Oily Omega Myths Persist
Harvard study reinforces misleading, sickening myths about fats and health
Harvard study reinforces misleading, sickening myths about fats and health
"Galileo's Middle Finger” — a recent book by science historian Alice Dreger, Ph.D. — delves into several damaging distortions of science.
Her stories of disinformation are distressing, but left this reader heartened by the eventual triumph of verifiable truth.
Another damaging set of distortions could easily mislead — and sicken — millions of Americans and people worldwide.
Despite their institution's lofty reputation, the latest distortions of the true picture come from respected researchers at Harvard university.
Their recent paper about the role of fats in human health made headlines, and it furthers a persistent myth.
However, as we'll explain, leading researchers into various fats' impact on health strongly dispute the Harvard group's assertions.
- Harvard team's mathematical analysis of two large population (epidemiological) studies estimated that diets higher in saturated fats raise the risk of death and disease modestly.
- The Harvard team's analysis estimated that diets higher in unsaturated fats — including monounsaturated fats (olive oil) and polyunsaturated omega-3 fats* and omega-6 fats**, especially omega-6s — reduce the risk of death and disease the most.
- Scientists expert in fats and health stress that the totality of evidence – including from more-reliable clinical trials — largely exonerates saturated fats, but strongly links America's very high average intake of omega-6 polyunsaturated vegetable fats to higher risks for disease and death.
**From nuts, seeds, and cheap vegetable oils (corn, soy, safflower, sunflower, cottonseed)
About the Harvard study
The recent paper comes from a Harvard School of Public Health team led by professor Walter Willett, M.D.
For their new study, Dr. Willett's team analyzed data from two large epidemiological studies that collected diet and health data from 126,233 people — 83,349 female nurses and 42,884 male health practitioners — and followed them for 32 to 36 years.
(Epidemiological studies use statistical math to look for associations between people's self-reported diets and their long-term health outcomes.)
Unfortunately, media outlets tend to regard his groups' conclusions and pronouncements as scientific gospel, without adding the full scientific context.
All of the participants were healthy at the start of each study, and were initially surveyed about their diets at the outset in 1980 and again every two to four years after.
The participants' health records were also examined to determine how many had died from any cause — including heart, metabolic, and brain diseases — before the end of each study.
The Harvard team estimated the effect of "replacing" five percent of daily calories from saturated fats with the same number of calories from each of three other common kinds of fat.
Because no such actual "replacements" occurred — as they would have in controlled clinical trials that test the effects of different diets — the Harvard group based their conclusions on mathematical projections.
Based on those mathematical "substitution analyses", the Harvard team concluded that people who "replaced" five percent of their daily calories from saturated fats with calories from unsaturated fats — especially polyunsaturated fats from plant foods and oils or seafood — were less likely to die prematurely from any cause.
Using mathematical projections, the Harvard group made the following estimates.
Unhealthy fat-intake profiles:
- Diets higher in trans fats were called the most dangerous. Every 2% higher intake of trans fat was linked to a 16% higher risk of premature death. (They come primarily from partially hydrogenated vegetable oils.)
- Diets higher in saturated fats (mostly from animal foods) were linked to a greater risk of premature death. Every 5% increase in total calories from saturated fat was linked to an 8% higher risk of premature death.
- Diets higher in any unsaturated fats — including polyunsaturated and monounsaturated* fats — were linked to an 11% to 19% lower risk of premature death.
- Diets higher in polyunsaturated omega-6 and omega-3 fats were linked to the lowest risk of premature death.
- Diets higher in polyunsaturated omega-3 fats were linked to a modestly lower risk of death from any cause, especially from dementia and other brain diseases of aging.
*Americans get polyunsaturated omega-6 fats primarily from plant oils, nuts, and seeds, while they get polyunsaturated omega-3 fats primarily from canola oil, fish (the best source by far), walnuts, and flax seeds. They get monounsaturated fatty acids (MUFAs) primarily from olive oil, canola oil, "high-oleic” versions of sunflower or safflower oil, and avocados.
Interestingly, mathematically "replacing" 5% of daily calories from saturated fat with calories from carbohydrates (sugars and starches) was estimated to reduce the risk of death only slightly.
