the readiness to change questionnaire: reliability and validity of a swedish version and a...
TRANSCRIPT
The Readiness to Change Questionnaire:Reliability and validity of a Swedish versionand a comparison of scoring methods
Lars Forsberg1*, Soren Ekman2, Jan Halldin3 and Sten Ronnberg4
1Department of Clinical Neuroscience, Section of Dependency Disorders,Karolinska Institutet, Sweden
2Department of Surgery at Danderyd Hospital, Sweden3Department of Public Health Sciences, Division of Social Medicine,Karolinska Institutet, Sweden
4Department of Social Work at Stockholm University, Sweden
Objectives. The aim of this study was to investigate the relative merits of threemethods of scoring the Swedish version of the Readiness to Change Questionnaire(RTCQ), either by assigning a stage in the Prochaska and DiClemente (1986) Stages ofChange Model or by treating the scores as a continuous readiness to change variable.Assigning a stage of change was achieved with both the quick method and the refinedmethod.
Design and methods. Out of 563 patients screened at an emergency surgical wardfor risky alcohol consumption, 165 met risk criteria and responded to the RTCQ. Thethree scoring methods were examined with regard to internal consistency, test–retestreliability, construct and predictive validity.
Results. All three methods of treating the RTCQ scores had satisfactory reliability.Since stages of change (quick method) were significantly but modestly correlated toalcohol consumption and to change-related behaviours at intervention, and moderatelycorrelated to alcohol problems, the quick method had reasonable construct validity.The refined method had higher construct validity; however, this method left 32% of thepatients without a stage assignment. The continuous readiness scale had higherconstruct validity than the quick method, but was not in par with the refined method.No scoring method was found to have predictive validity.
Conclusions. The RTCQ scores treated as a continuous readiness scale were a viablealternative to the original ways of assigning a stage of change to a patient. The SwedishRTCQ is reliable and has reasonable construct validity, but its predictive validity needsfurther investigation.
* Correspondence should be addressed to Lars Forsberg, Karolinska lnstitutet, Department of Clinical Neuroscience,Section of Dependency Disorders, Magnus Huss Clinic M4:02, Karolinska Hospital, 171 76 Stockholm, Sweden(e-mail: [email protected]).
335
British Journal of Health Psychology (2004), 9, 335–346
q 2004 The British Psychological Society
www.bps.org.uk
During the last 15–20 years the Stages of Change Model (Prochaska & DiClemente,
1986), conceptualizing motivational processes in behaviour change, has received
increasing attention (Miller & Heather, 1998, pp. 49–60). Clinicians and researchers in
health care and social services have found the model helpful when applied to risky
lifestyle behaviours such as eating disorders, smoking and excessive alcohol
consumption. It is believed that treatment outcome can be improved when there is amatch between the patient’s stage of change and the type of treatment offered
(Prochaska & DiClemente, 1986). The model (Prochaska & DiClemente, 1986) has five
stages of change: precontemplation, characterized by lack of problem recognition;
contemplation, characterized by uncertainty; preparation, when a decision to change is
taken, goals are set and plans are made on how to succeed; action, when the individual
experiments with alternative behaviours; and maintenance, along-term reinforcement
of the individual’s new healthier behaviours, so that they stabilize.
There are several tests assessing stages of readiness to change. In the alcoholtreatment field there is the short Readiness to Change Questionnaire (RTCQ; Rollnick,
Heather, Gold, & Hall, 1992), designed for clients with alcohol problems who do not
seek alcohol treatment. It provides scores for three scales representing three of the
stages of change, omitting the preparation and maintenance stages. Several methods of
allocating a patient to a stage of change in RTCQ were examined. Rollnick et al. reported
that the quick method has satisfactory reliability for all three scales in male excessive
alcohol consumers with low alcohol dependence who have sought health care but not
alcohol treatment. Using the quick method, the test has both concurrent (Rollnick et al.,1992) and predictive (Heather, Rollnick, & Bell, 1993) validity. However, the refined
method of stage allocation was shown to have higher validity in terms of prediction of
drinking behaviour as compared to the quick method. Furthermore, the refined method
could also allocate patients to the preparation stage. When testing the matching
hypothesis derived from the Stages of Change Model, only the refined method was
successful in matching risk drinkers to either brief motivational or skill-based
intervention (Heather, Rollnick, Bell, & Richmond, 1996). Patients not ready to change
were more successful with brief motivational intervention than with skill-basedintervention. Thus, this scoring method seems superior to the quick method and could
be used for treatment matching.
