Published in:
01-10-2016 | Original Research
BMI Trajectories as a Harbinger of Pre-Diabetes or Underdiagnosed Diabetes: an 18-Year Retrospective Cohort Study in Taiwan
Published in: Journal of General Internal Medicine | Issue 10/2016
Login to get accessAbstract
Background
Although prior studies have examined BMI trajectories in Western populations, little is known regarding how BMI trajectories in Asian populations vary between adults with and without diabetes.
Objective
To examine how BMI trajectories vary between those developing and not developing diabetes over 18 years in an Asian cohort.
Design
Multilevel modeling was used to depict levels and rates of change in BMI for up to 18 years for participants with and without self-reported physician-diagnosed diabetes.
Participants
We used 14,490 data points available from repeated measurements of 3776 participants aged 50+ at baseline without diabetes from a nationally representative survey of the Taiwan Longitudinal Study on Aging (TLSA1989-2007).
Main Measures
We defined development of diabetes as participants who first reported diabetes diagnoses in 2007 but had no diabetes diagnoses at baseline. We defined the reference group as those participants who reported the absence of diabetes at baseline and during the entire follow-up period.
Key Results
When adjusted for time-varying comorbidities and behavioral factors, higher level and constant increases in BMI were present more than 6.5 years before self-reported diabetes diagnosis. The higher BMI level associating with the development of diabetes was especially evident in females. Within 6.5 years prior to self-reported diagnosis, however, a wider range of decreases in BMI occurred (βdiabetes = 1.294, P = 0.0064; βdiabetes*time = 0.150, P = 0.0327; βdiabetes*time
2 = −0.008, P = 0.0065). The faster rate of increases in BMI followed by a greater decline was especially prominent in males and individuals with BMI ≧24.
Conclusions
An unintentional decrease in BMI in sharp contrast to the gradually rising BMI preceding that time may be an alarm for undiagnosed diabetes or a precursor to developing diabetes.