The aim of this study was to examine the reliability and validity of the Korean version of the Impaired Control Scale (K-ICS), a scale to screen patients with alcohol use disorder.
Participants were 173 inpatients with alcohol use disorder (AUD), and 174 normal controls (NC). Both AUD and NC groups completed the K-ICS as well as the Alcohol Dependence Scale (ADS), the Alcohol Abstinence Self-Efficacy Scale (AASES), the Brief Self-Control Scale (BSCS), and the Hospital Anxiety and Depression Scale (HAD).
The internal consistencies of K-ICS were good in both AUD and NC. A good convergent validity was clearly shown by significant correlations with the ADS and the AASES, respectively. But the K-ICS had no or weak correlations with the BSCS and the HAD. The ROC curve analyses indicated that the optimal cut-off points of failed control (FC) and predicted control (PC) were estimated as >15 and >13, respectively. Hierarchical multiple regression analysis suggested that FC is a robust predictor of the severity of AUD.
The K-ICS, especially FC subscale of it appears to be a valid and reliable measure of impaired control among both clinical and non-clinical sample.
Alcohol use disorder (AUD) is characterized by a very high relapse rate. For example, the AUD relapse mostly occurs within 3–4 months after treatment, with 80% of inpatients suffering a relapse within six months after being discharged, while no more than 14.5% and 12.4% of them are known to remain abstinent from alcohol for more than 1 and 2 years, respectively [
Thus, examining the cause of the high relapse rate of AUD based on theories and empirical findings relevant to self-regulation may have significant clinical implications. Researchers on self-regulation define human self-regulation as “a process to change or override a dominant reactive tendency to regulate one’s own emotion, thinking or behavior.” [
Although definitions of self-regulation vary, they include goals in common as an important aspect of self-regulation [
Heather et al. [
Moreover, a study on adolescent’s alcohol consumption [
Given the ICS has consistent reliability and excellent validity, predicts alcohol use problems in many studies, and is widely used to verify the effects of addiction treatment, a standardized Korean version of ICS would have substantial clinical benefits. Thus, this study adapted the ICS in Korean, and verified its reliability and validity, so as to present the cut-off points for distinguishing the AUD patient group from the general population based on the K-ICS, and shed light on the clinical benefits of the K-ICS by determining its predictability about the severity of symptoms in the AUD patient group.
The subjects of this study were AUD patients (n=173) and normal controls (n=174). Both groups consisted of males aged 20–69. The AUD patients were recruited at four sites including 2 alcohol addiction centers run by an alcohol clinic in Gwangju Metropolis and a psychiatric hospital in Gyeonggido, respectively, and an alcohol addiction center attached to a general hospital in Daejeon Metropolis. The AUD patients were inpatients who met the criteria for diagnosing AUD as per DSM-56 and were diagnosed with AUD by psychiatrists. Among those who were willing to participate in the study, those who met at least one of the following criteria were excluded: 1) expressing severe aggressiveness and hostility, 2) being unable to read and write Korean, 3) being unable to respond to the self-report questionnaire due to organic issues such as visual impairment and brain injury, and 4) having other psychiatric disorders such as major depressive disorder, bipolar disorder, and so on.
