Somkin 1997.

Methods	Randomised trial
Participants	7077 female health maintenance organization members (aged 50 ‐ 74 years with no prior mammogram in the previous 30 months or aged 20 ‐ 64 years with no prior Pap smear in the previous 36 months) in Northern California (USA)
Interventions	A letter sent to participants described some common barriers to screening and provided women with a telephone number to call and schedule a 'VIP mammogram' or a 'VIP Pap smear' (patient reminder group), the patient reminder plus a chart reminder (yellow form) placed manually by a chart room clerk (patient plus provider reminder group), and control/usual care group
Outcomes	Proportion of patients with mammograms and Pap smear done Assessed at: 6 months
Notes	1 relevant comparison (reminders plus other QI intervention (patient reminders) compared with other QI intervention) Funding: This study was funded by the Innovation Program, Clinical Services Branch, The Permanente Medical Group, Oackland, California
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	The sequence generation procedure is not described: Quote: "Separately for the mammography and Pap smear samples, we randomly selected 594 eligible women during each of 6 months (March to August 1994) and randomized them to receive one of the following..."
Allocation concealment (selection bias)	Unclear risk	Not reported
Blinding of participants and personnel (performance bias) All outcomes	Low risk	None of them was blinded to the intervention group, but the impact on outcome measurement seems to be limited because the outcome was routinely‐collected data
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Although this was unclear the outcome was assessed from a routinely‐used computerised system
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	This is not mentioned, although the follow‐up seems to be almost complete (?)
Selective reporting (reporting bias)	Low risk	Outcomes assessed were the relevant ones
Other bias	High risk	Authors incorporated a number of covariates in the analysis but unclear if this considered clustering appropriately. Risk of contamination?