Letter by Kelland and Roberts Regarding Article, “Lipid Management to Reduce Cardiovascular Risk: A New Strategy Is Required”
To the Editor:
We agree with Drs Superko and King that only by fully appreciating the differences between relative and absolute risk reduction can we determine the best treatment for our patients.1 Using the number needed to treat (NNT; the reciprocal of the absolute risk reduction)2 is a clinically meaningful way of doing this, although, as the authors point out, it is more difficult when comparing studies with varying lengths of follow-up. Superko and King attempt to overcome this problem by using the NNT per year, which they appear to have calculated simply by dividing the NNT by the length of follow-up (see Table 2 in their article). For the 2 years of the Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT) study, they quote a NNT of 58.8, and thus they quote the NNT per year as 29.4. This is flawed: The NNT will usually tend to rise if the time from start of treatment is reduced. Sackett et al3 suggested a simple correction for length of follow-up, in which the observed NNT is multiplied by the ratio of the actual average duration of follow-up to the duration of interest. Thus, for the PROVE-IT study, the NNT per year should be 117.6. However, this calculation assumes constant treatment efficacy over the entire length of the study. The greatest relative risk reduction in cardiovascular events due to lipid lowering might actually occur in the initial months after initiating treatment rather than in later years. It is more accurate therefore to calculate the NNT for each individual year directly, which requires knowing either the event rate or the hazard ratio and the event rate in the control group at the end of each year of the study.4
The authors go onto compare low-density lipoprotein cholesterol–lowering trials with low-density lipoprotein cholesterol–lowering combined with high-density lipoprotein cholesterol–raising trials. For example, they quote a NNT for the HDL Atherosclerosis Treatment Study (HATS) trial of 5.7. Although this appears impressively low, they fail to mention that the 95% confidence intervals for this study range from 3.0 to 86.4, reflecting the small sample size (n=34).
Although their article1 is designed to spark debate, and although we agree there is more to reducing cardiovascular risk than merely reducing low-density lipoprotein cholesterol, it is nevertheless important to back up the opinions expressed with data summarized with a sound statistical approach.
Superko HR, King S 3rd. Lipid management to reduce cardiovascular risk: a new strategy is required. Circulation. 2008; 117: 560–568.
Cook RJ, Sackett DL. The number needed to treat: a clinically useful measure of treatment effect. BMJ. 1995; 310: 452–454.
Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical Epidemiology: A Basic Science for Clinical Medicine. 2nd ed. Boston, Mass: Little Brown; 1991.
Altman DG, Andersen PK. Calculating the number needed to treat for trials where the outcome is time to an event. BMJ. 1999; 319: 1492–1495.