The normal control group was recruited at diverse sites, e.g., local communities, religious groups and companies, through public notices seeking study participants. The criteria for selecting the normal control group included those who had not been diagnosed with AUD, scored no more than 11 out of 40 on the Korean Version of Alcohol Use Disorder Test (AUDIT-K) in accordance with the criterion as suggested by Lee et al. [
To measure the alcohol self-regulation failure, the ICS [
To screen the alcohol use problems in the normal control group, the AUDIT-K, Korean version of AUDIT [
To determine the severity of alcohol problems in the normal control group and to verify the convergent validity of the K-ICS, the Alcohol Dependence Scale (ADS) developed by Skinner and Allen [
The Alcohol Abstinence Self-Efficacy Scale (AASES) is a self-report test composed of 20 question items developed by Diclemente et al. [
To verify the discriminant validity of the K-ICS, we used the Brief Self-Control Scale (BSCS) designed to measure the general self-control. The BSCS is a self-report scale composed of 13 question items adapted to Korean respondents by Hong et al. [
Zigmond and Snaith [
Collected data was analyzed in the following order. First, to verify the homogeneity of the two groups, the demographics including age and education and the age at first alcohol use underwent independent samples t-test. Second, to determine the validity of the constructs of the K-ICS, the exploratory factor analysis was conducted, and the item-total correlation and internal consistency were calculated. As for the size of the factor loading, whether to select each item was determined based on 0.40 as suggested by Kline [
IBM® SPSS® version 23.0 (IBM Corp., Armonk, NY, USA) and Medcalc® verion 17.6 (MedCalc Software, Ostend, Belgium) were used for data and ROC analyses, respectively. The statistical significance level was <0.05 based on the two-sided test.
The mean age at first alcohol use showed no significant difference between NC and AUD groups (
The exploratory factor analysis was conducted of the three subscales of the K-ICS (
The item-total correlation of each item exceeded 0.40 in both NC and AUD groups. Also, in the NC group, the AC, FC and PC showed high internal consistency of 0.853, 0.903, and 0.919, respectively, which indicates high reliability. In the AUD group, the internal consistency of AC, FC, and PC was 0.770, 0.907, and 0.931, respectively.
In the NC group, the AC was correlated with neither FC nor PC, whereas the FC and PC were strongly correlated (
In the AUD group, the AC showed a weak negative correlation with the FC and PC. Also, a relatively strong correlation was found between the FC and PC. The FC showed a relatively strong positive correlation with the ADS, whereas the PC had a weak correlation with the ADS. That is, the convergent validity of the FC was largely high, whereas that of the PC was rather low. Meanwhile, the weak negative correlation between the AC and ADS suggests the AC could be a variable independent of the ADS. Also, the FC had a relatively strong negative correlation with the AASES, whereas the PC had a weak negative correlation with the AASES. The AC had a weak positive correlation with the AASES.
As in the NC group, the AC had a weak positive correlation with the BSCS, whilst the HAD-D had a weak negative correlation with the HAD-A (in the AUD group). In addition, the FC had a relatively strong negative correlation with the BSCS, whilst the PC had a weak negative correlation with the BSCS. Thus, the FC converges on general self-regulation, whereas the PC has a relatively low degree of convergence. Meanwhile, both FC and PC showed largely weak correlations with HAD-D and HAD-A, which indicates the good discriminant validity.
Prior to the ROC curve analysis, we verified the differences in the three subscale scores on the K-ICS between AUD and NC groups. To that end, we conducted the analysis of co-variance with the age and education level set as covariates. As shown in
Since the AC showed no significant inter-group difference in the foregoing analysis of covariance (ANCOVA), the ROC curve analysis was applied to the FC and PC, not AC (
To determine the criterion-related validity of the K-ICS, we conducted the hierarchical multiple regression analysis, where the ADS was set as the dependent variable, while age, education, general self-regulation (BSCS) and abstinence self-efficacy were set as control variables. As shown in
Developed by Heather et al. [
Therefore, the present study adapted the ICS in Korean to test the impaired self-regulation in AUD, and verified its reliability, validity and clinical utility. The present findings highlighted the following.
First, each item of the three subscales of the K-ICS showed a very high factor loading and a satisfactory explanatory power in both NC and AUD groups. In brief, the K-ICS appropriately reflected the constructs of AC, FC, and PC.
Second, the AC, FC, and PC showed good internal consistency overall in both groups. In particular, the FC (NC: 0.903, AUD: 0.907) and PC (NC: 0.919, AUD: 0.931) had the internal consistency higher than that reported in Heather’s research on cross-validation [
Third, the K-ICS had good convergent and discriminant validity overall. Particularly, in both NC and AUD groups, the FC showed a strong correlation with the ADS, and a relatively strong correlation with the AASES, indicating the high convergent validity. In contrast, the FC showed a relatively weak correlation with the BSCS, inconsistent with the general self-regulation. In addition, in both groups, the FC showed a very weak correlation with each of the two subscales, i.e., depression and anxiety, of HAD which specifies emotional problems, and indicated good discriminant validity.
