Predictors of follow-up and assessment of selection bias from dropouts using inverse probability weighting in a cohort of university graduates

Dropouts in cohort studies can introduce selection bias. In this paper, we aimed (i) to assess predictors of retention in a cohort study (the SUN Project) where participants are followed-up through biennial mailed questionnaires, and (ii) to evaluate whether differential follow-up introduced selection bias in rate ratio (RR) estimates. The SUN Study recruited 9907 participants from December 1999 to January 2002. Among them, 8647 (87%) participants answered the 2-year follow-up questionnaire. The presence of missing information in key variables at baseline, being younger, smoker, a marital status different of married, being obese/overweight and a history of motor vehicle injury were associated with being lost to follow-up, while a self-reported history of cardiovascular disease predicted a higher retention proportion. To assess whether differential follow-up affected RR estimates, we studied the association between body mass index and the risk of hypertension, using inverse probability weighting (IPW) to adjust for␣confounding and selection bias. Obese individuals had a higher crude rate of hypertension compared with␣normoweight participants (RR = 6.4, 95% confidence interval (CI): 3.9–10.5). Adjustment for confounding using IPW attenuated the risk of hypertension associated to obesity (RR = 2.4, 95% CI: 1.1–5.3). Additional adjustment for selection bias did not modify the estimations. In conclusion, we show that the follow-up through mailed questionnaires of a geographically disperse cohort in Spain is possible. Furthermore, we show that despite existing differences between retained or lost to follow-up participants this may not necessarily have an important impact on the RR estimates of hypertension associated to obesity.     

Impact of mis-specification of the treatment model on estimates from a marginal structural model

Inverse probability of treatment weighted (IPTW) estimation for marginal structural models (MSMs) requires the specification of a nuisance model describing the conditional relationship between treatment allocation and confounders. However, there is still limited information on the best strategy for building these treatment models in practice. We developed a series of simulations to systematically determine the effect of including different types of candidate variables in such models. We explored the performance of IPTW estimators across several scenarios of increasing complexity, including one designed to mimic the complexity typically seen in large pharmacoepidemiologic studies.Our results show that including pure predictors of treatment (i.e. not confounders) in treatment models can lead to estimators that are biased and highly variable, particularly in the context of small samples. The bias and mean-squared error of the MSM-based IPTW estimator increase as the complexity of the problem increases. The performance of the estimator is improved by either increasing the sample size or using only variables related to the outcome to develop the treatment model. Estimates of treatment effect based on the true model for the probability of treatment are asymptotically unbiased.We recommend including only pure risk factors and confounders in the treatment model when developing an IPTW-based MSM.     

A flexible parametric approach for estimating continuous‐time inverse probability of treatment and censoring weights

Marginal structural Cox models are used for quantifying marginal treatment effects on outcome event hazard function. Such models are estimated using inverse probability of treatment and censoring (IPTC) weighting, which properly accounts for the impact of time‐dependent confounders, avoiding conditioning on factors on the causal pathway. To estimate the IPTC weights, the treatment assignment mechanism is conventionally modeled in discrete time. While this is natural in situations where treatment information is recorded at scheduled follow‐up visits, in other contexts, the events specifying the treatment history can be modeled in continuous time using the tools of event history analysis. This is particularly the case for treatment procedures, such as surgeries. In this paper, we propose a novel approach for flexible parametric estimation of continuous‐time IPTC weights and illustrate it in assessing the relationship between metastasectomy and mortality in metastatic renal cell carcinoma patients. Copyright © 2016 John Wiley & Sons, Ltd.     

Bayesian estimation of the average treatment effect on the treated using inverse weighting

We develop a Bayesian approach to estimate the average treatment effect on the treated in the presence of confounding. The approach builds on developments proposed by Saarela et al in the context of marginal structural models, using importance sampling weights to adjust for confounding and estimate a causal effect. The Bayesian bootstrap is adopted to approximate posterior distributions of interest and avoid the issue of feedback that arises in Bayesian causal estimation relying on a joint likelihood. We present results from simulation studies to estimate the average treatment effect on the treated, evaluating the impact of sample size and the strength of confounding on estimation. We illustrate our approach using the classic Right Heart Catheterization data set and find a negative causal effect of the exposure on 30-day survival, in accordance with previous analyses of these data. We also apply our approach to the data set of the National Center for Health Statistics Birth Data and obtain a negative effect of maternal smoking during pregnancy on birth weight.     

