INTRODUCTION
Patients with insomnia may exhibit bias toward negative sleep-related information [1]. Sleep disturbances often lead these individuals to go to bed earlier in an attempt to fall asleep more quickly and remain in bed longer in hopes of extending total sleep duration [2]. Meanwhile, based on the concept of goal adjustment capacity, people tend to lower their aspirations as a form of psychological self-protection, particularly when faced with unattainable goals or significant life disruptions [3]. However, persistently pursuing an unattainable goal often leads to negative emotional states such as depression or anxiety. This type of discrepancy is frequently observed in individuals with insomnia, who often express a strong desire to sleep for at least 5 hours when asked about their preferred sleep duration. However, they unconsciously aimed to sleep for a longer time (i.e., from 9 PM to 7 AM), a duration that exceeded their stated preference. A discrepancy between the desired total sleep time (dTST) and the desired time in bed (dTIB) was previously identified and developed as an index: the discrepancy between desired time in bed and desired total sleep time (DBST) index [4]. It is a concept of such cognitive dissonance between implicitly desired sleep time and desperately wanted sleep time. In clinical practice, physicians can evaluate this discrepancy by asking straightforward questions such as “How many hours do you want to sleep?” and “From when to when do you want to sleep?”
The DBST index is significantly correlated with insomnia severity among various groups [4-7], and changes in the DBST index are also correlated with changes in insomnia severity [8]. Lee et al. [4] reported that the DBST index was significantly correlated with insomnia severity as measured with the Insomnia Severity Index (ISI) among the general population. Ahn et al. [7] also reported this association among elderly adults. Among clinical samples of patients with insomnia, Kim et al. [6] also reported that the DBST index was significantly correlated with the ISI score. Among patients with cancer, who often have sleep disturbances [9], Cho et al. [5] reported the association of the DBST index with the ISI score and long sleep onset latency (SOL). The DBST index was also significantly associated with sleep-related cognition. The DBST index was significantly associated with preoccupation with sleep, measured with the Glasgow Sleep Effort Scale, among the general population [4] and older adults [7]. Dysfunctional beliefs about sleep were also reported to be associated with the DBST index [7]. An association was also reported between the DBST index and sleep-related appraisals [10].
However, there are several limitations to the DBST index. First, there was no significant association between the DBST index and insomnia severity among shift-working nurses [11]. Shift-working nurses may have responded to the questions assuming they do not work shifts, or the combination of severe insomnia and the younger age of shift-working nurses may have led them to report a longer desired sleep duration compared to patients with insomnia who tend to state a shorter dTST. Thus, it potentially reduces the discrepancy. Second, it is difficult to explain negative values of the DBST index. Some patients say that they want 8 hours of dTST, but they respond with 6 hours (from 12 AM to 6 AM) of dTIB. In this case, the DBST index is estimated as -2. Higher discrepancy reflects higher insomnia severity, but it is not clear what a negative number means in terms of insomnia severity. Third, calculating the DBST index by subtracting dTST from dTIB may limit the range of scores compared to when the two factors are summed. The restricted range may not accurately reflect the real association with a certain variable such as sleep-related cognition. Fourth, a higher proportion of participants reported the DBST index as “0” because they calculated their dTIB based on their response to dTST. For instance, if a patient indicates a desire to sleep for 6 h, they might report their dTIB as spanning from 12 AM to 6 AM, aligning their dTIB with their stated dTST (6 hours).
Therefore, a modified approach to estimating the DBST index is needed, while maintaining the concept of discrepancy between dTST and dTIB. In this study, a questionnaire-based version of the DBST index was developed, designed to more easily measure an individual’s discrepancy between dTST and dTIB, while addressing the limitations of the original DBST index.