As the Harvard team put it, "Our analyses provide strong evidence that using PUFAs [polyunsaturated fatty acids] and/or MUFAs [monounsaturated fatty acids] as the replacement nutrients for SFAs [saturated fatty acids] can confer substantial health benefits,”
An article published by Harvard asserted that "this study is by far the most detailed and powerful examination of this topic.”
However, rather than relatively "hard" evidence from clinical trials, their conclusions rest on disputable mathematical analyses of comparatively "soft" epidemiological evidence.
And the authors' discussion of their conclusions ignored the totality of the evidence from laboratory, epidemiological, and clinical studies ... which, together, paint a very different picture.
Clinical and lab evidence contradicts Harvard group's assertions
The "soft” evidence derived from epidemiological studies can only be confirmed or refuted by clinical research.
Consequently, the Harvard team's assertions sound like scientifically unjustifiable certitude and overreach.
In fact, the results of most clinical trials show that — versus replacing saturated fats with polyunsaturated (omega-6-rich) vegetable oils — people get a bigger health boost from cutting their intake of omega-6 fats and upping their intake of omega-3s ... especially "long-chain" omega-3s from seafood.
And most of the many thousands of published lab experiments support the strong likelihood that diets high in omega-6 fats — such as the average American's diet — hurt most people far more than diets high in saturated fats.
For example, late Harvard professor Alexander Leaf, M.D., pointed to the sadly overlooked Lyons Heart and Diet Study ... a clinical trial published in 1999.
As he wrote, "The authors make the telling point that clinical dietary trials that lowered saturated fatty acids and raised [omega-6] PUFA intake in an effort to lower cholesterol failed to improve the overall clinical prognosis of their experimental groups; only the trials that also lowered intake of omega-6 PUFAs and increased omega-3 fatty acids successfully lowered cardiovascular and all-cause mortality in the experimental cohort [group].” (Leaf A 1999)
(Dr. Leaf served as chairman of medicine at Massachusetts General Hospital, led preventive medicine at Harvard Medical School, and was among the first practicing physicians elected to the National Academy of Sciences.)
Researchers at the U.S. National Institutes of Health uncovered similar risks in long-lost data from a three-year-long Australian clinical trial, called the Sydney Heart Study.
In that clinical trial, some of the 458 male volunteers, all of whom had recently experienced serious heart trouble ate diets that replaced saturated fats with omega-6-rich vegetable oils (Ramsden CE et al. 2013).
The NIH scientists' analysis of lost data from the Australian clinical trial showed that men who replaced saturated fats with omega-6-rich vegetable oils were more likely die from any cause, including cardiovascular diseases (see Heart Risks Raised by Omega-6 Excess).
Likewise, the results of other controlled clinical trials — and thousands of lab experiments — contradict the assertions made by Dr. Willett's Harvard team.
For example, see Vegetable Oils Debunked for Heart Disease, Are Vegetable Oils Heart Healthy?, Heart Group's Omega-6 Advice Takes a Huge Hit, Heart Association Appears Blind to Risks of America's "Omega-Imbalance, and Report Finds Americans Need More Omega-3s ... and Far Fewer Omega-6s.
We also encourage you to view our "Out of Balance” video, which features leading fats-and-health researchers.
In that video, they explain why the average American's unprecedentedly high intake of omega-6 fatty acids from vegetable oils — and low intake of omega-3s from plant foods and seafood — is essentially proven to hurt their heart and overall health.
For more information on that topic, see our "Omega-3/6 Balance: Hidden Health Risk” page and the "Omega-3/Omega-6 Balance” section of our news archive.
Scientists expert in fats and health dispute the Harvard analysis
In this writer's opinion, the Harvard team's discussion of the published evidence was selective and misleading.
There's been a growing struggle for the public mind, between the basically pro-omega-6/anti-saturated-fat position of Dr. Willett's Harvard group and scientists who stress the large preponderance of lab, epidemiological, and clinical evidence showing that Americans' historically — and extremely high — intake of omega-6 fats is bad for heart, brain, immune, and overall health.
Highly regarded scientists who dispute the Harvard group's view include ones whose expertise we seek out: Professor William E. "Bill” Lands, Ph.D., CAPT Joseph F. Hibbeln, M.D., and Douglas Bibus Ph.D.
(You'll find their credentials* at the end of this article. They have neither reviewed and nor approved our characterizations of the Harvard study and its scientific context.)
