Introduction

It should be self-evident that healthcare decisions should be based on the best possible evidence; that effective treatments should be used whilst ineffective or harmful ones should not. Yet it is only relatively recently that the principles of evidence-based medicine have been explicitly discussed and gained widespread acceptance. There is an increasing awareness that the evidence for and against medical treatments must be evaluated properly and with appropriate scientific rigour. So far this book has set out the rationale for randomized controlled trials (RCTs) and described their role as main building blocks in clinical research. Undoubtedly, RCTs provide the most appropriate methods for the unbiased primary assessment of competing therapies. However, individual trials do not always provide the solutions that we seek and we need to look across trials at the totality of evidence. Consequently, there has been a vast increase in the number of systematic reviews and meta-analyses that have been undertaken in recent years [1] (Fig. 11.1).

Systematic reviews and meta-analyses provide an objective way of assembling, assessing and summarizing the evidence from RCTs, thereby providing a structured foundation for evidence-based medicine. The main reasons for systematic review and meta-analysis concern the need for completeness, for unbiased assessment, for synthesis and for statistical power. Probably the most important of these is the need to look at all the relevant evidence, so that evaluations and recommendations are based on the results of all trials and not just the published or well known ones. Another driving force is the need for summary and synthesis. Healthcare professionals, researchers, policy makers and indeed members of the public are relying increasingly on the process of systematic review. Faced with unmanageable amounts of information, they require reliable summaries of research on which to base clinical and policy decisions [2] and to guide future research. There

2000 15001000500

1970s 1980s 1990s Fig. 11.1 Number of meta-analyses of randomized trials published during three decades.

simply isn't sufficient time to read all relevant primary publications. Even enthusiastic clinical teachers estimate their median reading time at only two hours per week, yet it is estimated that general physicians would need to read seventeen articles a day every day of the year in order to keep up to date in their field [3]. In the past, review articles have been the main means of synthesizing medical information, but there has been considerable criticism of this approach [4]. In contrast to reports of primary research, traditional narrative reviews rarely make their objectives and methods explicit and can represent little more than the subjective opinions of their, usually, influential authors. Reviewers are often in the difficult position of trying to evaluate conflicting or equivocal studies based on a qualitative and perhaps selective reading of the literature. Indeed, it has been suggested that:

'because reviewers have not used scientific methods, advice on some life-saving therapies has been delayed for more than a decade while other treatments have been recommended long after controlled research has shown them to be harmful [5].'

In contrast, systematic reviews use explicit methodology, aim to assess all relevant evidence objectively, and go to considerable lengths to achieve unbiased assessment [6]. Where appropriate, meta-analysis offers a quantitative means of combining this evidence and, in doing so, of increasing statistical power and precision.

11.1.1 Definitions

A number of terms are used concurrently to describe the process of systematic review and integration of research evidence. These include systematic review, meta-analysis, research synthesis, overview and pooling [7] and there is sometimes confusion in the way the terms are used. Throughout this chapter we use the term systematic review to denote the entire process of locating, assembling and appraising trials, and meta-analysis to describe the process of quantitative synthesis whereby the results of RCTs are combined.

Systematic review: Means of reviewing clearly formulated questions, using explicit methodology, to minimize bias in the location, selection, critical evaluation and synthesis of research evidence.

Meta-analysis: Means of quantitatively combining the results of research studies to provide overall summary statistics.

Thus, a systematic review may or may not include a meta-analysis element, depending on whether it is appropriate or possible, and a good quality meta-analysis should always be done in the context of a systematic review. This distinction is important as there are many examples of meta-analyses that are not products of systematic review. Throughout the chapter, discussions are restricted to reviews and meta-analyses of randomized trials, we do not cover similar explorations of epidemiological studies, although many of the issues discussed are also relevant to such projects.

0 0

Post a comment