Epidemiology最新文献

Epidemiologic analyses that aim to quantify exposure effects on disease progression are not uncommon. Understanding the implications of these studies, however, is complicated, in part because different causal estimands could, at least in theory, be the target of such analyses. Here, to facilitate interpretation of these studies, we describe different settings in which causal questions related to disease progression can be asked, and consider possible estimands. For clarity, our discussion is structured around settings defined based on two factors: whether the disease occurrence is manipulable or not, and the type of outcome. We describe relevant causal structures and sets of response types, which consist of joint potential outcomes of disease occurrence and disease progression, and argue that settings where interventions to manipulate disease occurrence are not plausible are more common, and that, in this case, principal stratification might be an appropriate framework to conceptualize the analysis. Further, we suggest that the precise definition of the outcome of interest, in particular of what constitutes its permissible levels, might determine whether potential outcomes linked to disease progression are definable in different strata of the population. Our hope is that this paper will encourage additional methodological work on causal analysis of disease progression, as well as serve as a resource for future applied studies.

Institutions of higher education faced a number of challenges during the COVID-19 pandemic. Chief among them was whether or not to reopen during the second wave of COVID-19 in the fall of 2020, which was controversial because incidence in young adults was on the rise. The migration of students back to campuses worried many that transmission within student populations would spread into surrounding communities. In light of this, many colleges and universities implemented mitigation strategies, with varied degrees of success. Washington State University, located in the city of Pullman in Whitman County, WA, is an example of this type of university-community co-location, where the role of students returning to the area for the fall 2020 semester was contentious. Using COVID-19 incidence in Pullman, WA, reported to the Whitman County Health Department, we retrospectively study the transmission dynamics that occurred between the student and community subpopulations in fall 2020. We develop a two-population ordinary differential equations mechanistic model to infer transmission rates within and across the university student and community subpopulations. We use results from Bayesian parameter estimation to determine if exponential transmission of COVID-19 occurred in Pullman, WA, and the magnitude of cross-transmission from students to community members. We find these results are consistent with the estimation of the time-varying reproductive number that outbreak potential was minimal and resolved quickly, and conclude that the students returning to Washington State University-Pullman did not place the surrounding community at disproportionate risk of COVID-19 during fall 2020 when mitigation efforts were in place.

Background: Some studies examining associations between maternal injuries and preterm birth reported null or counterintuitive protective effects, especially for 3rd-trimester injuries, likely due to time-related biases.

Methods: This retrospective cohort study comprised all births occurring at the Atrium Health Wake Forest Baptist health system between 2018 and 2024. We ascertained maternal injuries using validated diagnostic codes and defined preterm birth as gestational age at delivery <37 weeks. We estimated associations between maternal injuries and preterm birth with two approaches. We used logistic regression for time-fixed analysis (injury at any point in pregnancy yes/no and preterm birth yes/no) and Cox proportional hazards models for time-varying analysis (i.e., time-varying injury definition, restricted follow-up to periods when pregnancies were at risk of preterm birth).

Results: Among 58,897 births, 1,801 women (3.1%) experienced maternal injuries during pregnancy. With the time-varying approach, maternal injuries were associated with increased risk of preterm birth (adjusted hazard ratio [HR]: 1.16; 95% confidence interval [CI] = 1.01, 1.32). Trimester-specific analyses showed positive associations for all trimesters, with higher effect estimates observed for 2nd and 3rd trimester injuries (adjusted HRs: 1.17; 95% CI = 0.97, 1.42) and 1.22 (95% CI = 0.92, 1.61), respectively. With time-fixed analyses, associations for any injury were underestimated, compared with time-varying analyses, and results for 3rd trimester injuries showed counterintuitive negative associations (adjusted odds ratio: 0.73 [0.54, 0.98]).

Conclusions: Time-related biases typically underestimate associations between maternal injuries and preterm birth, particularly for 3rd - trimester injuries. Rigorous study design and analytical methods that account for time-related biases are crucial in studies investigating adverse outcomes after maternal injuries.

Background: Incidence rates and cumulative incidences estimated using registers (e.g., electronic healthcare records) might be biased by including cases diagnosed before the inception of the register as being at risk. Washout periods can identify and exclude prevalent cases from analyses, but the impact of washout duration on estimates is unknown. We estimated risks of mental disorders according to different washout period durations.

