Screening for Unhealthy Alcohol and Other Drug Use by Health Educators: Do Primary Care Clinicians Document Screening Results?

Abstract

Objectives:

Health educators are increasingly being used to deliver preventive care including screening and brief intervention (SBI) for unhealthy substance use (SU) (alcohol or drug). There are few data, however, about the “handoff” of information from health educator to primary care clinician (PCC). Among patients identified with unhealthy SU and counseled by health educators, the objective of this study was to examine (1) the proportion of PCC notes with documentation of SBI and (2) the spectrum of SU not documented by PCCs.

Methods:

Before the PCC-patient encounter, health educators screened for SU, assessed severity (Alcohol, Smoking, and Substance Involvement Screening Test), and counseled patients. They also conveyed this information to the PCC before the PCC-patient encounter. Researchers reviewed the electronic medical record for PCC documentation of SBI performed by the health educator and/or the PCC.

Results:

Among patients with the health educator–identified SU, only 69% (342/495) of PCC notes contained documentation of screening by the health educator and/or the PCC. Documentation was found in all encounters with patients with likely dependent SU, but only 62% and 59% of encounters with patients with risky alcohol and drug use, respectively. Documentation of cocaine or heroin use was higher than that of alcohol or marijuana use but still not universal. Although all SU-identified patients had received a brief intervention (from a health educator and possibly a PCC), only 25% of PCC notes contained documentation of a brief intervention.

Conclusions:

Among patients screened and counseled by health educators for unhealthy SU, SBI was often not documented by PCCs. These results suggest that strategies are needed to integrate SBI by primary care team members to advance the quality of care for patients with unhealthy SU.

Unhealthy substance use (SU) (includes the spectrum of risky alcohol use through dependence and/or any illicit drug use through dependence) is often neither identified nor addressed in general medical settings (Hasin et al., 1990; Saitz et al., 1997). This failure occurs despite the known impact of alcohol and drug use on many common health problems (Turner, 2009) and the development of validated screening instruments to detect risky alcohol (National Institute on Alcohol Abuse and Alcoholism [2007] criteria; Bohn et al., 1995; Smith et al., 2009) and drug use (Smith et al., 2010). Although there is considerable evidence supporting the effectiveness of screening and brief intervention (SBI) counseling for decreasing risky alcohol consumption (Fleming et al., 1997; US Preventive Services Task Force, 2004; Kaner et al., 2009) and, to a lesser extent, illicit drug use (Bernstein et al., 2005), uptake of SBI has been slow (Aalto et al., 2002; Tabor et al., 2008). Reasons for underutilization of SBI include lack of time, lack of training, and uneasiness in discussing unhealthy SU with patients (Merrill et al., 2002; McCormick et al., 2006).

Team-based care in which medical staff other than the primary care clinician (PCC, ie, physician, nurse practitioner, or physician assistant) screen and counsel patients has been proposed to lessen the burden of multiple preventive agendas (Grumbach and Bodenheimer, 2004; Yarnall et al., 2009). In this model of care, medical staff, including nurses, medical assistants, and health educators (HEs), share the responsibility of delivering preventive care. Many preventive services such as checking blood pressure are delivered by non-PCC medical providers. The results of such tests are conveyed to the PCC, who acknowledges or records them and may act on them. If the blood pressure is elevated, the result is “handed off” to the PCC, who then assesses and confirms the result and acts on it as needed. Similarly, a positive screen for SU may be identified by medical staff other than the PCC, who then conveys the results to the PCC. A team-based approach such as one using an HE specifically trained to conduct SBI may result in a higher proportion of patients screened and counseled about SU (Babor et al., 2005).

Although such an approach may result in more screens and brief interventions (BI) being performed, it is unclear whether this division of labor will facilitate or discourage PCCs from addressing SU. A collaborative approach to SBI may facilitate PCC management of SU by obviating the need for PCCs to conduct SBI, allowing the PCC to focus on coordination and management of medical conditions affected by alcohol and drug use. Alternatively, screening by other clinical staff may “take it out of the hands” of the PCC, thereby diminishing PCC willingness to address SU aware of the fact that another staff member is performing that function. How or whether the “handoff” of information occurs may have implications for the care the patient receives, given that many patients are taking medications with potentially harmful interactions with alcohol or drug use (Saitz et al., 2003a; Jalbert et al., 2008). Patients who consume at-risk amounts of alcohol are more likely to need assistance with adherence to medications for common medical problems such as hyperlipidemia, hypertension (Bryson et al., 2008), and diabetes (Ahmed et al., 2006). Coordinating care is an important task highlighted in general by the National Committee for Quality Assurance and others in patient-centered medical homes (www.NCQA.org) and specifically for individuals with SU conditions (Institute of Medicine, 2006). Yet coordinating care may be more challenging for PCCs when clinical information is obtained by other clinical staff.

