• So now will people say losing weight has no value?

    Despair in the NEJM. “Cardiovascular Effects of Intensive Lifestyle Intervention in Type 2 Diabetes“:


    Weight loss is recommended for overweight or obese patients with type 2 diabetes on the basis of short-term studies, but long-term effects on cardiovascular disease remain unknown. We examined whether an intensive lifestyle intervention for weight loss would decrease cardiovascular morbidity and mortality among such patients.


    In 16 study centers in the United States, we randomly assigned 5145 overweight or obese patients with type 2 diabetes to participate in an intensive lifestyle intervention that promoted weight loss through decreased caloric intake and increased physical activity (intervention group) or to receive diabetes support and education (control group). The primary outcome was a composite of death from cardiovascular causes, nonfatal myocardial infarction, nonfatal stroke, or hospitalization for angina during a maximum follow-up of 13.5 years.

    The rationale for this study was simple. The researchers wanted to see if patients with type 2 diabetes who lost weight might have better cardiovascular outcomes. They got more than 5000 such people and randomized them to a normal education group, or a diet and physical activity program. The main outcome of interest was reduced death from cardiovascular causes for up to 13.5 years.

    The good news is that people in the intervention group lost more weight than those in the control group (8.6% vs 0.7% at one year, and 6% vs 3.5% at study end). Furthermore, those in the intervention group saw greater improvements in glycated hemoglobin, fitness, and risk factors. The bad news is that after almost a decade, there was just no difference in the real outcome of interest. The number of deaths from cardiovascular causes was just no different.

    They actually stopped the trial early, once a “futility analysis” showed it just wasn’t going to work, even if they waited a few more years.

    Now some will point to the “failure” of this study as proof that weight reduction doesn’t save lives. That’s true, if the only outcome you care about is death. But here in the real world, we care about many more things, which are also important. People who lose weight can see improvement in lots of domains, including depression, physical functioning, sleep apnea, and plain old quality of life.

    By focusing on the big outcomes while ignoring many of the smaller interim gains, we can lose sight of what matters. I’m not saying the study authors did this. I fear that some interpreting the results might, however.


    • Sounds familiar.

      Primary end point underpowered or needs more time to declare.

      Secondary end points with meaningful benefits.

      Wellness, personal responsibility, self-directed care a failure. Abandon approach, right?


      • Many of the other outcomes were primary endoints in prior published studies. It’s just that now (10 years later) we can look at death. Those other studies were not “secondary” endpoints. They’ve already been analyzed and released! Years ago!

        • I am getting cute for the Oregonians. 🙂

          For this study, primary composite was CV outcomes though. They did measure other things (risk factors).

        • @Aaron – I’d think the difference in secondary outcomes over 10 years is just as interesting compared to earlier years. Weight loss, for example looks much better at year 1 than at year 10, so I’m curious how those other outcomes fared over time. For example, since some weight returned by the end of the trial, did depression and sleep apnea rates also rebound? And to what degree? Were people more depressed than when they started due to the regain? Plausible if unlikely, but the study doesn’t say, so I find that ascribing a positive spin is unwarranted (on this study alone — obviously other studies bolster the secondary results at different timeframes).

          I don’t blame the study for this, but I think it’s unfair to compare positive benefits from year one to unknown outcomes at year 10, particularly when ignoring negatives like amputation and fractures numbers (I have separate issues with the “adjudicated fractures), or possible secondary causes… for example the study gave “meal replacement products […] at no cost”… free food alone would make some people happier, and certainly makes a diet easier to maintain.

          I’m not saying losing weight has no value, as the title predicts, but I think it’s an equal fallacy to read into a study things that are not there, Take this article in the atlantic: http://www.theatlantic.com/health/archive/2013/01/the-problem-with-all-of-this-overweight-people-live-longer-news/266756/
          doing a similar but opposite thing with the recent “overweight people live longer” study. I find that just as alarming.

          I’m particularly fond of “…proof that weight reduction doesn’t save lives. That’s true, if the only outcome you care about is death.” When saving lives, death is a pretty important variable. 🙂 (just making fun of you there… it’s a funny sounding sentence; your point is well taken).

          Good read, though, thanks for the article.

          • I think that an intervention – applied once – has diminishing returns over time. Absolutely. I have to “reapply” my weight watchers mentality every few years in order to keep weight off. This is a lifestyle change, not an “intervention” as it’s always treated…

    • Assuming “the smaller interim gains” to be true, which did it? …loss of weight … exercise … human intervention … something else.

    • To dismiss this data, the best exercise intervention that will probably ever be done, isn’t fair, either. Especially because your post repeatedly impugns the study that the outcome is to “save lives” and matters if the “only outcome you care about is death.” This isn’t the outcome of the study!

      821 met the primary outcome, there were only 109 cardiovascular deaths. 390 of the outcomes were angina. This is actually a fairly soft outcome.

