Effect of reducing total fat intake on body weight: systematic review and meta-analysis of randomised controlled trials and cohort studies

BMJ 2012; 345 doi: http://dx.doi.org/10.1136/bmj.e7666 (Published 6 December 2012)
Cite this as: BMJ 2012;345:e7666

Recent rapid responses

Rapid responses are electronic letters to the editor. They enable our users to debate issues raised in articles published on bmj.com. Although a selection of rapid responses will be included as edited readers' letters in the weekly print issue of the BMJ, their first appearance online means that they are published articles. If you need the url (web address) of an individual response, perhaps for citation purposes, simply click on the response headline and copy the url from the browser window.

Displaying 1-10 out of 15 published

Dear Frank, Deirdre, Dariush, Frank and Walter,

Thank you for your replies and further points. We reply, as ever, below.

1. Habitual dietary intake which promotes long term health vs. weight loss diets for people trying to lose weight
We agree that it is important to understand what sorts of diet are useful in helping people to lose weight or maintain their weight loss. However, this was not the question we asked. We aimed to assess the effect of total fat intake in the long term without the intention to lose weight. This is a valid question and one which the World Health Organization was keen for us to address. We look forward to reading a new systematic review of RCTs which assesses what types of diets are most useful for weight loss, and to finding out whether those diets which are best for weight loss are also best for promoting long term health. World-wide we would be extremely surprised if most people are trying to lose weight or maintain weight loss, although in some countries, a significant proportion of people are, at any one point in time, worried about their weight, and some of these people will be on a weight loss diet.

2. Cohort studies
We are surprised that you feel that the cohort data contradict the RCT data. We found that, of 16 assessments on the effects of total fat intake on subsequent weight change in seven cohorts, 11 showed no statistically significant effect, and five (31%) showed a statistically significant positive effect. None showed a statistically significant negative effect, although one of the 4 comparisons between total fat intake and change in waist circumference did (1). Although this is not a ringing endorsement, it is certainly more positive than negative in support of the RCT data. In the two included cohort studies of which one of you is a key author, which are high quality studies and include large numbers of participants (the Health Professionals Follow up study (2), and the Nurses’ Health Study (3)) the evidence is mixed. In 19,478 male health professionals aged 45 to 75, multivariate regression analyses was used to determine whether total fat intake was predictive of 4-year weight change, and it was found that a change of adjusted fat intake of 10g/d predicted 0.10kg of weight change over 4 years, but only in younger men (p<0.001 for ages 45-54 and 55-64 years, p>0.05 for age 65+) (2). In 31,940 women (nurses) aged 30+ the correlation between total fat (g/d) and weight gain over the subsequent 4 years (beta -0.0007, t -0.4) was not statistically significant. These analyses which used high quality cohort data appear to lend weak support to the RCT data – but they certainly do not suggest that there is a negative relationship between total fat intake and weight change. (Please see the published supplementary data for the fat and weight systematic review for more detail on the included cohort studies (4)).

3. WHI
Yes, there was a more dramatic effect seen early on in this study, but stable and significantly lower body weights were found in those eating less fat long term. The WHI trialists did not have the benefit of an overview of every long term trial of dietary fat reduction for at least 6 months in people not trying to lose weight, or they may have been more enthusiastic about the statistically significant weight loss they observed. This is a very consistent result – not just seen in their study, but across the board.

4. Lipid effects
We did not aim to provide data on the effect of a low fat diet on blood lipids. We checked lipid outcomes in studies that were included in our review simply to ensure that there were no adverse effects on lipids or blood pressure that might need to be balanced against any effects seen on body weight. Our review shows clearly that there is no suggestion of any such adverse effects in people eating less fat in real life settings in the long term (the trials Mensink and Katan included were of much shorter duration (5)).

5. Reduced and modified fat patterns
We do understand that mono-unsaturated fats can come from animal sources, or from vegetable sources. There is no problem in substituting saturated fats for some starchy foods and some monounsaturated fats if the latter are from vegetable sources such as olive oil. Being able to make this change depends, of course, on the appropriate foods being available, which they certainly are in Europe and New Zealand. You state that an appropriate source of monounsaturated fats is not available in the USA, which we find surprising but acknowledge your expertise in this area. We hope that policy makers in the USA (and countries in a similar position) find our review, and the WHO recommendations which this review feeds into, useful in levering various mechanisms by which the supply of appropriate foods rich in monounsaturated fats may be increased.

We understand the principle of energy balance, and that if a person consumes more energy then they expend, regardless of the source of that energy intake, they will gain weight.

The evidence is clear. There is high quality, consistent evidence that reduction of total fat intake has been achieved in large numbers of both healthy and at risk trial participants over many years. Lower total fat intake leads to small but statistically significant and clinically meaningful, sustained reductions in body weight from six months to over eight years in adults with baseline fat intakes of 28-43% of energy intake. Evidence supports a similar effect in children and young people.

With best wishes,
Lee Hooper and Carolyn Summerbell on behalf of all the authors.

Reference List

(1) Klesges RC, Klesges LM, Haddock CK, Eck LH. A longitudinal analysis of the impact of dietary intake and physical activity on weight change in adults. American Journal of Clinical Nutrition 1992;55:818-22.
(2) Coakley EH, Rimm EB, Colditz GA, Kawachi I, Willett WC. Predictors of weight change in men: results from the health professionals follow-up study. International Journal of Obesity 1998;22:89-96.
(3) Colditz GA, Willett WC, Stampfer MJ, London SJ, Segal MR, Speizer FE. Patterns of weight change and their relation to diet in a cohort of healthy women. American Journal of Clinical Nutrition 1990;51:1100-5.
(4) Hooper L, Abdelhamid A, Moore HJ, Douthwaite W, Skeaff CM, Summerbell CD. Effect of reducing total fat intake on body weight: systematic review and meta-analysis of randomised controlled trials and cohort studies. British Medical Journal 2012 Dec 7;345:e7666.
(5) Mensink RP, Katan MB. Effect of Dietary Fatty Acids on Serum Lipids and Lipoproteins. A meta-analysis of 27 trials. Arteriosclerosis and Thrombosis 1992;12:911-9.

