Causal Inference in Studying the Long-term Health Effects of Disasters: Challenges and Potential Solutions

2021 ◽  
Author(s):  
Koichiro Shiba ◽  
Takuya Kawahara ◽  
Jun Aida ◽  
Katsunori Kondo ◽  
Naoki Kondo ◽  
...  
Keyword(s):  
Causal Inference ◽  
Health Effects ◽  
Selection Bias ◽  
Proportional Hazards ◽  
Cox Regression ◽  
Causal Effect ◽  
Time Varying ◽  
Average Causal Effect ◽  
Effect Estimation

Abstract Two frequently encountered but underrecognized challenges for causal inference in studying the long-term health effects of disasters among survivors include: (a) time-varying effects of disasters on a time-to-event outcome and (b) selection bias due to selective attrition. We review approaches to overcome these challenges and show application of the approaches to a real-world longitudinal data of older adults who were directly impacted by the 2011 earthquake and tsunami (n=4,857). To illustrate the problem of time-varying effects of disasters, we examined the association between degree of damage due to the tsunami and all-cause mortality. We compared results from Cox regression assuming proportional hazards versus adjusted parametric survival curves allowing for time-varying hazard ratios. To illustrate the problem of selection bias, we examined the association between proximity to the coast (a proxy for housing damage from the tsunami) and depressive symptoms. We corrected for selection bias due to attrition in the two post-disaster follow-up surveys (conducted in 2013 and 2016) using multivariable adjustment, inverse probability censoring weighting, and survivor average causal effect estimation. Our results demonstrate that the analytic approaches ignoring time-varying effects on mortality and selection bias due to selective attrition may underestimate the long-term health effects of disasters.

Abstract TMP83: Long-term Risk of Venous Thromboembolism After Stroke

Stroke ◽  
2017 ◽  
Vol 48 (suppl_1) ◽  
Author(s):  
Santosh B Murthy ◽  
Alexander E Merkler ◽  
Gino Gialdini ◽  
Abhinaba Chatterjee ◽  
Costantino Iadecola ◽  
...  
Keyword(s):  
Ischemic Stroke ◽  
Venous Thromboembolism ◽  
Proportional Hazards ◽  
Cox Regression ◽  
Multivariable Analysis ◽  
First Year ◽  
Cumulative Rate ◽  
Incident Cases

Background: There are few data on the long-term risk of venous thromboembolism (VTE) among stroke survivors. We aimed to compare the incidence of VTE amongst patients with ischemic stroke versus those with intracerebral hemorrhage (ICH). Methods: We identified all adults discharged from nonfederal acute care hospitals in CA, NY, and FL between 2005 and 2012 with previously validated ICD-9-CM codes for ischemic stroke and ICH. Our primary outcome of VTE was defined as pulmonary embolism or deep vein thrombosis. To capture incident cases of VTE, we excluded patients with a VTE prior to or during the index stroke. Kaplan-Meier survival statistics were used to calculate the cumulative rate of incident VTE. Cox regression was used to compare the risk of VTE after stroke while adjusting for demographics, vascular risk factors, and Elixhauser comorbidity index. As there was a violation of the proportional-hazards assumption, we calculated separate hazard ratios (HR) for each year of follow-up. Results: We identified 834,660 patients with stroke, of whom 712,440 (85.3%) had ischemic stroke and 112,220 (14.7%) had ICH. Over a mean follow-up of 2.8 (+/-2.4) years, 19,937 (2.4%) developed VTE. After 7 years, the cumulative rate of VTE was 4.7% (95% confidence interval [CI], 4.5-4.9%) in patients with ICH and 4.4% (95% CI, 4.3-4.5%) in patients with ischemic stroke. In multivariable analysis, VTE risk was higher in the first year after ICH compared to ischemic stroke (HR 1.51; 95% CI, 1.43-1.58). However, following the first year, the hazard of VTE was higher among patients with ischemic stroke versus those with ICH (Figure). Conclusions: The risk of VTE after stroke varies by stroke type and time. Patients with ICH have a higher risk of VTE in the first year after stroke as compared to those with ischemic stroke while patients with ischemic stroke have a higher risk beyond 1 year.

scholarly journals P1462IMPACT OF HEART RATE AND BETA-BLOCKER USE ON THE LONG-TERM SURVIVAL OF CHRONIC HEMODIALYSIS PATIENTS

2020 ◽  
Vol 35 (Supplement_3) ◽  
Author(s):  
Takuhiro Moromizato ◽  
Kunitoshi Iseki ◽  
OCTOPUS Study Group
Keyword(s):  
Blood Pressure ◽  
Heart Rate ◽  
Cox Regression ◽  
Chronic Hemodialysis ◽  
Time Varying ◽  
Term Survival ◽  
Blood Pressures ◽  
Β Blocker ◽  
Time Varying Covariates

