Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta-analysis of observational studies
BMJ 2015; 351 doi: https://doi.org/10.1136/bmj.h3978 (Published 12 August 2015) Cite this as: BMJ 2015;351:h3978
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Dear Drs. Freeman and Esselstyn Jr.,
Thank you for your insightful commentary.
First, in response to the comment about adjustment in observational studies—as we note in our discussion1, we agree that the validity of using “most-adjusted” models, which account both for potential confounders and causal intermediates is debatable2, 3. In the studies we meta-analyzed, several investigator groups adjusted for other risk factors for the health outcomes (such as body-mass-index, family history, and smoking), as well as changes in risk factors on the causal pathway between diet and disease, such as serum lipids and blood pressure. These adjustments certainly attenuate the observed relationships between SFA and the outcomes. To allow readers to ascertain the impact of these adjustments on the interpretability of the pooled risks, we present both models in our supplementary materials and GRADE tables.
Models adjusted for potential confounders and intermediate variables underestimate associations because of over-controlling for the effect of causal intermediates; unadjusted models overestimate associations, because estimates reflect other determinants of the health outcomes. Notably, in our study those in the highest categories of saturated fat intakes tended to exercise less, smoke more, and eat less fibre. So whether the stronger associations between saturated fat and the health outcomes in the unadjusted models more fully reflect the independent contribution of saturated fat is debatable. We would posit that the least adjusted models reflect the contribution to risk of several unhealthy behaviours which tend to travel together (higher saturated fat diets, smoking, physical inactivity, and lower fibre intake).
To assess the potential impact of over-adjustment, we assessed “intermediately-adjusted models”, i.e. those that adjusted for the most-relevant confounders (smoking, age, sex, and total energy), but not potential causal intermediates (blood pressure or anti-hypertensive medications, serum lipids or lipid-lowering medications) for associations between saturated fat and cardiovascular outcomes, for which we had a reasonable number of studies. In these sensitivity analyses, the adjusted RR for saturated fat and CHD mortality is 1.21 (95% CI: 0.93 to 1.58 in 8 studies); for total CHD is 1.05 (95% CI: 0.93 to 1.19 in 11 studies) and for ischemic stroke is 0.87 (95% CI: 0.76 to 1.00 in 2 studies), which would not meaningfully change our conclusions based on the fully-adjusted models. This demonstrates that the overall results of our synthesis are robust and are not substantially affected by different approaches to covariate adjustment. They are also consistent with the pooled analysis of 11 American and European cohort studies that show that replacement of saturated fat by carbohydrate was not associated with decreased risk of CHD4.
Second, the linear association reported in the Seven Countries’ Study for saturated fat and mortality is striking, and represents an important contribution to the study of fat and mortality. However, it bears noting that this model regressed average cohort values for saturated fat on age-adjusted all-cause mortality rates5. Such an ecological association must be interpreted with caution as it is hypothesis-generating at best, and does not imply causality. Because the unit of analysis in this regression model was the cohort (n=16), the investigators could only adjust for a limited number of confounding variables (vitamin C and smoking). For example, the confounding effects of socioeconomic status and occupation, physical activity, total energy, or family history could not be considered due to the sparseness of data. Further, the representativeness of the dietary data (which represented a fraction of the entire country-specific cohorts) may have been suboptimal.
Finally, the cited study of the efficacy of large dietary changes to reverse CAD is impressive. However, the results likely reflect more than the effect of simply reducing saturated fat. It is important to note that many diets which are naturally low in saturated fat—such as plant-based diets, contain several foods that are highly nutritious and generally regarded as cardioprotective, and avoid highly refined carbohydrates and trans fats. Therefore, when adopting such diets, people do more than just eliminate foods higher in saturated fat. For example, in the cited study6, involving participants who voluntarily asked for counseling on plant-based nutrition (i.e. a self-selected and highly-motivated group), some cardioprotective components were increased (whole plant foods such as whole grains, legumes, lentils, flaxseed, and fruit), others were decreased (dairy, nuts, plant oils, fish), and less healthful foods were reduced (meat, sugary drinks, excess salt, and caffeine). This study did not include a control group, which presents a challenge for establishing causality and ascribing the benefits solely to dietary changes. In Dr. Ornish’s study7, the combination of intensive dietary changes (very low fat, plant-based nutrition) and stress management yielded remarkable results. The magnitude of the response to comprehensive diet and lifestyle changes in both of these studies serve to underscore the importance of considering more than a single nutrient--a “whole diet” and lifestyle approach is likely to be the most impactful path to optimal health.
Sincerely,
Russell de Souza
Sonia Anand
Joseph Beyene
References
1. de Souza RJ, Mente A, Maroleanu A, Cozma AI, Ha V, Kishibe T, et al. Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta-analysis of observational studies. BMJ 2015;351:h3978.
2. Stamler J. Diet-heart: a problematic revisit. Am J Clin Nutr 2010;91:497-9.
3. Scarborough P, Rayner M, van Dis I, Norum K. Meta-analysis of effect of saturated fat intake on cardiovascular disease: overadjustment obscures true associations. Am J Clin Nutr 2010;92:458-9; author reply 59.
4. Jakobsen MU, O'Reilly EJ, Heitmann BL, Pereira MA, Balter K, Fraser GE, et al. Major types of dietary fat and risk of coronary heart disease: a pooled analysis of 11 cohort studies. Am J Clin Nutr 2009;89:1425-32.
5. Kromhout D, Bloemberg B, Feskens E, Menotti A, Nissinen A. Saturated fat, vitamin C and smoking predict long-term population all-cause mortality rates in the Seven Countries Study. Int J Epidemiol 2000;29:260-5.
6. Esselstyn CB, Jr., Gendy G, Doyle J, Golubic M, Roizen MF. A way to reverse CAD? J Fam Pract 2014;63:356-64b.
7. Ornish D, Scherwitz LW, Doody RS, Kesten D, McLanahan SM, Brown SE, et al. Effects of stress management training and dietary changes in treating ischemic heart disease. JAMA 1983;249:54-9.
Competing interests: This study was funded by WHO, which defrayed costs associated with preparing the draft manuscript, including information specialist and technical support and article retrieval costs. This systematic review was presented by RJdeS at the 5th Nutrition Guidelines Advisory Group (NUGAG) meeting in Hangzhou, China (4-7 March, 2013), the 6th NUGAG meeting in Copenhagen, Denmark (21-24 Oct, 2013), and the 7th NUGAG meeting in Geneva, Switzerland (9-12 Sept, 2014). WHO covered travel and accommodation costs for RJdeS to attend these meetings. The research questions for the review were discussed and developed by the WHO Nutrition Guidance Expert Advisory Group (NUGAG) Subgroup on Diet and Health and the protocol was agreed by the WHO NUGAG Subgroup on Diet and Health; however, neither WHO nor the WHO NUGAG Subgroup on Diet and Health had any role in data collection or analysis. Drs. Beyene and Anand declare no competing interests.