The psychometric properties of the RTCQ, when using the quick method, have been
examined by a number of researchers. In a US study, low internal consistency reliability
was reported for the precontemplation and contemplation subscales of the RTCQ,
possibly due to the sample used; it was composed of alcohol-dependent patients seeking
alcohol care (Gavin, Sobell, & Sobell, 1998). In contrast, in a Dutch study (Defuentes-
Merillas, Dejong, & Schippers, 2002) of patients seeking alcohol treatment, the internalconsistency of the three scales was reported to be acceptable and having both
discriminant and concurrent validity. In both a Spanish (Rodriguez-Martos et al., 2000)
and German (Hannover et al., 2002) study, the internal consistency of the three scales
was found to be satisfactory. The German study found construct validity, but the Spanish
study found unsatisfactory concurrent validity. Unfortunately, none of the studies
examined the predictive validity of the RTCQ.
Budd and Rollnick (1996) found theoretical weaknesses in the model of discrete
stages of change and a re-analysis of the original data in the Rollnick et al. (1992) studyfavoured a continuous ‘readiness to change’ variable instead. Budd and Rollnick
proposed that the two alternative models should be examined further with regard to
reliability and validity, contrasting their relative merits. However, to our knowledge such
Lars Forsberg et al.336
studies have not been carried out. In a comprehensive review (Carey, Purmine, Maisto, &
Carey, 1999) of the psychometric properties of tests assessing stages of change, including
the RTCQ, Carey et al. are in favour of a continuous readiness to change construct.
Scores from a valid test assessing stages of change should correlate with change-
related behaviours such as quit attempts, thoughts about quitting, and decisions to
reduce or cut drinking (construct validity; Nunnally, 1967). Moreover, alcohol problemseverity and alcohol consumption have been related to motivation to change. Hence,
patients report more reasons to change drinking with more advanced stages of change
and stage of change is positively associated with alcohol problem severity (Kahler,
2001). Internal motivation for alcohol treatment has been found to be greater in patients
with more severe alcohol problems than in patients with fewer problems (DiClemente,
Bellino, & Neavins, 1999). Vik, Culbertson, and Sellers (2000) report that heavy drinking
students in the contemplation stage drank more and had more negative consequences
than students in the precontemplation and action stages. Thus, a valid readiness tochange test should also correlate with level of severity of alcohol problems. The Stages
of Change Model predicts that patients in the action stage would more likely reduce
alcohol consumption than patients in the contemplation stage and those in the
contemplation stage more likely than patients in the precontemplation stage. In the
Project Match study (Project Match Research Group, 1998), motivation to change was
the most potential predictor of drinking outcome throughout the three-year post-
treatment period. Thus, predictive validity is also important in considering the merits of
a test assessing motivation.The aim of the present study was to investigate the reliability in terms of test–retest
and internal consistency as well as construct (Nunnally, 1967) and predictive validity of
the RTCQ scores, treated either according to the Stages of Change Model or as one
continuous readiness variable (Likert scale). Assigning a stage of change was in its turn
achieved with both the quick method and the refined method. Construct validity was
examined by relating RTCQ scores to severity of alcohol problems, amount of alcohol
consumed, and whether or not the patients would set a goal and form a strategy to
reduce drinking at intervention. Predictive validity was examined by relating RTCQscores to changes in alcohol intake as assessed six months later.
Method
ParticipantsThe study was carried out as part of a randomized controlled study about brief
interventions (Forsberg, Ekman, Halldin and Ronnberg, 2000) approved by the Research
Ethical Committee at Karolinska Hospital (No. 92-327). Patients admitted consecutively
for emergency care in a surgical ward at Danderyd Hospital in Stockholm were asked to
participate in a study regarding, ‘alcohol and well-being’. Data were collected on l6 to
73-year-old patients, excluding those who were terminally ill, did not speak or
understand Swedish, or used drugs illicitly. Patients were informed that their results
would not be registered in their hospital records, that they would receive feedback andwere at liberty to withdraw from the study whenever they wished. Out of 697 patients
approached, 563 (81%) participated in the study. There were no differences in gender
and age between participants and non-participants.
