Introduction
Attention-deficit/hyperactivity disorder (ADHD) is a heterogeneous disorder associated with multiple neuropsychological deficits, such as executive deficits, delay-related difficulties (e.g., excessive temporal discounting, delay aversion), and deficits in temporal processing [
1‐
4]. These ADHD symptoms are dissociable, with substantial subgroups of patients affected in only one domain [
5,
6]. The preference for smaller, sooner (SS) over larger, later (LL) rewards, reward-delay impulsivity, is one of the strongest motivational markers in ADHD [
7‐
11]. This tendency is found across development, including in adulthood [
9,
12,
13]. Such a pattern of choice between rewards is related to a range of negative outcomes, including criminality, substance use, and greater functional impairment, such as in time and money management [
14] as well as in learning and study strategies [
15].
Sonuga-Barke suggested that this tendency to choose SS over LL rewards in ADHD is driven by two components: steeper temporal discounting of the future reward and acquired delay aversion (a negative emotional reaction to the imposition of delay) [
16]. Laboratory studies support a role for both exaggerated temporal discounting and delay aversion. In terms of delay aversion, ADHD individuals show excessive alerting to delay-related cues, which is similar to that of anxious children in the face of physical and social threats [
17]. They also display higher levels of delay-related frustration during long, boring tasks [
18,
19]. Neuroimaging studies have demonstrated delay aversion in ADHD, with increased involvement of emotional brain regions (i.e., the amygdala) in response to delay-related cues [
20,
21]. In terms of temporal discounting, many studies using real-time and hypothetical tasks have shown that individuals with ADHD discount the value of future rewards compared to controls [
9,
10]. The important roles of both ADHD components are supported by experimental studies that manipulated prereward delay and total delay independently. In a study by Marco et al. (2009), ADHD individuals exhibited similar preference for the SS reward and LL reward when trial length (i.e., total delay) was the same for both options and increased preference for the SS reward when that choice was linked to the reduction of delay by removing the postreward delay period [
8].
Measurements of temporal discounting and delay aversion typically involve laboratory tasks, most of which use paradigms that involve the choice between SS and LL rewards, specifically, the simple choice paradigm (SCP) and the temporal discounting paradigm (TDP). The delay parameters in these tasks may need to be extended to very high levels to assess performance in adulthood, given adult tolerance for long periods of delay. As a result, Clare et al. (2010) developed a short questionnaire, the Quick Delay Questionnaire (QDQ), to yield separate estimates of the two constructs in adults [
22]. The QDQ contains two subscales: delay aversion and delay discounting. The reliability and validity of the QDQ have been verified. In 2017, Thorell and colleagues developed a Swedish version of the scale and further verified the reliability and validity of this scale in a sample that included both patients with ADHD and clinical and nonclinical controls [
14]. In addition, the QDQ was found to be mainly related to hyperactivity and impulsivity symptoms rather than inattention symptoms. There was a significant association of QDQ scales with impairments in activities of daily living. The QDQ has been adapted into several other versions to assess delay-related difficulties in children with ADHD [
23,
24].
Some studies have also found that there may be differences in the level of reward-delay impulsivity in different ADHD subtypes and genders [
25,
26]. Scheres et al.‘s study reported that steep temporal reward discounting was observed in children and adolescents with ADHD-C but not in those with ADHD-I [
26]. Other studies have shown that girls with ADHD show a greater delay discounting than boys with ADHD [
27,
28]. A recent meta-analysis explored gender differences in the performance of two paradigms measuring the reward-delay impulsivity between normally developing healthy people and ADHD, and also showed that female ADHD were more inclined to choose SS rewards than male ADHD, showing greater reward-delay impulsivity [
25]. However, it is unclear whether there are also gender, subtype differences in reward-delay impulsivity levels in adults with ADHD.
At present, there are few studies on ADHD delay-related behaviours in China. Studies have mainly focused on children [
29‐
31]. The research questions of this study are whether QDQ is also applicable to the Chinese population, and what are the characteristics of the reward-delay impulsivity in Chinese adults with ADHD? Our research hypotheses are that QDQ adult self-assessment is well adapted to Chinese cultural background; adult ADHD may have abnormal reward-delay impulsivity, and ADHD subgroups may have different reward-delay impulsivity levels. The aim of this study was to develop a Chinese version of the QDQ (C-QDQ) and verify the reliability and validity of the C-QDQ in the Chinese population, and explore the ecological characteristics of reward-delay impulsivity in Chinese adults with ADHD. The novelty of this study is that it will be the first time to introduce QDQ scale in China to make up for the lack of self-assessment tools for adult reward-delay impulsivity in China. More specifically, we aimed to:
(1) Verify the reliability and validity of the C-QDQ (i.e., internal consistency and test-retest reliability as well as content validity, construct validity, and convergent validity).
(2) Explore the differences of reward-delay impulsivity between ADHD and healthy controls, as well as ADHD groups with different sex and subtypes.
Discussion
The presented study translated and examined the validity and reliability of the C-QDQ, a rating tool used to measure adult delay-related behaviours. The data were collected from participants with ADHD and healthy controls. The comprehensive evaluation showed that this scale has satisfactory reliability and validity.
