Instrumental Variable Analysis: Is The Cure Worse Than The Disease?

Main Category: Public Health
Article Date: 02 Jun 2009 - 6:00 PDT

email to a friend   printer friendly   view / write opinions   rate article

Causal inference is challenging in all non-experimental studies because of the possibility of hidden bias. Hidden bias may exist as a result of failure to control for unobservable factors, such as doctors' practice/prescription patterns.

The use of instrumental variable (IV) technique in outcomes research has increased in recent years because even in the presence of hidden bias, such methods may consistently estimate the average causal effects."However, like many techniques borrowed from one discipline and applied to another, there is a tendency to apply this method blindly." claimed Onur Baser, PhD, President with Stationed Research and Assistant Professor of Surgery with University of Michigan, and a sole author of the study.

In this paper, I provide an overview of the instrumental variable approach, examine possible tests to check the prerequisite conditions, and illustrate how weak instruments may produce inconsistent and inefficient results.

The immediate question that arises is if the IV method is superior to risk adjustment methods such as propensity score matching or multivariate regression in the sense that these methods both cover observable and unobservable factors, why not always use the IV method?

Says Dr. Baser, "If the correlation between the instrument and treatment is weak, this may lead to large inconsistencies in IV estimates with the bias in the same direction as that of OLS estimates. Since IV estimates also have larger standard errors than those of OLS estimates, "… the cure can be worse than the disease".

This will be discussed in Value in Health, the official journal of the International Society for Pharmacoeconomics and outcomes Research.

Value in Health (ISSN 1098-3015) publishes papers, concepts, and ideas that advance the field of pharmacoeconomics and outcomes research and help health care leaders to make decisions that are solidly evidence-based. The journal is published bi-monthly and has a regular readership of over 4,000 clinicians, decision-makers, and researchers worldwide.

ISPOR is a nonprofit, international organization that strives to translate pharmacoeconomics and outcomes research into practice to ensure that society allocates scarce health care resources wisely, fairly, and efficiently.

Source
ISPOR