20

@GoogleFacts wrote in a tweet:

The chance of an obese person attaining normal body weight is 1 in 210 for men and 1 in 124 for women, according to a new study.

Are those numbers accurate?

Martin Schröder
  • 548
  • 1
  • 6
  • 16
Christian
  • 33,271
  • 15
  • 112
  • 266

1 Answers1

46

No, this is not quite true.

The factoid is based on a real, peer-reviewed scientific study, that is relatively new.

Probability of an Obese Person Attaining Normal Body Weight: Cohort Study Using Electronic Health Records, American Journal of Public Health, September 2015, Vol 105, No. 9, doi: 10.2105/AJPH.2015.302773

They based their evidence on looking at nine years of medical records of adults in the UK, from 2004-2014.

In simple obesity (body mass index = 30.0–34.9 kg/m2), the annual probability of attaining normal weight was 1 in 210 for men and 1 in 124 for women, increasing to 1 in 1290 for men and 1 in 677 for women with morbid obesity (body mass index = 40.0–44.9 kg/m2 ).

There are some caveats that should be considered when thinking about these figures.

  • These are the annual figures, not lifetime figures. That makes the quote very misleading.
  • The numbers cited by Google Facts are only for obese people, not morbidly obese.
  • This is all people who were recorded by their research database (7% of the UK population), not merely people who were attempting to lose weight. If you are trying to lose weight, you might expect a larger chance of success.
  • They excluded patients who had bariatric surgery, and those who didn't have a weight recorded at least 3 times by their doctors.
Oddthinking
  • 140,378
  • 46
  • 548
  • 638
  • 9
    If I'm reading the study correctly, it also appears that obese people were actually *better* at losing weight than the merely overweight (which makes sense; if you're obese then you're perhaps more inclined to care about your weight and try to do something about it). It should also be noted that BMI is an imperfect metric. – Kevin Feb 15 '16 at 05:51
  • 1
    ... how on earth is the probability *increasing* when it goes from 1/210 to 1/1290? – Bakuriu Feb 15 '16 at 10:05
  • 2
    @Bakuriu: I believe Kevin is referring to the chance of losing 5% weight (rather than the chance of reaching a 'normal' BMI) as described in the study. – Oddthinking Feb 15 '16 at 10:18
  • @Kevin: I did consider adding a note saying "BMI isn't always useful measure for individuals; it was designed as a proxy for population studies." (citation-needed), but given this was a population study, it seemed unnecessary. – Oddthinking Feb 15 '16 at 10:20
  • 2
    I suppose I could dig into the study to answer this, but does "annual probability" here mean the probability of someone dropping from an "obese" BMI to a "normal" BMI all within this year, or does it mean the (larger) probability that someone who has ever been "obese" in the past, drops to "normal" during this year? As an analogy, we could say that the daily probability of someone walking from SF to NY is zero (because it takes longer than that) or we could say it's very slightly more than zero (because each day on average, very slightly more than zero people complete such a walk). – Steve Jessop Feb 15 '16 at 10:40
  • @SteveJessop: The latter. They looked at readings that may well be more than a year apart, and annualised the rate. (I guess in theory, someone may have made the transition several times between doctor visits) – Oddthinking Feb 15 '16 at 14:17
  • 4
    *"This is all people [...], not merely people who were attempting to lose weight"* - this can have a huge impact. The tweet is formulated in a way to suggest that it is very difficult to lose weight no matter how much you try, but the sample possibly includes a lot of people who don't even want to lose weight or don't even try. Many slightly overweight people are either happy with their weight or don't make significant efforts in losing weight. That the probability for women is larger makes it self-evident: women have a larger social pressure to be underweight to be considered attractive. – vsz Feb 15 '16 at 15:08