Validation of a Korean version of the Insomnia Catastrophizing Scale and examination of the relation between catastrophizing and sleep-related cognition

Abstract

This study explored the reliability and validity of the Korean version of the Insomnia Catastrophizing Scale (ICS) for assessing catastrophic thoughts related to sleep, shedding light on one of the key cognitive dimensions of insomnia, among the general Korean population. The study analyzed secondary data from 399 participants who completed an online survey conducted during September 26–30, 2022. The survey contained various rating scales, namely, the ICS, Dysfunctional Beliefs and Attitudes about Sleep-16, Glasgow Sleep Effort Scale, Dysfunctional Beliefs about Sleep-2, Insomnia Severity Index, and Patient Health Questionnaire-9. Among the participants, 51.1% were male, with 13.3% reporting past psychiatric symptoms and 12.3%, current psychologic distress. Confirmatory factor analysis confirmed the construct validity of both the nighttime (ICS-N) and daytime (ICS-D) subscales, with good model fits observed. Item response theory analysis indicated strong discrimination ability for both scales. Cronbach’s alphas showed the reliability of internal consistency for ICS-N (alpha = 0.950) and ICS-D (alpha = 0.961). The Korean version of the ICS, including both subscales, demonstrated reliability and validity in assessing sleep-related catastrophic thoughts among the general Korean population.

Introduction

Insomnia refers to the inability to fall asleep at night and the frequent waking up during the night. Insomnia may be acute, which often has clear triggers, or chronic, which typically has more complex causes and lasts more than 3 months. According to Spielman, chronic insomnia is caused by predisposing, precipitating, and perpetuating factors [1]. A chronic insomnia disorder, influenced by these three factors, forms a kind of solid dysfunctional thinking about sleep. As a perpetuating factor, repeated experiences of insomnia lead to negative beliefs about sleep becoming even more ingrained. During the course of each night, patients experience sleep anxiety repeatedly. Patients may also experience cognitive distortions, such as “I will not sleep well tonight” and “If I keep doing this, I will waste the rest of the night.”

Patients with insomnia have a tendency to catastrophize. Catastrophizing refers to the overestimation of the probability that a catastrophe will occur, such as “If this insomnia continues, I will surely lose my job.” [2]. An individual who overemphasizes the probability of a catastrophic event can exaggerate the possible consequences of occurrence [3]. In the context of insomnia, insomnia patients overemphasize the negative effects of sleep loss on their health and social functioning, leading to sleep disturbances [4, 5]. Patients tend to catastrophize the consequences of sleeplessness, particularly on their daytime functioning in terms of the psychobiological inhibition model, which explains stressors makes physiologic and psychologic arousal that leads to occurrence of insomnia [6].

Catastrophizing associated with insomnia can be measured with the Insomnia Catastrophizing Scale (ICS) [7], which consists of 20 items divided into two subscales (nighttime [ICS-N] and daytime [ICS-D]). The questionnaire has been shown to possess acceptable psychometric properties and is associated with significant levels of insomnia. Based on Beck’s common cognitive distortions of insomnia, ICS addresses two specific thought processes in insomnia, worry and rumination, with respect to dwelling on the worst possible outcome [4, 8]. ICS has been found to be a reliable and valid measure of catastrophizing in individuals with insomnia. The 11 items of the ICS-N revealed good factor loadings (0.65–0.84) and reliability of internal consistency (α = 0.92), and the 6 items of the ICS-D showed good factor loadings (0.74–0.84) and reliability of internal consistency (α = 0.91). Both subscales showed good convergent validity with the other rating scales [7]. It has been validated in various languages, such as Turkish [9], Portuguese [10], and Japanese [5]. However, validation among the Korean population has not been explored yet. While applying the Cognitive Behavioral Therapy for Insomnia (CBT-I), alleviating the sleep-related catastrophic thoughts is important to reduce insomnia severity [5]. In the clinical setting in Korea, developing the appropriate rating scale to assess sleep-related catastrophic thought might be useful to help insomnia patients. Therefore, our study aimed to explore the reliability and validity of the Korean version of the ICS among the general Korean population.

