Identifying Harmful Drinking Using a Single Screening Question in a Psychiatric Consultation-Liaison Population

Psychosomatics 2011:52:362–366

© 2011 The Academy of Psychosomatic Medicine. Published by Elsevier Inc. All rights reserved.

Original Research Reports Identifying Harmful Drinking Using a Single Screening Question in a Psychiatric Consultation-Liaison Population Suena H. Massey, M.D., Lorenzo Norris, M.D., Melissa Lausin, M.D., Chinyere Nwaneri, B.S., Daniel Z. Lieberman, M.D.

Background: Harmful drinking is common in medical inpatients, yet commonly missed due in part to time pressures. A screening question about past year heavy drinking recommended by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) has been validated in primary care and emergency room settings. We tested the psychometric properties of a modified single screening question (SSQ) in hospitalized patients referred to a consultation-liaison service. Methods: A psychiatry attending (n ⫽ 40), a psychiatry resident (n ⫽ 30) and a medical student (n ⫽ 30) administered the SSQ, followed by a self-report 10-item Alcohol Use Disorders Identification Test (AUDIT) to a sample of

100 consultation-liaison patients who were able to give informed consent for participation. Results: Using the AUDIT as a reference, the sensitivity and specificity of the SSQ to detect harmful drinking in this sample were .96 and .82, respectively. Gender differences in specificity were not found. The single question also had a strong correlation with dependence (rb ⫽ .457, p ⬍ .001), and harmful use (rb ⫽ .620, p ⬍ .001) subscales of the AUDIT. Conclusion: The SSQ about past year heavy drinking can rapidly identify harmful drinking in alert nonpsychotic consultation-liaison patients. (Psychosomatics 2011; 52:362–366)

H

Test (MAST) are better suited to time-limited clinical encounters, but may fail to detect harmful subsyndromal drinking problems.7–11 Furthermore, differences in performance of the CAGE and TWEAK have been detected among women and minority groups.11,12 More recently, the National Institute on Alcohol Abuse and Alcoholism (NIAAA) developed a single screening question (SSQ) about past year consumption, preceded by a prescreening question to provide a segue to

armful drinking is common among medical inpatients. Excessive alcohol consumption can have adverse effects on multiple organ systems and can complicate the management of diabetes, hypertension, obesity, and psychiatric conditions. Despite the high prevalence and serious risks associated with heavy drinking, it remains largely unidentified and untreated.1 Because a full diagnostic interview for alcohol problems is often impractical within the constraints of a typical patient encounter, screening instruments have been developed to rapidly identify patients at risk.2 The Alcohol Use Disorders Identification Test (AUDIT),3 one of the most extensively studied instruments4 has shown acceptable screening accuracy in medical inpatients,5 but is rarely used in clinical settings due to its length.6 Shorter screening instruments such as the Cut down, Annoyed, Guilty, Eye Opener (CAGE), Tolerance, Worried, Eye Openers, Amnesia, Kut down (TWEAK), and the Michigan Alcohol Screening 362

www.psychosomaticsjournal.org

Received October 30, 2010; revised January 4, 2011; accepted January 6, 2011. From Dept. of Psychiatry and Behavioral Sciences, George Washington University School of Medicine and Health Sciences (SHM, LN, ML, DZL); and Howard University College of Medicine (CN). Send correspondence and reprint requests to Suena H. Massey, M.D., Dept. of Psychiatry and Behavioral Sciences, George Washington University School of Medicine and Health Sciences, 2150 Pennsylvania Avenue, NW 8-411 Washington, DC 20037. e-mail: [email protected] © 2011 The Academy of Psychosomatic Medicine. Published by Elsevier Inc. All rights reserved.

