6 (SD 2.4) minutes, with 80.4% (168/209) of patients rating the test as good or excellent. For high-acuity symptomatology (defined via retrospective ophthalmologist recommendation for assessment within 24 hours), sensitivity was similar between eye+dot and nurse triage (58/76, 76.3%; 95% CI 65.2%-85.3% and 65/76, 85.5%; 95% CI 75.6%-92.5%, respectively; P=.19). However, eye+dot specificity was significantly higher (132/206, 64.1%; 95% CI 57.1%-70.6% vs 47/206, 22.8%; 95% CI 17.3%-29.2%; P<.001). Of 224 urgent nursing dispositions, 203 (90.6%) patients were downrated (considered suitable for nonurgent assessment) by ophthalmologist assignment, and 93 (41.5%) were downrated by eye+dot. In total, 90.8% (256/282) of the patients were considered suitable for ESS assessment through eye+dot triage. A total of 56.6% (145/256) of these patients subsequently received only medical advice without specialist investigations or management at their EEC visit, suggesting suitability for community eye care.

CONCLUSIONS: Eye+dot-automated triage has similar sensitivity but superior specificity to nursing triage for identifying high-acuity symptomatology. This small study suggests its potential to improve patient scheduling in EECs throughout the working week and improve the use of community services. A larger study is planned to establish the utility of the triage tool and develop an implementation model to scale up and spread the technology.

PMID:41980184 | DOI:10.2196/81343

JMIR Form Res. 2026 Apr 14;10:e79427. doi: 10.2196/79427.

ABSTRACT

BACKGROUND: The use of online physician rating platforms has significantly increased and has been shown to influence physician selection. There are limited data on the use of these platforms for rating surgeons.

OBJECTIVE: In this study, we sought to assess the geographic distribution of and patterns in rating scores of surgeons in the United States. Additionally, we examined rating volumes across different surgical specialties and the association between peer-nominated and patient-initiated ratings on online rating platforms in the United States.

METHODS: We conducted a cross-sectional study by identifying 201,154 surgeons in the United States via the National Plan and Provider Enumeration System records and Doctors and Clinicians downloadable file. We assessed surgeon coverage on 3 online rating platforms and their geographic use patterns. We described the rating scores and volumes across different surgical specialties and assessed the relationship between rating platforms by comparing peer-nominated and patient-initiated online ratings.

RESULTS: A total of 78.86% (158,630/201,154) of the surgeons had ratings on at least 1 of the 3 patient-initiated websites across 11 specialties. Plastic surgeons, neurosurgeons, and orthopedic surgeons had the highest mean number of patient-initiated ratings. Surgeons with “Top Doctor” recognition from peers (23,171/201,154, 11.52%) were associated with an increased median patient-initiated rating (Healthgrades: 4.36, IQR 3.88-4.71 vs 4.20, IQR 3.64-4.64, P<.001, and r=0.09; Vitals: 4.30, IQR 4.00-4.60 vs 4.20, IQR 3.80-4.50, P<.001, and r=0.09; RateMDs: 4.20, IQR 3.80-4.50 vs 3.80, IQR 3.60-4.60, P<.001, and r=0.16). Geographic analysis indicated that 91.06% (295,816,471/324,870,510) of the US population lives in a county with a surgeon rated 10 times or more.

CONCLUSIONS: Both patient-initiated and peer-nominated rating platforms have a comprehensive coverage of surgeons in the United States, but this coverage differs significantly between surgical specialties. Further work should assess how publicly available online ratings drive surgeon selection and their association with patient experience and postoperative outcomes.

PMID:41980177 | DOI:10.2196/79427

Addiction. 2026 Apr 14. doi: 10.1111/add.70412. Online ahead of print.

ABSTRACT

BACKGROUND AND AIMS: Prescription opioid-related harm remains a significant public health concern. This study aimed to evaluate the efficacy of the Opioid Safety Toolkit, a co-designed, interactive online resource, in increasing naloxone uptake and healthcare provider discussions among adults prescribed opioids for pain.

DESIGN: Parallel-group, open-label, randomised controlled trial.

SETTING: Community-based, online recruitment across Australia.

PARTICIPANTS: Adults (n = 314) prescribed opioids for non-cancer pain.

INTERVENTIONS: Participants were randomised to receive either the Opioid Safety Toolkit (intervention, n = 152), which included interactive and tailored educational content on opioid safety, or an active control website presenting evidence-based opioid safety information (n = 162). Both groups were followed for four weeks.

MEASUREMENTS: The primary outcome was self-reported naloxone requests four weeks post-intervention. Other outcomes were intentions to access naloxone immediately post-intervention, and healthcare provider discussions about opioid safety at four weeks, opioid safety knowledge (immediately after the intervention and at four weeks), satisfaction with resources and naloxone possession at four weeks.

FINDINGS: Participants in the intervention group were more likely to have requested naloxone at four weeks compared with controls [21.7% vs 9.9%, odds ratio (OR) = 2.5, 95% confidence interval (CI) = 1.3, 4.8; P = 0.005], and more likely to report intentions to access naloxone immediately post-intervention compared with controls (41.4% vs 15.4%, OR = 3.9, 95% CI = 2.3, 6.6; P < 0.001). Participants in the intervention group were not more likely to have healthcare provider discussions at four weeks compared with controls (OR = 1.1, 95% CI = 0.7, 1.8; P = 0.620). Post-intervention opioid overdose knowledge was statistically significantly higher in the intervention group compared with control group (Mean score 16.6, 95% CI = 15.5, 17.7 vs control mean score 13.3, 95% CI = 12.3, 14.3). Satisfaction with the resource was higher in the intervention group compared with control group (Mean = 20.0, 95% CI = 18.7, 21.3 vs Mean = 18.0, 95% CI = 16.7, 19.3, P = 0.035).

