Diet or Drugs?

When contemplating the ideal therapy for chronic conditions like obesity and diabetes, shouldn't this always be the context?

Here’s a thought experiment: Imagine you can achieve equivalent benefits from a drug therapy as a change in diet, even a radical one, which is the better approach?

Changing how you eat will take significant effort and you’ll have to live with the requisite restrictions for a lifetime. The drug requires only that you take a pill or give yourself an injection, and you’ll have to live with that for a lifetime.

Can you make an informed decision without knowing the long-term risks and complications of both? And how do we make this choice when we don’t know?

The recent revolution in weight-loss drugs has seemingly rendered these questions moot. As the New York Times reported this week, a world of physicians are now making this decision with the new GLP-1 drugs—Wegovy, Ozempic, and their ilk—for their own weight and health. “Physicians Really Are Healing Themselves, With Ozempic,” as the Times headline proclaimed. And the subhead: “At cardiology conferences and diabetes meetings, doctors can’t help noticing that thin seems to be very in.”

Dietary therapy goes unmentioned in the article, other than the implication that some version of it, or maybe many versions, must have failed all those physicians or they wouldn’t now be appearing at conferences looking suddenly so lean and healthy, and decidedly happy about it. Judging by the before-and-after photos of Dr. John Buse from the University of North Carolina, his choice seems entirely understandable.

“There are no studies documenting the percentage of doctors taking the drugs,” the Times says. But physicians “are a good litmus test for drugs that are highly effective.”

If doctors who read the papers describing clinical-trial results are rushing to get a new drug, that is an indication that it’s really promising….

Many newly thin cardiologists and diabetes specialists, like so many of their patients, had risk factors for heart disease. Or their blood sugar was creeping up. Or just the physical strain of carrying excess weight made everyday life burdensome. They say they like their new looks but also their new health and energy. In a way, they feel like members of a club.

One operative question, though, is whether these doctors are “healing themselves”, as the headline asserted. Surely, they’re leaner and healthier, and the evidence suggests they’re reducing their future risk of disease.

But if we think of obesity as a symptom of an underlying disorder—something other than eating too much—then Ozempic and its ilk have treated the symptom. They’re healthier than they were, but whether these physicians will have long-term consequences or complications, from either taking the drug or having to discontinue it, remains unknown.

Can we now say these physicians are healthy, per the Times headline? Or as healthy as they might be? Those are different questions that can only be answered in the context of diet.

The very same article could have been written about ketogenic diets or what I’ll now call ketogenic therapy to capture the way physicians in my world have come to think of them. The catch, of course, is that ketogenic therapy is not new and so not news, and it comes with the accumulated intellectual baggage of 60 years of use and promotion outside established medicine.

In my book, The Case for Keto, published four years ago, I made the point that the medical profession is now full of physicians—tens of thousands is a reasonable estimate—who realized that conventional dietary therapy was not working for their patients or for them (just as it had implicitly failed those physicians who are now embracing the GLP-1 drugs), and so tried low-carbohydrate/high-fat ketogenic diets and experienced benefits very similar to what the Times describes is happening now to these physicians turning to the drugs.

The key difference is that physicians who embraced ketogenic therapy had to challenge conventional thinking on diet and health, whereas those embracing GLP-1 drugs simply follow the evolving medical consensus. They do not have to go against it.

Here’s how I described the conversion narrative (borrowing Malcolm Gladwell’s terminology) in The Case for Keto:

Almost invariably, these physicians had a personal stake as well… To accept the possibility that the conventional thinking on diet and weight is misconceived and so fails your patients, it helps to have experienced that failure yourself. Some of these physicians had been vegetarians for decades. Some had been vegans. Many are athletes, even ultra- endurance athletes. They prided themselves on eating “healthy” and yet found they had become fatter, diabetic, or pre-diabetic despite doing everything “right.” They were telling their patients to eat low-fat diets, mostly plants, not too much (control their portion sizes), and to exercise. They were following that advice themselves—and it wasn’t working.

Their rate of success in getting obese patients to lose meaningful amounts of weight with this diet and exercise prescription—as Deborah Gordon, a family medicine physician in Ashland, Oregon, described it to me—was “close to zero.” So these doctors did what we would hope any thoughtful person would do, and certainly our physicians, in these circumstances: They kept their minds open and went searching for a better approach.

When I was interviewing physicians for The Case for Keto, ketogenic therapy was about the only choice they had. The efficacy of the GLP-1 drugs had yet to become news, so these physicians saw themselves with little option but to challenge conventional wisdom and look elsewhere for effective therapy.

