The American Heart Association (AHA) released it's Presidential Advisory on Dietary Fats and Cardiovascular Disease this week and the responses have been all over the place (see medscape, science-based medicine, HNR, Gary Taubes, Cardiobrief). Here's my summary/take:
There's nothing terribly new about this advisory; it's relatively the same line of thinking/evidence that the AHA has used to support it's recommendations in the past, so regular followers of the field will not read this report and be too shocked. The most novel aspect of it is an updated meta-analysis of 4 diet-heart trials which the AHA generated. The new meta analysis comes about because there have been multiple meta-analyses which have attempted to address this issue over the past several years, and they don't all align nicely - see here, here, here . Why? Because the study inclusion criterion that are used are often not the same. Shorter term trials have consistently shown that saturated fatty acids raise LDL-cholesterol, and monounsaturated and polyunsaturated fats lower LDL-cholesterol. Attempts to translate this effect on LDL to cardiovascular disease outcomes have been done in the past, often taking advantage of confined populations (e.g. hospitals, mental homes), and are of varying quality. In their rendition, the AHA puts forth some more rigorous inclusion criteria regarding the duration of the trial, confirmation of LDL lowering, and measures of adherence to the intervention diet; this cuts down the number of trials to 4, a substantial difference from previous meta-analyses. The criteria are reasonable and do eliminate some problematic trials which are pretty poor tests of the diet-lipid hypothesis.
The problem with all of the diet-heart trials, regardless of threshold criterion, is that they aren't pure fat replacement trials (replacement is key here because fat has calories, so to control for alterations in the caloric load, you've got to replace saturated fats with calories from either MUFAs, PUFAs or carbohydrates). One could (and many have) spend(t) pages narratively describing what we think happened in these trials and the clearest picture is that 1) PUFA (n6/n3) rich oils tended to replace SFAs; 2) outside of this, these old trials are messy as hell.
Trials that the AHA excluded were of a short duration, altered carbohydrate intake as well as fats, had low event rates and/or adherence issues. To give an example, the Minnesota Coronary Experiment was excluded. Like many trials from this era, it used a convenience population (i.e.mental hospitals); AHA excluded it because the intervention only lasted on average about 1 year, due to a very high discharge rate, which substantially reduced its power . While not noted by the AHA, the trial had fun little findings such as smokers being the least likely to die (see page 60 of this pdf). The AHA had good reasoning for eliminating the trials that they eliminated but they get into a bit of trouble with the 4 trials that they included. Their descriptions of them are quite lacking; although the 4 trials are better and meet the AHA's basic criteria, they are far from perfect. For example, the Finnish Mental Hospital Study was included in the meta analysis. This trial wasn't truly randomized (it was cluster randomized but only had 2 clusters) and has been excluded from previous Cochrane reviews. Within hospitals, there was a high dropout rate (only about 50 percent retention) and the study's use of a cross over design is considered not ideal for a disease that develops over years; additional concerns have been raised about the amounts of trans fats in the control diets (would likely inflate the intervention's effect size), and differences between groups in cardiotoxic medications and the sugar content of the diet. The Oslo trial was also included, and this intervention (based on self-report from a very small subgroup from the trial) led to numerous dietary changes, including a sizable increase in fish consumption and likely increases in fruit/vegetables/whole grains and reduced trans fat intakes. Furthermore, food was provided in Oslo, whereas the control group purchased and consumed their own food, leaving an imbalance between the intervention and control groups' level of interaction with the study. The effects of all of these changes on the outcomes observed are nearly impossible to quantify since many weren't adequately measured.
