- Email: [email protected]
A brief readiness to change drinking algorithm: Concurrent validity in female VA primary care patients
Addictive Behaviors 30 (2005) 389 – 395
Short communication
A brief readiness to change drinking algorithm: Concurrent validity in female VA primary care patients Amee J. Eplera,*,1, Daniel R. Kivlahanb,c,d, Kristen R. Bushb, Dorcas J. Dobiec,d, Katharine A. Bradleye,f a
Department of Psychological Sciences, University of Missouri-Columbia, Psychology Building, Room 105, 200 South Seventh Street, Columbia, MO 65211, USA b Center of Excellence in Substance Abuse Treatment and Education, VA Puget Sound Health Care System, Seattle, WA, USA c Mental Illness Research, Education, and Clinical Center, VA Puget Sound Health Care System, Seattle, WA, USA d Department of Psychiatry and Behavioral Sciences, University of Washington, Seattle, WA, USA e Health Services Research and Development and Primary and Specialty Medical Care Service, VA Puget Sound Health Care System, Seattle, WA, USA f Departments of Medicine and Health Services, University of Washington, Seattle, WA, USA
Abstract Brief primary care interventions for alcohol use should be tailored to patients’ readiness to change; however, validated measures of readiness to change are too lengthy to be practical in most primary care settings. We compared a readiness to change drinking algorithm (RTC Algorithm) based on three standardized questions to a validated 12-item readiness to change questionnaire (Rollnick RTCQ) in 85 hazardous drinking female Veterans Affairs (VA) patients. Results from comparisons of mean Rollnick RTCQ scale scores across RTC Algorithm categories suggest good concurrent validity. Regular assessment using the RTC Algorithm questions may help primary care providers tailor alcohol-related discussions with hazardous drinking patients. D 2004 Elsevier Ltd. All rights reserved. Keywords: Alcoholism; Alcohol abuse; Behavior change; Motivation; Primary health care; Military veterans
* Corresponding author. Tel.: +1-208-755-0222. E-mail address: [email protected] (A.J. Epler). 1 Previously at Health Services Research and Development, VA Puget Sound Health Care System, Seattle, WA, USA. 0306-4603/$ – see front matter D 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.addbeh.2004.05.015
390
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
1. Introduction In 1982, Prochaska and DiClemente described a model of change specifically intended for use in behavior change therapy. Since then, a number of instruments have been developed to measure readiness to change for a variety of risk behaviors, including alcohol consumption (Carey, Purnine, Maisto, & Carey, 1999). Several of these have been shown to predict outcomes after alcohol treatment (Heather, Rollnick, & Bell, 1993; Isenhart, 1997). Measure of readiness to change drinking may also be useful in primary care settings for tailoring brief alcohol counseling with patients who screen positive for hazardous or problem drinking (Samet & O’Connor, 1998). Primary care providers may be more likely to counsel patients about alcohol use if they are aware of their patients’ own concerns about their drinking. Only one measure of readiness to change, the Rollnick Readiness to Change Questionnaire or RTCQ, has been developed specifically for general medical settings (Rollnick, Heather, Gold, & Hall, 1992). The Rollnick RTCQ was designed to measure three dimensions of readiness to change (Rollnick et al., 1992), and predicts change over time in alcohol consumption (Heather et al., 1993). The 12-item questionnaire’s length makes the Rollnick RTCQ difficult to incorporate into routine primary care practice. Although briefer readiness to change algorithms have been developed and used in a variety of settings, they are based on
Fig. 1. Illustration of algorithm used to categorize hazardous drinkers’ readiness to change based on responses to the three questions of the RTC Algorithm. Frequencies (n) indicate the number of hazardous drinkers who ‘‘advance’’ along each individual pathway.
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
391
information gathered during unstructured assessments, making them difficult to interpret and evaluate (Carey et al., 1999). Therefore, a brief standardized measure of readiness to change drinking for primary care settings would be useful. In 1993, based on a review of available literature and face validity, three readiness to change drinking questions were developed for use in self-administered surveys (Bradley, McDonell, Stanfeld, & Fihn, 1994). Three pieces of information were considered important for counseling patients about their drinking: (1) Does the patient recognize that his/her drinking may be a problem? (2) Is the patient interested in changing his/her drinking? and (3) Has the patient recently changed his/her drinking? In the present study, these same readiness to change drinking questions (Fig. 1) and the 12-item Rollnick RTCQ were administered to female Veterans Affairs (VA) patients as part of a study of alcohol screening tests. The purpose of this report is to describe a scoring algorithm (RTC Algorithm) based on three readiness to change drinking questions and evaluate its concurrent validity using the Rollnick RTCQ as a comparison standard.
