Synthesis

systematic reviews and meta-analyses

Introduction

The research papers published in scientific and professional peer-reviewed journals are the most important source of research for driving evidence-based health care. Each research publication contributes to the overall knowledge base for describing, understanding and solving health problems. However, an individual research project is usually insufficient for firmly deciding the truth of a research hypothesis or determining the efficacy of a clinical intervention. In order to have more soundly based evidence we need to compare, contrast and synthesize the results of research papers dealing with the same topic. Consistent results across diverse research studies provide a sounder basis for justifying the evidence-based delivery of health services.

The term ‘research literature’ refers to the set of publications containing the network of theories, hypotheses, practices and evidence representing the current state of knowledge in a specific area of the health sciences. A literature review contains both the critical evaluation of the individual publications and the identification of emergent trends and patterns of evidence based upon these studies. The literature review is a synthesis of the available knowledge in an area and therefore constitutes the strongest foundation for initiating further advances in theory and practice.

The overall aim of this chapter is to outline the basic strategies used for synthesizing evidence and producing a literature review.

The specific aims of the chapter are to:

1. Identify the basic methodological principles relevant to writing a health sciences review.

2. Describe the process for conducting a narrative review.

3. Describe the process for conducting a systematic review.

4. Explain how to interpret the findings of a published meta-analysis.

5. Explain the uses and limitations of systematic reviews and meta-analyses.

Basic principles

The first thing to consider is that writing a literature review is a demanding intellectual challenge. The facts do not ‘speak’ for themselves. Rather, the evidence has to be extracted, critically evaluated, organized and synthesized into a logical, coherent representation of the current state of knowledge. For example, consider the review by Olanow (2004) titled ‘The scientific basis for the current treatment of Parkinson’s disease’. This relatively brief 15-page review is based on only 75 references, although there are thousands of research papers, articles and reports available on the anatomy, physiology and treatment of Parkinson’s disease. In writing the review, the author had to make a series of expert judgements regarding which were the nine key papers, containing the most salient, up-to-date information for understanding the medical treatment of Parkinson’s disease. Many papers were rejected.

Second, the outcome of the review process is influenced by the theoretical orientation and professional background of the reviewer. Olanow, a leading neurologist, provided an authoritative review written from a biomedical perspective. In contrast, a physiotherapist working in neurological rehabilitation might take a different conceptual approach to the causes and treatment of Parkinson’s disease. He or she might place more emphasis on psychological and social factors as integral components of the aetiology and treatment of Parkinson’s disease.

Third, the selection, analysis, critique and synthesis of the materials is an active, interpretive process drawing on the personal experiences, interests and values of the reviewer. Even if their professional backgrounds are identical, there are no guarantees that two reviewers interpreting the evidence from the same set of publications will arrive at exactly the same conclusions. Depending on how these reviewers approached the subject matter, they might emphasize different aspects of the evidence or select different patterns in the data as being important, resulting in different syntheses. In Chapter 1, we discussed the post-positivist position that theories and preconceptions can influence our perceptions of what is happening in the world, and therefore shape the way in which we construct knowledge.

Last, the notion that we all have our experiences, opinions and prejudices does not imply that ‘anything goes’ when writing health sciences reviews. On the contrary, we need to apply the principles of scientific methodology to ensure that we provide an accurate overview of the literature. In other words, there are principles that we must follow in preparing a literature review.

As stated before, in preparing literature reviews and evaluating research findings, a multiplicity of papers must be considered, at least according to the following general steps:

1. Identify relevant literature; select key papers.

2. Critically evaluate key papers, as discussed in this section. You might decide to discard some papers if irreparable problems are discovered.

3. Identify general patterns of findings in the literature. Tabulate findings, where appropriate.

4. Identify crucial disagreements and controversies.

5. Propose valid explanations for the disagreements. Such explanations provide a theoretical framework for resolving controversies and proposing future research.

6. Provide a clear summary concerning the state of the literature, identifying progress, obstacles and further research.

Narrative reviews of research evidence

There are several approaches to conducting health sciences reviews. For example, the previously mentioned review by Olanow (2004) can be classified as a ‘narrative’ review. This approach entails producing a ‘story-like’ overview of the state of current evidence and theories. A narrative approach to reviewing literature means that the review is conducted in a story-telling fashion. Narrative reviews are basically of two kinds: those where the review constitutes the entirety of the paper, and those which are only part of the paper, and are integrated into the introductory section of a paper that reports a research study. Narrative reviews have an important place in the evidence base for health practice. For instance:

• A narrative review is an essential part of the ‘Introduction’ section of all reports of research studies. The review outlines the current state of knowledge, identifies gaps in that knowledge, justifies the research in terms of its capacity to advance knowledge and practice, and provides that background and the rationale for the research questions.

• A narrative review is also an essential first step for constructing models and theories. The review is used to identify and synthesize available evidence and then to demonstrate the ‘fit’ between the proposed conceptual framework (theory) and the facts.

• A narrative review can also be of use for the purpose of identifying current best practice in preventing and treating a particular health problem.

The quality of narrative reviews varies, given that the analysis, critique and synthesis of the material is an active process, which draws on the creativity and intellectual style of the reviewer. There is no guarantee that two people interpreting the evidence from the same set of papers will produce identical syntheses. Depending on how individuals approach the subject matter of a field, they may emphasize different aspects of the evidence or select different patterns in the data, resulting in a potentially different overall conclusion. In the extreme, some health workers hold biased theoretical and ideological positions to such an extent that the meaning of the evidence could be completely distorted (e.g. a medical scientist employed by a tobacco company, who reviews literature relating to the health risks). There are various sources explaining how to conduct reviews (see, for example, http://www.prisma-statement.org/).