Methods

Participants and procedure

This study used data came from a previous study on insomnia severity in the general Korean population that conducted a survey and follow-up survey [11]. In the previous study, Chung et al. conducted a survey among participants aged 18–60 years old: a survey (T1) on September 26–30, 2022 that collected 400 responses and another survey (T2) on November 5–16, 2022 that collected 234 responses. For the present study, we used the T1 survey dataset (N = 400). We conducted their surveys via an online survey platform of EMBRAIN, a professional survey company in Seoul, Korea [12]. We estimated the sample size based on the centrum limit theorem [12]: we allocated 50 samples for eight cells (sex × four age groups: 20, 30, 40, and 50 s), thereby deriving 400 samples. The company sent enrollment emails to 7195 participants among 1,640,000 registered panelists; of them, 1021 accessed and 460 completed the survey [12]. We received the first 400 responses without identifying information after excluding incomplete or above the wo standard deviation of mean time of response speed [12]. In the present study, we found one participant who responded inappropriately to the questions in the sleep indices. Therefore, we included only 399 responses in the final analysis. The survey questionnaire included questions on age, sex, marital status, psychiatric history, current psychiatric distress, and rating scales. All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional Review Board of Asan Medical Center (2022-1283), and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The IRB waived obtaining written informed consent form from the participants. Rather, participants can begin the survey when they select “yes” to the question of agreement at the beginning of the survey.

Rating scales

ICS

The ICS is a self-report rating scale that consists of daytime (ICS-D) and nighttime (ICS-N) subscales [7]. The ICS-N and ICS-D includes 11 and 6 items, respectively, for assessing catastrophizing thoughts related to sleep disturbance. We developed a Korean version of ICS through translation and back-translation, with approval from the original authors of the ICS. First, we asked two bilingual experts to separately translate the English version of the ICS to Korean. Second, we asked two other bilingual experts to translate this Korean version to English without referring to the original ICS. Lastly, we asked a third party to compare and verify the versions. After completing the translation and back-translation process, corresponding author (S.C.) finally checked the translated version.

Dysfunctional Beliefs and Attitudes about Sleep-16 (DBAS-16)

The DBAS-16 is a self-report rating scale consisting of 16 items for assessing sleep-related dysfunctional beliefs [13]. Items are scored from 0 (strongly disagree) to 10 (strongly agree). High final average scores of all 16 items indicate high levels of dysfunctional beliefs about sleep. We applied the validated Korean version of DBAS-16 [14]. In this study, the Cronbach’s alpha was 0.912.

Glasgow Sleep Effort Scale

The seven-item GSES is a self-report rating scale for assessing preoccupation with sleep [15]. Items are scored on a three-point Likert scale (0: not at all, 1: to some extent, 2: very much). A higher total summed score indicates a greater level of preoccupation with sleep. We applied the validated Korean version of the scale [16]. The Cronbach’s alpha for our sample was 0.827.

Dysfunctional Beliefs about Sleep-2 (DBS-2)

The DBS-2 is a self-report rating scale for assessing dysfunctional beliefs about sleep [17]. The two items are rated on a 0–10 scale, with a higher averaged score indicating a higher level of dysfunctional beliefs about sleep. We applied the original Korean version, and split-half coefficients among this sample was 0.844.

Insomnia Severity Index (ISI)

The ISI is a self-report rating scale with seven items for assessing the severity of insomnia [18]. The items are scored on a five-point Likert scale (0–4), and the summed total score reflects the severity degree of insomnia. We applied the validated Korean version [19], and Cronbach’s alpha among this sample was 0.810.

Patient Health Questionnaire-9 (PHQ-9)

The PHQ-9 is a self-report rating scale for measuring the severity of depression [20]. The scale’s nine items can be scored on a four-point Likert scale. Higher total scores indicate worse levels of depression. We applied the validated Korean version[21], and Cronbach’s alpha among this sample was 0.905.