What was measurable in many of these trials was the change in serum cholesterol (not all trials at this time measured cholesterol fractions). The AHA notes that the reduction in cholesterol achieved in these 4 trials is consistent with the expected reduction in events, a similar finding to the meta-regression performed in the most recent Cochrane Review on this issue. This supports the notion cholesterol reduction is a major contributor to the reduction in events and by extension, the reduction in saturated and increase in polyunsaturates explains much of the intervention effects; however, we'd expect other aspects of the intervention arm (such as an increase in fish intake) would be highly correlated with this reduction in cholesterol; since many weren't measured, we can't take these into account in multivariate analyses. Note, however, that the majority of co-interventions that occurred do contribute to the same causal pathway of reduced ApoB containing lipoproteins (e.g. fiber and fish lower triglycerides, trans fats and cholesterol lowering also lower LDL). Despite these considerations, rigorous statistics can never adequately account for issues with the intervention delivery and improper randomization.
The controversial aspects of this report largely relate to the AHA's inadequate discussion of these limitations. While the AHA focused heavily on the limitations of what they excluded, they could've stood to be as critical with the trials they included in the meta-analysis. The only major limitations I noted from the AHA were concomitant reductions in dietary cholesterol and the fact that cholesterol levels were very high at baseline in these trials, given that these studies occurred before the widespread use of statins. I was a bit uncomfortable reading their descriptions and was waiting for the disgruntled responses. They arrived.
Other aspects of the advisory were a bit too casual/certain for me - the frank assertion by the AHA that the Sydney Heart Study's PUFA intervention arm contained a high level of trans fats which account for the increased risk of deaths is likely, but it's not quite certain since trans-fats weren't measured by the investigators.The effect this had on the results is not quantifiable now, but the AHA seems certain that this is a major factor in the study. Note that there are other limitations to this study that could've been highlighted beyond this assertion, given the changes in body weight, smoking cessation, and physical activity that occurred during Sydney. Their assertion is a microcosm of the whole report - where the AHA sees certainty, many others see shades of (suggestive) probably/maybe/grey.
Why the AHA didn't more rigorously critique the trials included in their meta-analysis is open to speculation. I'm quite sure that the members of the AHA advisory knew of the limitations of these trials; I have spoken to some members in person and have seen their own publications where they note similar limitations. The failure to note these limits makes it hard to defend accusations that this is a lot more advocacy than a science-based summary of the evidence. A more rigorous approach would've included a table of the risks of bias and assessment of methodologies, as is typically done in a systematic review of the evidence. The Cochrane review of Saturated Fats and CVD rated the evidence from these old trials as 'moderate' that reductions in saturated fat yield reductions in CVD events:
The controversy surrounding saturated fats and this advisory are a symptom of the controversies surrounding all of nutrition and chronic disease. Getting past prospective cohorts, surrogate outcomes, and animal evidence is really difficult; the field is excellent at generating hypotheses and really struggles to test these with large, well-conducted randomized controlled trials. We'll likely never have a world where there is perfect adherence to a single dietary intervention, and while interventions with multiple components are nice proofs of concept that nutrition does things when it comes to chronic disease, interpreting these is clearly messy. If you lower the evidence threshold from near flawless randomized controlled trials and allow the inclusion of prospective cohort evidence, surrogate outcomes, and primate interventions, you see a pattern emerge that replacement of SFA with MUFA/PUFA lower LDL and that is associated with reductions in CVD events; the AHA considered all lines of evidence in this advisory to support the causal inference that the effects of dietary fats on LDL meaningfully impact cardiovascular disease. If you maintain an evidence threshold that is firmly rooted in demonstrated causality with hard clinical outcomes, as is typical of evidence-based medicine's standards, you can forever run interference for any nutrition recommendation ever, including this advisory and other AHA recommendations to reduce industrial trans-fatty acids, sugar, and refined grains (we have no well-designed RCTs here either). It's unsurprising that low-carbohydrate enthusiasts critical of this report haven't been quite as critical of some of these reports as they have others. I suspect the perceived 'naturalness' of the dietary recommendation at hand influences the vigor of the response heavily here, given the relative silence/lack of controversy surrounding calls to reduce trans fats and sugar despite a lack of randomized controlled trials.