Fourth, the inter-group difference in the AC was not significant in the ANCOVA. This finding disagrees with Marsh et al. [
Fifth, in the ROC curve analysis of FC and PC, but not AC because of the inter-group differences in the subscales of the K-ICS, the AUD group scored over 0.90 in the FC and PC, which indicates a high discriminant power between the AUD and NC groups. Also, the optimal cutoff scores for the FC and PC were minimum 15 and 13, respectively.
Finally, in the analysis of criterion-related validity with the effects of control variables eliminated, the K-ICS showed an incremental explanatory power by 21% or so. Notably, even after the effects of control variables were taken into account, the FC showed an excellent predictive power about the ADS. Yet, the AC and PC showed no significant incremental explanatory power about the ADS. That is, in predicting the alcohol use problems, behavioral variables such as self-monitoring of one’s past behavior are likely to be a stronger predictor of the PC than cognitive variables such as presumptions. This is why the FC was primarily used among the three subscales of the ICS in previous studies on treatment effects and neuroimaging [
The present study has the following limitations. First, other variables such as family history and medication that could impact on the subjects’ self-regulation of alcohol were not duly considered. Second, the self-report ratings in the AUDIT-K and ADS used to select the normal controls may have triggered some positive biases. Hence, future studies need to conduct indepth interviews with mental health specialists in selecting the normal controls. Third, this study used the ADS as the only criterion variable of the K-ICS. Thus, future studies need be based on actual behavioral measurements such as alcohol consumption when subjects sleep out or go out to test the convergent validity of the K-ICS. Fourth, this study did not measure the test-retest reliability of the K-ICS, failing to secure the consistency of the K-ICS over time, although the consistency of results over time may be less required of the K-ICS which measures individuals’ behavioral aspects or states for a certain period of time instead of their general tendencies or traits.
Despite the foregoing limitations, the present study should be noted in that it verified the objectivity and validity of the K-ICS as an instrument for measuring the impaired self-regulation of alcohol in both normal control and clinical samples in research on AUD. Particularly, the finding that the predictive power of FC over the past six months outstripped that of PC in predicting the severity of alcohol dependence suggests behavior-oriented approaches need be taken in research on AUD in lieu of cognitive approaches.
The present study adapted the ICS widely used abroad to measure the impaired self-regulation of alcohol into the KICS and verified its reliability, validity and clinical utility. Hence, the proposed cutoff scores of the FC and PC are deemed conducive to duly screening people for alcohol problems, regularly monitoring clinical patients for impaired self-regulation of alcohol, and ultimately preventing the relapse of AUD. Furthermore, the limitations of the present study warrant further studies.
Receiver operating characteristic curves of the Korean version of the Impaired Control Scale-Failed Control.
Receiver operating characteristic curves of the Korean version of the Impaired Control Scale-Predicted Control.
