Dr. Joel Gelfand elucidates the concept of number needed to screen in "Behind the Numbers", a new Journal Watch Dermatology feature aimed at clarifying issues in epidemiologic design and biostatistics.


As dermatologists, we often ask ourselves, "Should I do a whole body skin check for melanoma on all of my patients?" The study by Breitbart and colleagues raises many important statistical and epidemiological issues that public health officials and individual clinicians must consider to determine if systematic skin cancer screening will effectively lower skin cancer mortality (the gold standard by which screening programs are judged). In this discussion, we will focus on melanoma mortality and the concept of number needed to screen, defined as the number of people that need to be screened during a given time period to prevent one melanoma-related death.

To calculate number needed to screen,1 we need to know the background risk, or prevalence, of the particular cancer to be detected in the population to be screened, and the mortality rate of the cancer in patients screened and not screened. In the study by Breitbart, 306,288 patients needed to be screened to detect 585 cases of melanoma or lentigo maligna melanoma. The authors note a prevalence of melanoma in their population of 1.6 per 1000 patients. Because we do not know what percentage of these melanomas would have been fatal if detected by routine care as opposed to the screening procedure, what follows are estimates. Based on National Cancer Institute data, it is estimated that 12.5% of melanomas are fatal. If we assume that the screening procedure was 100% effective in reducing mortality from melanoma, then about 5000 patients would need to be screened to prevent one death. In reality, no screening procedure is 100% effective, and a 20% risk reduction is often used as a rule of thumb to determine effectiveness. Thus, it may be that 25,000 patients need to be screened to prevent one death from melanoma.

How does melanoma screening compare with other screening procedures? Not very well. For example, about 800 patients need hemoccult screening to prevent one colon cancer death, about 1500 women aged 50 to 59 require screening mammography to prevent one breast cancer death, and about 420 patients need to be screened for cholesterol to prevent one cardiovascular death (assuming the resulting intervention is pravastatin). This relative ineffectiveness may be mitigated somewhat by the noninvasive nature of a skin check compared with a mammogram or blood draw and the relative ease of doing a skin biopsy as opposed to a colon or breast biopsy.

So should society invest in screening programs for melanoma?2 To draw a conclusion, beyond looking at number needed to screen, we must also assess the quality of the evidence showing that screening is effective in reducing morbidity and mortality, the benefits and harms of screening, and the cost-effectiveness of the screening program.

In clinical practice, the decision to screen all patients for melanoma or only those at increased risk is at the discretion of the individual physician. By looking at the number needed to screen, we can make more informed decisions as to what makes the most sense in the unique circumstances of our specific practice.


Citation(s):

1. Rembold CM. Number needed to screen: Development of a statistic for disease screening. BMJ 1998 Aug 1; 317:307.

2. Barratt A et al. Chapter 22.3, Moving from evidence to action: Recommendations about screening. In: Users' Guides to the Medical Literature. The McGraw-Hill Companies, Inc. 2008.