Epidemiology最新文献

Background: Payday loans refer to high-interest, short-term loans. These loans can provide immediate financial relief for individuals with limited access to traditional credit. However, the predatory nature of payday loans may portend increased financial strain and adverse public health consequences.

Methods: We examine whether state-level temporal variation in payday loan restrictions over a 20-year period (2000-2019) corresponds with a reduction in preterm births: a leading cause of infant mortality in the US. Between 2000 and 2019, 10 US states and the District of Columbia imposed restrictions on payday lending at varied time points. We use data on preterm births provided by the CDC Wonder database (2000-2019) and apply staggered difference-in-difference and event study approaches to examine whether preterm births (per 100 live births) declined among states that imposed payday lending restrictions, relative to states that never imposed restrictions. We also control for state-specific time propensity of preterm births, derived through time-series analysis.

Results: Results indicate a decline in the preterm births by approximately 0.22 per 100 live births (95% CI: -0.31, -0.13) within the first 3 years of payday loan restrictions, which corresponds to more than 4512 fewer than expected preterm births.

Conclusion: Our findings are consistent with the hypothesis that state-level payday lending restrictions may precede a reduction in preterm births.

Background: Little is known about how alcohol policies experienced in adolescence are associated with later health. We assess whether age of exposure to stricter alcohol policies is associated with later alcohol-attributable hospitalizations and mortality. We take advantage of an alcohol advertising ban and alcohol tax increases introduced in 1975-1977 with relatively stable alcohol policies before and after.

Methods: We used Finnish register data on birth cohorts 1950-1964 (1,175,878 individuals) to assess cohort-wise hazard ratios for the first incidence of alcohol-attributable hospitalization and mortality, and mortality due to external and other causes at ages 21-54 years.

Results: Men who were aged 19 to 25 at the time of the restrictive reform had similar risks for alcohol-attributable hospitalization and mortality to the reference group of those aged 18 - legal drinking age - at the time of reform. For those underage at the time, hospitalization and mortality rates were incrementally smaller cohort by cohort. For example, men who were 17 at the time of the reform had lower hazard ratios of alcohol-attributable hospitalization: 0.91 (95% CI 0.87; 0.95) as did those who were 13 [0.85 (95% CI 0.81; 0.89)]. The findings were similar for external-cause mortality, and similar yet more uncertain for women. In contrast, mortality from other causes declined continuously from cohort to cohort.

Conclusions: Our findings are consistent with the hypothesis that stricter alcohol policies in adolescence reduce harmful alcohol consumption patterns extending into adulthood and manifesting as lower alcohol-related harm to health.

Background: Health selection into neighborhoods describes unhealthy people moving disproportionately to lower-income neighborhoods, producing observable socioeconomic gradients sometimes falsely attributed to neighborhood effects on health. We investigated residential mobility outcomes and their relationship to baseline health using population-level data linkages in Ontario, Canada.

Methods: We included Canadian Community Health Survey (CCHS) respondents ages 25 to 64 between 2005 and 2014 (n=93,235). We assessed baseline health using self-reported health and multimorbidity. We captured moves using health administrative data and the Canadian census. We fit multinomial logistic regression models with a six-category residential mobility outcome: (1) non-movers in low-income neighborhoods; (2) non-movers in high-income neighborhoods; (3) movers from low-income to low-income; (4) movers from low-income to high-income; (5) movers from high-income to low-income; and (6) movers from high-income to high-income. We adjusted models for CCHS cycle, age, sex, household income, immigrant status, and residential instability.

Results: Compared to those with very good or excellent health, respondents reporting fair or poor health at baseline had higher odds of moving from low- to low-income neighborhoods (aOR=1.73, 95%CI 1.46-2.05), moving from high- to low-income (aOR=1.64, 95%CI 1.35-1.98), moving from low- to high-income (aOR=1.26, 95%CI 1.04-1.54), and not moving within low-income (aOR=1.36, 1.23-1.51) relative to not moving within high-income. Results were consistent for objective health measures, comparing respondents with at least four chronic conditions to those with one or none.

Conclusions: In a large, population-based study, both subjective and objective measures of health had a strong relationship with residential mobility outcomes.

