Simoni‐Wastila 2004.

Methods	Controlled interrupted time series study
Participants	Prescribers, and continuously enrolled Medicaid beneficiaries who were 19 years or older in 3 categories: Aid to Families with Dependent Children; Old Age Assistance; and Aid to the Permanently and Totally Disabled
Interventions	A regulatory restrictive prescribing policy, using a triplicate prescription programme for benzodiazepine use in Medicaid patients with chronic psychiatric and neurologic disorders. Triplicate prescription programmes (TPPs) require physicians to order covered medicines using multiple copy forms. The intervention aimed to reduce inappropriate drug use.
Outcomes	Medicine use in terms of use of benzodiazepines in per cent or per 100 enrolees
Notes	Although TPPs do not directly restrict physicians' freedom to prescribe, it has been suggested that they can reduce access to appropriate medicines by creating a "chilling effect" that discourages prescribing of the covered medicines due to physicians' fears of sanctions from the insurer
Risk of bias
Bias	Authors' judgement	Support for judgement
Intervention independent of other changes	Low risk	No other interventions were reported for the study time period. p323: control state similar to intervention state but for policy “New Jersey is a proximate state with a similar medicine reimbursement program but without any regulations targeting BZs during the 3‐year study period.”
Shape of the intervention effect pre‐specified	Low risk	p323: point of analysis identical with point of intervention
Intervention unlikely to affect data collection	Low risk	Data were extracted from monthly enrolment, medical claims, and medicine claims files from the computerized Medicaid Management Information System of the 2 study states for the years 1988 through 1990 and the year 1995
Knowledge of the allocated interventions adequately prevented during the study	Low risk	p328: assessment not blinded, but objective outcomes (BZ use and prescription rate)
Incomplete outcome data	Unclear risk	There were some prescriptions excluded, but the data on this were not reported in the paper (p325: some prescriptions written in 1988 were not filled immediately, January 1989 was excluded from estimates of TPP effects). No further information on the completeness of the outcome data was provided
Selective outcome reporting	Unclear risk	As we did not have access to a study protocol, it is not clear what outcomes were planned to be assessed at the protocol stage, and whether there are any that were not reported in the published study
Other risks of bias	Low risk	Appropriate design and analysis was performed. p325: appropriate time‐series analysis: “Autocorrelation and first‐order autoregressive effects were assessed and corrected in all models using the Statistical Analysis System Autoreg procedure (SAS Institute, Inc., Cary, North Carolina). Standardized regression coefficients from these models were used to compare post‐TPP changes in levels of and trends in use of BZs and other psychoactive medicines in the study state.” p323: “Using an interrupted time series with comparison series design, this study assessed changes in the use of BZs and other psychoactive medicines by clinically vulnerable patients in the New York Medicaid program and in a control state (New Jersey Medicaid) 12 months before and 24 months after the addition of BZs to the New York TPP."

BZ: Benzodiazepines

CBA: Controlled before‒after

HMO: Health Maintenance Organizations

TPP: Triplicate prescription programmes

USA: United States of America