Premium
The error of using returns‐to‐work to measure the outcomes of health care
Author(s) -
Baldwin Marjorie L.,
Johnson William G.,
Butler Richard J.
Publication year - 1996
Publication title -
american journal of industrial medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.7
H-Index - 104
eISSN - 1097-0274
pISSN - 0271-3586
DOI - 10.1002/(sici)1097-0274(199606)29:6<632::aid-ajim7>3.0.co;2-l
Subject(s) - medicine , measure (data warehouse) , work (physics) , health care , actuarial science , environmental health , data mining , mechanical engineering , business , engineering , economics , economic growth , computer science
Abstract This article uses data from The Survey of Ontario Workers With Permanent Impairments , the world's largest survey of injured workers, to show that, as currently used, return‐to‐work is a misleading measure of the effectiveness of health care. The article discusses examples of two serious limitations on the use of return‐to‐work to measure the outcomes of health care, where health care refers to all the medical and rehabilitative services provided to a worker following a workplace injury. The first limitation is that return‐to‐work, like many other outcomes of health care, is influenced by factors that are not directly related to health care. Using a logit model to estimate the determinants of first absences from work after an injury, we find that socioeconomic characteristics, economic incentives, and job characteristics have a significant influence on return‐to‐work. The second limitation on return‐to‐work as an outcome measure is that the first return‐to‐work after an injury, like a hospital discharge, frequently marks the end of only the first of several episodes of work disability caused by the original injury. Using first post‐injury returns‐to‐work as a proxy for recovery, we would assume that 85% of the Ontario workers recovered from their injury when, in fact, 61% had subsequent spells of work disability. We identified four mutually exclusive patterns of post‐injury work and work disability. Multinomial logit estimates of the determinants of the patterns show that health care is only one of several influences on return‐to‐work. The results also demonstrate that if return‐to‐work is used to measure outcomes, it must be evaluated over a time horizon that permits multiple spells of work disability. © 1996 Wiley‐Liss, Inc.