This study investigates care integration by examining PCC electronic medical record (EMR) notes of encounters with patients identified and counseled by HEs as having SU. Health educators used validated screening questions, conducted BI, and, if indicated, facilitated access to specialty addiction treatment. The study objectives were to assess the following among patients identified by HEs as having SU (1) the proportion of PCC EMR notes with documentation of SBI and (2) the spectrum of SU not documented by PCCs in terms of risk level and substance type (ie, alcohol, drug).

METHODS

Overview

This study was a retrospective review of PCC EMR notes of encounters with patients who screened positive for SU by HEs. We used data from the Massachusetts Screening, Brief Intervention, Referral and Treatment (MASBIRT) program, a Substance Abuse Mental Health Services Administration–funded clinical program to the Massachusetts Department of Public Health, to conduct universal SBI in general medical settings. MASBIRT employed and trained HEs whose primary (sole) responsibility was to conduct universal SBI for SU. Health educator training consisted of a standardized curriculum on alcohol and drug use, motivational interviewing, substance abuse treatment, and strategies for integrating into primary care settings, including sensitivity to patient flow issues. Training of HEs explicitly focused on integrating with the primary care medical team in part, because SBIRT services were embedded directly into the workflow of the clinic. The HE was instructed to introduce him- or herself to patients as a member of the patient’s medical “team.” Training also included self-assessment of videotaped encounters with standardized patients and direct observation of HEs in clinical encounters. After the training, HEs were periodically observed screening and counseling patients to reinforce fidelity to screening instruments and key elements of a BI. The study received approval from the Boston University School of Medicine institutional review board.

The SBI Clinical Protocol

Screenings were conducted by HEs while patients waited to be seen by their PCCs. Health educators worked closely with the medical staff to provide universal screening with minimal disruption to patient flow and spent much of their time in one location to foster familiarity between the HE and primary care staff. Patients who reported any drug (Smith et al., 2010) or heavy episodic alcohol use (more than 3 or 4 drinks in a day for men and women, respectively) (National Institute on Alcohol Abuse and Alcoholism [2007] criteria; Bohn et al., 1995; Smith et al., 2009) were then administered the Alcohol, Smoking and Substance Involvement Screening Test (ASSIST) (Humeniuk et al., 2008) to provide a measure of alcohol or drug use and level of risk associated with use. An ASSIST score of 27 or greater may indicate dependence. After administering the ASSIST, HEs conducted BI and, after discussion with the PCC, provided referrals to treatment if warranted. Brief interventions included personalized feedback, advice to change, and support of a patient’s perception of readiness and confidence to change. Health educators then completed a paper “Provider Communication Form” to convey screening results and brief counseling efforts to PCCs. This information was presented to the PCC along with a paper copy of the patient’s vital signs as was routine in this practice. The HE did not enter any record of MASBIRT screening and/or counseling into the PCC visit note (the EMR).

Study Sample

The study sample included patients of an urban, academic medical center’s primary care practice with HE-identified SU use during the period January 1 to June 30 of 2008.

Data Collection

The MASBIRT program tracks screening results for evaluation purposes in a database separate from the EMR used by PCCs. The MASBIRT database manager provided a list of all unique adult primary care patients who screened positive for SU at visits during the study period. Each patient contributed just 1 visit note to the analyses. Information was also provided about specific substances identified and ASSIST scores, all of which were collected by HEs for clinical care. A trained research assistant blinded to the specific results of MASBIRT screening reviewed PCC notes for documentation related to alcohol or drug use.

