Article Text


177 Use of propensity scores in occupational health?
  1. A D Descatha1,
  2. Leclerc2,
  3. Herquelot3
  1. 1Garches, France
  2. 2Inserm, Villejuif, France
  3. 3UVSQ Inserm, Villejuif, France


Objectives Intervention studies are needed in occupational health in order to test if work improvements are effective, although not always possible. In the last decade, propensity score analysis has been widely used in clinical settings for treatments for which randomised controlled trials are unlikely to be conducted, such as socioeconomic factors in public health. It gives an estimate of a treatment effect in making treated and untreated group comparable for confounding factors by modelling treatment probability. We aimed to describe the frequency of propensity score cited in occupational health literature and its use.

Methods Using the key words “occupational” and “propensity score”, studies were selected from PubMed database (last 20 years). Frequency of citation was compared to two other requests “clinical” “propensity score” - “public health” not “occupational” “propensity score”. Embase and Web of sciences were also used to find extra papers. Relevant information was extracted including use of propensity score.

Results Twenty-four papers were found in PubMed with propensity score in occupational health [0.01% of all citations of this area, versus 1210 (0.06%), and 2731 (0.07%) in clinical or other public health domains respectively P < 0.001]. Among papers in occupational health, 12 were relevant, published since 2005, mostly from US teams (n = 10). The analyses included many subjects (median 7314), coming from pre-existing databases. The studies dealt with treatment evaluation (n = 2), economic evaluation (n = 2), and risk factors analysis (n = 5). Three papers studied implementation of programs in occupational setting, two for economic or quality purpose, one for effectiveness of multidisciplinary primary prevention program including work adaptation.

Conclusions Few studies used propensity score analysis, two used this method for evaluating interventions (safety or coaching programs) and only one considered work adaptations/rehabilitation. More widespread use of this methodology in large workers datasets might give information of efficiency of work adaptation when intervention studies are not suitable.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.