Competing interests: Our competing interest is that we are the authors of this systematic review.

Lee Hooper, Senior Lecturer in Research Synthesis and Nutrition

Asmaa Abdelhamid, Helen J Moore, Wayne Douthwaite, C Murray Skeaff, Carolyn D Summerbell

Norwich Medical School, University of East Anglia, Norwich NR4 7TJ, UK

Click to like:

Dear Lee, Carolyn, Murray, and co-authors,
We thank you for your detailed and prompt response. A few points require further clarification.

1. Ad lib vs. weight loss trials
The distinction between ad lib and weight loss trials is not so clear as most people at any one time point are trying to lose and maintain weight. In fact, low-carbohydrate dietary interventions are often non-calorie restricted, and allow participants to consume foods high in protein and fat ad libitum.1 Further, while the intervention may not intend to promote weight loss, participants may be motivated to participate in a dietary intervention as an opportunity or motivation for personal weight loss goals. Therefore, it is important to consider evidence from weight loss and maintenance trials.

2. Cohort study confounding
While observational cohort studies are susceptible to confounding and bias if not carefully conducted, this may not explain discrepancies between the observational and trial findings. In the general population, participants with higher fat intake tend to have poorer lifestyle characteristics, such as low physical activity, higher intakes of refined carbohydrates, etc, which, if concerned about confounding, would actually bias results to favor low-fat diets even more. Rather, cohort studies do not support lower fat intake for long-term weight loss and maintenance.

3. WHI
Results from the WHI trial do not demonstrate that the “full effect was established early in the trial and maintained at 7 years”; rather participants in the low-fat diet group had appreciable mean weight loss (2.2 kg) at 1 year, which was almost entirely regained by year 4, and maintained at or just below their baseline weight through the rest of follow-up. The full effect, therefore, was not maintained. Of note, WHI investigators themselves refrained from overstating their findings to suggest long-term benefits of low-fat diets on body weight, concluding instead that the low-fat diet did not adversely affect weight gain.

4. Lipid effects
The meta-analysis concluded that there was no adverse effect on blood lipids (or even a benefit) of low fat trials is problematic because some of the trials that measured lipids were aiming to improve blood lipids and included a replacement of sat fat with unsaturated fat. This conclusion is very different from the results of carefully controlled feeding studies (e.g., Katan meta-analysis) that show that fat reduction increases TG’s and reduces HDL cholesterol.

5. Reduced/modified fat diet and starchy foods
Suggesting a reduced and modified fat permanent eating pattern is impractical and possibly even harmful advice to bestow on the general population. Saturated and monounsaturated fats are often derived from similar food sources such as meats and dairy. It would be difficult to achieve the reduction of one and not the other, from common food sources. “Starchy foods” are not healthy replacement for even “bad” fats, as many less healthy foods fall into this category, including potatoes, refined grains, and white breads, which have all been associated with weight gain.2 From a public health perspective, a simple reduced fat strategy would not work. Instead, emphasizing quality of fat and carbohydrates and overall diet quality is more scientifically sound and also more practical.

1. Shai I, Schwarzfuchs D, Henkin Y, Shahar DR, Witkow S, Greenberg I, et al. Dietary Intervention Randomized Controlled Trial (DIRECT) Group. N Engl J Med 2008; 359(3):229-41.
2. Mozaffarian D, Hao T, Rimm EB, Willett WC, Hu FB. Changes in diet and lifestyle and long-term weight gain in women and men. N Engl J Med 2011; 364(25):2392-404.

Competing interests: None declared

Frank Hu, Professor

Deirdre K. Tobias, Dariush Mozaffarian, Frank Sacks, Walter Willett

Harvard School of Public Health, 665 Huntington Ave

Click to like:

Dear Frank, Deirdre, Dariush, Frank, Walter and Gary,

Thank you all for your thoughts on our review. Yes, we did find that dietary fat reduction was associated with a slightly lower relative body weight, but not that this diminished over time (the meta-regression found no effect of study duration) (1). And no, most of the cohort studies did not support the consistent results of the 33 RCTs, but the RCTs deal much better with the very high levels of confounding of lifestyle, socioeconomic and dietary factors associated with lower fat intake in the general population, and so are more likely to tell us usefully the effect of lower fat intake. This is why the GRADE assessment relies on the RCTs (as they are many and consistent), and not on the cohort studies – evidence from cohort studies is much less able to determine causality.

Point 1. We excluded studies that aimed to cause weight loss in any of the participants as these studies were more likely to overstate any effect size. We were not interested in diets for people trying to lose weight, but on the effect of fat intake in people eating normally.

RCTs in which weight loss and subsequent weight maintenance are the primary endpoints are conceptually and methodically different from RCTs in which the macronutrient composition of the diet is modified without intention to lose weight. The cornerstone of dietary advice for weight loss is restriction of total energy intake, macronutrient composition is of lesser importance; therefore, all participants in weight loss trials are counselled, first and foremost, to eat less and they intend to do so. In this setting, the effects of the macronutrient composition of the diet are tested in participants motivated to lower their total energy intake and to lose weight. It is interesting that the results of weight loss trials show that the strongest predictor of successful weight loss and subsequent weight maintenance is compliance with the weight loss intervention, regardless of its’ macronutrient composition (2).