Dear Dr. Ornish,
Thank you for your insightful comments.
First, in response to the comment about adjustment in observational studies—as we note in our discussion1, we agree that the validity of using “most-adjusted” models, which account both for potential confounders and causal intermediates is debatable2, 3. In the studies we meta-analyzed, several investigator groups adjusted for other risk factors for the health outcomes (such as body-mass-index, family history, and smoking), as well as changes in risk factors on the causal pathway between diet and disease, such as serum lipids and blood pressure. These adjustments certainly attenuate the observed relationships between SFA and the outcomes. To allow readers to ascertain the impact of these adjustments on the interpretability of the pooled risks, in our supplementary materials and GRADE tables, we present both models.
Models adjusted for potential confounders and intermediate variables underestimate associations because of over-controlling for the effect of causal intermediates; unadjusted models overestimate associations, because estimates reflect other determinants of the health outcomes. Notably, in our study those in the highest categories of saturated fat intakes tended to exercise less, smoke more, and eat less fibre. So whether the stronger associations between saturated fat and the health outcomes in the unadjusted models more fully reflect the independent contribution of saturated fat is debatable. We would posit that the least adjusted models reflect the contribution to risk of several unhealthy behaviours which tend to travel together (higher saturated fat diets, smoking, physical inactivity, and lower fibre intake).
To assess the potential impact of over-adjustment, we assessed “intermediately-adjusted models”, i.e. those that adjusted for the most-relevant confounders (smoking, age, sex, and total energy), but not potential causal intermediates (blood pressure or anti-hypertensive medications, serum lipids or lipid-lowering medications) for associations between saturated fat and cardiovascular outcomes, for which we had a reasonable number of studies. In these sensitivity analyses, the adjusted RR for saturated fat and CHD mortality is 1.21 (95% CI: 0.93 to 1.58 in 8 studies); for total CHD is 1.05 (95% CI: 0.93 to 1.19 in 11 studies) and for ischemic stroke is 0.87 (95% CI: 0.76 to 1.00 in 2 studies), which would not meaningfully change our conclusions based on the fully-adjusted models. This demonstrates that the overall results of our synthesis are robust and are not substantially affected by different approaches to covariate adjustment. They are also consistent with the pooled analysis of 11 American and European cohort studies that show that replacement of saturated fat by carbohydrate was not associated with decreased risk of CHD4.
Second, we fully agree with you that the choice of replacement nutrient is important, and we have tried to make this point in several different ways in our paper (in the abstract as well as extensive discussion). Our approach to our study was to address two questions sequentially—first what is the impact of higher saturated fat on the health outcomes? This was the primary question we sought to answer, as developed in conjunction with the World Health Organization. However we recognize that simply giving advice on “eating less” saturated fat is not useful unless some alternative nutrient recommendation is made—and this is precisely how we interpret our main “null” association. We thus proceeded to provide context for this “null” association in our discussion and supplementary material, which considers replacement nutrients. We elected not to conduct a meta-analysis to answer these questions because the available observational evidence for replacement nutrients is sparse, yet remains important. We thus discussed both observational4, 5 and interventional6, 7 studies, to provide a more complete and accurate picture of the evidence base.
Third, we agree that there may be some important differences in the associations across sources of saturated fats—indeed this may contribute to the overall heterogeneity of our findings for saturated fat. Our mandate, was to address total saturated fats primarily; and then to examine the evidence for specific types. In our supplementary material, we have reviewed the evidence for specific sources of saturated fats, which may be found in Appendix 3 for dairy fats (15:0 and 17:0; generally neutral), myristic acid (14:0; generally neutral), palmitic acid (16:0; generally neutral); stearic acid (18:0; generally neutral). However, we would interpret these findings with some caution because of the small number of studies contributing to these analyses. Further, individual saturated fatty acids may share many common food sources, which is challenging for food-based nutrition guidelines. We also summarize here the results of individual studies of exchanging sources of saturated fat for one another8, 9. More work is needed to understand the roles of different saturated fats on health, and more importantly, how to craft public health messages surrounding them.
Finally, we would also draw attention to the fact that using the GRADE rating scale, which takes into account the limitations of the available evidence, we rated our certainty in the null association between saturated fats and the outcomes as overall, “very low”. This means that our estimate of association is very uncertain, in light of the many limitations of the included studies—including those which you have raised. Furthermore, our meta-analysis is observational, and thus cannot infer causality. We fully acknowledge that future work, and other lines of evidence (i.e. randomized controlled trials of replacement nutrients) have an important impact on how we interpret the associations between saturated fat and mortality, CHD/CVD, and type 2 diabetes, and this body of evidence should not be ignored. Thank you for your interest in our paper, and we appreciate the opportunity to respond.
Sincerely,
Russell de Souza
Joseph Beyene
Sonia Anand
References
1. de Souza RJ, Mente A, Maroleanu A, Cozma AI, Ha V, Kishibe T, et al. Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta-analysis of observational studies. BMJ 2015;351:h3978.
2. Stamler J. Diet-heart: a problematic revisit. Am J Clin Nutr 2010;91:497-9.
3. Scarborough P, Rayner M, van Dis I, Norum K. Meta-analysis of effect of saturated fat intake on cardiovascular disease: overadjustment obscures true associations. Am J Clin Nutr 2010;92:458-9; author reply 59.
4. Jakobsen MU, O'Reilly EJ, Heitmann BL, Pereira MA, Balter K, Fraser GE, et al. Major types of dietary fat and risk of coronary heart disease: a pooled analysis of 11 cohort studies. Am J Clin Nutr 2009;89:1425-32.
5. Farvid MS, Ding M, Pan A, Sun Q, Chiuve SE, Steffen LM, et al. Dietary linoleic acid and risk of coronary heart disease: a systematic review and meta-analysis of prospective cohort studies. Circulation 2014;130:1568-78.
6. Mozaffarian D, Micha R, Wallace S. Effects on coronary heart disease of increasing polyunsaturated fat in place of saturated fat: a systematic review and meta-analysis of randomized controlled trials. PLoS Med 2010;7:e1000252.
7. Hooper L, Martin N, Abdelhamid A, Davey Smith G. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev 2015;6:CD011737.
8. Hu FB, Stampfer MJ, Manson JE, Ascherio A, Colditz GA, Speizer FE, et al. Dietary saturated fats and their food sources in relation to the risk of coronary heart disease in women. Am J Clin Nutr 1999;70:1001-8.
9. de Oliveira Otto MC, Mozaffarian D, Kromhout D, Bertoni AG, Sibley CT, Jacobs DR, Jr., et al. Dietary intake of saturated fat by food source and incident cardiovascular disease: the Multi-Ethnic Study of Atherosclerosis. Am J Clin Nutr 2012;96:397-404.
