User login
You’re invited to a dinner party but you struggle to make small talk. Do not worry; that will invariably crop up over cocktails. Because all journalism has been reduced to listicles, here are four ways to seem clever at dinner parties.
1. The Predinner Cocktails: A Lesson in Reverse Causation
Wine connoisseurs sniff, swirl, and gently swish the wine in their mouths before spitting out and cleansing their palates to better appreciate the subtlety of each vintage. If you’re not an oenophile, no matter. Whenever somebody claims that moderate amounts of alcohol are good for your heart, this is your moment to pounce. Interject yourself in the conversation and tell everybody about reverse causation.
Reverse causation, also known as protopathic bias, involves misinterpreting the directionality of an association. You assume that X leads to Y, when in fact Y leads to X. Temporal paradoxes are useful plot devices in science fiction movies, but they have no place in medical research. In our bland world, cause must precede effect. As such, smoking leads to lung cancer; lung cancer doesn’t make you smoke more.
But with alcohol, directionality is less obvious. Many studies of alcohol and cardiovascular disease have demonstrated a U-shaped association, with risk being lowest among those who drink moderate amounts of alcohol (usually one to two drinks per day) and higher in those who drink more and also those who drink very little.
But one must ask why some people drink little or no alcohol. There is an important difference between former drinkers and never drinkers. Former drinkers cut back for a reason. More likely than not, the reason for this newfound sobriety was medical. A new cancer diagnosis, the emergence of atrial fibrillation, the development of diabetes, or rising blood pressure are all good reasons to reduce or eliminate alcohol. A cross-sectional study will fail to capture that alcohol consumption changes over time — people who now don’t drink may have imbibed alcohol in years past. It was not abstinence that led to an increased risk for heart disease; it was the increased risk for heart disease that led to abstinence.
You see the same phenomenon with the so-called obesity paradox. The idea that being a little overweight is good for you may appeal when you no longer fit into last year’s pants. But people who are underweight are so for a reason. Malnutrition, cachexia from cancer, or some other cause is almost certainly driving up the risk at the left-hand side of the U-shaped curve that makes the middle part seem better than it actually is.
Food consumption changes over time. A cross-sectional survey at one point in time cannot accurately capture past habits and distant exposures, especially for diseases such as heart disease and cancer that develop slowly over time. Studies on alcohol that try to overcome these shortcomings by eliminating former drinkers, or by using Mendelian randomization to better account for past exposure, do not show a cardiovascular benefit for moderate red wine drinking.
2. The Hors D’oeuvres — The Importance of RCTs
Now that you have made yourself the center of attention, it is time to cement your newfound reputation as a font of scientific knowledge. Most self-respecting hosts will serve smoked salmon as an amuse-bouche before the main meal. When someone mentions the health benefits of fish oils, you should take the opportunity to teach them about confounding.
Fish, especially cold-water fish from northern climates, have relatively high amounts of omega-3 fatty acids. Despite the plethora of observational studies suggesting a cardiovascular benefit, it’s now relatively clear that fish oil or omega-3 supplements have no medical benefit.
This will probably come as a shock to the worried well, but many studies, including VITAL and ASCEND, have demonstrated no cardiovascular or cancer benefit to supplementation with omega-3s. The reason is straightforward and explains why hormone replacement therapy, vitamin D, and myriad purported game-changers never panned out. Confounding is hard to overcome in observational research.
Prior to the publication of the Women’s Health Initiative (WHI) Study, hormone replacement therapy was routinely prescribed to postmenopausal women because numerous observational studies suggested a cardiovascular benefit. But with the publication of the WHI study, it became clear that much of that “benefit” was due to confounding. The women choosing to take hormones were more health conscious at baseline and healthier overall.
