Letters to the Editor
Absolute vs. Relative Numbers in Evaluating Drug Therapy
Am Fam Physician. 2001 May 15;63(10):1913-1916.
to the editor: Authors of articles in American Family Physician (AFP) often refer to the benefits of drug treatment only in relative terms. This practice has been shown to be misleading and affects physicians' willingness to prescribe1,2 and patients' acceptance of therapy.3
In particular, in the article “Bisphosphonates: Safety and Efficacy in the Treatment and Prevention of Osteoporosis,”4 the following statements were made:
1. The use of alendronate over three years… [produced] a 50 percent decrease in the risk of new vertebral, hip and wrist fractures in women with at least one preexisting vertebral fracture at baseline.”
While these numbers are correct, physicians need to realize that the absolute reduction in new vertebral fractures was 7 percent, hip fracture was 1.1 percent and wrist fracture was 1.9 percent over the three-year period.5
2. ”Treatment with 5 mg of risedronate… decreased the risk of new vertebral fractures by 41 percent and the cumulative incidence of nonvertebral fractures by 39 percent.”
The absolute reductions were 5 percent and 3.2 percent for nonvertebral and vertebral fractures, respectively.6
Editors of AFP need to be aware of these facts and should require authors who present benefits of drug therapy as relative reductions to also present the benefits as absolute numbers so that the information presented to their readership can truly be used to make informed decisions about drug therapy.
In addition, I would be interested to know if the authors present only relative numbers to their patients when they are discussing the benefits of drug therapy.
In my opinion, the following type of information should be presented to a woman with previous vertebral fractures and osteoporosis who is being considered for bisphosphonate therapy.
1. Mrs. Jones, your risk of developing a nonvertebral fracture over the next three years is approximately 8 percent.
2. If you take a drug daily for the next three years, that risk can be reduced from 8 percent to around 5 percent, or a difference of approximately 3 percent.
3. In general, results from studies have shown that these drugs are well tolerated (no difference in side effects between these drugs and placebo), although reports of esophageal irritation have been published.
4. The cost of the drug for this benefit will be approximately $2,000 over the next three years.
5. While the benefit may be greater than this over a longer time period, studies have not been conducted beyond three years.
If, after this discussion, the patient wishes to try the drug therapy, great; if not, great.
I strongly encourage physicians and authors to present benefits in absolute terms so patients can truly make informed decisions about drug therapy.
1. Bucher HC, Weinbacher M, Gyr K. Influence of method of reporting study results on decision of physicians to prescribe drugs to lower cholesterol concentration. BMJ. 1994;309:761–4.
2. Bobbio M, Demichelis B, Giustetto G. Completeness of reporting trial results: effect on physicians' willingness to prescribe. Lancet. 1994;343:1209–11.
3. Naylor DC, Chen E, Strauss B. Measured enthusiasm: does the method of reporting trial results alter perceptions of therapeutic effectiveness? Ann Intern Med. 1992;117:916–21.
4. Greenspan SL, Harris ST, Bone H, Miller PD, Orwoll ES, Watts NB, et al. Bisphosphonates: safety and efficacy in the treatment and prevention of osteoporosis. Am Fam Physician. 2000;61:2731–6.
5. Black DM, Cummings SR, Karpf DB, Cauley JA, Thompson DE, Nevitt MC, et al. Randomised trial of effect of alendronate on risk of fracture in women with existing vertebral fractures. Fracture Intervention Trial Research Group. Lancet. 1996;348:1535–41.
6. Harris ST, Watts NB, Genant HK, McKeever CD, Hangartner T, Keller M, et al. Effects of risedronate treatment on vertebral and nonvertebral fractures in women with postmenopausal osteoporosis: a randomized controlled trial. Vertebral Efficacy with Risedronate Therapy (VERT) Study Group. JAMA. 1999;282:1344–52.
in reply: I agree with Dr. McCormack that it is somewhat misleading to report only the relative reduction in risk in clinical trials of osteoporosis treatment. It is also somewhat misleading to report the “absolute” reduction (which, interestingly enough, Dr. McCormack reports as percentages, rather than absolute numbers). Perhaps the number needed to treat (NNT) to prevent an event would be clearer.
All of these numbers can be misconstrued when we move from the medical literature to the consultation or examination room and deal directly with a patient. I pity the patient on the receiving end of Dr. McCormack's scripted “presentation.” How does he know Mrs. Jones' risk of nonvertebral fractures? Certainly not from clinical trials, where subjects are highly motivated and healthier than the general population.
