NNT (Number Needed to Treat)

•

Updated on February 25, 2021

As you’ve likely noticed, we’re getting back in touch with our more surly, snarky side this week. It can be fun now and then to channel one’s inner Fuji…. (Frankly, we can’t imagine MDA without it, and we hope you agree.) From energy drinks, we turn to one of our more popular objects of exasperated investigation and sarcastic commentary. What would we do without you, Big Pharma? Suffice it to say, there’s never a dearth of material with you around. And we thank you for that.

Today’s tirade covers how a drug’s effectiveness is reported to the public and to physicians. We’ve all seen the commercials that beam promises like “reduces risk of heart attack by 40%.” Sure, the numbers can sound impressive, but is there more to the story than we’re hearing in those ads? (I’m sure you can’t imagine the answer here.) Should we chalk up our lingering skepticism to smart consumer attitudes or gratuitous cynicism? (Another leading, rhetorical question for good measure.)

Let’s examine. Everyone got their pencils, coffee, popcorn (just kidding), and stress balls ready? Let’s take this thing apart! First off, a bit of review. Some weeks ago we brought you perspective on the “harm-benefit” ratio in pharmaceutical testing. With the myriad of, say, cholesterol targeting drugs out there, you’d think that all that testing to get a new pill on the market would have to prove it’s better than the older, usually cheaper varieties already on the market. We said that would be a big, fat no. As long as they show some degree of effectiveness above that of a sugar pill, that’s all they need to prove. (O.K. – get marketing on it asap! Tell the people what they need!) Allowing companies to measure the efficacy of new pharmaceuticals against placebos instead of “current approved drugs,” we reported, has resulted in a barrage of new drugs no more effective than those already on the market and (this is the truly important part) a dangerous number of new formulations that are more risky than beneficial! “Holy buckets” is right.

Hmmm, so how can we really judge the effectiveness and safety of these pharmaceutical treatments? If we can’t take the ads – and their claims – purely at face value, what are we supposed to look at? We think that’s an excellent question.

First off, the ads aren’t lying per se. They’re simply reporting on their drug’s effectiveness in a certain way, a way that (by chance) sounds the most advantageous. In drug trials you have what’s called relative risk, and this is often the number we hear on the ads. In a research group of a 1000 people, if 20 people are expected to suffer a heart attack in a given study period and only 10 actually do, whatever pharmaceutical drug being tested on these folks can be credited with halving the risk of heart attack. This “50% reduction” refers to the relative risk. But when you put it in the larger context of the entire group (10 in 1000 – or 1 in 100), suddenly the results don’t seem as dramatic. This means that 100 people needed to be treated with the pharmaceutical drug to stave off heart attack in one lone person. This is the NNT, the number needed to treat, to prevent heart attack. NNT is expressed as a number needed to treat to prevent a single medical event (e.g. heart attack, death, etc.). With adequate research analysis, an NNT can be offered for a given group of people with common risk factors (e.g. NNT among those with high blood pressure to prevent non-fatal stroke, NNT among those who have had a heart attack to prevent a fatal heart attack). An NNT, particularly when it is offered in both a general clinical and risk sub-group format, can provide a more realistic view of a drug’s effectiveness. It can give you and your doctor a clearer picture of what a drug is likely or unlikely to do for you.

Unfortunately, NNT isn’t reported as often as it should be. Despite encouragement for better reporting in randomized controlled trials, NNT is often not included in published studies. We raise this point because we think NNT offers a crucial part of the picture in determining and accepting a given treatment. Relative risk likely can’t tell you much. The full picture depends more on NNT analysis. We suggest doing some serious digging and questioning when considering a nonessential drug treatment.

Now, we’re always taught to consider both sides of an argument. What are the objections to the NNT as a significant and telling measure of a drug’s effectiveness? As mentioned, NNT looks at the power of a treatment to delay a particular negative medical event (e.g. heart attack, need for blood transfusion, death). But the NNT is measured within a set time frame and likely doesn’t/can’t follow the pattern of delay over a prolonged course of treatment/impact (perhaps many years or even decades) in the entire research group. Critics of NNT’s use also say that reporting a drug’s effectiveness in term of NNT will dissuade people who may benefit from the drug. (But perhaps it should in certain cases….)

The deal is this. When you begin to look at the NNTs of popular pharmaceuticals, you begin to see huge differences. An NNT chart (click image below for close up) from a pharmaceutical industry piece in Business Week offers a nice (and telling) picture of this range.

The NNT for antibiotic prescriptions used to treat h. pylori, bacteria strain that can cause ulcers, is 1.1. For every 11 people taking the prescriptions, 10 will be successfully treated. Those are good odds. As you move into the lifestyle “blockbuster” drugs such as Lipitor, you discover that dozens if not hundreds or even a thousand people need to be treated for 1 person to avoid a particular medical event (e.g. heart attack). The number of people who benefit (relative to those treated) is so small for many of these drugs that you have to ask (and can safely assume) that many if not most of these lucky few (and very probably more in the research group) may have benefited the same or more from key lifestyle alterations!