The participants were interviewed using the screening questionnaires Mm-MAST
(Kristenson & Trell, 1982), CAGE (Ewing, 1984), and the Trauma Scale (Skinner, Holt,
Schuller, Roy, & Israel, 1984). The cut-off limit for alcohol problems was two affirmative
Swedish RTCQ: Reliability and validity 337
answers in any of the questionnaires. Positive cases were further assessed with regard to
alcohol consumption, and those having risk consumption according to either of two
criteria were subject to further analysis. The first criterion of risk consumption was
regularly drinking $162 grams of absolute alcohol per week on average for men and
$82 grams for women (Ashley, Ferrence, Room, Rankin, & Single, 1994). The second
criterion was peak consumption during the last 12-month period, for males $1.05
grams of absolute alcohol per kg body weight, and for females $0.90 grams (Babor &
Grant, 1991). Of the patients, 165 (29%) were classified as risk consumers (See Table 1)
and randomized to either extended alcohol counselling (modified Drinkers Check Up;
Forsberg et al., 2000) or a brief assessment and feedback. Both groups were followed
up. The baseline and follow-up assessments consisted of the Time Line Follow Back
Table 1. Descriptive data for the sample
Males(n ¼ 118)
Females(n ¼ 47)
Age16–30 yrs 35% (41) 51% (24)31–59 yrs 53% (62) 45% (21)$60 13% (15) 4% (2)
Civil status: Married/as couple 66% (78) 64% (30)Working status
Employed/studying 78% (92) 74% (37)Retired 12% (14) 11% (5)Unemployed 9% (11) 9% (4)Missing data 1% (1) 1% (1)
Alcohol diagnosis during the last 5 yrs 9% (11) 6% (3)No. of affirmative answers in CAGE
2 19% (23) 15% (7)3–4 22% (25) 13% (6)
No. of affirmative answers in Mm MAST2–3 50% (59) 64% (30)$4 40% (47) 26% (12)
Frequency of binge drinking(men, corresponding to 2 bottles of wine;women, 1 bottle of wine)Never 0% (0) 4% (2)#once per mth 7% (8) 7% (3)2–4 times per mth 53% (63) 49% (23)1–2 times per wk 29% (34) 32% (15)$3 times per wk 11% (13) 9% (4)
IQR1 Median IQR3 IQR1 Median IQR3
Weekly consumption in grams alcohol 39 103 184 28 66 102Peak amount in grams
alcohol/kg body weight1.50 1.95 2.88 1.26 1.67 2.59
Typical amount in gramsalcohol/kg body Weight
0.65 0.95 1.70 0.49 0.85 1.42
Number of sober days in a month 16 23 27 20 25 27
Lars Forsberg et al.338
(TLFB; Sobell & Sobell, 1995) for a 14-day period and frequency and amounts of alcohol
consumed during the past 12-month period (six months at follow-up) and the RTCQ.
The preliminary somatic diagnoses of the risk consumers (N ¼ 165) were classified
in 11 categories: suspicion of appendicitis (27.3%; n ¼ 45), acute abdomene (22.4%;
n ¼ 37), trauma (20.6%; n ¼ 34), GI-bleeding (7.3%; n ¼ 12), proctological ailments
(4.8%; n ¼ 8), ulcer/vomiting (4.8%; n ¼ 8), diverticulitis (3.0%; n ¼ 5), pancreatitis(1.8%; n ¼ 3), gallstone disease/jaundice (1.2%; n ¼ 2), intestinal obstruction (1.2%;
n ¼ 2), and others (5.5%; n ¼ 9).
RTCQThe second and first author of this article translated the RTCQ into Swedish. The
understanding of each item was tested on staff and patients. The first negatively
formulated item in RTCQ was positively reformulated following comments on the first
version. Slightly different wordings of the items were tested repeatedly on staff and
patients until the understanding of the items was clear (see the Appendix).
Since many patients were exhausted, RTCQ was administered as an interview. The
interviewer and the patient had a copy each of the RTCQ and the interviewer read theitems, marking the alternatives that best suited the patient. Each of the three subscales
of the questionnaire measuring precontemplation, contemplation and action contained
four statements each to which the patients responded with one of the following five
alternatives: totally agree/þ2, agree partly/þ1, unsure/0, partly disagree/–1, totally
disagree/–2. Thus, the patients’ score on each scale ranged from 28 to þ8.
The quick method of allocating stage of change was based on the highest score
obtained either on the precontemplation, contemplation or action scales. In the event of
a tie between two scale scores, the most advanced stage was chosen (Heather et al.,1993, Rollnick et al., 1992).
