Question explored with the scientific record
Why is absolute risk more useful than relative risk?
The short version: absolute risk tells you your actual chance of being helped or harmed; relative risk only tells you the ratio, which can make a tiny effect look enormous.
The evidence is clear on this. A 1998 BMJ analysis showed that a drug can produce a 50% relative risk reduction while the absolute risk drops from 2% to 1% — a one-in-a-hundred benefit [3]. The same paper demonstrated that the number needed to treat (NNT) to save one life varied from 29 to 200 depending on the population studied, even when the relative reduction looked similar [3]. A 2000 review in Diabetic Medicine showed that framing the same data as relative versus absolute risk changed how patients and doctors perceived the benefit, and argued that both must be presented together [1].
The table below shows how the same 50% relative reduction maps to very different absolute benefits, using the cholesterol drug data from the 1998 BMJ paper [3]:
| Scenario | Baseline risk (placebo) | Risk on drug | Absolute risk reduction | Relative risk reduction |
|---|---|---|---|---|
| High-risk population | 20% | 10% | 10 percentage points | 50% |
| Moderate-risk population | 2% | 1% | 1 percentage point | 50% |
| Low-risk population | 0.2% | 0.1% | 0.1 percentage points | 50% |
The mechanism is simple: relative risk is a ratio that strips away the baseline. If the disease is rare, a 50% reduction may mean preventing one event in a thousand people. Absolute risk preserves that denominator. The 2013 ulcerative colitis study found that patients shown relative risk were far more willing to take medication than those shown the same data as a Cates plot (absolute numbers), even though the underlying benefit was identical [5]. That is not informed consent. That is framing.
My call: absolute risk is more useful because it gives the actual magnitude of benefit or harm, and relative risk without it is systematically misleading. Confidence: high.
Sources used 11
-
Evidence‐based patient information in diabetes
Unbiased, evidence-based patient information in diabetes requires presenting both benefits and non-benefits and using multiple data framings (absolute and relative risks) with confidence intervals to help patients make informed decisions.
DOI: 10.1046/j.1464-5491.2000.00383.x -
Improving the communication of multifactorial cancer risk assessment results for different audiences: a co-design process
An eight-step co-design process with patients, the public, and healthcare professionals produced a dynamic CanRisk report that makes multifactorial breast/ovarian cancer risk information more accessible to individuals while guiding clinicians across settings.
DOI: 10.1007/s12687-024-00729-4 -
Communicating the risk reduction achieved by cholesterol reducing drugs
A critical analysis of how the risk reductions from cholesterol-lowering drugs are presented in trials and advertising (favoring relative over absolute risk), demonstrating how framing shapes perception and urging the use of absolute risks with clear acknowledgement of uncertain…
DOI: 10.1136/bmj.316.7149.1956 -
Dabigatran, Rivaroxaban, or Apixaban versus Warfarin in Patients with Nonvalvular Atrial Fibrillation: A Systematic Review and Meta-Analysis of Subgroups
NOACs (dabigatran, rivaroxaban, and apixaban) show similar protection to warfarin for nonhemorrhagic stroke and systemic embolic events in nonvalvular atrial fibrillation, but substantially reduce intracranial bleeding, with absolute benefits varying by regional warfarin quality…
DOI: 10.1155/2013/640723 -
Conveying medication benefits to ulcerative colitis patients and effects on patient attitudes regarding thresholds for adherence
The study investigates ulcerative colitis patients' understanding and preferences for four statistical formats (RR, AR, NNT, and Cates plot) to convey maintenance medication benefits, and identifies the minimum benefit thresholds patients would require to adhere to therapy for b…
DOI: 10.1016/j.crohns.2012.11.006 -
Outcomes of Citalopram Dosage Risk Mitigation in a Veteran Population
Reduction of prescribed citalopram dosages to 40 mg/day or less after FDA and VA safety communications was associated with higher all-cause hospitalizations or death in veterans previously treated with higher dosages, with no decrease in mortality or arrhythmia-related outcomes.
DOI: 10.1176/appi.ajp.2016.15111444 -
Grundlegende Maßzahlen zur Analyse 2-stufiger Merkmalsausprägungen: Risiko, Odds, Logits, Relatives Risiko, Odds Ratio
A methodological primer that defines and contrasts risk, odds, and logits for dichotomous data in rehabilitation research, illustrating how relative risk and odds ratio summarize associations and how logit-based measures provide stable, base-rate–invariant comparisons across dif…
DOI: 10.1055/s-0043-109930 -
“I know what you told me, but this is what I think:” Perceived risk of Alzheimer disease among individuals who accurately recall their genetics-based risk estimate
Post hoc analysis of the REVEAL randomized trial shows that among individuals who accurately recall their APOE-based lifetime Alzheimer’s disease risk, roughly half still perceive their personal risk as discordant with the communicated estimate, with baseline risk perception, se…
DOI: 10.1097/GIM.0b013e3181cef9e1 -
Blood Leukocyte AHRR Methylation and Risk of Non–smoking-associated Cancer: A Case-cohort Study of Non–Hodgkin Lymphoma
Lower AHRR cg23576855 methylation in pre-diagnostic blood leukocytes is linked to a higher, time-dependent risk of future non-Hodgkin lymphoma, likely reflecting a tumor-related response rather than a long-term susceptibility signal.
DOI: 10.1158/2767-9764.crc-23-0151 -
Impact of Obeticholic acid Exposure on Decompensation and Mortality in Primary Biliary Cholangitis and Cirrhosis
In patients with compensated PBC cirrhosis, OCA use was associated with increased hepatic decompensation but not liver-related mortality or transplantation.
DOI: 10.1002/hep4.1720 -
P-524 Indication and limitation of repeat hepatectomy for recurrent hepatocellular carcinoma
In a set of conference abstracts, repeat hepatectomy was associated with better five-year survival than other treatments for recurrent hepatocellular carcinoma, IL-6 rose acutely after hepatic resection, and serum hyaluronic acid rose after hepatectomy.
DOI: 10.1016/0928-4346(95)90823-p