The AHA's report lacks the humility one might expect from the field of nutrition; there are a lot of shades of 'maybe' and 'probably' across the evidence, including recommendations outside of the realm of chronic disease (i.e. preventing nutrient deficiencies). What's most concerning to me is that the fatty acid links to CVD are probably the most evolved body of evidence (sodium is a contender) that we have linking nutrition to chronic disease, in terms of having many controlled trials with multiple surrogates, non-human primate evidence, multiple prospective cohorts, and attempted trials with hard outcomes. Every other link is based on even further logical jumps and assumptions about the data. After looking at these older trials, and even more recent trials such as the Women's Health Initiative, I don't expect funding agencies to be throwing money at large dietary interventions with hard outcomes any time soon, and I imagine this plays into why some individuals in the field haven't just completely disregarded these old trials.
Expect to see varied opinions on what this all means for nutrition policy. To some, the critics and the dairy industry's responses to the evidence linking fatty acids to CVD will be another example of doubt brokering; to others, the critics will be seen as stalwarts of evidence-based medicine. Until the field of nutrition establishes solid criteria for the level of evidence required to become policy, nutrition and chronic disease will be continue to be a hot mess.
Other random things since I know i'll get comments:
1) The AHA made some commentary on coconut oil; health gurus hadn't told folks that coconut oil is amazing back in the 60s and 70s so most folks didn't consume high amounts of it. Consequently, these trials didn't test the effects of coconut oil on CVD outcomes. AHA recommends against it due to the effects on LDL. Coconut oil has largely gotten a free pass because it contains about half lauric acid, a 12 carbon saturated fat that often gets called a medium chain triglyceride, even though we clinically think of those as C8 and C10 fatty acids. Whether lauric is actually unique in its effects on CVD due to its ability to potently increase HDL-C levels remains to be seen, and further work will need to consider it's net impact on HDL activity/LDL relative to other fats. We'll likely never have a large trial for this, and you'd get big debates about whether to use extra virgin or refined coconut oils. Based on surrogates, coconut oil is meh. Based on outcomes, coconut oil is ???.
2) A lot of the controversy around saturated fats has come from analyses where self-reported saturated fat intake presented as a proportion of self-reported calorie intake doesn't independently predict cardiovascular diseases. There are a number of conceivable reasons why one dietary factor which impacts one risk factor doesn't independently predict CVD risk. The most recent meta-analysis of this observational data stated that "the certainty of the estimates for the association between saturated fats and all outcomes was very low". This meta-analysis discussed other analyses which showed benefits from replacement but did not itself model the effects of replacement; the focus of the AHA and nutrition-lipid folks has largely shifted to focusing on replacement, so we don't compare high saturated fat diets to high sugar/refined grain calories, which have less beneficial effects on lipid surrogates (but we still have no trials for hard outcomes on this topic).
3) Some analyses have suggested that dairy fat is unique in its effect on cardiovascular disease and AHA discussed this evidence, noting that the observational studies have not been consistent with regard to markers (C15/C17) of dairy fat exposure on CVD risk. All of the evidence suggesting a benefit has not considered replacement or comparator calories. C15 and C17's roles as markers of dairy fat exposure are questionable, as these fats are found in fish and result from elongation of propionate generated by fermentable fibers. In addition, the correlation between reported dairy fat consumption and C15/C17 is low, and is not specific to low fat or high fat dairy. We're currently lacking even surrogates and animal feeding studies in this realm to attempt to build a causal story that C15/C17/ dairy fats have some unique beneficial effects. I'm not surprised AHA isn't weighting this too heavily; it's another area where the data is quite underdeveloped and where replacement modeling/surrogates still support consumption of unsaturated fats.
4) Taubes noted that the focus of this whole conversation is on CVD events, and doesn't focus on CVD mortality or total mortality. He makes a passing comment about getting Alzheimer's or Cancer, which seems to suggest that he thinks evidence the AHA guidance will impact the risk for these in a negative fashion (i'm quite sure he's pulling on the advice to replace with linoleic acid here). The evidence linking fatty acids to all of these outcomes is extremely weak, and cohort evidence has not consistently supported these concerns. Folks may certainly weight theoretical concerns differently and find them to outweigh the moderate quality evidence suggesting a benefit but that's not how current evidence-based decision making works.
