Practical application of nonrandomized research to patient care: a case study of nesiritide.

Prospective, randomized clinical trials are the gold standard for establishing cause-and-effect relationships between therapeutic interventions and specific outcomes. Unfortunately, these types of studies are not always available to evaluate important clinical end points such as mortality. Meta-analyses and observational studies are often used in an attempt to fill the gap in the literature when no prospective, randomized trials exist to answer a research question. A rapid increase in the number of published meta-analyses and observational studies has occurred in recent years. In the absence of prospective, randomized studies, it is essential for the clinician to understand the important issues involved in interpreting these other types of studies. When evaluating a meta-analysis, the clinician should assess for several different forms of bias, clinical heterogeneity, and use of appropriate methodology. When evaluating an observational study, sources of bias and confounding should be identified. If the potential for confounding is present, the methods used to minimize the impact of confounders should be evaluated for appropriateness. Mortality associated with nesiritide in the treatment of acute decompensated heart failure is a contemporary example of the use of nonrandomized research to address a research question that is unanswered by prospective, randomized studies. We review the details of a recent meta-analysis and observational evaluation of mortality associated with nesiritide and discuss the issues that are important in critical evaluation of nonrandomized study designs.