Model‐based standardization to adjust for unmeasured cluster‐level confounders with complex survey data

Cai, Zhuangyu; Brumback, Babette

doi:10.1002/sim.6504

Cited by 4 publications

(4 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Specific methods for use in survival settings also exist and are the subject of ongoing research (26,33,61), but they are not discussed here. Also not covered here is the special case of uncontrolled confounding in multilevel or mixed model settings (10,32,39). Throughout, we focus on study settings in which the exposure-outcome association or effect is quantified using risk difference, mean difference, risk ratio (as in cohort studies), or odds ratio (as in case-control studies).…”

Section: Scopementioning

confidence: 99%

Bias Analysis for Uncontrolled Confounding in the Health Sciences

Arah

2017

Annu. Rev. Public Health

View full text Add to dashboard Cite

Uncontrolled confounding due to unmeasured confounders biases causal inference in health science studies using observational and imperfect experimental designs. The adoption of methods for analysis of bias due to uncontrolled confounding has been slow, despite the increasing availability of such methods. Bias analysis for such uncontrolled confounding is most useful in big data studies and systematic reviews to gauge the extent to which extraneous preexposure variables that affect the exposure and the outcome can explain some or all of the reported exposure-outcome associations. We review methods that can be applied during or after data analysis to adjust for uncontrolled confounding for different outcomes, confounders, and study settings. We discuss relevant bias formulas and how to obtain the required information for applying them. Finally, we develop a new intuitive generalized bias analysis framework for simulating and adjusting for the amount of uncontrolled confounding due to not measuring and adjusting for one or more confounders.

show abstract

Section: Scopementioning

confidence: 99%

Bias Analysis for Uncontrolled Confounding in the Health Sciences

Arah

2017

Annu. Rev. Public Health

View full text Add to dashboard Cite

show abstract

“…The parametric G-formula can provide valid PAF or GIF estimates by overcoming the limitations of the conventional methods for PAF and GIF estimation through generating a counterfactual population and using appropriate models. 48 , 49 Vangen-Lønne et al 50 applied the parametric G-formula to investigate the effect of joint interventions for complete or partial elimination of stroke risk factors on the 18-year cumulative stroke risk. Their findings showed that the risk of stroke would be reduced by 28% if SBP decreased to less than 140 mmHg in all individuals.…”

Section: Discussionmentioning

confidence: 99%

<p>Estimation of Generalized Impact Fraction and Population Attributable Fraction of Hypertension Based on JNC-IV and 2017 ACC/AHA Guidelines for Cardiovascular Diseases Using Parametric G-Formula: Tehran Lipid and Glucose Study (TLGS)</p>

Saatchi

Mansournia

Khalili

et al. 2020

RMHP

View full text Add to dashboard Cite

An area of interest to health policymakers is the effect of interventions aimed at risk factors on decreasing the number of new cardiovascular disease (CVD) cases. The aim of this study was to estimate the generalized impact fraction (GIF) and population attributable fraction (PAF) of hypertension (HTN) for CVD in Tehran. Patients and Methods: In this population-based cohort study, 8071 participants aged ≥30 years were followed for a median of 16 years. A survival model was used to estimate the 10-and 18-year risk of CVD. JNC-IV and 2017 ACC/AHA guidelines were used to categorize blood pressure (BP). PAF and GIF were estimated in different scenarios using the parametric G-formula. Results: Of 7378 participants included in analyses, 22.7% and 52.3% were classified as hypertensive according to the JNC-IV and 2017 ACC/AHA guidelines, respectively. According to the 2017 ACC/AHA, the 10-year risk of CVD was 5.1% (4.3-6.0%), 8.9% (6.7-12.0%), and 7.1% (6.1-8.4%) for normal BP, elevated BP, and stage 1 HTN, respectively, and 20.8% (18.8-23.0%) for stage 2 of the 2017 ACC/AHA and JNC-IV. The PAF of stage 2 vs stage 1 and vs normal BP for CVD was 17.4% (11.5-21.8%) and 20.4% (14.6-26.4%), respectively. The GIF of 30% reduction in the prevalence of stage 2 HTN to stage 1 and to normal BP for CVD was 5.1% (3.4-6.6%) and 6.1% (4.4-8.0%), respectively. Based on JNC-IV, the PAF and GIF of 30% for CVD were 17.8% (12.7-22.9%) and 5.4% (4.0-6.9%), respectively. Conclusion: By reducing the prevalence of HTN by 30%, a remarkable number of new CVD cases would be prevented. In an Iranian population, the comparison of HTN cases with normal BP showed no association between stage 1 HTN and CVD, whereas elevated BP was a significant risk factor for the incidence of CVD.

show abstract

“…Standard GLMM software, such as SAS PROC GLIMMIX or the R glmer function, can be used to estimate the vector of parameters θ = ( α 0 , β , γ , α X , α Z , α N , τ ) in the between‐within model. Brumback et al offered an example of a different choice for q (·), letting it depend on the maximum of X ij across j = 1,…, N i rather than on

{\overset{X}{true}}_{i}

.…”

Section: Outcome‐modeling Approachmentioning

confidence: 99%

“…However, when there is a categorical confounder that clusters the observations into a large number of small categories, model‐based standardization can fail due to the Neyman‐Scott problem, wherein maximum likelihood estimates of the parameters of the statistical model can be inconsistent. Recent research has extended the exposure‐modeling approach to address this problem by incorporating random effects; our goal is to investigate incorporation of random effects into the outcome‐modeling approach and to compare this to the exposure modeling approaches, particularly that of Skinner and D'Arrigo …”

Section: Introductionmentioning

confidence: 99%

Model‐based standardization using an outcome model with random effects

Wang

Brumback

Alrwisan

et al. 2019

Statistics in Medicine

Self Cite

View full text Add to dashboard Cite

Model-based standardization uses a statistical outcome model or exposure model to estimate a population-average association that is unconfounded by selected covariates. With it, one can compare groups using a distribution of confounders identical in each group to that of a standard population. We develop an approach based on an outcome model, in which the mean of the outcome is modeled conditional on the exposure and the confounders. In our approach, there is a confounder that clusters the observations into a very large number of categories. We treat the parameters for the clusters as random effects. We use a between-within model to account for the association of the random effects not only with the exposure but also with the cluster population sizes.We review alternative approaches presented in the literature, and we compare the outcome-modeling approach to recently proposed exposure-modeling approaches incorporating random effects. To illustrate, we use 2014 to compare proportions of acute respiratory tract infection diagnoses with an antibiotic prescription for emergency department versus outpatient visits, adjusting for confounding by unmeasured patient level variables and measured diagnosis-level variables. We also present results of a simulation study. KEYWORDS causal inference, confounding, generalized linear mixed models, marginal effect, model-based standardization 3378

show abstract

Model‐based standardization to adjust for unmeasured cluster‐level confounders with complex survey data

Cited by 4 publications

References 17 publications

Bias Analysis for Uncontrolled Confounding in the Health Sciences

Bias Analysis for Uncontrolled Confounding in the Health Sciences

<p>Estimation of Generalized Impact Fraction and Population Attributable Fraction of Hypertension Based on JNC-IV and 2017 ACC/AHA Guidelines for Cardiovascular Diseases Using Parametric G-Formula: Tehran Lipid and Glucose Study (TLGS)</p>

Model‐based standardization using an outcome model with random effects

Contact Info

Product

Resources

About