5) The concerns over saturated fats are probably largely driven by the recommendation to replace them with linoleic acid. Many in the basic sciences/lipid community are not fans of this recommendation for a lot of preclinical reasons (increased tissue susceptibility to oxidation, reduced endogenous n-3 elongation and competition for space within membranes). Replacement of saturated fats with MUFAs still reduces LDL to a lesser degree without garnering these concerns and will likely become a more common reality in an era when low PUFA/high oleic oils are taking over the market.
6) Yes this report focuses on nutrients and not foods. AHA did focus on consuming a low SFA diet to lower LDL in the context of specific 'healthy' dietary patterns, which I appreciated. As the food supply continues to alter the nutrient contents of its foods (i.e. genetic modification of high linoleic oils to be low linoleic high oleic), nutrients will continue to be important to consider, especially for nutrients which have demonstrated substantial impacts on (surrogate) outcomes. The call to focus on foods not nutrients tends to come from those most enthused with the hundreds of associations drawn between food exposures and disease outcomes in observational evidence, which probably isn't the solution to nutrition's problems here. It's notable that the issue with interpreting many dietary trials is the sheer complexity of the interventions; foods themselves contain many nutrients/bioactives which may or may not be consumed in the background diets of the control or intervention groups. It's important in trials to have a hypothesis which we are testing, not to throw 15 interventions at one group and hope something sticks. I am certainly a fan of food-based interventions whereby a food matrix effect has been demonstrated and can be isolated from the effects of its many nutrients; this is often lost in the vigorous support of 'foods not nutrients'.
There's nothing terribly new about this advisory; it's relatively the same line of thinking/evidence that the AHA has used to support it's recommendations in the past, so regular followers of the field will not read this report and be too shocked. The most novel aspect of it is an updated meta-analysis of 4 diet-heart trials which the AHA generated. The new meta analysis comes about because there have been multiple meta-analyses which have attempted to address this issue over the past several years, and they don't all align nicely - see here, here, here . Why? Because the study inclusion criterion that are used are often not the same. Shorter term trials have consistently shown that saturated fatty acids raise LDL-cholesterol, and monounsaturated and polyunsaturated fats lower LDL-cholesterol. Attempts to translate this effect on LDL to cardiovascular disease outcomes have been done in the past, often taking advantage of confined populations (e.g. hospitals, mental homes), and are of varying quality. In their rendition, the AHA puts forth some more rigorous inclusion criteria regarding the duration of the trial, confirmation of LDL lowering, and measures of adherence to the intervention diet; this cuts down the number of trials to 4, a substantial difference from previous meta-analyses. The criteria are reasonable and do eliminate some problematic trials which are pretty poor tests of the diet-lipid hypothesis.
The problem with all of the diet-heart trials, regardless of threshold criterion, is that they aren't pure fat replacement trials (replacement is key here because fat has calories, so to control for alterations in the caloric load, you've got to replace saturated fats with calories from either MUFAs, PUFAs or carbohydrates). One could (and many have) spend(t) pages narratively describing what we think happened in these trials and the clearest picture is that 1) PUFA (n6/n3) rich oils tended to replace SFAs; 2) outside of this, these old trials are messy as hell.
