Changes in body mass index (BMI) across a person’s life can help to gauge the likelihood of undiagnosed diabetes, say researchers.
In a cohort of 16,226 Japanese men and 7026 women aged 30 to 75 years, people’s lifetime maximum BMI and change in BMI between age 20 years and the lifetime maximum were both strongly associated with undiagnosed diabetes.
Indeed, in patients aged 50 years or older, these two indices predicted diabetes more strongly than current BMI.
“This is possibly because both weight histories are less likely to be influenced by an age-dependent decline in the association between obesity and diabetes”, observe lead study author Hirohito Sone (Niigata University Faculty of Medicine, Japan) and colleagues.
The prevalence of undiagnosed diabetes was 4.0% in men and 1.7% in women. Lifetime maximum BMI was the strongest predictor of diabetes in both genders, with each standard deviation (SD) increment raising the likelihood 1.58-fold in men and 1.65-fold in women. The associations were independent of variables including age, parental history of diabetes, hypertension and dyslipidaemia.
Change in BMI between age 20 years and lifetime maximum was also significantly associated with diabetes, at odds ratios (ORs) of 1.47 and 1.63 in men and women, respectively, as was current BMI, at corresponding ORs of 1.47 and 1.61.
The most useful BMI measure varied with age, however. In patients aged 50 years or older, lifetime maximum BMI and change between age 20 and lifetime maximum were the strongest predictors of diabetes, whereas in younger patients, lifetime maximum and current BMI were equally predictive.
The highest rates of undiagnosed diabetes were seen among people in the top tertile for both maximum BMI and BMI change (indicating a large and rapid weight gain), the team reports in Diabetic Medicine.
However, the effect of BMI history was modified by current BMI, in that a low current BMI was protective. For example, the prevalence of undiagnosed diabetes among men in the top tertile of maximum BMI was 1.81% if they were also in the top tertile of current BMI, but just 0.04% if they were in the bottom tertile.
The researchers conclude: “These results suggest that guidelines for screening for diabetes might be made more effective by introducing past high levels of BMI into the screening assessment.”