In response to our request for his comments, Dr. Lands said that, although the studies analyzed by the Harvard team were of relatively high quality, "[the Harvard study's authors act] as if the evidence had been obtained from an actual replacement intervention [controlled clinical trial].”
He stressed the point that "substitution analyses" (mathematical projections) based on data from epidemiological studies are not reliable substitutes for controlled clinical trials.
Regrettably, the Harvard study's authors repeatedly imply that the people surveyed for the two epidemiological studies had actually replaced 5% of the saturated fats in their diets with the same proportion of unsaturated fats … but that wasn't the case.
In addition to contradicting the conclusions of the Harvard team, the results of clinical trials contradict two aspects of conventional wisdom:
- High blood cholesterol levels — and/or diets high in saturated fats — cause cardiovascular disease (CVD).
- Replacing saturated fats with polyunsaturated omega-6 vegetable fats is the best way to reduce the risk of heart disease and death.
For more about the declining credibility of the "saturated fat theory of heart disease”, see Major Study Exonerates Saturated Fat — which concerns a landmark 2014 evidence review — and Heart-Diet Myths Get a Busting.
Dr. Willett criticized that 2014 evidence review in "Dietary fat and heart disease study is seriously misleading”, but a fast-growing scientific consensus based on the totality of the evidence now considers the saturated fat theory of heart disease greatly exaggerated and oversimplified.
Public comments from Harvard repeat the distortions
Study co-author Walter Willet, M.D. was interviewed for a Harvard podcast titled "Clearing Up the Confusion Over Fat”.
Unfortunately, Dr. Willett's statements in that interview mis-characterize the totality of the evidence, and will mislead people.
These are two key excerpts from that podcast; we've noted the point at which each statement occurs, and provided our responses:
Dr. Willett at 12:00 minutes: "There is a myth going around that omega-6 polyunsaturated fats are toxic and pro-inflammatory and bad for us. That's based on zero, zero data really. And this study showed that omega-6 polyunsaturated fats were the best of all. In fact, they were what was driving the majority of the benefit of overall polyunsaturated fat.”
Vital Choice response: Dr. Willett is attacking a straw man. No experienced fatty acid researcher has or would claim that omega-6 fats are inherently toxic or pro-inflammatory.
What experts like Drs. Lands, Hibbeln, and Bibus say — and what the evidence overwhelmingly shows — is that the extremely excessive intake of omega-6 fats in the standard American diet is pro-inflammatory and promotes major chronic diseases.
As Dr. Lands says, dietary omega-6 fats have a "narrow therapeutic window" of benefit. In other words, omega-6 intakes that rise substantially above the (much lower) average levels of our recent and ancient ancestors is unhealthful.
So it behooves prominent scientists to stress that point, but Dr. Willett's group has consistently failed to do that.
For more on the evidence of harm from America's omega-6 overload, see our page titled "Omega-3/6 Balance: Hidden Health Risk” and the "Omega-3/Omega-6 Balance” section of our news archive.
Dr. Willet at 14:25 minutes: ".... Definitely one is going to be better off in terms of the risk of dying prematurely if we use one of those [omega-6-rich] liquid [vegetable] vegetable oils instead of butter or lard or hard margarines.”
Vital Choice response: This assertion virtually ignores the totality of the laboratory, epidemiological, and clinical evidence, and treats debatable mathematical analyses of epidemiological evidence with far more weight than they deserve.
Although the Harvard team's analysis yielded results similar to those of some other epidemiological studies, all such studies are of inherently limited reliability.
In reality, the totality of the evidence points to two clear conclusions:
- Most Americans need to reduce omega-6 fat intake well below the current average.
- People should consume no more than 4 parts omega-6 fat to 1 part omega-3 fat, which was the ratio in human diets until the advent of the saturated fat theory of heart disease in the 1960s.
There is no evidence that the human species has adapted to thrive on the current intake ratio of 10 or 20 parts omega-6 fats to 1 part omega-3 fats.
In fact, such an adaptation may not be possible even over long periods of time, because omega-6 fats exert far more powerful effects on the human immune system, compared with omega-3 fats ... a point repeatedly stressed by Dr. Lands.
Again, you can learn more about this critical — but often-overlooked body of evidence — at "Omega-3/6 Balance: Hidden Health Risk”, from articles in the "Omega-3/Omega-6 Balance” section of our news archive, and by viewing our "Out of Balance” video.