Trials that the AHA excluded were of a short duration, altered carbohydrate intake as well as fats, had low event rates and/or adherence issues. To give an example, the Minnesota Coronary Experiment was excluded. Like many trials from this era, it used a convenience population (i.e.mental hospitals); AHA excluded it because the intervention only lasted on average about 1 year, due to a very high discharge rate, which substantially reduced its power . While not noted by the AHA, the trial had fun little findings such as smokers being the least likely to die (see page 60 of this pdf). The AHA had good reasoning for eliminating the trials that they eliminated but they get into a bit of trouble with the 4 trials that they included. Their descriptions of them are quite lacking; although the 4 trials are better and meet the AHA's basic criteria, they are far from perfect. For example, the Finnish Mental Hospital Study was included in the meta analysis. This trial wasn't truly randomized (it was cluster randomized but only had 2 clusters) and has been excluded from previous Cochrane reviews. Within hospitals, there was a high dropout rate (only about 50 percent retention) and the study's use of a cross over design is considered not ideal for a disease that develops over years; additional concerns have been raised about the amounts of trans fats in the control diets (would likely inflate the intervention's effect size), and differences between groups in cardiotoxic medications and the sugar content of the diet. The Oslo trial was also included, and this intervention (based on self-report from a very small subgroup from the trial) led to numerous dietary changes, including a sizable increase in fish consumption and likely increases in fruit/vegetables/whole grains and reduced trans fat intakes. Furthermore, food was provided in Oslo, whereas the control group purchased and consumed their own food, leaving an imbalance between the intervention and control groups' level of interaction with the study. The effects of all of these changes on the outcomes observed are nearly impossible to quantify since many weren't adequately measured.
What was measurable in many of these trials was the change in serum cholesterol (not all trials at this time measured cholesterol fractions). The AHA notes that the reduction in cholesterol achieved in these 4 trials is consistent with the expected reduction in events, a similar finding to the meta-regression performed in the most recent Cochrane Review on this issue. This supports the notion cholesterol reduction is a major contributor to the reduction in events and by extension, the reduction in saturated and increase in polyunsaturates explains much of the intervention effects; however, we'd expect other aspects of the intervention arm (such as an increase in fish intake) would be highly correlated with this reduction in cholesterol; since many weren't measured, we can't take these into account in multivariate analyses. Note, however, that the majority of co-interventions that occurred do contribute to the same causal pathway of reduced ApoB containing lipoproteins (e.g. fiber and fish lower triglycerides, trans fats and cholesterol lowering also lower LDL). Despite these considerations, rigorous statistics can never adequately account for issues with the intervention delivery and improper randomization.
The controversial aspects of this report largely relate to the AHA's inadequate discussion of these limitations. While the AHA focused heavily on the limitations of what they excluded, they could've stood to be as critical with the trials they included in the meta-analysis. The only major limitations I noted from the AHA were concomitant reductions in dietary cholesterol and the fact that cholesterol levels were very high at baseline in these trials, given that these studies occurred before the widespread use of statins. I was a bit uncomfortable reading their descriptions and was waiting for the disgruntled responses. They arrived.
Other aspects of the advisory were a bit too casual/certain for me - the frank assertion by the AHA that the Sydney Heart Study's PUFA intervention arm contained a high level of trans fats which account for the increased risk of deaths is likely, but it's not quite certain since trans-fats weren't measured by the investigators.The effect this had on the results is not quantifiable now, but the AHA seems certain that this is a major factor in the study. Note that there are other limitations to this study that could've been highlighted beyond this assertion, given the changes in body weight, smoking cessation, and physical activity that occurred during Sydney. Their assertion is a microcosm of the whole report - where the AHA sees certainty, many others see shades of (suggestive) probably/maybe/grey.