METHODS
Participants and Procedure
Two anonymous online surveys were conducted to develop the questionnaire version of the DBST index. The Study I survey was conducted via a professional survey platform (ENBRAIN) from June 27 to July 9, 2025. The sample size was estimated at 600 by allocating 50 samples for each of 12 cells (sex×six age groups) [12]. A total of 6,612 enrollment emails were sent to 1.8 million general population registrants; 1,536 accessed the survey and 664 completed it. A total of 600 responses were included in the final analysis after excluding the fastest 5% of respondents in each quota (based on response time) and those whose time spent between questions exceeded three times the overall average. The study protocol received approval from the Institutional Review Board of the Asan Medical Center (2025-0804).
The Study II survey was conducted via a professional survey platform (ENBRAIN) from June 18 to 24, 2025. In Study II, participants who responded “yes” to the question “Have you experienced difficulty falling asleep or maintaining sleep in the past three months?” were enrolled. The sample size was estimated at 600 by allocating 50 samples per each of the 12 cells (sex×six age groups) [12]. A total of 7,558 enrollment emails were sent to 1.8 million registered individuals from the general population registrants; 1,882 accessed the survey and 640 completed it. Similar to Study I, a total of 600 responses were provided to the researchers after excluding the fastest 5% of respondents in each quota based on response time as well as those whose time spent between questions exceeded three times the overall average. The study protocol was approved by the Institutional Review Board of the Asan Medical Center (2025-0607).
From two surveys, a total of 1,031 participants from the initial pool of 1,200 were included in the analysis for this study, following the exclusion of 169 individuals whose bedtimes fell outside the range of 9 PM to 1 AM or whose wake-up times occurred after 9 AM, to prevent the possible effect of the circadian rhythm sleep wake disorders.
Measures
Developing the questionnaire version of the DBST index
Questions reflecting the concept of the DBST index were developed as follows: Q1, “I desperately need at least five hours of sleep”; Q2, “I want to fall asleep early in the evening and sleep well into the late morning”; Q3, “I do not want to go to bed early in the evening”; and Q4, “I do not want to wake up late in the morning.” Q1 was formulated to describe agreement with dTST and Q2 with dTIB. Considering reverse scoring, Q3 and Q4 were designed to reflect preferences for earlier bedtimes and later wake-up times, respectively, which together may contribute to an increase in dTIB. Participants were asked to rate their level of agreement to each question on a Likert scale of 0 (strongly disagree) to 10 (strongly agree). The final score was calculated as the average score of all questions that were ultimately selected.
Sleep indices and the DBST index
In the clinic, physicians should routinely ask patients about their usual bedtime, time to fall asleep, or finally getting-out-of-bed time. Responses from patients were used to estimate the time variables such as bedtime, sleep onset time, and wake-up times. Time variables were transformed into numerical variables for statistical analysis or estimating duration (e.g., 15 minutes to 0.25, 30 minutes to 0.50). In this study, we excluded 169 individuals whose bedtimes fell outside the range of 9:00 PM to 1:00 AM or whose wake-up times occurred after 9:00 AM. Consequently, PM time variables were transformed into numeric values ranging from 9.0 to 12.0, while AM time variables were transformed into values ranging from 12.0 to 21.0. Using these time variables, duration variables such as SOL or TIB were calculated [13]. SOL can be estimated as [sleep onset time - bedtime], and TIB as [wake-up time - bedtime]. Additionally, participants’ total time in bed over a 24-hour period (TIB/d) [14] was evaluated to investigate the amount of time they spent lying down during the day and night.
The DBST index was calculated as the difference between dTIB and dTST [4]. Patients were asked about their desired sleep duration and period using the questions “How many hours do you want to sleep per day?” and “From what time to what time do you want to sleep?” The DBST index was then calculated as [dTIB - dTST].
ISI
The ISI is a seven-item instrument designed to assess the severity of insomnia symptoms [15]. Participants rated each statement using a five-level response system, resulting in a total score ranging from 0 to 28. Higher scores indicate more severe insomnia symptoms. In this study, the Korean adaptation of the ISI was used [16], which demonstrated a Cronbach’s alpha of 0.874, indicating strong internal consistency.