*The credentials of our scientific advisors:
- Dr. William E. Lands, Ph.D., is one of the world's leading fatty acid researchers. He was Professor of Biochemistry in Medical Schools at the University of Michigan (1955-1980) and the University of Illinois (1980-1991) where he studied the metabolism of fats, phospholipids, and prostaglandins. Writing as Lands, WE (and recently as Lands, B) he authored over 300 papers and the book, Fish, Omega-3 and Human Health, 2nd Edition, updated in 2005 from the 1985, Fish and Human Health. One of the world's 1,000 most cited scientists in 1965-1978, he received numerous awards including the 1969 Glycerine Research Award, the 1979 Verhagen Lectureship at Rotterdam University, honorary membership in the Australian Rheumatism Association, the Pfizer Biomedical Research Award (1985), the AOCS-Supelco Lipid Research Award (1997), the ISSFAL Lifetime Achievement Award (2006) , AOCS-Holman Lifetime Achievement Award (2010) and selection as a Fellow of the American Association for the Advancement of Science, the Society for Free Radical Biology and Medicine and the American Society for Nutrition. After retiring from university teaching, he directed the basic research program at the National Institute on Alcohol Abuse and Alcoholism (1990-1997) and served there as Senior Scientific Advisor to the Director (1997-2002). He has a personal and corporate wellness site at http://efaeducation.org and a community informal education site at http://fastlearner.org/. Now, he is fully retired and can be contacted at email@example.com.
- Dr. Joseph Hibbeln, M.D., is a clinical psychiatrist at the National Institute of Health, a recognized expert on the role of omega-3 fats in brain health and child development, and the author or co-author of more than 130 peer-reviewed research papers. He currently serves as acting chief of the Section of Nutritional Neuroscience within the Institute's Laboratory of Membrane Biochemistry and Biophysics.
- Dr. Douglas Bibus, Ph.D., is a University of Minnesota biochemist who studied and assisted in the laboratory of Ralph T. Holman, Ph.D., the biochemist who discovered the existence, chemical structure, and importance of omega-3 fatty acids and named them omega-3s. He is a two-time winner of the American Chemical Society's Award in Analytical Chemistry and the author of more than 20 peer-reviewed research papers.
- Breslow JL. n-3 fatty acids and cardiovascular disease. Am J Clin Nutr. 2006 Jun;83(6 Suppl):1477S-1482S. Review. Circulation. 1999 Feb 16;99(6):733-5. Accessed at http://circ.ahajournals.org/content/99/6/733.long
- de Lorgeril M et al. Mediterranean Diet, Traditional Risk Factors, and the Rate of Cardiovascular Complications After Myocardial Infarction : Final Report of the Lyon Diet Heart Study Circulation1999;99:779-785 (Free full text)
- de Lorgeril M, Reanud S, Mamelle N, Salen P, Martin JL, Monjuad I, Gidolet J, Touboul P, Delaye J. Mediterranean alpha-linolenic acid-rich diet in secondary prevention of coronary heart disease. Lancet. 1994;343:1454-1459.
- de Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye J, Mamelle N. Mediterranean diet, traditional risk factors, and the rate of cardiovascular complications after myocardial infarction: final report of the Lyon Diet Heart Study. Circulation. 1999 Feb 16;99(6):779-85
- de Lorgeril M, Salen P. The Mediterranean diet in secondary prevention of coronary heart disease. Clin Invest Med. 2006 Jun;29(3):154-8.
- de Lorgeril M. Essential polyunsaturated fatty acids, inflammation, atherosclerosis and cardiovascular diseases. Subcell Biochem. 2007;42:283-97. Review.
- De Meester F. Progress in lipid nutrition: the Columbus Concept addressing chronic diseases. World Rev Nutr Diet. 2009;100:110-21. doi: 10.1159/000235716. Epub 2009 Aug 17. Review.
- Dolecek TA, Granditis G. Dietary polyunsaturated fatty acids and mortality in the Multiple Risk Factor Intervention Trial (MRFIT). World Rev Nutr Diet. 1991;66:205-16.
- Dolecek TA. Epidemiological evidence of relationships between dietary polyunsaturated fatty acids and mortality in the multiple risk factor intervention trial. Proc Soc Exp Biol Med. 1992 Jun;200(2):177-82.
