Multiple sclerosis (MS) is a chronic autoimmune condition primarily affecting women and often diagnosed during childbearing years. This study assessed the impact of the lookback period and calculation method on MS preval...Multiple sclerosis (MS) is a chronic autoimmune condition primarily affecting women and often diagnosed during childbearing years. This study assessed the impact of the lookback period and calculation method on MS prevalence in three healthcare data sources including women of childbearing age (from Italy, Norway and Wales) and three data sources including pregnant women (from France, Finland and Spain). Women aged 15 to 49 years from 2005 to 2019 were included, data from pregnant women were collected around the pregnancy period. MS cases were identified based on at least one MS diagnosis or one dispensation for an MS-specific medication. All data sources provided inpatient diagnoses and medication data; outpatient diagnoses were available in Norway and Finland, and primary care diagnoses in Norway, Finland and Wales. We assessed MS case detection rate by lookback period, and compared three methods for estimating yearly MS prevalence: period prevalence (PP), average point prevalence (APP) and person-time prevalence (PTP). The estimated lookback periods to identify 95% of MS cases ranged from 6 to 9 years. APP and PTP provided lower prevalence estimates than PP, especially when the lookback to identify MS was short. In women of childbearing age, MS prevalence increased over time with all calculation methods and the highest MS prevalence was observed in Norway (PP of 402 per 100,000 in 2019). Finland showed the highest MS prevalence in pregnant women (PP of 218 per 100,000 in 2018). This study highlights the importance of sufficient lookback and available data to accurately estimate MS prevalence.
Winckler JM, Sørensen KK, Strandberg-Larsen K
… +4 more, Torp-Pedersen C, Kesmodel US, Andersen MP, Broccia M
Eur J Epidemiol
· 2025 Sep · PMID 40705230
·
Full text
Children with heavy prenatal alcohol exposure may experience increased healthcare needs and access barriers. We aimed to quantify their healthcare utilisation patterns compared to all other Danish children in this nation...Children with heavy prenatal alcohol exposure may experience increased healthcare needs and access barriers. We aimed to quantify their healthcare utilisation patterns compared to all other Danish children in this nationwide cohort study of children born 1997-2020, followed until Dec 31, 2022. Data was derived from eight national registers. Heavy prenatal alcohol exposure was defined by hospital contacts with 100% alcohol-attributable diagnoses given to mother or child, redeemed prescriptions for drugs to treat alcohol dependence, or enrollment into treatment clinics, one year before or during pregnancy. We estimated age- and sex-specific rates and incidence rate ratios (IRRs) for use of somatic and psychiatric hospitals, and use of general practice (GP), and risk ratio (RR) for participation in the GP-based preventive child health programme. Of 1,457,962 children followed for 17,778,705 person-years, 5898 (0.4%) were heavily prenatal alcohol exposed. Exposed children had higher use of GP and hospitals compared to reference children, particularly notable for psychiatric hospital contacts with IRR for respectively ages 0-5, 6-11 and 12-18 years of 3.55 (2.98-4.24), 2.68 (2.41-2.98), and 2.19 (1.95-2.46); and for planned outpatient contacts 2.01 (1.92-2.10), 1.29 (1.21-1.37), and 1.20 (1.12-1.28). Despite higher healthcare use, participation in the preventive child health programme had a RR of 0.69 (95% CI 0.67-0.72) for exposed compared to reference children. The higher use of hospitals and GP, but lower participation in the child health programme underscores the long-term consequences and societal burden of heavy prenatal alcohol exposure, and leaves a need for awareness and adequate support to ensure health equity.
