Multinomial Regression Research Articles

Impact of social determinants of health on hepatocellular carcinoma surveillance, treatment, and health care costs.

The impact of clinical factors and social determinants of health on treatment patterns and health care costs among patients with HCC is unknown. Using 100% Medicare Fee-For-Service claims and a commercial multipayor claims database, we identified patients diagnosed with HCC from January 1, 2017, to December 31, 2020. Surveillance receipt was defined 12 months prior to HCC diagnosis, whereas treatment and health care costs were assessed post-HCC diagnosis. Multinomial logistic regression was used to assess the association between demographics, social determinants of health, and surveillance or HCC treatment. Multivariable generalized linear regression was used to identify factors associated with total health care costs. Of the 32,239 patients with HCC (mean age 68 y, 67% male, 73% White), 70% received surveillance and only half (51%) received any treatment. Curative treatment receipt was higher among those with prior surveillance (24% with CT/MRI and 18% with ultrasound vs. 9% with no surveillance). Curative treatment was independently associated with HCC surveillance and inversely associated with Black race, lower education level, and diagnosis in the year 2020 (COVID-19 year). Higher health care costs were independently associated with Black race, low English proficiency, living alone, and diagnosis in 2018-2020, and inversely associated with CT/MRI-based surveillance. Race and social determinants of health were independently associated with curative treatment receipt and health care costs. Increasing access to high-quality HCC surveillance may improve treatment receipt and reduce health disparities among patients with HCC.

Lifestyle-specific association between adversity dimensions with group-based mental health trajectories.

Childhood adversity (CA) has a substantial correlation with mental health problems. Keeping a healthy lifestyle is essential for mental health interventions; it is unclear, however, how healthy lifestyle affect the relationship between CA and persistent mental health problems. This longitudinal study (n=1112, 54.5% male) collected the data on CA (measured through three dimensions: threat, deprivation and unpredictability), mental health problems, and lifestyle factors. Group-based multi-trajectory modeling (GBMTM) was utilized to estimate trajectories for three mental health problems (i.e., depression, ADHD and overanxiety). Close friendships, regular physical activity, appropriate sleep duration, shorter screen time, and healthy eating were combined to establish a healthy lifestyle score (which ranges from 0 to 5). Higher scores indicated a healthier lifestyle. Three trajectories of mental health problems were identified: persistently low risk (24.9%), persistently medium-high risk (50.0%), and persistently high risk (25.1%). Multinomial logistic regression showed that high adversity (high-threat: β=2.01, P<0.001; high-deprivation: β=1.03, P<0.001; high-unpredictability: β=0.83, P=0.001; high-overall adversity: β=1.64, P<0.001) resulted in a persistently high risk of mental health problems; these outcomes were maintained after robust control for covariates. Further lifestyle stratification, null associations were observed among children with a healthy lifestyle, irrespective of their gender; however, after controlling for covariates, the above associations remained relatively stable only among boys. The generalizability of our findings is restricted by 1) limited racial diversity and 2) missing data. This finding underscores the benefits of promoting a healthy lifestyle in children to prevent persistent mental health problems caused by CA.

A biopsychosocial approach to phenotyping people with knee osteoarthritis awaiting total knee arthroplasty: A secondary cohort analysis

