The use of biomarkers in surveillance, medical screening, and intervention

The relative benefit of an active treatment over a control is usually expressed as the relative risk, the relative risk reduction, or the odds ratio. These measures are used extensively in both clinical and epidemiological investigations. For clinical decision making, however, it is more meaningful to use the measure "number needed to treat." This measure is calculated on the inverse of the absolute risk reduction. It has the advantage that it conveys both statistical and clinical significance to the doctor. Furthermore, it can be used to extrapolate published findings to a patient at an arbitrary specified baseline risk when the relative risk reduction associated with treatment is constant for all levels of risk.

To develop the number needed to screen, a new statistic to overcome inappropriate national strategies for disease screening. Number needed to screen is defined as the number of people that need to be screened for a given duration to prevent one death or adverse event. Number needed to screen was calculated from clinical trials that directly measured the effect of a screening strategy. From clinical trials that measured treatment benefit, the number needed to screen was estimated as the number needed to treat from the trial divided by the prevalence of heretofore unrecognised or untreated disease. Directly calculated values were then compared with estimate number needed to screen values. Standard literature review. For prevention of total mortality the most effective screening test was a lipid profile. The estimated number needed to screen for dyslipidaemia (low density lipoprotein cholesterol concentration >4.14 mmol/1) was 418 if detection was followed by pravastatin treatment for 5 years. This indicates that one death in 5 years could be prevented by screening 418 people. The estimated number needed to screen for hypertension was between 274 and 1307 for 5 years (for 10 mm Hg and 6 mm Hg diastolic blood pressure reduction respectively) if detection was followed by treatment based on a diuretic. Screening with haemoccult testing and mammography significantly decreased cancer specific, but not total, mortality. The number needed to screen for haemoccult screening to prevent a death from colon cancer was 1374 for 5 years, and the number needed to screen for mammography to prevent a death from breast cancer was 2451 for 5 years for women aged 50-59. These data allow the clinician to prioritise screening strategies. Of the screening strategies evaluated, screening for, and treatment of, dyslipidaemia and hypertension seem to produce the largest clinical benefit.

In diagnostic accuracy studies, the contrast of interest can be one of the following: one single test contrast; comparing two or more single tests; further testing in addition to previous diagnostics; and comparing alternative diagnostic strategies. The clinical diagnostic problem under study must be specified. Studies of "extreme contrasts" (as early phase evaluations) and studies in "clinical practice" settings (assessing clinical value) should be distinguished. Design options are (1) survey of the total study population, (2) case-referent approach, or (3) test-based enrollment. Data collection should generally be prospective, but ambispective and retrospective approaches are sometimes appropriate. In addition to determinants of primary interest [the test(s) under study] possible modifiers of test accuracy and confounding variables must be specified. The reference standard procedure should be independent from the test results. Applying a reference standard can be difficult in case of classification errors, lack of a clear pathophysiologic concept, incorporation bias, or invasive or complex investigations. Possible solutions are: an independent expert panel, and the delayed type cross-sectional study (clinical follow-up). Also, a prognostic criterion can be chosen. For studies to be relevant for practice, inclusion criteria must be based on "intention to diagnose" or "intention to screen." The recruitment procedure is preferably a consecutive series of presenting patients or a target population screening, respectively. Sample size estimation should be routine. Analysis has to be focused on the contrast of interest. Estimating test accuracy and prediction of outcome need different approaches. External (clinical) validation requires repeated studies in other, similar populations. Also, systematic reviews and meta-analysis have a role. To enable readers of diagnostic research reports to evaluate whether methodological key issues were addressed, authors are advised to follow the STARD guidelines.