Statistical analysis

First, we examined the construct validity of the ICS-N and ICS-D using confirmatory factor analysis (CFA). We checked for normality assumption using skewness and kurtosis of all items based on the value within the range from −2 to + 2 [22]. We examined data suitability and sampling adequacy using Kaiser–Meyer–Olkin (KMO) test and Bartlett’s test of sphericity. We then checked the model fitness of CFA through comparative fit index (CFI), Tucker–Lewis index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR) values [23]. We performed multi-group CFA to investigate whether the ICS-N and ICS-D can measure catastrophizing thoughts related to sleep disturbance in the same way across sex (male vs. female) and severity of insomnia (ISI ≥ 8 vs. ISI < 8) variables. Second, we applied a graded response model (GRM), an item response theory model for polytomous items, to explore the validity of measurement scales. In GRM, item fits, slope, and threshold parameters were calculated. Third, we confirmed internal consistency reliability using Cronbach’s alpha and McDonald’s omega. We performed Pearson’s correlation analysis to explore the convergent validity of the ICS-N and ICS-D with other existing rating scales. We posited that the ICS scale may be positively correlated with DBAS-16, GSES, DBS-2, ISI, and PHQ-9. We used JASP version 0.14.1.0 (JASP Team, Amsterdam, The Netherlands) and R version 4.0.5 [24] using Rstudio version 2023.06.1 + 524 [25] to conduct the statistical analyses. CFA and Mutligroup CFA were performed using the R package lavaan v 0.6-18 [26], and GRM was performed using the R package mirt v 1.41 [27].

Results

Table 1 shows the characteristics of the 399 participants. A total of 204 respondents (51.1%) were male, 53 (13.3%) reported experiencing past psychiatric symptoms, and 49 (12.3%) had current psychologic distress.

Table 1.

Baseline demographic and clinical characteristics of the study subjects

Variable	Participants (N = 399) Mean ± SD, N (%)
Male, n (%)	204 (51.1%)
Age (years)	41.1 ± 10.2
Marital status
Single	172 (43.1%)
Married, with kids	185 (46.4%)
Married, without kids	37 (9.3%)
Others	5 (1.3%)
Psychiatric history
Have you experienced or have you been treated for depression, anxiety, or insomnia? (Yes)	53 (13.3%)
Currently, do you think that you are depressed or anxious, or do you need help regulating your mood state? (Yes)	49 (12.3%)
Symptoms rating
Insomnia Catastrophizing Scale (ICS)
ICS-N	21.1 ± 12.6 (0–55)
ICS-D	12.3 ± 7.7 (0–30)
Dysfunctional Beliefs and Attitudes about Sleep-16 (DBAS-16)	5.3 ± 1.5 (1.3–10)
Glasgow Sleep Effort Scale (GSES)	11.3 ± 2.9 (7–21)
Dysfunctional Beliefs about Sleep-2 (DBS-2)	7.1 ± 2.1 (0–10)
Insomnia Severity Index (ISI)	11.4 ± 5.3 (0–28)
Patient Health Questionnaire-9 (PHQ-9)	6.4 ± 5.5 (0–27)

Confirmatory factor analysis

Table 2 presents the normality assumption for all items. The KMO value of 0.958 and Bartlett’s test of sphericity (p < 0.001) showed that the data were suitable and sampling was adequate for conducting CFA. We conducted CFA using items of the ICS-N and ICS-D separately as a single factor-model, in accordance with the original report [7], and we observed good fits for ICS-N (CFI = 0.999, TLI = 0.999, RMSEA = 0.016, and SRMR = 0.044, Table 3) and ICS-D (CFI = 1.000, TLI = 1.000, RMSEA = 0.000, and SRMR = 0.029). Multi-group CFA showed that both ICS-N and ICS-D had scalar level invariance across sex (male vs/ female; ∆CFI = 0.000, ∆CFI = 0.000, respectively) and insomnia variables (ISI ≥ 8 vs. ISI < 8; ∆CFI = 0.000, ∆CFI = 0.000, respectively). These results indicate that both ICS-N and ICS-D assess same construct across sex and groups having insomnia or not and confirm the validity of assessing mean differences in true scores between these groups.

Table 2.