Outcomes

Study outcomes included the following:

1. “Documentation of SU” was defined as any mention of alcohol or drug use in the PCC note detected by the HE and/or PCP. References to MASBIRT screening without confirmation by the PCC, past SU, or phrases such as “no drug use” were considered to meet criteria for documentation. Criteria for SU documentation did not require that the PCC confirm or address HE screening results with the patient.

2. “Concordant documentation” was defined as mention of the same substance type (alcohol, marijuana, cocaine, or opioid) identified by HE screening and documentation of some level of unhealthy use. Because the instrument used by HEs to assess SU (the ASSIST questionnaire) is not a diagnostic tool, the severity of SU documented in PCC notes did not have to match HE assessment to establish concordance. If the PCC assessment did not agree with HE screening results, documentation was considered to be concordant if the note acknowledged HE screening along with documentation of their own assessment, similar to clinical situations in which a PCC takes into account the result of a screening test but, after doing his or her own assessment, does not agree with the results. In this case, the PCC generally documents the result of the screening test and supporting documentation of their assessment.

3. “Documentation of BI” included any of the following: advising patients to abstain from use or not to exceed recommended limits, providing feedback on the link between alcohol/drug use and health problems, or referral to treatment (Saitz et al., 2003b). PCC notes were considered to contain BI documentation regardless of whether it was performed by the PCC or HE.

Statistical Analyses

We report the proportion of PCC notes with any documentation of SU and the proportion with documentation concordant with HE screening. For descriptive purposes, frequency and concordance of documentation are presented by HE-identified substance type (ie, alcohol, marijuana, cocaine, and heroin), along with severity as determined by the ASSIST questionnaire. “Likely dependence” was defined as an ASSIST score of 27 or greater. All other unhealthy use was termed “risky use.”

RESULTS

Among 3905 unique primary care patients screened by an HE during the 6-month study period, 13% (495/3905) screened positive for SU comprising this study’s sample. Patients had the following characteristics: male (63%), self-identified as black (58%), average age 43.8 years (standard deviation = 13.8), many with a chronic medical condition potentially affected by alcohol or drug use, such as hypertension (32%), diabetes mellitus (12%), depression (13%), or hepatitis C infection (9%).

Among the sample of 495 primary care patients who screened positive for SU by an HE, 55% (270/495) screened positive for unhealthy alcohol use only, 30% (148/495) for drug use only, and 15% (77/495) for both alcohol and drug use. Among the 225 patients who screened positive for drug use (either alone or in combination with alcohol use), 76% (171/225) screened positive for marijuana, 13% (30/225) for cocaine, and 13% (29/225) for opioids, primarily heroin. As expected in a primary care population of patients with unhealthy alcohol use, substantially all screened positive for risky use (97%, 338/347) and 3% (9/347) for likely dependence. Among patients with any drug use, 93% (209/225) reported risky use and 7% (16/225) for likely dependence.

Of the 495 primary care patients with HE-identified SU, 69% (342/495) had documentation related to alcohol or drug use. Of these 342 patients, 58% had documentation of SU concordant with HE screening (ie, mention of the same substance type and any level of unhealthy use).

Primary care clinician documentation was found in 100% of patients who screened positive by an HE for likely dependent alcohol use (Table 1); however, documentation was found in only 64% (217/338) of those with risky use. Similarly, SU documentation was found in 100% of patients who screened positive for likely dependent drug use but only 59% (123/209) of patients with risky drug use.

TABLE 1.

Primary Care Clinician Documentation of Unhealthy Alcohol or Drug Use in Patients Screened by a Health Educator by Severity of Use

	Patients Identified by HE Screening	PCC Notes With Any Documentation of Alcohol or Drug use, n (%)*	PCC Notes With Concordant Documentation, n (%)†
Unhealthy alcohol use‡
Risky use	338	217 (64%)	114 (34%)
Likely dependence	9	9 (100%)	8 (89%)
Unhealthy drug use§
Risky use	209	123 (59%)	80 (38%)
Likely dependence	16	16 (100%)	13 (81%)

^*Mention of alcohol or drug use in any part of the PCC note. Percentage equals the number of PCC notes with documentation of SU/the number of patients identified by HE screening.

^†Mention of the same substance type (alcohol, marijuana, cocaine, or opioid) identified by HE screening with any level of unhealthy use. Percentage equals the number of PCC notes concordant with HE screening/the number of patients identified by HE screening.