In our paper, we addressed a different question; the effect of the macronutrient composition of the diet, specifically total fat, on body weight in trials in which participants were not seeking actively to restrict total energy intake or to lose weight. Dietary recommendations for people who need to lose weight are important, but were not our focus. Systematic reviews of trials of different methods of weight reduction are to be welcomed, but this was not the remit provided by WHO.

Point 2. We agree that the Women’s Health Initiative (WHI) trial provided the majority of RCT participants, as it was an extraordinarily large trial – which was why we ran the meta-analysis using random effects as the main analysis (effect size −1.57kg (95% CI −1.97 to −1.16, 27 RCTs), and ran sensitivity analyses using fixed effects analysis (effect size −1.04kg, 95% CI −1.18 to −0.90, 27 RCTs), and also omitting WHI (effect size −1.65kg, 95% CI −2.10 to −1.21, 26 RCTs) - see table 3 in the full published paper for these results. Whatever way you run the analysis participants cutting down on fat have significantly lower weight than controls. The main analysis used the random effects model as the included RCTs had different types of interventions – though of course they all reduced total fat in the intervention groups, but it was important to check that the significant effect did not rest solely on this choice of model – which it does not. The significant effect remains regardless of the pooling method.

You state that the effect size in the Women’s Health Initiative (WHI) was “0.7 kg at 7.5 years or <0.1 kg per year” – the effect was, as you say, 0.7kg lower weight in intervention participants compared to control participants at the final assessment. But this does NOT equate to 0.1kg per year – the full effect was established early in the trial and maintained at 7 years. The point is that at 7 years the intervention group is STILL significantly lighter – this is a long term effect.

Point 3 (and Gary’s point). We agree that the findings of short duration trials are less useful than those of longer duration. This was why we only included studies that continued for at least 26 weeks, and used the latest duration data available for each included trial. The meta-regression clarified that study duration (once a trial is at least 26 weeks long) does not relate to the size of the weight loss – the size of the effect does vary according to the degree of fat reduction, and according to the level of fat intake at baseline, but not by study duration. For this reason it is valid to run a sensitivity analysis to remove studies with more attention to the intervention group. The effect size does not alter in the sensitivity analysis, showing clearly that the weight loss was not due to greater attention or time or encouragement in the intervention group. The effect size in the main analysis was −1.57kg (95% CI −1.97 to −1.16, 27 RCTs), the effect size in the studies that did not provide more attention to the intervention group (Gary’s studies where all participants “receive identical care”) was −1.42kg (95% CI −2.12 to −0.73, 8 RCTs).

Using secular trends to single out a decrease in energy from dietary fat as the major factor causing the rise in obesity rates in the US is tenuous epidemiologically. Secular trends in China show unequivocally that the rise in energy from dietary fat parallels the rise in obesity rates (3); however, this evidence suffers from the same limitations as that in the US. No doubt the changes in rates of obesity in China and US are attributable to secular changes across a range of dietary and non-dietary factors (including levels of physical activity).

In the US between 1971 and 2000 obesity levels rose from 15 to 31% and total fat (in g/d) stayed very constant, although energy intakes rose. It was not fat reduction that caused the rise in obesity over this period, but increased energy intakes as Americans ate increased portion sizes, more salty snacks, soft drinks and pizza (4). Although people didn’t reduce the amount of fat they ate, the proportion of calories from fat fell (as they ate more calories). This made it possible for observers to suggest that the fall in the percentage of energy from fat caused the weight rise (as the time trend seemed confirmatory). The much more obvious cause of the rise in obesity was the rise in calories.

Conversely, the evidence that cutting down on the amount of fat we eat causes a fall in weight is very strong (1). It is a consistent message from 33 long term randomised controlled trials (the best scientific way of testing cause and effect) conducted in men and women, on several continents, for up to 11 years duration, in people who were well and those who had chronic illness. Evidence in children supports the adult evidence.

The results of our meta-analysis revealed that the typical “compensatory increase in carbohydrates” in the lower fat treatment groups was associated with reduced not increased cardiovascular risk profile. This suggests that the dietary changes that accompanied the lower fat diets – in those not trying to lose weight – were favourable not deleterious. We believe this result based on RCTs is worthy of consideration.

We agree with your point on modifying fat intake to reduce the risk of cardiovascular disease – this was the finding of our own Cochrane systematic review of RCTs (updated last year) on the effects of fats on cardiovascular events and mortality (5). Neither reducing total fat (replacing fats, usually mainly with carbohydrates) nor modifying fat (replacing saturates and processed fats with mono- and poly-unsaturates) showed statistically significant effects on mortality or cardiovascular mortality (in randomised controlled trials of at least 6 months duration) in over 65,000 participants.
Combining RCTs that either reduced or modified fat did reduce cardiovascular events by 14% (RR 0.86, 95% CI 0.77 to 0.94, 24 trials, 65,508 participants of whom 7% had a cardiovascular event). When we separated out (subgrouped) trials into those which reduced total fat, modified fat, or did both the results were as follows, compared those who continued their usual fat intake:
a) Reduced fat intake: RR 0.97, 95% CI 0.87 to 1.08, 50,655 participants, 3632 cardiovascular events, p 0.55
b) Modified fat intake: RR 0.82, 95% CI 0.66 to 1.02, 11,660 participants, 855 cardiovascular events, p 0.07
c) Reduced and Modified fat intake: RR 0.77, 95% CI 0.57 to 1.03, 3193 participants, 400 cardiovascular events, p 0.08

While effects of b and c were not quite statistically significant (we clearly lost some power in subgrouping) they are of marginal significance (the reduced fat diet, a, was not even close to statistical significance). The suggestion is that a combination of reducing and modifying fat is as beneficial in reducing the risk of cardiovascular events as the modified fat intake. This is a way of eating that will both protect against cardiovascular disease and gain the long term weight benefits of a low fat diet.

