Nick Freemantle

Methodological difificulties in rehabilitation research

This paper presents an overview of some of the pitfalls and suggests ways of improving the quality of research into rehabilitation after stroke. The aims of rehabilitation are outlined and methodological problems inherent in this area discussed, including spontaneous recovery, multidimensional outcomes, definition of treatment and placebo effects. Major weaknesses found in the rehabilitation literature are identified and recommendations for improvement made, including the need for comparable controls, adequate numbers, appropriate outcome measures, clear definition of therapy, generalizability, and a concern for the cost-effectiveness of stroke rehabilitation packages. The field of rehabilitation is still relatively new and idiosyncratic in form and content. Thus it still provides opportunities for research into its relative effectiveness before current practices in rehabilitation become entrenched in medical practice and folklore. Adequate funding support for improved, multicentre clinical trials in rehabilitation is essential.

Examining the effectiveness of treatments for depression in general practice

This paper examines the importance of systematic reviews to decision-makers in the UK National Health Service (NHS), outlines Effective Health Care which is a specific initiative to provide systematic reviews in an accessible form, describes the findings of a recent review of the effectiveness of treatment of depression in primary care, and highlights the importance of the Cochrane Collaboration in future work on the effectiveness of treatments in mental health.

Assisted conception on the NHS.

EDITOR, -David T Baird wonders why governments have reservations about assisted conception techniques for managing infertility. Surely one of the reasons is the poor quality of evidence for the effectiveness of some treatments, which we review in the most recent issue of Effective Health Care.2 Many subfertility treatments have not been evaluated by randomised controlled trials. For example, there is no published report of a randomised controlled trial comparing in vitro fertilisation and embryo transfer with an untreated control group (for example, subjects in whom treatment is delayed). Similarly, the view that medical treatment of amenorrhoea is highly effective is based entirely on retrospective reviews. Estimating the increase in pregnancy rate over the often appreciable spontaneous pregnancy rate that would have occurred in the absence of treatment is often difficult.3 The relative effectiveness of in vitro fertilisation and embryo transfer and other assisted conception techniques is not clear for various reasons. For severe bilateral occlusion of the fallopian tubes in vitro fertilisation and embryo transfer is the only possible treatment, but for women with at least one patent and healthy fallopian tube there is controversy over which technique is best. Well designed randomised controlled trials are needed to answer questions about the best technique for particular indications and patient characteristics.4 Different treatments may have different effects on the monthly fecundity and cumulative pregnancy rates. For example, prednisolone treatment of antibodies to sperm may need longer follow up before improvement in male fertility is observed. With assisted conception the effects are more rapid. Therefore, to compare treatments time must be incorporated into the analysis. Many studies, however, do not report the duration of follow up or number of cycles of treatment. Often life table analysis cannot be used appropriately because of the lack of information about nonrandom drop out of patients. Small studies cannot reliably answer questions of efficacy and may result in potentially effective interventions being dismissed prematurely. To increase the size of their study some authors group together patients with various causes of infertility, but this makes interpreting the results difficult. Another approach is to use meta-analysis to pool the results of small studies. Variability in the selection of patients, treatment, and measurement of outcomes is so great among studies, however, that aggregating results can be misleading. Lack of good quality evidence on the effectiveness of treatment for subfertility has contributed to health authorities often sceptical stance.5 These treatments must be evaluated thoroughly before they become generally available on the NHS.