To determine whether nonresponse to a mailed health survey predicts mortality in patients with rheumatoid arthritis (RA) and in a community sample in Finland.
A 5-page health questionnaire was administered in 2000. Two years later the vital status of the subjects was ascertained from the Population Registry.
A total of 1095 (73%) patients with RA and 1530 (77%) community control subjects returned a completed questionnaire. Over the 2-year period, the number of deaths was 57 (5.2%) in RA responders and 37 (9.3%) in RA nonresponders (p = 0.004). The corresponding figures in community controls were 34 (2.2%) and 23 (4.9%) (p = 0.002). In a Cox regression model adjusted for age and sex, RA patient and community control nonresponders were respectively 1.65 (95% CI 1.07 to 2.55) and 2.89 (95% CI 1.69 to 4.94) times more likely to die over the 2 years compared to the responders.
Nonresponders to a mailed health survey were more likely to die over 2 years compared to responders. The possible nonresponse bias should be kept in mind in the interpretation of the results of studies that are based on mail questionnaires only.