The relative efficacy of ketogenic therapy versus the GLP-1 drugs is hard to ascertain because no clinical trials compare them directly. The efficacy of the new GLP-1 drugs was established in a series of clinical trials known as STEP (Semaglutide Treatment Effect in People with obesity). In these trials, the average weight loss for patients on semaglutide (the active component of Wegovy and Ozempic) was 15%-17% after 68 weeks. Weight loss tended to plateau after a year.

As for ketogenic therapy, we have a report published just this week of the experience of 50 employees of a manufacturing company using a telemedicine intervention to maximize compliance with ketogenic therapy. In this study, the subjects lost, on average, 15.5% of their body weight after a year. The New York physician who designed the intervention, Tro Kalayjian, achieved and sustained a 150-pound weight loss with ketogenic therapy.

The STEP trials were randomized with a placebo control. The telemedicine approach was neither. The difference introduces considerable uncertainty in the comparison, but the results nonetheless suggest the two approaches are roughly similar in efficacy for weight loss.

Now that the drugs are so ubiquitous, the questioning of medical consensus seems no longer necessary. When conventional dietary approaches fail either physicians or their patients, the drugs are available to ensure that neither need subject themselves to the kind of learning experience that might lead them to try ketogenic therapy.

This drugs-vs-diet dichotomy was driven home to me last month when I met two physicians who ran a large medical group in the midwest. One of them told me that he was a huge fan of my work. He had read Good Calories, Bad Calories 15 or so years ago, and it had changed his life. Since then, he said, he had prescribed ketogenic therapy to several hundred of his obese and diabetic patients and they had seemingly thrived on it.

His partner in this medical group then told me his story. He was tall and lean, and I never would have guessed that he had ever struggled with his weight. But he had, he said, until he went on Mounjaro, a cousin to Wegovy and Ozempic, and lost 100 pounds. Now he was effortlessly maintaining that weight loss. He looked terrific. He said he loved the drug and, like the physicians in the Times article, felt like he had been given a new life, and now a healthy one and, most likely, a much longer one.

When I asked him why his partner was so enamored with my approach and dietary therapy, considering how happy he was with the drug, he said about his colleague, “Well, he’s a doctor.”

“Aren’t you a doctor, too?” I asked.

He said, “I’m an internist. I prescribe pills.”

He was not entirely serious, but there was truth in what he was saying.

Why physicians favor drugs over diet

Some large proportion of physicians will always prefer drug therapy to dietary alternatives. They have convinced themselves that their patients won’t stay on diets long-term and that their patients would prefer drugs to diets.¹ Many assuredly do. (In my studies on the history of obesity and diabetes research, I noticed physicians making this assumption a century ago.)

They have faith in the pharmaceutical industry to have done its job, and they feel they have more control over the outcome. They certainly have more faith in the pharmaceutical industry and the FDA regulators than they do the promoters of fad diets (like myself). Their patients either take the drugs prescribed or they don’t, and if the patients don’t, they can discuss why at the next appointment and what to do about it.

Undoubtedly, drugs are the easier option for both physicians and patients. No lengthy explanations are required; no follow-ups with dietitians or misunderstanding about what to eat or not eat (no, a potato is a carbohydrate); it doesn’t require a learning curve from the patient. It requires very little effort. Take the pills or don’t. And if the patients aren’t going to follow a dietary prescription, then the drugs de facto must be the better option.

Even in the late 1990s, when I first began researching diet and disease for the journal Science, influential physicians told me that they advised patients to lower their cholesterol by diet, knowing that many to most would fail—either because the diets were ineffective or unsustainable. Then the physicians could prescribe the appropriate medications, having felt they had fulfilled their ethical obligation to suggest lifestyle therapy first. They didn’t have to think of themselves as merely pill-pushers.

Gina Kolata, the author of the Times article, discussed just this thinking in a response to one of the comments:

…these doctors -- like so many other people -- have tried repeatedly to get their weight and risk factors under control with diet and exercise and have not succeeded. It is expected that the high cost of the drugs -- and even the fact that they must be injected -- is temporary because so many companies are now developing other versions, some of which are pills. It's analogous to cholesterol. Many people -- and I am one of them -- have high cholesterol no matter what they eat and no matter how thin they are. So we have to take statins for the rest of our lives

One of the factors in making these kinds of decisions is certainly agency: not just who do we trust, but who do we trust more if or when either the therapy stops working or complications arise. Do we want to control this situation ourselves and with the help of physicians who have experience with ketogenic therapy, or rely on the physicians who prescribed the medications and the pharmaceutical industry? That choice is preordained once we decide whether to challenge conventional thinking on diet and disease or embrace it.