Psychosomatics 52:4, July-August 2011

Massey et al. a discussion about alcohol use in primary care and mental health settings.13 “Do you sometimes drink beer, wine, or other alcoholic beverages?” is recommended as an introduction. If the answer is no, the screen is considered negative and education about safe drinking limits should follow. If the answer is “yes,” a clinician then asks “Have you had five or more drinks (four or more for women) in a single day in the past year?” An answer of “yes” is considered a positive screen and should be followed by more comprehensive assessment. The definition of at-risk drinking reflected by cutoffs of five or more drinks per day for men and four or more drinks per day for women was based on data from the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC)14,15 showing a sharply increasing risk for alcohol use disorders when consumption met or surpassed these daily limits at any time in the past year. The risk of medical complications and social harm also increased at this level, and the associations were not significantly affected by differences in demographic or socioeconomic characteristics. Furthermore, less than 1% of individuals who never exceeded the daily limits identified in the SSQ had an alcohol use disorder.16 While the NIAAA screening question is well supported by epidemiologic data,14,15 its validity may be limited to primary care settings17 and trauma patients seen in the emergency room18,19 as it has not been extensively tested in other clinical settings. In the current survey study, we report on the performance of a modified SSQ based on cutoffs recommended by the NIAAA, in a sample of inpatients on a psychiatric consultation-liaison service. METHODS

changes, and determination of capacity to make decisions regarding medical care. Referring teams were not informed about the study prior to requesting a psychiatric consultation. Recruitment Patients who receive a psychiatric consultation at this hospital are tracked on a password-protected census, which is updated daily; patients are removed from the census when they are discharged from the hospital. Recruitment was conducted by reviewing the census and approaching every patient on the census for participation. Participants were recruited in two phases. The first phase of recruitment (August to September 2008) was conducted by an attending consultation-liaison psychiatrist (n ⫽ 40) who reviewed the census at the end of each work day, and approached patients for participation in the early evening or on weekends. In reviewing a preliminary data analysis, while results were favorable, it was believed that validity could be strengthened by increasing the sample size. Furthermore, we sought to vary the operator characteristics to provide greater generalizability to findings. Thus, a second phase of recruitment and data collection of the remaining subjects was conducted from August 2009 to April 2010 by a resident and a medical student. In this phase, investigators reviewed the census on Fridays and approached each patient for participation over the weekend. Participants were approached for participation and interviewed only once by either the resident or the medical student, not both. The resident and medical student investigators were not involved in the psychiatric consultations, nor were they identifiable by patients as part of their consulting psychiatric team.

Sample Participants were adult patients hospitalized on medical, surgical, and intensive care units at the George Washington University Hospital, who were seen and followed by the psychiatry consultation-liaison service during their admission. Patients who were unable to provide informed consent due to delirium, psychosis, agitation, or other forms of mental status alteration were excluded from participation. While reasons for each consultation were not recorded, the most common reasons for psychiatric consultation in this hospital are evaluation of suicide risk during or following hospital admission, management of mood, anxiety or psychotic disorders during admission, management of agitation, evaluation of mental status Psychosomatics 52:4, July-August 2011

Procedures All investigators began recruitment by introducing themselves as part of a research team, presenting an information sheet to patients describing a research study about alcohol use and health. Patients were informed that information obtained from the study would be collected anonymously, and would not be shared with their medical team unless they chose to do so. Furthermore, patients were told that the choice to participate would in no way affect their access to care. Recruitment and data collection were not embedded in the psychiatric consultation, but performed separately. Consent for participation was obtained orally to protect confidentiality. www.psychosomaticsjournal.org

