The enigma of increased non-cancer mortality after weight loss in healthy men who are overweight or obese

Objective. To study effects on non-cancer mortality of observational weight loss in middle-aged men stratified for body mass index (BMI), taking a wide range of possible confounders into account. Design. Prospective, population based study. Setting. Male population of Malmo, Sweden. Participants. In all 5722 men were screened twice with a mean time interval of 6 years in Malmo, southern Sweden. They were classified according to BMI category at baseline (<21, 22-25, overweight: 26-30, and obesity: 30+ kg m(-2) ) and weight change category until second screening (weight stable men defined as having a baseline BMI +/- 0.1 kg m(-2) year(-1) at follow-up re-screening). Main outcome measures. Non-cancer mortality calculated from national registers during 16 years of follow-up after the second screening. Data from the first year of follow-up were excluded to avoid bias by mortality caused by subclinical disease at re-screening. Results. The relative risk (RR; 95% CI) for non-cancer mortality during follow-up was higher in men with decreasing BMI in all subgroups: RR 2.64 (1.46-4.71, baseline BMI <21 kg m(-2) ), 1.39 (0.98-1.95, baseline BMI 22-25 kg m(-2) ), and 1.71 (1.18-2.47, baseline BMI 26+ kg m(-2) ), using BMI-stable men as reference group. Correspondingly, the non-cancer mortality was also higher in men with increasing BMI, but only in the obese group (baseline BMI 26+ kg m(-2) ) with RR 1.86 (1.31-2.65). In a subanalysis, nonsmoking obese (30+ kg m(-2) ) men with decreased BMI had an increased non-cancer mortality compared with BMI-stable obese men (Fischer's test: P =0.001). The mortality risk for nonsmoking overweight men who increased their BMI compared with BMI-stable men was also significant (P =0.006), but not in corresponding obese men (P =0.094). Conclusions. Weight loss in self-reported healthy but overweight middle-aged men, without serious disease, is associated with an increased non-cancer mortality, which seems even more pronounced in obese, nonsmoking men, as compared with corresponding but weight-stable men. The explanation for these observational findings is still enigmatic but could hypothetically be because of premature ageing effects causing so-called weight loss of involution.