Methods: This population-based cohort included all 6,478,162 individuals aged 1-80 years living in Denmark in 2010-2021. Using hospital contacts in 2010-2021, we estimated incidence rates and cumulative incidence of mental disorders according to different washout period durations (0, 1, 2, 5, 15, and 41 years) based on hospital contacts prior to 2010.

Results: Without a washout period, the lifetime cumulative incidence of any mental disorder was 49.4% (95% confidence interval [CI]: 49.2%, 49.5%) for females and 45.1% (95% CI: 45.0%, 45.2%) for males. Estimates decreased when we increased the washout, reaching a lifetime incidence of 40.3% (95% CI: 40.1%, 40.4%) for females and 36.6% (95% CI: 36.5%, 36.8%) for males when using all available data (41 years of washout). Without a washout period, estimates for specific mental disorder types were up to 60% higher than those obtained with the maximum washout period, but the bias in absolute terms depended on the underlying risks.

Conclusions: While including all cases identifiable in a register may decrease uncertainty, the inclusion of prevalent cases as being at risk may lead to substantially overestimated measures. We highlight the need for caution when using administrative registers and electronic healthcare databases.

Background: Sequential positivity is often a necessary assumption for drawing causal inferences, such as through marginal structural modeling. Unfortunately, verification of this assumption can be challenging because it usually relies on multiple parametric propensity score models, unlikely to all be correctly specified. Therefore, we propose a new algorithm, called sequential Positivity Regression Tree (sPoRT), to overcome this issue and identify the subgroups found to be violating this assumption, allowing for insights about the nature of the violations and potential solutions.

Methods: We present different versions of sPoRT based on either stratifying or pooling over time under static or dynamic treatment strategies. This methodologic development was motivated by a real-life application of the impact of the timing of initiation of HIV treatment with and without smoothing over time, which we also use to demonstrate the method.

Results: The illustration of sPoRT demonstrates its easy use and the interpretability of the results for applied epidemiologists. Furthermore, an R notebook showing how to use sPoRT in practice is available at github.com/ArthurChatton/sPoRT-notebook.

Conclusions: The sPoRT algorithm provides interpretable subgroups violating the sequential positivity violation, allowing patterns and trends in the confounders to be easily identified. We finally provided practical implications and recommendations when positivity violations are identified.

Background: Intimate partner violence (IPV) is an important global health issue for which measurement error limits public health action. Although most national IPV prevalence estimates come from general health surveys like the Demographic and Health Surveys (DHS), such data probably underestimate prevalence compared with violence-focused surveys.

Methods: Using violence-focused surveys conducted in the same country and year (±1) as validation data, we explored two methods of bias adjustment to address measurement error in DHS prevalence estimates. In multidimensional bias analysis, we directly adjusted summary prevalence estimates, using a range of possible sensitivities (10%-100%) and specificities (95%-100%) to elucidate their reasonable bounds. In multiple overimputation, we reestimated all IPV observations, incorporating prior information on measurement error, and averaged prevalence estimates over 50 iterations.

Results: Multidimensional bias analysis revealed that an assumption of 95% specificity resulted in negative prevalence estimates in some cases, confirming that false positives are likely negligible. Reasonable sensitivities varied considerably across countries and IPV types, likely due to differences in the number of items used to assess IPV. Multiple overimputation-adjusted estimates were similar to survey estimates, except when unadjusted DHS estimates were <5% and highly discrepant. Past-year estimates were less discrepant than lifetime estimates, suggesting that recall bias may be a factor in underreporting.

Conclusion: This study examines measurement error due to IPV underreporting in specific contexts where external information exists, highlighting the need for more accurate IPV assessment using multiple items per domain and for internal validation studies to be incorporated into large-scale surveys.

Background: The Super Learner is an ensemble learning method that has been widely used with doubly robust causal effect estimators. It is recommended to deploy the Super Learner with a diverse library of algorithms. To our knowledge, however, the magnitude of the improvements gained by including many algorithms has not yet been systematically evaluated in common epidemiologic research settings.