Alping P, Frisell T, He A
… +3 more, Hillert J, Fink K, McKay KA
Eur J Epidemiol
· 2025 Aug · PMID 40705229
·
Full text
The Swedish Multiple Sclerosis (MS) Register is a nationwide clinical tool and research database. Its generalizability to the broader MS population has not been previously assessed. We aimed to quantify the register's co...The Swedish Multiple Sclerosis (MS) Register is a nationwide clinical tool and research database. Its generalizability to the broader MS population has not been previously assessed. We aimed to quantify the register's completeness and potential disparities between those who were included and those who were not. Using linked national registers, we identified persons with MS through December 2020 as those with ≥ 3 MS diagnoses in the National Patient Register. We assessed the MS register's completeness and compared characteristics between included and non-included persons, using overlap weighting to account for age differences. Among 21,320 prevalent (2020) and 7,355 incident (2011-2020) MS cases, the completeness of the MS register was 85% and 87%, respectively, but varied by county (67-94% in 2020). Prevalent cases not included in the register were older (mean 62.4 vs. 52.3 years) and more likely to have no salaried income (55.3% vs. 30.5%); had fewer years of education (20.8% vs. 12.0% with ≤ 9 years); more often been hospitalized (41.5% vs. 32.2%); and collected more unique prescribed drugs (median 7 vs. 5 drugs per year). Incident cases were also older at first MS diagnosis (mean 50.3 vs. 39.2 years). All differences were attenuated by adjusting for age, although disparities remained. The Swedish MS Register demonstrates high completeness, but may not fully represent older persons with MS, those diagnosed later in life, and those of lower socioeconomic status. These limitations should be considered when interpreting findings based on the register's data.
Guxens M, Botella N, Stafoggia M
… +13 more, Canto M, Petricola S, Valentín A, Lertxundi A, Fernández-Somoano A, Freire C, García-Altés A, Diez E, Marí-Dell'Olmo M, Iñiguez C, López MJ, Ramis R, Binter AC
Eur J Epidemiol
· 2025 Sep · PMID 40705228
·
Full text
We aimed (i) to assess the relationship of pregnancy-average particulate matter (PM) exposure with birthweight, birthweight at term, low birthweight at term, small for gestational age, and preterm birth, (ii) to identify...We aimed (i) to assess the relationship of pregnancy-average particulate matter (PM) exposure with birthweight, birthweight at term, low birthweight at term, small for gestational age, and preterm birth, (ii) to identify critical windows of susceptibility to PM exposure across pregnancy on birthweight and small for gestational age, and (iii) to assess the presence of socioeconomic inequalities on these associations. We established a population-based, nationwide cohort using the Spanish birth registry between 2004 and 2016 (N = 3,678,445). We estimated daily PM and PM concentrations for the entire pregnancy at the maternal residential address at child's delivery using spatiotemporal land use random-forest models. Linear, logistic, and distributed lag linear models were used for the different analysis. All models were stratified by maternal educational level and area-level deprivation index. Mean PM and PM concentrations during pregnancy were 25.1 and 12.7 µg/m, respectively. Higher pregnancy-average PM concentrations were associated with lower birthweight and increased odds of preterm birth (-7.1 g [95%CI -8.5; -5.7] and OR 1.04 [95%CI 1.02; 1.05], respectively, per 10 µg/m increase in PM). Similar results were found for PM, in particular for levels above 10 µg/m. These associations were stronger in infants born to mothers with lower education, particularly when combined with residence in more deprived areas. We observed some windows of susceptibility to PM for birthweight, mainly in the third trimester of pregnancy, with a similar pattern across socioeconomic levels. We did not observe windows of susceptibility to PM. Structural policies to reduce exposure to current PM levels in pregnant women and socioeconomic inequalities are needed.
Guan L, Feng L, Khoo ALY
… +8 more, Ye KX, Ho R, Ng TP, Larbi A, Kennedy BK, Koh WP, Chong YS, Maier AB
Eur J Epidemiol
· 2025 Sep · PMID 40682752
·
Full text
The global population is ageing rapidly. While genetics, lifestyle, and environment are known contributors to healthspan, most insights are drawn from Western cohorts, leaving Asian populations underrepresented despite u...The global population is ageing rapidly. While genetics, lifestyle, and environment are known contributors to healthspan, most insights are drawn from Western cohorts, leaving Asian populations underrepresented despite unique biological, lifestyle, and cultural factors. The SG90 cohort study aimed to fill knowledge gaps in healthy ageing by identifying modifiable medical, biological, lifestyle, psychological, behavioural, and social factors that contribute to longevity in the oldest-old. The study recruited 1,158 participants aged 85 and above from the Singapore Chinese Health Study (SCHS) and Singapore Longitudinal Aging Study (SLAS) between 2015 and 2021. Data collection involved face-to-face interviews to obtain sociodemographic, lifestyle, sleep, functional status, quality of life, medical conditions and healthcare economics information, along with clinical assessments covering physical examinations, anthropometry, physical performance, cognition, and mental health. Biospecimens, including blood, saliva, stool, urine, toenails, hair, and skin tape strips were collected to support extensive multi-omic and cellular analyses. Participants, primarily female (64.5%) and Chinese (97.5%) with a median age of 87 years [interquartile range (IQR): 86-89], were mostly non-smokers (72.1%) and infrequent alcohol consumers (94.9%), with 66.5% exercising regularly. Functional assessments indicate high independence, with median Basic activities of daily living (BADL) and Instrumental ADL (IADL) scores of 20 (IQR: 19-20) and 14 (IQR: 11-16), respectively. 36% of participants rated their self-reported health as good to excellent. The SG90 cohort study offers a comprehensive clinical and biological data resource on healthy ageing among Asia's oldest-old, laying a foundation for targeted interventions to promote healthy longevity and quality of life.
