Endocrine Abstracts

Introduction: Weight loss is a cornerstone in the management of prediabetes and diabetes mellitus. The association between weight and all-cause mortality is controversial and even paradoxical. Studies usually assess baseline weight. Studies assessing weight changes over time are limited. We present preliminary results of our attempt to model weight changes in patients with prediabetes and diabetes assessing the relationship with all-cause mortality.

Methods: We evaluated a retrospective cohort of Olmsted county residents with prediabetes (n=15868) and diabetes (n=10744) seen at Mayo Clinic. They had vital signs before and after 01-01-2005 and were alive by 01-01-2011. Body mass index (BMI) was calculated to assess weight changes and classified as Low or High using 32 as cutoff. The study period, 01-01-2000 to 01-01-2011 was divided in 3 stages: Baseline from 01-01-2000 to 12-31-2004 (5 years), Follow-up 1 from 01-01-2005 to 12-31-2007 (3 years) and Follow-up 2 from 01-01-2008 to 01-01-2011 (3 years). For each patient we defined six 3-step BMI trajectories by using BMI Low (L), High (H) or any value (X): 1) L-L-L: always low BMI. 2) H-H-H: always high BMI. 3) L-H-L: start and end with Low but had a High BMI at some point. 4) H-L-H: start and end with High but had a Low BMI at some point. 5) H-X-L: start with High but end with Low BMI. 6) L-X-H: start with Low but end with High BMI.

Results: As expected, diabetes mortality rates were higher than prediabetes, but their overall distribution by BMI trajectories was similar (Table). Subjects with high BMI (H-H-H) had lower mortality rate compared with low BMI (L-L-L). Subjects that ended with Low BMI, regardless of initial BMI level (L-H-L and H-X-L), had higher mortality than subjects ending with high BMI (H-L-H and L-X-H).

Conclusion: Our preliminary results evaluating BMI trajectories suggest elevated mortality in subjects with BMI below 32 when compared with BMI above 32, and subjects that lost weight when compared with those that gained weight. These results are contrary to our expectation and against our conceived pathophysiological interaction between obesity, diabetes and its complications. The fact that similar findings have been described for other clinical conditions, impose the need for additional research to find a suitable explanation related to methodology or clinical cause.