Demographic and clinical characteristics of subjects
Variables | Group | M | SD | p-value |
---|---|---|---|---|
Age (years) | NC | 41.86 | 12.19 | <0.001 |
AUD | 50.03 | 8.88 | ||
Education (years) | NC | 15.29 | 2.55 | <0.001 |
AUD | 12.49 | 2.44 | ||
Age of first alcohol drinking (years) | NC | 18.37 | 4.16 | 0.566 |
AUD | 18.71 | 6.24 |
AUD: alcohol use disorder, NC: normal control, M: mean, SD: standard deviation
The factor loadings and items-total correlations of the K-ICS
Factors | No. of items | Factor loadings |
% of variance |
Item-total correlations |
Cronbach’s α |
||||
---|---|---|---|---|---|---|---|---|---|
NC | AUD | NC | AUD | NC | AUD | NC | AUD | ||
AC | 1. | 0.647 | 0.531 | 55.84 | 42.34 | 0.601 | 0.469 | 0.853 | 0.770 |
2. | 0.795 | 0.737 | 0.726 | 0.626 | |||||
3. | 0.761 | 0.530 | 0.674 | 0.458 | |||||
4. | 0.887 | 0.846 | 0.779 | 0.706 | |||||
5. | 0.613 | 0.543 | 0.573 | 0.458 | |||||
FC | 1. | 0.674 | 0.611 | 50.13 | 51.15 | 0.673 | 0.580 | 0.903 | 0.907 |
2. | 0.644 | 0.622 | 0.636 | 0.592 | |||||
3. | 0.708 | 0.745 | 0.713 | 0.707 | |||||
4. | 0.711 | 0.702 | 0.717 | 0.657 | |||||
5. | 0.673 | 0.696 | 0.627 | 0.658 | |||||
6. | 0.806 | 0.866 | 0.732 | 0.814 | |||||
7. | 0.807 | 0.807 | 0.728 | 0.768 | |||||
8. | 0.817 | 0.851 | 0.737 | 0.810 | |||||
9. | 0.528 | 0.599 | 0.481 | 0.575 | |||||
10. | 0.659 | 0.581 | 0.607 | 0.556 | |||||
PC | 1. | 0.669 | 0.609 | 55.56 | 58.52 | 0.660 | 0.587 | 0.919 | 0.931 |
2. | 0.749 | 0.797 | 0.740 | 0.770 | |||||
3. | 0.731 | 0.804 | 0.724 | 0.768 | |||||
4. | 0.807 | 0.848 | 0.800 | 0.814 | |||||
5. | 0.639 | 0.761 | 0.593 | 0.732 | |||||
6. | 0.820 | 0.855 | 0.762 | 0.817 | |||||
7. | 0.861 | 0.867 | 0.806 | 0.830 | |||||
8. | 0.845 | 0.780 | 0.796 | 0.747 | |||||
9. | 0.495 | 0.590 | 0.447 | 0.572 | |||||
10. | 0.762 | 0.680 | 0.728 | 0.663 | |||||
Total items: | 0.950 | 0.933 |
AUD: alcohol use disorder, NC: normal control, AC: attempted control, FC: failed control, PC: predicted control
Correlations among variables of interest
1. | 2. | 3. | 4. | 5. | 6. | 7. | 8. | |
---|---|---|---|---|---|---|---|---|
1. AC | -0.214 |
-0.216 |
-0.157 |
0.387 |
0.186 |
-0.279 |
-0.141 |
|
2. FC | -0.038 | 0.496 |
0.589 |
-0.471 |
-0.426 |
0.275 |
0.333 |
|
3. PC | -0.142 | 0.840 |
0.367 |
-0.388 |
-0.362 |
0.308 |
0.311 |
|
4. ADS | -0.113 | 0.603 |
0.566 |
-0.411 |
-0.311 |
0.392 |
0.438 |
|
5. AASES | 0.240 |
-0.369 |
-0.414 |
-0.327 |
0.399 |
-0.360 |
-0.378 |
|
6. BSCS | 0.099 | -0.336 |
-0.262 |
-0.266 |
0.386 |
-0.448 |
-0.476 |
|
7. HAD-D | 0.012 | 0.237 |
0.178 |
0.192 |
-0.051 | -0.266 |
0.760 |
|
8. HAD-A | 0.026 | 0.261 |
0.210 |
0.248 |
-0.042 | -0.309 |
0.608 |
p<0.05,
p<0.01,
p<0.001,
alcohol inpatients’.