Beslay M, Geissbühler Y, Beau AB
… +22 more, Messina D, Benevent J, Ballardini E, Barrachina-Bonet L, Cavero-Carbonell C, Coldea A, García-Villodre L, Geldhof A, Gini R, Hellwig K, Jordan S, Leinonen MK, Lopez-Leon S, Manfrini M, Martikainen V, Mitter VR, Neville AJ, Nordeng H, Puccini A, Vukusic S, Morris JK, Damase-Michel C
Eur J Epidemiol
· 2025 Aug · PMID 40679704
·
Full text
Prevalence of Multiple Sclerosis (MS) has increased over the last decades, primarily among women of childbearing age. Several algorithms for identifying MS have been described in the literature, providing heterogeneous p...Prevalence of Multiple Sclerosis (MS) has increased over the last decades, primarily among women of childbearing age. Several algorithms for identifying MS have been described in the literature, providing heterogeneous prevalence estimates. We compared five algorithms to identify MS in women of childbearing age and estimated MS prevalence by time period and age-group. The study population included women aged 15 to 49 years-old between 2005 and 2019, from three data sources including all women (from Italy, Norway, and Wales), and three including pregnant women only (from France, Finland, and Spain; data collected around pregnancy). Five algorithms were tested: MS1 to MS3 combined MS diagnoses and MS-medicine prescriptions/dispensations, requiring 1, 2, or 3 occurrences, respectively; MS4 and MS5 used only MS diagnoses, requiring at least 2 occurrences (MS4 allowed just 1 if diagnosis was from inpatient care). In 2015-2019, MS prevalence based on MS1 ranged from 109 to 359 per 100,000 women: 109 in France, 121 in Spain, 195 in Wales, 232 in Finland, 264 in Italy, and 359 in Norway. More restrictive algorithms led to greater disparity, with MS3 ranging from 53 in Spain to 325 in Norway, and MS5 from 21 in France to 345 in Norway. All algorithms showed expected prevalence trends by time and age among women of childbearing age, though lower than in the literature. Overall, MS1 provided prevalence estimates most closely aligned with existing literature. This study offers key insights into choosing algorithms for identifying MS in women of childbearing age and in pregnant women.
While previous literature suggests that plant-based diets may be associated with a lower risk of breast cancer, evidence remains inconsistent. In this study, we investigated the association between adherence to plant-bas...While previous literature suggests that plant-based diets may be associated with a lower risk of breast cancer, evidence remains inconsistent. In this study, we investigated the association between adherence to plant-based diets and breast cancer risk in the European Prospective Investigation into Cancer and Nutrition (EPIC) study. Overall plant-based diet index (PDI), healthful (hPDI), and unhealthful PDI (uPDI) were calculated, and multivariable Cox models were used to estimate hazard ratios (HR) and 95% confidence intervals (CI) of breast cancer for the three indices. Mediation analysis was performed to assess the role of body mass index (BMI) and waist circumference (WC) in the association between hPDI and postmenopausal breast cancer risk. Over a median follow-up of 14.9 years, 10,805 incident invasive breast cancer cases were identified among 258,343 women. In the multivariable model, not adjusted for BMI, higher adherence to hPDI was inversely associated with breast cancer risk, with HR per 1-SD increase [95% (CI)] of 0.97 (0.94, 0.99). The corresponding HRs (95% CI) per 1-SD increase for overall PDI and uPDI were 0.98 (0.96, 1.00) and 1.01 (0.99, 1.03), respectively. The associations between hPDI and postmenopausal breast cancer were partly explained by BMI and WC, which mediated 30% and 52% of this association, respectively. Higher adherence to hPDI was associated with a slightly lower total breast cancer risk. For postmenopausal breast cancer, this association was partly explained by lower BMI or WC. These findings suggest that promoting healthful plant-based diets could support breast cancer risk reduction.
