The European Journal of Public Health Advance Access originally published online on March 19, 2009
The European Journal of Public Health 2009 19(5):548-553; doi:10.1093/eurpub/ckp022
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Miscellaneous |
Self-reporting weight and height: misclassification effect on the risk estimates for acute myocardial infarction
Andreia Oliveira1,2,3, Elisabete Ramos1,2,3, Carla Lopes1,2,3 and Henrique Barros1,2,3
1 Department of Hygiene and Epidemiology of University of Porto Medical School, Porto, Portugal.
2 Cardiovascular Research & Development Unit of University of Porto Medical School, Porto, Portugal.
3 Public Health Institute of Porto University, Porto, Portugal.
Correspondence: Andreia Oliveira, Department of Hygiene and Epidemiology, University of Porto Medical School, Alameda Prof. Hernâni Monteiro, 4200-319 Porto, Portugal, tel: +351 225513652, fax: +351 225513653, e-mail: acmatos{at}med.up.pt
Received October 28, 2008 , accepted February 3, 2009
Background: The accuracy of self-reported weight and height to measure obesity has been evaluated, but no information is available on the possible error effects of self-reporting when estimating the association between body mass index (BMI) categories and the occurrence of acute myocardial infarction (AMI). We aim to evaluate if two different sources of information on height and weight (reported vs. measured) result in different risk estimates for non-fatal AMI events. Methods: A population-based case–control study was conducted with 732 cases of first AMI and 1914 community controls, recruited from the same catchment area of hospitals. As part of an interview, participants self-reported weight and height immediately before having it measured. Data were analysed separately by sex and age strata (
45; >45 years). Results: Women under-reported their weight and over-reported their height, and the mean differences between measured and self-reported data were significantly larger in controls. Male controls also under-reported their weight, but cases over-reported it. After adjustment, in younger women the use of self-reported data underestimated the AMI risk according to BMI categories, but in older ones the self-reporting overestimated AMI risk, although with no statistical significance. In younger men, the association between AMI and self-reported obesity (BMI 
30 kg/m2) was overestimated in 
50% (measured: OR = 2.05, 95% CI 1.08–3.87; self-reported: OR = 3.06, 95% CI 1.56–6.00). In older participants, a significant association was only found for overweight men when using self-reported data. Conclusions: Self-reporting of height and weight produced a differential misclassification and biased risks for AMI according to BMI, affecting not only the magnitude, but also the estimates direction.
Keywords: bias, body height, body weight, case–control studies, myocardial infarction