This “reduced and modified fat” way of eating should be seen as a permanent eating pattern (not a short term diet) as benefits in reducing cardiovascular events become apparent from about 2 years after dietary changes are initiated. A reduced and modified fat diet can be achieved by cutting down on saturated and trans fats, and replacing with starchy foods, while ensuring that the fats that are eaten are mainly poly- and mono-unsaturated oils.

Gary suggests that the data are dubious due to a lack of blinding. It is of course correct that blinding is the ideal in reducing bias. However humans are very good at identifying fat – even in the included institutional studies (where all food was provided to both intervention and control participants) and shop studies (where participants purchased food in a study shop blinded to whether they were provided with full or reduced fat foods) participants could almost certainly tell whether they were eating the control or the reduced fat versions. Even these trials that tried to blind the intervention were not ideal, but were the best evidence we will be able to find. They certainly provide higher quality evidence of causality than cohort studies that are confounded by many other dietary, lifestyle and socioeconomic factors.

The evidence is clear. There is high quality, consistent evidence that reduction of total fat intake has been achieved in large numbers of both healthy and at risk trial participants over many years. Lower total fat intake leads to small but statistically significant and clinically meaningful, sustained reductions in body weight in adults in studies with baseline fat intakes of 28-43% of energy intake and durations from six months to over eight years. Evidence supports a similar effect in children and young people.

With best wishes for a healthy 2013, Lee Hooper, Carolyn Summerbell and Murray Skeaff on behalf of all the authors.

1. Effect of reducing total fat intake on body weight: systematic review and meta-analysis of randomised controlled trials and cohort studies. Hooper L, Abdelhamid A, Moore HJ, Douthwaite W, Skeaff CM, Summerbell CD. BMJ 2012; 345 doi: http://dx.doi.org/10.1136/bmj.e7666

2. Dansinger ML, Gleason JA, Griffith JL, Selker HP, Schaefer EJ. Comparison of the Atkins, Ornish, Weight Watchers, and Zone diets for weight loss and heart disease risk reduction: a randomized trial. JAMA: The Journal of the American Medical Association 2005, 293(1), 43–53. doi:10.1001/jama.293.1.43

3. Zhai F, Wang H, Du S, He Y, Wang Z, Ge K, Popkin BM. Prospective study on nutrition transition in China. Nutrition Reviews 2009, 67, S56–S61. doi:10.1111/j.1753-4887.2009.00160.x

4. Trends in Intake of Energy and Macronutrients – United States, 1971-2000. Wright JD, Kennedy-Stephenson J, Wang CY, McDowell MA, Johnson CL. National Centre for Health Statistics CDC, accessed 4th Dec 2012. http://www.cdc.gov/mmwr/preview/mmwrhtml/mm5304a3.htm

5. Reduced or modified dietary fat for preventing cardiovascular disease. Hooper L, Summerbell CD, Thompson R, Sills D, Roberts FG, Moore H, Davey Smith G. Cochrane Database of Systematic Reviews 2011 issue 7. Art No CD002137. DOI: 10.1002/14651858.CD002137.pub2.

Competing interests: We are the authors of the systematic review being discussed, but have no further competing interests.

Lee Hooper, Senior Lecturer in Research Synthesis and Nutrition

Asmaa Abdelhamid, Helen J Moore, Wayne Douthwaite, C Murray Skeaff, Carolyn D Summerbell

Norwich Medical School, University of East Anglia, Norwich NR4 7TJ, UK

Click to like:

In their systematic review and meta-analysis of 33 randomized controlled trials (RCTs; n=57,735), Hooper et al. reported that dietary fat reduction was associated with a slightly lower relative body weight (difference = 1.6 kg, which diminished over time).1 The authors concluded that “lower total fat intake leads to small but statistically significant and clinically meaningful, sustained reductions in body weight in adults…” This conclusion, however, was not supported by findings from their parallel meta-analysis of 10 observational cohort studies (n=107,624). Further, the meta-analysis of the RCTs was conceptually and methodologically unsound, rendering the conclusion unjustified.

First, the authors deliberately excluded RCTs aimed at testing effects of macronutrient composition on long-term weight loss and subsequent weight maintenance. The exclusion of such trials is questionable, given that the most commonly recommended method for both prevention of weight gain, weight loss, and weight maintenance is through the alteration of macronutrient composition. With this exclusion criterion, several long-term, rigorously conducted RCTs designed specifically to test the effects of altering macronutrient composition on body weight were not included in the meta-analysis. These trials have consistently shown that reducing total fat intake and replacing it with carbohydrate is not more effective for weight loss and maintenance than altering other macronutrients;2-5 Rather, adherence to a higher-fat Mediterranean-style diet for 2 years compared to low-fat diets conferred significant benefits on body weight and cardiovascular risk factors, which were sustained during an additional 4 years of follow-up.6-7 Inclusion of these trials in the meta-analysis would likely substantially alter the overall results and conclusions.

