Problematic alcohol use is a spectrum ranging from excessive consumption without consequences to alcohol use disorder (AUD) (1) The United States spends $26 billion annually to treat alcohol use and its consequences (2). Although two third of patients with problematic alcohol use see a healthcare provider annually, only 24% have their drinking assessed (3). Prevalence estimates for problematic alcohol use in critically ill patients vary from 19% to 40% (45), and of ICU patients who are regular drinkers up to 65% meet criteria for AUD (6). In total, around 1.4 million patients with problematic alcohol use are discharged from U.S. ICUs annually (7), and 40–56% of these patients will return to the hospital within 1 year (89).
Screening, Brief Intervention, and Referral to Treatment (SBIRT) (10) is a paradigm to assess and address problematic drinking. SBIRT has been extensively studied in healthcare settings (10–13). SBIRT involves using a validated questionnaire to screen all patients for alcohol use. When problematic use is identified, whether or not it rises to the level of AUD, a brief intervention is used to motivate change. Motivational interviewing (MI) is a common brief intervention used in SBIRT (1214). In a typical MI conversation, a provider and patient spend 15–30 minutes weighing the pros and cons of drinking, developing a goal, and formulating a plan. Ideally, the patient will choose to reduce alcohol consumption or achieve abstinence. Although SBIRT produces modest reductions in alcohol consumption and consequences, there is significant controversy regarding its efficacy in inpatient settings (1516).
SBIRT has been studied in primary care, trauma, and emergency departments (1617), but not in ICU survivors. Several characteristics of this population make it difficult to extrapolate findings from other settings (18). First, providers may not offer SBIRT because substance use is perceived to be less important than patients' other medical conditions (13), although in fact severity of illness is the best predictor of ICU survivors' readiness to change (6). Second, SBIRT implementation is inconsistent: in a study at three hospitals using SBIRT, only 72% of patients with problematic alcohol use were successfully identified, only 50% received an intervention, and just 34% had any follow-up (6). Finally, semistructured interviews with ICU patients and families suggest barriers due to mental health or cognitive concerns, including temporary delirium while hospitalized. Patients reported that caring providers and family members' opposition to drinking were key facilitators of change. Additionally, patients said their willingness to stop drinking was increased based on religious faith, events related to hospitalization, and accumulating health problems (18). As a whole, these findings suggest missed opportunities for SBIRT with critically ill patients and suggest that an alternative approach is needed.
RECOVERY NAVIGATOR: A NOVEL SBIRT INTERVENTION FOR CRITICALLY ILL PATIENTS
Based on these findings, we created a Recovery Navigator role to deliver SBIRT in the ICU. The navigator was a paraprofessional trained in MI, who assessed barriers and facilitators and provided a tailored intervention to ICU patients. A specially trained Recovery Navigator helped to ensure systematic SBIRT delivery despite the barriers reported by ICU clinicians.
This intervention combined several theoretically pertinent components (Fig. 1): first, the Recovery Navigator was trained on MI (19) using a well-established learning approach (20). Second, the navigator employed principles from self-determination theory, which integrates well with MI (21) and suggests the importance of patient-provider relatedness, patient autonomy, and perceived competence. Third, to enhance autonomy and competence, the navigator used decision support tools including a written summary of the conversation. Fourth, to facilitate referrals, the navigator built relationships with community organizations that treat AUD. Finally, the navigator was trained to capitalize on factors that make critical illness a “teachable moment” for patients with problematic alcohol use, such as patients' recognition of the seriousness of their medical condition, family support networks mobilized in response to a health crisis, or religious support. Based on these components, the Recovery Navigator intervention is best described as a MI and shared decision-making (MI-SDM) model of care.
The Recovery Navigator was hospital based, which facilitated initial contact in the ICU. Because some patients had delirium, we hypothesized that longitudinal contact after hospital discharge might be beneficial. Repeated contact would permit time for patients' cognition to improve, allow patients to weigh the pros and cons of alcohol use, and let patients consider treatment. Conversely, ICU patients might be motivated enough for a single session to help due to factors like medical severity and mobilized social support. Because the need for follow-up contact was unclear, we designed a pilot trial of both single-session and multisession MI-SDM.