2. Methods This study is based on data collected as part of the Veteran Women’s Alcohol Problems Study conducted at VA Puget Sound Health Care System in 2000. Women were eligible for an interview study if they resided in Washington State and had received care at VA Puget Sound. Of 2548 eligible women, 393 (15%) completed the in-person Alcohol Use Disorders and Associated Disabilities Interview Schedule (AUDADIS; Grant, Harford, Dawson, Chou, & Pickering, 1995) and a preinterview self-administered questionnaire, which included general health questions, alcohol screening measures followed by the three readiness to change questions, (Bradley et al., 1994) the 12-item Rollnick RTCQ (Rollnick et al., 1992), and several mental health and lifestyle measures. Demographic characteristics and alcohol consumption were obtained by AUDADIS interviewers blinded to results from the preinterview questionnaire. Informed consent was obtained from all interview participants and the University of Washington Institutional Review Board approved the study. Demographic characteristics obtained from the AUDADIS interview (Grant et al., 1995) included age, race, education, and materials status. The AUDADIS was used to determine whether participants met criteria for past-year hazardous drinking among women: drinking on average greater than seven drinks per week or more than three drinks on an occasion (National Institute on Alcohol Abuse and Alcoholism, 1995). In addition, participants were evaluated using the Diagnostic and Statistical Manual Fourth Edition (DSM-IV) criteria to identify active alcohol abuse or dependence (American Psychiatric Association, 1994). The RTC Algorithm (Fig. 1) was used to categorize past-year hazardous drinkers into precontemplation, contemplation, or action. First, hazardous drinkers who indicated a recent decrease in drinking on RTC Algorithm questions were assigned to the action category. Second, remaining hazardous drinkers who had thought about drinking less or had tried but been unable to drink less or were possibly, probably, or sure they drank more than they should were assigned to the contemplation category. Finally, all remaining hazardous drinkers, who
392
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
reported no interest in changing their drinking and no recognition that they drank more than they should, were assigned to the precontemplation category. We used standard scoring for the 12-item Rollnick RTCQ (Rollnick et al., 1992). Values for missing responses to the RTCQ (n = 11) were imputed by using the average of the other three items in the corresponding scale. In addition, each hazardous drinker was assigned to a single stage based on the stage of change with the highest four-item Rollnick RTCQ scale score (Heather et al., 1993; Rollnick et al., 1992). Analyses for this study included only participants who met criteria for past-year hazardous drinking and completed the self-report survey. All analyses used SPSS software (SPSS for Windows, 1999). The RTC Algorithm categories were compared to mean Rollnick RTCQ scale scores and stage of change to evaluate concurrent validity. Three sets of analyses were conducted. First, Kendall’s tau-b (a=.05) was used to test the association of readiness to change stages based on the RTC Algorithm and the Rollnick RTCQ. Second, within-groups analyses were conducted to determine whether RTC Algorithm categories were consistent with the profile of Rollnick RTCQ scale scores. To compare mean scores for each Rollnick RTCQ scale (e.g., precontemplation, contemplation, and action) within hazardous drinkers assigned to each RTC Algorithm category, we used a separate general linear model (GLM) for repeated measures (a=.05). Third, between-groups analyses were conducted using a one-way ANOVA (a=.05) to compare mean Rollnick scale scores between participants assigned to different readiness to change categories by the RTC Algorithm. Pairwise comparisons of mean Rollnick RTCQ scale scores between and within groups of women with the same RTC Algorithm category are also reported.
