Davis 2004.

Methods	Three‐arm cluster randomised trial (12 months follow‐up)
Participants	68 general practices (1133 patients) in Tayside, Scotland, UK Mean age across the arms ranged from 49 to 50 years; 47% were male
Interventions	Control: postal dissemination of a nationally developed clinical guideline Intermediate intervention: postal dissemination of guideline plus workshops and protocol documents Intensive intervention: intermediate intervention plus epilepsy nurse specialist to assist practices in the running of epilepsy review clinics
Outcomes	Primary outcome (SF‐36): Health‐related quality of life at baseline and 12 months after the intervention Secondary outcomes (five different epilepsy‐specific instruments, all of which have been previously published): Perceived severity of seizures (ICTAL & PERCEPT) at baseline and 12 months after the intervention Perceived adverse drug effects (ADEP) at baseline and 12 months after the intervention Impact of epilepsy on patients' lives (IMPACT) at baseline and 12 months after the intervention Sense of mastery over illness (MASTERY) at baseline and 12 months after the intervention Cognitive function (COGFUNC) at baseline and 12 months after the intervention
Funding	Support from Glaxo‐Wellcome, Janssen‐Cilag, Novartis, Parke‐Davis, Sanofi, and UCB‐Pharma allowed the provision of hospitality at the workshop sessions
Notes	—
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated cluster‐randomisation of GP practices
Allocation concealment (selection bias)	Low risk	A researcher not connected with the trial conducted allocation at randomisation
Blinding (performance bias and detection bias) All outcomes	Unclear risk	None of the GP practices (or staff), participants or assessors appeared to have been blinded. For some outcomes (from questionnaires as opposed to medical records), this may have introduced bias
Incomplete outcome data (attrition bias) All outcomes	High risk	Dropout rates were high in the trial, with only 72% completing the programme
Selective reporting (reporting bias)	Low risk	All outcomes detailed in the Methods were reported in the results
Other bias	Unclear risk	Power calculations and the required sample size were reported (although it is noted that the numbers of participants in each group fell short of that desired). The statistical analysis was appropriate for the cluster‐randomised design. There was no obvious possibility of contamination
Overall risk of bias	Unclear risk	Lack of clarity regarding blinding and significant levels of dropout