Abstract Background and Aims Increase in resting heart rate might influence mortalities of dialysis patients, and the use of β-blocker might improve their survival probability. However, the influence of heart rate and benefits of β-blocker on their survival are difficult to quantify because of following obstacles: prone to measurement errors; inherent association of heart rate with blood pressures, comorbidities, and medication use; and a necessity of repeated measurements of vital signs and medication use. Therefore, at the design process of our previous randomized control trial on the Olmesartan Clinical Trial in Okinawan patients under OKIDS (OCTOPUS), we included the repeated measures design to quantify the influence of vital sign values on the survival retrospectively. We combined the repeated measurement data and additional the long-term prognosis information of the participants obtained after the OCTOPUS with aim of investigating the influence of time varying covariates: heart rates, blood pressures, and β-blocker use, on the long-term survival of hemodialysis patients. Method We investigated 461 adult OCTOPUS participants who received chronic hemodialysis and antihypertensive medications in Okinawa. The OCTOPUS trial, which was conducted between June 2006 and June 2011, did not detect the survival benefit of angiotensin receptor blocker (ARB)NDT 2013, but the study and the additional follow-up of participants’ prognosis provided us with information to investigate influence of predictors on long-term survival in the population. Throughout the OCTOPUS trial, study participants were measured pre-dialysis blood pressures, pre-dialysis resting heart rates, and their medication use for one week at their dialysis centers every six months after their participations. Following the trial, we collected the prognosis information of all participants until July 31st, 2018. Finally, we merged the multiple-measured data during the OCTOPUS with the prognosis data. Mean values of three measurements of blood pressures and heart rates and β-blocker use were introduced to the Cox-regression model as time-varying covariates with essential non-time varying covariates, which include age, gender, and diabetes. Results In this retrospective cohort study, 221 (47.9%) out of 461 participants deceased, and the median follow-up length was 10.21 years. Initial mean resting heart rate and pre-dialysis mean blood pressure were 78(±10) per minute and 159.5(±14) mmHg, respectively. 10% of participants were prescribed β-blocker initially. The resting heart rate of all participants significantly decreased by 1.75 and 2.45 per minutes after two and four years respectively. β-blocker could significantly decrease the mean heart rate by 3.54 and 2.90 per minutes after two and four years. With our Cox-regression with the time varying covariates, increase of heart rate was significantly associated with higher mortality (P=0.002), but the use of β-blocker was not associated with the mortality. (P=0.691) Additionally, we could not detect the interaction of heart rate and β-blocker use on the mortality. (P= 0.796) Although lower blood pressure was significantly associated with higher mortality in our initial Cox-regression analysis, an introduction of interaction term of heart rate and blood pressure remove the significance of influence of blood pressure on the survival. Conclusion In hypertensive chronic hemodialysis patients, higher heart rate is associated with higher mortality. However, use of beta-blocker was not associated with improvement of their mortality.

scholarly journals Causal inference under over-simplified longitudinal causal models

2021 ◽  
Vol 0 (0) ◽  
Author(s):  
Lola Étiévant ◽  
Vivian Viallon
Keyword(s):  
Causal Inference ◽  
Causal Effect ◽  
Sufficient Conditions ◽  
Repeated Measurements ◽  
Causal Models ◽  
Causal Effects ◽  
Sensitivity Analyses ◽  
Time Varying ◽  
Weighted Averages

Abstract Many causal models of interest in epidemiology involve longitudinal exposures, confounders and mediators. However, repeated measurements are not always available or used in practice, leading analysts to overlook the time-varying nature of exposures and work under over-simplified causal models. Our objective is to assess whether – and how – causal effects identified under such misspecified causal models relates to true causal effects of interest. We derive sufficient conditions ensuring that the quantities estimated in practice under over-simplified causal models can be expressed as weighted averages of longitudinal causal effects of interest. Unsurprisingly, these sufficient conditions are very restrictive, and our results state that the quantities estimated in practice should be interpreted with caution in general, as they usually do not relate to any longitudinal causal effect of interest. Our simulations further illustrate that the bias between the quantities estimated in practice and the weighted averages of longitudinal causal effects of interest can be substantial. Overall, our results confirm the need for repeated measurements to conduct proper analyses and/or the development of sensitivity analyses when they are not available.

scholarly journals The Bayesian Causal Effect Estimation Algorithm

2015 ◽  
Vol 3 (2) ◽  
pp. 207-236 ◽  
Author(s):  
Denis Talbot ◽  
Geneviève Lefebvre ◽  
Juli Atherton
Keyword(s):  
Model Selection ◽  
Causal Inference ◽  
Bayesian Model Averaging ◽  
Causal Effect ◽  
Model Averaging ◽  
Estimation Algorithm ◽  
Continuous Exposure ◽  
Statistical Model Selection ◽  
Effect Estimation ◽  
Adjusted Model

AbstractEstimating causal exposure effects in observational studies ideally requires the analyst to have a vast knowledge of the domain of application. Investigators often bypass difficulties related to the identification and selection of confounders through the use of fully adjusted outcome regression models. However, since such models likely contain more covariates than required, the variance of the regression coefficient for exposure may be unnecessarily large. Instead of using a fully adjusted model, model selection can be attempted. Most classical statistical model selection approaches, such as Bayesian model averaging, do not readily address causal effect estimation. We present a new model averaged approach to causal inference, Bayesian causal effect estimation (BCEE), which is motivated by the graphical framework for causal inference. BCEE aims to unbiasedly estimate the causal effect of a continuous exposure on a continuous outcome while being more efficient than a fully adjusted approach.

Sign in / Sign up

Export Citation Format

Share Document