The goal of this study was to engage critically ill patients with problematic alcohol use in treatment. A three-arm pilot randomized trial was designed to test the Recovery Navigator strategy: usual care, a single MI-SDM session, or repeated MI-SDM sessions. This preliminary study was intended to determine the feasibility and acceptability of MI-SDM and to suggest whether a future full-scale efficacy study is warranted. We considered patient recruitment, demographics, randomization, attrition, treatment fidelity, and descriptive results on potential outcomes.
MATERIALS AND METHODS
Setting and Participants
This Institutional Review Board–approved study was conducted in two urban medical ICUs. Patients were screened with the Alcohol Use Disorders Identification Test (AUDIT)-Concise (22). Adult men (≥ 18 yr) with a score greater than or equal to 4 or women with a score greater than or equal to 3 were enrolled. Patients were eligible only after their critical illness and any delirium were resolved. Patients were excluded if pregnant, unable to speak and read English, unable to provide consent, or expected to survive less than 6 months.
Screening and Enrollment
Medical ICU patients were screened for eligibility daily by a research coordinator, who administered the AUDIT-C and obtained informed consent.
Patients were randomized in blocks of six for equal groups, stratified by hospital. The coordinator randomized patients using a computer-generated algorithm.
It was impossible to blind patients or the navigator, but the research assistant who collected outcome data was blinded to treatment allocation, and a different person randomized patients. Patients were instructed not to tell the research assistant which treatment they received, and there was no evidence that the blind was broken.
Initial Patient Contact
The study coordinator who enrolled patients collected baseline data and provided a $20 gift card. Patients randomized to either intervention group then completed their first session with the navigator. In single-session MI-SDM, patients worked with the navigator to develop a recovery plan including referral to treatment if appropriate; there was no additional contact with the navigator. In multisession MI-SDM, patients developed plans that also included one or more follow-up navigator sessions.
At enrollment, the coordinator collected contact information and scheduled follow-up visits. Reminders were sent before 3- and 6-month follow-ups, with cab vouchers or bus tokens if needed. The 6-month follow-up could also be completed by phone.
Intervention and Comparison Groups
About half of patients received SBIRT, usually including a short discussion about alcohol use (not with the Recovery Navigator) and a list of treatment resources. Usual care could include a range of strategies from brief advice to MI, but was not standardized. The study team had no contact with patients beyond collecting data.
One hospital-based MI-SDM session was delivered by the Recovery Navigator, a White woman with a master’s degree in social work and prior experience treating alcohol use and using MI. The navigator asked permission to discuss alcohol use, then reviewed AUDIT results and AUD symptoms. She used MI to help patients weigh pros and cons of drinking and offered feedback about connections between the patient’s alcohol use and current illness. Finally, she asked patients to verbalize a plan. If this included a desire for treatment, she helped patients compare options including self-change and facilitated referrals.
Patients completed a similar first session with the same Recovery Navigator. Then the navigator 1) followed up with patients after discharge; 2) provided several MI-SDM sessions on a schedule tailored to patient preferences; and 3) offered instrumental support (e.g., transportation vouchers) to help patients execute their plans. Sessions were suggested weekly in the first month, then monthly, for at least six visits over 3 months. Follow-up sessions were similar to the first, using MI if patients continued to drink and facilitating treatment decision-making if not. The navigator provided referrals from Substance Abuse and Mental Health Services Administration’s treatment locator (https://findtreatment.samhsa.gov/) and contacted resources for warm handoffs.
Feasibility indicators included: 1) recruitment, defined as the number of patients enrolled per month; 2) patients' willingness to be randomized; 3) baseline equivalence of groups; 4) attrition within each group; and 5) number and length of MI-SDM visits.
The Client Satisfaction Questionnaire (CSQ)-8 was used to assess acceptability. This scale’s eight items are rated on a four-point Likert-type scale and load on a single dimension. The instrument has excellent internal consistency (α = 0.83–0.93) (23).