The impact of selection bias on vaccine effectiveness estimates from test-negative studies

IntroductionEstimates of vaccine effectiveness (VE) from test-negative studies may be subject to selection bias. In the context of influenza VE, we used simulations to identify situations in which meaningful selection bias can occur. We also analyzed observational study data for evidence of selection bias.MethodsFor the simulation study, we defined a hypothetical population whose members are at risk for acute respiratory illness (ARI) due to influenza and other pathogens. An unmeasured “healthcare seeking proclivity” affects both probability of vaccination and probability of seeking care for an ARI. We varied the direction and magnitude of these effects and identified situations where meaningful bias occurred. For the observational study, we reanalyzed data from the United States Influenza VE Network, an ongoing test-negative study. We compared “bias-naïve” VE estimates to bias-adjusted estimates, which used data from the source populations to correct for sampling bias.ResultsIn the simulation study, an unmeasured care-seeking proclivity could create selection bias if persons with influenza ARI were more (or less) likely to seek care than persons with non-influenza ARI. However, selection bias was only meaningful when rates of care seeking between influenza ARI and non-influenza ARI were very different. In the observational study, the bias-naïve VE estimate of 55% (95% CI, 47-–62%) was trivially different from the bias-adjusted VE estimate of 57% (95% CI, 49-–63%).ConclusionsIn combination, these studies suggest that while selection bias is possible in test-negative VE studies, this bias in unlikely to be meaningful under conditions likely to be encountered in practice. Researchers and public health officials can continue to rely on VE estimates from test-negative studies.     

The impact of side effect framing on COVID-19 booster vaccine intentions in an Australian sample

ObjectiveTo evaluate the effect of presenting positively attribute-framed side effect information on COVID-19 booster vaccine intention relative to standard negatively-framed wording and a no-intervention control.Design and participantsA representative sample of Australian adults (N = 1204) were randomised to one of six conditions within a factorial design: Framing (Positive; Negative; Control) × Vaccine (Familiar (Pfizer); Unfamiliar (Moderna)).InterventionNegative Framing involved presenting the likelihood of experiencing side effects (e.g., heart inflammation is very rare, 1 in every 80,000 will be affected), whereas Positive Framing involved presenting the same information but as the likelihood of not experiencing side effects (e.g., 79,999 in every 80,000 will not be affected).Primary outcomeBooster vaccine intention measured pre- and post-intervention.ResultsParticipants were more familiar with the Pfizer vaccine (t(1203) = 28.63, p <.001, Cohen’s d_z = 0.83). Positive Framing (M = 75.7, SE = 0.9, 95% CI = [73.9, 77.4]) increased vaccine intention relative to Negative Framing (M = 70.7, SE = 0.9, 95% CI = [68.9, 72.4]) overall (F(1, 1192) = 4.68, p =.031, η_p² = 0.004). Framing interacted with Vaccine and Baseline Intention (F(2, 1192) = 6.18, p =.002, η_p² = 0.01). Positive Framing was superior, or at least equal, to Negative Framing and Control at increasing Booster Intention, irrespective of participants’ pre-intervention level of intent and vaccine type. Side effect worry and perceived severity mediated the effect of Positive vs. Negative Framing across vaccines.ConclusionPositive framing of side effect information appears superior for increasing vaccine intent relative to the standard negative wording currently used.Pre-registrationSee: aspredicted.org/LDX_2ZL.     

A marginal structural model for multiple‐outcome survival data:assessing the impact of injection drug use on several causes of death in the Canadian Co‐infection Cohort

It is often the case that interest lies in the effect of an exposure on each of several distinct event types. For example, we are motivated to investigate in the impact of recent injection drug use on deaths due to each of cancer, end‐stage liver disease, and overdose in the Canadian Co‐infection Cohort (CCC). We develop a marginal structural model that permits estimation of cause‐specific hazards in situations where more than one cause of death is of interest. Marginal structural models allow for the causal effect of treatment on outcome to be estimated using inverse‐probability weighting under the assumption of no unmeasured confounding; these models are particularly useful in the presence of time‐varying confounding variables, which may also mediate the effect of exposures. An asymptotic variance estimator is derived, and a cumulative incidence function estimator is given. We compare the performance of the proposed marginal structural model for multiple‐outcome data to that of conventional competing risks models in simulated data and demonstrate the use of the proposed approach in the CCC. Copyright © 2013 John Wiley & Sons, Ltd.