Skip to main content
Log in

An Introduction to Sensitivity Analysis for Unobserved Confounding in Nonexperimental Prevention Research

  • Published:
Prevention Science Aims and scope Submit manuscript

Abstract

Despite the fact that randomization is the gold standard for estimating causal relationships, many questions in prevention science are often left to be answered through nonexperimental studies because randomization is either infeasible or unethical. While methods such as propensity score matching can adjust for observed confounding, unobserved confounding is the Achilles heel of most nonexperimental studies. This paper describes and illustrates seven sensitivity analysis techniques that assess the sensitivity of study results to an unobserved confounder. These methods were categorized into two groups to reflect differences in their conceptualization of sensitivity analysis, as well as their targets of interest. As a motivating example, we examine the sensitivity of the association between maternal suicide and offspring’s risk for suicide attempt hospitalization. While inferences differed slightly depending on the type of sensitivity analysis conducted, overall, the association between maternal suicide and offspring’s hospitalization for suicide attempt was found to be relatively robust to an unobserved confounder. The ease of implementation and the insight these analyses provide underscores sensitivity analysis techniques as an important tool for nonexperimental studies. The implementation of sensitivity analysis can help increase confidence in results from nonexperimental studies and better inform prevention researchers and policy makers regarding potential intervention targets.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. The original study estimated hazard ratio of 1.80 with a 95 % confidence interval of 1.19, 2.74.

  2. Note that this is not a loss of generality; if the unobserved confounder is negatively associated with exposure status, we could simply redefine the unobserved confounder to meet this scenario.

References

Download references

Acknowledgments

The authors wish to acknowledge the NARSAD Young Investigator Award to Dr. Holly C. Wilcox, and Dr. Holly C. Wilcox for allowing us to use the motivating example. We also thank the National Institute of Drug Abuse for the training support for S. Janet Kuramoto (1F31DA0263182), the National Institute of Mental Health (NIMH) Prevention Research T32 Training Grant for the training support for Weiwei Liu (T32 MH18834), and NIMH for the support of Elizabeth Stuart’s time (K25 MH083846). This work was performed while Weiwei Liu was a postdoctoral fellow and S. Janet Kuramoto was a student at Johns Hopkins Bloomberg School of Public Health.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Weiwei Liu.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

ESM 1

(DOCX 72 kb)

Rights and permissions

Reprints and permissions

About this article

Cite this article

Liu, W., Kuramoto, S.J. & Stuart, E.A. An Introduction to Sensitivity Analysis for Unobserved Confounding in Nonexperimental Prevention Research. Prev Sci 14, 570–580 (2013). https://doi.org/10.1007/s11121-012-0339-5

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11121-012-0339-5

Keywords

Navigation