Fidelity was enhanced via MI training, plus interactions with 11 trial patients that were rated by expert MI coders. Feedback was shared with the navigator. After randomization began, 50% of sessions were recorded and reviewed using the MI skills code (MISC) (24). Patients also completed four theory-related scales: 1) the 10-item Scale To Assess the therapeutic Relationship (STAR) alliance measure (25); 2) the Autonomy-Healthcare Climate scale (26); 3) the Perceived Competence Scale (27); and 4) a Treatment Motivation Scale.
Potential outcomes for a larger-scale efficacy study were assessed to confirm the utility of measures and the potential of MI-SDM to generate change. The primary outcome was the 30-day Alcohol Timeline Follow-Back (28), which measures abstinence from alcohol, days abstinent, and heavy drinking days. Secondary outcomes were healthcare utilization (inpatient readmission, outpatient visits); alcohol treatment; and psychiatric treatment, all based on the Treatment Services Review tool (16). Alcohol-related problems were assessed via the Short Inventory of Problems-2, revised (SIP-2R) (29), Hospital Anxiety and Depression Scale (30), Patient-Reported Outcomes Measurement Information System Applied Cognition scale (31), and Impact of Events Scale (32) for trauma. Patients' self-reported legal problems and employment were also assessed. Finally, social support was measured with the important people interview (33).
We examined descriptive data including group-level means or proportions and CIs. We considered between-group differences in outcomes based on a completers analysis rather than intent-to-treat (ITT), both because attrition was considered separately and because usual care had the most attrition and was artificially disadvantaged using ITT. However, we did reanalyze results using ITT in a sensitivity analysis, and conclusions were unchanged. Consistent with the approach to pilot studies proposed by Leon et al (34), we did not calculate a priori power or use inferential tests. Instead, we examined the magnitude and direction of any between-group differences to draw conclusions about the usefulness of measures and the desirability of future studies on MI-SDM.
From January 2017 to June 2018, we screened 111 patients and enrolled 47. Of 64 who refused randomization, major reasons were stigma (e.g., worried about family reactions), being too busy, or wanting to solve problems independently. The first 11 patients were used to refine MI-SDM and are not presented; the remaining 36 were randomized and included in all analyses.
Table 1 shows demographics by group. Participants' median age was 44 (range, 28–74), 60% were male, and 50% were White non-Hispanic. Their median AUDIT-C score was 10 (range, 3–12), and 24/36 had scores suggesting AUD. There were no pretreatment differences (ps > 0.12), except for higher frequency of chronic obstructive pulmonary disorder or pneumonia (×2 = 6.55; p = 0.04) and more comorbidities (M = 2.67) in single-session MI-SDM than in the other two groups (M = 1.42–1.91; F(2, 32) = 3.42; p = 0.45). However, Acute Physiology and Chronic Health Evaluation severity scores did not differ between groups, nor did baseline alcohol use (ps > 0.48). Overall, randomization produced roughly equivalent groups although single-session MI-SDM recipients were slightly more medically ill.
Patient flow is shown in Figure 2: 42% of patients accepted help, which is notable among patients not seeking alcohol treatment. There was 31% attrition by 6 months, distributed evenly across groups (×2 = 2.25; p = 0.32). Completers were equivalent to dropouts on most demographics (ps > 0.09), although women were more likely to remain in the study (×2 = 9.97; p = 0.002). Despite attrition, all single-session MI-SDM patients received one session, and multisession MI-SDM participants completed an average of 4.92/6 sessions (sd = 5.98). Session length was similar for single-session (M = 40.8 min) and multisession MI-SDM (M = 42.2 min).
The mean CSQ-8 score was high (M = 27.8/32; range, 2–32), with no differences across groups. CSQ-8 scores were not strongly correlated with age, gender, race, housing, income, education, employment, or comorbidities (rs < 0.29). Satisfaction was lower for Latino/Latina patients (r = –0.32) and higher for patients with AUD (r = 0.21). No adverse events were reported in any group.