Those of us who opt for ketogenic therapy have been convinced by the argument that refined carbohydrates and sugars² are the causes of the chronic disorders that we’re trying to resolve, and so abstaining from these foods is likely enough to be beneficial, that it’s an obvious choice. In that sense, the disorders we’re trying to resolve—diabetes and obesity most obviously—are responses to the carbohydrate content of the diet. If we don’t eat those foods, we don’t manifest the disorders. We’re not healthier, by this thinking (right or wrong), we’re healthy.

Drug therapy, whatever the efficacy of the drug, will always be treating the symptoms of the disorder, not removing the cause.

This has certainly been the case with type 2 diabetes, as I described in my recent book Rethinking Diabetes:

Both type 1 and type 2 are still considered progressive chronic diseases, meaning the conditions of patients is expected to worsen, and more drugs and/or insulin will be required in response. A 2017 review, written by four leading diabetes specialists and based on a symposium on diabetes care held at the annual meeting of the American Diabetes Association, concluded that “first and foremost” it had to be recognized that type 2 diabetes would inevitably get worse with time. It is “characterized by a progressive deterioration of glycemic control,” the authors wrote, “glycemic” referring to blood sugar levels.

By 2018, a trial conducted by the San Francisco start-up Virta Health had demonstrated that type 2 diabetes could be put into remission by continuous remote care (telemedicine and smartphones) that focused on ketogenic therapy or what Virta Health called nutritional ketosis. The results strongly supported the hypothesis that the symptoms of type 2 diabetes are a response to the carbohydrate content of the diet.³ This was an inference that could be made by a diet trial, but not a trial of drug therapy. From my book:

…the results from the Virta Health trial represent yet another challenge to a core belief about type 2 diabetes: decades of observational studies and clinical trials of drug therapy for diabetes had consistently concluded that type 2 diabetes is a chronic, progressive, and degenerative condition—­ as had, most notably the UKPDS trial. Diabetologists had concluded based on their drug trials that the primary challenge to better treatment—­ “the most frustrating barrier,” as diabetes authorities had described it in 2017—­ is the reluctance of physicians to either add more medications or increase doses of those medications that patients were already taking. The Virta Health trial had demonstrated that this conclusion was wrong. It had demonstrated that the degeneration of the diabetic condition could be reversed, even that the disease itself could be put into a drug-­free remission with continuous remote care and a diet that restricted carbohydrates and replaced those calories mostly with fat.

The belief that type 2 diabetes is necessarily a chronic, progressive disease—­ that it inevitably gets worse—­ can now be defended only in the context of the accepted standard of care for treatment. One way to look at the Virta Health results is that for many individuals with type 2 diabetes, the disease manifests itself only in the context of a carbohydrate-­ rich diet of the kind the ADA has been prescribing for more than half a century. It is a response to the carbohydrate content of the diet. For those with type 2 diabetes eating such a diet, the symptoms and the progression of the disease can only be managed by drug therapy. For those eating a mostly carbohydrate-­ free, fat-­ rich diet—­ marked by the state of nutritional ketosis—­ the disease ceases to progress and the symptoms essentially vanish.

Regrettably, this cannot be said about obesity, for two reasons: 1) We have no similar trials for obesity alone, and 2) the conventional thinking on the cause of obesity.

Achieving a healthy weight with ketogenic therapy is not considered remission of obesity, because it’s assumed that either the ketogenic therapy will stop working or the patient (or physician) will not maintain the diet. (It’s also commonly, and incorrectly, assumed that the diet works merely by reducing calorie intake—something that is unsustainable for most people. I’ll discuss this assumption in future posts.)

In the absence of clinical trials demonstrating long term safety and health in a diet-vs-drug comparison—and assuming the two methods can be equally effective, which may or may not be true and would depend on a host of variables—we’re left with the question of which is safest?

Why the news reports favor drugs

Here the drugs have decided public relations advantage not, because they necessarily are more beneficial, but because the long-term effects can be more easily studied.

On the one hand, the pharmaceutical industry is willing to fund clinical trials that can follow thousands or tens of thousands of patients long enough to establish benefits on what are called hard endpoints—heart attacks, for instance, or deaths from heart attacks, rather than a soft endpoint or risk factor like LDL cholesterol. Physicians, and so the journalists covering the field, have always assumed that ketogenic therapy is unsafe because the saturated-fat-rich nature of the diet will increase LDL cholesterol and cause heart disease.