3. Results Of the 393 female VA patients who completed the AUDADIS interview and preinterview questionnaire, 85 (22%) met NIAAA criteria for past-year hazardous drinking and completed both RTC measures. The study sample was mostly over 40 (M = 42 years, S.D. = 11), white (78%), married (51%), and reported some college or technical training (69%). Thirty-five (41%) hazardous drinkers also met DSM-IV criteria for an alcohol disorder. Mean Rollnick RTCQ scale scores in the study population (N = 85) were as follows: precontemplation 2.3 (S.D. = 4.1); contemplation 3.5 (4.2); and action 1.6 (4.6). The RTC algorithm categorized 37 (44%), 15 (18%), and 33 (39%) women into precontemplation, contemplation, and action, respectively. Readiness to change categories assigned by the RTC Algorithm and the Rollnick RTCQ were reliably correlated (Kendall’s tau-b=.59; P < .001; Table 1), with 59 of the 85 participants (69%) assigned to the same category (Table 1). Among participants classified into the precontemplation category by the Rollnick RTCQ (n = 56), 36 (64%) were categorized as precontemplation by the RTC Algorithm, 9 (16%) were categorized as contemplation because they indicated some recognition that their drinking was excessive or they had considered change, and 11 (20%) were categorized as action because they
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
393
Table 1 Readiness to change drinking based on the RTC Algorithm compared to the Rollnick RTCQ stage of change Rollnick RTCQ stage of changea
RTC algorithm categories Precontemplation category
Contemplation category
Action category
Precontemplation stage Contemplation stage Action stage Total
36 (64.3) – 1 (4.2) 37 (43.5)
9 3 3 15
11 2 20 33
a
Total (16.1) (60.0) (12.5) (17.6)
(19.6) (40.0) (83.3) (38.8)
56 5 24 85
(100) (100) (100) (100)
Rollnick RTCQ stage of change was determined using the highest four-item scale score (see Methods).
indicated that they had cut down or quit drinking in the past 3 months. Among participants categorized as contemplation or action by the RTCQ (n = 29), 28 (97%) were also categorized as contemplation or action by the RTC Algorithm. Within-groups analyses revealed that mean Rollnick RTCQ scale scores differed among participants categorized into each algorithm group. Mean Rollnick RTCQ scale scores within the algorithm precontemplation group (n = 37) differed significantly from each other [GLM for repeated measures; F(2,35) = 127.5; P < .001; Fig. 2]. The mean Rollnick RTCQ precontemplation scale score (M = 5.1, S.D. = 2.9) was significantly higher than both the mean contemplation (M = 6.5, S.D. = 2.1; P < .001) and action (M = 5.0, S.D. = 3.0;
Fig. 2. Mean scores for each scale of the Rollnick 12-item RTCQ for hazardous drinkers assigned to the precontemplation, contemplation, and action categories based on the RTC Algorithm. Error bars represent the standard error of the means for each Rollnick scale.
394
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
P < .001) scale scores in these participants. Among the small number of participants categorized as contemplation (n = 15) by the RTC Algorithm, the mean Rollnick RTCQ scale scores did not differ significantly overall [ F(2,13) = 3.8; P=.051]. Among participants categorized as action (n = 33) by the RTC Algorithm, the mean Rollnick RTCQ scale scores differed significantly overall [ F(2,31] = 13.1; P < .001]. However, only the mean Rollnick RTCQ action (M = 2.3, S.D. = 3.9) and contemplation (M = 1.0, S.D. = 4.6; P < .001) scale scores differed significantly in these participants. Between-groups analyses revealed that mean Rollnick RTCQ scale scores were significantly different between participants categorized into each algorithm group (Fig. 2). Mean Rollnick RTCQ precontemplation [ANOVA; F(2,84) = 21.3; P < .001], contemplation [ F(2,84) = 27.8; P < .001], and action scores [ F(2,84) = 43.7; P < .001] were significantly different across participants categorized into precontemplation, contemplation, and action by the RTC Algorithm. Pairwise comparison of the algorithm groups revealed that mean Rollnick precontemplation scores were significantly higher for those categorized as precontemplation by the RTC Algorithm category (M = 5.1, S.D. = 2.9) compared to those categorized as contemplation (M = 0.2, S.D. = 1.8; P < .001) or action (M = 0.2, S.D. = 4.3; P < .001); mean Rollnick RTCQ contemplation scores were significantly lower for those categorized as precontemplation by the RTC Algorithm (M = 6.5, S.D. = 2.1) compared to those categorized as contemplation (M = 1.6, S.D. = 2.1; P < .001) or action (M = 1.0, S.D. = 4.6; P < .001); Rollnick RTCQ action scores also were significantly lower for those categorized as precontemplation by the RTC Algorithm (M = 5.0, S.D. = 3.0) compared to those categorized as contemplation (M = 2.1, S.D. = 2.3; P=.005) or action (M = 2.3, S.D. = 3.9; P < .001) and these latter two groups also differed reliably ( P < .001).
