From a fellow dietitian:

"I do believe there are significant risks for mortality with increased BMI, irrespective of diet. I have seen studies supporting that, so I don’t think it’s a slam dunk for embracing any level of obesity. See the latest BMJ systematic review for example:

http://www.bmj.com/content/bmj/353/bmj.i2156.full.pdf"

My response:

My fingers get twitchy when I see this kind of research used to justify recommending weight loss, but I see it all the time and I think it's important to unpack the details.

The meta-analysis you've posted is a good one and can tell us many things. Mainly it shows that a group of larger people is, on average, likely to have a shorter lifespan than a group of smaller people. Interestingly, this one shows that smoking appears to be worse for you if you have a high BMI. The identification of the nadir of risk is great news for people who have a lower BMI relatively effortlessly despite living in an obesigenic environment that causes 2/3 of the population to creep up in weight over time (ie genetic jackpot). The recent influx of big BMI and mortality meta-analyses interestingly comes in response to Flegal's paper from 2013 which found that being overweight conferred the lowest risk of death in an NHANES dataset, thus suggesting that larger bodies might not be as bad as originally thought. Anyway, I digress. There are far more interesting things for dietitians in this study.

Firstly, where are our intentional weight losers (ie including our clients)? Well, we don't know. Unless they are running separate analysis for those who intentionally lost weight into a lower BMI category we cannot tell whether they have the same risks as so-called 'never-fat' people and so we cannot say that weight loss delivers on it's promise. This study found that using highest adult BMI when analysing mortality resulted in even lower risks for the lower BMI categories. So this might mean that all weight loss is bad, or it might mean that there are so few healthy people losing weight, keeping it off and not dying early that the disease-related weight loser stats drown them out. Think about that for a bit: there are more people actively dying from disease to the point that they are losing weight than there are people who have successfully, permanently lost weight. Probably a bit of both is true. Bottom line: population studies showing how bad obesity is do not justify recommending weight loss; they just show that being overweight/obese confers risk. They say nothing about the merits of no longer being obese or about the risks of becoming obese. Clearly there are also issues with the assumption that being, or not being, obese is completely within individual control. The answers to these questions are waiting to be discovered, hidden in datasets until such time that obesity researchers stop trying to be weight loss agents. https://www.cambridge.org/core/journals/nutrition-research-reviews/article/a-review-and-meta-analysis-of-the-effect-of-weight-loss-on-all-cause-mortality-risk/26226C6DF1BA32BEB00AAC87FC416667" target="_blank">Here is another paper looking at the mortality risks related to intentional weight loss (spoiler: overweight people who intentionally lost weight had an 11% increase in mortality). 

Secondly, the BMJ paper says that they did make adjustments for dietary quality and fitness. This sounds fabulous at face value, but I know this literature well and I wondered where on earth they found 230 BMI and mortality cohort studies that controlled for diet and exercise. So I nerded out in the supplementary material. Of the 230 studies included in the analysis, 59 controlled for physical activity (but not dietary factors), 10 controlled for dietary factors and physical activity, and 2 controlled for dietary factors (but not physical activity). People in these studies represented 7.7%, 3.8% and 0.062% of the total people studied respectively. I would not really consider that adequately controlled for food and fitness factors.

Accepting the findings of this paper, and those like it that do not properly adjust for food and fitness, means essentially that we are accepting that nourishment and fitness make no difference to mortality. Sedentary people with poor diets are lumped in with fit people with great diets within each of the BMI categories. Most of the studies included did not adjust for SES either. Does that seem fair or accurate? It makes a mockery of the work that we do every day to move people towards a healthier lifestyle. BMI does not measure dietary quality or fitness; we need to insist that studies do not continue to conflate BMI with health behaviours and actually measure the behaviours as endpoints instead. Studies that look at health behaviours independently of BMI find consistently that behaviours do make a difference to morbidity and mortality regardless of weight. Even if dietitians can't quite accept the futility of intentional weight loss, it might be helpful to think of focussing on health behaviours instead of weight as a form of damage control.

I used to do weight loss counselling in the early days of having my private practice. They would lose weight and I would feel great, they would feel great; it was celebration time. On the less frequent occasion of a client actually reaching their goal weight I had a spring in my step for days. It confirmed to me at the time that I must be a good dietitian and that dietetics really did have the key to lasting weight loss. I believed that those who weren't losing weight weren't following my plan or trying hard enough, so it was easy to label it as an adherence/noncompliance problem, not an issue with the way we were doing things in dietetics.

Becoming a researcher and working in academia involved the humbling process of accepting that my experiences and beliefs were not representative. Only seeing the 'honeymoon' phase of weight loss and devaluing the importance of my clients' previous (failed) weight loss attempts had lead me to have a distorted view of the effectiveness of dietetic counselling for weight loss. Unless a dietitian is following all of their clients for 5+ years it is likely that they suffer this same confirmation bias. In the absence of good quality research showing the 5 year outcomes of dietitian-delivered, individualised weight loss counselling, we have to accept that we are probably the same or worse than the well-funded, multidisciplinary, beautifully designed weight loss studies described in the literature which were used by the NHMRC in their review to conclude that long-term weight loss for the majority of people is unattainable.