CONCLUSIONS: We found good evidence that, compared with a gold-standard opioid information website, the Opioid Safety Toolkit increased naloxone requests among Australian adults prescribed opioids for non-cancer pain. We also observed consistent effects across secondary outcomes, with the Toolkit increasing intentions to access naloxone, enhancing opioid overdose knowledge and yielding higher satisfaction ratings, although it did not increase healthcare provider discussions at four weeks.

PMID:41980176 | DOI:10.1111/add.70412

J Air Waste Manag Assoc. 2026 Apr 14. doi: 10.1080/10962247.2026.2657443. Online ahead of print.

ABSTRACT

In recent decades, ethylene oxide (EtO) has been a widely used industrial sterilant and chemical intermediate that has faced increasing scrutiny related to its carcinogenic potential. This study evaluated residential and occupational exposure to EtO in a valley surrounding two point-source emissions in close proximity (~11 meters apart). Air samples were collected in residential and industrial areas. Eight locations, ranging from 100 to 1700 meters from the center of the point-sources in varying directions, had average concentrations between 0.290 and 3.212 µg/m³ (0.16 to 1.8 ppb) with peak levels reaching 26.4 µg/m³ (15 ppb). Exposure scenarios were developed based on daily activity patterns, long-term residency, and estimates derived from historical emissions data. Under the most conservative assumptions, including 40 years of occupational exposure during the periods of highest recorded emissions around the facility, the maximum estimated cumulative lifetime exposure was 591 ppm-days. When compared with epidemiology studies of EtO-exposed workers from similar facilities (studies used by the EPA and IARC in their cancer risk assessments), the highest cumulative exposures observed (13,500+ ppm-days) were at least 23-fold higher than our maximum estimated lifetime exposure value (591 ppm-days). Importantly, these high-exposure groups showed no statistically significant cancer incidence, particularly for breast and lymphohematopoietic cancers. When compared to regulatory values and health-based benchmarks adjusted to cumulative exposures, estimated exposures were substantially below levels associated with increased cancer incidence in epidemiological cohorts for the community surrounding the sterilization facility, even to the most susceptible populations.Implications: This study provides a data-driven, site-specific framework for evaluating long-term human health risks from ethylene oxide (EtO) emissions, using ambient air monitoring, historical emissions data, and conservative EPA-aligned exposure assumptions. Despite assessing one of the highest-emitting sterilization facilities in the U.S. estimated exposures remained well below levels linked to increased cancer risk in epidemiological studies. These findings challenge the assumption that facility proximity alone poses significant health risk and support the development of risk-based, proportionate air quality policies. The approach offers regulatory agencies a more transparent and scientifically grounded basis for EtO risk characterization under the Clean Air Act, TSCA, and state air toxics programs.

PMID:41980174 | DOI:10.1080/10962247.2026.2657443

WMJ. 2026;125(1):162-166.

ABSTRACT

BACKGROUND: The University of Wisconsin’s Rural and Urban Community Health Scholars (RUSCH) pathway program was developed to prepare undergraduate students interested in addressing health disparities in Wisconsin for successful medical school matriculation.

METHODS: Post-completion enrollment outcomes and demographics of participants who completed RUSCH from 2010 through 2024 were analyzed to assess medical school and health professions school matriculation, with associations evaluated using chi-square tests.

RESULTS: Seventy-four percent of participants enrolled in a health professions degree program, with 49% enrolling in medical school, most at institutions within Wisconsin. Men were more likely than women, and non-Hispanic participants were more likely than Hispanic participants, to enroll in medical school following RUSCH completion.

DISCUSSION: RUSCH participation was associated with success in pursuing a health profession degree; however, demographic differences in enrollment outcomes need to be addressed.

PMID:41980157

WMJ. 2026;125(1):158-161.

ABSTRACT

INTRODUCTION: The adoption of artificial intelligence (AI) in image generation raises concerns about potential bias, as these technologies may not accurately reflect the populations represented in the images they create. This study examined whether AI-generated images of medical students accurately represent the diversity of the current US medical student population.

METHODS: Using the DALL-E (Open AI) image-generation algorithm, we created 300 images with the text prompt “medical student.” Two researchers independently analyzed images for demographic indicators, including perceived sex, race/ethnicity, age group, setting, and attire. Descriptive statistics summarized the data, and subgroup analyses assessed differences in portrayals by sex and race/ethnicity. Demographic proportions in the virtual cohort were graphically compared with Association of American Medical Colleges enrollment data.

RESULTS: Of the 300 generated images, 227 (76%) were females and 223 (74%) were White, indicating overrepresentation compared with actual medical school demographics. Black and Latino/Hispanic students were more commonly depicted in scrubs compared to White students, who were often portrayed in white coats or collared shirts (P = .002). No images represented Native American/Alaskan Native or Native Hawaiian/Pacific Islander students.

CONCLUSIONS: AI-generated images of medical students demonstrated significant demographic disparities, indicating potential bias in these technologies. Such biased portrayals may perpetuate stereotypes and hinder diversity efforts. Future research should identify and address these biases to promote more equitable and inclusive applications of AI tools.

PMID:41980156