'Patients with multiple risk factors for heart disease who took statin drug X reduced their risk of heart attack by 30 percent.”
That statement of relative risk reduction is a common framework used to describe the potential benefits of a drug, device, or other interventions for patients, whether in advertising campaigns or in a health care setting. But relative risk reduction, as most clinicians know, has a tendency to exaggerate the benefits of a given therapy.
“A 50 percent relative risk reduction may mean that you will need to treat 5000 patients so one will benefit from the treatment, or it may mean you only need to treat five patients,” said Diogo Mendes, PhD, of CHAD, the Centre for Health Technology Assessment and Drug Research at AIBILI, the Association for Innovation and Biomedical Research on Light and Image, in Coimbra, Portugal.
Consider a 2017 article in the New England Journal of Medicine on the clinical outcomes of treatment with the PCSK9 inhibitor evolocumab (Repatha) in patients with cardiovascular disease. The authors wrote that patients on evolocumab experienced “a 20% reduction in the risk of the more clinically serious key secondary composite end point of cardiovascular death, myocardial infarction, or stroke.” (New Engl J Med. 2017;376:1713; https://bit.ly/3anympx.)
But that's the relative risk reduction. The absolute risk reduction was 1.5 percent: Patients taking evolocumab reduced their two-year risk of cardiovascular death, heart attack, and stroke from 7.4 percent to 5.9 percent. That's a significant benefit, but it sounds a lot smaller than a 20 percent reduction to most people. And in a health care environment where shared decision-making is so important, it's essential that clinicians present the benefits and risks of a potential treatment in a way that patients and their families can understand and reasonably evaluate.
NNT and NNH
One useful alternative to relative risk reduction is the number needed to treat (NNT) framework, and its corollary, the number needed to harm (NNH). First introduced in the clinical literature in the late 1980s, the NNT quantifies the number of people who need to receive a drug or intervention for just one person to receive a benefit or prevent an adverse outcome. It's calculated by dividing 100 by the absolute risk reduction (100/ARR=NNT). The NNT for evolocumab (100/1.5), for example, is approximately 67, so 67 patients would need to be treated with evolocumab to prevent a heart attack, stroke, or cardiovascular death in one person.
Unadjusted NNTs are good predictors of quality-adjusted life year (QALY) gains from a given therapy, according to a 2006 study from the University of Toronto. (J Clin Epidemiol. 2006;59:224.) It found that “an NNT of 5 or less (≤5) was probably associated with a meaningful health benefit ... [while] ... an NNT of 15 or more (≥15) was quite certain to be associated with, at most, a small net health benefit.”
The BMJ requires randomized, controlled trials to report the absolute rather than the relative risks and NNTs with 95% confidence intervals (CIs), but Dr. Mendes and colleagues noted that few authors express their findings in terms of NNT or ARR. “Relative effect measures, such as relative risk (RR) or odds ratio (OR), are more commonly seen in the scientific literature,” they wrote. (BMC Med 2017;15, 112; https://bit.ly/2ycbueH.)
A group of emergency physicians found the NNT such a useful tool that they created a website nearly a decade ago focused on evaluating patient-important benefits and harms of given therapies based on NNT and NNH. TheNNT.com now includes hundreds of reviews, each with a color-coded summary borrowed from the traditional stoplight system. Therapies rated green have clear evidence of patient-important benefits that outweigh associated harms, such as steroids for asthma attacks. “If you give steroids to 8 patients with asthma attack in the emergency department, you prevent one from having to be admitted to the hospital,” TheNNT.com notes. “There are definitely side effects to steroids—high blood sugar, hyperactivity—but these, to us, seem minor in comparison.”
Therapies rated yellow require more study, typically because the data are not conclusive or substantial enough to give a clear recommendation. Red ratings suggest that harms and benefits may be relatively equivalent or equivocal, with benefits not clearly outweighing harms. And black ratings have clear associated harms with little to no tangible benefits, such as liberal transfusions for patients with anemia.