BackgroundPrevious research showed chronic post-total knee arthroplasty (TKA) pain in 20% of people with knee osteoarthritis (KOA). Various preoperative biopsychosocial-related factors have been described, but phenotyping people with KOA awaiting TKA based on these factors is still lacking. This could be relevant to understanding differences in TKA surgery responses. ObjectiveTo identify phenotypes in people with KOA awaiting TKA and differences in post-TKA pain based on preoperative biopsychosocial factors. MethodsPeople with KOA awaiting TKA in 4 hospitals in Belgium and the Netherlands were included. A cross-sectional latent profile analysis was conducted on structural, metabolic, functional, pain-related, psychological and social variables. Concurrent validity was tested using 3-step multinomial logistic regression. The difference in one-year post-TKA pain was examined with linear mixed model analysis. ResultsTwo hundred and seventeen participants were included in the latent profile analysis with a mean (SD) age of 65.5 (7.7) years, including 109 women. A model with 2 phenotypes differed in 14 out of 21 variables. Participants with phenotype 2 (28%) had a higher body mass index (BMI), higher chance of having less structural damage (KOA grade), lower mean quadriceps strength and physical function (Knee Society Scoring System functional and 30-second chair stand test), higher pain intensity, number of pain locations, and indices of central sensitisation (temporal summation, central sensitisation inventory score, and lower pressure pain thresholds), higher pain catastrophising, anxiety and depression, and higher post-TKA pain intensity compared to phenotype 1 (72%). Concurrent validity was confirmed in 3 out of 4 variables. ConclusionsPhenotype 2 (28%) with nociplastic pain characteristics in combination with worse psychological factors, BMI, functional and structural factors, and phenotype 1 (72%) not representing these characteristics were identified. Phenotype 2 had worse pain intensity scores after TKA compared to phenotype 1. Attention to the characteristics of phenotype 2 is warranted concerning post-TKA pain. Database registrationThe protocol is registered at ClinicalTrials.gov (NCT05380648).

Sustainable Steps in the Snow: Exploring factors associated with active school commuting in Finland

Physical inactivity of children and youth is a major health problem. Active commuting could increase the amount of physical activity and be a more environmentally friendly traffic mode choice compared with motorized vehicles. Parents have an integral role in determining the options a child has for commuting. We explored parental barriers and motivators associated with the active commute of their children.This cross-sectional study used the CLIMATE NUDGE Survey data of parents with school-aged children. A set of 25 possible influencing factors included demographic factors, social aspects, and pro-environmental attitudes, and questions about the motivating role of journey characteristics and health effects. The data were analyzed with two multinomial logistic regression analyses (N = 320): first, comparing those who either never or sometimes commute actively to those who always do so, and second, those who always or sometimes commute actively to those who never do so.Results indicated the strongest determinants to differentiate those who always and those who never use active commute, were commute distance and parental beliefs about social commute norms. The longer the commute distance was, the more likely the child was to belong to the never active commuter group and less likely to belong to the always active commuter group. The more parents believed their child’s peers commuted actively, the more likely it was that their child always commuted actively and less likely they never did so.We found several factors associated with active commuting, yet somewhat surprisingly neither environmental attitudes nor climate change related reasons were significantly associated with active commuting. To conclude, instead of highlighting environmental aspects to promote active commute, we recommend highlighting social norms and perception of the commute distance, and addressing weather-related barriers and health related motivators.

Home Health Care and Place of Death in Medicare Beneficiaries With and Without Dementia.

Home health care supports patient goals for aging in place. Our objective was to determine if home health care use in the last 3 years of life reduces the risk of inpatient death without hospice. We analyzed the characteristics of 2,065,300 Medicare beneficiaries who died in 2019 and conducted multinomial logistic regression analyses to evaluate the association between the use and timing of home health care, dementia diagnosis, and place of death. Receiving any home health care in the last 3 years of life was associated with a lower probability of inpatient death without hospice (Pr 23.3% vs 31.5%, p < .001), and this effect was stronger when home health care began prior to versus during the last year of life (Pr 22.5% vs 24.3%, p < .001). Among all decedents, the probability of death at home with hospice compared to inpatient death with hospice was greater when any home health care was used (Pr 46.0% vs 36.5%, p < .001), and this association was strongest among beneficiaries with dementia who started home health care at least 1 year prior to death (Pr 55.6%, p < .001). Use of home health care during the last 3 years of life was associated with reduced rates of inpatient death without hospice, and increased rates of home death with hospice. Increasing affordable access to home health care can positively affect end-of-life care outcomes for older Americans and their family caregivers, especially those with dementia.