Buchwald J, Lehtimäki T, Raitakari O
… +3 more, Salomaa V, Kaprio J, Pirinen M
Eur J Epidemiol
· 2025 Sep · PMID 40646428
·
Full text
Faster nicotine metabolism, defined as the nicotine metabolite ratio (NMR), is known to associate with heavier smoking and challenges in smoking cessation. However, the broader health implications of genetically determin...Faster nicotine metabolism, defined as the nicotine metabolite ratio (NMR), is known to associate with heavier smoking and challenges in smoking cessation. However, the broader health implications of genetically determined nicotine metabolism are not well characterized. We performed a hypothesis-free phenome-wide association study (PheWAS) of over 21,000 outcome variables from UK Biobank (UKB) to explore how the NMR (measured as the 3-hydroxycotinine-to-cotinine ratio) associates with the phenome. As the exposure variable, we used a genetic score for faster nicotine metabolism based on 10 putative causal genetic variants, explaining 33.8 % of the variance in the NMR. We analysed ever and never smokers separately to assess whether a causal pathway through nicotine metabolism is plausible. A total of 57 outcome variables reached phenome-wide significance at a false discovery rate of 5 %. We observed expected associations with several phenotypes related to smoking and nicotine, but could not replicate prior findings on cessation. Importantly, we found novel associations between genetically determined faster nicotine metabolism and adverse health outcomes, including unfavourable liver enzyme and lipid values, as well as increased caffeine consumption. These associations did not appear to differ between ever and never smokers, suggesting the corresponding pathways may not involve nicotine metabolism. No favourable health outcomes were linked to genetically determined faster nicotine metabolism. Our findings support a possibility that a future smoking cessation therapy converting fast metabolizers of nicotine to slower ones could work without adverse side effects and potentially even provide other health-related benefits.
Örtqvist AK, Söderling J, Magnus MC
… +3 more, Urhoj SK, Håberg SE, Stephansson O
Eur J Epidemiol
· 2025 Jul · PMID 40643819
·
Full text
Pregnant and puerperal women are at increased risk of venous thromboembolism (VTE) owing to hemostatic changes in preparation for childbirth. The objective of this study was to investigate if COVID-19 infection was assoc...Pregnant and puerperal women are at increased risk of venous thromboembolism (VTE) owing to hemostatic changes in preparation for childbirth. The objective of this study was to investigate if COVID-19 infection was associated with VTE in pregnancy or 12 weeks postpartum when considering (prophylactic or therapeutic) anticoagulant use. This population-based register study included all women in Sweden and Norway giving birth after 22 gestational weeks, with conception dates from March 2020 to 2022. A PCR-verified COVID-19 test was used as the exposure, and a VTE diagnosis during pregnancy or 12 weeks postpartum was the outcome. Non-infected women consisted of those testing negative and untested individuals. Cox regression analyses, with COVID-19 infection as a time-varying exposure, and adjusted for maternal characteristics and anticoagulant use, provided overall hazard ratios. To evaluate whether there was a particular increased risk of VTE shortly after testing positive for COVID-19, we estimated time-specific risk of VTE in the first 2, 4, 8, 12, and 16 weeks following COVID-19 infection. Data from each country were first analyzed separately and then meta-analyzed. Among 323,868 participants, 46,048 (14.2%) had COVID-19 during pregnancy, and 80 (0.2%) were diagnosed with VTE. Pregnant women with COVID-19 had a higher VTE incidence rate compared to non-infected (4.9 vs. 2.9 per 1000 person-years; adjusted overall hazard ratio [aHR] 1.26, 95% Confidence Interval [CI] 0.80-2.00). The highest risk was within two weeks of infection (aHR 4.63, 95% CI 2.71-7.90) but remained elevated up to 12 weeks post-infection (aHR 1.86, 95% CI 1.17-2.94). In the postpartum period, 8,515 (2.6%) had COVID-19, and 6 (0.07%) were diagnosed with VTE (aHR 5.17, 95% CI 2.50-10.69). Although VTE post-COVID-19 infection was rare, the infection was associated with increased VTE risk during pregnancy and postpartum, even after adjusting for anticoagulant use. These findings should contribute to the individual risk assessment when evaluating the need for prophylactic anticoagulants in pregnancy and postpartum.