^‡Unhealthy alcohol use defined as any heavy episodic alcohol use (more than 3 or 4 drinks in a day for men and women with ASSIST score less than 27). Likely dependence defined as ASSIST score of 27 or greater.

^§Unhealthy drug use is defined as any illicit drug use. “Risky” drug use is defined as any drug use with ASSIST score less than 27; likely dependence as ASSIST score 27 or greater.

HE, health educator; PCC, primary care clinician; SU, substance use.

Concordant documentation was quite high for those with likely dependent alcohol or drug use (89% [8/9]) and 81% [13/16], respectively), although total numbers were small (Table 1). For those with HE-identified risky alcohol and drug use, concordant documentation was only 34% (114/338) and 38% (80/209), respectively.

Table 2 presents documentation results by substance type. Cocaine and opioid use was documented in 83% of PCC notes, while documentation of alcohol or marijuana use was 65% and 34%, respectively. Concordant documentation for patients with HE-identified cocaine and opioid use was 63% (19/30) and 59% (17/29), and for patients with alcohol and marijuana, 35% (122/347) and 30% (51/171), respectively.

TABLE 2.

Documentation of Unhealthy Alcohol or Drug Use in Primary Care Clinician Notes of Patients Screened by a Health Educator by Substance Type^*

	Patients Identified by HE Screening	PCC Notes With Any Documentation of Alcohol or Drug Use, n (%)†	PCC Notes With Concordant Documentation, n (%)‡
Alcohol	347	226 (65)	122 (35)
Marijuana	171	58 (34)	51 (30)
Cocaine	30	25 (83)	19 (63)
Opioid	29	24 (83)	17 (59)

^*Unhealthy alcohol use defined as any heavy episodic alcohol use (more than 3 or 4 drinks in a day for men and women). Unhealthy drug use is defined as any illicit drug use.

^†Any mention of alcohol or drug use anywhere in the PCC note. Percentage equals the number of PCC notes with documentation of SU/the number of patients identified by HE screening.

^‡Concordant documentation was defined as mention of the same substance type (alcohol, marijuana, cocaine, or opioid) identified by HE screening and documentation of any level of unhealthy use. Percentage equals the number of PCC notes concordant with HE screening/the number of patients identified by HE screening.

HE, health educator; PCC, primary care clinician; SU, substance use.

Although HEs had counseled all patients, only 25% of PCC notes documented counseling by the HE or PCC in the form of advice to drink within recommended limits or to abstain (19%), referral for further treatment (10%), and/or feedback linking alcohol/drug to health problems (8%).

DISCUSSION

In this study of primary care patients screened positive for unhealthy SU and counseled by an HE, a substantial portion of PCC EMR notes did not contain any documentation related to alcohol or drug use. Most of the unhealthy alcohol use not documented was risky use, which is more likely to respond to BI than alcohol dependence (Fleming and Manwell, 1999; D’Onofrio and Degutis, 2002; Saitz, 2010). Documentation of cocaine or heroin use was higher than that of alcohol or marijuana use but still not universal. Most patients with unhealthy SU had no PCC EMR documentation of counseling by the PCC or HE even though HEs had delivered a BI to all patients.

These results are relevant to the rising interest in patient-centered medical homes in which care is delivered by a coordinated, proactive primary care team (National Committee for Quality Assurance, 2011). These results suggest that beyond division of labor, team-based care requires effective forms of communication between team members (Rodriguez et al., 2007). In one study, PCC unwillingness to value the results of nurse or medical assistant in screening was a barrier to team-delivery of preventive care (Aspy et al., 2008). Active efforts to coordinate team activities and improve communication between team members are needed to fully realize the benefits of team-based care (Bodenheimer et al., 2002; Grumbach and Bodenheimer, 2004). This may be particularly important when, rather than existing clinical staff, SBI services are delivered by “contracted specialists” or in this study, HEs hired and trained specifically to conduct SBI services (Bernstein et al., 2009).

We found little documentation of marijuana use. Primary care clinicians may not have documented marijuana use for a number of reasons including the perception that marijuana use is less likely to lead to abuse and dependence (Anthony et al., 1994) compared with other drug use and the concern about documentation of illicit drug use in the EMR. Although universal screening for marijuana use is not recommended by the US Preventive Services Task Force, knowing about a patient’s marijuana use may be helpful for the management of respiratory, psychiatric, and behavioral problems potentially worsened by marijuana use (Degenhardt and Hall, 2001; Moore et al., 2005; Bonn-Miller et al., 2010).

