Citation

He H, Lu N, Stephens B, Xia Y, Bossarte RM, Kane CP, Tang W, Tu XM. Stat. Methods Med. Res. 2019; 28(2): 503-514.

Affiliation

Division of Biostatistics and Bioinformatics, Department of Family Medicine and Public Health, University of California, San Diego, CA, USA.

Copyright

(Copyright © 2019, SAGE Publishing)

DOI

10.1177/0962280217729843

PMID

28933251

Abstract

Large-scale public health prevention initiatives and interventions are a very important component to current public health strategies. But evaluating effects of such large-scale prevention/intervention faces a lot of challenges due to confounding effects and heterogeneity of study population. In this paper, we will develop metrics to assess the risk for suicide events based on causal inference framework when the study population is heterogeneous. The proposed metrics deal with the confounding effect by first estimating the risk of suicide events within each of the risk levels, number of prior attempts, and then taking a weighted sum of the conditional probabilities. The metrics provide unbiased estimates of the risk of suicide events. Simulation studies and a real data example will be used to demonstrate the proposed metrics.

Language: en

Keywords

Causal inference; effective sample size; metrics; multiple events; population heterogeneity; potential outcome