Competing interests: This study was funded by WHO, which defrayed costs associated with preparing the draft manuscript, including information specialist and technical support and article retrieval costs. This systematic review was presented by RJdeS at the 5th Nutrition Guidelines Advisory Group (NUGAG) meeting in Hangzhou, China (4-7 March, 2013), the 6th NUGAG meeting in Copenhagen, Denmark (21-24 Oct, 2013), and the 7th NUGAG meeting in Geneva, Switzerland (9-12 Sept, 2014). WHO covered travel and accommodation costs for RJdeS to attend these meetings. The research questions for the review were discussed and developed by the WHO Nutrition Guidance Expert Advisory Group (NUGAG) Subgroup on Diet and Health and the protocol was agreed by the WHO NUGAG Subgroup on Diet and Health; however, neither WHO nor the WHO NUGAG Subgroup on Diet and Health had any role in data collection or analysis. Drs. Beyene and Anand declare no competing interests.
Dear Dr. Blackburn,
Thank you for your insightful comments.
As we note in our discussion1, we agree that the validity of using “most-adjusted” models, which account both for potential confounders and causal intermediates is debatable2, 3. In the studies we meta-analyzed, several investigator groups adjusted for other risk factors for the health outcomes (such as body-mass-index, family history, and smoking), as well as changes in risk factors on the causal pathway between diet and disease, such as serum lipids and blood pressure. These adjustments certainly attenuate the observed relationships between SFA and the outcomes. To allow readers to ascertain the impact of these adjustments on the interpretability of the pooled risks, in our supplementary materials and GRADE tables, we present both models.
Models adjusted for potential confounders and intermediate variables underestimate associations because of over-controlling for the effect of causal intermediates; unadjusted models overestimate associations, because estimates reflect other determinants of the health outcomes. Notably, in our study those in the highest categories of saturated fat intakes tended to exercise less, smoke more, and eat less fibre. So whether the stronger associations between saturated fat and the health outcomes in the unadjusted models more fully reflect the independent contribution of saturated fat is debatable. We would posit that the least adjusted models reflect the contribution to risk of several unhealthy behaviours which tend to travel together (higher saturated fat diets, smoking, physical inactivity, and lower fibre intake).
To assess the potential impact of over-adjustment, we assessed “intermediately-adjusted models”, i.e. those that adjusted for the most-relevant confounders (smoking, age, sex, and total energy), but not potential causal intermediates (blood pressure or anti-hypertensive medications, serum lipids or lipid-lowering medications) for associations between saturated fat and cardiovascular outcomes, for which we had a reasonable number of studies. In these sensitivity analyses, the adjusted RR for saturated fat and CHD mortality is 1.21 (95% CI: 0.93 to 1.58 in 8 studies); for total CHD is 1.05 (95% CI: 0.93 to 1.19 in 11 studies) and for ischemic stroke is 0.87 (95% CI: 0.76 to 1.00 in 2 studies), which would not meaningfully change our conclusions based on the fully-adjusted models. This demonstrates that the overall results of our synthesis are robust and are not substantially affected by different approaches to covariate adjustment. They are also consistent with the pooled analysis of 11 American and European cohort studies that show that replacement of saturated fat by carbohydrate was not associated with decreased risk of CHD4.
Thank you for your interest in our paper, and we appreciate the opportunity to respond.
Sincerely,
Russell J de Souza
Joseph Beyene
Sonia S Anand
References
1. de Souza RJ, Mente A, Maroleanu A, Cozma AI, Ha V, Kishibe T, Uleryk E, Budylowski P, Schunemann H, Beyene J, Anand SS. Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: Systematic review and meta-analysis of observational studies. BMJ. 2015;351:h3978
2. Stamler J. Diet-heart: A problematic revisit. Am J Clin Nutr. 2010;91:497-499
3. Scarborough P, Rayner M, van Dis I, Norum K. Meta-analysis of effect of saturated fat intake on cardiovascular disease: Overadjustment obscures true associations. Am J Clin Nutr. 2010;92:458-459; author reply 459
4. Jakobsen MU, O'Reilly EJ, Heitmann BL, Pereira MA, Balter K, Fraser GE, Goldbourt U, Hallmans G, Knekt P, Liu S, Pietinen P, Spiegelman D, Stevens J, Virtamo J, Willett WC, Ascherio A. Major types of dietary fat and risk of coronary heart disease: A pooled analysis of 11 cohort studies. Am J Clin Nutr. 2009;89:1425-1432
Competing interests: This study was funded by WHO, which defrayed costs associated with preparing the draft manuscript, including information specialist and technical support and article retrieval costs. This systematic review was presented by RJdeS at the 5th Nutrition Guidelines Advisory Group (NUGAG) meeting in Hangzhou, China (4-7 March, 2013), the 6th NUGAG meeting in Copenhagen, Denmark (21-24 Oct, 2013), and the 7th NUGAG meeting in Geneva, Switzerland (9-12 Sept, 2014). WHO covered travel and accommodation costs for RJdeS to attend these meetings. The research questions for the review were discussed and developed by the WHO Nutrition Guidance Expert Advisory Group (NUGAG) Subgroup on Diet and Health and the protocol was agreed by the WHO NUGAG Subgroup on Diet and Health; however, neither WHO nor the WHO NUGAG Subgroup on Diet and Health had any role in data collection or analysis. Drs. Beyene and Anand declare no competing interests.
Trans fats consumption associated with low intake vitamins, antioxidants and minerals and their effect on health.
We have read the article by Dr de Souza RJ and colleagues in your recent issue of the journal which shows that increased consumption of trans fats is associated with increased risk of mortality and coronary artery disease(1). A decrease in fruit and vegetable consumption is associated with poor health and increased risk of non communicable diseases (NCDs). Low intake of fruits and vegetables is seen in families with low socioeconomic status and this is often due to the increase in cost of fruits and vegetables (2). Vegetables contain generally 90-96% water and fruits water content is between 80 and 90% (3). When food is cooked, cooking medium and the duration of heating can decrease the vitamin (fat-soluble and water-soluble vitamins) content of foods as vitamins are susceptible to heat. Cooking method that can result in the loss of minerals. This generates undesirable compounds (like nitrosamines) during frying, baking, grilling, smoking and roasting . Cooking can also increase the formation and bio-availability of some antioxidants, such as lycopene. (2) and decrease water content (4). Increased consumption of trans fats decreases plasma high-density lipoprotein (HDL) and increases low-density lipoprotein (LDL)-cholesterol. Trans fat intake can increase the body fat (5).
