Vegetarian girls came out lighter in a survey of 8,799 Austrian pupils. But across all the girls there, underweight was more common than overweight

Here's the tidy version of a paper published in Nutrients on 30 October: teenagers who exercise often, and teenagers who skip meat, are less likely to be overweight. Both claims come from a large nationwide survey of 8,799 Austrian secondary school pupils. Both hold up less neatly once you read the tables, and the first table contains a number the authors don't go on to analyse.

The survey, called From Science 2 School, invited all 2,688 secondary schools in Austria. About 1.1% of the country's pupils took part, aged 9 to 22, and 63.1% were girls. They typed in their own height and weight, which were converted to percentiles against German reference curves for age and sex. Diet was a label, confirmed by control questions on specific foods: 92.8% omnivorous, 5.6% vegetarian, 1.6% vegan. And the data aren't new. Some 98% of answers came in before 10 March 2020, so this is a picture of pupils more than six years ago.

Exercise gives the steadier result. Against pupils with no regular leisure exercise, those active one to three days a week had 26% lower odds of overweight or obesity, and those active four to seven days had 48% lower odds. The link held in boys (odds ratio 0.40) and girls (0.53). The strongest figure was in 10 to 12 year olds, at 0.25, though by our arithmetic only about 90 children that age reported no exercise, so the comparison group is small.

Then comes the odd part. Average BMI percentile barely differed between pupils who exercised and those who didn't: 49.5 against 50.6, p = 0.205. That test compared any exercise with none, while the odds split pupils by days a week, so the two results don't directly clash. The authors suggest exercise may guard against the top end of the weight range rather than shift everyone down, and they note that BMI can't tell muscle from fat. They also concede the arrow could point the other way: a heavier child may simply be less likely to take up sport.

Now the vegetarians. Their average percentile was 40.5 against 50.3 for omnivores, with 41% lower odds of overweight or obesity. But look at who they were. By our count from the paper's table, 87% of vegetarians were girls, against 61% of omnivores, and girls overall sat well below boys, at 45.3 against 57.3. So part of that 10-point gap may just reflect who goes vegetarian. Among girls alone it narrows to 7.5 points. Among boys it wasn't significant, and the odds ratio pointed the other way, at 1.21, though with only 64 vegetarian boys that's very uncertain. No single age group reached significance on those odds either. Vegans looked similar at 0.57, but the range ran from 0.30 to 1.09. The authors put that down to the small group, which is plausible but unproven, and their conclusion still speaks of plant-based diet types.

Here's the number from that first table. In this survey, 15.4% of girls were underweight and 9.7% were overweight or obese, or 855 girls against 534. Boys ran the other way, at 240 against 503. The paper calls its results more favourable body weight outcomes and never reports underweight by diet or exercise. So we can't say vegetarian girls were more often underweight, and they may not have been. We also can't assume a lower number is all good news. Teens often under-report weight and over-report height, as the authors note, which could inflate the underweight count and shrink the overweight one.

An older survey of 2,516 young people in Minnesota is a useful companion. There, young adult vegetarians were less likely to be overweight or obese, but current vegetarians of both ages more often reported binge eating with loss of control, and adolescents who had given up vegetarianism more often reported extreme weight-control behaviours. It's only an association, and the Austrian survey didn't ask about any of this, but it's a reason not to read lighter as simply healthier in this age group.

How much do diet and exercise explain here? Very little, on the paper's own figures. The models accounted for only 1.3% to 4.5% of the variation in weight status by a rough statistical measure, and diet type for 0.6% of the variation in BMI percentile. The analysis describes no adjustment for family income or parental education, and it doesn't capture what vegetarian pupils actually ate. It's a snapshot, so it can't show cause. What's left is still worth having: being active most days goes with lower odds of overweight in boys and girls alike, as you'd expect from other work. A well-planned vegetarian diet can suit teenagers, according to the Academy of Nutrition and Dietetics. But this survey doesn't make going vegetarian a way to manage a teenager's weight, and a sudden diet change alongside weight loss is worth a conversation with a doctor.