Assessing the Predictive Validity of the Stages of Change Readiness and Treatment Eagerness Scale (SOCRATES) in Alaska Native and American Indian People

Abstract

Objectives:

The objective of this study was to examine the predictive validity of the Stages of Change Readiness and Treatment Eagerness Scale (SOCRATES) among Alaska Native and American Indian (ANAI) people with an alcohol use disorder.

Methods:

The sample was 170 ANAI adults with an alcohol use disorder living in Anchorage, Alaska who were part of a larger alcohol intervention study. The primary outcome of this study was alcohol use as measured by mean urinary ethyl glucuronide (EtG). EtG urine tests were collected at baseline and then up to twice a week for four weeks. We conducted bivariate linear regression analyses to evaluate associations between mean EtG value and each of the three SOCRATES subscales (Recognition, Ambivalence, and Taking Steps) and other covariates such as demographic characteristics, alcohol use history, and chemical dependency service utilization.

We then performed multivariable linear regression modeling to examine these associations after adjusting for covariates.

Results:

After adjusting for covariates, mean EtG values were negatively associated with the Taking Steps (P = 0.017) and Recognition (P = 0.005) subscales of the SOCRATES among ANAI people living in Alaska. We did not find an association between mean EtG values and the Ambivalence subscale (P = 0.129) of the SOCRATES after adjusting for covariates.

Conclusions:

Higher scores on the Taking Steps and Recognition subscales of the SOCRATES at baseline among ANAI people predicted lower mean EtG values. This study has important implications for communities and clinicians who need tools to assist ANAI clients in initiating behavior changes related to alcohol use.

Alcohol use disorder (AUD) is a prominent health problem with serious biological, physiological, and social consequences.¹ Despitehighratesofalcoholabstinence,^2–4 Alaska Native and American Indian (ANAI) people throughout the US experience more adverse alcohol-related health outcomes compared to the general US population.^5,6 Alcohol-related mortalityamongANAIpeopleistwicethatofthegeneralpopulation.⁵ Rates of specific causes of alcohol-related mortality such as alcoholic liver disease⁶ and hypothermia⁷ are higher among ANAI populations than among other racial groups. Despite these alcohol-related disparities, challenges continue to persist on how best to engage and provide culturally acceptable treatment for ANAI people with AUD.^8–10

Identifying predictors of treatment outcomes is valuable as it enables clinicians to improve patient and family counseling and identify target areas for treatment.¹¹ Among the general population, several studies have identified predictors of treatmentoutcomesforthosewithAUD.^12–14 Motivation is one of the most consistent predictors of alcohol treatment outcomes.¹¹ Motivation is central to the Transtheoretical Model of Change, which maintains that to change a behavior individuals must possess an adequate level of motivation to move through five non-linear steps of change, that is, precontemplation, contemplation, preparation, action, and maintenance.¹⁵ From this model, the Stages of Change Readiness and Treatment Eagerness Scale (SOCRATES) was developed to specifically assess the motivation of individuals with AUD to change behavior.¹⁶

The SOCRATES has been used and studied for more than two decades,^17–22 but few studies have examined its predictive utility with racial/ethnic minority populations. Although studies of factors associated with treatment completion and outcomes amongANAIpeoplehavebeendone,^12,23,24 these studies did not account for motivation to change. For instance, one study found recovery-oriented social support and not having a difficult living situation were predictors of abstinence at treatment discharge.¹² Another study reported gender was an important predictor for successful treatment discharges among Alaska Native people, with women being more likely than men to successfully complete the treatment program.²³ Lastly, a study found men, those with legal problems, and those with a longer length of stay were more likely to accept a referral to substance abuse treatment following detoxification discharge.²⁴

The alcohol-related disparities within ANAI communities highlight the need to investigate the predictive validity of the SOCRATES to ensure it can be used interchangeably across different racial/ethnic backgrounds.^25–27 Cultural norms and practices influence the relevance of specific constructs for particular groups, the range of behaviors and responses that are indicators of that construct, and even how individuals understand and interpret items intended to assess the constructs.²⁵ Therefore, the adequacy of a measure for one culture does not guarantee its adequacy or appropriateness for another cultural group. Furthermore, the tremendous diversity of ANAI communities, small population sizes of distinct ANAI cultural groups, and varying cultural contexts and worldviews likely influence the validity of measures such as the SOCRATES for ANAI contexts.²⁷