In the refined method of allocating stage of change, only stage score profiles that are
logically consistent with the stages of change model are used. Hence, this method
discards participants with a stage profile with:
. positive scores on precontemplation and action and a negative score on
contemplation;
. positive as well as negative scores in all three scales;
. positive scores on both precontemplation and contemplation; and
. negative scores in precontemplation and contemplation and positive scores in action.
Participants who scored 0 or negatively on precontemplation and positively on
contemplation and action were defined as reflecting the preparation stage of change if
the contemplation score was higher than the action score. If the action score was as
high or higher than the contemplation score, participants were allocated to the action
stage (Heather et al., 1993).All 12 items of the readiness score were combined to form a continuous Likert scale.
The signs of the precontemplation items were changed to represent readiness to
change, which formed a scale ranging from 224 to þ24.
Reliability and validity testsTo investigate the test–retest reliability of the RTCQ, 18 patients responded to the
questionnaire twice. These patients were chosen simply because they were in-patients
and available for the second interview one or two days after the first. Another
Swedish RTCQ: Reliability and validity 339
interviewer most often performed the second interview. Cronbach’s alpha coefficients
were calculated to estimate the internal consistency reliability of the scales. However,
the internal consistency was not possible to calculate for the refined method, since this
method uses stage score profiles instead of scale scores to allocate patients to a stage.
The session protocols of one of the randomized groups (n ¼ 85), the modified
Drinkers Check Up, with motivational interviewing and feedback about risk
consumption, were examined regarding notes about patient considerations to reduce
drinking. To estimate the construct validity (Nunnally, l967), the RTCQ results were
correlated with the patients’ self-reports to reduce or cut drinking (‘goal to change’) and
a strategy to do so (‘strategy to change’); both variables were dichotomous. Construct
validity was also estimated by correlating RTCQ results to amount of alcohol consumed,
severity of alcohol problems and alcohol diagnoses. The five measures of alcohol
consumption were:
(1) weekly consumption in grams of pure alcohol estimated by the TLFB;
(2) typical amount in grams of pure alcohol per kg body weight (‘On a typical day
when you drink alcohol, how much do you drink?’);
(3) peak amount in grams of pure alcohol per kg body weight (‘On the occasion youdrank the most, what and how much did you drink?);
(4) frequency of binge drinking (‘How often have you on one and the same occasion
drunk the equivalent or more of [for males] two bottles/[for females] one bottle of
wine?’);and
(5) sober days (‘In an average month, on how many days do you abstain from alcohol?’).
Indices of alcohol problems were the number of affirmative answers in CAGE and in
Mm-MAST at intervention and a DSM-diagnosis of alcohol dependence or alcohol abusein the last five-year period. The diagnoses were taken from a medical-care register,
where all in-patient care within Stockholm County is recorded. The construct validity
was estimated by calculating Kendall tau correlation coefficients with discrete two or
three level variables, while Spearman rank correlation coefficients were calculated with
continuous data. One-tailed testing at 5% significance level was chosen because of the
hypothesized direction of the correlation.
The predictive validity (Nunnally, 1967) of the three methods to treat RTCQ scores
was also examined. Stages of change and readiness scores were each correlated with the
differences in alcohol intake between baseline and a six-month follow-up with regard to
four measures of alcohol consumption. This change score was calculated as a ratio of the
difference in alcohol intake and the mean of consumption at baseline and at follow-up,
thus compensating for regression towards the mean. Frequency of binge drinking was
coded as ‘increasing–equal–decreasing’ between baseline and follow-up, thus making it
possible to calculate a correlation with RTCQ scores. Follow-up data was available for
122 out of 165 patients (74%).
Results
Quick methodUsing the quick method of stage allocation, out of 165 patients, 92 (56%) were
classified to the precontemplation stage, 40 (24%) to the contemplation stage and 33
(20%) to the action stage. The median score was 3.00 for the precontemplation
scale (IQR1 ¼ 23:00; IOR3 ¼ 6:50), 22.00 for the contemplation scale
(IQR1 ¼ 25:00; IQR3 ¼ 4:00) and 24.00 for the action scale (IQR1 ¼ 28:00;
Lars Forsberg et al.340
IQR3 ¼ 4:00), where IQR ¼ integrative range: The internal consistency reliability was
estimated to be .78 for the precontemplation scale,.80 for the contemplation scale and
.80 for the contemplation scale and .74 for the action scale. Of 18 patient, 16 were
allocated to the same stage of change both times.