Prevalence and risk factors associated with undiagnosed hypertension among adults aged 15–49 years in India: insights from the National Family Health Survey (NFHS-5)

ObjectivesTo estimate the prevalence and identify the factors associated with undiagnosed hypertension in India. Study designA secondary data analysis using the National Family Health Survey (NFHS-5) covering the period 2019–2021. MethodsInformation on hypertension among individuals aged 15–49 years was extracted from the survey dataset. We estimated the prevalence of undiagnosed hypertension using physical measurements along with self-reported data from the survey. A log-binomial model with survey-adjusted Poisson regression was used to estimate the prevalence ratio between undiagnosed and diagnosed hypertension. Multinomial logistic regression analysis examined the factors associated with diagnosed hypertension (vs healthy) and undiagnosed hypertension (vs healthy). All the analyses were survey-weight adjusted and stratified by gender. ResultsThe survey-adjusted prevalence of undiagnosed hypertension was 8.75% (8.62%–8.87%) and was higher among males [13.56% (13.03%–14.12%)] than in females [8.14% (8.03%–8.25%)]. The proportion of individuals with undiagnosed hypertension among total hypertension was 44.99% (44.44%–45.55%) and was higher in males [65.94% (64.25%–67.60%)] than in females [42.18% (41.66%–42.71%)]. ConclusionsOur findings revealed that age, higher body mass index, no access to health care, and having no comorbidities were risk factors for undiagnosed hypertension. One in twelve people had undiagnosed hypertension, and of those with hypertension, one in two were undiagnosed, with males being disproportionately affected. Targeted public health interventions are crucial to improve hypertension screening, particularly among middle-aged and obese individuals without comorbidities.

Effect of Local Fire Risks and Preventive Measures on the Installation of Residential Firefighting Systems in Detached House

This study analyzes the major factors affecting the installation of residential firefighting systems in detached houses. In particular, it investigates the dependence of local fire risk and local government prevention efforts, along with housing and household characteristics, on the type of installation. This study used data from the 2020 Population and Housing Census, the first nationwide survey on residential firefighter system installation, and employed multinomial logistic regression analysis. The analysis results revealed that less than half of households had installed all firefighting systems, where the installation rate differed significantly by region. Housing and household characteristics also influenced the installation type. Local fire risk as well as the local government’s fire prevention policies positively affect the installation of residential firefighting systems.

Factors of unmet dental care needs due to the time, economic, and physical constraints among older people

Objectives: We aim to identify factors associated with unmet dental care needs in older adults and explore the impact of time, economic and physical constraints on each. Methods: The analysis was based on the 2023 Community Health Survey. Statistical analyses were conducted using chi-square test and multinomial logistic regression for 65,952 people aged 65 years and older. Results: First, we looked whether subjects experienced unmet dental care needs and found that 8.4% did. Second, according to the type of unmet dental care needs, 54.1% of the time, economic, and physical constraints were economic constraints. Analyzing each of the time, economic, and physical constraints, we found that time constraints were associated with younger age. For economic constraints, lower education level and health security type of recipients in medical aid program were more likely to experience unmet dental care needs. In terms of physical constraints, those living in rural areas were more than three times more likely to experience unmet dental care needs than those living in urban areas. Conclusions: The results showed that time constraints were related to age, economic constraints were related to household income level, and physical constraints were related to residential area.

Female Reproductive System and Thyroid Dysfunction: Findings from a 12-Year Follow-Up in the Tehran Thyroid Study.

Background: The impact of thyroid dysfunction (TD) on the female reproductive system has been extensively documented. While there is evidence suggesting that alteration in female reproductive status may affect thyroid function, conflicting results have prevented definitive conclusions. This study aimed to investigate the associations of parity, spontaneous abortion (mentioned as abortion throughout this study), and menopause status with the prevalence and incidence of TD. Methods: From the Tehran thyroid study population, 2711 participants were included in the cross-sectional analysis to explore associations between female reproductive status and TD. Overall, 2191 participants with euthyroid were included in the survival study and followed up in 3-year intervals. Multinomial logistic regression was adopted in cross-sectional analysis and multivariable Cox proportional hazard model was used to determine associations between the incidence of TD with parity, abortion, and menopause status, adjusting for age, smoking, body mass index, and thyroid peroxidase antibodies positivity. Results: At the baseline, multiple parities (≥4) were significantly associated with overt hypothyroidism (odds ratio [OR] = 1.12; confidence interval [CI] 1.0-1.26) and subclinical hyperthyroidism (OR = 1.11 [CI 1.03-1.21]). Furthermore, multiple abortions were associated with overt hyperthyroidism (OR = 2.09 [CI 1.02-4.26]). Over the course of the study, multiple parities were significantly associated with the incident subclinical and clinical hypothyroidism. Conversely, a history of abortion was associated with a reduced risk of incident overt hypothyroidism. We found no significant association between menopause status and the prevalence or incidence of either hypothyroidism or hyperthyroidism. Conclusions: Our results suggest that the female reproductive system may be associated with thyroid function. Parity and abortion are associated with the occurrence of TD. A deeper understanding of the underlying mechanisms of the cellular and molecular alterations in signaling cascades during pregnancy is necessary to fully elucidate these associations.

