Persell 2016e.

Study characteristics
Methods	Cluster RCT
Participants	Large adult primary care practice affiliated with an academic medical center, Chicago, USA (Northwestern Medical Faculty Foundation) 342 patient visits, 8 physicians
Interventions	CDSS triggered by antibiotic prescription for acute respiratory infection that prompted clinicians to provide free‐text justification that would be included in medical record
Outcomes	Process adherence (prescribing)
Co‐Interventions	Educational: Single educational session for providers in control and intervention groups Beyond Clinician Education: Audit and feedback in intervention group; Financial incentives directed at providers in control and intervention groups
CDSS Features ‐ Acknowledgement of CDSS Required	Yes ‐ required acknowledgement of the CDSS but not documentation of action taken
CDSS Features ‐ Other	Conveyed patient‐specific information, decision support was complex, developed by study investigators, included supporting information on‐screen, interruptive, makes care recommendation, possible to execute desired action, 'push' mode of delivery, targeted overuse, user workflow considered in design
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment (selection bias)	Low risk
Incomplete outcome data (attrition bias) All outcomes	Low risk
Baseline characteristics similar?	Unclear risk	N/A
Unit of analysis error	Low risk	N/A