4. Discussion In this sample of female VA patients who met criteria for past-year hazardous drinking, the three RTC Algorithm questions appear to identify dimensions of readiness to change drinking similar to those of the validated Rollnick RTCQ. Over 95% of participants in the contemplation or action categories based on the Rollnick RTCQ were also categorized as contemplation or action based on the RTC Algorithm. Of patients who indicated some readiness to change on the RTC Algorithm, only 42% were classified as precontemplation by the Rollnick RTCQ. Some limitations of this study should be noted. First, this study was based on a relatively small convenience sample of participants who agreed to be interviewed and included only female VA patients. Second, participants completed all measures for a research study with a guarantee of confidentiality rather than in a clinical setting where participants would expect results to be shared with their providers. Finally, this cross-sectional study cannot address the predictive validity of these three questions. Given these limitations, additional studies are warranted in males and other primary care populations, as well as studies of the feasibility and usefulness of routinely administering these questions in primary care settings. Research exploring the stability of responses to these questions over time is a necessary next step.
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
395
Despite the need for additional research, the RTC Algorithm has several important strengths. Primary care providers who want a measure of readiness to change to help guide brief interventions with hazardous drinkers (Samet & O’Connor, 1998) may prefer a measure that helps them identify any ambivalence toward drinking or consideration of change among hazardous drinkers, even if this ‘‘contemplation’’ is not the predominant attitude, as required by the Rollnick RTCQ. In addition, primary care providers may like the fact that the three RTC Algorithm questions identify patients who have recently changed their drinking, whereas the Rollnick RTCQ can categorize as precontemplation those at-risk drinkers who have recently cut down or quit drinking. Finally, the RTC Algorithm’s brevity makes it practical for use with primary care patients who screen positive for hazardous drinking.
Acknowledgements This research was supported by a grant from the Department of Veteran’s Affairs, Health Services Research and Development Service (GEN 97-022). A. Epler is currently supported by a training grant (T32 #AA13526-01) from the National Institute on Alcohol Abuse and Alcoholism (NIAAA). Dr. Bradley is currently supported by a K23 Career Development Award (#AA00313) from NIAAA and is a Robert Wood Johnson Foundation Generalist Physician Faculty Scholar.
References American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Bradley, K. A., McDonell, M. B., Stanfeld, B. K., & Fihn, S. D. (1994). Problem and hazardous drinkers’ readiness to change (abstract). Journal of General Internal Medicine, 9(April suppl. 2), 85. Carey, K. B., Purnine, D. M., Maisto, S. A., & Carey, M. P. (1999). Assessing readiness to change substance abuse: A critical review of instruments. Clinical Psychological Sciences and Practice, 6(3), 245 – 266. Grant, B. F., Harford, T. C., Dawson, D. A., Chou, P. S., & Pickering, R. P. (1995). The alcohol use disorder and associated disabilities interview schedule (AUDADIS): Reliability of alcohol and drug modules in a general population sample. Drug and Alcohol Dependence, 39, 37 – 44. Heather, N., Rollnick, S., & Bell, A. (1993). Predictive validity of the readiness to change questionnaire. Addiction, 88, 1667 – 1677. Isenhart, C. E. (1997). Pretreatment readiness for change in male alcohol dependent subjects: Predictors of oneyear follow-up status. Journal of Studies on Alcohol, 58(4), 351 – 357. National Institute on Alcohol Abuse and Alcoholism. (1995). The physician’s guide to helping patients with alcohol problems. Bethesda, MD: National Institutes of Health. Rollnick, S., Heather, N., Gold, R., & Hall, W. (1992). Development of a short ‘readiness to change’ questionnaire for use in brief, opportunistic interventions among excessive drinkers. British Journal of Addiction, 87, 743 – 754. Samet, J. H., & O’Connor, P. G. (1998). Alcohol abusers in primary care: Readiness to change behavior. American Journal of Medicine, 105(4), 302 – 306. SPSS for Windows. (Version 10.0.7). (1999). Chicago, IL: SPSS, Statistical software.