Ekroos S, Toffol E, Heikinheimo O
… +2 more, Haukka J, Arvas M
Eur J Epidemiol
· 2025 Aug · PMID 40638010
·
Full text
The WHO aims to reduce iron deficiency anaemia globally. Use of modern hormonal contraception (HC) could offer protection against anaemia in premenopausal women, but population-level effectiveness is unclear. We aim to q...The WHO aims to reduce iron deficiency anaemia globally. Use of modern hormonal contraception (HC) could offer protection against anaemia in premenopausal women, but population-level effectiveness is unclear. We aim to quantify the effect of HC on anaemia. This nested case-control study includes over half the fertile-aged female population of Finland in 2017. Data on HC use from the national Prescription Center were combined with data on anaemia diagnosis from national care registries. Cases (anaemia diagnosis during follow-up, 2019-2020) were matched with up to five controls by age and municipality. After calculating the minimally sufficient adjustment set, adjusted odds ratios were derived in a conditional multivariable regression model accounting for matching. 3 100 cases of anaemia were matched with 13 143 controls. The minimally sufficient adjustment set included age, education level, obesity, abnormal uterine bleeding, and cancer. Compared to non-users of HC, risk of anaemia was lower in users of combined oral contraceptives containing ethinylestradiol (0·74 [0·66-0·83]) or oestradiol (0·49 [0·35-0·68]), progestin-only oral contraceptives (0·42 [0·35-0·51]), LNG-IUDs (0·64 [0·43-0·94]), and contraceptive vaginal rings (0·68 [0·49-0·94]). Individual product effects ranged from 0·77 [0·66-0·90] for drospirenone and ethinylestradiol to 0·40 [0·32-0·48] for desogestrel-only. Benefits of HC use extend to anaemia protection on population level. Anaemia protection should be included in guidelines on HC to support clinical decision making.
Ivansson EL, Johansson T, Karlsson T
… +1 more, Johansson Å
Eur J Epidemiol
· 2025 Sep · PMID 40632378
·
Full text
Osteoporotic fractures, largely resulting from reduced estrogen levels after menopause and subsequent bone loss, are a leading cause of disability among older women. Although oral contraceptive pills (OCPs) contain estro...Osteoporotic fractures, largely resulting from reduced estrogen levels after menopause and subsequent bone loss, are a leading cause of disability among older women. Although oral contraceptive pills (OCPs) contain estrogen, their long-term impact on bone health and osteoporosis risk remain uncertain. Here, we assessed the effect of OCP use on bone mineral density (BMD) and osteoporosis using data from 257,185 women from the UK Biobank, born 1936-1970. Time-dependent Cox regression was used to estimate adjusted hazard ratios (HRs) and 95% confidence intervals (CI) for osteoporosis, while multivariable linear regression was used to assess the effect of OCP use on BMD, measured as T-scores in standard deviation units based on quantitative ultrasound of the calcaneus. By the end of follow-up in 2020, 7.6% of the participants had received an osteoporosis diagnosis. The rate of osteoporosis was lower among ever OCP users (HR = 0.86; 95% CI 0.83-0.89; P = 2.8 × 10). OCP use was also associated with a higher BMD T-score (0.052; 0.038-0.067; P = 2.1 × 10) with an increasing effect with longer use. Use of OCPs for 0-1 years had no significant effect on BMD (P = 0.081). However, longer durations were associated with increased BMD T-scores compared to never users: 2-5 years (0.046; 0.027-0.065, P = 2.2 × 10), 6-10 years (0.062; 0.043-0.080; P = 3.5 × 10), 11-15 years (0.062; 0.042-0.081; P = 3.2 × 10) and 16 + years (0.064; 0.044-0.083; P = 1.2 × 10). We found prior OCP use to be associated with higher BMD and a reduced risk of osteoporosis, potentially offering long-term benefits and suggesting that OCP use could reduce osteoporotic complications in older women.