- Gebauer SK, Psota TL, Harris WS, Kris-Etherton PM. n-3 fatty acid dietary recommendations and food sources to achieve essentiality and cardiovascular benefits. Am J Clin Nutr. 2006 Jun;83(6 Suppl):1526S-1535S. Review
- Ghosh S, Novak EM, Innis SM. Cardiac proinflammatory pathways are altered with different dietary n-6 linoleic to n-3 -linolenic acid ratios in normal, fat-fed pigs. Am J Physiol Heart Circ Physiol (2007)293: H2919-H2927.
- Griffin BA. How relevant is the ratio of dietary n-6 to n-3 polyunsaturated fatty acids to cardiovascular disease risk? Evidence from the OPTILIP study. Curr Opin Lipidol. 2008 Feb;19(1):57-62. Review.
- Harris WS, Mozaffarian D, Rimm E, Kris-Etherton P, Rudel LL, Appel LJ, Engler MM, Engler MB, Sacks F. Omega-6 fatty acids and risk for cardiovascular disease: a science advisory from the American Heart Association Nutrition Subcommittee of the Council on Nutrition, Physical Activity, and Metabolism; Council on Cardiovascular Nursing; and Council on Epidemiology and Prevention. Circulation. 2009 Feb 17;119(6):902-7. Epub 2009 Jan 26.
- Hibbeln JR et al. Healthy intakes of n-3 and n-6 fatty acids: estimations considering worldwide diversity.Am J Clin Nutr 2006 83: S1483-1493. Free Full Text.
- Hibbeln JR, Nieminen LR, Blasbalg TL, Riggs JA, Lands WE. Healthy intakes of n-3 and n-6 fatty acids: estimations considering worldwide diversity. Am J Clin Nutr. 2006 Jun;83(6 Suppl):1483S-1493S.
- Hibbeln JR, Nieminen LR, Lands WE. Increasing homicide rates and linoleic acid consumption among five Western countries, 1961–2000. Lipids 2004;39:1207–13.
- Lai CQ et al. Dietary Intake of n-6 Fatty Acids Modulates Effect of Apolipoprotein A5 Gene on Plasma Fasting Triglycerides,Remnant Lipoprotein Concentrations, and Lipoprotein Particle Size: The Framingham Heart Study.Circulation (2006)113: 2062-2070.
- Lands B. False profits and silent partners in health care. Nutr Health. 2009;20(2):79-89. Review.
- Lands B. Measuring blood fatty acids as a surrogate indicator for coronary heart disease risk in population studies. World Rev Nutr Diet. 2009;100:22-34. doi: 10.1159/000235708. Epub 2009 Aug 17. Review.
- Lands B. Planning primary prevention of coronary disease.Curr Atheroscler Rep. 2009 Jul;11(4):272-80. Review.
- Lands WE. Dietary fat and health: the evidence and the politics of prevention: careful use of dietary fats can improve life and prevent disease. Ann N Y Acad Sci. 2005 Dec;1055:179-92. • Leaf A. Dietary prevention of coronary heart disease: the Lyon Diet Heart Study.
- Leaf A. Dietary Prevention of Coronary Heart Disease: The Lyon Diet Heart Study.Circulation 1999;99:733-735.
- Multiple Risk Factor Intervention Trial Group. The Multiple Risk Factor Intervention Trial (MRFIT). A national study of primary prevention of coronary heart disease. (1976) Journal of the American Medical Association, 235(8): 825-827.
- Okuyama H, Ichikawa Y, Sun Y, Hamazaki T, Lands WE. The cholesterol hypothesis--its basis and its faults. World Rev Nutr Diet. 2007;96:1-17.
- Okuyama H, Ichikawa Y, Sun Y, Hamazaki T, Lands WE. Why isn't the causal relationship between linoleic acid and mortalities from coronary heart disease and stroke revealed by clinical studies? World Rev Nutr Diet. 2007;96:105-18.
- Ramsden CE, Faurot KR, Carrera-Bastos P, Cordain L, De Lorgeril M, Sperling LS. Dietary fat quality and coronary heart disease prevention: a unified theory based on evolutionary, historical, global, and modern perspectives. Curr Treat Options Cardiovasc Med. 2009 Aug;11(4):289-301.