Financial Determinants of Adoption of Agricultural Technologies in Irish Potato Production in Ol Kalou Sub County, Kenya: Application of Multinomial Logistic Regression

Irish potato is a significant source of food and income for many households worldwide. The demand for the crop has steadily increased around the globe. Kenyan Irish potato farmers have the potential to produce about 30 tonnes per Ha. However, small-scale Irish farmers in the country realize low output ranging from 4-8 tonnes per Ha due to limited uptake of agricultural technologies. This study aimed to analyze the financial factors influencing the uptake of agricultural technologies in Irish potato production in Ol Kalou Sub County. The agricultural technologies under investigation were chemical fertilizer, certified seeds, fungicides, and farm machinery. Production and innovation diffusion theories guided the study. Descriptive cross-sectional research design was used to obtain data from a study population of 21,942 smallholder Irish potato farmers in Ol Kalou Sub County. A multiple-stage sampling technique was employed to give a sample size of 385 respondents small scale Irish potato farmers, where data was collected through a semi-structured questionnaire. Data collected was analyzed using multinomial logistic regression through SPSS version 28 and STATA version 17. The model indicated that off-farm income, access to credit, and production risk had a positive and significant influence on adopting agricultural technologies. Conversely, production cost had a negative and significant influence on the adoption of agricultural technologies, while the availability of subsidies had no significant influence. The Marginal effect analysis showed that the availability of off-farm income increased adoption by 22.00%, the increase in credit increased adoption by 2.00%, the availability of mitigation measures of production risk increased adoption by 19.00%, while the increase in production cost decreased adoption by 7.00%. The study concluded that off-farm income, credit facilities, production risk and cost of production influence the adoption of agricultural technologies. The study recommended that the government should develop policy regulations such as diversification of income, grants that reduce production costs, credit incentives, and crop insurance.

Multi-label Random Subspace Ensemble Classification1

In this work, we develop a new ensemble learning framework, multi-label Random Subspace Ensemble (mRaSE), for multi-label classification. Given a base classifier (e.g., multinomial logistic regression, classification tree, K-nearest neighbors), mRaSE works by first randomly sampling a collection of subspaces, then choosing the best ones that achieve the minimum cross-validation errors and, finally, aggregating the chosen weak learners. In addition to its superior prediction performance, mRaSE also provides a model-free feature ranking depending on the given base classifier. An iterative version of mRaSE is also developed to further improve the performance. A model-free extension is pursued on the iterative version, leading to the so-called Super mRaSE, which accepts a collection of base classifiers as input to the algorithm. We show the proposed algorithms compared favorably with the state-of-the-art classification algorithm including random forest and deep neural network, via extensive simulation studies and two real data applications. The new algorithms are implemented in an updated version of the R package RaSEn.

The Interaction of Race and Pregnancy Intentions on Breastfeeding Initiation Rates and Duration in the United States.