Item-level properties of the Korean version of the ICS

Items	Mean	SD	Skewness	Kurtosis	Corrected item-total correlation	Factor loading
A) ICS-N
Item 1	1.65	1.30	0.509	−0.317	0.752	0.772
Item 2	1.56	1.31	0.435	−0.728	0.710	0.727
Item 3	2.61	1.36	−0.390	−0.691	0.692	0.709
Item 4	2.24	1.43	0.028	−0.817	0.729	0.747
Item 5	1.82	1.34	0.210	−0.878	0.793	0.814
Item 6	2.13	1.51	0.112	−1.023	0.795	0.816
Item 7	1.87	1.47	0.284	−1.049	0.841	0.867
Item 8	2.00	1.50	0.094	−1.161	0.793	0.816
Item 9	1.63	1.40	0.450	−0.751	0.833	0.860
Item 10	1.48	1.39	0.472	−1.034	0.802	0.828
Item 11	2.03	1.42	0.113	−1.055	0.784	0.808
B) ICS-D
Item 1	2.22	1.29	−0.023	−0.739	0.835	0.853
Item 2	1.82	1.38	0.228	−0.911	0.873	0.892
Item 3	1.99	1.42	0.178	−0.938	0.876	0.894
Item 4	2.12	1.43	0.055	−0.989	0.895	0.918
Item 5	2.02	1.46	0.062	−1.096	0.905	0.930
Item 6	2.15	1.42	0.043	−1.018	0.876	0.898

Table 3.

Scale-level properties of the Korean version of the ICS

Psychometric properties	ICS-N	ICS-D	Suggested cut off
Cronbach’s alpha	0.950	0.961	≥ 0.7
McDonald’s omega	0.951	0.962	≥ 0.7
Standard error of measurement	2.82	1.52	Smaller than SD /2
Average variance extracted	0.637	0.806	≥ 0.5
Composite reliability	0.951	0.961	≥ 0.7
Split-half reliability	0.965	0.967	≥ 0.7
IRT reliability	0.950	0.953	≥ 0.7
Model fit of confirmatory factor analysis
χ2 (df, p value)	48.475 (44, 0.279)	7.128 (9, 0.624)	Non-significant
CFI	0.999	1.000	> 0.95
TLI	0.999	1.000	> 0.95
RMSEA	0.016	< 0.001	< 0.08
SRMR	0.044	0.029	< 0.08

Item response theory

Table 4, Figs. 1, and 2 presents the GRM output (item fits, slope, and threshold parameters) of both ICS-N and ICS-D. Concerning ICS-N, non-significant S-χ² values suggested that all the items had good fit. All the items had a very high slope ranging from 1.950 (Item 3) to 4.424 (Item 9). Threshold parameters showed Item 3 as the least difficult item in which a higher latent trait (nighttime insomnia catastrophizing) was required to endorse 5th and 6th response options. Items 2, 9, and 10 were the most difficult items, in which a higher latent trait was required to endorse 2nd to 6th response options. Concerning ICS-D, the corresponding S-χ² values of all the items were non-significant at p = 0.01, which suggested that all the items had good fit. Similar to ICS-N, all the ICS-D items had a very high slope, ranging from 3.379 (Item 1) to 6.217 (Item 5). Results on the threshold parameters showed that a higher latent trait (daytime insomnia catastrophizing) was required to endorse 2nd to 6th response options across all items. `

Table 4.

Slope parameters and threshold parameters of the Korean version of the ICS

Items	Item fit			Slope parameter (a)	Threshold parameter (b)
	S-χ2	df	pvalue		b1	b2	b3	b4	b5
A) ICS-N
Item 1	55.153	61	0.687	2.772	−0.844	−0.018	0.831	1.526	2.200
Item 2	80.304	70	0.471	2.318	−0.800	0.085	0.782	1.593	2.757
Item 3	96.993	73	0.117	1.950	−1.736	−0.972	−0.157	0.730	2.197
Item 4	80.311	76	0.632	2.276	−1.299	−0.447	0.209	1.172	1.872
Item 5	55.488	56	0.632	2.955	−0.885	−0.114	0.568	1.363	2.306
Item 6	63.843	65	0.632	3.023	−0.984	−0.260	0.281	0.998	1.756
Item 7	57.640	50	0.471	4.223	−0.751	−0.019	0.413	1.047	1.888
Item 8	104.231	66	0.022	3.037	−0.799	−0.178	0.340	1.054	2.047
Item 9	45.443	49	0.680	4.424	−0.541	0.060	0.615	1.297	1.928
Item 10	53.659	52	0.632	3.673	−0.381	0.217	0.640	1.362	2.617
Item 11	88.644	63	0.099	2.773	−1.049	−0.180	0.342	1.115	2.206
B) ICS-D
Item 1	47.236	26	0.024	3.379	−1.432	−0.545	0.171	1.071	2.109
Item 2	39.135	28	0.095	4.142	−0.825	−0.124	0.420	1.267	2.058
Item 3	45.113	28	0.042	3.936	−0.927	−0.212	0.354	1.103	1.929
Item 4	34.417	21	0.050	5.330	−1.016	−0.342	0.241	0.939	1.792
Item 5	25.606	20	0.179	6.217	−0.848	−0.231	0.229	1.002	1.827
Item 6	40.938	22	0.024	4.941	−1.078	−0.336	0.248	0.886	1.887

