Implementation of substance use screening in rural federally-qualified health center clinics identified high rates of unhealthy alcohol and cannabis use among adult primary care patients

This ancillary study was designed with the same elements as its parent study, which was previously described [19]. Implementation was guided by the Knowledge to Action (KTA) framework, which can inform the selection and implementation of new clinical practices [26, 27]. Study activities focused on the KTA ‘action cycle,’ and consisted of adapting, implementing, and evaluating the use of substance use screening tools with implementation outcome measures. Although the implementation process in these ancillary study clinics was similar to that of the parent study, all of the study activities described herein were conducted at the Maine FQHC clinic sites. To inform the screening implementation approach, barriers were assessed through focus groups and interviews with stakeholders [22], and the EHR tools developed to support screening were tailored through multiple rounds of usability testing conducted with providers and staff at the three participating clinics.

The screening rollout was intentionally staggered, with Clinic 1 initiating first (November 2018), followed by the remaining clinics in spring 2019 (Clinic 2 in April 2019, Clinic 3 in May 2019). This schedule allowed for technical and logistical issues to be addressed in the first clinic, prior to rolling out the program to additional clinics. Following the initiation of screening at each clinic, another round of focus groups with clinic providers was conducted to understand issues related to early implementation [24].

Implementation outcomes were assessed for 12 months following initiation of screening, using EHR data gathered at the clinic level. Data collection ended in May 2020. Our primary outcome was adoption of screening, measured as the screening rate for alcohol and drug use. Secondary outcomes were the prevalence of unhealthy substance use detected via screening, and provider adoption of the EHR-integrated clinical decision support tool. The Institutional Review Boards at New York University Grossman School of Medicine and Dartmouth College approved the study. Standards for Quality Improvement Reporting Excellence (SQUIRE) reporting guidelines for healthcare quality improvement studies were followed [28].

The study was conducted in three clinics of a single rural FQHC located near Bangor, Maine. Recreational cannabis was legalized in Maine prior to the start of the study, in late 2016. Medicaid expansion was implemented in Maine in January 2019, which was during the study period. The participating FQHC is the largest in the state, with a total of nine primary care clinics (eight with adult patients, one pediatrics only) serving approximately 60,000 patients a year.

The three study clinics were chosen by the FQHC based on their geographic proximity to one another, size (medium-large clinics were preferred), use of a shared EHR system, and availability of an on-site clinical champion. Patients served by these FQHC clinics include rural, poor, unhoused, and medically underserved populations. Prior to the study, none of the participating clinics were systematically screening patients for alcohol or drug use. Table 1 identifies key characteristics of the study clinics. Primary care providers (PCPs) were physicians (MD and DO), nurse practitioners (NP), and physician assistants (PA). All clinics had integrated behavioral health services that primarily focused on mental health care (not substance use), and offered office-based buprenorphine treatment for opioid use disorder. All clinics used a Centricity™ EHR (GE Healthcare, Chicago, Illinois).

Clinics used screening program elements similar to those of the parent study, which have been demonstrated to increase adoption of screening and interventions by health care providers [29‐32]. These elements are summarized in Table 2 and include an EHR-integrated clinical decision support system (CDSS) consisting of a clinical reminder and counseling script, and use of a validated screening tool. The approach to delivering these elements of the screening program, including the visit types that were eligible for screening, frequency of clinical reminders, and content of the counseling script, were based on what was learned in the focus groups and through usability testing. A notable difference from the parent study was the screening tool used by the clinics. The parent study used the single-item screening questions for alcohol and drugs, followed by AUDIT-C and/or DAST-10 [19], whereas the ancillary study clinics agreed to use the Tobacco, Alcohol, Prescription Medication and Other Substance Use (TAPS) Tool [33, 34]. The TAPS Tool is a 2-part screener and brief assessment, the structure and scoring of which were previously reported [33, 35]. TAPS-1 identifies any past-year use of tobacco, alcohol (exceeding daily limit of 5 drinks for men or 4 drinks for women), prescription medications (non-medical use), or illicit drugs (including cannabis). TAPS-2 is administered for each substance class endorsed in TAPS-1, and identifies current (past 3 months) use and risk level for up to 9 types of substances: tobacco, alcohol (above daily limits), cannabis, opioids, illicit stimulants (cocaine, methamphetamine), heroin, and non-medical use of prescription sedatives, opioids and stimulants. Responses are summed within each substance type to generate a substance-specific risk score, ranging from 0 to 3 for tobacco and other drugs, and 0–4 for alcohol. TAPS score of 0 indicates no current use of that substance, a score of 1 indicates moderate risk use (called ‘problem use’ in the TAPS validation study), and a score of 2 indicates higher risk use [33]. For example, an individual who reports at least 1 day of alcohol use above the recommended daily limit on TAPS-1 would be administered the TAPS-2 alcohol items. If they reported on the TAPS-2 no alcohol use above the daily limits in the past 3 months they receive a score of 0 for alcohol; with any use above the daily limits their score is 1; and use above the daily limits plus one problem related to alcohol generates a substance specific alcohol score of 2.