ANAI people in Alaska are diverse regarding cultural practices and beliefs. Thus, we aimed to evaluate the predictive validity of SOCRATES subscale scores on drinking outcomes for a culturally diverse sample of urban ANAI adults with AUD in Alaska. The sample in the present study was part of a larger alcohol-related intervention study.²⁸ The primary outcome of this study was alcohol use as measured by mean urinary ethyl glucuronide (EtG). Our goal was to determine if and which of the SOCRATES subscales predicted mean EtG values. The results of our study have important implications for communities and clinicians who need tools to help ANAI clients in initiating behavior changes related to alcohol use. Our findings will also demonstrate if the SOCRATES is an appropriate measure under conditions of greater cultural diversity than was typical in prior studies. Moreover, this is the first study, to our knowledge, that used an alcohol biomarker to evaluate the predictive validity of the SOCRATES in an underrepresented population.

METHODS

Setting

Data collection for this research project occurred at Southcentral Foundation, a tribally-owned and -operated healthcare facility in Anchorage, Alaska that provides primary healthcare services to more than 65,000 ANAI people residing in southcentral Alaska. Southcentral Foundation provides services to all Indian Health Service beneficiaries. The Southcentral Foundation service population is approximately 85% Alaska Native (aloneorin combination) and 15% American Indian (alone or in combination). In addition to primary care services, Southcentral Foundation offers other specialty care including alcohol-related services and programs. Anchorage is an urban setting and is the most populous city in Alaska (pop. 291,538) and has a large ANAI community (8.9% ANAI alone)²⁹ representing 229 federally-recognized tribes from Alaska and others across the US. In Alaska, ANAI people are extremely diverse. For instance, there are20differentAlaskaNativelanguagesandroughlyeightbroad culturalgroups.³⁰ The different cultural groups have varied social practices and beliefs. The study received tribal and institutional review board approvals.

Sample and Recruitment

The sample was 170 self-identified ANAI adults that enrolled and participated in the larger alcohol intervention study. Data analyzed in this study are part of a larger alcohol related intervention study; more detail regarding participant recruitment procedures and study events can be found in a previously published paper.²⁸ Participants were recruited through advertisements in an outpatient treatment center for substance use disorders and primary care clinics as well as through radio and social media advertising. Inclusion criteria included: (1) self-identifying as ANAI; (2) being age 18 years or older; (3) meeting criteria for current alcohol dependence, as defined by the Diagnostic and Statistical Manual of Mental Disorders, fourth edition, text revision (DSM-IV-TR); (4) having consumed four or more standard drinks on one or more occasions in the last 30 days; (5) being able to read and speak English; and (6) being able to provide written, informed consent. Exclusion criteria screened out individuals who reported using illicit drugs, prescription amphetamines, or opioids on more days than alcohol within the last 90 days, those at risk of dangerous alcohol withdrawal, and those with medical or psychiatric conditions that would not allow them to participate. Eligible individuals who wished to participate in the study were scheduled for an in-person, individual baseline interview with a member of the research team where they provided written, informed consent.

Treatment Description

Data collection occurred from June 2016 to September 2019. In the larger alcohol intervention study,²⁸ all participants completed a 4-week lead-in phase prior to randomization (warm-up), then 12 weeks of either a contingency management intervention for alcohol abstinence, or a control condition where participants received reinforcers for attending study visits, regardless of alcohol abstinence. For the present study, we used only data from the 4-week warm-up phase to avoid confounding with the treatment condition.

