Possible pitfalls in the prediction of weight gain in middle-aged normal-weight individuals: Results from the NDB-K7Ps-study-2

Abstract

BackgroundThe prevalence of obesity has not decreased worldwide and obesity-related morbidities have been increasing steadily. However, few studies have investigated factors contributing to weight gain in normal-weight individuals. Thus, in this community-based cohort study, we aimed to investigate factors contributing to weight gain in normal-weight participants. MethodsClinical variables and 10 % increase in weight over 10 years (10 %IBW10Y) were retrospectively investigated in apparently healthy 615,077 normal-weight (body mass index (BMI) 21.0–24.9 kg/m2) participants aged 40–64 years who had regularly undergone health checkup. Machine learning and logistic regression analysis (nested case-control study) were used to predict 10 %IBW10Y. ResultsIn total, 6.8 % of men and 8.9 % of women had 10 %IBW10Y (P < 0.0001). The prevalence of obesity (BMI ≥25.0 kg/m2) after 10 years and weight gain were higher in participants with 10 %IBW10Y (72.3 %, 8.9 kg) (case-group) versus those without 10 %IBW10Y (11.5 %, −0.4 kg) (control-group), respectively. Machine learning showed positive contributing factors to 10 %IBW10Y were, in descending order, age early 40 s, current smoking, female sex, low serum triglyceride (≤59 mg/dL), and low glycated hemoglobin (HbA1c) (≤4.9 %). Age early 60 s, non-smoking, male sex, high triglyceride (≥200 mg/dL), and HbA1c 6.0 %−6.9 % were negative contributing factors. Logistic regression analysis showed similar results except for high HbA1c (≥7.5 %) as a positive contributing factor. ConclusionsIn middle-aged individuals with normal weight who undergo regular health check-ups, certain favorable features (female sex, low triglyceride, and low HbA1c), as well as smoking habits that are subject to change in the future, which could lead to weight gain, may be overlooked.250 ＜250 words