Background: Unintended pregnancy increases adverse perinatal health outcomes. Limited studies exist on the influence of pregnancy intention on breastfeeding. This study examines the interaction of Race and couples' pregnancy intention on breastfeeding initiation and duration. Methods: Using the National Survey of Family Growth (2017-2019), we categorized breastfeeding into three categories (never breastfed, breastfed for 6 months or less, and breastfed for more than 6 months). Multinomial logistic regression and probability marginal effects were estimated across racial categories (Hispanics, non-Hispanic [NH] White, Black, and NH-others or multiracial). Results: The probability (Pr.) of never breastfeeding was the greatest among NH-Black women for all pregnancy intentions-both intended (Pr.: 0.34), mother-only (Pr.: 0.27), father-only (Pr.: 0.45), ambivalent (Pr.: 0.55), and neither intended (Pr.: 0.37). The probability of breastfeeding for 6 months or less is highest for Hispanics for both intended (Pr.: 0.41), NH-White for mother-only intended (Pr.: 0.49), Hispanics for father-only intended (Pr.: 0.56), ambivalent pregnancy intentions (Pr.: 0.60), and NH-others or multiracial for neither intended pregnancies (Pr.: 0.48). The probability of breastfeeding for more than 6 months is highest for NH-others or multiracial for both intended (Pr.: 0.50), ambivalent (Pr.: 0.49), and neither intended (Pr.: 0.36), while highest for NH-Black for mother-only intended (Pr.: 0.43). Conclusions: Our study demonstrates the interaction of Race and couples' pregnancy intention on breastfeeding initiation and duration. Owing to the high volume of unintended pregnancies in the United States, the current study might help maternal and child healthcare providers understand the interaction between race and pregnancy intentions on breastfeeding initiation and duration in the United States.

Cardiometabolic risk factor clusters in older adults using latent class analysis on the Bushehr elderly health program

Metabolic syndrome (MetS), comprising obesity, insulin resistance, hypertension, and dyslipidemia, increases the risk of type II diabetes mellitus and cardiovascular disease. This study aimed to identify the prevalence and determinants of specific clusters of the MetS components and tobacco consumption among older adults in Iran. The current study was conducted in the second stage of the Bushehr Elderly Health (BEH) program in southern Iran—a population-based cohort including 2424 subjects aged ≥ 60 years. Latent class analysis (LCA) was used to identify MetS and tobacco consumption patterns. Multinomial logistic regression was conducted to investigate factors associated with each MetS class, including sociodemographic and behavioral variables. Out of 2424 individuals, the overall percentage of people with one or more components of MetS or current tobacco use was 57.8% and 20.8%, respectively. The mean (SD) age of all participants was 69.3(6.4) years. LCA ascertained the presence of four latent classes: class 1 (“low risk”; with a prevalence of 35.3%), class 2 (“MetS with medication-controlled diabetes”; 11.1%), class 3 (“high risk of MetS and associated medication use”; 27.1%), and class 4 (“central obesity and treated hypertension”; 26.4%). Compared to participants with a body mass index (BMI) < 30, participants with BMI ≥ 30 were more likely to belong to class 3 (OR 1.91, 95% CI 1.31–2.79) and class 4 (OR 1.49, 95% CI 1.06–2.08). Polypharmacy was associated with membership in class 2 (OR 2.07, 95% CI 1.12–3.81), class 3 (OR 9.77, 95% CI 6.12–15.59), and class 4 (OR 1.76, 95% CI 1.07–2.91). The elevated triglyceride-glucose index was associated with membership in class 2 (OR 12.33, 95% CI 7.75–19.61) and class 3 (OR 12.04, 95% CI 8.31–17.45). Individuals with poor self-related health were more likely to belong to class 3 (OR 1.43; 95% CI 1.08–1.93). Four classes were identified among older adults in Iran with distinct patterns of cardiometabolic risk factors. Segmenting elderly individuals into these cardiometabolic categories has the potential to enhance the monitoring and management of cardiometabolic risk factors. This strategy may help reduce the severe outcomes of metabolic syndrome in this susceptible population.

Sociodemographic disparities in depression and anxiety symptom severity among young adults in Sweden