A similar phenomenon occurred during COVID. Patients with low serum vitamin D levels had worse outcomes, prompting many to suggest vitamin D supplementation as a possible treatment. Trials did not support the intervention because we’d overlooked the obvious. People with vitamin D deficiency have underlying health problems that contribute to the vitamin D deficiency. They are probably older, frailer, possibly with a poorer diet. No amount of statistical adjustment can account for all those differences, and some degree of residual confounding will always persist.
The only way to overcome confounding is with randomization. When patients are randomly assigned to one group or another, their baseline differences largely balance out if the randomization was performed properly and the groups were large enough. There is a role for observational research, such as in situations where ethics, cost, and practicality do not allow for a randomized controlled trial. But randomized controlled trials have largely put to rest the purported health benefits of over-the-counter fish oils, omega-3s, and vitamin D.
3. The Main Course — Absolute vs Relative Risk
When you get to the main course, all eyes will now be on you. You will almost certainly be called upon to pronounce on the harms or benefits of red meat consumption. Begin by regaling your guests with a little trivia. Ask them if they know the definition of red meat and white meat. When someone says pork is white meat, you can reveal that “pork, the other white meat,” was a marketing slogan with no scientific underpinning. Now that everyone is lulled into a stupefied silence, tell them that red meat comes from mammals and white meat comes from birds. As they process this revelation, you can now launch into the deeply mathematical concept of absolute vs relative risk.
Many etiquette books will caution against bringing up math at a dinner party. These books are wrong. Everyone finds math interesting if they are primed properly. For example, you can point to a study claiming that berries reduce cardiovascular risk in women. Even if true — and there is reason to be cautious, given the observational nature of the research — we need to understand what the authors meant by a 32% risk reduction. (Side note: It was a reduction in hazard, with a hazard ratio of 0.68 (95% CI, 0.49-0.96), but we won’t dwell on the difference between hazard ratios and risk ratios right now.)
This relative risk reduction has to be interpreted carefully. The authors divided the population into quintiles based on their consumption of anthocyanins (the antioxidant in blueberries and strawberries) and compared the bottom fifth (average consumption, 2.5 mg/d) with the top fifth (average consumption, 25 mg/d). The bottom quintile had 126 myocardial infarctions (MIs) over 324,793 patient-years compared with 59 MIs over 332,143 patient-years. Some quick math shows an approximate reduction from 39 to 18 MIs per 100,000 patient-years. Or to put it another way, you must get 4762 women to increase their berry consumption 10-fold for 1 year to prevent one heart attack. Feel free to show people how you calculated this number. They will be impressed by your head for numbers. It is nothing more than 39 minus 18, divided by 100,000, to get the absolute risk reduction. Take the reciprocal of this (ie, 1 divided by this number) to get the number needed to treat.
Describing risks in absolute terms or using number needed to treat (or harm) can help conceptualize statistics that are sometimes hard to wrap your head around.
4. Dessert — Funding
By the time the coffee is served, everyone will be hanging on to your every word. This is as it should be, and you should not be afraid of your newfound power and influence.
Dessert will probably involve some form of chocolate, possibly in cake format. (Anyone who serves fruit as dessert is not someone you should associate with.) Take the opportunity to tell your follow diners that chocolate is not actually good for you and will not boost brain performance.
The health benefits of chocolate are often repeated but rarely scrutinized. In fact, much of the scientific research purporting to show that chocolate is good for you did not actually study chocolate. It usually involved a cocoa bean extract because the chocolate manufacturing process destroys the supposedly health-promoting antioxidants in the cocoa bean. It is true that dark chocolate has more antioxidants than milk chocolate, and that the addition of milk to chocolate further inactivates the potentially healthy antioxidants. But the amount of sugar and fat that has to be added to chocolate to make it palatable precludes any serious consideration about health benefits. Dark chocolate may have less fat and sugar than milk chocolate, but it still has a lot.
But even the cocoa bean extract doesn’t seem to do much for your heart or your brain. The long-awaited COSMOS study was published with surprisingly little fanfare. The largest randomized controlled trial of chocolate (or rather cocoa bean extract) was supposed to settle the issue definitively.
