International journal of epidemiology最新文献

Background: Biobanks typically rely on volunteer-based sampling. This results in large samples (power) at the cost of representativeness (bias). The problem of volunteer bias is debated. Here, we (i) show that volunteering biases associations in UK Biobank (UKB) and (ii) estimate inverse probability (IP) weights that correct for volunteer bias in UKB.

Methods: Drawing on UK Census data, we constructed a subsample representative of UKB's target population, which consists of all individuals invited to participate. Based on demographic variables shared between the UK Census and UKB, we estimated IP weights (IPWs) for each UKB participant. We compared 21 weighted and unweighted bivariate associations between these demographic variables to assess volunteer bias.

Results: Volunteer bias in all associations, as naively estimated in UKB, was substantial-in some cases so severe that unweighted estimates had the opposite sign of the association in the target population. For example, older individuals in UKB reported being in better health, in contrast to evidence from the UK Census. Using IPWs in weighted regressions reduced 87% of volunteer bias on average. Volunteer-based sampling reduced the effective sample size of UKB substantially, to 32% of its original size.

Conclusions: Estimates from large-scale biobanks may be misleading due to volunteer bias. We recommend IP weighting to correct for such bias. To aid in the construction of the next generation of biobanks, we provide suggestions on how to best ensure representativeness in a volunteer-based design. For UKB, IPWs have been made available.

Background: Model-estimated air pollution exposure products have been widely used in epidemiological studies to assess the health risks of particulate matter with diameters of ≤2.5 µm (PM2.5). However, few studies have assessed the disparities in health effects between model-estimated and station-observed PM2.5 exposures.

Methods: We collected daily all-cause, respiratory and cardiovascular mortality data in 347 cities across 15 countries and regions worldwide based on the Multi-City Multi-Country collaborative research network. The station-observed PM2.5 data were obtained from official monitoring stations. The model-estimated global PM2.5 product was developed using a machine-learning approach. The associations between daily exposure to PM2.5 and mortality were evaluated using a two-stage analytical approach.

Results: We included 15.8 million all-cause, 1.5 million respiratory and 4.5 million cardiovascular deaths from 2000 to 2018. Short-term exposure to PM2.5 was associated with a relative risk increase (RRI) of mortality from both station-observed and model-estimated exposures. Every 10-μg/m3 increase in the 2-day moving average PM2.5 was associated with overall RRIs of 0.67% (95% CI: 0.49 to 0.85), 0.68% (95% CI: -0.03 to 1.39) and 0.45% (95% CI: 0.08 to 0.82) for all-cause, respiratory, and cardiovascular mortality based on station-observed PM2.5 and RRIs of 0.87% (95% CI: 0.68 to 1.06), 0.81% (95% CI: 0.08 to 1.55) and 0.71% (95% CI: 0.32 to 1.09) based on model-estimated exposure, respectively.

Conclusions: Mortality risks associated with daily PM2.5 exposure were consistent for both station-observed and model-estimated exposures, suggesting the reliability and potential applicability of the global PM2.5 product in epidemiological studies.

Background: Maternal colonization by the bacterium Group B streptococcus (GBS) increases risk of preterm birth, a condition that has an important impact on the health of children. However, research studies that quantify the effect of GBS colonization on preterm birth have reported variable estimates of the effect measure.

Methods: We performed a simulated cohort study of pregnant women to assess how timing of exposure (GBS colonization) assessment might influence results of studies that address this question. We used published data on longitudinal maternal GBS colonization and on the distribution of preterm births by gestational age to inform parameters used in the simulations.

Results: Assuming that the probability of preterm birth is higher during weeks when pregnant women are colonized by GBS, our results suggest that studies that assess exposure status early during pregnancy are more likely to estimate an association between GBS colonization and preterm birth that is closer to the null, compared with studies that assess exposure either at birth or during gestational weeks matched to preterm births. In sensitivity analyses assuming different colonization acquisition rates and diagnostic sensitivities, we observed similar results.

Conclusions: Accurate quantification of the effect of maternal GBS colonization on the risk of preterm birth is necessary to understand the full health burden linked to this bacterium. In this study, we investigated one possible explanation, related to the timing of exposure assessment, for the variable findings of previous observational studies. Our findings will inform future research on this question.

Background: Colorectal cancer (CRC) is the third-most-common cancer worldwide and its rates are increasing. Elevated body mass index (BMI) is an established risk factor for CRC, although the molecular mechanisms behind this association remain unclear. Using the Mendelian randomization (MR) framework, we aimed to investigate the mediating effects of putative biomarkers and other CRC risk factors in the association between BMI and CRC.

Methods: We selected as mediators biomarkers of established cancer-related mechanisms and other CRC risk factors for which a plausible association with obesity exists, such as inflammatory biomarkers, glucose homeostasis traits, lipids, adipokines, insulin-like growth factor 1 (IGF1), sex hormones, 25-hydroxy-vitamin D, smoking, physical activity (PA) and alcohol consumption. We used inverse-variance weighted MR in the main univariable analyses and performed sensitivity analyses (weighted-median, MR-Egger, Contamination Mixture). We used multivariable MR for the mediation analyses.

Results: Genetically predicted BMI was positively associated with CRC risk [odds ratio per SD (5 kg/m2) = 1.17, 95% CI: 1.08-1.24, P-value = 1.4 × 10-5] and robustly associated with nearly all potential mediators. Genetically predicted IGF1, fasting insulin, low-density lipoprotein cholesterol, smoking, PA and alcohol were associated with CRC risk. Evidence for attenuation was found for IGF1 [explained 7% (95% CI: 2-13%) of the association], smoking (31%, 4-57%) and PA (7%, 2-11%). There was little evidence for pleiotropy, although smoking was bidirectionally associated with BMI and instruments were weak for PA.

Conclusions: The effect of BMI on CRC risk is possibly partly mediated through plasma IGF1, whereas the attenuation of the BMI-CRC association by smoking and PA may reflect confounding and shared underlying mechanisms rather than mediation.

Background: Surveillance data and vaccination registries are widely used to provide real-time vaccine effectiveness (VE) estimates, which can be biased due to underreported (i.e. under-ascertained and under-notified) infections. Here, we investigate how the magnitude and direction of this source of bias in retrospective cohort studies vary under different circumstances, including different levels of underreporting, heterogeneities in underreporting across vaccinated and unvaccinated, and different levels of pathogen circulation.

Methods: We developed a stochastic individual-based model simulating the transmission dynamics of a respiratory virus and a large-scale vaccination campaign. Considering a baseline scenario with 22.5% yearly attack rate and 30% reporting ratio, we explored fourteen alternative scenarios, each modifying one or more baseline assumptions. Using synthetic individual-level surveillance data and vaccination registries produced by the model, we estimated the VE against documented infection taking as reference either unvaccinated or recently vaccinated individuals (within 14 days post-administration). Bias was quantified by comparing estimates to the known VE assumed in the model.

Results: VE estimates were accurate when assuming homogeneous reporting ratios, even at low levels (10%), and moderate attack rates (<50%). A substantial downward bias in the estimation arose with homogeneous reporting and attack rates exceeding 50%. Mild heterogeneities in reporting ratios between vaccinated and unvaccinated strongly biased VE estimates, downward if cases in vaccinated were more likely to be reported and upward otherwise, particularly when taking as reference unvaccinated individuals.

Conclusions: In observational studies, high attack rates or differences in underreporting between vaccinated and unvaccinated may result in biased VE estimates. This study underscores the critical importance of monitoring data quality and understanding biases in observational studies, to more adequately inform public health decisions.