Article Text

Download PDFPDF

The effect of shift work on ischaemic heart disease
  1. G Yadegarfar1,
  2. R McNamee2
  1. 1
    Department of Biostatistics and Epidemiology, School of Public Health, Isfahan University of Medical Sciences, Isfahan, Iran
  2. 2
    School of Epidemiology and Health Sciences, Manchester University, Manchester, UK
  1. G Yadegarfar, Department of Epidemiology and Biostatistics, School of Public Health, Isfahan University of Medical Sciences, Hezar-Jerib Road, Isfahan, Iran; g_yadegarfar{at}

Statistics from

We would like to clarify some of the issues raised by Professor Knutsson (Occup Environ Med 2008;65:152) in his recent commentary concerning our article “Shift work, confounding and death from ischaemic heart disease”.1 Firstly, it is not true that previously “a similar study was carried out at the same factory”. As we state in our article, the subjects of the present study were from a different cohort from that investigated by McNamee et al2 in 1996. The cohorts of the two studies had the same employer but correspond to two different factory sites approximately 100 miles apart. Thus, the recent study provides new evidence which is independent of the 1996 publication.

We agree with Professor Knutsson that one should not adjust for intermediate variables that might lie on the causal chain between shift work and outcome. In our article we noted that such adjustments would be “inappropriate” and we adjusted only for pre-employment “metabolic” variables (typically measured about 1 month before taking up employment at the company). It is true that both day and shift workers in our cohort may have had some prior experience of shift work with a different employer. Thus Professor Knutsson appears to argue that our adjustment for pre-entry metabolic measures might still constitute a form of over-adjustment. While we cannot rule this out, it is important to note that the adjustment had almost minimal impact, changing the relative risk for shift work from 1.09 to 1.10; thus an unadjusted analysis would have given virtually the same results throughout the article. Furthermore, in unreported analyses, there was no evidence of interaction between shift work and the baseline variables.

Adjustment for “social class” (inferred through job titles) reduced the adjusted relative risk to 1.04. Professor Knutsson considered the meaning of social class so derived, for an industrial population, speculating that workers assigned to different social classes might differ more in terms of other “work exposures” than in personal characteristics such as living conditions: therefore it would have been ideal to consider directly the impact of such other exposures. In fact we addressed other exposures in our article; for example, we noted that excessive noise was a possible risk factor for ischaemic heart disease3 and that workers were exposed to noise in some degree. Adjustment for noise did not greatly change the relative risk of death for shift workers in comparison with day workers (OR 1.05, 90% CI 0.84 to 1.32).

In summary, we do not believe that over- or under-adjustment is a major issue in explaining the results of our study. However, we agree with Professor Knutsson that it is important in shift work research to distinguish genuine confounders from intermediate variables: in work being prepared for publication, we are investigating longitudinal changes in body mass index and blood pressure during employment in our cohort members.



  • Competing interests: None.

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.

Linked Articles