There were significant relationships between stage of change and indices of alcohol
problems (yes responses in CAGE and Mm-MAST; see Table 2). However, the
correlations were low between stages of changes and indices of alcohol consumption
measures, an alcohol diagnosis, ‘goal to change’ and ‘strategy to change’ (see Table 2).
Thus, the method had modest construct validity. There was no significant correlation
between stage of change and changes in any of the measures of alcohol consumption at
the six-month follow-up. There was a tendency that patients in the contemplation stage
reduced weekly consumption (t test, p , :07) more often than patients in the action and
precontemplation stages. Thus, the evidence of predictive validity was weak and needs
to be further established.
Refined methodUsing the refined method of stage allocation, out of 165 patients, 70 (42%) were
classified to the precontemplation stage, 16 (10%) to the contemplation stage, 13 (8%)
to the preparation stage, and 13 (8%) to the action stage. However, 53 (32%)
patients were not classifiable. The test–retest reliability was satisfactory, since of the 18
patients the same four patients were unclassified both times and of the remaining
Table 2. Correlations (tau) between assigned stage of change versus continuous readiness score in
relation to alcohol consumption, severity of alcohol problems and patient intentions to reduce drinking
at intervention (N ¼ 165)
Stage of change Readiness score
Quick method Refined methodb Trichotomized Continuousa
Weekly consumption in gramsalcohol
.19** .32** .23** .22**
Peak amount in gramsalcohol/kg body weight
.12* .22** .15** .14**
Typical amount in gramsalcohol/kg body weight
.13* .20** .18** .15**
Frequency of binge drinking(for men corresponding to2 bottles of wine,for women 1 bottle
.22** .27** .23** .21**
Alcohol diagnosis during thelast 5 years
.21** .33** .26** .25**
Affirmative answers in CAGE .47** .61** .56** .54**Affirmative answers in Mm-MAST .34** .47** .46** .40**Setting a goal to
reduce drinking.24*c .49**e .23*c .25**c
Forming a strategy toreduce drinking
.26**d .62**f .36**d .38**d
**Correlation significant , .01 level (1-tailed); *Correlation significant , .05 level (1-tailed).aSpearman rank correlation; bN ¼ 12; cN ¼ 81; dN ¼ 82; eN ¼ 55; fN ¼ 56:
Swedish RTCQ: Reliability and validity 341
14 patients, 13 were allocated to the same stage of Change both times. The refined
method resulted in higher correlations between stages of change on the one hand and
indices of alcohol problems, measures of alcohol consumption, goal to change and
strategy to change on the other (see Table 2). Thus, construct validity was found.
However, there was no significant correlation between stage of change and changes in
any of the measures of alcohol consumption at the six-month follow-up. Thus, nopredictive validity could be found.
The Kendall tau correlation (Spearman rank correlation in parentheses) between the
quick method and the refined method was .97 (.99). Thus, the refined method did not
have unique information as compared to the quick method.
Continuous modelUsing the continuous readiness score, the median readiness score was 27.00
(IQR1 ¼ 216:00; IQR3 ¼ 6:00). The test–retest reliability was .82 (Kendall tau) and the
Spearman rank correlation was .86. The internal consistency reliability of thecontinuous scale was .88 Items deleted from the scale did not increase the reliability.
The average inter-item correlation was .38. The corresponding inter-item correlation in
the quick method was .47 for the precontemplation scale, .51 for the contemplation
scale and .50 for the action scale.
When the continuous readiness score was treated as a trichotomous discrete variable
to be comparable to the analyses of stage of change, 15 of the 18 patients were in the
same third range of the variable both times. Thus, the continuous model had test–retest
reliability. The trichotomized continuous variable was significantly correlated (tau) with
indices of alcohol problems and strategy to change (see Table 2). However, the
significant correlations to alcohol consumption and goal to change were low. The
correlations were of about the same size when using the continuous variable without
trichotomizing it (see Table 2). Thus, reasonable construct validity was found. The
continuous readiness score was weakly but significantly correlated (Spearman rank
correlation) with reduced weekly alcohol consumption (r ¼ 0:15; p , :05) at the six-
month follow-up. This relationship disappeared when using the trichotomized variable.