COSMOS showed no cardiovascular or neurocognitive benefit to the cocoa bean extract. But the health halo of chocolate continues to be bolstered by many studies funded by chocolate manufacturers.
We are appropriately critical of the pharmaceutical industry’s involvement in drug research. However, we should not forget that any private entity is prone to the same self-interest regardless of its product’s tastiness. How many of you knew that there was an avocado lobby funding research? No matter how many industry-funded observational studies using surrogate endpoints are out there telling you that chocolate is healthy, a randomized trial with hard clinical endpoints such as COSMOS should generally win the day.
The Final Goodbyes — Summarizing Your Case
As the party slowly winds down and everyone is saddened that you will soon take your leave, synthesize everything you have taught them over the evening. Like movies, not all studies are good. Some are just bad. They can be prone to reverse causation or confounding, and they may report relative risks when absolute risks would be more telling. Reading research studies critically is essential for separating the wheat from the chaff. With the knowledge you have now imparted to your friends, they will be much better consumers of medical news, especially when it comes to food.
And they will no doubt thank you for it by never inviting you to another dinner party!
Labos, a cardiologist at Hôpital, Notre-Dame, Montreal, Quebec, Canada, has disclosed no relevant financial relationships. He has a degree in epidemiology.
A version of this article appeared on Medscape.com.
You’re invited to a dinner party but you struggle to make small talk. Do not worry; that will invariably crop up over cocktails. Because all journalism has been reduced to listicles, here are four ways to seem clever at dinner parties.
1. The Predinner Cocktails: A Lesson in Reverse Causation
Wine connoisseurs sniff, swirl, and gently swish the wine in their mouths before spitting out and cleansing their palates to better appreciate the subtlety of each vintage. If you’re not an oenophile, no matter. Whenever somebody claims that moderate amounts of alcohol are good for your heart, this is your moment to pounce. Interject yourself in the conversation and tell everybody about reverse causation.
Reverse causation, also known as protopathic bias, involves misinterpreting the directionality of an association. You assume that X leads to Y, when in fact Y leads to X. Temporal paradoxes are useful plot devices in science fiction movies, but they have no place in medical research. In our bland world, cause must precede effect. As such, smoking leads to lung cancer; lung cancer doesn’t make you smoke more.
But with alcohol, directionality is less obvious. Many studies of alcohol and cardiovascular disease have demonstrated a U-shaped association, with risk being lowest among those who drink moderate amounts of alcohol (usually one to two drinks per day) and higher in those who drink more and also those who drink very little.
But one must ask why some people drink little or no alcohol. There is an important difference between former drinkers and never drinkers. Former drinkers cut back for a reason. More likely than not, the reason for this newfound sobriety was medical. A new cancer diagnosis, the emergence of atrial fibrillation, the development of diabetes, or rising blood pressure are all good reasons to reduce or eliminate alcohol. A cross-sectional study will fail to capture that alcohol consumption changes over time — people who now don’t drink may have imbibed alcohol in years past. It was not abstinence that led to an increased risk for heart disease; it was the increased risk for heart disease that led to abstinence.
You see the same phenomenon with the so-called obesity paradox. The idea that being a little overweight is good for you may appeal when you no longer fit into last year’s pants. But people who are underweight are so for a reason. Malnutrition, cachexia from cancer, or some other cause is almost certainly driving up the risk at the left-hand side of the U-shaped curve that makes the middle part seem better than it actually is.
Food consumption changes over time. A cross-sectional survey at one point in time cannot accurately capture past habits and distant exposures, especially for diseases such as heart disease and cancer that develop slowly over time. Studies on alcohol that try to overcome these shortcomings by eliminating former drinkers, or by using Mendelian randomization to better account for past exposure, do not show a cardiovascular benefit for moderate red wine drinking.
