Dormuth 2004.

Methods	Study design: C‐RCT Unit of allocation: health areas Stratification by: number of physicians per area Type of comparison: PEM only vs. nothing group A: control: delayed intervention group B: mailing of therapeutic letters
Participants	Physicians Clinical speciality: general practice/family medicine Level of training: fully trained Setting/country: not clear/Canada
Interventions	The PEM consisted of 12 issues of the 'Therapeutics Letter' distributed between October 1994 and December 1997. Therapeutics Letter was a 2‐ to 4‐page colour‐printed bulletin mailed to most practicing physicians in British Columbia. Therapeutics Letter is a publication issued by the Therapeutics Initiative of the University of British Columbia. The letters included were those that had a clear message which could be predicted to result in a change to prescribing behaviour
Outcomes	12 process outcomes: proportion of newly treated patients receiving the analysis drug (cimetidine) proportion of newly treated patients receiving the analysis drug (metronidazole/amoxicillin or tetracycline) proportion of newly treated patients receiving the analysis drug  (ASA/ibuprofen/naproxen) proportion of newly treated patients receiving the analysis drug (isosorbide dinitrate) proportion of newly treated patients receiving the analysis drug (thiazide diuretics) proportion of newly treated patients receiving the analysis drug (inhaled corticosteroids) proportion of newly treated patients receiving the analysis drug (calcium‐channel blockers) proportion of newly treated patients receiving the analysis drug (long‐acting benzodiazepines) proportion of newly treated patients receiving the analysis drug (hormones) proportion of newly treated patients receiving the analysis drug (calcium‐channel blockers) proportion of newly treated patients receiving the analysis drug (clonazepam/alprazolam/diazepam) proportion of newly treated patients receiving the analysis drug (finasteride)
Notes	ES not computable No intervention: increase of 10% in the number of patients with prescriptions PEM: decrease of 15% in the number of patients with prescriptions
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote, pg. 1058: "one local health area in each pair was randomly selected and assigned (blindly by M.M. using the RAND function on Excel) to be in the control group"
Allocation concealment (selection bias)	Low risk	Quote, pg. 1058: "one local health area in each pair was randomly selected and assigned (blindly by M.M. using the RAND function in Excel) to be in the control group"
Baseline characteristics similar (selection bias)	Low risk	Quote, pg. 1059: "characteristics of the intervention and control physicians in 1991 are displayed in Table 2. The physicians and their patient populations were well balanced for these characteristics." TABLE 2, pg. 1058: "shows physician characteristics in 1994. Characteristics measured are percentage of general practitioners, mean age in years, percentage of men, mean number of visits from patients aged 66 years or more, mean age in years of patients aged 66 years or more and percentage of men/women/sex unknown of patients aged 66 years or more" COMMENT: the baseline characteristics of the intervention and control groups were reported and similar
Baseline outcome measurements similar (selection bias)	Low risk	Based on the large total number of prescriptions, baseline outcomes for the number of newly treated patients are similar across groups
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Quote, pg. 1058: "no requests to be excluded were received" COMMENT: the study does not specifically report on all physicians randomized by area at the beginning of the study remaining in the prescribing database throughout the study. Perhaps physicians retired, moved to a new area, or died
Blinding of outcome assessment (detection bias) All outcomes	Low risk	The outcome was objective
Contamination protection (contamination bias)	Low risk	Quote, pg. 1058: "the intervention and control groups were created by grouping an approximate 10% sample of prescribing physicians in 24 local health areas in a paired, cluster randomized design into 12 pairs based on the number of physicians in each area." such that all physicians within 1 local health area would be clustered
Selective reporting (reporting bias)	Low risk	All relevant outcomes in the methods section were reported in the results section
Other bias	Low risk	There was no evidence of other risks of bias