After the training phase, 98% of navigator utterances were MI consistent. Median MISC global ratings on a 1–7 scale were as follows: seven for acceptance (range, 5–7), six for empathy (range, 4–7), five for MI spirit (range, 4–6), and six for client self-exploration (range, 4–7). These results suggest high fidelity to both the techniques and the spirit of MI.
Patients rated both MI-SDM groups more autonomy supportive than usual care (Table 2), with slightly higher scores for multisession than single-session MI-SDM. There was a small difference on the Perceived Competence Scale, favoring the two MI-SDM conditions. Patient-provider relatedness and readiness for treatment were similar across groups. The group differences on autonomy and competence support are consistent with self-determination theory.
Group means and CIs for each outcome are shown in Table 2. All differences were in the expected direction. Based on descriptive statistics, patients had more days abstinent, less heavy drinking, fewer alcohol problems, lower depression, anxiety, and posttraumatic stress disorder symptoms, fewer cognitive problems, more days worked, fewer legal problems, and less healthcare utilization in the MI-SDM groups than with usual care. CIs were large and overlapped because of the small sample, so these preliminary results cannot be used to draw conclusions about MI-SDM’s efficacy. Even though the effects were clinically meaningful in magnitude (e.g., 5–9 fewer heavy drinking days per month), we cannot be certain whether they are replicable because of this study’s small sample size. Interestingly, there was no evidence that more patients received substance use treatment after MI-SDM, and no difference in rehospitalization. Any benefits, therefore, may be due to MI-SDM itself rather than the navigator’s facilitation of substance use treatment.
Results suggest that MI-SDM delivered by a Recovery Navigator is feasible and acceptable for brief substance use intervention in intensive care. Diverse urban hospital ICU patients with problematic alcohol use were willing to receive MI-SDM. About two thirds had AUD, and all had significant illness. Although attrition was 31%, patients received interventions as intended, on average one contact lasting 40 minutes in single-session MI-SDM, or five sessions lasting 42 minutes in multisession MI-SDM. Thus, the intervention was feasible even though some patients did not provide follow-up data. Patients lost to follow-up and completers were similar on all variables including alcohol use severity, which reduces concern about bias due to attrition. Patients were equally satisfied with both MI-SDM conditions and with usual care.
The intervention was delivered with high fidelity to MI principles based on observational data and patient ratings. Preliminary data suggested that autonomy and competence are particularly important to ICU patients with problematic alcohol use, and group means confirmed that MI-SDM was different from usual care on these variables. The two MI-SDM groups were also associated with higher patient ratings of support from significant others 6 months after enrollment. These findings support the theoretical basis for MI-SDM. Contrary to expectations, the groups did not differ on relatedness or readiness for treatment.
Patients receiving MI-SDM had more abstinence from alcohol, fewer heavy drinking days, and fewer alcohol-related difficulties than usual care patients based on descriptive statistics. None of these findings were statistically significant due to small sample size, and the efficacy of MI-SDM is therefore still unknown, but differences between means were in the expected direction and clinically meaningful in magnitude. A full-scale randomized trial is needed to determine whether these promising initial results of MI-SDM are replicable. Both single-session and multisession MI-SDM are potentially viable for future studies.
This was a pilot study with a small sample. Between-group differences did not meet conventional criteria for significance but were large enough in magnitude to be clinically meaningful. The current evidence is thus suggestive but not conclusive, and the primary finding of this study is that MI-SDM is viable for further testing. To detect differences on the days abstinent measure with α of 0.05, the same three groups, and similar intervention effect sizes, we would have needed 46 participants per group to achieve 80% power. Achieving 80% power to detect effects on other outcomes would have required similar sample sizes: 41 participants per group for heavy drinking days, 30 for abstinence, and 28 for the SIP-2R measure of alcohol-related problems, These sample sizes should be readily achievable in a single-site randomized controlled trial. Attrition was a partial limitation as described above. We used a completers analysis rather than ITT, but a sensitivity analysis using ITT only exaggerated MI-SDM’s effects. The fact that patients who dropped out were similar to those who remained is reassuring, but differences related to attrition might have emerged in a larger sample. Similarly, even though randomization produced equivalent groups at baseline, a larger sample might have revealed failures of randomization. These limitations can be addressed in a full-scale trial.
