#138 The skinny on BMI and mortality
CLINICAL QUESTION
What is the association between body mass index (BMI) and mortality?
BOTTOM LINE
Normal (20-25) to overweight (25-30) BMI carry the lowest risk of mortality, with ~25 appearing lowest (in elderly ~27.5). Mortality increases when BMI is below “low-normal” (BMI <20) and obese (BMI ≥30), more at the extremes.
Reading Tools for Practice Article can earn you MainPro+ Credits
Join NowAlready a CFPCLearn Member? Log in
EVIDENCE
≥8 systematic reviews of observational studies in general population evaluating all-cause mortality at various BMI ranges.1-10 Focusing on the largest and most recent:
- Most studies1 (97 studies, 2.88 million participants) compared to normal (BMI 18.5-24.9), relative risks (RR):
- Overweight (BMI 25-29.9): RR=0.94.
- Obese Grade I (BMI 30-35): RR=0.95.
- Obese Grade ≥II (BMI >35): RR=1.29.
- Most participants2 (eight studies, 5.8 million participants) comparing to high normal BMI (22.5-25), hazard ratios (HR) for men:
- Low (BMI <18.5): HR=1.88.
- Low normal (BMI 18.5-20): HR=1.39.
- Mid normal (BMI 20-22.5): HR=1.15.
- High normal (BMI 22.5-25): HR=1.00.
- Low overweight (BMI 25-27.5): HR=0.97.
- High overweight (BMI 27.5-30): HR=1.04.
- Obesity Grade I (BMI 30-35): HR=1.18.
- Third largest3 (19 studies, 1.46 million participants) compared to BMI 22.5-24.9 for women:
- BMI <18.5-20: Increase mortality (HR=1.25).
- BMI 20-27.4: Very similar risk throughout range (HR=1.03-1.05).
- BMI >27.5: Mortality increases with BMI, examples:
- BMI 27.5-30: HR=1.14.
- BMI 40-50: HR=2.13.
- Others found similar.4-8
- Meta-analysis in specific populations:
- Diabetes:9 Similar to above.
- Elderly (age ≥65):1,10,11 Overweight lower risk (best ~27.5 BMI).10
- Pre-existing CVD,12-15 COPD,16 hemodialysis:17 Overweight and Grade I obesity similar risk12 or reduced risk13-17 relative to normal weight BMI.
- Confidence intervals not presented above: Trends of risk are more informative. Highest risk occurs at extremes of BMI with lowest risk occurring around 25 (27.5 in elderly). Minimal differences in HR/RR around 1 (e.g. 0.9-1.1) are likely of little clinical importance.
- Observational studies cannot prove causation.
- BMI indicates weight for height: Weight (in kilograms) divided by height (in metres) squared. BMI does not indicate fitness level.18
- Guidelines recommend the use of BMI as an assessment for obesity and intervention in individuals who are overweight and obese.19,20
Latest Tools for Practice
#377 How to slow the flow IV: Combined oral contraceptives
In premenopausal heavy menstrual bleeding due to benign etiology, do combined oral contraceptives (COC) improve patient outcomes?
#376 Testosterone supplementation for cis-gender men: Let’s (andro-)pause for a moment (Update)
What are the benefits and harms of testosterone supplementation in healthy cis-gender men or those with age-related low testosterone?
#375 Pharm for Fibro: Can antidepressants ease the pain?
Do antidepressants reduce pain in patients with fibromyalgia?
This really is not terribly helpful because there is no control for reverse causality in lower BMI ranges. Many life-limiting illnesses cause weight loss. This could lead to the type of interpretation error that had physicians recommending daily moderate alcohol as a health promoting intervention.