      Evidence that exercise is great for depression for primary prevention or prevents sleep apnea isn’t _that_ strong

      There are people who don’t enjoy going to the gym. It’s time-consuming and expensive, so it’s not a small thing we ask of everyone. If I have to start telling my patients they should spend 150 minutes per week exercising as an OSA prevention strategy, I’m not sure that’s worthwhile. I’m not saying exercise is useless, but in the big scheme of things we push for and fund, noting that the biggest incentives we have are now truly under question is a big deal.

      • Wait, who’s dismissing the data? I totally believe their results. I just don’t think that preventing death should be the reason we encourage people to lost weight.

        Put this in the same camp as preventive medicine for me. Lots of people think it saves money. I’m not so sure. I think improves outcomes and quality of life, though, and I totally think it’s still worthwhile.

        • You keep writing “death.” The study also looked at heart attack, stroke, and hospitalization for angina and found no effect. That’s a lot more than death.

          • That’s because it’s what I imagine others will focus on. But you’re right, non-fatal stroke and MIs were also included.

            Maybe I’m not making myself clear. I don’t dispute the results. I totally believe them. I just don’t think that should change the reason that obesity and weight loss are important.

    • Their outcome was very limited in its scope, cardiovascular morbidity. I agree with Dr. Carroll not focusing on other outcomes means this study may(will) be misconstrued to the public. You would think a measure of Quality of LIfe in this study would have been included or a way to measure other types of medical related deaths.

    • A couple of things:

      1) I think it’s prudent to keep in mind that the patients being studied were diabetics – a patient population notorious for its resistance to medical therapy for primary cardiovascular prevention. Look at the data for aspirin in diabetics and non-diabetics, for example. Whether the lessons from this study should provide guidance for our counseling of non-diabetic patients remains to be seen.

      2) I think Aaron maybe overstates how underwhelming the results are. Looking at table 2, there are some trends toward benefit in the weight-loss group for death from any cause, fatal or nonfatal MI, and heart failure. The futility vis-a-vis the composite endpoint seems to be driven mainly from stroke and unstable angina.

    • Perhaps the intervention regime was too extreme? It mentions that they aimed to restrict caloric intake to 1200 to 1800 calories. I could see 1,800, but the lower bound 1200 seems too low, even for a carefully prescribed diet, to be healthy.

    • The mean age of these patients at the start of the study was nearly 60, which means a lot of these patients were 70+ after 13.5 yrs. It does not say for how long these patients had been diagnosed with diabetes before the study began. I wonder if there is anyway to tease out of the data the impact of earlier intervention. It could be that the study did not show a significant difference in outcomes, because the damage had already be done.

      also I wonder what the results look like if you try and stratify the data into different levels of weight loss. Do you start to see significant differences in outcomes at a given weight loss threshold?

      • What do you mean by “significant?” If I am reading this appendix correctly: http://www.nejm.org/doi/suppl/10.1056/NEJMoa1212914/suppl_file/nejmoa1212914_appendix.pdf
        The study was only designed to be able to detect at least an 18% reduction in CV events. But reading further, it wasn’t even able to do that because the rate of CV events was much lower in the control group than they had expected. The appendix goes on to explain that even without the unexpectedly good control group, 18% was actually on the upper end of what qualifies for clinically significant based on other literature, which ranges from 15% to 20% (why a 14% reduction is excluded from their definition of “public health benefits” is not something I understand).

        Anyway, I’m not as optimistic about the believably of this result as Aaron Carroll seems to me. Perhaps the study does tell us that the effect of weight loss is smaller than previously estimated from observational studies (which produced estimates in the 25% to 33% reduction range), but unless I’ve misread something, it doesn’t seem to suggest that there aren’t huge benefits to weight loss–it simply was never designed to be able to detect these benefits.

    • The average weight loss vs. the control group over the length of the study was 4%. Starting from an average BMI of 36, the intervention group averaged about 1.5 points lower than the control. The authors say the study had sufficient power based on the benefit of 18% that was targeted in the study’s design. Isn’t that a really large expected benefit? If you could expect an 18% reduction in risk going from 36 to 34.5, what would you get going from 36 to 25?

      Also, shouldn’t there be a bigger distinction drawn between proving there is no benefit, and not proving that there is a benefit? Saying that losing weight doesn’t reduce the risk of death based on this study is like flipping a coin 10 times, getting 7 heads, and concluding the coin must be fair.

    • why only look at CV mortality?
      why not look at all mortality?
      obesity does complex things to the body. We have more understanding of the CV mechanism, but I doubt it is the only one.

      agreed with early commentators that leaving out a QOL metric is a major flaw (though, that would have raised study costs if it took the form of survey questions)

    • http://www.nejm.org/doi/full/10.1056/NEJMe1306987

      as the accompanying editorial points out, patients in the control arm were more likely to need insulin as well as statins and additional antihypertensive medications; this makes sense, but use of these medications (which have proven benefits) may have biased the results. losing weight clearly has value from a cost perspective, if nothing else. quality of life (not just because the patients in the weight-loss group are thinner – not having to use insulin is a plus in my book too) can’t be overlooked either