“We really believe in information symmetry, and, in this case, health information symmetry, which means we want both physicians and patients to have access to the same data, or at least summarized data, in an unbiased, accurate, and easy-to-understand form,” explained editor-in-chief Shahriar Zehtabchi, MD, a professor and the vice chair of academic affairs in emergency medicine at SUNY Downstate Medical Center in Brooklyn, NY. “When we have this information readily available, both for doctors and patients, they can discuss options together in a shared decision-making model more effectively. The NNT framework makes it much easier for patients to understand what physicians are saying and what research has shown.”
Take antibiotics for COPD exacerbation, for example. TheNNT.com gives this therapy a green light, noting that the NNT for reducing short-term mortality is 8, while the NNH for diarrhea is 20, the most common side effect. On the other hand, antibiotics for treating acute sinusitis in adults get a red rating, with an NNT of 17 for a cure at one to two weeks and an NNH of 8 for adverse events, primarily diarrhea and other GI side effects.
“Calculating the NNT and NNH obviously becomes more complicated when a disease is more complex or outcomes are hard to define or measure,” Dr. Zehtabchi said. “For example, if you're talking about patients' quality of life or some other patient-reported outcomes, that may not be as clear-cut as other outcomes such as mortality. When determining the outcome is not standardized and clearly measured, the decisions become harder. But for many treatments or interventions, NNT and NNH make it easier for the doctor to explain the research data to the patient and easier for the patient to understand and make a decision.”
Dozens of Specialties
Although TheNNT.com was launched by emergency physicians, it now encompasses more than two dozen specialties, with representatives from family medicine, internal medicine, cardiology, public health, and pharmacology on its editorial and scientific advisory boards. “Everything published on the site is peer-reviewed by our external reviewers and board members,” Dr. Zehtabchi said.
Because it was created by emergency physicians, TheNNT.com takes into account the fast pace of decision-making in the emergency department. “You know how our work environment is,” said Dr. Zehtabchi. “No one has time to go and pull a systematic review or an original randomized controlled trial, read and appraise and criticize the paper. NNT allows a physician in a critical care area to quickly find an answer for a question they have and discuss it with the patient and/or family in a way they will understand, then dive deeper later on if they want to.”
Engagement with TheNNT.com has grown significantly in recent years. Its Twitter account (@theNNTgroup) had 5500 impressions in January 2019 and 6500 the following month. A year later, impressions had grown to 53,800 and 65,600. Several journals have partnered with TheNNT.com to publish their articles regularly. Academic Emergency Medicine now publishes two of its reviews in every issue in a section called “Brass Tacks,” and the journal American Family Physician publishes one NNT article each month in a section called “Medicine by the Numbers.”
Dr. Mendes said despite the utility of the NNT and NNH, a metric is only as good as the calculation methods used to arrive at the numbers. “We have reviewed studies published in general internal medicine journals, and have found that about 30 percent of studies in high-impact journals reported NNTs that were not correctly calculated,” he said. “Application of the NNT is never a problem because it is always useful additional information about the effectiveness of a treatment that a physician should know in order to better support clinical decisions, but when a physician is reading a paper, they should have the proper knowledge to understand if the NNT as reported in the paper was correctly calculated or not.”
He explained that the NNT must always be presented with background information, not just as a number. “Sometimes you have papers that simply report ‘NNT=5.’ That doesn't mean anything if the authors don't provide you with a period of treatment, a confidence interval, and the control group used to calculate the NNT,” he said. “Even more important, NNT presented without NNH does not give full context. You should always compare NNT with NNH.”
TheNNT.com receives messages every day asking for additional content, Dr. Zehtabchi said. “Often, these emails are from patients asking, ‘My doctor has prescribed this medicine, and I don't see it on your website.’ But we limit our reviews to topics for which there is a high-quality systematic review that examines the evidence. If we find a methodologically robust systematic review (with or without meta-analysis), we solicit experts to write an NNT summary on the topic. If there are no systematic reviews, we generally keep the topic on our radar until enough trials are conducted and a systematic review is published that can be summarized and used as a source for the NNT article.”
Share this article on Twitter and Facebook.
Access the links in EMN by reading this on our website, or in our free iPad app, both available at www.EM-News.com.
Comments? Write to us at email@example.com.
Ms. Shawis a freelance writer with more than 20 years of experience writing about health and medicine. She is also the author of Having Children After Cancer, the only guide for cancer survivors hoping to build their families after a cancer diagnosis. You can find her work atwww.writergina.com.