The interventional effects approach to causal mediation analysis is increasingly common in epidemiologic research given its potential to address policy-relevant questions about hypothetical mediator interventions. Multiple imputation is widely used for handling multivariable missing data in epidemiologic studies. However, guidance is lacking on best practice for using multiple imputation when estimating interventional mediation effects, specifically regarding the role of missingness mechanism in the performance of the method, how to appropriately specify the multiple imputation model when g-computation is used for effect estimation, and appropriate variance estimation. To address this gap, we conducted simulations based on the Victorian Adolescent Health Cohort Study. We considered seven missingness mechanisms, involving varying assumptions regarding the influence of an intermediate confounder, a mediator, and/or the outcome on missingness in key variables. We compared the performance of complete-case analysis, six multiple imputation approaches by fully conditional specification, differing in how the imputation model was tailored, and a "substantive model compatible" multiple imputation-fully conditional specification approach. We evaluated MIBoot (multiple imputation, then bootstrap) and BootMI (bootstrap, then multiple imputation) approaches for variance estimation. All multiple imputation approaches, apart from those clearly diverging from best practice, yielded approximately unbiased estimates when none of the intermediate confounder, mediator, and outcome variables influenced missingness in any of these variables and non-negligible bias otherwise. We observed the largest bias for interventional effects when each of the intermediate confounders, mediators, and outcomes influenced their own missingness. BootMI returned variance estimates with smaller bias than MIBoot.

Background: The population attributable fraction corresponds to the reduction of the outcome had individuals (counter-to-fact) not experienced the exposure scaled by the observed incidence. Estimators proposed by Levin and Miettinen implicitly assume the study population is a random sample of the target population, which is not always the case.

Methods: In our example, we estimate the reduction in AIDS or death among women diagnosed with HIV in the United States in 2008 had they not had a history of injection drug use. To transport risk estimates from 1,164 women in the Women's Interagency HIV Study to the 11,282 women diagnosed with HIV in the United States in 2008, we use inverse probability of treatment and the inverse odds of sampling weighting. We estimate the variance of the population attributable fraction with a nonparametric bootstrap and M-estimation using the sandwich variance estimator.

Results: The population attributable fraction estimated in the observed sample was 0.21 (95% CI: 0.13, 0.29). After transporting the population attributable fraction to the target population, it was 0.13 (95% CI: 0.065, 0.19).

Conclusions: Defining the target population and identification conditions allows for a clearer interpretation of the population attributable fraction.

Background: Recent work conceptually unifying selection and collider-restriction bias as threats to internal validity implies that their impact on observed associations should similarly align. We reviewed epidemiologic literature to summarize existing knowledge about the impact of selection and collider bias.

Methods: We systematically searched for peer-reviewed, methodologic articles and general epidemiology textbooks published in English from January 1, 2000 through July 12, 2024. We included sources that focused on internal validity and discussed the magnitude or direction of selection or collider bias. We abstracted conclusions about the likely magnitude and direction of bias, which stratum or strata are affected when restricting analyses to a subset, and the conditions under which the consequences of bias were evaluated.

Results: : We retained 33 of 5,508 identified articles and 12 of 205 textbooks for data abstraction. Overall, we found that collider bias articles conveyed its impact as minimal while selection bias sources described variable effects. We also observed that most collider bias sources evaluated bias under the sharp null (assuming no relationship between the exposure and outcome) and found differences between how selection and collider bias sources discussed the role of interaction and the strata affected.

Conclusions: Although collider-restriction and selection bias affecting internal validity are considered theoretically equivalent, conclusions differ about their consequences for study results. Investigating collider bias not under the sharp null and considering the role of both multiplicative and additive interaction between the causes of a collider may improve our ability to predict and quantify its impact on internal validity.

Background: Time-series models for count outcomes are routinely used to estimate short-term health effects of environmental exposures. The dispersion parameter is universally assumed to be constant over the study period.

Objective: To examine whether dispersion depends on time-varying covariates in a case study of emergency department (ED) visits in Atlanta during 1999-2009, and to evaluate approaches for addressing time-varying dispersion.