Short communication
A brief readiness to change drinking algorithm: Concurrent validity in female VA primary care patients Amee J. Eplera,*,1, Daniel R. Kivlahanb,c,d, Kristen R. Bushb, Dorcas J. Dobiec,d, Katharine A. Bradleye,f a
Department of Psychological Sciences, University of Missouri-Columbia, Psychology Building, Room 105, 200 South Seventh Street, Columbia, MO 65211, USA b Center of Excellence in Substance Abuse Treatment and Education, VA Puget Sound Health Care System, Seattle, WA, USA c Mental Illness Research, Education, and Clinical Center, VA Puget Sound Health Care System, Seattle, WA, USA d Department of Psychiatry and Behavioral Sciences, University of Washington, Seattle, WA, USA e Health Services Research and Development and Primary and Specialty Medical Care Service, VA Puget Sound Health Care System, Seattle, WA, USA f Departments of Medicine and Health Services, University of Washington, Seattle, WA, USA
Abstract Brief primary care interventions for alcohol use should be tailored to patients’ readiness to change; however, validated measures of readiness to change are too lengthy to be practical in most primary care settings. We compared a readiness to change drinking algorithm (RTC Algorithm) based on three standardized questions to a validated 12-item readiness to change questionnaire (Rollnick RTCQ) in 85 hazardous drinking female Veterans Affairs (VA) patients. Results from comparisons of mean Rollnick RTCQ scale scores across RTC Algorithm categories suggest good concurrent validity. Regular assessment using the RTC Algorithm questions may help primary care providers tailor alcohol-related discussions with hazardous drinking patients. D 2004 Elsevier Ltd. All rights reserved. Keywords: Alcoholism; Alcohol abuse; Behavior change; Motivation; Primary health care; Military veterans
* Corresponding author. Tel.: +1-208-755-0222. E-mail address: [email protected] (A.J. Epler). 1 Previously at Health Services Research and Development, VA Puget Sound Health Care System, Seattle, WA, USA. 0306-4603/$ – see front matter D 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.addbeh.2004.05.015
390
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
1. Introduction In 1982, Prochaska and DiClemente described a model of change specifically intended for use in behavior change therapy. Since then, a number of instruments have been developed to measure readiness to change for a variety of risk behaviors, including alcohol consumption (Carey, Purnine, Maisto, & Carey, 1999). Several of these have been shown to predict outcomes after alcohol treatment (Heather, Rollnick, & Bell, 1993; Isenhart, 1997). Measure of readiness to change drinking may also be useful in primary care settings for tailoring brief alcohol counseling with patients who screen positive for hazardous or problem drinking (Samet & O’Connor, 1998). Primary care providers may be more likely to counsel patients about alcohol use if they are aware of their patients’ own concerns about their drinking. Only one measure of readiness to change, the Rollnick Readiness to Change Questionnaire or RTCQ, has been developed specifically for general medical settings (Rollnick, Heather, Gold, & Hall, 1992). The Rollnick RTCQ was designed to measure three dimensions of readiness to change (Rollnick et al., 1992), and predicts change over time in alcohol consumption (Heather et al., 1993). The 12-item questionnaire’s length makes the Rollnick RTCQ difficult to incorporate into routine primary care practice. Although briefer readiness to change algorithms have been developed and used in a variety of settings, they are based on
Fig. 1. Illustration of algorithm used to categorize hazardous drinkers’ readiness to change based on responses to the three questions of the RTC Algorithm. Frequencies (n) indicate the number of hazardous drinkers who ‘‘advance’’ along each individual pathway.
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
391
information gathered during unstructured assessments, making them difficult to interpret and evaluate (Carey et al., 1999). Therefore, a brief standardized measure of readiness to change drinking for primary care settings would be useful. In 1993, based on a review of available literature and face validity, three readiness to change drinking questions were developed for use in self-administered surveys (Bradley, McDonell, Stanfeld, & Fihn, 1994). Three pieces of information were considered important for counseling patients about their drinking: (1) Does the patient recognize that his/her drinking may be a problem? (2) Is the patient interested in changing his/her drinking? and (3) Has the patient recently changed his/her drinking? In the present study, these same readiness to change drinking questions (Fig. 1) and the 12-item Rollnick RTCQ were administered to female Veterans Affairs (VA) patients as part of a study of alcohol screening tests. The purpose of this report is to describe a scoring algorithm (RTC Algorithm) based on three readiness to change drinking questions and evaluate its concurrent validity using the Rollnick RTCQ as a comparison standard.
