Hypnotics and mortality--partial confounding by disease, substance abuse and socioeconomic factors?

The aim of this Cohort study of 10 527 Danish men was to investigate the extent to which the association between hypnotics and mortality is confounded by several markers of disease and living conditions.

Exposure was purchases of hypnotics 1995-1999 ("low users" (150 or less defined daily dose (DDD)) or "high users" (151 or more DDD)). Follow-up for all-cause mortality was from 1 Jan 2000 to 19 June 2010. Cox proportional hazard models were used to study the association. Covariates were entered one at a time and simultaneously. Results were reported using hazard ratio (HR) and 95% confidence intervals (CI).

When covariates were entered one at a time, the changes in HR estimates showed that psychiatric disease, socioeconomic position and substance abuse reduced the excess risk by 17-36% in the low user group and by 45-52% in the high user group. Somatic disease, intelligence score and cohabitation reduced the excess risk by 2-11% in the low user group and 8-24% in the high user group. When adjusting for all covariates, the HR was reduced to 1.22 95% CI (0.97-1.54) in the low user group and 1.43 95% CI (1.11-1.85) in the high user group.

The results of this study point at psychiatric disease, substance abuse and socioeconomic position as potential confounding factors partly explaining the association between use of hypnotics and all-cause mortality.