The American Journal of Managed Care October 2006
Implementation of Evidence-based Alcohol Screening in the Veterans Health Administration
Background: Despite evidence-based guidelines, brief alcohol screening and counseling have not been routinely integrated into most primary care practices in the United States.
Objective: To describe the results of the implementation of evidence-based alcohol screening by the Veterans Health Administration (VA) in 2004, as the first step toward implementation of brief alcohol counseling.
Study Design: This observational study of outpatients from all 21 VA networks relied on the following 2 data sources from the VA Office of Quality and Performance: (1) Medical record reviews, designed to compare VA networks quarterly, evaluated whether established VA patients had documented screening for alcohol misuse and documented follow-up assessment for alcohol use disorders among those who screened positive for alcohol misuse (January-March 2005); and (2) Mailed patient satisfaction surveys from 2004, which oversampled patients new to the VA (response rate, >70%), included the Alcohol Use Disorders Identification Test-Consumption (AUDIT-C) questions and asked about past-year advice "to drink less or not to drink alcohol" from a VA provider.
Results: Based on 10 115 medical record reviews, 93% (range, 89%-96% across networks) of outpatients were screened for alcohol misuse, and 25% (range, 11%-36%) screened positive. Among screen-positive patients, 42% (range, 5%-84%) had documented follow-up assessment, but absolute numbers of screen-positive patients evaluated were small (27-80 patients per network). Based on 235 481 patient surveys, the prevalence of alcohol misuse was 22% (range, 15%-27% across networks), and 28% (range, 20%-36%) of screen-positive patients reported receiving alcohol-related advice. Alcohol-related advice increased as AUDIT-C scores increased.
Conclusion: The VA successfully implemented evidence-based alcohol screening, but the rate of follow-up among screen-positive patients remained low.
(Am J Manag Care. 2006;12:597-606)
Alcohol misuse is common and accounts for about as much death and disability globally as tobacco and hypertension.1 More than 8% of the US population experience alcohol abuse or dependence,2 and an additional 30% drink at risky levels (definitions of alcohol misuse are given in Table 1).3 Alcohol screening followed by brief alcohol counseling has been demonstrated to be efficacious in randomized controlled trials4,5 and has been identified as a national prevention priority,6-8 but efforts at widespread implementation of these practices have not been successful.9
The Veterans Health Administration (VA) recently replaced a program of annual screening for alcohol use disorders10 with a program of annual screening for alcohol misuse (risky drinking and alcohol use disorders) as the first step toward implementation of evidence-based brief alcohol counseling. This article describes the adoption and nationwide implementation of an alcohol misuse screening performance measure in 2004. Although no formal prospective evaluation of this program was conducted, we present results of performance monitoring regarding alcohol screening and follow-up for the 21 VA networks through March 2005, comparing results based on medical record reviews and patient surveys.
THE SPECTRUM OF ALCOHOL MISUSE
Since the early 1980s, there has been increasing recognition that most individuals who experience alcohol misuse are not alcohol dependent. Before then, medical providers focused predominantly on the diagnosis and referral of patients with alcohol dependence. In 1990, a seminal report noted that, because patients with nondependent risky and problem drinking outnumber those who meet diagnostic criteria for alcohol dependence, adverse outcomes from the former group constitute a greater burden to society.11
Efficacy of Brief Alcohol Counseling in Primary Care
In the 1980s and 1990s, recognition of the importance of addressing the full spectrum of alcohol misuse led to research aimed at decreasing risky and nondependent problem drinking. British researchers conducted the first randomized controlled trial to demonstrate the efficacy of brief alcohol counseling by primary care providers.12 Later, Project TrEAT (Trial for Early Alcohol Treatment) replicated the study in the United States.13 The counseling interventions in these trials consisted of 5 to 20 minutes of patient-centered counseling that typically included a detailed assessment of drinking and readiness to change, feedback linking alcohol use to health, explicit advice, negotiation of a drinking goal, and follow-up.4,6 Project TrEAT demonstrated a significant 11.5% decrease in the proportion of patients with risky drinking 12 months after brief alcohol counseling and a significant decrease in subsequent hospital utilization.13 After 4 years, for every $1.00 spent on screening and brief alcohol counseling, $4.30 on average were saved on inpatient and emergency care, whereas $39.00 were saved from the societal perspective.14 Meta-analyses have confirmed the efficacy of brief alcohol counseling for decreasing drinking among outpatients,4 although some authors have noted that low levels of recruitment for these trials may limit the generalizability and broader benefits of brief alcohol counseling in practice.15
Implementation of Brief Alcohol Counseling in Trials
Implementation of brief alcohol counseling in routine clinical settings has been hampered by the complexity of the procedures required to implement such counseling in randomized controlled trials. Primary care trials demonstrating its efficacy have used research staff to (1) screen for alcohol misuse and further assess patients who screen positive, (2) educate providers, and (3) ensure that patients who screened positive were offered appropriate counseling.
