Multiple-Model-based Robust Estimation of Causal Treatment Effect on a Binary Outcome with Integrated Information from Secondary Outcomes

Abstract–

An assessment of the causal treatment effect in the development and progression of certain diseases is important in clinical trials and biomedical studies. However, it is not possible to infer a causal relationship when the treatment assignment is imbalanced and confounded by other mechanisms. Specifically, when the treatment assignment is not randomized and the primary outcome is binary, a conventional logistic regression may not be valid to elucidate any causal inference. Moreover, exclusively capturing all confounders is extremely difficult and even impossible in large-scale observational studies. We propose a multiple-model-based robust (MultiMR) estimator for estimating the causal effect with a binary outcome, where multiple propensity score models and conditional mean imputation models are used to ensure estimation robustness. Furthermore, we propose an enhanced MultiMR (eMultiMR) estimator that reduces the estimation variability of MultiMR estimates by incorporating secondary outcomes that are highly correlated with the primary binary outcome. The resulting estimates are less sensitive to model mis-specification compared to those based on state-of-the-art doubly-robust methods. These estimates are verified through both theoretical and numerical assessments. The utility of (e)MultiMR estimation is illustrated using the Uniform Data Set (UDS) from the National Alzheimer’s Coordinating Center with the objective of detecting the causal effect of the short-term use of antihypertensive medications on the development of dementia or mild cognitive impairment. The proposed method has been implemented in an R package and is available at https://github.com/chencxxy28/eMultiMR.

KEYWORDS:

• Binary outcomes
• Causal inference
• Information integration
• Multiple-model-based robust estimation
• Secondary data

Supplemental Materials

The supplementary material provide the numerical procedure for solving the constrained optimization stated in (8) and (10), as well as the technical proofs for Lemma 1 and Theorem 1 presented in the main article.

R-package for eMultiMR: The proposed method has been implemented in an R package and is available at https://github.com/chencxxy28/eMultiMR.

Disclosure Statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.

Additional information

Funding

Wang’s work was supported by the startup funding from Department of Population and Quantitative Health Sciences at Case Western Reserve University. Long’s work was partially supported by Grants RF1AG063481 and R01AG071174 from the NIH. The contents of this article are solely the responsibility of the authors, and do not represent the official views of NIH. This article was prepared using the NACC database, which is funded by National Institute of Aging (NIA) of the NIH Grant U24 AG072122. NACC data are contributed by the NIA-funded ADRCs: P30 AG062429 (PI James Brewer, MD, PhD), P30 AG066468 (PI Oscar Lopez, MD), P30 AG062421 (PI Bradley Hyman, MD, PhD), P30 AG066509 (PI Thomas Grabowski, MD), P30 AG066514 (PI Mary Sano, PhD), P30 AG066530 (PI Helena Chui, MD), P30 AG066507 (PI Marilyn Albert, PhD), P30 AG066444 (PI John Morris, MD), P30 AG066518 (PI Jeffrey Kaye, MD), P30 AG066512 (PI Thomas Wisniewski, MD), P30 AG066462 (PI Scott Small, MD), P30 AG072979 (PI David Wolk, MD), P30 AG072972 (PI Charles DeCarli, MD), P30 AG072976 (PI Andrew Saykin, PsyD), P30 AG072975 (PI David Bennett, MD), P30 AG072978 (PI Neil Kowall, MD), P30 AG072977 (PI Robert Vassar, PhD), P30 AG066519 (PI Frank LaFerla, PhD), P30 AG062677 (PI Ronald Petersen, MD, PhD), P30 AG079280 (PI Eric Reiman, MD), P30 AG062422 (PI Gil Rabinovici, MD), P30 AG066511 (PI Allan Levey, MD, PhD), P30 AG072946 (PI Linda Van Eldik, PhD), P30 AG062715 (PI Sanjay Asthana, MD, FRCP), P30 AG072973 (PI Russell Swerdlow, MD), P30 AG066506 (PI Todd Golde, MD, PhD), P30 AG066508 (PI Stephen Strittmatter, MD, PhD), P30 AG066515 (PI Victor Henderson, MD, MS), P30 AG072947 (PI Suzanne Craft, PhD), P30 AG072931 (PI Henry Paulson, MD, PhD), P30 AG066546 (PI Sudha Seshadri, MD), P20 AG068024 (PI Erik Roberson, MD, PhD), P20 AG068053 (PI Justin Miller, PhD), P20 AG068077 (PI Gary Rosenberg, MD), P20 AG068082 (PI Angela Jefferson, PhD), P30 AG072958 (PI Heather Whitson, MD), P30 AG072959 (PI James Leverenz, MD).