2. The Hors D’oeuvres — The Importance of RCTs
Now that you have made yourself the center of attention, it is time to cement your newfound reputation as a font of scientific knowledge. Most self-respecting hosts will serve smoked salmon as an amuse-bouche before the main meal. When someone mentions the health benefits of fish oils, you should take the opportunity to teach them about confounding.
Fish, especially cold-water fish from northern climates, have relatively high amounts of omega-3 fatty acids. Despite the plethora of observational studies suggesting a cardiovascular benefit, it’s now relatively clear that fish oil or omega-3 supplements have no medical benefit.
This will probably come as a shock to the worried well, but many studies, including VITAL and ASCEND, have demonstrated no cardiovascular or cancer benefit to supplementation with omega-3s. The reason is straightforward and explains why hormone replacement therapy, vitamin D, and myriad purported game-changers never panned out. Confounding is hard to overcome in observational research.
Prior to the publication of the Women’s Health Initiative (WHI) Study, hormone replacement therapy was routinely prescribed to postmenopausal women because numerous observational studies suggested a cardiovascular benefit. But with the publication of the WHI study, it became clear that much of that “benefit” was due to confounding. The women choosing to take hormones were more health conscious at baseline and healthier overall.
A similar phenomenon occurred during COVID. Patients with low serum vitamin D levels had worse outcomes, prompting many to suggest vitamin D supplementation as a possible treatment. Trials did not support the intervention because we’d overlooked the obvious. People with vitamin D deficiency have underlying health problems that contribute to the vitamin D deficiency. They are probably older, frailer, possibly with a poorer diet. No amount of statistical adjustment can account for all those differences, and some degree of residual confounding will always persist.
The only way to overcome confounding is with randomization. When patients are randomly assigned to one group or another, their baseline differences largely balance out if the randomization was performed properly and the groups were large enough. There is a role for observational research, such as in situations where ethics, cost, and practicality do not allow for a randomized controlled trial. But randomized controlled trials have largely put to rest the purported health benefits of over-the-counter fish oils, omega-3s, and vitamin D.
3. The Main Course — Absolute vs Relative Risk
When you get to the main course, all eyes will now be on you. You will almost certainly be called upon to pronounce on the harms or benefits of red meat consumption. Begin by regaling your guests with a little trivia. Ask them if they know the definition of red meat and white meat. When someone says pork is white meat, you can reveal that “pork, the other white meat,” was a marketing slogan with no scientific underpinning. Now that everyone is lulled into a stupefied silence, tell them that red meat comes from mammals and white meat comes from birds. As they process this revelation, you can now launch into the deeply mathematical concept of absolute vs relative risk.
Many etiquette books will caution against bringing up math at a dinner party. These books are wrong. Everyone finds math interesting if they are primed properly. For example, you can point to a study claiming that berries reduce cardiovascular risk in women. Even if true — and there is reason to be cautious, given the observational nature of the research — we need to understand what the authors meant by a 32% risk reduction. (Side note: It was a reduction in hazard, with a hazard ratio of 0.68 (95% CI, 0.49-0.96), but we won’t dwell on the difference between hazard ratios and risk ratios right now.)
This relative risk reduction has to be interpreted carefully. The authors divided the population into quintiles based on their consumption of anthocyanins (the antioxidant in blueberries and strawberries) and compared the bottom fifth (average consumption, 2.5 mg/d) with the top fifth (average consumption, 25 mg/d). The bottom quintile had 126 myocardial infarctions (MIs) over 324,793 patient-years compared with 59 MIs over 332,143 patient-years. Some quick math shows an approximate reduction from 39 to 18 MIs per 100,000 patient-years. Or to put it another way, you must get 4762 women to increase their berry consumption 10-fold for 1 year to prevent one heart attack. Feel free to show people how you calculated this number. They will be impressed by your head for numbers. It is nothing more than 39 minus 18, divided by 100,000, to get the absolute risk reduction. Take the reciprocal of this (ie, 1 divided by this number) to get the number needed to treat.