Why the AHA didn't more rigorously critique the trials included in their meta-analysis is open to speculation. I'm quite sure that the members of the AHA advisory knew of the limitations of these trials; I have spoken to some members in person and have seen their own publications where they note similar limitations. The failure to note these limits makes it hard to defend accusations that this is a lot more advocacy than a science-based summary of the evidence. A more rigorous approach would've included a table of the risks of bias and assessment of methodologies, as is typically done in a systematic review of the evidence. The Cochrane review of Saturated Fats and CVD rated the evidence from these old trials as 'moderate' that reductions in saturated fat yield reductions in CVD events:
 |
| Hooper et al, 2015; noting limitations of the available trial evidence |
The controversy surrounding saturated fats and this advisory are a symptom of the controversies surrounding all of nutrition and chronic disease. Getting past prospective cohorts, surrogate outcomes, and animal evidence is really difficult; the field is excellent at generating hypotheses and really struggles to test these with large, well-conducted randomized controlled trials. We'll likely never have a world where there is perfect adherence to a single dietary intervention, and while interventions with multiple components are nice proofs of concept that nutrition does things when it comes to chronic disease, interpreting these is clearly messy. If you lower the evidence threshold from near flawless randomized controlled trials and allow the inclusion of prospective cohort evidence, surrogate outcomes, and primate interventions, you see a pattern emerge that replacement of SFA with MUFA/PUFA lower LDL and that is associated with reductions in CVD events; the AHA considered all lines of evidence in this advisory to support the causal inference that the effects of dietary fats on LDL meaningfully impact cardiovascular disease. If you maintain an evidence threshold that is firmly rooted in demonstrated causality with hard clinical outcomes, as is typical of evidence-based medicine's standards, you can forever run interference for any nutrition recommendation ever, including this advisory and other AHA recommendations to reduce industrial trans-fatty acids, sugar, and refined grains (we have no well-designed RCTs here either). It's unsurprising that low-carbohydrate enthusiasts critical of this report haven't been quite as critical of some of these reports as they have others. I suspect the perceived 'naturalness' of the dietary recommendation at hand influences the vigor of the response heavily here, given the relative silence/lack of controversy surrounding calls to reduce trans fats and sugar despite a lack of randomized controlled trials.
The AHA's report lacks the humility one might expect from the field of nutrition; there are a lot of shades of 'maybe' and 'probably' across the evidence, including recommendations outside of the realm of chronic disease (i.e. preventing nutrient deficiencies). What's most concerning to me is that the fatty acid links to CVD are probably the most evolved body of evidence (sodium is a contender) that we have linking nutrition to chronic disease, in terms of having many controlled trials with multiple surrogates, non-human primate evidence, multiple prospective cohorts, and attempted trials with hard outcomes. Every other link is based on even further logical jumps and assumptions about the data. After looking at these older trials, and even more recent trials such as the Women's Health Initiative, I don't expect funding agencies to be throwing money at large dietary interventions with hard outcomes any time soon, and I imagine this plays into why some individuals in the field haven't just completely disregarded these old trials.
Expect to see varied opinions on what this all means for nutrition policy. To some, the critics and the dairy industry's responses to the evidence linking fatty acids to CVD will be another example of doubt brokering; to others, the critics will be seen as stalwarts of evidence-based medicine. Until the field of nutrition establishes solid criteria for the level of evidence required to become policy, nutrition and chronic disease will be continue to be a hot mess.
Other random things since I know i'll get comments:
1) The AHA made some commentary on coconut oil; health gurus hadn't told folks that coconut oil is amazing back in the 60s and 70s so most folks didn't consume high amounts of it. Consequently, these trials didn't test the effects of coconut oil on CVD outcomes. AHA recommends against it due to the effects on LDL. Coconut oil has largely gotten a free pass because it contains about half lauric acid, a 12 carbon saturated fat that often gets called a medium chain triglyceride, even though we clinically think of those as C8 and C10 fatty acids. Whether lauric is actually unique in its effects on CVD due to its ability to potently increase HDL-C levels remains to be seen, and further work will need to consider it's net impact on HDL activity/LDL relative to other fats. We'll likely never have a large trial for this, and you'd get big debates about whether to use extra virgin or refined coconut oils. Based on surrogates, coconut oil is meh. Based on outcomes, coconut oil is ???.