Risky alcohol or drug use was documented less frequently than more severe alcohol or drug use. Our results align with a study that found that physicians are less likely to screen and counsel patients with risky drinking than patients with alcohol use disorders (Burman et al., 2004). Possible explanations of this finding include the following: (1) risky use is less clinically obvious than SU disorders and typically requires the use of a screening test; (2) PCCs do not recognize the value of identifying and counseling for risky drinking or illicit drug use (Aspy et al., 2008); or (3) PCCs may be more reluctant to document alcohol or drug use that is not as severe as abuse or dependence.

Regardless of whether SBI is documented in primary care notes and PCCs are aware of patients’ unhealthy SU, it can be argued that using the “clinical team approach” model may result in the delivery of an effective preventive counseling to more patients than PCC-centered models of SBI. However, PCC awareness of SU is necessary for the management of common medical conditions affected by alcohol or drug use. A substantial proportion of patients with risky alcohol use are taking medications that may result in harmful interactions (Saitz et al., 2003a; Jalbert et al., 2008). Risky drinking is also associated with lower adherence to medications for hyperlipidemia and hypertension (Bryson et al., 2008) and suboptimal self-care preventive practices (Merrick et al., 2008). Furthermore, PCC knowledge of the patient’s SU change plan allows PCCs to track change in alcohol or drug consumption in future visits and, if indicated, encourage further reduction.

Future considerations for examining this model of care include whether HE screening for unhealthy alcohol or drug use may be seen as “separate” from other preventive care, potentially affirming the stigma of SU and its separation from the rest of primary care. It is possible that PCCs may be more likely to note SU screening results if “bundled” with screening for other behaviors such as exercise, nutrition, or even depression (Funderburk et al., 2007) or if SBI services were provided by existing staff such as medical assistants. Also, because there are International Classification of Disease codes for SU disorders but not for less-severe forms of unhealthy SU, PCC notation may be facilitated by the addition of International Classification of Disease codes for “unhealthy,” “risky,” or “heavy” use. It is possible that patients may have concerns about documenting risky alcohol or drug use in their medical record.

Several limitations to this study should be considered. First, failure to document SU screening results and/or BI in the PCC note may not indicate that the PCC was unaware of screening results or that SU was not addressed. Second, we examined only 1 visit note and PCCs may have addressed unhealthy SU at a future visit. However, it is less likely that the PCC would remember to address this newly identified problem (eg, risky drinking) on future visits without documentation in the visit note on the specific day that SBIRT was offered.

A third limitation of our study is that we did not measure clinical outcomes; therefore, it is unknown whether there is additional clinical benefit associated with documenting HE screening results by PCCs, augmenting BIs, or reinforcing the patient’s plan for change at future visits. Fourth, this study did not examine PCC notes of patients who were not screened by HEs so there is no comparison of rates of SU documentation without HE screening. The scope of this study was limited to examining only HE-screened patients. Finally, this study examined a model of SBI in which screening results are communicated by paper and PCC notes are recorded in an EMR. However, challenges with coordination of care by PCCs for patients screened and counseled by an HE are likely to exist when documentation of SBI is entered directly into an EMR by the HE. Future studies may examine PCC awareness of screening results when non-PCC staff document directly into an EMR.

Our study has several strengths. This study examines a model of care that may become more commonplace to decompress the PCC workload by allowing screening and counseling to be performed by other clinical team members. We used data from real-world clinical care to assess the reality of implementation of SBI in primary care. Also, this study examines SBI for illicit drug use in primary care, whereas the majority of research on SBI has focused exclusively on alcohol use.

Inattention to unhealthy SU is a missed opportunity for PCCs to address patient’s SU and improve management of medical conditions affected by alcohol and drug use. The emphasis on team-based care stems, in part, on the premise that it may be unrealistic for PCCs to address all preventive medical care mandates (Yarnall et al., 2003). With growing use of clinical teams sharing preventive screening, this study suggests that strategies are needed to integrate screening and counseling by primary care team members with PCC management of SU.