Measures and Materials

The primary outcome measure was alcohol use defined by the mean urinary EtG value²⁸ for each participant. EtG is an alcohol metabolite^31,32 that can be detected in urine for up to five days after drinking alcohol depending on the amount of alcohol consumed.³³ The EtG analysis returns a value between 0 and 2000ng/mL, with higher values indicating higher alcohol use. EtG values of 100ng/mL or less are indicative of no drinking in the last five days, values of 101ng/mL to 499ng/mL correspond to some light drinking (men 4 standard drinks, women 3 standard drinks) in the last 48hours or heavy drinking (men >4 standard drinks, women >3 standard drinks) in the last 5 days, and values greater than or equal to 500ng/mL are indicative of heavy drinking in the last 48hours.³⁴ Urine samples were collected during the baseline interview then up to twice a week for four weeks, which totaled up to eight urine samples per person. During the baseline interview, we also collected self-reported demographic characteristics (sex, gender, age, race, ethnicity, housing status, education, and religious preference). Members of the research team administered several clinical measures at the same time for all participants during the baseline interview (ie, the start of the 4-week warm-up phase), including the focus of this report, the SOCRATES.¹⁶

SOCRATES-Alcohol Version

The SOCRATES is a 19-item measure that is comprised of two versions (ie, 8A=alcohol or 8D=drugs) used to assess motivation to change substance use behavior.¹⁶ For the present study, the alcohol version of the measure was used. The SOCRATES measures motivation to change alcohol use behavior on three subscales: Recognition, Ambivalence, and Taking Steps. Items are rated on a 5-point Likert scale (“strongly disagree” to “strongly agree”). Recognition (score range of 7–35) measures the level of awareness individuals express regarding the problems alcohol use is causing in their lives and a desire to change (eg, “I really want to make changes in my drinking,” and “If I don’t change my drinking soon, my problems are going to get worse”). Ambivalence (score range of 4–20) measures the level of uncertainty individuals have regarding the problems that alcohol use is causing for themselves or others in their lives (eg, “Sometimes I wonder if my drinking is hurting other people,” and “There are times when I wonder if I drink too much”). Taking Steps (score range of 8–40) measures the level of positive behavior change individuals have initiated and are making in relation to their alcohol use (eg, “I have already started making some changes in my drinking,” and “I was drinking too much at one time, but I’ve managed to change my drinking”). Total scores for each subscale were calculated and scores reflect the level of Recognition, Ambivalence, or Taking Steps for individuals.

Analytical Plan

The dependent variable was the mean EtG value for each participant during the warm-up phase. We first conducted bivariate linear regression analyses to evaluate associations between mean EtG value and each of the three SOCRATES subscales (Recognition, Ambivalence, and Taking Steps), demographic characteristics, alcohol use histories (ie, number of years of heavy alcohol use as defined by men 5 standard drinks or women 4 standard drinks per day, age of first heavy alcohol use, and number of days of heavy alcohol use in the last 30 days since the baseline interview) and chemical dependency utilization (ie, whether or not an individual received chemical dependency services for alcohol in the month before the baseline interview). We then performed multivariable linear regression modeling to examine these associations after adjusting for covariates. Covariates with a P value <0.25 were retained in the multivariable analysis.³⁵ Alpha was set at 0.05, and all data analyses were performed using R software.³⁶

RESULTS

Table 1 describes the sample(N=170)and shows similar distributions of men and women. The mean number of years of heavy alcohol use was more than a decade, and the mean age at first heavy alcohol use was under 20 years. The mean EtG across participants was 635.4 (SD=665.6)ng/mL with a minimum of 0.0ng/mL and a maximum of 2000.0ng/mL.

TABLE 1.

Participant Demographics, Chemical Dependency Service Utilization, and Alcohol Use Characteristics (N = 170)

Variable	Mean (SD) or Percentage
Age (years)	46.4 (11.6)
Male	51%
Education completed (years)	12.7 (2.1)
Employed (full-time, part-time, self)	49%
Married	13%
Used chemical dependency services in last 30 days	15%
Heavy alcohol use (years)	16.6 (12.8)
Age at first heavy alcohol use (years)	19.2 (7.7)
Heavy alcohol use in last 30 days (days)	9.9 (9.2)

Table 2 describes participants’awarenessofalcohol-related problems, ambivalence about changing their drinking, and motivation to change alcohol use, as measured by the SOCRATES subscales. Overall, participants scored low on Recognition (mean=29.3, SD=4.6, range=16–35), medium on Ambivalence (mean=14.7, SD=3.3, range=4–20), and medium on Taking Steps (mean=33.6, SD=4.7, range=16–40) using interpretative ranges developed through Project MATCH for people seeking treatment for alcohol problems.³⁷

TABLE 2.