2. Methods This study is based on data collected as part of the Veteran Women’s Alcohol Problems Study conducted at VA Puget Sound Health Care System in 2000. Women were eligible for an interview study if they resided in Washington State and had received care at VA Puget Sound. Of 2548 eligible women, 393 (15%) completed the in-person Alcohol Use Disorders and Associated Disabilities Interview Schedule (AUDADIS; Grant, Harford, Dawson, Chou, & Pickering, 1995) and a preinterview self-administered questionnaire, which included general health questions, alcohol screening measures followed by the three readiness to change questions, (Bradley et al., 1994) the 12-item Rollnick RTCQ (Rollnick et al., 1992), and several mental health and lifestyle measures. Demographic characteristics and alcohol consumption were obtained by AUDADIS interviewers blinded to results from the preinterview questionnaire. Informed consent was obtained from all interview participants and the University of Washington Institutional Review Board approved the study. Demographic characteristics obtained from the AUDADIS interview (Grant et al., 1995) included age, race, education, and materials status. The AUDADIS was used to determine whether participants met criteria for past-year hazardous drinking among women: drinking on average greater than seven drinks per week or more than three drinks on an occasion (National Institute on Alcohol Abuse and Alcoholism, 1995). In addition, participants were evaluated using the Diagnostic and Statistical Manual Fourth Edition (DSM-IV) criteria to identify active alcohol abuse or dependence (American Psychiatric Association, 1994). The RTC Algorithm (Fig. 1) was used to categorize past-year hazardous drinkers into precontemplation, contemplation, or action. First, hazardous drinkers who indicated a recent decrease in drinking on RTC Algorithm questions were assigned to the action category. Second, remaining hazardous drinkers who had thought about drinking less or had tried but been unable to drink less or were possibly, probably, or sure they drank more than they should were assigned to the contemplation category. Finally, all remaining hazardous drinkers, who
392
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
reported no interest in changing their drinking and no recognition that they drank more than they should, were assigned to the precontemplation category. We used standard scoring for the 12-item Rollnick RTCQ (Rollnick et al., 1992). Values for missing responses to the RTCQ (n = 11) were imputed by using the average of the other three items in the corresponding scale. In addition, each hazardous drinker was assigned to a single stage based on the stage of change with the highest four-item Rollnick RTCQ scale score (Heather et al., 1993; Rollnick et al., 1992). Analyses for this study included only participants who met criteria for past-year hazardous drinking and completed the self-report survey. All analyses used SPSS software (SPSS for Windows, 1999). The RTC Algorithm categories were compared to mean Rollnick RTCQ scale scores and stage of change to evaluate concurrent validity. Three sets of analyses were conducted. First, Kendall’s tau-b (a=.05) was used to test the association of readiness to change stages based on the RTC Algorithm and the Rollnick RTCQ. Second, within-groups analyses were conducted to determine whether RTC Algorithm categories were consistent with the profile of Rollnick RTCQ scale scores. To compare mean scores for each Rollnick RTCQ scale (e.g., precontemplation, contemplation, and action) within hazardous drinkers assigned to each RTC Algorithm category, we used a separate general linear model (GLM) for repeated measures (a=.05). Third, between-groups analyses were conducted using a one-way ANOVA (a=.05) to compare mean Rollnick scale scores between participants assigned to different readiness to change categories by the RTC Algorithm. Pairwise comparisons of mean Rollnick RTCQ scale scores between and within groups of women with the same RTC Algorithm category are also reported.