Describing risks in absolute terms or using number needed to treat (or harm) can help conceptualize statistics that are sometimes hard to wrap your head around.
4. Dessert — Funding
By the time the coffee is served, everyone will be hanging on to your every word. This is as it should be, and you should not be afraid of your newfound power and influence.
Dessert will probably involve some form of chocolate, possibly in cake format. (Anyone who serves fruit as dessert is not someone you should associate with.) Take the opportunity to tell your follow diners that chocolate is not actually good for you and will not boost brain performance.
The health benefits of chocolate are often repeated but rarely scrutinized. In fact, much of the scientific research purporting to show that chocolate is good for you did not actually study chocolate. It usually involved a cocoa bean extract because the chocolate manufacturing process destroys the supposedly health-promoting antioxidants in the cocoa bean. It is true that dark chocolate has more antioxidants than milk chocolate, and that the addition of milk to chocolate further inactivates the potentially healthy antioxidants. But the amount of sugar and fat that has to be added to chocolate to make it palatable precludes any serious consideration about health benefits. Dark chocolate may have less fat and sugar than milk chocolate, but it still has a lot.
But even the cocoa bean extract doesn’t seem to do much for your heart or your brain. The long-awaited COSMOS study was published with surprisingly little fanfare. The largest randomized controlled trial of chocolate (or rather cocoa bean extract) was supposed to settle the issue definitively.
COSMOS showed no cardiovascular or neurocognitive benefit to the cocoa bean extract. But the health halo of chocolate continues to be bolstered by many studies funded by chocolate manufacturers.
We are appropriately critical of the pharmaceutical industry’s involvement in drug research. However, we should not forget that any private entity is prone to the same self-interest regardless of its product’s tastiness. How many of you knew that there was an avocado lobby funding research? No matter how many industry-funded observational studies using surrogate endpoints are out there telling you that chocolate is healthy, a randomized trial with hard clinical endpoints such as COSMOS should generally win the day.
The Final Goodbyes — Summarizing Your Case
As the party slowly winds down and everyone is saddened that you will soon take your leave, synthesize everything you have taught them over the evening. Like movies, not all studies are good. Some are just bad. They can be prone to reverse causation or confounding, and they may report relative risks when absolute risks would be more telling. Reading research studies critically is essential for separating the wheat from the chaff. With the knowledge you have now imparted to your friends, they will be much better consumers of medical news, especially when it comes to food.
And they will no doubt thank you for it by never inviting you to another dinner party!
Labos, a cardiologist at Hôpital, Notre-Dame, Montreal, Quebec, Canada, has disclosed no relevant financial relationships. He has a degree in epidemiology.
A version of this article appeared on Medscape.com.
You’re invited to a dinner party but you struggle to make small talk. Do not worry; that will invariably crop up over cocktails. Because all journalism has been reduced to listicles, here are four ways to seem clever at dinner parties.
1. The Predinner Cocktails: A Lesson in Reverse Causation
Wine connoisseurs sniff, swirl, and gently swish the wine in their mouths before spitting out and cleansing their palates to better appreciate the subtlety of each vintage. If you’re not an oenophile, no matter. Whenever somebody claims that moderate amounts of alcohol are good for your heart, this is your moment to pounce. Interject yourself in the conversation and tell everybody about reverse causation.
Reverse causation, also known as protopathic bias, involves misinterpreting the directionality of an association. You assume that X leads to Y, when in fact Y leads to X. Temporal paradoxes are useful plot devices in science fiction movies, but they have no place in medical research. In our bland world, cause must precede effect. As such, smoking leads to lung cancer; lung cancer doesn’t make you smoke more.
But with alcohol, directionality is less obvious. Many studies of alcohol and cardiovascular disease have demonstrated a U-shaped association, with risk being lowest among those who drink moderate amounts of alcohol (usually one to two drinks per day) and higher in those who drink more and also those who drink very little.