p values adjusted for false discovery rate

Fig. 1.

Item characteristic curve for the ICS-N. Here, X axis indicated the latent trait or ability (nighttime insomnia catastrophizing) and Y axis indicated probability. P1 denotes the curve for the response option ‘0’, P2 denotes the curve for the response option ‘1’, P3 denotes the curve for the response option ‘2’, P4 denotes the curve for the response option ‘3’, P5 denotes the curve for the response option ‘4’, and P6 denotes the curve for the response option ‘5’

Fig. 2.

Item characteristic curve for the ICS-D. Here, X axis indicated the latent trait or ability (daytime insomnia catastrophizing) and Y axis indicated probability. P1 denotes the curve for the response option ‘0’, P2 denotes the curve for the response option ‘1’, P3 denotes the curve for the response option ‘2’, P4 denotes the curve for the response option ‘3’, P5 denotes the curve for the response option ‘4’, and P6 denotes the curve for the response option ‘5’

Reliability and evidence based on relations to other variables

Cronbach’s alpha, McDonald’s omega, composite reliability, split-half reliability, and IRT reliability were 0.950, 0.951, 0.951, 0.965, and 0.950 for ICS-N and 0.961, 0.962. 0.961, 0.967, and 0.953 for ICS-D, respectively (Table 3). The ICS-N score was significantly positively correlated with moderate to high degree with ICS-D (r = 0.76, p < 0.01), DBAS-16 (r = 0.66, p < 0.01), GSES (r = 0.74, p < 0.01), ISI (r = 0.62, p < 0.01), and PHQ-9 (r = 0.53, p < 0.01, Table 5), and positively with low degree with DBS-2 (r = 0.21, p < 0.01). The ICS-D score was significantly correlated positively with moderate to high degree with DBAS-16 (r = 0.69, p < 0.01), GSES (r = 0.59, p < 0.01), ISI (r = 0.43, p < 0.01), and PHQ-9 (r = 0.41, p < 0.01, Table 5), and positively with low degree with DBS-2 (r = 0.23, p < 0.01).

Table 5.

Pearson correlation coefficients of variables for all participants (n = 399)

Variables	Age	ICS-N	ICS-D	DBAS-16	GSES	DBS-2	ISI
ICS-N	−0.07
ICS-D	−0.14**	0.76**
DBAS-16	−0.09	0.66**	0.69**
GSES	−0.12*	0.74**	0.59**	0.60**
DBS-2	−0.04	0.21**	0.23**	0.37**	0.15**
ISI	−0.03	0.62**	0.43**	0.50**	0.55**	0.22**
PHQ-9	−0.05	0.53**	0.41**	0.40**	0.46**	0.07	0.53**

ICS Insomnia Catastrophizing Scale, DBAS-16 Dysfunctional Beliefs and Attitudes about Sleep-16, GSES Glasgow Sleep Effort Scale, DBS-2 Dysfunctional Beliefs about Sleep-2, ISI Insomnia Severity Index, PHQ-9 Patient Health Questionnaire-9

**p < 0.01, *p < 0.05

Discussion

Our study demonstrated the reliability and validity of the Korean version of the ICS-N and ICS-D as rating scales for measuring sleep-related catastrophizing thoughts. Based on the CFA, the single models of the ICS-D and ICS-N showed good model fit. The Korean version of the ICS-N and ICS-D had good item discrimination indices in both classical and item response theory approaches. In addition, GRM showed that both scales provided sufficient information about sleep-related catastrophizing thoughts.