In the background of this picture is the crass, but nonetheless practical consideration of money. How much are you paying for the chance to be one person in a hundred? How much is the government or insurance company paying for you to be part of this lottery? How much are you paying for someone else to be? (We warned you about the crass part. Call us insensitive, but there it is.) We’re happy to pay, mind you, for truly necessary treatments that work for us and for other people, but when it comes to drugs with NNT of 200+? Something about that sticks in our craw. Cads we are, all of us here, for suggesting such a vile sentiment. Can’t we find a better way to spend our health dollars for the good of ourselves and others?

But consider for a second the following. Since we’re on the subject of cholesterol targeting medications, let’s look at the fact that they are the top selling therapeutic class of drugs (by sales $) in the U.S. In 2007, they accounted for $18.4 billion dollars in pharmaceutical sales. (Proton pump inhibitors were next and antipsychotics a close third with each clocking in over $13 billion, by the way.) Let’s look at some hard examples. As the Business Week piece shows, you have to prescribe cholesterol targeting medication to 16-23 people who have heart disease or a history of heart attack in order to prevent a single heart attack in one person. To save one of those people’s lives, the number jumps to 48. When you look at patients without heart disease but risk factors like high blood pressure, you need to “treat” up to 250 people to prevent a heart attack or stroke in one person. To prevent a death in this group, 500 people have to take this medication for 5 years straight.

Again, $18.4 billion dollars for one class of blockbuster medication in one year alone. Is it unseemly to wonder how many of these people were given these prescriptions after only presenting a single risk factor – a scenario that would be put then squarely in that 250-500+ NNT realm? What about Avandia (blood sugar reducer) prescription, which has an NNT of 1000+ to prevent a single heart attack and other serious medical events related to diabetes? Is it fair to wonder how many have tried adopting (and been consistently and strongly advised by doctors to adopt) effective lifestyle changes (e.g. diet, exercise, stress reduction)?

But there’s more. What about the risk of side effects? What risks are you taking on for the slim chance of being perhaps that 1 in 250 people? How many of those 250 people being treated with a cholesterol targeting drug will suffer muscle pain, mild dementia, gastrointestinal problems, etc. in order for one person to significantly benefit from the drug? Are you comfortable with that gamble as an individual in that group? Are we comfortable as a society saying that these odds are acceptable risks for the millions of people, including children now, who are prescribed these medications each year? Speaking of dramatic acronyms, there’s another that (not surprisingly) gets even less press. It’s a disconcerting little nugget called NNH: number needed to harm. (Now’s a good time to sip your drink and use that stress ball.) NNH in this context represents the number of people who need to take a certain medication for one person to experience a negative outcome. The “harm” here can vary, encompassing everything from experiencing a moderate to serious side effect, the development of a related condition, or death.

The number of people who experience side effects (some worse than the condition being “treated”) is substantial, and the number is growing. Remember the harm-benefit ratio post we reviewed earlier? Donald Light, Professor of Comparative Health Policy at the University of Medicine and Dentistry of New Jersey, explains “[O]ne in seven new drugs is superior to existing drugs, but two in every seven new drugs result in side effects serious enough for action by the [FDA].” These actions Light is referring to come after a drug’s release. And many physicians dismiss patient complaints of side effects (“nearly a third of the time”!) because too many aren’t fully aware of the side effects of these drugs. Researchers found that 60% of patients “said they felt their doctors did not appreciate the impact the symptoms had on their quality of life” when they reported side effects common enough to be listed in the medication’s information materials.

A side note about the side effect numbers and NNH… Before any study is completed, a number of people routinely drop out and aren’t counted in the final tabulations. This is typical and entirely expected with any human research group. And it’s inevitable that some of these people drop out due to their experience (understood at the time or not understood) of related side effects. Because they didn’t complete the study, they can’t be counted in the official study numbers, and it’s unlikely we ever hear or read anything about them.

We spend a lot of time complaining about Big Pharma (no, seriously?). But our problem isn’t (entirely) with the development, distribution and sales of truly essential medications. It’s with the culture of pill popping suggested by marketing ploys (and fueled by many influences in our society), a culture that diminishes the importance of basic health knowledge and the power of individual lifestyle choices. Manipulation for profit. We think people deserve better – more (and more accurate and relevant) information, higher standard testing and disclosure, and better opportunities to discuss medical concerns and diverse treatment options with their care providers. Exploring the NNTs of suggested prescriptions is part of this picture – for the patient and the care provider. Perhaps it’s a sad state of affairs when we have to bring suspicion, a certain dose of cynicism, and a magnifying glass to our consideration of medical treatment. We’d argue, however, that it’s the smart thing to do.

Now it’s your turn to weigh in with your insights and responses. Tell us what you think.