Menopause Weight Gain: Planetary and Low-Insulinemic Diets Led

Somewhere around forty-eight, a familiar arithmetic stops working. The weight that used to answer to a fortnight of smaller dinners and a few more walks now sits where it is, and the number on the scale has drifted up by a couple of kilograms a year without anything obvious having changed. The advice arriving from every direction is the same advice as always, only louder: eat well, move more. What it never says is which of the dozen named ways of eating currently on offer is the one worth the effort.

That is the gap the week's coverage stepped into, and the study underneath it is unusually well suited to the question. Rather than testing one diet against a control, researchers led by Tong Xia at the Harvard T.H. Chan School of Public Health scored the same women against twelve different dietary patterns at once and asked which ones tracked the least weight gain through the menopausal transition. The answer is narrower and less flattering to the usual headline diets than the coverage suggested.

The analysis draws on the Nurses' Health Study II, a prospective cohort that has followed American female nurses since 1989. This paper takes 38,283 of those women, mean age 45.6 years (SD, 3.0), and watches a twelve-year window centred on menopause, using data running to 2019. Diet was recorded every four years with a validated food frequency questionnaire rather than a single baseline snapshot, so a woman whose eating changed across the window is scored on how she actually ate as she went.

Each woman's diet was then converted into twelve separate scores: the plant-based diet index and its healthy and unhealthy variants, the Mediterranean diet, Dietary Approaches to Stop Hypertension, the Planetary Health Diet Index, a low-carbohydrate score along with its healthy and unhealthy variants, an empirical dietary inflammatory pattern, an empirical dietary index for hyperinsulinemia, and ultraprocessed food intake. Two outcomes were measured: annual change in self-reported body weight, in kilograms per year, and whether a woman crossed into obesity. Generalised estimating equations handled the weight change, Cox proportional hazards models the obesity risk.

The adjustment list matters for reading what follows. The models account for age, race and ethnicity, marital status, income, postmenopausal hormone therapy use, parity, smoking, alcohol, total energy intake, physical activity and baseline body mass index. Over 340,122 person-years of follow-up, 5,214 women developed obesity, and the cohort as a whole gained an average of 0.80 kg per year (SD, 1.00). The data analysis was performed between November 2024 and May 2025.

The largest reduction in weight gain went to the low-insulinemic pattern, which the paper measures as the reverse of its hyperinsulinemia index. Comparing the top fifth of that score with the bottom fifth, the difference was 0.28 kg per year less weight gained (95% CI, -0.30 to -0.26). Set against the cohort's average 0.80 kg per year, that is roughly a third of the annual drift. Multiplied out across the full twelve-year window it would come to something over three kilograms, though the paper reports the annual rate rather than a twelve-year total.

For crossing into obesity, the ranking flipped order. The lowest risk belonged to the Planetary Health Diet Index, with a hazard ratio of 0.46 (95% CI, 0.42 to 0.51), followed by the low-insulinemic score at 0.51 (95% CI, 0.46 to 0.56). Those are large numbers by the standards of nutritional epidemiology, and the tight confidence intervals reflect a cohort with a great many person-years in it rather than a small study getting lucky.

The two winning patterns turn out to describe similar plates. The paper reports that the hyperinsulinemia index correlated most strongly with red and processed meats, sodium, and French fries, while the planetary health index correlated most strongly with nuts, unsaturated fats, whole grain carbohydrates and vegetable protein. Scoring well on either means much the same food: nuts, legumes, fruits, vegetables and whole grains in, red and processed meat, salt and potatoes down.

The most interesting design choice sits in the exposure list. Low-carbohydrate eating was not scored once but three times, as an overall score and then split into healthy and unhealthy versions according to whether the protein and fat displacing the carbohydrate came from plants or from animals. A study that thought carbohydrate quantity was the operative variable would have had no reason to build that split.

What the winning scores have in common points the same way. Neither the planetary health index nor the hyperinsulinemia index is a rule about a macronutrient; both are rules about which foods. A diet can be low in carbohydrate and score badly on the insulin index by leaning on processed meat, or high in carbohydrate and score well on whole grains and legumes. The abstract read for this piece names the leaders rather than publishing the full twelve-way league table, so where the low-carbohydrate scores actually landed against the Mediterranean or DASH patterns is not something this account can report.

This is a cohort study, not a trial. Nobody was assigned a diet; the women who scored highest on the planetary health index chose to eat that way, and people who make that choice differ from people who do not in ways no adjustment list fully absorbs. The paper's authors control for income, smoking, alcohol, physical activity and baseline weight precisely because those differences are known to exist, and residual confounding survives that kind of adjustment routinely.

The measurements have give in them, too. Body weight is self-reported, and food frequency questionnaires are validated instruments rather than exact ones. The cohort is overwhelmingly white, American, female and professionally health-literate, which is what makes forty years of follow-up possible and also what limits how far the numbers travel. And the outcomes here are weight and obesity alone: nothing in this study speaks to hot flashes, sleep, bone density or cardiovascular risk.

The value is not in the effect size, which is modest, but in the comparison. Most nutrition evidence around menopause tests one pattern against usual eating, which leaves a woman comparing studies that were never run against each other. Scoring twelve patterns in one cohort, on one set of women, with one set of outcomes, is the design that answers a ranking question instead of adding another entry to the pile.

The finding also cuts against the framing that dominates this territory. Nothing in the result points toward eating less; the patterns that won are defined by composition, and the paper's own conclusion frames them as optimised weight management rather than restriction. For a woman being told that the answer to midlife weight is a smaller plate, a large cohort finding that the operative variable was what the plate held is a different piece of information, and a more useful one.

The honest headline here is a ranking, not a discovery. Two patterns came out in front, they describe nearly the same plate, and the plate is the one nutritional epidemiology has been pointing at for two decades: nuts, legumes, vegetables, whole grains, and not much processed meat or salt. What makes this study worth attention is the head-to-head design, which is rare and expensive and settles an argument that separate trials cannot. What it does not deliver is a large effect. A third of a kilogram a year is real at the scale of a population and modest at the scale of one woman's bathroom scale, and the obesity hazard ratios, striking as they look, belong to women who chose these diets rather than were assigned them. The useful takeaway is narrower than the coverage: among the named diets competing for attention in midlife, the ones that tracked best were defined by food quality rather than by carbohydrate arithmetic or portion size, and that is a finding about composition, not about eating less.