In this study, we conducted the CFA for the single-factor model of both ICS-N and ICS-D subscales separately, in accordance with the original report [7]. Both subscales showed a good fit for the model. Item response theory results showed that items are able to discriminate between varying levels of the underlying latent trait, supporting the structural validity of the scales in assessing sleep-related catastrophizing thoughts. It is noteworthy that the threshold parameters indicated a range of difficulty across items, providing insight into the scale’s responsiveness to various degrees of catastrophizing. These results were consistent with the Japanese version of the ICS [5]. Best to the authors knowledge, this is the second study that assessed the psychometric properties of the ICS using a more advanced item response theory approach.

The ICS-N and ICS-D showed good convergent validity with the other rating scales measuring dysfunctional sleep-related cognition, namely, DBAS-16, GSES, and DBS-2. Thus, the result shows the catastrophizing thoughts related to insomnia might be one of core features parallel to dysfunctional beliefs about sleep (DBAS-16 and DBS-2) or preoccupation with sleep (GSES). The original study [7] reported that the two ICS subscales showed a good convergent validity with sleep-related worry measured with the Anxiety and Preoccupation about Sleep Questionnaire, cognitive pre-sleep arousal measured with cognitive subscale from the Pre-Sleep Arousal Scale, or anxiety level measured with the anxiety subscale of the Hospital Anxiety and Depression Scale. However, the study did not report the convergent validity of two ICS subscales in relation to other scales of sleep-related cognition. In the validation study for the Japanese version of the ICS [5], ICS showed a good convergent validity with DBAS-16.

In the psychobiological inhibition model proposed by Espie [6], catastrophic thought may act on cognitive arousal on the same level with dysfunctional beliefs about sleep. However, sleep-related catastrophic thought is a concept differ from dysfunctional beliefs about sleep, though the two share the same concepts at some point. Catastrophic thoughts focus on immediate consequences of sleeplessness, while dysfunctional beliefs are general attitudes toward sleeplessness long-term in nature. On the other hand, preoccupation with sleep is placed under the similar level with sleep-related catastrophic thought and dysfunctional beliefs about sleep [6] as these are related with sleep-related cognitive de-arousal. Thus, in this study, we explored the convergent validity of the Korean version of the ICS scale in relation to the concepts of dysfunctional beliefs about sleep (DBAS-16) and preoccupation with sleep (GSES).

Our study had several limitations. First, the study used data from an anonymous online survey, which might lead to selection bias due to approachability of online. However, numerous sleep-related rating scales have been validated using online surveys [5, 10]. Despite the limitations of the online survey, it is often employed in research. Second, this study targeted the general population, not a clinical sample of patients with insomnia. However, we conducted multi-group CFA to examine whether the ICS assess in the same way across groups with insomnia or not to control this issue. Further study is needed on patients with insomnia to explore the usability of the scale in clinical practice.

In conclusion, the ICS-N and ICS-D scales, with demonstrated good reliability and validity, can be valuable tools to be applied in both clinical practice and research. The scales can measure the concept of catastrophizing in the context of insomnia, a cognitive factor that plays a critical role in the development and maintenance of insomnia. The Korean version will be beneficial in exploring the characteristics of insomnia, developing treatment strategies, and contributing to future research efforts aimed at improving the sleep health of Korean adults.

Author contribution

Conceptualization: Junseok Ahn, Oli Ahmed, Young Rong Bang, Seockhoon Chung, Markus Jansson-Fröjmark‬. Data curation: Junseok Ahn, Young Rong Bang, Seockhoon Chung. Formal analysis: Oli Ahmed, Seockhoon Chung‬. Methodology: Junseok Ahn, Oli Ahmed, Young Rong Bang, Seockhoon Chung‬. Writing—original draft: all authors. Writing—review and editing: all authors.‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬‬

Declarations

Conflict of interest

The authors have no competing interests to declare.

Ethical statement

All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional Review Board of Asan Medical Center (2022–1283), and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. The IRB waived obtaining written informed consent form from the participants. Rather, participants can begin the survey when they select “yes” to the question of agreement at the beginning of the survey.