Search

In a systematic review (SR), the available evidence for a clinical problem is systematically and comprehensively collected from different studies, the likelihood of bias is assessed and the results are summarised in a reproducible manner. The results of an SR can be used in either individual patient care or the formulation of a practice guideline. Throughout the world, (national) organisations for guideline development use systems for classifying the validity of evidence according to the study design ('levels of evidence'), and this is also the case in the Netherlands. In formulating a treatment recommendation, consideration must be given to the various factors which determine the strength of the evidence: study design, consistency of the results (if more studies are available), quality of the individual studies, magnitude and precision of the reported effect and the clinical relevance of the outcome measures. If insufficient studies with the same outcome measure are available, problems can arise in an SR. In practice, problems can also arise with respect to estimating the cogency of an SR compared to another review or primary studies. For the same subject, there might be discordant reviews, differences between reviews and megatrials might exist or just one randomised clinical trial might be available. These problems can often be solved with the aid of a systematic analysis. This requires both methodological and clinical expertise.

Objective: To inventory the current state of the art regarding the effectiveness of conservative treatment of chronic low back pain., Design: Systematic reviews., Method: The relevant literature from the period January 1966-September 1999 was retrieved via Medline, Embase, PsychLit and the Cochrane Library and via reference lists in the articles found. The methodological quality of the studies was assessed using criteria for internal validity. On the basis of the number of examinations, their quality and the consistency of the findings, conclusions were subdivided into four levels of strength of scientific evidence., Results: There was strong evidence that exercise therapy and multidisciplinary treatment programmes were effective in chronic low back pain, and moderate evidence that non-steroidal anti-inflammatory drugs (NSAIDs), back schools and behavioural therapy were effective in chronic low back pain. There was also strong evidence that traction was not effective in chronic low back pain. Strong evidence for effectiveness of many other commonly used interventions was lacking.

Objective: To inventory the current state of the art regarding the effectiveness of conservative treatment of acute low back pain., Design: Systematic reviews., Method: The relevant literature from the period January 1966-September 1999 was retrieved via Medline, Embase, Psychlit and the Cochrane Library and via reference lists in the articles found. The methodological quality of the studies was assessed using criteria for internal validity. On the basis of the number of examinations, their quality and the consistency of the findings, conclusions were subdivided into four levels of strength of scientific evidence., Results: Approximately two-thirds of the randomised trials in the field of back pain were of poor methodological quality. There was strong evidence that advice to stay active, non-steroidal anti-inflammatory drugs (NSAIDs) and muscle relaxants were effective in acute low back pain. There was also strong evidence that bed rest and specific exercises were not effective in acute low back pain. Strong evidence for effectiveness of many other commonly used interventions was lacking.

Structured searching and selection of studies is an important component of a systematic review. It is recommended to record the various steps in a protocol in advance. The thoroughness of the searching and selection will partially depend on the available resources, like manpower and funds. A search action should be based on an unequivocally formulated research or clinical question, that is operationalized into clear inclusion and exclusion criteria. The actual start of a search strategy is a search in preferably multiple databases like Medline and EMBASE-Excerpta Medica. Additional search actions can be performed in trial registers and printed indexes and by correspondence with experts and hand searching of journals. Storage of the search results in a bibliographic database is recommended. Various methodological problems may play a role in searching and selecting studies for a review: studies may selectively not be published, results are only partially presented in the publication, studies are selectively included in reference lists, and reviewers themselves may make systematic errors in the selection process.