Abstract Background Young adults are disproportionately affected by mental health problems such as depression and anxiety. While previous studies have identified sociodemographic predictors using the four-item Patient Health Questionnaire (PHQ-4), differences in symptom severity remain less explored. This study aimed to evaluate the sociodemographic predictors of different levels of depression and anxiety symptoms among young adults in Sweden. Methods Survey data from a population-based sample of Swedish adults aged 20-21 (n = 3,226) was used. Symptom severity was categorized into three levels: normal/mild, moderate, and severe, based on established cutoff scores on the PHQ-4. Gender, parental education and birth country, and upper secondary education were assessed through linked registry information. Multinomial logistic regression was used. Results Females had a higher risk of moderate (RR = 1.8, 95% CI 1.5, 2.2) and severe (RR = 2.3, 95% CI 1.8, 3.1) symptoms (compared to no or mild symptoms) relative to males. Among males, the risk of moderate symptoms was higher without a secondary degree (RR = 1.9, 95% CI 1.0, 3.4) compared to an academic degree. Males with parents born abroad faced elevated risks of moderate (RR = 1.9, 95% CI 1.3, 2.9) and severe (RR = 2.2, 95% CI 1.3, 3.7) symptoms relative to males with Swedish-born parents. Among females, parental tertiary education was associated with a lower risk of moderate symptoms (RR = 0.7, 95% CI 0.5, 0.9) compared to lower education. Further, females having a vocational (RR = 1.5, 95% CI 1.1, 2.1) or no secondary degree (RR = 3.6, 95% CI 2.1, 6.1) had an increased risk of severe symptoms compared to females with an academic degree. Conclusions This study identified sociodemographic predictors for moderate and severe symptoms of depression and anxiety among young men and women in Sweden. Mental health promotion and prevention should specifically target young adults at elevated risk of experiencing moderate to severe symptoms. Key messages • Understanding sociodemographic predictors of depression and anxiety symptoms in young adulthood is essential for public health work. • Mental health promotion and prevention needs to include appropriate interventions for young men and women from diverse socioeconomic and cultural backgrounds.

Epidemiology of hepatitis C from 2015 to 2022 in a central Italian region: the OPT-HepaC project

Abstract Background Chronic hepatitis C virus (HCV) infection is a major global public health problem. This study investigated the evolving landscape of HCV in Tuscany (Italy) from 2015 to 2022, considering demographic shifts, clinical profiles, treatment regimens, and outcomes. Methods The study was conducted in order to characterize HCV-positive patients who started treatment between 2015 and 2022 in Tuscan prescribing centers. The study was part of OPT-HepaC, a project funded by Regional Health Service. Data on demographics, clinical history and outcomes were collected. Three different temporal eras were analysed: the period before andfter the universal availability of direct-acting antivirals (DAA) and the COVID-19 era. Treatment’s effectiveness was evaluated by the Sustained Virological Response (SVR12), assessed 12 weeks after the end of treatment. Outcomes included SVR12, non-SVR12 and lost to follow-up (LTFU). ANOVA, Kruskal-Wallis test and multinomial logistic regression were used for statistical analysis. Results 6882 HCV patients (age: 58.88±14.21) were treated by the centres participating at the study. Males (56.4%) and Italian (91.8%) were predominant. Sociodemographic and variable characteristics vary across the time. While the overall SVR12 remained stable over time (χ2(2)=0.357, p = 0.837), the LTFU rate significantly increased during the COVID-19 era. Facilitating factors for the non-SVR12 and LTFU outcomes were male gender (Odds Ratio (OR)=2.134, p = 0.006)/cirrhosis (OR = 1.713, p = 0.040)/HIV co-infection (OR = 3.251, p = 0.005) and DAA treatment of third generation (OR = 4.187, p = 0.006), respectively. Conclusions The study provides a comprehensive view of the HCV landscape, highlighting demographic changes, treatment regimens and outcomes. The COVID-19 pandemic affected service delivery, delaying HCV diagnosis/treatment and influencing the LTFU rate. Despite challenges, the region maintained high SVR12 rates, emphasizing the importance of sustained efforts in HCV care. Key messages • The study highlights the dynamic nature of HCV management in Tuscany. • The study emphasises the need for continuous adaptation to public health challenges, providing insights for future public health strategies.