363

Identifying Harmful Drinking Using Single Question As participants were informed that they would be asked about their alcohol use during the consent process we did not administer the prescreening question, and started with, “How many times in the past year have you had X or more drinks in a day?” where X is five for men and four for women. A response of one or more times was considered a positive screen. Participants were then asked about their age, gender, race, ethnicity, and number of drinks consumed on the heaviest drinking day in the past 12 months. Next, participants completed a paper-and-pencil Alcohol Use Disorders Identification Test (AUDIT). The items included in the AUDIT reflect three dimensions of drinking: alcohol consumption (items 1–3), alcohol dependence, such as difficulty in controlling drinking, neglect of alternative interests, and physiologic withdrawal (items 4 – 6), and adverse consequences from drinking (items 7–10). Analyses The sensitivity and specificity of the SSQ to predict harmful drinking, defined by a positive AUDIT score of 8 or higher, were calculated. A one-way classification with Yates correction was employed to assess the association between a positive SSQ and a positive AUDIT. Biserial correlation was used to assess the relationship between the SSQ and total AUDIT score, and the score of each of the three AUDIT subscales. PASW 17.0 (SPSS) was used for all analyses. RESULTS Out of an estimated 350 patients who were seen by the service during recruitment periods, 147 (42%) were approached for participation. Twenty-six (18% of those approached) were ineligible for participation due to mental status alteration and 21 (14% of those approached) refused participation. One hundred patients (68% of those approached, or 29% of total recruitment pool) provided data for the analysis. Demographic characteristics were as follows: 57% men, 43% women; 58% African-American, 40% non-Hispanic white, 1% American Indian/Alaskan Native, 1% Asian/Pacific Islander. The mean age was 50.7 years (standard deviation [SD] ⫽ 17.3 years). The SSQ was positive in 39% of participants. The SSQ was positive in 42% of males, 35% of females, 60% of non-Hispanic whites, and 26% of African-Americans. Participants with a positive screen (n ⫽ 39) reported a mean of 92 heavy drinking days in the past year (SD ⫽ 364

www.psychosomaticsjournal.org

TABLE 1.

Characteristics of Participants with a Positive SSQ and Negative AUDIT (n ⴝ 13)

Percentage Male

53.8%

Age (years) Past year heavy drinking days Drinks on past year heaviest drinking day AUDIT total score Consumption subscale Dependence subscale Harmful use subscale

Min 18

Max 78

Mean 49.4

SD 18.9

1 4

175 24

16.6 7.3

47.7 5.4

2 2 0 0

7 6 2 5

4.4 3.5 0.2 0.7

1.5 1.3 0.6 1.4

AUDIT ⫽ Alcohol Use Disorders Identification Test; SSQ ⫽ single screening question.

126 days), a median of 15 heavy drinking days in the past year, an average of 10 drinks on the heaviest drinking day in the past year (SD ⫽ 9 drinks), and a mean AUDIT score of 14 (SD ⫽ 11). Individuals who screened negative (n ⫽ 61) reported an average of one drink on the heaviest drinking day in the past year (SD ⫽ 2 drinks) and had a mean AUDIT score of 1 (SD ⫽ 2). Participants who screened positive but had a negative AUDIT (n ⫽ 13, 7 males, 6 females) had a mean score of 4 on the AUDIT (SD ⫽ 2), an average of 16 heavy drinking days in the past year (SD ⫽ 48), and an average of 7 drinks on their heaviest drinking day (SD ⫽ 5 drinks). Characteristics of the false positive group are shown in Table 1. The sensitivity and specificity of the SSQ to detect harmful drinking were .96 (95% confidence interval [CI] .79 –1) and .82 (95% CI .71–.90), respectively. Sensitivity by gender was: men .94 (95% CI .71–99) and women 1 (95% CI .63–1). Specificity of the SSQ did not differ by gender. A positive SSQ was significantly correlated with a positive AUDIT score ␹2 ⫽ 47.793 (df ⫽ 1, n ⫽ 100, P ⬍ 0.001). A positive screen was also significantly correlated with the AUDIT total score, and all three AUDIT subscales (P ⬍ 0.001). Biserial correlation coefficients for AUDIT total score, consumption, dependence, and harmful use subscales were .658, .730, .457, and .620, respectively. While this study was not designed to detect differences in operating characteristics between investigators, an exploratory analysis revealed no statistically significant differences in age, heavy drinking days, total drinks on heaviest drinking day, and total AUDIT score among participants recruited during different phases (Table 2). Psychosomatics 52:4, July-August 2011

Massey et al.

TABLE 2.

Characteristics of Participants By Recruitment Periods C-L Psychiatrist

Age (years) Past year heavy drinking days Drinks on past year heaviest drinking day AUDIT total score Percentage male Sensitivity of SSQ Specificity of SSQ

Resident and Medical Student

Mean

SD

Mean

SD

P

54.5 45.8

15.1 18.4

48.4 29.2

18.4 75.5

0.087 0.373

4.5

5.6

4.8

8.3

0.841

6.7 57.5% 1 .793

10.1

6.2 56.7% .938 .841

9.2

0.802

C-L ⫽ Consultation-Liaison; AUDIT ⫽ Alcohol Use Disorders Identification Test; SSQ ⫽ single screening question.

