The challenge of systematic reviews of diagnostic and staging studies in cancer

PURPOSE: Most diagnostic and staging studies focus on test performance characteristics while others may address long term clinical outcomes such as disease-free or overall survival, quality of life or cost. However, controversy remains as to which outcomes to measure and when long term outcomes based on large and costly randomized trials are required. METHODS: The comparative advantages for reviews of diagnostic and staging studies based on test performance characteristics in observational studies versus the assessment of outcomes in randomized controlled trials are presented. RESULTS: The measurement of diagnostic test performance is based on sensitivity, specificity, predictive value, the diagnostic odds ratio and Receiver Operating Characteristic (ROC) analysis. Alternatively, the clinical impact of a diagnostic test should consider the overall benefits and harms generally measured in terms of disease-free or overall survival, toxicity or quality of life and costs. Methods are available to systematically summarize these measures across studies, assess clinical and methodological heterogeneity and to evaluate the quality of studies. The advantages and disadvantages of studies of test performance or overall clinical impact are highlighted in the context of two examples: screening mammography and sentinel node biopsy. CONCLUSION: There appears to be an important role for systematic reviews of both studies of test performance and of long-term clinical impact in the evaluation of new diagnostic and staging procedures.