Mills 1999a.

Methods	Controlled before‐and‐after study (1 year follow‐up) Mean age of participants was 53 and 54 years in intervention and control arms respectively; 52% were male
Participants	574 patients with epilepsy from 14 general practices in northwest Bristol, England, UK
Interventions	Epilepsy specialist nurse service in primary care
Outcomes	Primary outcomes Frequency of epilepsy attacks at baseline and 12 months later Numbers of participants using more than one antiepileptic drug at baseline and 12 months later Provision of information at baseline and 12 months later Use of and attitudes to care at baseline and 12 months later Secondary outcomes Perceived effect of epilepsy and its treatment on everyday life at baseline and 12 months later Use of and attitudes towards the epilepsy specialist nurse at baseline and 12 months later All outcomes were derived from self completion questionnaire based on the Living With Epilepsy survey instrument
Funding	Study funded by Avon Health Authority
Notes	—
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The pre and postintervention periods for study and control practices were the same for the intervention and control groups, and the study and control sites were comparable with respect to distributions of practice size, doctor:population ratio, socio‐economic status, and mean distance from hospital. Practices were not, however, randomised to intervention and control arms
Allocation concealment (selection bias)	Unclear risk	Details of allocation concealment were not reported
Blinding (performance bias and detection bias) All outcomes	High risk	Authors do not report if any of the participants, clinicians or assessors were blinded. The subjective nature of the outcomes measured (all by self reported questionnaire) means this may have introduced bias
Incomplete outcome data (attrition bias) All outcomes	High risk	Dropout rates were high: 50.9% completed both baseline and final questionnaires
Selective reporting (reporting bias)	Low risk	All outcomes detailed in the methods were reported in the results
Other bias	Unclear risk	Power calculations and required sample size were reported. Though the unit of allocation was the clinic, statistical analysis did not account for clustering by clinic, and was thus not appropriate. Some significant differences were reported between intervention and control at baseline. There was no obvious possibility of contamination
Overall risk of bias	High risk	Quasi‐randomisation, no apparent blinding and significant dropout rate