Polycystic ovary syndrome and labor market attachment in Swedish women

Abstract Background Polycystic ovary syndrome (PCOS) is a common endocrine disorder in women of fertile age and is associated with several burdensome comorbidities. We use Swedish longitudinal register data to identify and compare clusters of employment trajectories among working age women, in order to investigate how PCOS affects labor market attachment. Data and Methods A cohort of 157,356 women born in 1975-1977 were followed annually between the ages of 30 and 39. Labor market states were constructed by combining detailed data on various types of earned income. PCOS was identified through national inpatient and outpatient registers. Sociodemographic covariates (civil status, region of origin, highest attained education, number of children) were measured at the age of 29 years. Sequence analysis was used to identify clusters of typical labor market trajectories. Multinomial logistic regression was employed to assess the associations between being diagnosed with PCOS and belonging to the identified clusters. Results Women with PCOS spent less time in employment and were more dependent on sickness benefits during the follow-up time than those without PCOS. Five clusters of trajectories were identified: stable employment, education into employment, labor market exclusion, continuous unstable position, long-term sickness. Compared to being in stable employment cluster, women diagnosed with PCOS were 97% more likely to experience long-term sickness (RRR (relative risk ratio): 1.97 [CI: 1.90-2.05]), and had 11% higher risk of belonging to education into employment (RRR: 1.11 [CI: 1.07-1.15]) and a 4% higher risk (RRR: 1.04 [CI: 1.00-1.09]) of experiencing labor market exclusion. The magnitude of the association increased for education into employment and labor market exclusion after adjusting for covariates. Conclusions PCOS can lead to disadvantaged labor market outcomes and better strategies are needed in order to prevent economic exclusion among women diagnosed with the condition. Key messages • PCOS leads to disadvantaged labor market outcomes. • Women with PCOS may experience economic exclusion.

Past-year mental health issues: discrepant survey &amp; administrative data

Abstract Introduction Accurately measuring mental health disorder prevalence is crucial for public health planning. Administrative and survey data are common methods, but often yield differing results. This study compares case identification of mental health issues across these data sources in Slovenia. Methods We linked the 2019 European Health Interview Survey (EHIS) with three Slovenian health administrative databases: the National Hospital Health Care Statistics Database, Outpatient Prescription Drugs Database, and Absence From Work Database. Case identification utilized self-report of any mental health issue in the past 12 months and healthcare utilization records in 12 months preceding participation in EHIS. Multinomial logistic regression was used to examine the association between socio-demographic factors and case identification across data sources. Results Significant differences were found in 12 month prevalence of any mental health issue estimates between data sources. Only 45.9% of self-reported cases were identified in administrative data, and 36.6% of administrative cases self-reported mental health issues (Kappa = 0.302). Socio-demographics, including age, gender, education, and employment status, were significantly associated with the likelihood of identification in specific data sources. Discussion Our findings underscore the importance of data source and assessment tool selection on mental health prevalence estimates. Variations in identification across population subgroups could reflect differences in healthcare access and self-disclosure biases. Linking data sources and considering their inherent limitations is crucial for accurate burden estimation. Key messages • Data source and assessment methods significantly impact how we identify individuals with mental health issues in the past year. • Combining data sources and understanding their limitations is essential for improving mental health burden estimation.

Sickness absence patterns during the COVID-19 pandemic among Finnish public sector employees

Abstract Background The patterns of sickness absence (SA), that is, the clustering of differing lengths of episodes and total number of days absent, have not been examined. We examined these SA patterns and factors associated with the patterns during the COVID-19 pandemic. Methods We used survey data from employees of four large Finnish cities in 2020 linked with SA data in 2021 retrieved from employer registers (n = 39 791). SA patterns were defined through cluster analysis (K-means) including number of SA days, short (1-9 days), and long (10-365 days) SA episodes within one year. The associations of employee and work characteristics in 2020 with SA clusters in 2021 were analysed with multinomial regression. Results Four SA patterns (=clusters) were identified: 1) Optimal SA pattern (n = 31 320, 79%), 2) Recurring short SA episodes (n = 5149, 13%), 3) Recurring long SA episodes (n = 2964, 7%), and 4) Very high levels of SA with long episodes (n = 358, 1%). Compared to optimal SA pattern, female gender, higher body mass index, former smoking, abstinence from alcohol, not being in high occupational socioeconomic position, on-site work during COVID-19, and COVID-19 infection were associated with all suboptimal SA patterns. Recurring short SA episodes were more likely in younger employees (odds ratio=0.98, 95% confidence interval 0.97-0.98), those with a temporary job (1.10, 1.00-1.20), current smokers (1.34, 1.21-1.48), those using passive commute modes (1.03, 1.01-1.05 [car]; 1.02, 1.00-1.04 [public transport]), and those who had experienced a team reorganization due to COVID-19 (1.27, 1.10-1.46). The two most adverse SA patterns were more likely among older employees, and less likely among those cycling to work (0.91, 0.84-0.98; 0.97, 0.95-0.99). Conclusions We identified three differential suboptimal SA patterns. Compared to the optimal SA pattern, several individual-level risk factors, but also some COVID-19 related factors were found to associate with suboptimal SA patterns. Key messages • We identified three suboptimal sickness absence patterns during COVID-19 pandemic. • Individual characteristics, but also COVID-19 related factors were associated with the observed sickness absence patterns.