Increase in trans fat consumption leads to impairment of memory behavioral irritability and aggression in adults (6.7). In India the fat content of vegetable oils from street vendors has high levels of saturated fat and trans fat. People are consuming snacks having increased content of trans fat as part of daily diets. Trans fat consumption in India exceeds WHO recommendations for transfat intake (8). Trans fats intake is also associated with breast cancer, shortening of pregnancy period, risks of preeclampsia, disorders of nervous system, colon cancer, diabetes, obesity and allergy(9). A recent study shows that eating while walking leads to overeating compared to eating during other forms of distraction such as watching TV or during a conversation with a friend(10).The distraction, including eating at our desks also can lead to weight gain (10). People eat more often commercially prepared foods while walking, watching TV , and working at desk.
Consumption of snacks with trans fat is more often associated with low intake of dietary fibre, vegetables and fruits leading to low intake of vitamins, antioxidants and minerals. The deficiency of these can also contribute to metabolic disorders and cardiovascular disease .
References
1.de Souza RJ, Mente A, Maroleanu A, Cozma AI, Ha V, Kishibe T, Uleryk E, Budylowski P, Schünemann H, Beyene J, Anand SS. Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta-analysis of observational studies.
BMJ. 2015 Aug 11;351:h3978. doi: 10.1136/bmj.h3978.
2. Increasing fruit and vegetable consumption to reduce the risk of non communicable diseases
http://www.who.int/elena/titles/bbc/fruit_vegetables_ncds/en/
3. Bastin S , Henken K Water Content of Fruits and Vegetables. University of Kentucky - Cooperative Extension Service - December 1997. https://www2.ca.uky.edu/enri/pubs/enri129.pdf
4. The Why, How and Consequences of cooking our food. EUFIC REVIEW 11/2010
http://www.eufic.org/article/en/expid/cooking-review-eufic/
5. Bendsen NT, Chabanova E, Thomsen HS, Larsen TM, Newman JW, Stender S, Dyerberg J, Haugaard SB, Astrup A. Effect of trans fatty acid intake on abdominal and liver fat deposition and blood lipids: a randomized trial in overweight postmenopausal women.
Nutr Diabetes. 2011 Jan 31;1:e4. doi: 10.1038/nutd.2010.4.
6. Golomb BA, Bui AK A Fat to Forget: Trans Fat Consumption and Memory
PLoS One. 2015 Jun 17;10(6):e0128129. doi: 10.1371/journal.pone.0128129. eCollection 2015.
7.Golomb BA1, Evans MA, White HL, Dimsdale JE. Trans fat consumption and aggression PLoS One. 2012;7(3):e32175. doi: 10.1371/journal.pone.0032175. Epub 2012 Mar 5.
8. Downs SM1, Singh A2, Gupta V3, Lock K4, Ghosh-Jerath S. The need for multisectoral food chain approaches to reduce trans fat consumption in India
BMC Public Health. 2015 Jul 22;15(1):693. doi: 10.1186/s12889-015-1988-7.
9. Dhaka V, Gulia N, Ahlawat KS, Khatkar BS. Trans fats—sources, health risks and alternative approach - A review. Journal of food science and technology. 2011;48(5):534-541. doi:10.1007/s13197-010-0225-8.
10. Jane Ogden J , Oikonomou E and Alemany G. Distraction, restrained eating and disinhibition: An experimental study of food intake and the impact of ‘eating on the go’ Journal of Health Psychology Ogden, August 2015 DOI:10.1177/1359105315595119
Competing interests: No competing interests
Much work has already shown adverse outcomes from too much saturated fat in the diet, in addition to the development of diabetes, heart disease, and other cardiovascular events. As such, your work published here has gone too far in that anytime one over-adjusts for important variables, the real data and trends become hidden. it is analogous to over-smoothing an image — the granular image becomes so smooth it obscures some of the salient features.
As the authors indicate in their own work, a 5% increase in saturated fat increased mortality by nearly 5% — this is a highly important trend that was not included in the final conclusions.
Furthermore, recent work (1) reported experience eliminating saturated fats in 200 participants with significant cardiovascular disease. They were requested to eat whole food plant based nutrition and avoid oils, meat, fish, poultry, dairy products, eggs, and sugary drinks. After 3.75 years of follow up highlights of the 177 adherent participants included:
1 Program adherence 89.3%
2 Elimination of major cardiac events (heart attack, stroke, death) 99.4%
3 Resolution of angina 93%
4 Evidence of disease reversal via angiogram, carotid ultrasound, stress test, resolution of claudication and erectile dysfunction.
5 Twenty-seven participants successfully avoided previously recommended interventions of stents and bypass surgery.
This study, the pioneering work of Ornish and more recently others employing similar techniques, makes it inconceivable to embrace saturated fats and disagrees with the findings of De Souza, et al.
Caldwell B. Esselstyn, Jr., M.D.
Director Cardiovascular Disease Prevention and Reversal Program
Cleveland Clinic Wellness Institute
1950 Richmond Road
Lyndhurst, Ohio 44124
Andrew Freeman, MD, FACC
Co-chair, American College of Cardiology Nutrition Workgroup
Director of Clinical Cardiology, National Jewish Health, Denver, CO
1. Esselstyn CB Jr, Gendy G, Doyle J, Golubic M, Roizen MF. A Way to Reverse CAD. The Journal of Family Practice. July 2014 Vol 63, No 7 page 257
Competing interests: No competing interests
With great interest we read the meta-analysis by the Souza and colleagues titled "Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta-analysis of observational studies", published in the British Medical Journal [1]. In their comprehensive article, the authors systematically reviewed observational studies investigating the associations between saturated fat (SFA) and trans unsaturated fat on health status (including: all-cause mortality, cardiovascular disease (CVD incidence and mortality), coronary heart disease (CHD incidence and mortality), ischemic stroke, and type 2 diabetes mellitus). The authors concluded that there is no association between SFA (comparing the highest category vs. lowest category) and all health outcomes, mentioning the heterogeneous evidence including methodological limitations. However, it should be noted that subgroup analysis showed a significant association between higher intakes of SFA and CHD mortality (i.e. for studies conducted in America, participants: <60 years of age, <25% smoking prevalence).
There is a plethora of systematic reviews and meta-analyses investigating the effects of SFA on CVD and CHD [2], reporting contradictory results especially when comparing data from observational studies with those from randomized controlled trials [3]. Recently, Hooper et al. 2015 [4] conducted a meta-analysis of intervention trials and observed a 17% reduced risk of combined cardiovascular events comparing low SFA vs. high SFA diets. No significant effects could be observed for all-cause mortality, CVD/CHD mortality, myocardial infarction, and stroke. The authors concluded that the "evidence supports the reduction of saturated fat to reduce cardiovascular events within the timescale of these dietary trials”[4]. Apart from the differences in the primary findings, the meta-analyses by Souza et al. [1] and by Hooper et al. [4] disagree with respect to their evaluation of the quality of evidence according to the GRADE group [5] which was rated “very low” by Souza et al. [1], but “moderate” by Hooper et al. [4]. Applying the GRADE guidelines in Nutritional Sciences has several limitations, since RCTs in nutrition research are often prone to inherent methodological constraints. E.g., they sometimes cannot be controlled with "true" placebos, but rather by a limitation of certain aspects of nutrient compositions, food groups or dietary patterns. Other limitations include lack of double blinding, poor compliance and adherence, cross-over bias and drop-out. Thus, in the field of nutritional epidemiology, well-designed prospective cohort studies will provide important evidence as well [6]. Consequently, the quality of evidence statements as recommended by the GRADE working group, where observational studies started with a initial quality of a body of evidence “low”, should be reconsidered for prospective cohort studies in Nutritional Sciences [5].