Patient Health Questionnaire-2
The Patient Health Questionnaire-2 (PHQ-2) [17] is a condensed version of the PHQ-9 [18], designed to evaluate the severity of depressive symptoms. Participants rated each item on a four-point scale, ranging from 0 (not at all) to 3 (nearly every day), with total scores ranging from 0 to 6. Higher scores indicate greater severity of depression. In this study, the Korean version of the PHQ-2 [19] demonstrated strong internal consistency, as evidenced by a Cronbach’s alpha of 0.805.
Dysfunctional Beliefs and Attitudes about Sleep-6
The Dysfunctional Beliefs and Attitudes about Sleep-6 (DBAS-6) [20] is a shortened version of the original DBAS-16 [21], developed using machine learning to assess individuals’ levels of dysfunctional beliefs about sleep. Participants rated items on a scale from 0 (strongly disagree) to 10 (strongly agree). The final score is calculated as the average of all items, with higher averages indicating more pervasive maladaptive sleep-related beliefs. In this study, the Korean version of the DBAS-6 [20] demonstrated a Cronbach’s alpha of 0.780, indicating high internal consistency.
Statistical Analysis
Descriptive statistics for demographic variables and rating scale scores were presented as mean±standard deviation for continuous variables and as numbers and percentages for categorical variables. Differences between the samples from Study I and Study II were compared using Student’s t-test for continuous variables and the chi-squared test for categorical variables. Statistical significance was defined at a two-tailed p-value <0.05. Pearson’s correlation coefficients were calculated to examine relationships between the newly developed items for the modified questionnaire version of the DBST index and the DBST index itself, clinical characteristics, sleep indices, and rating scales scores in each Study. After selecting significantly meaningful items, a brief questionnaire was created by averaging their scores. Pearson’s correlation coefficients were re-examined to explore the relationship between the developed questionnaire and clinical variables in each study. To identify variables that contribute to the questionnaire scores, linear regression analyses were conducted in each study. Finally, a mediation analysis was performed among all participants to explore whether the developed questionnaire directly influences insomnia severity and whether sleep-related cognition and depression mediate this relationship. All statistical analyses were performed using jamovi version 1.6.18.0 (https://www.jamovi.org).
RESULTS
A total of 1,031 participants from the initial pool of 1,200 were included in the analysis for this study, following the exclusion of 169 individuals whose bedtimes fell outside the range of 9 PM to 1 AM or whose wake-up times occurred after 9 AM. Demographic characteristics of participants are presented in Table 1. Among the two groups (Study I conducted within the general population and Study II conducted among individuals reporting insomnia), the proportion of participants using hypnotics was significantly higher in Study II. Sleep onset time was significantly delayed, SOL was longer, and TIB was shorter in Study II compared to Study I participants. The DBST index and rating scales scores were significantly higher in Study II compared to Study I.
In Study I, the DBST index was significantly correlated with the ISI (r=0.17, p<0.01), DBAS-6 (r=0.18, p<0.001), and PHQ-2 (r=0.13, p<0.01) (Table 2). Q1 and Q2 were significantly correlated with the DBST index (r=0.15 and r=0.17, respectively; all p<0.001). In Study II, the DBST index showed similar correlation patterns with the rating scales. Q1 and Q2 were also significantly correlated with the DBST index. Among both Study I and II, Q3 and Q4 were not significantly correlated with the DBST index. DBSTq-2, an averaged score of the selected Q1 and Q2, was significantly correlated with the DBST index in Study I (r=0.19, p<0.01) and Study II (r=0.19, p<0.01) (Table 3). The DBSTq-2 score was significantly correlated with the ISI (r=0.46, p<0.01), PHQ-2 (r=0.33, p<0.01), and DBAS-6 (r=0.40, p<0.01) in Study I. Similar results were observed in Study II. Among all participants, the score density of the DBST index and DBSTq-2 is presented in Fig. 1, and correlations with the ISI are presented in Fig. 2.
