Brace yourselves for a study which produced results which are, to say the least, counterintuitive. A group of Finnish researchers found in an epidemiological study on twins that those overweight people who intended to lose weight who lost weight were most at risk of dying over the study period. (the higher the hazard ratio the greater the risk of dying)
Weight loss in the obese improves risk factors for cardiovascular diseases and diabetes. However, several studies have shown inconsistent long-term effects of weight loss on mortality. We investigated the influence on mortality of intention to lose weight and subsequent weight changes among overweight individuals without known co-morbidities.
Methods and Findings
In 1975, a cohort of individuals reported height, weight, and current attempts (defined as “intention”) to lose weight, and in 1981, they reported current weight. Mortality of the 2,957 participants with body mass index ≥ 25 kg/m2 in 1975 and without pre-existing or current diseases was followed from 1982 through 1999, and 268 participants died. The association of intention to lose weight in 1975 and actual weight change until 1981 with mortality was analysed while controlling for behavioural and psychosocial risk factors and hypertension as possible confounders. Compared with the group not intending to lose and able to maintain stable weight, the hazard ratios (with 95% confidence intervals) in the group intending to lose weight were 0.84 (0.49–1.48) for those with stable weight, 1.86 (1.22–2.87) for those losing weight, and 0.93 (0.55–1.56) for those gaining weight. In the group not intending to lose weight, hazard ratios were 1.17 (0.82–1.66) for those who did lose weight, and 1.57 (1.08–2.30) for those gaining weight.
Deliberate weight loss in overweight individuals without known co-morbidities may be hazardous in the long term. The health effects of weight loss are complex, possibly composed of oppositely acting processes, and need more research.
That is the abstract of the study. Since it was published in PLoS Medcine you can click through and read the whole study. The body of the study has an extensive discussion of previous relevant research, more details of this study, and analyses of what it all means.
But don't rush to conclusions. Epidemiologist Meir Stampfer of the Harvard School for Public Health says epidemiology studies of the effect of weight on mortality need to correct for a number of problems.
Epidemiologic studies of the relation between overweight and mortality typically must address three principal concerns . First, in many populations, cigarette smokers tend to be leaner than nonsmokers. Because cigarette smoking is such a strong risk factor for mortality, failure to adjust for this adequately can lead to confounding, with the erroneous conclusion that leanness carries increased risk of death. Statistical adjustment for smoking is often insufficient to account for this difficulty. Smoking can lead to medical conditions, sometimes sub-clinical, that are associated with decreased body weight, such as chronic obstructive pulmonary disease. The presence of symptoms or diagnosed conditions may induce smokers to quit. Moreover, the intensity of smoking is related to both risk of death and body mass index. For these reasons, the best way to assess the impact of overweight on risk of mortality is simply to exclude current and past smokers. Kaprio and colleagues' study differentiated only current smoker or nonsmoker. Thus, never-smokers were included in the same category as past smokers, regardless of how much the past smokers had smoked or their reasons for quitting.A second difficulty in some epidemiologic studies is the inclusion of intermediary factors as co-variates. Weight loss improves hypertension and diabetes, so including these as co-variates would tend to attenuate the apparent benefit of weight loss. In the present study, the authors appropriately excluded people with diabetes, and adjustment for hypertension appeared to have little impact. The third and most difficult issue in studies of overweight and mortality is reverse causation, the impact of disease on body weight. This can occur either through the biological impact of a condition (diagnosed or preclinical) or as an inducement to attempt to lose weight as a means to improve health. The authors' keen recognition of this problem provides a significant strength to the present study. The authors appropriately excluded individuals with a wide range of conditions to identify an apparently healthy cohort. This critical step is often ignored. Sometimes, investigators also exclude deaths that occur in the first few years after follow-up, to reduce the impact of reverse causation. Such lagged analyses can be helpful, but some chronic conditions of long duration, such as depression, chronic lung disease, and heart failure (conditions that often may not reach the level of clinical diagnosis) can lead to lower body weight and higher mortality risk. Hence, that strategy (not employed by Kaprio and colleagues) may not fully avoid the problem.
Stampfer points out that the people who lost more weight may not have done so as a result of a desire to lose weight. They may have intended to lose weight. But when they finally did lose weight it might have been due to diseases that eventually killed them. Plus the study size was not big enough and the degree of obesity was fairly small.