Another important point raised by de Souza et al. [1] is the substitution of SFA by unsaturated fats (PUFA and/or MUFA). Recently, we performed a systematic review on dietary fatty acids in the secondary prevention of coronary heart disease including 12 trials with 7,150 patients [7]. Following meta-analyses as well as uni- and multivariate meta-regression, it was concluded that recommendations of higher intakes of PUFA in replacement of SFA was not associated with risk reduction of all-cause mortality, CVD, and CHD in subjects with CHD. Moreover, a meta-analysis of cohort studies investigating the effects of MUFA, oleic acid, and olive oil showed that MUFA of mixed animal and vegetable sources per se did not yield any significant effects on health status (all-cause mortality, CVD, CHD, and stroke) [8], while olive oil seemed to be associated with reduced risk. Additionally, a high MUFA:SFA ratio was inversely associated with all-cause and CVD mortality [8]. The high MUFA:SFA ratio (high intake of extra virgin olive oil) is one of the main components of the Mediterranean dietary pattern [9], which was associated with reduced risk of various non-communicable diseases [10-12].
A meta-analysis of cohort studies showed that only processed meat (all-cause mortality, CVD mortality, cancer mortality), and meat (all-cause mortality) were associated with increased risk, whereas no significant association could be observed for butter, cheese and milk [13]. Therefore, future meta-analyses should not focus primarily on fatty acids (i.e. SFA, MUFA, PUFA), but on specific foods (such as butter, milk, meat, and salmon), or food patterns which are the main sources of SFA.
References
1. de Souza RJ, Mente A, Maroleanu A, et al. Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta-analysis of observational studies, 2015.
2. Astrup A, Dyerberg J, Elwood P, et al. The role of reducing intakes of saturated fat in the prevention of cardiovascular disease: where does the evidence stand in 2010? The American journal of clinical nutrition 2011;93(4):684-8 doi: 10.3945/ajcn.110.004622[published Online First: Epub Date]|.
3. Hoenselaar R. Saturated fat and cardiovascular disease: the discrepancy between the scientific literature and dietary advice. Nutrition 2012;28(2):118-23 doi: 10.1016/j.nut.2011.08.017[published Online First: Epub Date]|.
4. Hooper L, Martin N, Abdelhamid A, et al. Reduction in saturated fat intake for cardiovascular disease. The Cochrane database of systematic reviews 2015;6:CD011737 doi: 10.1002/14651858.cd011737[published Online First: Epub Date]|.
5. Balshem H, Helfand M, Schunemann HJ, et al. GRADE guidelines: 3. Rating the quality of evidence. Journal of clinical epidemiology 2011;64(4):401-6 doi: 10.1016/j.jclinepi.2010.07.015[published Online First: Epub Date]|.
6. Satija A, Yu E, Willett WC, et al. Understanding nutritional epidemiology and its role in policy. Advances in nutrition 2015;6(1):5-18 doi: 10.3945/an.114.007492[published Online First: Epub Date]|.
7. Schwingshackl L, Hoffmann G. Dietary fatty acids in the secondary prevention of coronary heart disease: a systematic review, meta-analysis and meta-regression. BMJ open 2014;4(4):e004487 doi: 10.1136/bmjopen-2013-004487[published Online First: Epub Date]|.
8. Schwingshackl L, Hoffmann G. Monounsaturated fatty acids, olive oil and health status: a systematic review and meta-analysis of cohort studies. Lipids in health and disease 2014;13:154 doi: 10.1186/1476-511X-13-154[published Online First: Epub Date]|.
9. Trichopoulou A, Costacou T, Bamia C, et al. Adherence to a Mediterranean diet and survival in a Greek population. The New England journal of medicine 2003;348(26):2599-608 doi: 10.1056/NEJMoa025039[published Online First: Epub Date]|.
10. Schwingshackl L, Hoffmann G. Mediterranean dietary pattern, inflammation and endothelial function: a systematic review and meta-analysis of intervention trials. Nutrition, Metabolism and Cardiovascular Diseases 2014(0):(accepted manuscript; doi: http://dx.doi.org/10.1016/j.numecd.2014.03.003 ) doi: http://dx.doi.org/10.1016/j.numecd.2014.03.003%5Bpublished Online First: Epub Date]|.
11. Schwingshackl L, Hoffmann G. Adherence to Mediterranean diet and risk of cancer: A systematic review and meta-analysis of observational studies. International journal of cancer. Journal international du cancer 2014 doi: 10.1002/ijc.28824[published Online First: Epub Date]|.
12. Schwingshackl L, Missbach B, Konig J, et al. Adherence to a Mediterranean diet and risk of diabetes: a systematic review and meta-analysis. Public health nutrition 2015;18(7):1292-9 doi: 10.1017/S1368980014001542[published Online First: Epub Date]|.
13. O'Sullivan TA, Hafekost K, Mitrou F, et al. Food sources of saturated fat and the association with mortality: a meta-analysis. American journal of public health 2013;103(9):e31-42 doi: 10.2105/ajph.2013.301492[published Online First: Epub Date]|.
Competing interests: No competing interests
In the very same week it is reported that medication used in the treatment of diabetes now accounts for some 10 per cent of spending by the National Health Service here in the UK, I am somewhat baffled that the media – in reporting a study containing the most convincing evidence to date that saturated fats are not harmful to health – have largely overlooked what is perhaps the most significant conclusion in this study (1): namely that trans-palmitoleic acid, which is present in dairy products, offers protection from type 2 diabetes, the form of the disease responsible for the relentless increase in the incidence of diabetes in the UK and other westernised societies.
For years, misguided ‘experts’ have advised the public to reduce their consumption of saturated fats, so much so that dairy products have been all but demonised. It is now high time the record was set straight. The shift to vegetable oils in place of dairy products has caused a massive increase in the ratio of omega-6 to omega-3 polyunsaturates in western diets. This shift is associated with pro-inflammatory conditions and it is no coincidence that it has been accompanied by an equally alarming increase in the incidence of our major inflammatory diseases, namely type 2 diabetes and cardiovascular disease (2,3).