- Ramsden CE, Hibbeln JR, Lands WE. Letter to the Editor re: Linoleic acid and coronary heart disease. Prostaglandins Leukot. Essent. Fatty Acids (2008), by W.S. Harris. Prostaglandins Leukot Essent Fatty Acids. 2009 Jan;80(1):77; author reply 77-8. Epub 2009 Jan 14.
- Ramsden CE, Hibbeln JR, Lands WE. Letter to the Editor re: Linoleic acid and coronary heart disease. Prostaglandins Leukot. Essent. Fatty Acids (2008), by W.S. Harris. Prostaglandins Leukot Essent Fatty Acids. 2009 Jan;80(1):77; author reply 77-8. doi: 10.1016/j.plefa.2008.12.002. Epub 2009 Jan 14.
- Ramsden CE, Hibbeln JR, Majchrzak SF, Davis JM. n-6 fatty acid-specific and mixed polyunsaturate dietary interventions have different effects on CHD risk: a meta-analysis of randomised controlled trials.
- Ramsden CE, Hibbeln JR, Majchrzak-Hong SF. All PUFAs are not created equal: absence of CHD benefit specific to linoleic acid in randomized controlled trials and prospective observational cohort. World Rev Nutr Diet. 2011;102:30-43. doi: 10.1159/000327789. Epub 2011 Aug 5.
- Ramsden CE, Ringel A, Feldstein AE, Taha AY, MacIntosh BA, Hibbeln JR, Majchrzak-Hong SF, Faurot KR, Rapoport SI, Cheon Y, Chung YM, Berk M, Mann JD. Lowering dietary linoleic acid reduces bioactive oxidized linoleic acid metabolites in humans. Prostaglandins Leukot Essent Fatty Acids. 2012 Oct-Nov;87(4-5):135-41. doi: 10.1016/j.plefa.2012.08.004. Epub 2012 Sep 5.
- Ramsden CE, Zamora D, Leelarthaepin B, Majchrzak-Hong SF, Faurot KR, Suchindran CM, Ringel A, Davis JM, Hibbeln JR. Use of dietary linoleic acid for secondary prevention of coronary heart disease and death: evaluation of recovered data from the Sydney Diet Heart Study and updated meta-analysis. BMJ. 2013 Feb 4;346:e8707. doi: 10.1136/bmj.e8707.
- Sepulveda JL, Tanhehco YC, Frey M, Guo L, Cropcho LJ, Gibson KM, Blair HC. Variation in human erythrocyte membrane unsaturated Fatty acids: correlation with cardiovascular disease. Arch Pathol Lab Med. 2010 Jan;134(1):73-80. doi: 10.1043/2008-0795-OAR1.1.
- Simopoulos AP, Leaf A, Salem N. Workshop on the Essentiality of and Recommended Dietary Intakes for Omega-6 and Omega-3 Fatty Acids.J Am Coll Nutr 1999 18: 487-489.
- Simopoulos AP. The importance of the omega-6/omega-3 fatty acid ratio in cardiovascular disease and other chronic diseases.Exp Biol Med (Maywood). 2008 Jun;233(6):674-88.
- Stamler J, Neaton JD, Cohen JD, Cutler J, Eberly L, Grandits G, Kuller LH, Ockene J, Prineas R; MRFIT Research Group. Multiple risk factor intervention trial revisited: a new perspective based on nonfatal and fatal composite endpoints, coronary and cardiovascular, during the trial. J Am Heart Assoc. 2012 Oct;1(5):e003640. doi: 10.1161/JAHA.112.003640. Epub 2012 Oct 25.
- Wang C, Harris WS, Chung M, Lichtenstein AH, Balk EM, Kupelnick B, Jordan HS, Lau J. n-3 Fatty acids from fish or fish-oil supplements, but not alpha-linolenic acid, benefit cardiovascular disease outcomes in primary- and secondary-prevention studies: a systematic review. Am J Clin Nutr. 2006 Jul;84(1):5-17. Review.
- Wang DD, Li Y, Chiuve SE, Stampfer MJ, Manson JE, Rimm EB, Willett WC, Hu FB. Association of Specific Dietary Fats With Total and Cause-Specific Mortality. JAMA Intern Med. 2016 Jul 5. doi: 10.1001/jamainternmed.2016.2417. [Epub ahead of print]