Although 47 of the 111 patients approached for this study agreed to participate, 64 refused, with major reasons being stigma or desire to handle the problem themselves. Additionally, we identified screening for substance use as a gap in regular ICU care even at hospitals that have policies to promote it. Both of these findings suggest a need to improve recruitment in future studies or routine clinical interventions to address substance use among critically ill patients. More automated screening procedures such as automatic flags for alcohol problems in electronic health records and prompts or queries to identify patients with problematic alcohol use could facilitate future recruitment. Future trials also might randomize at the unit level to capitalize on clinician enthusiasm for addressing substance use in units where it exists. It might also be possible to train ICU clinicians on basic MI strategies for use in linking patients to studies or interventions that address substance use.
Finally, groups differed on only three of five hypothesized mechanisms of action for MI-SDM. MI-SDM supported patients' autonomy and perceived competence, as well as family engagement. However, relatedness and treatment readiness were similar across groups, and MI-SDM patients were no more likely to receive alcohol treatment. In fact, the navigator reported that many patients declined referrals precisely because they valued autonomy and wanted to change on their own. Alternately, patients might have seen MI-SDM itself as a form of treatment. The balance between autonomy support and treatment facilitation is an important issue for future research.
ICUs serve many patients with problematic alcohol use, but providers do not adequately address alcohol use. A Recovery Navigator was a feasible and acceptable way to offer SBIRT in the ICU setting. About 40% of patients offered the intervention accepted it, which is a positive result among patients who may not recognize they have a problem. Once patients enrolled, they were highly satisfied with the support they received. MI-SDM promoted patients' autonomy and competence and patients in MI-SDM reported less alcohol use and improvement on symptoms and role-functioning metrics, although between-group comparisons were nonsignificant due to small sample size. MI-SDM did not increase referral to treatment. MI-SDM should be tested in a full-scale clinical trial to address alcohol-related health risks in a vulnerable ICU population.
1. Saitz R. Clinical practice. Unhealthy alcohol
use.N Engl J Med2005352596–607
2. Harwood HJ. Updating estimates of the economic costs of alcohol
abuse in the United States: Estimates, update methods, and data2000Bethesda, MDNational Institute on Alcohol
Abuse and Alcoholism
3. Weisner C, Matzger H. Missed opportunities in addressing drinking behavior in medical and mental health services.Alcohol
Clin Exp Res2003271132–1141
4. Clark BJ, Williams A, Feemster LM, et al.; NHLBI ARDS Network InvestigatorsAlcohol screening scores and 90-day outcomes in patients with acute lung injury.Crit Care Med2013411518–1525
5. Marik P, Mohedin B. Alcohol
-related admissions to an inner city hospital intensive care unit
6. Clark BJ, Smart A, House R, et al. Severity of acute illness is associated with baseline readiness to change in medical intensive care unit
patients with unhealthy alcohol
Clin Exp Res201236544–551
7. Kim MM, Barnato AE, Angus DC, et al. The effect of multidisciplinary care teams on intensive care unit
mortality.Arch Intern Med2010170369–376
8. Clark BJ, Keniston A, Douglas IS, et al. Healthcare utilization in medical intensive care unit
survivors with alcohol
Clin Exp Res2013371536–1543
9. Clark BJ, Rubinsky AD, Ho PM, et al. Alcohol
screening scores and the risk of intensive care unit
admission and hospital readmission.Subst Abus201620161–8
10. McQueen J, Howe TE, Allan L, et al. Brief interventions for heavy alcohol