Methods: Using the double generalized linear model (DGLM) framework, we jointly modeled the Poisson log-linear mean and dispersion to estimate associations between ED visits for respiratory diseases and daily ozone concentrations. We conducted a simulation study to evaluate the impact of time-varying overdispersion on health effect estimation when constant overdispersion is assumed and developed analytic code for implementing DGLM using R.

Results: We found dispersion to depend on calendar date and meteorology. Assuming constant dispersion, the relative risk (RR) per interquartile range increase in 3-day moving ozone exposure was 1.037 (95% CI: 1.024, 1.050). In multivariable dispersion model, the RR was reduced to 1.029 (95% CI: 1.020, 1.039), but with a large (26%) reduction in log RR standard error. The positive associations for ozone were robust against different dispersion model specifications. Simulation study results also demonstrated that when time-varying dispersion is present, it can lead to larger standard error assuming constant dispersion.

Conclusion: When the outcome exhibits large dispersion in a time-series analysis, allowing for covariate-dependent time-varying dispersion can improve inference, particularly by increasing estimation precision.

Background: Prior studies examining the association of cannabis use with nicotine abstinence did not distinguish between individuals co-using nicotine and cannabis versus those who switched from nicotine to exclusive cannabis use; these may have different effects on nicotine abstinence. We examined associations of cannabis use uptake with subsequent nicotine abstinence approximately 1 year later among adults using cigarettes and/or e-cigarettes.

Methods: Using six waves of the Population Assessment of Tobacco and Health Study (2013-2021), we assessed transitions from exclusive nicotine use pre-baseline (time t) to (1) exclusive cannabis use, (2) nicotine-cannabis co-use, (3) non-use of both nicotine and cannabis, and (4) continued exclusive nicotine use at baseline (t+1) as exposure variables. Analyses examined associations with nicotine abstinence (from both cigarettes and e-cigarettes) at 1-year follow-up (t+2).

Results: Among 8,382 adults (19,618 observations) reporting exclusive nicotine use pre-baseline, 1% transitioned to exclusive cannabis use, 9% to nicotine-cannabis co-use, and 9% to non-use of both drugs; 81% were still using nicotine exclusively at baseline. Transition to nicotine-cannabis co-use (6%) vs. exclusive nicotine use (10%) was inversely associated with nicotine abstinence at follow-up (adjusted relative risk [aRR]=0.68, 95%CI=0.55-0.83). Transition to exclusive cannabis use (72%) was positively associated with nicotine abstinence compared to continued exclusive nicotine use (10%) aRR=4.66, 95%CI=3.83-5.67), and with similar nicotine abstinence at follow-up (72%) compared to non-use of both drugs (65%; aRR=0.98, 95%CI=0.81-1.18).

Conclusions: Co-use of nicotine and cannabis was associated with lower nicotine abstinence. Switching to exclusive cannabis use was associated with similar or greater nicotine abstinence.

Background: Cognitive change is an important factor in understanding dementia. Estimating effects of exposures on cognitive change requires choosing an analytical timescale, typically time on study or current age. There is limited consensus regarding timescale choice in epidemiologic cognitive aging research.

Methods: Using a coordinated analytic approach in ten cohorts of older adults, we evaluated whether estimated effects of two exposures on memory change differed depending on timescale (time on study or current age). We modeled effects of APOE ε4 genotype (a time-invariant exposure) and diabetes (a potentially time-varying/acquired exposure evaluated at baseline) using mixed-effects models with linear and non-linear time specifications for both timescales.

Results: Among APOE ε4 carriers, model-estimated memory scores at baseline (time on study timescale) or at each cohort's median baseline age (current age timescale) were lower and average rate of decline was faster than among APOE ε4 noncarriers. Model-estimated memory scores at baseline or at median baseline age were generally similar across baseline diabetes status, with variability across cohorts in the diabetes-memory change association. In some cohorts, trends in diabetes-memory change associations differed in direction across timescales.

Conclusions: Timescale was largely inconsequential for estimated effects of APOE genotype, but yielded differences in estimated diabetes effects, raising questions about the appropriate scale. Current age scale may be problematic because diabetes was measured heterogeneously in age across individuals, a common issue in aging cohorts. Our work demonstrates approaches to evaluate alternative timescales, including in multi-cohort analyses, and highlights potential implications for estimated exposure effects on cognitive change.