Linear regression analysis was performed to explore whether the DBST index, ISI, PHQ-2, and DBAS-6 contribute to the DBSTq-2 in each study (Table 4). We observed that the DBST index (β=0.10, p=0.012), ISI (β=0.31, p<0.001), and DBAS-6 (β=0.22, p<0.001) contributed to the DBSTq-2 in Study I and in Study II (DBST index [β=0.11, p=0.004], ISI [β=0.43, p<0.001], and DBAS-6 [β=0.17, p<0.001]). Mediation analysis was performed to explore whether DBAS-6 and PHQ-2 mediated the association between the DBST index/DBSTq-2 and the ISI. The DBST index directly influenced the ISI, and DBAS-6 and PHQ-2 mediated the association among participants in Study I (Table 5 and Fig. 3A). DBSTq-2 also directly influenced insomnia severity, and both DBAS-6 and PHQ-2 mediated this association (Table 6 and Fig. 3B).
Receiver operating characteristic (ROC) analysis determined that the optimal cutoff point for the DBSTq-2 to identify participants reporting insomnia was ≥5.25, with an area under the curve (AUC) of 0.73 (95% confidence interval [CI], 0.70–0.76), specificity of 75.66% (95% CI, 71.25%–79.69%), sensitivity of 63.50% (95% CI, 59.55%–67.33%), and overall accuracy of 68.45% (95% CI, 65.51%–71.28%). ROC for the DBST index shows the optimal cutoff point to identify participants reporting insomnia was ≥1.10, with an AUC of 0.57 (95% CI, 0.53–0.60), specificity of 79.71% (95% CI, 75.5%–83.5%), sensitivity of 32.24% (95% CI, 28.6%–36.11%), and overall accuracy of 51.55% (95% CI, 48.45%–54.65%).
DISCUSSION
In this study, the two-item brief questionnaire version of the DBSTq-2 effectively represents the concept of the DBST index while minimizing its limitations. The DBSTq-2 demonstrated significant correlations with the ISI, DBAS-6, and PHQ-2 among the general population and individuals reporting insomnia. Furthermore, among all participants, the DBST and DBSTq-2 directly influenced insomnia severity, with DBAS-6 and PHQ-2 mediating these associations.
Q1 (“I desperately need at least five hours of sleep”) and Q2 (“I want to fall asleep early in the evening and sleep well into the late morning”) were selected as items of the DBSTq-2 in this study. The DBSTq-2 can be rated based on a 10-point Likert scale from 0 (strongly disagree) to 10 (strongly agree). When an individual strongly desires even a minimal amount of sound sleep, they are likely to assign a higher rating to Q1. Moreover, if an individual unconsciously desires extended sleep duration, they are also likely to assign higher ratings to Q2. When assessing the participant’s DBST index, the discrepancy range is calculated by subtracting dTST from dTIB, which may result in a restricted range. However, by utilizing the DBSTq-2 and summing the scores of its items, this limitation is mitigated, resulting in a broader and more informative scoring range (Fig. 1). Furthermore, the correlation coefficient between the DBSTq-2 and the ISI was higher than that between the DBST index and the ISI (Fig. 2). The DBSTq-2 not only encapsulates the conceptual framework of the DBST index but also demonstrates comparable results when correlated with the ISI, DBAS-6, and PHQ-2, even in the results of linear regression analysis and mediation analysis. The Q3 (“I do not want to go to bed early in the evening”) and Q4 (“I do not want to wake up late in the morning”) were designed with reverse scoring to create a questionnaire that facilitates fluent information for the questionnaire version of the DBST index. However, Q3 and Q4 were not correlated with the DBST index in this study; thus, they were not included in the questionnaire version.
The DBSTq-2 was significantly and inversely correlated with desired bedtime, but not with desired wake-up time, in Studies I and II. A positive correlation between desired wake-up time and DBSTq-2 had been hypothesized but was not observed. In study I, DBSTq-2 was positively correlated with dTIB but not with dTST. In Study II, however, the DBSTq-2 was positively associated with dTIB and inversely associated with dTST. The lack of a significant association between the DBSTq-2 and dTST may stem from the fact that participants in Study I were from the general population and exhibited relatively low insomnia severity compared to those in Study II. The finding that the DBSTq-2 demonstrates an inverse correlation with dTST and a positive correlation with dTIB supports the notion that the DBSTq-2 effectively encapsulates the core concept of the DBST index.
