Search

Background: Previous studies have reported conflicting results about the associations of adiposity with risk of all-cause mortality in Hispanic populations. Moreover, the relevance of different markers of adiposity for cause-specific mortality may differ between Hispanic and other populations. Methods: Between 1998 and 2004, 159 755 adults aged ≥35 years from Mexico City were enrolled into a prospective study. Follow-up for cause-specific mortality was until 1st January 2018. Cox regression was used to estimate the relative risks (RRs) of death from specific causes associated with general adiposity (total body mass [BMI] and fat mass index [FMI]) and with abdominal adiposity (waist circumference, waist-hip ratio [WHR], and waist-height ratio [WHtR]), both overall and independently of each other, after adjustment for confounders. To minimise bias from reverse causation, the main analyses excluded those with screen-detected or diagnosed diabetes, other chronic diseases, and deaths in the first 5 years of follow-up. Excess attributable mortality and survival curves were derived from the RRs combined with national mortality rates. Results: Among those aged 35-89 years when recruited, mean (standard deviation [SD]) BMI was 29.5 (4.9) kg/m2 and 34% were obese (>30 kg/m2). Both BMI and FMI had J-shaped associations with all-cause mortality, whereas waist circumference, WHR and WHtR were all linearly and positively related to mortality throughout the ranges studied. Each 5 kg/m2 higher BMI above 25 kg/m² was associated with 32% higher risk of death at ages 40-74 years, chiefly due to a 38% higher risk of death from vascular or metabolic diseases. Across the full ranges, 1 SD higher levels of the adiposity markers were associated with 21% higher mortality at ages 40-74 years for BMI, 22% higher mortality for FMI, 25% higher mortality for waist circumference, 17% higher mortality for WHR, and 26% higher mortality for WHtR. For deaths from a vascular or metabolic cause, these estimates were 29%, 31%, 32%, 24% and 37%, respectively. Adjustment of BMI, WHR and WHtR for other markers of adiposity resulted in only a modest attenuation of the strength of the associations of these markers with mortality. By contrast, adjustment of waist circumference for the other markers of adiposity strengthened its association with mortality. Similarly, after adjustment for lean mass index, the association of FMI with all-cause mortality was strengthened. The various adiposity markers also displayed associations with several non-vascular non-metabolic causes of death. In particular, all of the adiposity markers were strongly related to death from respiratory and infective causes, but none were strongly associated with cancer mortality. Associations were broadly similar in men and women, somewhat stronger at younger than at older ages, and consistent according to levels of other confounders. Sensitivity analyses replicated the results of the main findings. Among those with diabetes, adiposity (particularly BMI) was inversely related to mortality, reflecting a reversal of causality caused by weight loss resulting from poor glycaemic control. The excess mortality associated with BMI above 30 kg/m² compared with BMI below 25 kg/m² accounted for 14% of all vascular and metabolic deaths. Applying the findings to current national Mexican death rates, compared with a BMI of 25 kg/m² life expectancy from age 40 years was reduced by about 3 years at a BMI of 30 kg/m², but by almost 10 years at a BMI of 40 kg/m². Conclusions: Consistent with findings in Western populations, both general, and especially, abdominal adiposity, were strongly predictive of all-cause and cause-specific mortality in this Mexican cohort, thereby refuting the 'Hispanic obesity paradox'. The failure of previous studies of Hispanic populations to detect such associations is likely due to their failure to account for reverse causality bias (arising from weight loss due to poorly controlled diabetes). Further studies are needed to elucidate the causal relevance of the associations with different markers of adiposity, and to determine the mechanisms underlying such associations.