Because trials of ketogenic therapies neither last long enough, nor include enough participants (thousands or more), they cannot establish whether the effect on LDL cholesterol translates to an increased risk of heart disease in this context of carbohydrate restriction. The absence of evidence from RCTs on this point makes it difficult to challenge the belief that elevated LDL cholesterol always translates to increased heart disease risk. The argument can be made, as Virta Health does (see image below), that overall heart disease risk decreases significantly, at least in patients with type 2 diabetes, even as LDL cholesterol might increase. But that doesn’t mean heart disease risk couldn’t go down further still if the diet prescribed didn’t raise LDL.

On the other hand, with hundreds of thousands if not millions of patients already taking these GLP-1 drugs, researchers can use established databases of electronic health records—here, for instance, a nationwide database, and here, a Veterans Affairs database—to compare the health status of patients who have been on the GLP-1 drugs to those who haven’t. This is the kind of evidence that fueled a recent Times report suggesting that the GLP-1 drugs might work to prevent Alzheimer’s Disease. (Headline: “To Protect Against Alzheimer’s, Researchers Look to Weight Loss Drugs.” Subhed: “Could medications like Ozempic benefit the brain? Here’s what scientists are learning.”)

These observational studies, of course, don’t tell us whether these associations between drugs and disease are causal. In the VA study, for instance, the researchers compared diabetic patients on the GLP1 drugs to those on what’s known as usual care, which includes a host of older, more traditional diabetic medications. Here’s what they found:

Compared to usual care, GLP-1RA use was associated with a reduced risk of substance use and psychotic disorders, seizures, neurocognitive disorders (including Alzheimer’s disease and dementia), coagulation disorders, cardiometabolic disorders, infectious illnesses and several respiratory conditions. There was an increased risk of gastrointestinal disorders, hypotension, syncope, arthritic disorders, nephrolithiasis, interstitial nephritis and drug-induced pancreatitis associated with GLP-1RA use compared to usual care.

The list of disorders associated with reduced risk is very impressive. Even if we assume these associations are causal, that the drugs do prevent or delay the appearance of these diseases, we still don’t know whether ketogenic therapy might be more effective, perhaps safer, without the increased risk of those other disorders that also associated with the use of these drugs. (One explanation for the existing associations is that the chronic diseases that appear to go down with GLP-1 use are disorders that are caused or exacerbated by the drugs employed in usual care. This is a plausible scenario, for instance, with insulin therapy for type 2 diabetes, as I discuss at great length in Rethinking Diabetes.)

These observational studies include no diet comparison, no diet-only control, because the electronic health databases tell us what drugs have been prescribed, not what diets are followed. (And if they did tell us what diets, if any, were prescribed, whether or not the patient complied with the dietary prescription might itself be a confounder in interpreting any associations observed.) Ultimately to know which is safer and more effective in a diet vs drug comparison, which approach has the greatest benefit-to-risk ratio for all these numerous chronic disorders, clinical trials comparing the two are necessary.

Meanwhile, we seemingly get weekly articles in the media reporting on the numerous chronic conditions that are likely to be improved or even prevented with the GLP-1 drugs: the recent Alzheimer’s article in the Times, for instance, or this one from last week, reporting on the results of a small clinical trial confirming anecdotal accounts that GLP-1 drugs reduce the craving for alcohol.

When the Times reported back in December 2023 on the many disorders that researchers have hopes of treating or preventing with the new drugs (headline: “What’s Next for Ozempic?”), these are virtually all disorders on which physicians are already testing ketogenic diets.

Many of these disorders are associated with obesity, insulin resistance, and type 2 diabetes, which suggests that any therapy that improves those—as ketogenic diets do—will also improve the conditions that are associated with them.

Here’s the illustration for the Times article:

Search “ketogenic diet” on the website clinicaltrials.gov and among the several hundred hits⁴, you’ll find trials assessing the safety and efficacy of virtually every imaginable chronic disease or disorder: A very much non-exhaustive list, in no particular order, includes autism and autism spectrum disorder, Alzheimer’s disease, major depressive disorder, multiple sclerosis, alcoholism, brain metastases, traumatic brain injury, rheumatoid arthritis, glioblastoma, anorexia nervosa, Parkinson’s disease, heart failure, non-alcoholic fatty liver disease, metastatic pancreatic cancer, polycystic ovary syndrome, and, of course, obesity, insulin resistance, and diabetes.

Some of these trials may be the product of wishful thinking, but it’s difficult to imagine that the relevant clinical investigators do not have at least anecdotal evidence to justify their investment of time, money, and effort into the trials, or, for that fact, that ethics boards would approve the trials without such justification. Over the coming years, we’ll see how these play out.

