Instrumental variable analyses. Exploiting natural randomness to understand causal mechanisms

Abstract

Instrumental variable analysis is a technique commonly used in the social sciences to provide evidence that a treatment causes an outcome, as contrasted with evidence that a treatment is merely associated with differences in an outcome. To extract such strong evidence from observational data, instrumental variable analysis exploits situations where some degree of randomness affects how patients are selected for a treatment. An instrumental variable is a characteristic of the world that leads some people to be more likely to get the specific treatment we want to study but does not otherwise change those patients' outcomes. This seminar explains, in nonmathematical language, the logic behind instrumental variable analyses, including several examples. It also provides three key questions that readers of instrumental variable analyses should ask to evaluate the quality of the evidence. (1) Does the instrumental variable lead to meaningful differences in the treatment being tested? (2) Other than through the specific treatment being tested, is there any other way the instrumental variable could influence the outcome? (3) Does anything cause patients to both receive the instrumental variable and receive the outcome?

Figure 1.

Causal diagram for instrumental variables. (A) General framework. We need to study whether a treatment causes differences in outcomes. (B) A confounder affects treatment and outcome, so we cannot do a simple comparison between treated and untreated patients. (C) An instrumental variable can provide a solution. (D) The instrumental variable only works if the only connection between the instrumental variable and the outcomes is through the treatment.

Figure 2.

A first example relating instrumental variables to randomized control trials (RCTs). (A) RCT with perfect compliance. Note the absence of an arrow between the confounder and the treatment. (B) RCT with imperfect compliance. If harder to ventilate patients with worse lung disease are more likely to be protocol deviations, then confounding re-enters.

Figure 3.

Using an instrumental variable to examine whether having a third child influences parents’ productivity at work.

Figure 4.

Using an instrumental variable to examine whether referral to long-term acute care hospitals (LTACs) results in improved 1-year survival for critically ill patients.

Figure 5.

Attempts to use an instrumental variable analysis to understand the impact of availability of percutaneous coronary intervention (PCI) on acute myocardial infarction (AMI) outcomes. (A) Question as originally formulated. (B) Reality of alternate pathway. (C) Reformulated question that could be answered with instrumental variable.