But one must ask why some people drink little or no alcohol. There is an important difference between former drinkers and never drinkers. Former drinkers cut back for a reason. More likely than not, the reason for this newfound sobriety was medical. A new cancer diagnosis, the emergence of atrial fibrillation, the development of diabetes, or rising blood pressure are all good reasons to reduce or eliminate alcohol. A cross-sectional study will fail to capture that alcohol consumption changes over time — people who now don’t drink may have imbibed alcohol in years past. It was not abstinence that led to an increased risk for heart disease; it was the increased risk for heart disease that led to abstinence.
You see the same phenomenon with the so-called obesity paradox. The idea that being a little overweight is good for you may appeal when you no longer fit into last year’s pants. But people who are underweight are so for a reason. Malnutrition, cachexia from cancer, or some other cause is almost certainly driving up the risk at the left-hand side of the U-shaped curve that makes the middle part seem better than it actually is.
Food consumption changes over time. A cross-sectional survey at one point in time cannot accurately capture past habits and distant exposures, especially for diseases such as heart disease and cancer that develop slowly over time. Studies on alcohol that try to overcome these shortcomings by eliminating former drinkers, or by using Mendelian randomization to better account for past exposure, do not show a cardiovascular benefit for moderate red wine drinking.
2. The Hors D’oeuvres — The Importance of RCTs
Now that you have made yourself the center of attention, it is time to cement your newfound reputation as a font of scientific knowledge. Most self-respecting hosts will serve smoked salmon as an amuse-bouche before the main meal. When someone mentions the health benefits of fish oils, you should take the opportunity to teach them about confounding.
Fish, especially cold-water fish from northern climates, have relatively high amounts of omega-3 fatty acids. Despite the plethora of observational studies suggesting a cardiovascular benefit, it’s now relatively clear that fish oil or omega-3 supplements have no medical benefit.
This will probably come as a shock to the worried well, but many studies, including VITAL and ASCEND, have demonstrated no cardiovascular or cancer benefit to supplementation with omega-3s. The reason is straightforward and explains why hormone replacement therapy, vitamin D, and myriad purported game-changers never panned out. Confounding is hard to overcome in observational research.
Prior to the publication of the Women’s Health Initiative (WHI) Study, hormone replacement therapy was routinely prescribed to postmenopausal women because numerous observational studies suggested a cardiovascular benefit. But with the publication of the WHI study, it became clear that much of that “benefit” was due to confounding. The women choosing to take hormones were more health conscious at baseline and healthier overall.
A similar phenomenon occurred during COVID. Patients with low serum vitamin D levels had worse outcomes, prompting many to suggest vitamin D supplementation as a possible treatment. Trials did not support the intervention because we’d overlooked the obvious. People with vitamin D deficiency have underlying health problems that contribute to the vitamin D deficiency. They are probably older, frailer, possibly with a poorer diet. No amount of statistical adjustment can account for all those differences, and some degree of residual confounding will always persist.
The only way to overcome confounding is with randomization. When patients are randomly assigned to one group or another, their baseline differences largely balance out if the randomization was performed properly and the groups were large enough. There is a role for observational research, such as in situations where ethics, cost, and practicality do not allow for a randomized controlled trial. But randomized controlled trials have largely put to rest the purported health benefits of over-the-counter fish oils, omega-3s, and vitamin D.
3. The Main Course — Absolute vs Relative Risk
When you get to the main course, all eyes will now be on you. You will almost certainly be called upon to pronounce on the harms or benefits of red meat consumption. Begin by regaling your guests with a little trivia. Ask them if they know the definition of red meat and white meat. When someone says pork is white meat, you can reveal that “pork, the other white meat,” was a marketing slogan with no scientific underpinning. Now that everyone is lulled into a stupefied silence, tell them that red meat comes from mammals and white meat comes from birds. As they process this revelation, you can now launch into the deeply mathematical concept of absolute vs relative risk.