Eur J Epidemiol
· 2025 Aug · PMID 40627067
·
Full text
Based on compelling evidence from observational epidemiological studies, screening colonoscopy has since long been thought to strongly lower the burden of colorectal cancer (CRC), both by early detection of prevalent CRC...Based on compelling evidence from observational epidemiological studies, screening colonoscopy has since long been thought to strongly lower the burden of colorectal cancer (CRC), both by early detection of prevalent CRC and prevention of incident CRC through detection and removal of precancerous lesions. Widespread offer and use of screening colonoscopy went along with a dramatic decline in CRC incidence in screening age groups in the US, in contrast to an increase in incidence at younger ages and in countries not engaging in CRC screening. The recently published 10-year results from the NordICC trial, the first randomized clinical trial (RCT) reporting long-term effects of screening colonoscopy on CRC risk and mortality, has been widely interpreted as challenging the evidence for strong efficacy of screening colonoscopy. Such reasoning was based on the trust that randomization in this large-sized trial should have prevented any residual confounding that might have affected the observational epidemiological studies. However, randomization cannot prevent other potential biases which should be carefully addressed and minimized in both observational and interventional studies. We illustrate that such biases may have led to major underestimation of screening effects in the NordICC trial. The observed patterns underline the need for more rigorous efforts to prevent and correct for such biases, along with the need to derive more informative metrics of screening efficacy. Such metrics should include informative estimates of screening colonoscopy effects on both early detection of prevalent CRC cases and prevention of incident CRC cases. The momentum for CRC screening should by no means slowed by misinterpretation of the NordICC trial evidence.
Kang D, Park J, Kim H
… +8 more, Lee JE, Nam SJ, Kim SW, Yu J, Chae BJ, Ryu JM, Cho J, Lee SK
Eur J Epidemiol
· 2025 Aug · PMID 40627066
·
Full text
Existing evidence indicates that long-term effects of breast cancer treatment can significantly impact survivors' ability to fulfill their personal, familial, and social roles. However, few studies comprehensively integr...Existing evidence indicates that long-term effects of breast cancer treatment can significantly impact survivors' ability to fulfill their personal, familial, and social roles. However, few studies comprehensively integrate patient-reported outcomes (PROs) and real-world healthcare utilization data, particularly in Asian populations. Thus, we established a prospective cohort, the Breast Cancer Information Grand Round for Survivorship (BIG-S), to address these gaps and introduce this resource. The prospective BIG-S cohort recruited patients newly diagnosed with breast cancer at Samsung Medical Center starting in November 2018. Clinical data, recurrence, and healthcare utilization were systematically collected from electronic medical records by trained researchers, and body composition was measured using multifrequency bioelectrical impedance analysis. PROs were assessed following recommendations from the International Consortium for Health Outcomes Measurement (ICHOM). These included health-related quality of life, physical, psychological, social, and cognitive functions, symptoms, healthy behaviors, financial difficulty, spiritual well-being, and cancer adaptation, using validated questionnaires. A total of 2,749 patients participated, with an average age of 49.7 years. The mean quality-of-life score at diagnosis was 55.6, indicating moderate general well-being, and improved gradually to 68.2 at four years post-diagnosis. At baseline, participants showed high physical, cognitive, and role functioning scores but had relatively lower emotional and social functioning scores. Over four years, emotional and social functioning improved, whereas cognitive and role functioning declined. Survivors initially reported low sexual functioning, sexual enjoyment, and future perspectives, all of which significantly decreased during follow-up. Fatigue and insomnia were persistent throughout the observation period. Healthcare utilization initially concentrated on plastic surgery and rehabilitation medicine and increasingly shifted towards gynecology, family medicine, and psychiatry after two years. The BIG-S cohort uniquely integrates clinical data, PROs, and healthcare utilization patterns, offering comprehensive insights into breast cancer survivorship trajectories. Findings from BIG-S are expected to guide targeted interventions and inform tailored survivorship care strategies, especially for Asian breast cancer survivors.