The clinics chose to screen for substance use only during adult annual preventive care visits, which all adult primary care patients are intended to complete once per year. They opted for patient self-administered screening on tablet computers, conducted in the waiting room prior to the PCP visit, as the predominant screening approach. Tablets were managed by clinical staff, and the clinics were responsible for maintaining and operating them. Because the Centricity^™ EHR did not accommodate patient-facing questionnaires, the clinics contracted with an outside software development company (PatientLink© [PatientLink Enterprises, Oklahoma City, OK]) to build the tablet-based screener that would be used by patients. The software allowed patients to complete the TAPS Tool as well as depression screening (Patient Health Questionnaire (PHQ)-2/9) on the tablet, with results seamlessly uploaded into the EHR at the point of care. TAPS Tool screening results were visible to the PCP in a study-designed clinical decision support system (CDSS) integrated into the EHR. To access the CDSS, providers clicked on a tab labeled ‘TAPS’, located on the home screen just beneath the tab for viewing vital signs. The CDSS landing page presented a summary of the screening results, from which providers could click on an ‘intervention’ tab that provided guidance for conducting and documenting a brief motivational counseling intervention, an option for selecting patient education handouts, and an order set for treatment referrals. Provider use of the CDSS was recommended for patients with moderate or high risk use of alcohol and/or drugs.

All clinics used the same implementation strategy, elements of which are summarized in Table 2.

Clinical champions: Clinical champions can be integral to enacting practice change, by supporting, marketing, and overcoming barriers to implementation [36, 37]. In our study, each clinic had a clinical champion who was a practicing PCP at their site. Champions facilitated the integration of screening into local workflows, provided ongoing support and education to the clinical staff, identified problems related to adoption and workflow, and met approximately once/month with the local research team to monitor implementation.

Practice facilitation: Practice facilitation is an implementation strategy wherein trained facilitators work with clinic leaders and staff to implement evidence-based practices [38‐40]. Practice facilitation was led by one staff member from the FQHC’s quality improvement unit, who was a former medical assistant in one of the study clinics, and assisted by a research assistant. All clinics received practice facilitation for approximately 3 months prior to the start of the screening program, and on an ongoing basis throughout the first year.

Training: Training was led by a psychiatrist from the FQHC (TG or VA) who was a clinical leader of substance use services. All clinical staff received a one-hour group training session that covered screening and brief intervention for unhealthy substance use, and use of the EHR-integrated screening tools. Additional training for medical assistants and front desk staff was done by the practice facilitator. Medical assistants were trained on administering the screening instrument to patients who needed assistance and accepting results from the tablets into the EHR. Front desk staff were trained on using the clinical reminder, asking patients to complete screening prior to the visit, and on distribution and collection of the tablets.

Implementation data were extracted from the EHR for the first 12 months of screening at each clinic. Summary reports of the tracked implementation outcomes were generated weekly for the first 3 months of screening at each clinic, and then quarterly. Reports were sent to the investigators and then reviewed by the practice facilitator, clinical champion, and research team at each site to inform ongoing implementation. Implementation reports allowed the research team and practice facilitators to identify unanticipated technical or logistical issues arising in the implementation process (e.g. data extraction codes missing information on a specific substance, or workflow issues leading to sudden drops in screening rate at a specific clinic). Each data extract included the (1) number of patients eligible for screening (adults 18 years or older, who had an annual preventive care visit in the given time period), (2) number of patients screened, (3) TAPS Tool results indicating the prevalence of moderate and high risk use of tobacco, alcohol, cannabis, other illicit drugs, and prescription medications, and (4) provider use of the CDSS counseling script.

The FQHC sent EHR data extracts capturing screening implementation outcomes to the study’s data and statistical coordinating center (The Emmes Company) throughout the first 12 months. Demographic characteristics were extracted after the end of the study period. Descriptive statistics were calculated to characterize the rates and frequencies of the pre-defined implementation outcomes and patient demographics. Screening rate was calculated as the proportion of eligible patients presenting for an annual preventive care visit who completed the TAPS-1 screener. Prevalence of moderate and high risk use was calculated based on the TAPS score, using standard cutoffs (1 + for moderate risk, 2 + for high risk). Because the counseling script was only suggested when patients screened positive for moderate or high risk use of alcohol and/or drugs, the rate of counseling was calculated based on the moderate-high risk population. All analyses were performed using SAS version 9.4.

Our study has limitations stemming from its origins as an implementation feasibility study. Clinics were not randomized, and so it is not known to what extent the screening rates observed during the study period were attributable to the screening program versus reflecting an evolution of practice that may have occurred even without these implementation efforts. Patients with screening-eligible visits during the implementation year were more likely to have private insurance than in the general clinic population, suggesting some limitations in generalizing these findings to all patients within the health system. Maine’s Medicaid expansion occurred during the study, which could have introduced changes in primary care utilization that were not anticipated. All clinics used the same set of implementation strategies, and so we were unable to discern which strategies were most impactful. Similarly, because all clinics used the TAPS Tool for screening, we could not test the hypothesis that the screening tool was partially responsible for the higher prevalence of drug use detected here, in comparison to the parent study. The only measure of provider counseling was documentation in the study-provided CDSS, and it is possible that PCPs were having discussions of screening results without documenting it there. Similarly, treatment referrals were only tracked if they were made using the CDSS, and informal referrals or those made by other clinic staff, such as behavioral health providers, were not captured. Conducting chart reviews or using natural language processing to analyze clinical notes may have captured this information, but were outside the scope of our study. Finally, as clinics from a single FQHC in northern Maine, the study clinics are not representative of all rural primary care practices. Notably, the predominantly White patient population served by these clinics is not generalizable to the rural population of the U.S.