Many etiquette books will caution against bringing up math at a dinner party. These books are wrong. Everyone finds math interesting if they are primed properly. For example, you can point to a study claiming that berries reduce cardiovascular risk in women. Even if true — and there is reason to be cautious, given the observational nature of the research — we need to understand what the authors meant by a 32% risk reduction. (Side note: It was a reduction in hazard, with a hazard ratio of 0.68 (95% CI, 0.49-0.96), but we won’t dwell on the difference between hazard ratios and risk ratios right now.)
This relative risk reduction has to be interpreted carefully. The authors divided the population into quintiles based on their consumption of anthocyanins (the antioxidant in blueberries and strawberries) and compared the bottom fifth (average consumption, 2.5 mg/d) with the top fifth (average consumption, 25 mg/d). The bottom quintile had 126 myocardial infarctions (MIs) over 324,793 patient-years compared with 59 MIs over 332,143 patient-years. Some quick math shows an approximate reduction from 39 to 18 MIs per 100,000 patient-years. Or to put it another way, you must get 4762 women to increase their berry consumption 10-fold for 1 year to prevent one heart attack. Feel free to show people how you calculated this number. They will be impressed by your head for numbers. It is nothing more than 39 minus 18, divided by 100,000, to get the absolute risk reduction. Take the reciprocal of this (ie, 1 divided by this number) to get the number needed to treat.
Describing risks in absolute terms or using number needed to treat (or harm) can help conceptualize statistics that are sometimes hard to wrap your head around.
4. Dessert — Funding
By the time the coffee is served, everyone will be hanging on to your every word. This is as it should be, and you should not be afraid of your newfound power and influence.
Dessert will probably involve some form of chocolate, possibly in cake format. (Anyone who serves fruit as dessert is not someone you should associate with.) Take the opportunity to tell your follow diners that chocolate is not actually good for you and will not boost brain performance.
The health benefits of chocolate are often repeated but rarely scrutinized. In fact, much of the scientific research purporting to show that chocolate is good for you did not actually study chocolate. It usually involved a cocoa bean extract because the chocolate manufacturing process destroys the supposedly health-promoting antioxidants in the cocoa bean. It is true that dark chocolate has more antioxidants than milk chocolate, and that the addition of milk to chocolate further inactivates the potentially healthy antioxidants. But the amount of sugar and fat that has to be added to chocolate to make it palatable precludes any serious consideration about health benefits. Dark chocolate may have less fat and sugar than milk chocolate, but it still has a lot.
But even the cocoa bean extract doesn’t seem to do much for your heart or your brain. The long-awaited COSMOS study was published with surprisingly little fanfare. The largest randomized controlled trial of chocolate (or rather cocoa bean extract) was supposed to settle the issue definitively.
COSMOS showed no cardiovascular or neurocognitive benefit to the cocoa bean extract. But the health halo of chocolate continues to be bolstered by many studies funded by chocolate manufacturers.
We are appropriately critical of the pharmaceutical industry’s involvement in drug research. However, we should not forget that any private entity is prone to the same self-interest regardless of its product’s tastiness. How many of you knew that there was an avocado lobby funding research? No matter how many industry-funded observational studies using surrogate endpoints are out there telling you that chocolate is healthy, a randomized trial with hard clinical endpoints such as COSMOS should generally win the day.
The Final Goodbyes — Summarizing Your Case
As the party slowly winds down and everyone is saddened that you will soon take your leave, synthesize everything you have taught them over the evening. Like movies, not all studies are good. Some are just bad. They can be prone to reverse causation or confounding, and they may report relative risks when absolute risks would be more telling. Reading research studies critically is essential for separating the wheat from the chaff. With the knowledge you have now imparted to your friends, they will be much better consumers of medical news, especially when it comes to food.
And they will no doubt thank you for it by never inviting you to another dinner party!
Labos, a cardiologist at Hôpital, Notre-Dame, Montreal, Quebec, Canada, has disclosed no relevant financial relationships. He has a degree in epidemiology.
A version of this article appeared on Medscape.com.