Search

Diagnostic tests are critical components of effective health care. They help determine treatments that are most beneficial for a given patient. Their assessment is a complex process that includes such challenges as a dearth of studies that evaluate clinical outcomes and lack of data on use of the test in realistic clinical settings. The methodologic quality of studies of diagnostic tests also lags behind the quality of studies of therapeutic interventions. Statistical methods to combine diagnostic accuracy data are more complex and not as well developed, leading to difficulties in the interpretation of results. The Agency for Healthcare Research and Quality Technology Assessment Program has adopted a 6-level framework for evaluating diagnostic technologies. The model emphasizes the need for systematic reviews of diagnostic test studies to go beyond the assessment of technical feasibility and accuracy to examine the impact of the test on health outcomes. In this paper, we use examples from 3 Evidence-based Practice Center reports to illustrate 3 challenges reviewers may face when reviewing diagnostic test literature: finding relevant studies, assessing methodologic quality of diagnostic accuracy studies, and synthesizing studies that evaluate tests in different patient populations or use different outcomes.

Purpose: To compare luteinizing hormone-releasing hormone (LHRH) agonists with orchiectomy or diethylstibestrol, and to compare antiandrogens with any of these three alternatives. Data Sources: A search of the MEDLINE, Cancerlit, EMBASE, and Cochrane Library databases from 1966 to March 1998 and Current Contents to 24 August 1998 for articles comparing the outcomes of the specified treatments. The search was limited to studies on prostatic neoplasms in humans. Total yield was 1477 studies. Study Selection: Reports of efficacy outcomes were limited to randomized, controlled trials. Twenty-four trials involving more than 6600 patients, phase II studies that reported on withdrawals from therapy (the most reliable indicator of adverse effects), and all studies reporting on quality of life were abstracted. Data Extraction: Two independent reviewers abstracted each article by following a prospectively designed protocol. The meta-analysis combined data on 2-year overall survival by using a random-effects model and reported results as a hazard ratio relative to orchiectomy. Data Synthesis: Ten trials of LHRH agonists involving 1908 patients reported no significant difference in overall survival. The hazard ratio showed LHRH agonists to be essentially equivalent to orchiectomy (hazard ratio, 1.262 [95% CI, 0.915 to 1.386]). There was no evidence of difference in overall survival among the LHRH agonists, although CIs were wider for leuprolide (hazard ratio, 1.099 [CI, 0.207 to 5.835]) and buserelin (hazard ratio, 1.1315 [CI, 0.533 to 2.404]) than for goserelin (hazard ratio, 1.1172 [CI, 0.898 to 1.390]). Evidence from 8 trials involving 2717 patients suggests that nonsteroidal antiandrogens were associated with lower overall survival. The CI for the hazard ratio approached statistical significance (hazard ratio, 1.2158 [CI, 0.988 to 1.496]). Treatment withdrawals were less frequent with LHRH agonists (0% to 4%) than with nonsteroidal antiandrogens (4% to 10%). Conclusions: Survival after therapy with an LHRH agonist was equivalent to that after orchiectomy. No evidence shows a difference in effectiveness among the LHRH agonists. Survival rates may be somewhat lower if a nonsteroidal antiandrogen is used as monotherapy., Drugs that mimic luteinizing hormone-releasing hormone (LHRH) appear to be as effective as orchiectomy in treating men with prostate cancer. Prostate cancer can be stimulated by male hormones, so removing the testes or treating men with drugs that inhibit male hormone production are two treatment options. In a review of 24 studies covering more than 6,600 patients, survival rates were similar in men who were treated surgically and those who were treated with drugs that mimic LHRH. These drugs suppress male hormone production and include leuprolide, buserelin, and goserelin.