Second, of the 57,735 participants represented in the meta-analysis, 41,353 participants (72%) came from a single trial, the Women’s Health Initiative Dietary Modification Trial (WHI).8 Yet, WHI represented only 7% of the total weight because random-effects inverse variance weighting was used to compute the weighted mean difference across trial estimates. In the presence of highly varying trial sizes and findings, as seen here, random-effects models assign disproportionally higher weights to smaller and less-precise studies.9 For example, in the meta-analysis by Hooper et al., the four intervention comparisons reported in the DEER trial with only 367 participants, received 16% of the total weight (more than double the weight of the WHI).10 The consequence of such an approach is an over-contribution of the results from small and less precise studies. As expected, secondary fixed-effects model demonstrated a much smaller overall pooled effect (1.0 kg weight difference for low-fat diets vs. controls). This effect size is close to the difference in body weight between the low-fat group and control group observed in the WHI (0.7 kg at 7.5 years or <0.1 kg per year). These findings argue against any appreciable long-term benefit of advising low-fat diets on body weight.

Third, a fundamental limitation of trials of macronutrient change on body weight is that blinding is difficult and any substantial change in diet can lead to a small weight loss, thus creating a bias in favor of the more intensive intervention. Hooper et al. reported that the results were consistent when they removed trials where more attention was given to the participants in the low fat arm (only 8 of the original 33 trials remaining), but almost all of these were trials of less than one year. The generalizability of findings from short-term trials is questionable as six months is the typical time point in which transient weight loss tapers off and weight begins to rebound. Given that the effect of dietary fat reduction on body weight diminished over time in the Hooper et al. meta-analysis and that the analysis of long-term cohort studies did not support their conclusion, the very small effect in the Hooper et al. meta-analysis is likely to be a non-specific effect due to differential intervention intensity and participant compliance between the low-fat and control groups.

Secular U.S. trends in the last three decades demonstrate a correlation between substantial decreases in the percentage of dietary energy from total fat and the obesity epidemic. Large cohort studies have found that types of fat are more important than the total amount of fat in determining risk of weight gain11 and greater increase in waist circumference.12 The different associations with specific types of fat may reflect different biological actions of these fats on insulin resistance and adipose tissue. In addition, prospective cohort studies and RCTs alike have demonstrated that replacing unhealthy fats (i.e. saturated and trans fats) with polyunsaturated fats, rather than lowering the total amount of fat, is associated with a significant reduction in cardiovascular events,13 consistent with the effects on blood lipids.14

In dietary intervention studies and dietary practice, a reduction in fat typically results in compensatory increase in carbohydrates. Mounting evidence indicates that high consumption of refined carbohydrates and added sugar is associated with increased risk of weight gain, type 2 diabetes, and cardiovascular disease.15 Responding to readers’ comments on the meta-analysis, Hooper suggested “When we cut down on the amount of fat we eat from animal sources and very processed foods (saturated and trans fats), we can eat a bit more starchy food (bread, pasta, potatoes, rice, cereals) to make up the energy.” This recommendation can actually do more harm than good in the general population, especially in the setting of decreasing physical activity and increasing insulin resistance. These starchy foods are rapidly digestible and absorbable and typically have high glycemic index and glycemic load. Increasing consumption of refined carbohydrates such as white bread, white rice, and potatoes has been associated with greater weight gain and increased risk of type 2 diabetes,16-17 while a recent trial has demonstrated that reducing the glycemic index of a diet confers significant beneficial effects on energy metabolism and cardiovascular risk factors.18

In summary, the results from the current meta-analysis provide no compelling evidence for recommending a low-fat diet for the prevention of weight gain and obesity in the general population. A low-fat approach may actually be deleterious if types of fat and types of replacement macronutrients are not carefully considered, as was demonstrated by the escalating obesity and diabetes epidemics in the midst of the low-fat high-carbohydrate campaign of the past decades in the US. The overall body of scientific evidence clearly demonstrates that dietary recommendations should focus not on lowering the total fat content of the diet but rather on specific types of fats and carbohydrates and, more importantly, on specific foods and overall dietary patterns.