Why doesn't he also consider her risk for vertebral fractures and hip fractures? It has recently been shown that 20 percent of patients who experience a new vertebral fracture will have another in the year that follows.1 Patients who have a hip fracture have about the same likelihood of having another fracture in the next year or two.2
Dr. McCormack tells his patient about the cost of the drug but says nothing about the costs of not treating the problem. What about the costs of fractures, not just in dollars but also in suffering?
Strategies for risk reduction are well established for hypertension and hypercholesterolemia. Physicians don't talk to their patients with these conditions in the terms proposed by Dr. McCormack. Rather than discuss specific levels of risk or risk reduction, we tell our patients, “Your blood pressure is too high; you should be on medication to reduce it;” or “Your cholesterol level remains elevated despite diet and exercise; we need to add medication to bring it down.” If, in his patients with hypertension or hypercholesterolemia, Dr. McCormack takes the approach he advocates for patients who have osteoporosis, I doubt that many of his patients opt for therapy.
I agree with the request for including more complete information about the results of clinical trials. I strongly disagree with his proposal for using this information in clinical practice. I tell patients who have low bone density or a fragility fracture that they have osteoporosis.3 I tell them that patients who have osteoporosis should be treated.4 Effective agents, such as bisphosphonates, are safe and effective.5 If the patient wants further discussion, I am always happy to oblige; however, most patients want my advice, not a lesson in data analysis.
1. Lindsay R, Watts N, Roux C, Brown J, Barton I, Flowers K, et al. Increased risk of new vertebral fracture within 1 year of an incident vertebral fracture. Osteoporos Int. 2000;11(suppl 2):S112.
2. Colon-Emeric CS, Sloane R, Hawkes WG. The risk of subsequent fractures in community-dwelling men and male veterans with hip fracture. Am J Med. 2000;109:324–6.
3. Kanis JA, Melton LJ, Christiansen C. The diagnosis of osteoporosis. J Bone Miner Res. 1994;9:1137–41.
4. National Osteoporosis Foundation. Physician's guide to prevention and treatment of osteoporosis. Belle Mead, NJ: Excerpta Medica, 1998.
5. Greenspan SL, Harris ST, Bone H, Miller PD, Orwoll ES, Watts NB, et al. Bisphosphonates: safety and efficacy in the treatment and prevention of osteoporosis. Am Fam Physician. 2000;61:2731–6.
editor's note: Dr. McCormack's points are well taken. Our editorial policy at AFP is to clarify whether an author is discussing relative or absolute changes in percentages when discussing treatment effect. Our preference is to include, where relevant, the absolute change, because readers can use this information to calculate NNT. In addition, the absolute change often gives a better sense of the clinical relevance of a treatment effect. At times, relative changes may not be clinically important, even though the change in percentage may seem like a lot. For example, if drug X reduces mortality from 0.2 to 0.1 percent, this is a 50 percent relative reduction. On the surface, this may seem dramatic; however, the absolute reduction is only 0.1, a small decrease overall. Using the formula for NNT = 100/absolute risk reduction, one can calculate that 1,000 persons would need to be treated with drug X in order for one person's life to be prolonged (NNT = 100/0.1).
AFP's editors have two tasks in mind when reviewing a discussion of percentage change: first, to clarify whether the text is referring to relative or absolute change; and second, to present the information in a clinically meaningful way. In the latter case, this often means using absolute change rather than relative change. Where helpful, we ask authors to supply the NNT so that readers have another, more useful way of understanding the clinical significance of the discussion.
Send letters to Kenneth W. Lin, MD, MPH, Associate Deputy Editor for AFP Online, e-mail: email@example.com, or 11400 Tomahawk Creek Pkwy., Leawood, KS 66211-2680.
Please include your complete address, e-mail address, and telephone number. Letters should be fewer than 400 words and limited to six references, one table or figure, and three authors.
Letters submitted for publication in AFP must not be submitted to any other publication. Possible conflicts of interest must be disclosed at time of submission. Submission of a letter will be construed as granting the American Academy of Family Physicians permission to publish the letter in any of its publications in any form. The editors may edit letters to meet style and space requirements.
Copyright © 2001 by the American Academy of Family Physicians.
This content is owned by the AAFP. A person viewing it online may make one printout of the material and may use that printout only for his or her personal, non-commercial reference. This material may not otherwise be downloaded, copied, printed, stored, transmitted or reproduced in any medium, whether now known or later invented, except as authorized in writing by the AAFP. Contact firstname.lastname@example.org for copyright questions and/or permission requests.
Want to use this article elsewhere? Get Permissions