One fundamental advantage that diet trials, particularly ketogenic diet trials, have over drug trials is that they can inform us on the diet-disease relationship, which is crucial for prevention.

The simplest way to think of ketogenic therapy is as a way of eating that severely restricts carbohydrate-rich foods and replaces those calories with fat. If such a therapy prevents or delays the appearance of a chronic disease, or if it puts the chronic disease into remission—as with type 2 diabetes and pediatric epilepsy, most notably, and perhaps even schizophrenia and bipolar disorder, as suggested by numerous anecdotal accounts and this pilot trial published last May suggested—it gives us information about the potential role of carbohydrates and/or dietary fat in the disease process.⁵

Studies that compare drugs to diet can also tell us something about the mechanisms by which the drug works. I’ll discuss both of these issues in future posts.

Ultimately the drug bias in media coverage is a product of what health journalists consider news. And the GLP-1 drugs, wildly effective and exploding in use, are undeniably news. This is a story and the journalists have to cover it. And they can do so, as I’ve discussed in previous posts, without having to challenge either their preconceptions about the nature of a healthy diet or the consensus opinions of the medical community. It’s an easy story and an exciting one.

That the same could be said about dietary therapy, and that drug therapy always exists in the context of the safest and most effective dietary therapy, is not considered relevant because it’s not the dietary therapy that’s news.

One last note: I talked above about the question of agency and how that affects our decisions. With ketogenic therapy, there are likely to be no advances of any significance in the coming years. Once you’re eating this way, if problems come up or it becomes more difficult to maintain a healthy weight or control your blood sugar, your options for how to address the problem are limited. They might work, but they’re limited.

With drug therapy, there’s always the promise of new drugs coming down the pipeline. The Times article on physicians embracing the new drugs ended on just such a note. A physician who thinks the drugs changed his life, that “it’s the greatest thing [he] ever did,” comments that it’s just the beginning.

“There are 120 new agents coming along,” he noted, referring to drugs in clinical trials. “I look forward to ones that may have even better safety and effectiveness.”

It’s hard to argue with that logic, other than to say that the new drugs, too, will only be treating symptoms. They will not be putting these disorders into remission. Ketogenic therapy might be.

1. This is typically an assumption based on the observation that only a tiny percentage of people sustain significant weight loss on any diet. Because the physicians believe the diet must work if sustained, then the failure of the diet to achieve weight loss goals implies the failure of the patient, not the diet.

2. And, ok, maybe seed oils, although I remain skeptical.

3. The Virta trial was not randomized, but randomization is not necessary to establish that ketogenic therapy can reverse the course of diabetes and put it into remission.

4. Not all are clinical trials of ketogenic diets and many are on epilepsy, as the efficacy of ketogenic diets for epilepsy has long been accepted. Without doing a careful count, I’d estimate that perhaps 250 are trials of ketogenic diets as therapies for chronic disorders.

5. When I discussed the implications of what’s come to be called the carbohydrate-insulin model or hypothesis of obesity and chronic disease in Good Calories, Bad Calories (UK title: The Diet Delusion), I described the theoretical implications this way:

Both diabetes and metabolic syndrome are associated with an elevated incidence of virtually every chronic disease, not just heart disease. Moreover, the diabetic condition is associated with a host of chronic blood vessel-related problems known as vascular complications: stroke, a stroke-related dementia called vascular dementia, kidney disease, blindness, nerve damage in the extremities, and atheromatous disease in the legs that often leads to amputation. One obvious possibility is that the same metabolic/hormonal abnormalities that characterize the diabetic condition – in particular, elevated blood sugar, hyperinsulinemia and insulin resistance -- may also cause these complications and the associated chronic diseases. And otherwise healthy individuals, therefore, would be expected to increase their risk of all these conditions by the consumption of refined and easily-digestible carbohydrates, inflicting their damage first through their effects on blood sugar and insulin, and then indirectly through triglycerides, lipoproteins, fat accumulation and assuredly other factors as well.

This is a fundamental tenet of the carbohydrate hypothesis: If the risk of contracting any chronic disease or condition increases with metabolic syndrome and Type 2 diabetes, then it’s a reasonable hypothesis that insulin and/or blood sugar play a role in the disease process. And if insulin and blood sugar do play a pathological role, then it’s a reasonable hypothesis that the same conditions can be caused or exacerbated in healthy individuals by the consumption of refined and easily-digestible carbohydrates and sugars.