Thus, only weak predictive validity was found.The tau correlation (Spearman rank correlation in parentheses) between the
continuous readiness score and the quick method was .63 (.80), and between the
continuous method and refined method was .71 (.84).
Discussion
Both the quick method of stage allocation and scoring of RTCQ as a continuous
readiness scale showed satisfactory test–retest reliability. Compared to the quick
method with three stages of change, the internal consistency reliability was higher in the
continuous scale due to more items in the scale despite somewhat lower average inter-item correlations. The quick method showed lower construct validity than the
continuous scale model. Using the trichotomized continuous variable did not lower the
validity. Stage of change using the quick method did predict drinking behaviour after a
six-month period of time, albeit weakly. Thus, patients in the contemplation stage
reduced their weekly consumption more often than patients in the precontemplation
Lars Forsberg et al.342
and action stages. There might be weak predictive validity even for the continuous
variable, but it has to be investigate further.
The cut-off points of the trichotomous division of the variable varied depending on
the sample used. However, the trichotomous division of the continuous variable made
the comparisons between the scoring methods more accurate in the present study. An
objection to the continuous readiness variable model is that more research is needed todevelop norms, to give meaning to the scale scores. In contrast, a stage allocation
immediately informs the clinician of what kind of predictive relations according to the
stages of change model that the client has attained to.
The refined method of stage allocation had satisfactory test–retest reliability and
showed construct validity. However, the high correlations indicating construct validity
might be an artifact as a result of the selection of only 68% of the patients with consistent
response pattern. Thus, no less than 32% of the patients had responded to the RTCQ in a
way that was not consistent with the refined Stages of Change Model. In the original
study (Heather et al., l993) only 22% of the patients had response patterns, which did
not conform to the model. In another sample of alcohol-dependent patients seekingalcohol treatment, 32% could not be classified according to the refined method
(McMahon & Jones, 1996). In contrast, only 16% of the patients attending a
detoxification programme could not be classified to the Stages of Change Model; in the
same study, 28.1% in a sample of offenders with alcohol-related crimes were
unclassifiable (Defuentes-Merillas et al., 2002). Hence, scoring the RTCQ according to a
strictly interpreted Stage of Change Model often seems to leave a significant proportion
of patients without a stage assignment, which suggests that the refined method has
doubtful clinical value. However, in a revised version of the RTCQ (Heather, Luce, Peck,
Dunbar, & James, 1999), no less than 95% of the patients had score profiles that werelogically consistent.
In contrast to earlier findings, the refined method in the present study did not show
predictive validity (Heather et al., 1993). Thus, the refined method did not seem to be of
greater clinical use than the other two methods of treating the RTCQ scores.
The weak predictive validity of the Swedish RTCQ might be due to small changes in
alcohol consumption between baseline and the six-month follow-up. Reported weekly
alcohol consumption was reduced by 28% (37 grams of alcohol), the typical amount by
only 13% (0.15 grams of alcohol/kg body weight) and the peak amount of alcohol by
16% (0.37 grams of alcohol/kg body weight). Thus, improved criteria variables areneeded in future RTCQ validation research.
In another study of the Swedish RTCQ (Forsberg, Halldin, & Wennberg, 2003) a
confirmatory factor analysis found slightly better fit for a three-factor model than for one
continuous variable. However, it seems justifiable from a psychometric point of view to
score the Swedish RTCQ either as a stage of change using the quick method or as a
continuous readiness score.
The questionnaire was a clinically useful basis for discussing risk consumption and
readiness to change a patient’s alcohol use in a non-confronting way. Another potential
use is to examine difficult populations with regard to stage of change as a foundation ofhow to raise the often difficult to approach alcohol issue. In the present sample, 80% of
the patients with hazardous or harmful alcohol use were not ready to change and the
outcome in screening as well as in interventions might be improved if the screening
tests, the administration of the tests and the interventions were adapted to the initially
low awareness of own alcohol problems and readiness to change. Future research
should establish whether the outcome of alcohol screening and brief intervention is
Swedish RTCQ: Reliability and validity 343
improved by adapting the tests, test administration and interventions to patients not
ready to change.