1. Hooper L, Abdelhamid A, Moore HJ, Douthwaite W, Skeaff CM, Summerbell CD. Effect of reducing total fat intake on body weight: systematic review and meta-analysis of randomised controlled trials and cohort studies. BMJ 2012; 345:e7666.
2. Hession M, Rolland C, Kulkarni U, Wise A, Broom J. Systematic review of randomized controlled trials of low-carbohydrate vs. low-fat/low-calorie diets in the management of obesity and its comorbidities. Obes Rev 2009;10(1):36-50.
3. Sacks FM, Bray GA, Carey VJ, Smith SR, Ryan DH, Anton SD, et al. Comparison of weight-loss diets with different compositions of fat, protein, and carbohydrates. N Engl J Med 2009; 360(9):859-73.
4. Gardner CD, Kiazand A, Alhassan S, Kim S, Stafford RS, Balise RR, et al. Comparison of the Atkins, Zone, Ornish, and LEARN diets for change in weight and related risk factors among overweight premenopausal women: the A TO Z Weight Loss Study: a randomized trial. JAMA 2007; 297(9):969-77.
5. Foster GD, Wyatt HR, Hill JO, Makris AP, Rosenbaum DL, Brill C, et al. Weight and metabolic outcomes after 2 years on a low-carbohydrate versus low-fat diet: a randomized trial. Ann Intern Med 2010; 153(3):147-57.
6. Shai I, Schwarzfuchs D, Henkin Y, Shahar DR, Witkow S, Greenberg I, et al. Weight loss with a low-carbohydrate, Mediterranean, or low-fat diet.; Dietary Intervention Randomized Controlled Trial (DIRECT) Group. N Engl J Med 2008; 359(3):229-41.
7. Schwarzfuchs D, Golan R, Shai I. Four-year follow-up after two-year dietary interventions. N Engl J Med 2012;367(14):1373-4.
8. Howard BV, Manson JE, Stefanick ML, Beresford SA, Frank G, Jones B, et al. Low-fat dietary pattern and weight change over 7 years: the Women's Health Initiative Dietary Modification Trial. JAMA 2006; 295(1):39-49.
9. Al Khalaf MM, Thalib L, Doi SA. Combining heterogenous studies using the random-effects model is a mistake and leads to inconclusive meta-analyses. J Clin Epidemiol 2011; 64(2):119-23.
10. Stefanick ML, Mackey S, Sheehan M, Ellsworth N, Haskell WL, Wood PD. Effects of diet and exercise in men and postmenopausal women with low levels of HDL cholesterol and high levels of LDL cholesterol. N Engl J Med 1998;339(1):12-20.
11. Field AE, Willett WC, Lissner L, Colditz GA. Dietary fat and weight gain among women in the Nurses' Health Study. Obesity (Silver Spring) 2007; 15(4):967-76.
12. Koh-Banerjee P, Chu NF, Spiegelman D, Rosner B, Colditz G, Willett W, et al. Prospective study of the association of changes in dietary intake, physical activity, alcohol consumption, and smoking with 9-y gain in waist circumference among 16 587 US men. Am J Clin Nutr 2003; 78(4):719-27.
13. 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(3):e1000252.
14. Mensink RP, Zock PL, Kester AD, Katan MB. Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: a meta-analysis of 60 controlled trials. Am J Clin Nutr 2003; 77(5):1146-55.
15. Livesey G, Taylor R, Hulshof T, Howlett. Glycemic response and health--a systematic review and meta-analysis: relations between dietary glycemic properties and health outcomes. Am J Clin Nutr 2008; 87(1):258S-268S.
16. Mozaffarian D, Hao T, Rimm EB, Willett WC, Hu FB. Changes in diet and lifestyle and long-term weight gain in women and men. N Engl J Med 2011; 364(25):2392-404.
17. Hu EA, Pan A, Malik V, Sun Q. White rice consumption and risk of type 2 diabetes: meta-analysis and systematic review. BMJ 2012; 15;344:e1454.
18. Ebbeling CB, Swain JF, Feldman HA, Wong WW, Hachey DL, Garcia-Lago E, et al. Effects of dietary composition on energy expenditure during weight-loss maintenance. JAMA 2012; 307(24):2627-34.

Competing interests: None declared

Frank Hu, Professor

Deirdre K. Tobias, Dariush Mozaffarian, Frank Sacks, Walter Willett

Harvard School of Public Health, 665 Huntington Ave

Click to like:

The recent article by Hooper et al [1] addresses a critical question regarding the increasing rates of obesity worldwide: what role does dietary fat play in the regulation of body weight?

By analyzing only trials in which participants were randomized to a low-fat diet for outcomes other than weight reduction, the authors suggest that this will avoid any confounding by the conscious reduction of total calorie intake. They identify 33 randomised controlled trials (RCTs) that met Cochrane Collaboration criteria, and then conclude that reducing fat intake will cause small, long-term reductions in body weight (1.6 kg), body mass index and waist circumference. The “remarkably consistent” nature of this finding, the authors conclude, suggests that populations should indeed be advised to reduce fat consumed to below 30 percent of total calories.

A close analysis of the RCTs in this meta-analysis, however, reveals a common bias that leads us to question this conclusion. In at least 20 of the 33 trials – constituting greater than 95 per cent of all the participants – the researchers compared an intervention that included significant counseling to consume a low-fat diet (with the implication that this diet will make them healthier) to a usual diet intervention with little or, most often, no counseling.

The Women’s Health Initiative with almost 49,000 participants – 79 percent of all those in the Hooper et al analysis – is an ideal example.[2] The women assigned to the low-fat diet had 18 group sessions with a dietitian in the first year; four sessions in each year that followed. These group activities were also supplemented “by an intervention protocol consisting of 3 individual interviews that used reflective listening techniques…” The controls were given a copy of Dietary Guidelines for Americans and other health-related material, and that was all.

This disparity in counseling and support between control and intervention groups is known as performance bias. The Cochrane Collaboration defines it as “systematic differences between groups in the care that is provided, or in exposure to factors other than the interventions of interest.”[3] Hooper et al note that “sensitivity analysis assessed the effects of… trials not free of systematic differences in care,” implying that they could rule out the effect of this performance bias on the end result. It is not clear, though, how they could assume with any confidence that performance bias was not responsible for the results obtained.

Neither randomisation nor blinding can eliminate or control for performance bias and, as the authors note, virtually all of these trials were not blinded in any case. The overall effect is arguably the equivalent of a drug trial that not only forgoes blinding but a placebo control as well.

Whether the small reduction in weight observed in these RCTs was caused by the low-fat diet or some unintended consequence of the counseling is something that cannot be established from such uncontrolled trials. Nor can it be determined whether counseling participants to eat a different dietary prescription entirely – a low-carbohydrate diet, say, or low-glycemic index diet, or a low-sugar diet, or a Mediterranean diet -- would have led to greater weight loss and offer a more effective public health strategy to prevent obesity.

The only way to address these questions would be to indeed do, as Hooper et al suggest, “high quality trials,” but to be high-quality they would have to randomize participants to control and intervention groups that receive identical care. Only such well-controlled trials can tell us if counseling the public once again to consume less fat – as has been the advice in the U.S. and the U.K. since the 1980s, coincident with rising obesity rates in both countries -- will help prevent obesity.


[1] Hooper L, Abdelhamid A, Moore H, et al. Effect of reducing total fat intake on body weight: systematic review and meta-analysis of randomised controlled trials and cohort studies. BMJ 2012; 345:e7666

[2] Howard BV, Manson JE, Stefanick ML, et al. Low-fat dietary pattern and weight change over 7 years: the Women’s Health Initiative Dietary Modification Trial. JAMA 2006;295:39-49.

