Weight loss cut risk scores. A year later, much came back.

Randomised controlled trial (RCT). 212 people. 800 kcal/day diet until BMI < 25 (average 159 days), then followed 1 year.

A trial. Chance decided who did what.

Who: Sedentary, healthy premenopausal women aged 21-49 with overweight (BMI 27-30), normal glucose tolerance, non-smokers, Birmingham, Alabama, USA; 221 enrolled (115 African American, 106 European American), 212 with baseline risk data, 131 at post-weight loss, 105 at 1 year

• Diet only, after loss-5.2 ptspredicted risk
• Diet only, 1 year on+4.0 ptsas weight returned
• Diet + weights, 1 yr-2.4 ptsvs start

More to know, from our read of the paper:

• Outcomes are PREDICTED 10-year risks from risk equations (CMDS: age, BMI, blood pressure, glucose, lipids...), not actual diabetes or heart events. Say 'predicted risk' / 'risk score'
• Heavy dropout: 212 with baseline risk, 131 at post-weight loss, 105 at 1 year. Analysis assumes missing data are missing at random; authors flag possible bias
• Only healthy, sedentary premenopausal women with overweight (BMI 27-30); authors say results may not apply to men, older people, people with obesity or existing disease
• Heart (MACE) risk was tiny to begin with (~0.75-0.94% over 10 years); changes are fractions of a percentage point. Don't show MACE numbers as if large
• The one significant between-group difference: Diet-only had a larger MACE-risk drop during weight loss than Diet+Resistance (adjusted p = 0.022)

In a randomized trial, 221 sedentary premenopausal women with overweight ate 800 calories a day until they reached a BMI under 25, with cardio, weights, or diet only. Predicted 10-year diabetes risk fell in every group, for example from 13.6% to 8.5% on diet alone. A year later, as weight partly came back, it rose again (to 12.2% on diet alone). Only the diet-plus-weights group was still clearly below its starting risk, but the groups were not clearly different from each other. These are predicted risks, not actual diagnoses, and many women dropped out.

We read the abstract and the full text of this paper and checked every number on this page against it (17 facts, checked on 2026-10-29). The study design is named above. Cohort and other observational studies are described as linked, not proven. Found an error? Email theis@onestudyaday.com.