users admitted to general hospital wards.Cochrane Database Syst Rev20118CD005191
11. Graham LJ, Davis AL, Cook PF, et al. Screening, Brief Intervention, and Referral to Treatment in a rural Ryan white part C HIV clinic.AIDS Care201628508–512
12. Cook PF, Aagaard L, Bowler F, et al. Screening, Brief Intervention, and Referral to Treatment: Nurses helping Colorado training program.J Nurs Educ201857476–482
13. Rosenthal L, Barnes C, Aagaard L, et al. Initiating SBIRT, alcohol
, and opiate training for nurses employed on an inpatient medical-surgical unit: A quality improvement project.MEDSURG Nursing201827227–230
14. Grossberg P, Halperin A, Mackenzie S, et al. Inside the physician’s black bag: Critical ingredients of brief alcohol
15. Saitz R. Candidate performance measures for screening for, assessing, and treating unhealthy substance use in hospitals: Advocacy or evidence-based practice?Ann Intern Med201015340–43
16. Saitz R, Palfai TP, Cheng DM, et al. Brief intervention for medical inpatients with unhealthy alcohol
use: A randomized, controlled trial.Ann Intern Med2007146167–176
17. Saitz R. Alcohol
screening and brief intervention in primary care: Absence of evidence for efficacy in people with dependence or very heavy drinking.Drug Alcohol
18. Clark BJ, Jones J, Cook P, et al. Facilitators and barriers to initiating change in medical intensive care unit
survivors with alcohol
use disorders: A qualitative study.J Crit Care201328849–856
19. Rollnick S, Miller WR, Butler CC. Motivational Interviewing
in Health Care: Helping Patients Change Behavior2007New York, NYGuilford
20. Cook PF, Manzouri S, Aagaard L, et al. Results from 10 years of interprofessional training on motivational interviewing
.Eval Health Prof201740159–179
21. Markland D, Ryan RM, Tobin VJ, et al. Motivational interviewing
and self-determination theory.J Soc Clin Psychol200524811–831
22. Bush K, Kivlahan DR, McDonell MB, et al. The AUDIT alcohol
consumption questions (AUDIT-C): An effective brief screening test for problem drinking. Ambulatory care quality improvement project (ACQUIP). Alcohol
Use Disorders Identification Test.Arch Intern Med19981581789–1795
23. Larsen DL, Attkisson CC, Hargreaves WA, et al. Assessment of client/patient satisfaction: Development of a general scale.Eval Program Plann19792197–207
24. Moyers T, Martin T, Catley D, et al. Assessing the integrity of motivational interviewing
interventions: Reliability of the motivational interviewing
skills code.Behav Cog Psychother200331177–184
25. McGuire-Snieckus R, McCabe R, Catty J, et al. A new scale to assess the therapeutic relationship in community mental health care: STAR.Psychol Med20073785–95
26. Williams GC, Freedman ZR, Deci EL. Supporting autonomy to motivate patients with diabetes for glucose control.Diabetes Care1998211644–1651
27. Smith MS, Wallston KA, Smith CA. The development and validation of the perceived health competence scale.Health Educ Res19951051–64
28. Sobell LC, Sobell MB. Alcohol
Consumption Measures: Assessing Alcohol
Problems: A Guide for Clinicians and Researchers1995Vol 2Bethesda, U.S.Department of Health and Human Services75–99
29. Kiluk BD, Dreifuss JA, Weiss RD, et al. The short Inventory of Problems - Revised (SIP-R): Psychometric properties within a large, diverse sample of substance use disorder treatment seekers.Psychol Addict Behav201327307–314
30. Zigmond AS, Snaith RP. The Hospital Anxiety and Depression Scale.Acta Psychiatr Scand198367361–370
31. Saffer BY, Lanting SC, Koehle MS, et al. Assessing cognitive impairment using PROMIS(®) applied cognition-abilities scales in a medical outpatient sample.Psychiatry Res2015226169–172
32. Horowitz M, Wilner N, Alvarez W. Impact of event scale: A measure of subjective stress.Psychosom Med197941209–218
33. Groh DR, Olson BD, Jason LA, et al. A factor analysis of the important people inventory.Alcohol
34. Leon AC, Davis LL, Kraemer HC. The role and interpretation of pilot studies in clinical research.J Psychiatr Res201145626–629