[3] Higgins JP, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomized trials. BMJ 2011;343:d5928.

Competing interests: None declared

Gary A. Taubes, journalist

Peter Attia, physician

Nutrition Science Initiative, 6020 Cornerstone Ct. W. San Diego, CA 92121

Click to like:

Dear Dhastagir, Johannes, Uffe and Stephen,

Thank you all for your interesting discussion on our research!

Dhastagir – thank you, and yes, this about our culture of food, isn’t it?

Johannes – perhaps I should stress that we are not talking about weight reducing diets here, our paper is about how we eat permanently. If we ensure that how we eat day to day, year to year is a bit lower in fat and that the remaining fat is modified (so, we cut down on saturated and trans fats) permanently we know that this will reduce our weight and protect us from cardiovascular events. And no, we have no competing interests in this.

If advice to eat a lower fat diet is so bad, as some letters suggest, one would expect the results of the RCTs to show considerable weight gain on the lower fat diets as well as deterioration of lipids and blood pressure. This was not the case – maybe we should ask, "why?". To dismiss the result as irrelevant because it does not agree with our preconceived notions serves no purpose in solving the problem of obesity. Our meta-analysis provides important information based on "real" long term clinical trials. Maybe as scientists we should think about its implications for population nutritional recommendations. Evidently, people who switch to a lower fat diet make changes to their diet that help control body weight and other CVD risk factors. Maybe it is unfair to assume that those switching to lower fat diets increase their energy intake by increasing consumption of high GI foods loaded with free sugars. Of course it is important to know what is best to replace the fat, but our paper cannot answer this question.

Uffe – we did not include the study by Cornier as it aimed to reduce the weight of 21 heavy women (and was also too short, running for only 16 weeks). We are not reporting on how well weight loss trials work, or which work better – we are looking at long term eating patterns (which was why we excluded all studies that aimed to reduce weight in any of the participants).

Stephen – yes, the studies varied in exactly what they replaced the fat calories with. Generally there did seem to be an increase in carbohydrates across the studies, but it was usually not clear whether this was due to increases in refined or less refined carbohydrates, and the effect on protein did vary. But despite this the weight loss was remarkably consistent, and was greater when the reduction in fat was greater.

We agree with you all that the main dietary message for everyday eating for people and populations is not going to be “cut down on fat”. Of course not. However, there is VERY good evidence that it should be part of the message, and as Stephen suggests maybe the issue is what we begin to replace the saturated and trans fats we cut down on with.

Best wishes to all of you for the festive season, Lee and Murray (on behalf of all the authors)

Competing interests: None declared

Lee Hooper, Senior Lecturer in Research Synthesis and Nutrition

C Murray Skeaff

Norwich Medical School, University of East Anglia, Norwich NR4 7TJ, UK

Click to like:

I notice that the reduced fat trial had a negative or insignificant effect on HDL and no effect on triglycerides.

So the reduced fat approach does nothing to help with two of the important risk factors relating to CVD, high trigs being an independent risk factor and low HDL also playing a negative role.

Have you considered trialling a low carbohydrate/ moderate protein / high fat diet to see the effect that has on trigs and HDL? A number of studies have shown that when carbohydrate is reduced, triglycerides plummet, and HDL increases with fat intake. Could it be the carbohydrate that is the danger factor here rather than the fat?

Low carbohydrate appears to have a positive effect on LDL particle size, reducing Pattern B in favour of the non-atherogenic Pattern A LDL.

Additionally, this section seems rather vague, we actually have little idea whether CHO consumption rose or fell during the studies.

Your text:

Effect of reducing fat on intake of energy, carbohydrate, sugars, protein, and alcohol
Indications were that during the diet periods energy intake was usually lower in the low fat group than in the control or usual fat groups, and sugar intake was not measured often but where reported intakes seemed to be higher in the low fat arms (except in MeDiet33, see supplementary table 4). Carbohydrate intakes seemed almost universally higher in the low fat arms than in the usual fat arms, and protein intakes were sometimes higher and sometimes similar. There was no consistent pattern in alcohol intake. When subgroups were used to examine the effect of degree of energy reduction in the reduced fat group compared with the control group on weight loss, there was some indication that a greater degree of energy reduction in the reduced fat group was associated with greater weight reduction (see table 2, test for subgroup differences P=0.04, forest plot displayed in supplementary fig 4).

Any feedback would be welcomed.

Competing interests: None declared

Stephen Reed, Personal Trainer

none, Netherbury, DORSET

Click to like:

A misleading meta-analysis

In their meta-analysis1 Hooper et al have ignored the results from the study by Cornier et al.2 These authors found that insulin-resistent women lost more weight on a lowcarb/highfat diet, whereas insulin-sensitive women lost more weight on a highcarb/lowfat diet, and the differences were highly significant (p<0.001).2 This means that without separating the participants according to their degree of insulin sensitivity, it is impossible to draw any conclusion about specific effect of the macronutrients from a trial aimed at weight loss.

The benefit of a low-carb/high-fat diet without calorie restriction in insulin-resistant patients or patients with manifest type-2 diabetes has been demonstrated in at least six experiments (table 1)3-8. Although the fat intake covered about fifty per cent of the total calorie intake, these experiments lead to a much greater weight loss than was observed in the trials reviewed by Hooper et al.

Sceptics probably object that such high fat intakes may worsen the lipid profile. However, there were only insignificant changes of total and LDL cholesterol in these experiments; in four of them HDL increased significantly and in all of them triglycerides decreased significantly.