Abrahamowicz M, Beauchamp ME, Roberts EK
… +1 more, Taylor JMG
Eur J Epidemiol
· 2025 Jun · PMID 40608267
·
Full text
The hazard has been a central concept in the analysis and interpretation of time-to-event data for over 50 years. At any follow-up time, the hazard is the probability of the event happening in the next unit of time among...The hazard has been a central concept in the analysis and interpretation of time-to-event data for over 50 years. At any follow-up time, the hazard is the probability of the event happening in the next unit of time amongst those still at risk. Hazard ratios (HRs) between groups are frequently used to quantify the exposure/treatment's association with the failure time. In a highly cited paper, Hernán criticized HRs, asserting that their decreases over time may reflect simply a built-in selection bias, induced by an unmeasured susceptibility, and should not be interpreted as genuine changes in treatment effect. Hernán supports his arguments mainly by the results of a hormone therapy trial, where the HR for coronary heart events decreased largely during follow-up, with hazards crossing from harmful to protective treatment effect. However, he did not present simulations or algebraic derivations to demonstrate that these changes may reflect just an unmeasured susceptibility. We use simulations and real-world case studies to systematically explore this issue. The first simulation series reveals how the underestimation bias and changes over time in Cox proportional hazards model-based HRs depend on the joint impact of susceptibility on the hazard, its distribution, and the incidence of events; with important bias toward the null occurring only for a very strong susceptibility. Further simulations mimic the hormone therapy trial highlighted by Hernán, to demonstrate that the reported bias and crossing hazards are extremely unlikely to reflect just an unmeasured susceptibility, which suggests some biological reasons for decaying treatment HR, possibly including decreasing treatment adherence. Finally, we present real-world examples of interpretable and clinically plausible time-dependent HRs in cancer research. In conclusion, results of our simulations and real-world case studies suggest that concerns about HR limitations may be overstated, and we encourage researchers to model time-dependent HRs and consider potential biological and clinical causes thereof.
Toffol E, Partonen T, Heikinheimo O
… +3 more, But A, Latvala A, Haukka J
Eur J Epidemiol
· 2025 Aug · PMID 40601246
·
Full text
Earlier findings on the relationship between use of hormonal contraception (HC) and depressive symptoms and disorders are contradictory. Thus, we assessed the associations of use of different types of systemic hormonal c...Earlier findings on the relationship between use of hormonal contraception (HC) and depressive symptoms and disorders are contradictory. Thus, we assessed the associations of use of different types of systemic hormonal contraceptives in the six preceding months with the risk of depression in women aged 15-49 years. Data were obtained from national registers in Finland. All cases of depression in the years 2018-2019 were identified in a population-based cohort of women. We used a nested case-control design with 1:4 ratio (n = 117,360 cases) and applied multivariable conditional logistic regression models. During the follow-up a total of 23,480 new cases with the diagnosis of depression were observed (incidence rate: 21.7, 95% confidence interval = 21.5-22.0 per 1000 person-years). Use of HC in the six preceding months, specifically that of combined hormonal contraceptives (containing gestodene and ethinylestradiol, drospirenone and ethinylestradiol, and nomegestrol and estradiol), was significantly associated with a lower risk of depression compared to non-use when controlling for marital status, socioeconomic status, education, recent delivery, recent psychiatric hospitalization, chronic diseases, use of psychiatric medications (excluding antidepressants) and former use of HC (odds ratio: 0.90, 95% confidence interval = 0.85-0.95; 0.86, 95% confidence interval = 0.81-0.91, respectively). Current use of progestogen-only preparations (norethisterone, levonorgestrel, desogestrel) was not associated with depression. This pattern was evident in all age groups, including adolescent girls. HC use appeared not associated with an increased risk of depression in fertile-aged women and across all age groups, including adolescent girls.