Evidence for the beneficial effects of particular trans-fats of endogenous origin, as opposed to those of industrial origin, is widely documented in the literature (4-6). In addition to trans-palmitoleic acid (5), cis-9, trans-11 conjugated linoleic acid, the main isomer of CLA present in dairy products, is also believed to be protective against diabetes (6). Now that the health benefits of these trans-fats have been ‘demonstrated’ using statistics (the ‘blind’ method, so favoured by those forever offering inconsistent dietary advice), will the experts please come clean and admit to the wider public that they were wrong to demonise dairy products? I quote a statement from a 2010 paper by Mozaffarian and colleagues (5):
‘Our results suggest that efforts to promote exclusive consumption of low-fat and nonfat dairy products, that would lower population exposure to trans-palmitoleate, may be premature until the mediators of health benefits of dairy consumption are better established.’
It is due to such efforts to reduce saturated fat intake that the typical low-fat ‘healthy’ yoghurt contains some 16 grams of sugar, which is certainly not conducive to good health! In the intervening years, since the 2010 paper, studies have been published which support the assertion that the consumption of dairy products is associated with a lowered risk of type 2 diabetes (7, 8).
The public have not been credited with the ability to handle anything more than health messages that are so uncomplicated – so simple – as to be useless, if not outright counterproductive. Why did all trans-fats have to be demonised just because the artificial, industrial varieties are harmful? Many people still fret unnecessarily about the presence of trans-fats in their food, but the food industry has gone to considerable lengths to reduce these to amounts that are insignificant to those consuming a sensible, balanced diet. Foods containing fats of animal origin will always contain very small amounts of trans-fats, but these are a completely different kettle of fish to those that were once introduced in large quantities by artificial means.
Those who recommend the replacement of dietary animal fats with vegetable oils usually base their reasoning on the observation that this leads to a reduction in plasma low density lipoprotein (LDL), which – the argument goes – lowers the risk of cardiovascular disease. This reasoning is based, however, almost entirely on statistical ‘risk factors’ and fails to take proper account of the underlying biochemical mechanisms of atherosclerosis.
Whilst it is unclear precisely how atherosclerosis is initiated, it is evident that the ongoing process comprises a vicious cycle of inflammation and self-amplifying free-radical chain reactions, culminating in the modification of LDL to an oxidised form (‘oxLDL’) that is avidly taken up via the scavenger receptor pathway. The basic chemistry of LDL modification is ‘screaming out loud’ to us that polyunsaturated fatty acids, being far more susceptible to lipid peroxidation than their saturated relatives, should promote the process (9).
If we accept that a certain fraction of LDL in the bloodstream is going to be converted to oxLDL, then of course lowering total LDL will lower the absolute level of the modified lipoprotein. It is an unfortunate coincidence, then, that the dietary measures commonly recommended to lower total LDL (‘high in polyunsaturates/low in saturates’) can logically be assumed to promote oxLDL formation. Rather than intervening downstream to the LDL oxidation step, after the damage has been done, it would surely make more sense to consume diets which minimise LDL modification in the first place. Raised LDL per se is not the problem: prevent its oxidation and elevated levels should not be atherogenic.
Unfortunately such nuances will never be revealed by the adoption of a mechanism-blind statistical approach to the investigation of the links between diet and cardiovascular disease. No matter how ‘significant’ the statistics, how perfectly ‘controlled’ the sample, number crunching should always be subservient to mechanism-based biological chemistry. If we are to break free from this perpetual cycle of conflicting dietary advice, a major rethink is required as to what constitutes evidence worthy of forming the basis of dietary guidelines. Blind statistics it is not.
References
1. De Souza, R. J. et al. (2015) Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta analysis of observational studies. BMJ 351:h3978
2. Simopoulos, A. P. (1999) Essential fatty acids in health and chronic disease. Am. J. Clin. Nutr. 70 (suppl.), 560S – 569S
3. Blasbalg, T. L. et al. (2011) Changes in consumption of omega 3 and omega 6 fatty acids in the United States during the 20th century. Am. J. Clin. Nutr. 93, 950 – 962
4. Pariza, M. W. (2004) Perspective on the safety and effectiveness of conjugated linoleic acid. Am. J. Clin. Nutr. 79 (suppl.), 1132S – 1136S
5. Mozaffarian, D. et al. (2010) Trans-palmitoleic acid, metabolic risk factors, and new-onset diabetes in US adults. Ann Intern. Med. 153, 790 – 799
6. Castro-Webb, N., Ruiz-Narváez, E. A. and Campos, H. (2012). Cross-sectional study of conjugated linoleic acid in adipose tissue and risk of diabetes. Am. J. Clin. Nutr. 96, 175 – 181
7. O’Connor, L. M. et al. (2014) Dietary dairy product intake and incident type 2 diabetes: a prospective study using dietary data from a 7-day food diary. Diabetologia DOI 10.1007/s00125-014-3176-1
8. Zong, G. et al. (2014) Dairy consumption, type 2 diabetes, and changes in cardiometabolic traits: a prospective cohort study of middle-aged and older Chinese in Beijing and Shanghai. Diabetes Care 37, 56 – 63
9. Burkitt, M. J. (2001) A critical overview of the chemistry of copper-dependent low density lipoprotein oxidation: roles of lipid hydroperoxides, -tocopherol, thiols, and ceruloplasmin. Arch. Biochem. Biophys. 394, 117 –135
Competing interests: No competing interests
From a public health perspective, the question is, whether it will improve the health of a population when they are given instructions to restrict saturated fat. We should remember that today this applies to all ages - that infants in kindergartens and primary schools do not have access to whole milk because of concerns about saturated fat and heart disease in the adult population. There is an increased incidence of rickets today, which may in part be an unintended consequence of this instruction, as may increased incidences of diabetes and obesity (due to the replacement of natural foods relatively high in saturated fat with grains, refined carbohydrates, and other highly processed foods that may promote over-eating).
If so, this is troubling, because obesity and diabetes are clearly risk factors for cardiovascular disease.
From our point of view, it is easiest for people to control weight and blood sugar if they remove sugar, refined starch, and carbohydrate dense foods in general from their diet. Should they then worry about the saturated fat in the whole foods and minimally refined cooking fats they use?
The chain of causality accepted by another correspondent can be summarised as "1) SFA raises LDL, 2) LDL is associated with CVD, 3) lowering LDL is associated with lower rate of CVD".