1. Hooper L, Abdelhamid A, Moore HJ, Douthwaite W, Skeaff CM, Summerbell CD. Effect of reducing total fat intake on body weight: systematic review and meta-analysis of randomised controlled trials and cohort studies. BMJ 2012;345:e7666 doi: 10.1136/bmj.e7666
2. Cornier MA, Donahoo WT, Pereira R, Gurevich I, Westergren R, Enerback S et al. Insulin Sensitivity Determines the Effectiveness of Dietary Macronutrient Composition on Weight Loss in Obese Women. Obes Res 2005;13:703-9
3. Stern L, Iqbal N, Seshadri P, Chicano KL, Daily DA, McGrory J et al. The effects of low-carbohydrate versus conventional weight loss diets in severely obese adults: one-year follow-up of a randomized trial. Ann Intern Med 2004;140:778-85.
4. Vernon MC, Mavropoulos J, Transue M, Yancy WS, Westman EC. Clinical experience of a carbohydrate-restricted diet: effect on diabetes mellitus. Metab Syndr Relat Disord. 2003;1:233-7.
5. Yancy WS, Vernon MC, Westman EC. A pilot trial of a low-carbohydrate, ketogenic diet in patients with type 2 diabetes. Metab Syndr Relat Disord. 2003;1:239-43.
6. Yancy WS Jr, Foy M, Chalecki AM, Vernon MC, Westman EC. A low-carbohydrate, ketogenic diet to treat type 2 diabetes. Nutr Metab 2005;2:34.
7. Westman EC, Yancy WS Jr, Mavropoulos JC, Marquart M, McDuffie JR. The effect of a low-carbohydrate, ketogenic diet versus a low-glycemic index diet on glycemic control in type 2 diabetes mellitus. Nutr Metab 2008;5:36.
8. Volek JS, Phinney SD, Forsythe CE, Quann EE, Wood RJ, Puglisi MJ et al. Carbohydrate restriction has a more favorable impact on the metabolic syndrome than a low fat diet. Lipids 2009;44:297-309.

Competing interests: None declared

Uffe Ravnskov, MD, PhD, retired specialist in internal medicine and nephrology

independent researcher, Magle Stora Kyrkogata 9, 22350 Lund, Sweden

Click to like:

12 December 2012

Dear Lee,

I am sorry, but in your discussion you are missing the most important point I have made.

Imagine: If there was a drug XYZ, that lowers LDL-cholesterol by 10% and CHD by let´s say 15%, and which was proven to be effective in a meta-analysis of RCTs: Who then would prescribe this drug XYZ, if he knew there were potent statins, that lower LDL by 50% and CHD by >30%? Wouldn´t you always look for the best option?

Why then promoting “low-fat”? Why ignoring better dietary strategies in your whole discussion? This question has not been answered in this discussion yet.

People are not eating „fat“ nor „carbs“ alone, they are eating „meals“. We must offer them the best advice for mixed meals, not only to lose weight, but furthermore to lower their risk cardiovascular disease and diabetes. The best advice is: low energy density, low glycemic load, more protein and moderate fat, but good fat quality. Of course this has to be translated into the patients´s language, which we do in our prevention clinic every day (www.preventionfirst.de).

Even if “strong evidence” is showing a 1,6 kg weight-loss with a low-fat diet, this MUST NOT become a public health message, if plenty of other evidence is showing, that low-fat is doing worse than Mediterranean diets with moderate carbohydrate reduction and low-carb diets like (modern) Atkins diet, especially in insulin-resistant patients. “Eating more starchy foods” for them is detrimental, because it will lead to hyperinsulinemia, ß-cell-exhaustion, fatty liver, lower HDL-cholesterol, higher blood pressure and so on...

On the other hand, it´s billions of Euros and Dollars that are at stake in this debate, so everyone reading this discussion should be aware, that there is a very strong lobbyism from BIG SUGAR. They seem to as effective in brain-washing as BIG TOBACCO in the last century.

I personally have no conflicts of interest, and I have never been lecturing for any of the involved parties (unlike many of our so called “experts” in the field of nutrition).

Many doctors who are taking care for obese and/or diabetic patients will go for low-carb, because they see the immediate benefits in terms of weight loss, HbA1c, insulin demand and the much lower prescription rate for oral antidiabetic and antihypertensive drugs, which saves costs.

And, dear colleaguse, if you still have doubts – just try out low-carb for yourself and ignore the outdated recommendations!

Dr. med. Johannes Scholl
President of the Deutsche Akademie für Präventivmedizin e.V. (German Academy for Preventive Medicine)

Competing interests: None declared

Johannes G. Scholl, Physician

Prevention First Institut, Europastr. 10, 65385 Ruedesheim am Rhein

Click to like:

With scientific evidence the author says that if we cut down the amount of fat in the animal fat we eat and eat more starchy food it will help reduce our weight. Reduction in our weight even if it is small could be beneficial when such a diet is followed over a period of time. The message is cut your fat in the diet and see that your weight is controlled systematically.

In a fast food culture where people eat more animal fat in their burgers (sometimes people remove all of the salad served along with it) it may prove healthy. People who live in a corporate world who rely on such fast foods with less physical activity need to take notice of cutting down fat in the diet or increasing salad in their diet. In a mad mad mad world when people are in a hurry with a burger and carbonated drink in hand - we have little time to measure our calories and composition of our diet.

Yet the study has made everyone aware that fat in the diet needs to be taken care of.

Competing interests: None declared

dhastagir s sheriff, Professor

Faculty of Medicine, Benghazi University, Benghazi, Libya

Click to like: