+++
How Can We Summarize Benefits and Risks?
++
Evidence-based practice (EBP) requires that clinicians summarize and consider both the benefits and risks of treatment for patients. Furthermore, clinicians must consider patient values and preferences related to these competing benefits and risks to determine which management strategies are in patients' best interests (see Chapter 27, Decision Making and the Patient).
++
These activities require clear summaries of the magnitude of treatment effect. The relative risk reduction (RRR; the control event rate minus the treatment event rate divided by the control event rate), the absolute risk reduction (ARR; the control event rate minus the treatment event rate), and the number needed to treat (NNT) represent alternative ways of summarizing the effect of treatment (see Chapter 9, Does Treatment Lower Risk? Understanding the Results).In this chapter, we provide examples of trials that have reported NNT and discuss how best to interpret this measure and apply it in clinical decision making.
+++
The Number Needed to Treat in Weighing Benefit and Harm
++
The NNT, the number of patients the clinician must treat for a particular period to prevent 1 adverse -target event (such as a stroke) or to create a positive outcome (such as a patient free of dyspepsia), may be the most attractive single measure. Arithmetically, the NNT is the inverse of the ARR. Clinicians could therefore simply take the ARR from a randomized clinical trial (RCT), calculate its inverse, and derive an NNT for their patients. However, because patients often begin with very different risks of adverse outcomes, such an approach can be profoundly misleading.
++
Consider, for instance, the GUSTO (Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries) trial,1 which reported the mortality within 30 days of hospital admission for approximately 20 000 patients who received streptokinase and approximately 10 000 who received tissue plasminogen activator (tPA). In the patients receiving tPA, the risk of dying was 6.3%; in those receiving streptokinase, the risk was 7.3%. The relative risk of dying with tPA is therefore 6.3/7.3 (86%), the RRR is 1.0 − 0.86 (14%), the ARR is 7.3 − 6.3 (1%), and the NNT is 100/1 (100). When deciding on whether an individual patient required tPA, we could assume that we might treat 100 patients to prevent a single death.
Such an approach ignores the fact that in the acute phase of ST-elevation myocardial infarction, patients have very different risks of dying. The Thrombolysis in Myocardial Infarction risk score estimates that in the month after ST-elevation myocardial infarction, the likelihood of dying in low-risk patients (ie, those who are younger, with noncomplicated Killip I inferior wall myocardial infarction and absence of other adverse prognostic factors) is 4.4%, whereas 36% of high-risk patients (ie, older patients and those with Killip III-IV anterior wall myocardial infarction) will die.2 Thus, the impact of tPA assessed ...