Project MIMIC (Maximizing Implementation of Motivational Incentives in Clinics)

Provider-level measure of whether the provider delivered the target number of CM sessions (at least 10 sessions) to at least one patient based on based on electronic medical record review and data entered into a study-specific CM tracker tool for up to 25 charts per site (25 charts\*30 sites = 750 charts). Providers will report on patient encounters in the electronic medical record and the study-specific CM tracker tool, and for each encounter will report if CM was provided. Using patient level data, providers will be classified as 1 (delivered 10 or more sessions to at least 1 patient) or 0 (did not deliver 10 or more CM sessions to any patients). \*This measure was initially defined as a patient-level outcome. We altered the level at which CM Exposure was assessed because our initial approach excluded providers who were trained but never delivered CM. To follow intent-to-treat principles, we aggregate CM Exposure data at the provider-level in a manner that uses all available data.

Time frame: From baseline to 9 months post-baseline

Population: Table reports proportions with listwise deletion. The pre-specified plan uses mixed models with full maximum likelihood estimation (intent-to-treat) and data from all 95 ATTC providers and 91 E-ATTC.

Proportion of programs delivering any CM after removal of active support. This is calculated based on review of all patient charts over a 6-month interval. Providers report on patient encounters in the medical record, and for each encounter report if CM was provided. Programs are classified as 1 (reported delivering CM to at least 1 patient) or 0 (did not deliver CM to any patients). The proportion of programs delivering CM is then calculated; a higher proportion is a better outcome. \*The level at which CM Sustainment was assessed was altered from provider-level to program-level because of the frequency of programs failing to report applying CM among any patients, across any of its providers. In addition, there was such high staff turnover we could not assess at the provider-level using original provider IDs. To be able to use all available data from all programs' medical records, we report on the proportion of any programs delivering CM after removal of active support.

Time frame: 6-month time interval following Implementation time period

Population: Sustainment was measured at the organizational level (N=14 ATTC, N=14 E-ATTC) as a dichotomous variable indicating whether or not they continued delivering the intervention. The number of participants seen in each condition is unknown and therefore overall number of participants is entered as NA (not applicable).

Provider scores on the Contingency Management Competence Scale for Reinforcing Attendance (CMCS; Petry & Ledgerwood, 2010). Coders blind to treatment condition rate audio recorded CM sessions using the CMCS, which measures provider skill in CM delivery. CMCS contains 6 CM-specific skill items and 3 general skill items that are scored on a scale from 0 to 7. For each item, a score of 0 indicates an audio recording was not submitted, a score of 1 indicates the lowest possible skill and a score of 7 indicates the highest possible skill. Possible scale scores range from a minimum of 0 to 63. An average score will be calculated for each provider, with a minimum of 0 and maximum of 7. Providers will submit one audio recording per month for the duration of the 9-month Implementation phase. Each provider's highest CMCS score will be used in analysis. Higher scores indicate higher skill, which is a better outcome.

Time frame: From baseline to 9 months post-baseline

Population: Table reports means and standard deviations. The pre-specified plan uses mixed models with full maximum likelihood estimation (intent-to-treat) and data from all 95 ATTC providers and 91 E-ATTC providers.

Count of problems as reported using an adapted version of the Global Appraisal of Needs Substance Problems Scale (Dennis et al., 2002), which has been adapted to focus specifically on problems related to opioids. The scale contains 16 items that correspond to problems related to opioid use. Patients are asked the last time they had each problem with responses including past month, past year, lifetime, or never. A count of problems experienced over the past month will be calculated for each patient. The minimum possible score is 0 and the maximum possible score is 16. Higher scores indicate higher problems, which is a worse outcome. This will be calculated for all patients who complete follow-up.

Time frame: Assessed at 3 and 6-months from patient baseline assessment

Population: Table reports means and standard deviations with listwise deletion. The pre-specified plan uses mixed models with full maximum likelihood estimation (intent-to-treat) and data from all patients who reported any follow-up data (N = 592).

Days of abstinence as reported using calendar-based recall based on the Timeline Followback Interview method (Sobell & Sobell, 1992). Days of opioid abstinence will be calculated from 0 to 30 for each patient, with higher numbers indicating more days of abstinence (which is a better outcome). This will be calculated for all patients who complete follow-up.

Time frame: Assessed at 3 and 6-months from patient baseline assessment

Population: Table reports means and standard deviations with listwise deletion. The pre-specified plan uses mixed models with full maximum likelihood estimation (intent-to-treat) and data from all patients who reported any follow-up data (N = 592).

Implementation climate scale (Jacobs et al., 2014). This scale contains 6 items scored on a 1 to 5 scale. An average score across the 6 items will be calculated per provider. Possible scores on this outcome range from a minimum of 1 to a maximum of 6. Higher scores indicate a more positive implementation climate, which is a better outcome.

Time frame: From baseline to 9 months post baseline

Population: Table reports means and standard deviations with listwise deletion. The pre-specified plan uses mixed models with full maximum likelihood estimation (intent-to-treat) and data from N=54 ATTC providers and N=58 E-ATTC providers.

Measure of leadership engagement (Garner, unpublished data). The scale contains 4 items scored on a 1 to 5 scale. An average perceived leadership engagement scale will be calculated for each provider. Possible scores on this outcome range from a minimum score of 1 to a maximum score of 5. Higher scores indicate higher perceived leadership engagement, which is a better outcome.

Time frame: From baseline to 9 months post baseline

Population: Table reports means and standard deviations with listwise deletion. The pre-specified plan uses mixed models with full maximum likelihood estimation (intent-to-treat) and data from N=53 ATTC providers and N=58 E-ATTC providers.