Cut-off values for estimated cardiorespiratory fitness in terms of physical functioning among middle-aged to older adults.

Cardiorespiratory fitness is linked to various health outcomes. This study aimed to establish cut-off values for estimated cardiorespiratory fitness associated with poor physical functioning in middle-aged to older adults. A total of 2638 participants (1083 males), aged 40-69 years, performed a 6-min walking test (6MWT) to estimate cardiorespiratory fitness (VO<inf>2max)</inf> and completed a health-related questionnaire. Physical functioning was assessed using four outcomes: perceived health, perceived fitness, difficulties in 2-km walking, and difficulties in climbing several stairs. Receiver operator characteristics (ROC) curves were used to determine cut-off values for estimated VO<inf>2max</inf> and 6MWT distance linked to poor physical functioning based on the four outcomes. Analyses were done separately for males and females, and for 40- to 49-, 50- to 59-, and 60- to 69-year-old groups. Multinomial regression was used to analyze differences between fitness thirds in outcome variables. ROC analyses were significant for all outcomes (AUC 0.686-0.834, P<0.001). In the total group, cut-off values for estimated VO<inf>2max</inf> were 29.9-30.2 mL/kg/min, and for walking distance, 588.0-599.5 m. Cut-off values decreased with age, and mean differences between males and females were 3.3 mL/kg/min and 33 m. All cut-off values were within ±1.3 mL/kg/min of the lower fitness tertile. The odds ratios of having poor physical functioning in the middle and high fitness thirds were less than 0.50 compared to low fitness third. VO<inf>2max</inf> of 30 mL/kg/min and 600-m walking distance seemed reasonable cut-off values for poor physical functioning. Exceeding the currently proposed thresholds is a desirable goal to avoid consequences of insufficient fitness.

Social health gradient and risk factors among patients hospitalized for COVID-19 and pre-pandemic respiratory infections. A linked national individual case-control study in Belgium

IntroductionThe literature establishes a clear social gradient in health for transmissible respiratory diseases. However, this gradient’s extent remains largely unexplored in the context of COVID-19, and it is uncertain whether the pandemic has exacerbated this gradient. The study aims to compare the socio-economic profiles and comorbidities during the COVID-19 pandemic with a control population affected by viral pneumonia/respiratory disease in 2019.MethodsThis case-control study analyzed linked data from all patients hospitalized for COVID-19 in 2020 (n = 22,087) and for respiratory diseases in 2019 (n = 7,586). Socio-economic data from the social security database were linked to clinical data from the hospital registry. We analyzed the socio-demographic and clinical factors associated with COVID-19 hospitalization (control group, wave 1, and wave 2) using multinomial regressions and logistic regression models and the length of stay during hospitalization using binomial negative regressions.ResultsA social health gradient was observed in both the COVID-19 and control groups, with a significant increase across waves for COVID-19 (p-trend &amp;lt; 0.0001). Men, people over the age of 45, those with comorbidities, high population density, lower income, lower socio-economic status, and people living in Brussels capital were at higher risk of COVID-19 hospitalization and longer length of stay compared to the control group. Except for sub-Saharan Africans, all patients of foreign nationality had a significantly increased risk of hospitalization (p &amp;lt; 0.001), but a shorter length of stay compared to Belgians.ConclusionThe socio-health gradient for COVID-19 followed the same pattern as that observed in pre-pandemic respiratory diseases, intensifying in the second wave and among the most deprived groups. This study emphasizes the importance of collecting social data alongside clinical data for a better understanding of social health inequalities and for tailoring health prevention policies.