The Score Distribution of Motivation to Change for the SOCRATES Subscales: Recognition (Score Range of 7–35), Ambivalence (Score Range of 4–20), and Taking Steps (Score Range of 8–40)

Subscale	Mean (SD)	Median	Minimum	Maximum	Interpretation
Recognition	29.3 (4.6)	30	16	35	Low
Ambivalence	14.7 (3.3)	15	4	20	Medium
Taking Steps	33.6 (4.7)	34	16	40	Medium

All covariates from the bivariate linear regression analyses were retained in the multivariable linear regression analysis as all covariates had P values <0.25. In the bivariate and multivariable linear regression analyses, scores on the Taking Steps and Recognition subscales predicted subsequent mean urinary EtG values (Table 3). Individuals had lower alcohol use, as reflected by decreases in mean EtG values, if they scored higher on the Taking Steps and Recognition subscales at baseline. For every one-unit increase in the Taking Steps score, the mean EtG decreased by 26.8ng/mL after adjusting for other covariates. This means that an individual who scored very low on the Taking Steps subscale (ie, 8) had a mean EtG value 858ng/mL higher than an individual who scored very high on the Take Steps subscale (ie, 40), while holding other covariates constant. For every one-unit increase in the Recognition score, the mean EtG decreased by 31.4ng/mL after adjusting for other covariates. This means that an individual who scored very low on the Recognition subscale (ie, 7) had a mean EtG value 879ng/mL higher than an individual who scored very high on the Recognition subscale (ie, 35), while holding other covariates constant. Ambivalence subscale scores did not predict mean EtG values in bivariate or multivariable linear regression analyses.

TABLE 3.

Results of Bivariate and Multivariable Linear Regression Models Estimating the Association of Each Covariate With the Mean EtG Value for Each Participant

Variable	Bivariate			Multivariable
	Estimate	SE	P	Estimate	SE	P
Subscales
Recognition	−36.00	10.95	0.001†	−31.39	11.09	0.005†
Ambivalence	4.59	15.81	0.772	20.97	13.75	0.129
Taking steps	−50.92	10.35	<0.001†	−26.83	11.11	0.017*
Demographics
Age (years)	7.94	4.41	0.074	−1.49	3.96	0.707
Sex (ref: male)	−131.59	102.53	0.201	−133.59	89.78	0.139
Alcohol use history
Age at first heavy use (years)	11.63	6.68	0.083	11.62	5.77	0.046*
Heavy use in last 30 days (days)	33.26	5.03	<0.001†	32.19	4.92	<0.001†
Chemical dependency services in last 30 days (ref: yes)	355.30	142.40	0.014*	92.52	126.70	0.466

^*P<0.05.

^†P<0.01

The number of days of heavy alcohol use in the past 30 days at baseline also predicted mean EtG values in both the bivariate and multivariable linear regression analyses. For every day increase in the number of heavy days of alcohol use in the past 30 days, the mean EtG increased by 32.2ng/mL after adjusting for other covariates. Lastly, the age at first heavy alcohol use predicted mean EtG values in both the bivariate and multivariable linear regression analyses. For every year increase in age at first heavy alcohol use, the mean EtG increased by 11.6ng/mL after adjusting for other covariates.

DISCUSSION

We examined the predictive validity of motivation to change alcohol use behavior—as assessed by the SOCRATES scale—on drinking outcomes, as defined by mean EtG values, among a diverse sample of urban ANAI adults with AUD. We found that individuals who were aware of their alcohol-related problems (ie, Recognition subscale) and who were already taking action to reduce their alcohol consumption (ie, Taking Steps subscale) at baseline had lower mean EtG values than individuals who were less aware of their alcohol-related problems and/or not taking action to reduce their alcohol consumption. Of the three SOCRATES subscales, the Taking Steps and Recognition subscales predicted mean EtG values while the Ambivalence subscale did not. Taking Steps is the most behaviorally anchored of the three subscales. The measure asks respondents to indicate whether they are engaging in behavior change, for example, “I’m not just thinking about changing my drinking, I’m already doing something about it”, as opposed to merely being aware of a problem (Recognition) or wondering whether drinking alcohol is causing problems (Ambivalence).¹⁶ Our findings are consistent with previous studies that examined the predictive validity of the SOCRATES in other populations and found the Taking Steps subscale to be predictive of alcohol use.^19–22,38 Our results also demonstrate the SOCRATES is an appropriate measure for a culturally diverse sample of urban ANAI people, which implies the SOCRATES is an appropriate measure under conditions of greater cultural diversity than was typical of prior samples. An innovative aspect of our study is the utility of an alcohol biomarker (ie, ethyl glucuronide) for assessing the predictive validity of the SOCRATES in an underrepresented population.