DISCUSSION Harmful drinking is prevalent in hospitalized patients, yet commonly missed due in part to time pressures. In this study, we examined the utility of a single screening question (SSQ) to detect harmful drinking in hospitalized patients on a psychiatric consultation-liaison service. The SSQ had a high sensitivity (.96) and an adequate specificity (.82), despite a long time frame of inquiry (12 months), to detect harmful drinking, as defined by an AUDIT score of 8. Importantly, a positive screen was moderately correlated with a positive AUDIT on dependence and harmful use subscales, in addition to the consumption subscale. Our findings are consistent with similar existing literature. In a recent study on a larger nationally representative sample of emergency department patients who participated in NESARC,19 Stewart and colleagues found a similar sensitivity (.96) and specificity (.80) of the SSQ to detect harmful drinking, as defined by the Alcohol Use Disorders and Associated Disabilities Interview Schedule (AUDADIS). Thirteen participants screened positive using the SSQ, but had a negative AUDIT score (Table 1). As mean score on the consumption subscale was 3.5 while dependence and harmful use subscales are 0.2 and 0.7, respectively, we speculated that false positive participants were binge drinkers. However, we cannot rule out the possibility that participants minimized the consequences of their drinking thereby under-reporting symptoms on the AUDIT. The decision to incorporate any health screen into practice requires an assessment of the implications of information gained from screening, individual and healthcare system costs, risks of screening, and the prevalence of Psychosomatics 52:4, July-August 2011

disease of interest in those screened.20 The SSQ was originally developed for primary care patients, as part of a program of screening and brief intervention shown to be effective in reducing subsequent drinking.13 The efficacy of brief interventions delivered in hospitalized patients is unclear. There is recent evidence that more severe alcohol use disorders, which are less responsive to brief intervention, may be over-represented among inpatients who are heavy drinkers.21 However, the costs and risks of the SSQ are low, prevalence of drinking problems is high in hospital settings, and referral or transfer to alcohol treatment may be easier to facilitate in the hospital compared with the outpatient setting. Furthermore, hospitalization may represent an opportunity to increase motivation to change drinking behavior, especially when alcohol-related illnesses are highlighted by treating clinicians.22 Thus, rapid identification of alcohol problems in inpatients is important to facilitate brief intervention and also to expedite engagement in treatment to prevent further morbidity. Despite a modest positive predictive value of the SSQ based on this study, given the low cost of administration of the SSQ and the high prevalence of undetected harmful drinking among hospitalized patients requiring psychiatric consultation, an argument can be made for its incorporation into routine practice. As with any positive screening test, a positive SSQ or AUDIT needs to be followed by more comprehensive assessment. It is important to emphasize that results in this study provide validation of the SSQ in a fairly specific population of inpatients who elicited psychiatric consultation and were able to provide consent to participate in a research study. Some of the patients who were excluded may have exhibited delirium or other forms of mental status alteration directly attributable to alcohol use disorders. Since a screening interview is less likely to be accurate in a patient with delirium or psychosis, information from physical examination, corroborative history, and clinical testing is often crucial in identifying alcohol use disorders. The use of a running census during recruitment aided in the identification of each new consultation to be tracked during their hospital stay so that participants could be approached for participation on a less than daily basis. However, during the second phase of recruitment the census was only reviewed weekly, resulting in nonconsecutive recruitment. While we did not find any differences in age, gender, and reported drinking parameters between data collected during the two recruitment phases, future studies utilizing consecutive recruitment would ensure a more representative sample of inpatients over a shorter www.psychosomaticsjournal.org

365

Identifying Harmful Drinking Using Single Question period of time to minimize the effect of changes in the clinical population on psychometric properties of the screen. Furthermore, examining how patient factors (sociodemographic, primary diagnoses, reasons for consultation, and admission) and clinician factors (experience, personal characteristics, inter-rater reliability) affect performance of the screen would be informative.

ter discharge. Psychiatric consultation in the inpatient setting is an opportunity to intervene and facilitate treatment entry for patients with alcohol use disorders. The first step is identification, and a single screening question about past year heavy drinking can be a valuable part of a routine psychiatric consultation-liaison assessment.