However,
1) LDL cholesterol is a calculated figure derived from other measurements and the atherogenicity of the LDL particles depends on factors such as ApoB, ApoC3, small dense LDL particles, etc. which are controlled by the response to dietary carbohydrate than by the response to saturated fats, which tend to increase the proportion of larger and less atherogenic LDL particles. This beneficial response to saturated fat is increased when the carbohydrate content of the diet is lower. [1]
2) Mendelian randomisation shows that 50% of the relationship between LDL cholesterol and CHD risk is due to genetic, rather than diet or lifestyle, factors.[2] For this reason it is not unreasonable to adjust CHD data for cholesterol levels. CHD cases can have higher LDL than age-matched controls even when both groups have been eating the same diet for some time.[3]
3) Replacing saturated fat with carbohydrate lowers LDL cholesterol but is not associated with any reduction in CVD in sub-group meta-analysis of RCTs. On the other hand, cholesterol lowering associated with increased intake of PUFA is associated with a reduction in CVD events, but not fatal events or all-cause mortality.[4]. In prospective cohort studies, replacing carbohydrate from all sources with PUFA is associated with marginally more benefit than replacing saturated fat with PUFA.[5]
We have to get energy from somewhere, and the human capacity to derive energy from extra PUFAs without increased harm is probably limited. Ancel Keys et al in the Seven Countries Study found that the natural human diets they studied supplied between 3-7% energy from linoleate, with minimal contributions from other PUFAs, and did not consider that the minimal cholesterol lowering effect of this small variation in PUFA could accounted for the different heart disease rates between populations. [6] PUFAs are essential nutrients with anti-thrombotic effects and the foods that supply them also supply vitamin E and other tocopherols that might be expected to modify cardiovascular risk through CPK pathways. The essentiality of PUFAs and the presence of other essential nutrients in PUFA foods are confounders when analyzing the effect of replacing other nutrients, such as saturated fats, with PUFA.
Such limits on saturated fats as promulgated in the past have tended to increase the intake of trans fats, and in that part of the population convinced to eat low-fat diets may have even decreased the consumption of polyunsaturated fats.
We have more to gain today by preventing and reversing type 2 diabetes, obesity, and metabolic syndrome, all significant risk factors for CVD mortality, in the context of a nutritionally optimal diet that restricts carbohydrate-dense foods and processed foods, and we can do this most easily if we do not impose artificial limits on the population's intake of saturated fats, an instruction that has probably played a role in increasing the incidence of these conditions.
[1] Siri-Tarino P et al. (2015) Polyunsaturated Fats Versus Carbohydrates for Cardiovascular Disease Prevention and Treatment. Annu. Rev. Nutr. 2015. 35:517–43
[2] Holmes MW et al (2015) Mendelian randomization of blood lipids for coronary heart disease. Eur Heart J. 2015 Mar 1;36(9):539-50. doi: 10.1093/eurheartj/eht571. Epub 2014 Jan 27.
[3] Oliver MF, Boyd GS (1953) The Plasma Lipids in Coronary Artery Disease. Br Heart J. 1953 Oct;15(4):387-92
[4] Hooper L. et al (2015) Reduction in saturated fat intake for cardiovascular disease, The Cochrane Library, June 10 2015. DOI: 10.1002/14651858.CD011737
[5] Farvid MS et al (2014) Dietary linoleic acid and risk of coronary heart disease: a systematic review and meta-analysis of prospective cohort studies. Circulation. 2014 Oct 28;130(18):1568-78. doi: 10.1161/CIRCULATIONAHA.114.010236.
[6] Keys A et al (1980) Seven Countries. A multivariate analysis of death and coronary heart disease. Cambridge, MA; Harvard University Press, ISBN: 0-674-80237-3, 1980. 381 pp.
Competing interests: No competing interests
Questioning the Evidence for the Dietary Fat-Heart Theory: Deja Vue All Over Again
As every journalist knows, to generate interest on a well-worn topic you have to be a contrarian. Now meta-analyses to question the diet-heart are the latest fad – as evidenced in this paper (1). Nutrition has the dubious distinction of providing the greatest volume of re-cycled discoveries that are contrary to conventional wisdom. Most can be safely disregarded as fads, but some potentially have great public health relevance. Coronary heart disease (CHD) is still the commonest cause of death in industrialized countries, so our understanding of the causal process is of paramount importance. The recent determined challenges to the veracity of the dietary fat-heart theory are therefore a serious cause for concern.
The structure of the dietary-fat-heart theory is elegant and comprehensive. First recognized a century ago in animal experimental research, dietary cholesterol and saturated fat raise serum LDL-cholesterol, in turn promoting atheroma and thrombosis. This diet-serum lipids relationship has been replicated in a large number of meticulously conducted nutrition trials in humans (2). In cohort studies, like Framingham, and international comparisons, like the Seven Countries Study, elevated LDL-C powerfully predicts events in groups, and less precisely in individuals. A relatively small set of other lifestyle factors interacts with genes to modulate that risk. Current LDL-C lowering agents reduce events in the magnitude predicted by observational studies.
Then why the resurgent controversy? Skepticism appears to be driven by three concerns. First, a chorus of objections to dietary fats instead suggest that refined carbohydrates, in particular sugar, are the key nutrient raising LDL-C. However, this concern has been repeatedly addressed since the 1980s. The body of evidence from metabolic ward studies, observational studies and RCTs consistently suggests that sugars may indeed contribute to the rising prevalence of obesity and diabetes; however, they are only a weak risk factor for CHD (3). Second, the obesity/diabetes epidemic has seized huge attention in the public health nutrition debate. So much, in fact, that we have lost sight of the larger impact of saturated fat on serum cholesterol. Finally, contrarians argue that the absence of randomized controlled trials (RCT) demonstrating lower intake in saturated fatty acids reduces CHD leaves the dietary-fat heart theory “unproven” (4).
This last objection is a form of analytic fundamentalism that perennially haunts epidemiology like a bad debt. While the RCT is an invaluable tool to study the benefits/risks of medical interventions it is hardly the only logical design for testing hypotheses. Randomized experiments form only a small fraction of the evidence base across scientific disciplines. In evolution, geology, astrophysics and many others, the RCT plays no role whatsoever, yet strong theories with enormous predictive power have emerged. Has anyone seen an RCT on sympatric speciation or plate tectonics? Worse yet, if data from an RCT was obligatory evidence on which to base judgments, we would still be at a total loss as how to address humanity’s biggest challenge – climate change. Biomedicine also has its own exemplars- RCTs played no role in proving that smoking causes lung cancer, and the impact of air pollution on CHD and asthma has never been “proven” by that standard. Instead inter-locking steps of a causal pathway, leading through a series of inferences, more often link cause to consequence (5). When we look for the etiology of common diseases, lowly observational data usually provide the crucial information.
But the dietary-fat-heart sceptics face an even bigger challenge. At the zenith of the CHD epidemic, mortality in the US among persons 25+ was 746 per 100,000. Preventive campaigns based on the precepts of the dietary-fat-heart theory initiated a sharp and continuous decline in CHD death rates - a public health phenomenon that even cardiovascular epidemiologists refused to believe for at least 10 years. In 2013, the age-adjusted CHD mortality was 159/100,000, 77% lower than in 1968. Thus 1.5 million fewer deaths occurred than if the 1968 rate had persisted. Clearly, several interventions each played a role. Fewer cigarettes smoked, kick-started the CHD decline, and the shift to low-fat dairy products helped drive cholesterol down from 220 to 200 mg/dl. As shown in numerous countries, the precipitous decline in CHD mortality fits a model quantifying powerful declines in cholesterol, smoking and hypertension, plus medical interventions in more recent decades (7). It is sobering to recognize that the widespread use of aspirin in the US accounted for a larger mortality decline than surgery and angioplasty combined (7).