More work should be carried out to improve the reliability and validity of tests of
motivation to change. Hence, there are findings that indicate that the assessed readiness
to change does not represent all of the constructs of patient motivation (McMahon &
Jones, l996). Thus, motivation to change might be a wider construct than what is assessedby the tests. McMahon and Jones (1996) found that negative alcohol expectancies
predicted days to relapse, as did readiness to change, but that the two tests were
uncorrelated. The expectancy test and the readiness to change test each seem to have
their own strengths and to reflect different aspects of patient motivation. A revised
expectancy/motivation hypothesis has been proposed, which states preconditions for a
successful outcome of treatment (McMahon & Jones, 1992, 1996). Recently,
precontemplation has been differentiated in reluctance, rebellion, resignation and
rationalization (DiClemente & Velasquez, 2002), where each type might call for adifferent intervention strategy. Thus, precontemplation might rather be split into four
aspects to be appropiately assessed. The Readiness to Change and the Stages of Change
Models are complex constructs and each needs further study as well as tests measuring
the constructs.
Conclusion
The RTCQ scores treated as a continuous readiness scale are a viable alternative to the
original quick method of assigning a stage of change to a patient. The refined method of
stage allocation seems to be of limited clinical value, leaving a significant proportion of
the patients without a stage assignment. The Swedish RTCQ is reliable and hasreasonable construct validity, but needs further investigation to establish its predictive
validity.
Acknowledgements
The National Institute of Public Health provided this project with financial support. Professor
Hans Bergman has contributed with valuable points of view. Ulf Brodin Ph.D. has acted as
statistical consultant. We are also indebted to the staff of the surgical admission ward (Ward 61 at
Danderyd Hospital) under the leadership of Chief Nurse Mildred Rutstrom as well as to the staff on
the surgical unit and surgical wards.
References
Ashley, M. J., Ferrence, R., Room, R., Rankin, J., & Single, E. (1994). Moderate drinking and health:
Report of an international symposium. Canadian Medical Association Journal, 151, 1–16.
Babor, T. F., & Grant, M. (1991). Project on identification and management of alcohol-related
problems. Report on phase II: A randomized clinical trial of brief intervention in primary
health care. Geneva: World Health Organization.
Budd, R., & Rollnick, S. (1996). The structure of the readiness to change questionnaire: A test of
Prochaska & DiClemente’s transtheoretical model. British Journal of Health Psychology, 1,
365–376.
Carey, K., Purmine, D., Maisto, S., & Carey, M. (1999). Assessing readiness to change substance
abuse. Clinical Psychology in Science and Practice, 6, 245–266.
Lars Forsberg et al.344
Defuentes-Merillas, L., Dejong, C., & Schippers, G. (2002). Reliability and validity of the Dutch
version of readiness to change questionnaire. Alcohol and Alcoholism, 37, 93–99.
DiClemente, C., Bellino, L., & Neavins, T. (1999). Motivation for change and alcoholism treatment.
Alcohol Research and Health, 23, 86–92.
DiClemente, C., & Velasquez, M. (2002). Motivational interviewing and the stages of change. In
W. R. Miller & S. Rollnick (Eds.), Motivational interviewing: Preparing people for change
(pp. 201–216). London: Guilford Press.
Ewing, J. A. (1984). Detecting alcoholism. Journal of the American Medical Association, 252,
1905–1907.
Forsberg, L., Ekman, S., Halldin, J., & Ronnberg, S. (2000). Brief interventions for risk consumption
of alcohol at an emergency surgical ward. Addictive Behaviors, 25, 471–475.
Forsberg, S., Halldin, J., & Wennberg (2003). Psychometric propreties and factor structure of the
Readiness to Change Questionnaire. Alcohol and Alcoholism, 38(3), 276–280.
Gavin, D., Sobell, L., & Sobell, M. (1998). Evaluation of the Readiness to Change Questionnaire
with problem drinkers in treatment. Journal of Substance Abuse, l0, 53–58.
Hannover, W., Thyrian, J., Hapke, U., Rumpf, H.-J., Meyer, C., & John, U. (2002). The Readiness to
Change Questionnaire in subjects with hazardous alcohol consumption, alcohol misuse and
dependence in a general population survey. Alcohol and Alcoholism, 37, 362–369.
Heather, N., Luce, A., Peck, D., Dunbar, B., & James, I. (1999). Development of a treatment version
of the Readiness to Change Questionnaire. Addiction Research, 7, 63–83.
Heather, N., Rollnick, S., & Bell, A. (1993). Predictive validity of the Readiness to Change
Questionnaire. Addiction, 88, 1667–1677.
Heather, N., Rollnick, S., Bell, A., & Richmond, R. (1996). Effects of brief counselling among male
heavy drinkers identified on general hospital wards. Drug Abuse and Alcoholism Review, 15,
29–38.