Methods: We applied Super Learning with two doubly robust estimators, augmented inverse probability weighting (AIPW) and targeted minimum loss-based estimation (TMLE), to estimate the average treatment effect (ATE) of high periconceptional dietary fruit and vegetable density on the risk of preeclampsia among 7,923 women from the nuMoM2b study. Using a reference ensemble with a diverse library of algorithms, we compared estimates under different sets of algorithms included in the Super Learner to evaluate whether ATE estimates were sensitive to library choices.

Results: The doubly robust estimators fitted with the reference Super Learner ensemble suggested ≥2.5 cups/1,000 kcal of total fruit and vegetable density was associated with a lower risk of preeclampsia. ATE estimated on the risk difference scale by AIPW was -0.019 (95% confidence interval = -0.036, -0.003) and by TMLE was -0.023 (95% confidence interval = -0.039, -0.007). Excluding any individual algorithm from the reference ensemble had little impact on estimates from either AIPW or TMLE. However, relying on a single algorithm (e.g., extreme gradient boosting) yielded results that were much more variable.

Conclusion: Our empirical findings support recommendations to build ensemble learners for doubly robust estimators using a diverse array of flexible machine learning algorithms.

Background: The effects of life course financial hardship on fertility have not been well studied.

Methods: We examined the association between life course financial hardship and fecundability in Pregnancy Study Online (PRESTO), a preconception cohort study of US and Canadian pregnancy planners aged 21-45 years who identified as female (2013-2023; N = 6,377). We followed participants up to 12 months or until pregnancy. Participants reported financial hardship in childhood (≤11 years), adolescence (12-17 years), and adulthood (≥18 years) via three questions: not having enough money for living expenses; needing to borrow money for medical expenses; or receiving public assistance. We used inverse probability-weighted proportional probabilities models to estimate fecundability ratios (FRs) and 95% confidence intervals (CIs), accounting for time-dependent confounding and selection bias.

Results: Compared with no financial hardship, financial hardship during any life stage was associated with slightly reduced fecundability (FR = 0.93, 95% CI: 0.86, 1.0). Associations were similar for financial hardship during childhood and adolescence; however, those experiencing financial hardship during adulthood had lower fecundability (FR = 0.83, 95% CI: 0.77, 0.90). The association between adolescent financial hardship and fecundability was similar among those with and without childhood financial hardship. However, the association of adult financial hardship with fecundability was stronger among those who experienced hardship earlier in life (i.e., adult financial hardship among those with child/adolescent financial hardship: FR = 0.77; 95% CI: 0.64, 0.93).

Conclusion: Adulthood is a sensitive period for the effects of financial hardship on fecundability. Moreover, cumulative financial hardship across the life course was associated with greater reductions in fecundability.

Rigid prespecification can be impractical for noninterventional studies using secondary datasets, where data-driven flexibility is often required. Using target trial emulations comparing immunomodulator treatments for COVID-19, we piloted an adaptive strategy that accommodates warranted mid-course refinements within a prespecified framework. Our preregistered protocol outlined an initial study plan along with predetermined diagnostic thresholds and contingencies. Implementation proceeded through sequential phases, allowing researcher decisions to be guided by prespecified criteria under varying degrees of blinding to results. The adaptive approach led to alterations in the underlying target trial and to the analysis plan used for emulation, strengthening the plausibility of causal assumptions and improving the relevance of findings. During the initial baseline phase, indicated contingencies included sample restrictions, redefining treatments from class-level to product-specific comparisons, a revised propensity score model, and weight truncation. In the subsequent postbaseline phase, diagnostic checks triggered a modified causal contrast, inverse probability of censor weighting to address noncompliance, cause-specific hazard estimation to contextualize competing events, and additional reporting of hazard ratios for progressively truncated follow-up periods. For a secondary study objective, the adaptive framework allowed for some iterative attempts to improve validity while providing a clear stopping point. Similar approaches could lend transparent structure to the process of learning what causal questions the data are equipped to support. Beyond guarding against researcher bias, prespecification of adaptive protocols may promote more robust designs by encouraging investigators to be explicit about their assumptions, strategies for interrogating those assumptions, and specific criteria for determining when and how deviations may be required.