Alcohol Misuse Screening and Assessment. Trials have generally used the CAGE questionnaire (an acronym indicating cut down on drinking, annoyed by complaints about drinking, guilty about drinking, and had an eye-opener [drink] first thing in the morning) supplemented with 3 other questions12,13,16 or with the World Health Organization's (WHO) 10-item Alcohol Use Disorders Identification Test (AUDIT).17 The CAGE, a validated screening questionnaire for alcohol use disorders, has not been used alone in these trials because it does not screen for risky drinking. After screening, research staff further assessed patients who screened positive to confirm risky drinking and to exclude patients with severe problems indicating a need for more intensive interventions.
Provider Education. Many providers are unfamiliar with the broad target population for brief alcohol counseling.18,19 Therefore, researchers have educated providers about the shift from a narrow focus on alcohol dependence to a broader focus on the whole spectrum of alcohol misuse, as well as about specific components of brief patient-centered alcohol counseling.18,19
Ensuring That Brief Alcohol Counseling Occurred. Researchers have often scheduled a special patient visit with the primary care provider for alcohol counseling.13 Alternatively, patients have been referred to health counselors to ensure that brief alcohol counseling was offered to patients with alcohol misuse who were randomized to receive the intervention.20,21
Implementation Studies of Brief Alcohol Counseling
Published studies22-35 of efforts to implement brief alcohol counseling in nonresearch settings, most conducted outside the United States, have had disappointing results. Trials in England,22,24 Finland,25 Sweden,26 and Australia27 and the multicountry WHO study28 have tried (1) marketing programs to providers, (2) educating and coaching providers, (3) screening patients and prompting providers with results (sometimes along with educational patient handouts), and (4) providing small financial incentives. Except for a study29 combining the first 3 of these components, no strategy (to our knowledge) has markedly increased the rates of screening and brief alcohol counseling.25-27 In a large study,30 implementation was so limited that 20% was considered a high level of alcohol screening and 10% a high level of brief alcohol counseling. Sustained implementation was rare once research studies ended.29
Implementation studies31-35 of brief alcohol counseling in the United States have focused primarily on screening patients and on prompting providers with screening results or having other staff conduct brief alcohol counseling, and results have been marginal or negative. The most promising study16 trained providers for 3 hours and placed a detailed algorithm on the medical record as a prompt when a patient screened positive for alcohol misuse. Unfortunately, such detailed paper algorithms are unlikely to be practical in nonresearch clinical settings with multiple preventive agendas.36 A recent implementation study9 in 5 managed care settings screened fewer than 25% (median, 54% [range, 0%- 95%]) of patients and observed variability across sites in the rates of brief alcohol counseling among patients who screened positive for alcohol misuse.
No study, to our knowledge, has confirmed the effectiveness of alcohol screening and counseling integrated with other recommended preventive screening and counseling in a nonresearch setting, because no such program has been successfully implemented.37 Electronic medical record systems and computerized clinical reminders have been proposed38 but not evaluated as a means to implement routine alcohol screening and counseling. Finally, we know of no study that has evaluated a stepped approach to implementation, sequentially implementing alcohol screening, provider education, and incentives for follow-up brief alcohol counseling.
SYSTEMS FOR IMPLEMENTING PREVENTIVE CARE IN THE VA
VA Computerized Patient Record System (CPRS)
CPRS has been implemented at all 21 VA networks, each made up of 3 to 10 VA facilities and up to 47 community-based outpatient clinics, which are often freestanding (non-VA) clinics that contract to also provide care for VA outpatients. CPRS includes results reporting, note writing, and order entry components, as well as computerized clinical reminders. A version of CPRS software is being made available to non-VA practices.
CPRS Clinical Reminders
Computerized clinical reminders in CPRS prompt clinicians to provide evidence-based care, and some also incorporate decision support, educating clinicians and guiding them through evidence-based algorithms. These reminders insert prespecified text into the clinician's electronic progress note, thereby documenting that the desired care has been provided. Actions taken in one clinical reminder (eg, documenting a positive screen for alcohol misuse) can trigger another reminder (eg, following up on brief alcohol counseling).
The VA's current clinical reminder system is passive: reminders do not pop up automatically. Instead, providers must be motivated to open a "reminders" button after they open a progress note in CPRS to view all clinical activities that are due. Reminders remain due until a required clinical action turns it off. Although some VA clinical reminders have been developed and disseminated nationally, most have been developed locally or adapted from reminders developed at other facilities. At some facilities, completion of clinical reminders is required for all providers; at others, clinical reminders have been implemented as an optional tool or are used predominantly by nursing staff.39
VA Performance Measures
Since 1996, the VA has established mandatory performance measures for all VA networks. A national Performance Measures Work Group recommends to the VA's Under Secretary for Health which performance measures should be implemented each year, based on the importance of the condition in the VA population, the magnitude of the gap between current and "best" practices, the strength of supporting evidence for a practice, and the feasibility of nationwide monitoring. The Under Secretary for Health determines which measures will be included in the performance contracts of VA executives. The directors and key top managers of the 21 regional VA networks and 157 VA medical centers have personal salary incentives that are contingent on performance. All VA facilities, including 696 community outpatient clinics, must then implement practices to meet performance targets.
VA IMPLEMENTATION OF EVIDENCE-BASED ALCOHOL SCREENING
Identifying Alcohol Screening as a Priority