3. Results Of the 393 female VA patients who completed the AUDADIS interview and preinterview questionnaire, 85 (22%) met NIAAA criteria for past-year hazardous drinking and completed both RTC measures. The study sample was mostly over 40 (M = 42 years, S.D. = 11), white (78%), married (51%), and reported some college or technical training (69%). Thirty-five (41%) hazardous drinkers also met DSM-IV criteria for an alcohol disorder. Mean Rollnick RTCQ scale scores in the study population (N = 85) were as follows: precontemplation 2.3 (S.D. = 4.1); contemplation 3.5 (4.2); and action 1.6 (4.6). The RTC algorithm categorized 37 (44%), 15 (18%), and 33 (39%) women into precontemplation, contemplation, and action, respectively. Readiness to change categories assigned by the RTC Algorithm and the Rollnick RTCQ were reliably correlated (Kendall’s tau-b=.59; P < .001; Table 1), with 59 of the 85 participants (69%) assigned to the same category (Table 1). Among participants classified into the precontemplation category by the Rollnick RTCQ (n = 56), 36 (64%) were categorized as precontemplation by the RTC Algorithm, 9 (16%) were categorized as contemplation because they indicated some recognition that their drinking was excessive or they had considered change, and 11 (20%) were categorized as action because they
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
393
Table 1 Readiness to change drinking based on the RTC Algorithm compared to the Rollnick RTCQ stage of change Rollnick RTCQ stage of changea
RTC algorithm categories Precontemplation category
Contemplation category
Action category
Precontemplation stage Contemplation stage Action stage Total
36 (64.3) – 1 (4.2) 37 (43.5)
9 3 3 15
11 2 20 33
a
Total (16.1) (60.0) (12.5) (17.6)
(19.6) (40.0) (83.3) (38.8)
56 5 24 85
(100) (100) (100) (100)
Rollnick RTCQ stage of change was determined using the highest four-item scale score (see Methods).
indicated that they had cut down or quit drinking in the past 3 months. Among participants categorized as contemplation or action by the RTCQ (n = 29), 28 (97%) were also categorized as contemplation or action by the RTC Algorithm. Within-groups analyses revealed that mean Rollnick RTCQ scale scores differed among participants categorized into each algorithm group. Mean Rollnick RTCQ scale scores within the algorithm precontemplation group (n = 37) differed significantly from each other [GLM for repeated measures; F(2,35) = 127.5; P < .001; Fig. 2]. The mean Rollnick RTCQ precontemplation scale score (M = 5.1, S.D. = 2.9) was significantly higher than both the mean contemplation (M = 6.5, S.D. = 2.1; P < .001) and action (M = 5.0, S.D. = 3.0;
Fig. 2. Mean scores for each scale of the Rollnick 12-item RTCQ for hazardous drinkers assigned to the precontemplation, contemplation, and action categories based on the RTC Algorithm. Error bars represent the standard error of the means for each Rollnick scale.
394
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
P < .001) scale scores in these participants. Among the small number of participants categorized as contemplation (n = 15) by the RTC Algorithm, the mean Rollnick RTCQ scale scores did not differ significantly overall [ F(2,13) = 3.8; P=.051]. Among participants categorized as action (n = 33) by the RTC Algorithm, the mean Rollnick RTCQ scale scores differed significantly overall [ F(2,31] = 13.1; P < .001]. However, only the mean Rollnick RTCQ action (M = 2.3, S.D. = 3.9) and contemplation (M = 1.0, S.D. = 4.6; P < .001) scale scores differed significantly in these participants. Between-groups analyses revealed that mean Rollnick RTCQ scale scores were significantly different between participants categorized into each algorithm group (Fig. 2). Mean Rollnick RTCQ precontemplation [ANOVA; F(2,84) = 21.3; P < .001], contemplation [ F(2,84) = 27.8; P < .001], and action scores [ F(2,84) = 43.7; P < .001] were significantly different across participants categorized into precontemplation, contemplation, and action by the RTC Algorithm. Pairwise comparison of the algorithm groups revealed that mean Rollnick precontemplation scores were significantly higher for those categorized as precontemplation by the RTC Algorithm category (M = 5.1, S.D. = 2.9) compared to those categorized as contemplation (M = 0.2, S.D. = 1.8; P < .001) or action (M = 0.2, S.D. = 4.3; P < .001); mean Rollnick RTCQ contemplation scores were significantly lower for those categorized as precontemplation by the RTC Algorithm (M = 6.5, S.D. = 2.1) compared to those categorized as contemplation (M = 1.6, S.D. = 2.1; P < .001) or action (M = 1.0, S.D. = 4.6; P < .001); Rollnick RTCQ action scores also were significantly lower for those categorized as precontemplation by the RTC Algorithm (M = 5.0, S.D. = 3.0) compared to those categorized as contemplation (M = 2.1, S.D. = 2.3; P=.005) or action (M = 2.3, S.D. = 3.9; P < .001) and these latter two groups also differed reliably ( P < .001).