CONCLUSIONS Undetected harmful drinking in hospital patients may contribute to morbidity and treatment complications af-

Disclosures: Dr. Lieberman has received speaker’s honoraria from GlaxoSmithKline. The other authors have no interests to disclose.

References 1. McGlynn EA, Asch SM, Adams J, et al: The quality of health care delivered to adults in the United States. N Engl J Med 2003; 348:2635–2645 2. Babor TF, Higgins-Biddle JC: Alcohol screening and brief intervention: dissemination strategies for medical practice and public health. Addiction 2000; 95:677– 686 3. Saunders JB, Aasland OG, Babor TF, et al: Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption-II. Addiction 1993; 88:791– 804 4. Reinert DF, Allen JP: The alcohol use disorders identification test: an update of research findings. Alcohol Clin Exp Res 2007; 31:185–199 5. Berner MM, Kriston L, Bentele M, et al: The alcohol use disorders identification test for detecting at-risk drinking: a systematic review and meta-analysis. J Stud Alc 2007; 68:461– 473 6. MacKenzie DM, Langa A, Brown TM: Identifying hazardous or harmful alcohol use in medical admissions: a comparison of AUDIT, CAGE and brief MAST. Alcohol Alcohol 1996; 31:591– 599 7. Ewing JA: Detecting alcoholism. The CAGE questionnaire. J Am Med Assoc 1984; 252:1905–1907 8. Russell M: New assessment tools for risk drinking during pregnancy: TACE, TWEAK, and others. Alcohol Health Res World 1994; 18:55– 61 9. Selzer ML: The Michigan Alcoholism Screening Test: the quest for a new diagnostic instrument. Am J Psychiatry 1971; 127: 1653–1658 10. Fiellin DA, Reid MC, O’Connor PG: Screening for alcohol problems in primary care: a systematic review. Arch Intern Med 2000; 160:1977–1989 11. Cherpitel CJ: Screening for alcohol problems in the U.S. general population: a comparison of the CAGE and TWEAK by gender, ethnicity, and services utilization. J Stud Alc 1999; 60:705–711

366

www.psychosomaticsjournal.org

12. Steinbauer JR, Cantor SB, Holzer, et al: Ethnic and sex bias in primary care screening tests for alcohol use disorders. Ann Intern Med 1998; 129:353–362 13. Willenbring ML, Massey SH, Gardner MB: Helping patients who drink too much: an evidence-based guide for primary care physicians. Am Fam Physician 2009; 80:44 –50 14. Dawson DA, Grant BF, Li TK: Quantifying the risks associated with exceeding recommended drinking limits. Alcohol Clin Exp Res 2005; 29:902–908 15. Dawson DA, Li TK, Grant BF: A prospective study of risk drinking: at risk for what? Drug Alcohol Depend 2008; 95: 62–72 16. Smith PC, Schmidt SM, Allensworth-Davies D, et al: Primary care validation of a single-question alcohol screening test. J Gen Intern Med 2009; 24:783–788 17. Reed Jr DN, Saxe A, Montanez M, et al: Use of a single question to screen trauma patients for alcohol dependence. J Trauma 2005; 59:619 – 622 18. Canagasaby A, Vinson DC: Screening for hazardous or harmful drinking using one or two quantity-frequency questions. Alcohol Alcohol 2005; 40:208 –213 19. Stewart SH, Borg KT, Miller PM: Prevalence of problem drinking and characteristics of a single-question screen. J Emerg Med 2010; 39:291–295 20. Blumberg MS: Evaluating health screening procedures. Operations Res 1957; 5:351–360. 21. Saitz R, Freedner N, Palfai TP, et al: The severity of unhealthy alcohol use in hospitalized medical patients: the spectrum is narrow. J Gen Intern Med 2006; 21:381–385 22. Williams EC, Palfai T, Cheng DM, et al: Physical health and drinking among medical inpatients with unhealthy alcohol use: a prospective study. Alcohol Clin Exp Res 2010;34: 1257–1265

Psychosomatics 52:4, July-August 2011