Our findings have potential treatment implications for ANAI people with AUD. It is often challenging for clinicians to engage people in treatment who are ambivalent to change or do not believe change is needed.³⁹ Understanding an individual’s level of motivation to change their alcohol use behavior may allow clinicians to individualize treatment approaches to align with each person’s readiness to change behaviors related to alcohol use. In general, people who score high on the Taking Steps subscale and are within the action stage of the Transtheoretical Model of Change, should be advised to develop skills to implement behavior change in their treatment plan.³⁹ Additionally, they need to build awareness of various psychological (cognitive, behavioral, emotional) events that may work against their efforts at behavior change. These individuals need to learn ways to prevent major behavior reversals, such as returning to pre-change patterns and levels of alcohol use. People who score high on the Taking Steps subscale and are within the maintenance stage of the Transtheoretical Model of Change should sustain and strengthen any changes that they have made in the problem behavior (eg, drinking). These behavior changes need to be integrated into the individual’s lifestyle. Individuals who score low on both the Taking Steps and the Recognition subscales may be in the earlier stages of change (eg, precontemplation or contemplation); they would benefit from receiving motivational strategies from clinicians rather than focusing on behavioral change.³⁹ It is important to note that these are general treatment recommendations based on the Transtheoretical Model of Change and more research is necessary to identify specific treatment recommendations for ANAI people.

Our study has important limitations. First, these analyses included a modest sample size, which restricts their statistical power. We suggest that future studies examine the predictive validity of the SOCRATES with a larger number of ANAI adults with AUD to understand the generalizability of our findings. Second, our study primarily consisted of ANAI adults with moderate-severe AUD (ie, current alcohol dependence). Thus, our findings may not be applicable to individuals with mild AUD (ie, current alcohol abuse). Third, we did not distinguish between Alaska Native and American Indian people. We assume our sample consisted of primarily Alaska Native people given the high population of Alaska Native people in Anchorage, which may then limit the generalizability to American Indian communities, particularly those residing in the lower 48 states. Fourth, although our findings demonstrate that the SOCRATES is associated with mean EtG values among ANAI people, the SOCRATES was not created with a focus on ANAI people. Thus, the SOCRATES may not be capturing certain constructs that may potentially make the scale more predictive of drinking outcomes (eg, adding items related to the negative impact of alcohol use on engagement in cultural activities or on their role in their community). We suggest future studies develop and test a culturally-adapted SOCRATES for ANAI people. Lastly, because data was only collected at one urban site, our findings may not generalize to other ANAI populations, particularly rural communities. More research is recommended to determine the generalizability of our findings to ANAI people with AUD in other settings.

CONCLUSIONS

Our results suggest that action-oriented motivation to change alcohol use behavior (ie, the Taking Steps subscale of the SOCRATES) is related to drinking outcomes among a culturally diverse urban ANAI population in Alaska as indicated by mean EtG values over four weeks. Our findings imply the SOCRATES may be an appropriate measure for other culturally diverse groups such as ANAI people in Alaska. It may be beneficial to emphasize improving action-oriented motivation, such as including in the individuals’ treatment plan the development of more skills around healthy emotional functioning, to change alcohol use behavior among ANAI people with AUD. Future research should focus on demographic and clinical factors that are associated with scoring higher on the Taking Steps subscale of the SOCRATES, and how to improve motivation while individuals participate in treatment. Lastly, our study demonstrates the utility of an alcohol biomarker (ie, ethyl glucuronide) for examining the predictive utility of the SOCRATES in an underrepresented population.