Forty-four studies were included: 20 RCTs, 6 controlled before-and-after studies, 7 observational controlled studies, 9 interrupted time series and 2 economic analyses.
The studies were generally of poor quality. Problems included the lack of sample size calculations, the lack of appropriate statistical tests, and the lack of data suitable for extraction. Additional tables were available on the British Journal of General Practice website (accessed 29/09/2005).
Professional (16 studies): studies of educational guidelines generally showed some change in clinical behaviour. The direction of change in referral rates varied with the intervention: rates increased with training in assessing long-term mentally-ill patients (1 study), guidelines plus education session (1 study), and structured management sheet plus guidelines (1 study); no change with open access facility (1 study) and guidelines (2 studies); rates decreased after training in managing menorrhagia (1 study) and computerised decision support (1 study); and variation in referral rates reduced with checklist for glue-ear plus video (1 study). The effect sizes varied.
Organisational (22 studies): there were few good-quality studies of in-house primary care schemes. Decreases in referrals were found for an in-house referral scheme (1 study), whereas increased rates were found with antenatal care provided by GPs and midwives (1 study). Studies of in-house Helicobacter pylori testing showed mixed results: 2 RCTs found large increases in referral rates, while one interrupted time series found a large decrease. The results from 2 in-house physiotherapy studies were inconclusive. Nine of the 10 studies of specialist outreach schemes found at least a small effect on referral rates in the direction that seemed rational. Reductions in referrals were found for specialist interventions in mental health (3 studies), in-house counselling (2 studies) and in-house ophthalmology (1 study). Open access schemes showed reductions with open access physiotherapy (1 study) and an open access Helicobacter pylori screening (1 study), but found no change with open access to urological investigations (1 study).
Financial/regulatory (6 studies): the studies generally showed a change in the direction anticipated. In most cases this was a reduction in the referral rates.
Patient/public (4 studies): large increases in referral rates were found for campaigns about malignant melanoma (2 studies), while a small non significant increase was associated with educational material on general health for mothers of infants (1 study). The other study on antidepressant compliance was underpowered.
Mental health (11 studies): 4 studies of professional interventions showed an effect on referral rates, but the clinical appropriateness and generalisability were uncertain. Two educational interventions both showed increased referrals, while all of the organisational interventions appeared to show an effect on referral rates in the anticipated direction. The effect sizes varied.