4. Discussion In this sample of female VA patients who met criteria for past-year hazardous drinking, the three RTC Algorithm questions appear to identify dimensions of readiness to change drinking similar to those of the validated Rollnick RTCQ. Over 95% of participants in the contemplation or action categories based on the Rollnick RTCQ were also categorized as contemplation or action based on the RTC Algorithm. Of patients who indicated some readiness to change on the RTC Algorithm, only 42% were classified as precontemplation by the Rollnick RTCQ. Some limitations of this study should be noted. First, this study was based on a relatively small convenience sample of participants who agreed to be interviewed and included only female VA patients. Second, participants completed all measures for a research study with a guarantee of confidentiality rather than in a clinical setting where participants would expect results to be shared with their providers. Finally, this cross-sectional study cannot address the predictive validity of these three questions. Given these limitations, additional studies are warranted in males and other primary care populations, as well as studies of the feasibility and usefulness of routinely administering these questions in primary care settings. Research exploring the stability of responses to these questions over time is a necessary next step.
A.J. Epler et al. / Addictive Behaviors 30 (2005) 389–395
395
Despite the need for additional research, the RTC Algorithm has several important strengths. Primary care providers who want a measure of readiness to change to help guide brief interventions with hazardous drinkers (Samet & O’Connor, 1998) may prefer a measure that helps them identify any ambivalence toward drinking or consideration of change among hazardous drinkers, even if this ‘‘contemplation’’ is not the predominant attitude, as required by the Rollnick RTCQ. In addition, primary care providers may like the fact that the three RTC Algorithm questions identify patients who have recently changed their drinking, whereas the Rollnick RTCQ can categorize as precontemplation those at-risk drinkers who have recently cut down or quit drinking. Finally, the RTC Algorithm’s brevity makes it practical for use with primary care patients who screen positive for hazardous drinking.
Acknowledgements This research was supported by a grant from the Department of Veteran’s Affairs, Health Services Research and Development Service (GEN 97-022). A. Epler is currently supported by a training grant (T32 #AA13526-01) from the National Institute on Alcohol Abuse and Alcoholism (NIAAA). Dr. Bradley is currently supported by a K23 Career Development Award (#AA00313) from NIAAA and is a Robert Wood Johnson Foundation Generalist Physician Faculty Scholar.
References American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders (4th ed.). Washington, DC: Author. Bradley, K. A., McDonell, M. B., Stanfeld, B. K., & Fihn, S. D. (1994). Problem and hazardous drinkers’ readiness to change (abstract). Journal of General Internal Medicine, 9(April suppl. 2), 85. Carey, K. B., Purnine, D. M., Maisto, S. A., & Carey, M. P. (1999). Assessing readiness to change substance abuse: A critical review of instruments. Clinical Psychological Sciences and Practice, 6(3), 245 – 266. Grant, B. F., Harford, T. C., Dawson, D. A., Chou, P. S., & Pickering, R. P. (1995). The alcohol use disorder and associated disabilities interview schedule (AUDADIS): Reliability of alcohol and drug modules in a general population sample. Drug and Alcohol Dependence, 39, 37 – 44. Heather, N., Rollnick, S., & Bell, A. (1993). Predictive validity of the readiness to change questionnaire. Addiction, 88, 1667 – 1677. Isenhart, C. E. (1997). Pretreatment readiness for change in male alcohol dependent subjects: Predictors of oneyear follow-up status. Journal of Studies on Alcohol, 58(4), 351 – 357. National Institute on Alcohol Abuse and Alcoholism. (1995). The physician’s guide to helping patients with alcohol problems. Bethesda, MD: National Institutes of Health. Rollnick, S., Heather, N., Gold, R., & Hall, W. (1992). Development of a short ‘readiness to change’ questionnaire for use in brief, opportunistic interventions among excessive drinkers. British Journal of Addiction, 87, 743 – 754. Samet, J. H., & O’Connor, P. G. (1998). Alcohol abusers in primary care: Readiness to change behavior. American Journal of Medicine, 105(4), 302 – 306. SPSS for Windows. (Version 10.0.7). (1999). Chicago, IL: SPSS, Statistical software.