2) A lot of the controversy around saturated fats has come from analyses where self-reported saturated fat intake presented as a proportion of self-reported calorie intake doesn't independently predict cardiovascular diseases. There are a number of conceivable reasons why one dietary factor which impacts one risk factor doesn't independently predict CVD risk. The most recent meta-analysis of this observational data stated that "the certainty of the estimates for the association between saturated fats and all outcomes was very low". This meta-analysis discussed other analyses which showed benefits from replacement but did not itself model the effects of replacement; the focus of the AHA and nutrition-lipid folks has largely shifted to focusing on replacement, so we don't compare high saturated fat diets to high sugar/refined grain calories, which have less beneficial effects on lipid surrogates (but we still have no trials for hard outcomes on this topic).
3) Some analyses have suggested that dairy fat is unique in its effect on cardiovascular disease and AHA discussed this evidence, noting that the observational studies have not been consistent with regard to markers (C15/C17) of dairy fat exposure on CVD risk. All of the evidence suggesting a benefit has not considered replacement or comparator calories. C15 and C17's roles as markers of dairy fat exposure are questionable, as these fats are found in fish and result from elongation of propionate generated by fermentable fibers. In addition, the correlation between reported dairy fat consumption and C15/C17 is low, and is not specific to low fat or high fat dairy. We're currently lacking even surrogates and animal feeding studies in this realm to attempt to build a causal story that C15/C17/ dairy fats have some unique beneficial effects. I'm not surprised AHA isn't weighting this too heavily; it's another area where the data is quite underdeveloped and where replacement modeling/surrogates still support consumption of unsaturated fats.
4) Taubes noted that the focus of this whole conversation is on CVD events, and doesn't focus on CVD mortality or total mortality. He makes a passing comment about getting Alzheimer's or Cancer, which seems to suggest that he thinks evidence the AHA guidance will impact the risk for these in a negative fashion (i'm quite sure he's pulling on the advice to replace with linoleic acid here). The evidence linking fatty acids to all of these outcomes is extremely weak, and cohort evidence has not consistently supported these concerns. Folks may certainly weight theoretical concerns differently and find them to outweigh the moderate quality evidence suggesting a benefit but that's not how current evidence-based decision making works.
5) The concerns over saturated fats are probably largely driven by the recommendation to replace them with linoleic acid. Many in the basic sciences/lipid community are not fans of this recommendation for a lot of preclinical reasons (increased tissue susceptibility to oxidation, reduced endogenous n-3 elongation and competition for space within membranes). Replacement of saturated fats with MUFAs still reduces LDL to a lesser degree without garnering these concerns and will likely become a more common reality in an era when low PUFA/high oleic oils are taking over the market.
6) Yes this report focuses on nutrients and not foods. AHA did focus on consuming a low SFA diet to lower LDL in the context of specific 'healthy' dietary patterns, which I appreciated. As the food supply continues to alter the nutrient contents of its foods (i.e. genetic modification of high linoleic oils to be low linoleic high oleic), nutrients will continue to be important to consider, especially for nutrients which have demonstrated substantial impacts on (surrogate) outcomes. The call to focus on foods not nutrients tends to come from those most enthused with the hundreds of associations drawn between food exposures and disease outcomes in observational evidence, which probably isn't the solution to nutrition's problems here. It's notable that the issue with interpreting many dietary trials is the sheer complexity of the interventions; foods themselves contain many nutrients/bioactives which may or may not be consumed in the background diets of the control or intervention groups. It's important in trials to have a hypothesis which we are testing, not to throw 15 interventions at one group and hope something sticks. I am certainly a fan of food-based interventions whereby a food matrix effect has been demonstrated and can be isolated from the effects of its many nutrients; this is often lost in the vigorous support of 'foods not nutrients'.
A sobering line from the 2015 Cochrane meta-analysis:
ReplyDelete"we have not noted any new trials on the horizon and so perhaps the current evidence set is as definitive as we will achieve during the ’statin era’."
So the best we can hope for is likely more well-designed prospective cohort studies (with repeat diet assessments, diet biomarkers, high inter-individual variation in fat intake, substitution modeling, etc.) Most of the cohort studies published today are post-hoc analyses of studies with very different research questions. But then again they will have to be huge to have enough power given today's low incidence rates.