Kahler, C. (2001). Generation and recall of alcohol-related information in excessive drinkers:
Relation to problem severity, outcome expectancies, and stage of change. Psychology of
Addictive Behaviors, 2, 109–117.
Kristenson, H., & Trell, E. (1982). Indicators of alcohol consumption: Comparisons between a
questionnaire (Mm-Mast), interviews and serum g-glutamyl transferase (GGT) in a health
survey of middle-aged males. British Journal of Addiction, 77, 297–304.
McMahon, J., & Jones, B. (1992). The change process in alcoholics: Client motivation and denial
in the treatment of alcoholism within the context of contemporary nursing. Journal of
Advanced Nursing, 17, 173–186.
McMahon, J., & Jones, B. (1996). Post-treatment abstinence survivorship and motivation for
recovery: The predictive validity of the Readiness to Change (RCQ) and Negative Alcohol
Expectancy (NAEQ) Questionnaires. Addiction Research, 4, 161–176.
Miller, W., & Heather, N. (1998). Treating addictive behaviors: Processes of change (2nd ed.).
New York: Plenum Press.
Nunnally, J. (1967). Psychometric theory. New York: McGraw-Hill.
Prochaska, J., & DiClemente, C. (1986). Toward a comprehensive model of change. In W. Miller &
N. Heather (Eds.), Treating addictive behaviours: Processes of change (pp. 4–27). New York:
Plenum Press.
Project Match Research Group (1998). Matching alcoholism treatments to client heterogenity:
Project Match three-year drinking outcomes. Alcoholism, Clinical and Experimental
Research, 22, 1300–1311.
Rodriguez-Martos, A., Rubio, G., Auba, G., Santo-Domingo, J., Rorralba, L., & Campillo, M. (2000).
Readiness to Change Questionnaire: Reliability study of its Spanish version. Alcohol and
Alcoholism, 35, 270–275.
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.
Swedish RTCQ: Reliability and validity 345
Skinner, H. A., Holt, S., Schuller, R., Roy, J., & Israel, Y. (1984). Identification of alcohol abuse using
laboratory tests and a history of trauma. Annals of Internal Medicine, 101, 847–851.
Sobell, L., & Sobell, M. (1995). Alcohol consumption measures. In J. Allen & M. Columbus (Eds.),
Assessing alcohol problems: A guide for clinicians and researchers (pp. 55–73). Rockville,
MD: National Institute on Alcohol Abuse and Alcoholism.
Vik, P., Culbertson, K., & Sellers, K. (2000). Readiness to change drinking among heavy-drinking
college students. Journal of Studies on Alcohol, 61, 674–680.
Received 1 June 2002; revised version received 2 April 2003
Appendix: Items in the Swedish Readiness to Change Questionnaire showing thehypothesized three stages of change: precontemplation (P), contemplation (C) and action (A)
Items
1.P Jag tycker att jag dricker lagom mycket.I don’t think I drink too much.
2.A Jag forsoker dricka mindre nu an jag gjorde forr.I am trying to drink less than I used to.
3.C Jag tycker om att dricka men ibland dricker jag for mycket.I enjoy my drinking but sometimes
I drink too much.4.C Ibland tycker jag att jag borde minska mitt drickande.
Sometimes I think I should cut down on my drinking.5.P Det ar ingen mening med att tanka pa mina alkoholvanor.
It is a waste of time thinking about drinking.6.A Jag har nyligen minskat min alkoholkonsumtion.
I have just recently changed my drinking habits.7.A Vem som helst kan tala om att gora nagot at sitt drickande, men jag gor faktiskt
nagonting at det.Anyone can talk about wanting to do something about drinking, but I am actually
doing something about it.8.C Jag ar i den situationen att jag borde tanka pa att dricka mindre alkoholhaltiga
drycker.I am at the stage where I should think about drinking less alcohol
9.C Ibland tycker jag att min alkoholkonsumtion blir ett problem.My drinking is a problem sometimes.
10.P Jag behover inte andra mina alkoholvanor.There is no need for me to think about changing my drinking.
11.A Jag, haller faktiskt pa att andra mina dryckesvanor nu.I am actually changing my drinking habits right now.
12.P Det finns ingen anledning for mig att dricka mindre alkohol.Drinking less alcohol would be pointless for me.
Lars Forsberg et al.346