AC: attempted control, FC: failed control, PC: predicted control, ADS: Alcohol Dependence Scale, AASES: Alcohol Abstinence Self-Efficacy Scale, DSRS: Drinking Self-Regulation Scale, BSCS: Brief Self-Control Scale, HAD-D: Hospital Anxiety and Depression Scale-depression, HAD-A: Hospital Anxiety and Depression Scale-anxiety
Comparison of K-ICS scores between AUD and NC group
Variables | Group | M | SD | p-value |
η2 |
---|---|---|---|---|---|
AC | AUD | 10.09 | 5.21 | 0.355 | 0.003 |
NC | 10.02 | 4.17 | |||
FC | AUD | 6.07 | 5.75 | <0.001 | 0.557 |
NC | 25.75 | 8.53 | |||
PC | AUD | 5.35 | 5.96 | <0.001 | 0.409 |
NC | 21.38 | 9.78 |
analysis of covariance adjusted for age and education.
AUD: alcohol use disorder, NC: normal control, AC: attempted control, FC: failed control, PC: predicted control, M: mean, SD: standard deviation
The values of accuracy indices according to various cutoff scores of K-ICS-FC
Cut-off scores | Sensitivity | 95% CI | Specificity | 95% CI |
---|---|---|---|---|
>13 | 93.68 | 89.0–96.8 | 88.44 | 82.7–92.8 |
>14 | 92.53 | 87.6–96.0 | 92.49 | 87.5–95.9 |
>15 |
91.38 | 86.2–95.1 | 94.80 | 90.4–97.6 |
>16 | 89.66 | 84.1–93.8 | 95.95 | 91.8–98.4 |
>17 | 87.93 | 82.1–92.4 | 96.53 | 92.6–98.7 |
cut-off score recommended by this study.
K-ICS-FC: Korean version of the Impaired Control Scale-Failed Control, CI: confidence interval
The values of accuracy indices according to various cutoff scores of K-ICS-PC
Cut-off scores | Sensitivity | 95% CI | Specificity | 95% CI |
---|---|---|---|---|
11 | 81.29 | 74.6–86.8 | 86.47 | 80.4–91.2 |
>12 | 80.12 | 73.3–85.8 | 87.06 | 81.1–91.7 |
>13 |
78.95 | 72.1–84.8 | 91.18 | 85.9–95.0 |
>14 | 76.02 | 68.9–82.2 | 92.94 | 88.0–96.3 |
>15 | 72.51 | 65.2–79.1 | 94.12 | 89.4–97.1 |
cut-off score recommended by this study.
K-ICS-PC: Korean version of the Impaired Control Scale-Predicted Control, CI: confidence interval
Results of hierarchical multiple regression analysis of K-ICS on ADS among AUD group
Model | Predictors | B | SE | β | t | R2 | ΔR2 | F |
---|---|---|---|---|---|---|---|---|
Model 1 | Age | -0.163 | 0.089 | -0.146 | -10.827 | 0.208 | 8.587 |
|
Education | 0.014 | 0.326 | 0.003 | 0.043 | ||||
BSCS | -0.194 | 0.145 | -0.116 | -10.337 | ||||
AASES | -0.221 | 0.052 | -0.358 | -40.223 |
||||
Model 2 | Age | -0.125 | 0.078 | -0.112 | -10.600 | 0. 415 | 0.208 |
12.987 |
Education | 0.269 | 0.288 | 0.067 | 0.934 | ||||
BSCS | 0.288 | 0.131 | 0.017 | 0.219 | ||||
AASES | 0.067 | 0.053 | -0.134 | -10.544 | ||||
AC | 0.075 | 0.173 | 0.032 | 0.431 | ||||
FC | 0.572 | 0.097 | 0.500 | 50.903 |
||||
PC | 0.116 | 0.080 | 0.115 | 10.454 |
p<0.001.
ADS: Alcohol Dependence Scale, AC: attempted control, FC: failed control, PC: predicted control, AASES: Alcohol Abstinence Self-Efficacy Scale, DSRS: Drinking Self-Regulation Scale, BSCS: Brief Self-Control Scale, HAD-D: Hospital Anxiety and Depression Scale-depression, HAD-A: Hospital Anxiety and Depression Scale-anxiety, K-ICS: Korean version of the Impaired Control Scale, SE: standard error