Cairat M, Severi G, Huybrechts I
… +1 more, Fournier A
Eur J Epidemiol
· 2025 Jul · PMID 40601245
·
Full text
Concerns about the safety of titanium dioxide (TiO), including potential carcinogenicity, have prompted its ban in foods in the European Union, while remaining allowed as pharmaceutical excipient. We aimed to evaluate wh...Concerns about the safety of titanium dioxide (TiO), including potential carcinogenicity, have prompted its ban in foods in the European Union, while remaining allowed as pharmaceutical excipient. We aimed to evaluate whether ingesting increasing quantities of TiO through medicines is associated with higher cancer risk. Data were derived from the French National Health Data System, a nationwide medico-administrative database. A case-control study was nested within two cohorts: users of metformin (all doses) and users of 200 mg acebutolol, both available in TiO-containing and TiO-free formulations. During 2013-2021, 293,101 cancer cases were identified and matched to 2,930,633 controls. TiO exposure through metformin and acebutolol consumption was calculated based on drug claims from 2006 up to five years before the index date. Conditional logistic regression models estimated linear associations between TiO exposure and cancer risk. RRs of overall cancer per 1000 TiO-containing tablets and per 10,000 mg of TiO increments were both 1.00 (95% CI: 0.99-1.01). Analyses by cancer site also yielded RRs very close to 1.00 or slightly different but not statistically significant, except for breast (RR per 10,000 mg: 1.03, 95% CI:1.00-1.07) and lymphoid/hematopoietic (RR per 1000 tablets: 0.97, 95% CI: 0.95-1.00) cancers, which however lost significance after Bonferroni correction. There was a suggestion of non-linear positive association for central nervous system cancers. This first epidemiological study on TiO ingestion and cancer found no meaningful linear association between increasing TiO exposure through medicines and overall or site-specific cancer risk. Non-linear associations cannot be excluded.
Posthumus AM, Knobbe TJ, Kremer D
… +19 more, Gomes-Neto AW, Dielwart IJC, Jonker J, Doorenbos CSE, Eisenga MF, van Londen M, Douwes RM, Nieuwenhuis LM, Annema C, de Boer MT, de Borst MH, Damman K, Pol RA, Gan CT, Verschuuren EAM, Blokzijl H, de Meijer VE, Bakker SJL, TransplantLines Investigators
Eur J Epidemiol
· 2025 Aug · PMID 40601244
·
Full text
The TransplantLines Biobank and Cohort Study (NCT03272841) is an ongoing prospective study conducted at the University Medical Centre Groningen, The Netherlands. TransplantLines aims to identify risk factors and biomarke...The TransplantLines Biobank and Cohort Study (NCT03272841) is an ongoing prospective study conducted at the University Medical Centre Groningen, The Netherlands. TransplantLines aims to identify risk factors and biomarkers associated with health problems following solid organ transplantation and donation. Additionally, the study seeks to develop new interventions to reduce symptom burden and improve long-term outcomes, including health-related quality of life, cardiovascular complications, graft failure, and mortality. It includes recipients of (combined) heart, liver, lung, kidney, pancreas, and small bowel transplants, as well as living liver and kidney donors, and deceased (multi-)organ donors. The biobank contains a wide range of biomaterials including whole blood, serum, EDTA-plasma, buffy coat, 24-h urine samples, faeces, hair, nails, and tissues. Data collection includes physical and cognitive assessments, extensive laboratory analysis, metagenomic sequencing, and questionnaires. TransplantLines, initiated in 2015, consists of 5143 participants as of October 2024, among 2312 (45%) females. The mean age was 50 (± 16) years at transplantation, 55 (± 11) years at living donation and 56 (± 15) years at deceased donation. Both cross-sectional and longitudinal biomaterials and data are included. For recipients, longitudinal biomaterials and data were collected at: pre-transplantation, at transplantation, and at 3, 6, 12, 24, and 60 months post-transplantation. For living donors, data were collected at pre-donation, donation, 3 months post-donation, and/or 5 or 10 years post-donation.
Eur J Epidemiol
· 2025 Jun · PMID 40576931
·
Full text
Hazard ratios are routinely reported as effect measures in clinical trials and observational studies. However, many methodological works have raised concerns about the interpretation of hazard ratios as causal effects. T...Hazard ratios are routinely reported as effect measures in clinical trials and observational studies. However, many methodological works have raised concerns about the interpretation of hazard ratios as causal effects. These concerns are often related to three points: (i) depletion of susceptible individuals leads to selection bias and complicates the causal interpretation of the hazard ratio, (ii) the hazard ratio is not collapsible, and (iii) the conventional proportional hazards assumption rarely holds in medical studies. We discuss the relation between these three points. We ground our presentation on an example about effect of endocrine therapy in reducing the risk of recurrence or death in a population of patients with breast cancer. We also describe why survival curves and risk differences do not exhibit any of the undesirable properties of hazard ratios.