DBSTq-2 demonstrated a significant correlation with the PHQ-2 in Study II, whereas the original DBST index did not. This finding suggests that the DBSTq-2 may offer greater utility in exploring the concept of the DBST index among individuals with insomnia. This shows the rationale for developing a questionnaire version of the DBST index. There were no robust correlations between the DBST index and other measures, such as depression, anxiety, or sleep-related cognitions. A possible explanation may lie in the influence of circadian rhythm patterns. For example, the absence of an association between the DBST index and ISI among shift-working nurses suggests that this relationship may be diminished in individuals with circadian rhythm sleep–wake disorders. To mitigate this effect, a total of 1,031 participants were analyzed after excluding 169 individuals whose circadian rhythms might be delayed or advanced. Another potential explanation is the significant proportion of participants with a DBST index estimated as 0. Among 1,031 participants, 274 (26.6%) exhibited a DBST index value of 0. Third, the potential U-shape distribution of the DBST index can reduce its correlation with the ISI. Furthermore, the interpretation of negative DBST index values remains ambiguous. The DBSTq-2 exhibits stronger robustness in its correlation results with other measures in this study and can serve as a concise version of quick questions for physicians to assess patients’ sleep disturbances in clinical practice.
An optimal cutoff point for the DBSTq-2 to identify participants reporting insomnia was determined to be 5.25, with a specificity and sensitivity of 75.7% and 63.5%, respectively. The DBST index, on the other hand, has inherent limitation in defining a valid cutoff point. When ROC analysis was performed for the DBST index, a score of 1.1 was examined as an optimal cutoff point. However, the sensitivity, specificity, accuracy, and AUC were 32.24%, 79.71%, 51.55%, and 0.57, respectively. The findings indicate that the DBSTq-2 demonstrates greater efficacy in distinguishing individuals with insomnia compared to the DBST index.
The study has several limitations. First, the two surveys were conducted among the general population rather than among a clinical sample of individuals with insomnia. Although participants in Study II were enrolled based on self-reported insomnia symptoms within the past three months, it was not confirmed whether their symptoms reached a clinically significant threshold. Furthermore, the use of online anonymous surveys may introduce bias, as participants’ responses may be inaccurate or inconsistent. Therefore, the findings of this study should be validated in a clinically diagnosed insomnia population to ensure their robustness and applicability. Second, participants’ sleep–wake cycles were assessed based on subjective self-reports in an online survey rather than objective measures such as actigraphy or polysomnography. Furthermore, other sleep disorders such as restless leg syndrome or obstructive sleep apnea, which might influence participants’ sleep disturbance, could not be ruled out. Third, this study enrolled participants whose bedtime was between 9:00 PM to 1:00 AM and wake-up time before 9:00 AM. It is possible that circadian rhythm patterns influenced the DBSTq-2 results. Further studies are needed to confirm whether delayed or advanced sleep–wake cycles may affect the correlation between the DBSTq-2 and other psychologic measures. Fourth, the two studies were conducted independently as cross-sectional surveys. Consequently, the findings should be interpreted with caution, as the cross-sectional design limits the ability to establish causal relationships. Fifth, multiple significance tests were conducted in this study, which may elevate the risk of Type I error. This limitation should be acknowledged to ensure accurate interpretations.
In conclusion, a modified questionnaire version of the DBST index, the DBSTq-2, was developed to simultaneously enhance its applicability and address its limitations. The DBSTq-2 represents a more robust, concise, and clinically practical tool compared to the original DBST index, offering an efficient mean of evaluating insomnia severity. Its applicability across diverse groups and cultural contexts warrants further investigation.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation 
Print 