As a further step toward reducing the potential bias of the impact of disease on body weight, the authors identified individuals with intent to lose weight. Unintended weight loss is well known as an ominous clinical sign, usually signifying a serious illness. Just over a third of the overweight subjects in the Kaprio cohort had expressed intent to lose weight at baseline in 1975. It is interesting to note that despite this intention, as a group, the median weight change in the ensuing six years was a gain of 0.33 kg/m2, almost identical to the weight change in the group that had not expressed intent to lose weight (gain of 0.31 kg/m2). Thus, this study cannot fairly be characterized as an assessment of intentional weight loss and its subsequent effect on mortality. Because the changes in weight are so similar in these two groups, it is implausible to attribute the weight loss in the intent-to-lose group to that intention. These findings render the results particularly difficult to interpret.Other important limitations include the very small number of endpoints—only 268 total deaths in the cohort. When further subdivided according to intention to lose weight, and categories of weight change, the numbers are far from sufficient for reliable estimates. For example, the main conclusions are based on the subgroup of those with intent to lose weight who actually lost weight; this group had only 42 deaths, of which ten were violent. Another related difficulty is that this cohort, though overweight, is not drastically obese. The median body mass index (BMI) was 26.7 kg/m2, and fewer than 10% of people had a BMI greater than 30. With such a narrow range of BMI, coupled with the very small changes in weight during the six years of initial follow-up, even a very large study would not have sufficient power to detect plausible relative risks associated with weight change.
At the extreme it seems very unlikely that people who are morbidly obese (e.g. the people who are so large they can no longer even stand up) would suffer a net harm from losing weight. Also, people who have various risk factors that show up more among the obese such as hypertension and type II diabetes would probably benefit more from weight loss than those who do not have those risk factors who are equally overweight.
If you are overweight, then losing weight is good for your health, surely? Unfortunately, the evidence on which an answer to this seemingly simple question might be based is at best equivocal, and at worst very controversial. Previous work has shown that weight loss in obese people improves risk factors associated with cardiovascular diseases and diabetes, but studies are conflicting on the long-term effects of weight loss on mortality. A study in this month's PLoS Medicine by Jaakko Kaprio and colleagues on a Finnish dataset adds more evidence to this debate, but experts are divided on what can be concluded from it.
The major difficulty in getting clear results on this question is that it is virtually impossible to do a controlled trial to answer it. Hence, the evidence accumulated has come mostly from epidemiological studies, but it is notoriously difficult to remove all confounding factors from these studies. Kaprio and colleagues' study is another epidemiological study, but we should not simply dismiss the data as unreliable just because of the problems inherent to such a study design. Instead, we should consider their study in the light of all the other evidence available.
Suppose the Finnish study's results hold up in a larger study. How could this be? One possibility is that loss of non-fat cell mass while on a diet may cause harm.
The study leaves us with the question of how intentional weight loss could lead to excess mortality. The authors suggest that this could be due to the unavoidable loss of lean body mass, which according to several other studies may increase mortality, and which may outweigh the beneficial effects of losing fat mass in healthy individuals. The authors therefore conclude that “the long-term effects of weight loss are complex, and they may be composed of oppositely operating effects with net results reflecting the balance between these effects.”
The research paper (which is the first link above) discusses this idea at greater length.
Suppose loss of lean body mass during dieting inflicts real harm. Large amounts of high calorie burning aerobic exercise might be the solution. If one could burn off calories with exercise then perhaps the lean muscle mass loss could be avoided. Though possibly internal organ mass might still be lost while dieting.
Another possible solution might be slower diets. But in the longer term the ideal solution would be drugs for regulating appetite that avoid obesity in the first place. For those already obese we probably need drugs that signal only fat cells to burn fat while leaving other tissue types unchanged.
Once effective weight loss drugs make it to market then fairly well prospective studies on the effect of weight loss will become possible. Without weight loss drugs few people currently succeed in keeping off the weight that they lose. I've seen references (sorry, no URL) to studies which purport to show that yo yo dieting where the weight goes on and off repeatedly is more harmful than no dieting at all. This makes intuitive sense when you consider the discussion above about the loss of lean body mass during diets. Someone who repeatedly diets puts their body through many episodes of lean body mass loss. Also, someone who is binging to gain weight will put their body through higher calorie consumption periods and probably therefore periods of even higher lipid levels from consuming mass quantites. My guess is that while people from Remulac can consume mass quantities safely humans can not.
My advice: Concentrate less on weight loss and more on improving the quality of the food you eat. Also, change how you go about your daily activities to inject more exercise into what you do. Try to walk rather than drive where that is practical. Take stairs rather than elevators. Get a totally manual mower rather than a gas engine mower. If your dog wants you to take him or her for a run then do so. Dogs are great personal fitness trainers if you will only obey them.
|Share |||Randall Parker, 2005 June 30 05:22 PM Aging Diet Studies|