If the dietary-fat-heart theory was wrong, why, despite a steep increase in sugar consumption and obesity, have CHD mortality rates continued to drop 5% / year through 2013?
Health professionals have two essential social functions – to improve the quality of life of their patients and to reduce mortality. Over the last half century, the dietary-fat-heart theory has arguably contributed more to the health of the US population than any other single “theory” apart from tobacco control (another theory equally “unproven” by RCTs). We should continue to bet on the winners.
References
1. Intake of saturated and trans unsaturated fatty acids and risk of all cause mortality, cardiovascular disease, and type 2 diabetes: systematic review and meta-analysis of observational studies BMJ 2015; 351 doi: http://dx.doi.org/10.1136/
2. Clarke R, Frost C, Collins R, Appleby P, Peto R. Dietary lipids and blood cholesterol: quantitative meta-analysis of metabolic ward studies. BMJ. 1997;314:112-7.
3. Stanhope K, Medici V, Bremer AA, Lee V, Lam HD, Nunez MV, Guoxia X Chen GX, Nancy L Keim NL, Havel PJ. A dose-response study of consuming high-fructose corn syrup–sweetened beverages on lipid/lipoprotein risk factors for cardiovascular disease in young adults. Am J Clin Nutr 2015:10.3945 ; doi:10.3945/ajcn.114.100461
4. Zoë H, Julien SB, Stephen MC, Bruce Davies3, Nicholas S, James J D, Fergal G. Evidence from randomised controlled trials did not support the introduction of dietary fat guidelines in 1977 and 1983: a systematic review and meta-analysis. Open Heart 2015;2: doi:10.1136/openhrt-2014-000196.
5. Hill AB. The environment and disease: Association or causation? Proceedings Roy Soc Med 1965;58:295-300.
6. Cooper R, Stamler J, Dyer A, Garside D. The decline in mortality from coronary heart disease, U.S.A., 1968-1975. J Chron Dis 1978;31:709-720.
7. Ford ES1, Ajani UA, Croft JB, Critchley JA, Labarthe DR, Kottke TE, Giles WH, Capewell S. Explaining the decrease in U.S. deaths from coronary disease, 1980-2000. N Engl J Med. 2007;356:2388-98.
Source: CDC; WONDER. Accessed May 7, 2015
Competing interests: No competing interests
Does quality of dietary fats matter?
Meta-analyses have also become popular in nutrition research. However, statistics behind a random-effects meta-analysis is a nuanced one (1). When random-effects model for meta-analysis is fitted, one needs to estimate between-study variance (τ2 i.e. tau-squared) with one of the many estimators available (1) to obtain summary effect estimate and confidence intervals (CIs) for this effect estimate.
Routinely, the random-effects modeling is based on approach proposed by DerSimonian and Laird (DL) (2) and CIs for the summary effect estimate are constructed using standard normal distribution. However, this approach might not be the most reliable approach, at least in some situations (3,4). Different analytical choices might make difference since research findings are typically based on a statistical significance threshold.
For example, a recent Cochrane review showed in prespecified subgroup analysis that saturated fats (SFA) replaced with polyunsaturated fats (PUFA) reduced cardiovascular events (risk ratio [RR] 0.73; 95% CI 0.58 to 0.92, P=0.007, Tau2=0.06, I2=69%, 7 studies) (5,6). A Mantel-Haenszel method was used in the Cochrane review, however, an inverse-variance method yielded identical results as depicted above.
I used recent editorial by Hooper et al considering their Cochrane review (6) to extract the relevant data, and re-analyzed it with R (R Core Team 2014, R Foundation For Statistical Computing, Vienna, Austria, version 3.1.2) using meta package (7) to adjust confidence interval and p-value for the summary effect estimate with the t-distribution based method by Hartung and Knapp and by Sidik and Jonkman (HKSJ) (see ref. 3 for details).
As shown in Figure, RR for cardiovascular events was 0.73 (95% CI 0.50 to 1.08, two-tailed P=0.095) with the method by HKSJ. As evident, the CI became wider and the P-value larger. By all means, this re-analysis should not be taken at face value, however, it illustrates that statistics matter. Evidently, if one is willing to accept the statistically significant versus non-significant framework, then two different conclusions abound.
In simulated meta-analyses, for example with 7 unequal sized trials and statistical heterogeneity (I2= 50-90%), error rates exceeded the 5% threshold and were around 10% with the DL approach (3). In addition, of 185 statistically significant binary outcome Cochrane meta-analyses with DL, 26 % (48/185) were non-significant with HKSJ (3). Interestingly, this is exactly what happened with the two different analytical choices. The width of CI might be too narrow in the aforementioned Cochrane review or any other meta-analysis with a statistical heterogeneity thereof (4). Naturally, other between-study variance estimators than DL can be used.
Undoubtedly, a Bayesian framework can provide broader interpretation of statistically significant research findings with the calculation of posterior probability for true positivity (8).
After all, the strength of evidence depends on who evaluates it and which methodology is adopted by the guideline panels for grading the evidence. Maybe statistics is much more important than acknowledged.
1. Veroniki AA, Jackson D, Viechtbauer W, et al. Methods to estimate the between-study variance and its uncertainty in meta-analysis. Res Synth Methods 2015 Sep 2. doi: 10.1002/jrsm.1164.
2. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986; 7: 177-88.
3. IntHout J, Ioannidis JP, Borm GF. The Hartung-Knapp-Sidik-Jonkman method for random effects meta-analysis is straightforward and considerably outperforms the standard DerSimonian-Laird method. BMC Med Res Methodol 2014; 14: 25.
4. Cornell JE, Mulrow CD, Localio R, et al. Random-effects meta-analysis of inconsistent effects: a time for change. Ann Intern Med 2014; 160: 267-70.
5. Hooper L, Martin N, Abdelhamid A, Davey Smith G. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev 2015; 6: CD011737.
6. Hooper L, Martin N, Abdelhamid A. Cochrane corner: what are the effects of reducing saturated fat intake on cardiovascular disease and mortality? Heart 2015 Oct 9. doi: 10.1136/heartjnl-2015-308521.
7. Schwarzer Guido. meta: General Package for Meta-analysis. R-package version 4.3-0. 2015. https://cran.r-project.org/web/packages/meta/index.html. Cited October 20, 2015.
8. Pereira TV, Ioannidis JP. Statistically significant meta-analyses of clinical trials have modest credibility and inflated effects. J Clin Epidemiol 2011; 64: 1060-9.
jesper.m.kivela@helsinki.fi
Competing interests: I like statistics.