In terms of internal consistency, Cronbach’s α coefficient between 0.70 and 0.90 is ideal, with exceeding lower bound meaning a low reliability, and exceeding higher bound meaning too many similar items [
51,
52]. In terms of test-retest reliability, a coefficient of 0.75 indicates sufficient retest reliability [
53]. In Clare et al.‘s study, the test-retest coefficients of the total QDQ and the two subscale scores were all above 0.8, and the internal consistency was satisfactory [
22]. In Thorell and colleagues’ study, the internal consistency of the QDQ was found to be adequate for scores on the two QDQ subscales, whether for all participants or separately for the clinical and nonclinical sample (Cronbach’s αs ranging between 0.71–0.83) [
14].In the present study, both the internal consistency coefficient and test-retest coefficient of the C-QDQ and the two subscales were good. It is worth noting that in Clare et al.‘s study, the sample included was the general adult population recruited by the network platform, and no individuals were excluded from the sample for any reason, their participants with an average age of 23.81 years old, range 18–77 years old and 86% women [
22]. In Thorell and colleagues’ study, the sample included not only patients diagnosed with ADHD, health controls, but also patients diagnosed with clinical disorders other than ADHD [
14]. In our study, the sample mainly included ADHD and health controls, the median age of the total sample was 25 years, and 46.7% of participants were male. However, all studies recruited adults and calculated internal consistency including all participants, and the results were good, indicating that QDQ always has good reliability in the adult population.
For the content validity of the scale, it was evaluated by six mental illness experts, and it showed good content validity at both the scale level and item level. This finding is in line with the expectation of using this scale to measure delay-related behaviours in this study.
The CFA verified the two-factor structure of the original scale, and the results showed that the factor structure model fit the C-QDQ data well. Therefore, the scale maintained a good scale structure even in a different language. But it’s worth noting, the two C-QDQ subscales respectively evaluate different aspects of delay reward, while this study found a high correlation between the two subscales. The reason may be that delay aversion reflects the negative experience of waiting [
16], while delay discounting represents the decision-making attitude towards future reward, which themselves include waiting. Admittedly, this may also indicate that the discriminant validity of the C-QDQ is inadequate. In addition, we improved the model fit according to the modification indices based on the correlation between items. Although the correlation between items is not difficult to understand from the perspective of item content, but it may also indicate inadequate discriminant validity of the scale.
In terms of the convergent validity of the C-QDQ, according to previous research, individuals with high impulsivity are more inclined to choose SS rewards [
54]. We believe that both delay aversion and delay discounting are related to the BIS-11 total score, which reflects the overall impulsivity characteristics of individuals [
38]. As expected, both C-QDQ subscale scores were significantly positively correlated with the BIS-11 total score, indicating that people with higher trait impulsivity are more likely to show more prominent delay-related difficulties, including a higher delay discounting rate and greater delay aversion.
In terms of the correlation between the C-QDQ scores and two laboratory measures of delay-related difficulties, the present study found that the TCIP, which is regarded as a tool to measure delay aversion, was correlated not only with the delay aversion score but also with the delay discounting score. As a tool to measure delay discounting, the DDT was not significantly related to the delay discounting score and was not significantly related to the delay aversion score. The C-QDQ appeared to be somewhat contradictory in its correlation with the two laboratory tasks, but this may be due to structural differences in the measurement of different neuropsychological paradigms [
55]. First, the TCIP repeats the same choice, which may lead to greater boredom than the DDT task. Second, there is a true delay in the TCIP, as participants can only proceed to the next option after the delay period is over. According to the delay aversion hypothesis, individuals with ADHD will experience more negative emotions when they have to wait, prompting them to choose SS rewards [
56]. Previous studies have found a more pronounced preference for SS rewards related to actual delay experience and longer delay [
29,
30]. In this study, the proportion of SS choices in the TCIP was calculated to reflect participants’ negative emotions in the face of delay. However, it is worth noting that the participants may choose SS rewards for reasons other than emotions. For example, individuals may choose SS rewards because they are not interested in the experiment or because they are eager to end the experiment for personal reasons. TCIP performance is likely influenced by factors other than the delay-related negative emotions, while the delay aversion subscale measures only the delay-related negative emotions, which made the conclusion of this study more conservative. Therefore, there was a significant correlation between the delay aversion score and TCIP performance, which supported the role of delay-related negative emotions on impulsive choice.
TCIP performance was also associated with the delay discounting score, as the TCIP may essentially be a delay discounting task, even though it is usually thought to reflect delay aversion. In fact, a fixed delay discounting rate (k) is set in advance in the TCIP. If the participant has a 50% choice rate for SS rewards, it means that the participant has a discounting rate of k; if the percentage of SS rewards is greater than 50%, the discounting rate is higher than k. Delay discounting is the depreciation of the value of the future delayed reward, and in response to the choice, the participant will choose more SS rewards, which explains why the TCIP was also related to the delay discounting score. Therefore, the correlations between the C-QDQ subscales and TCIP may prove the ideal convergent validity not only of the delay aversion subscale, but also of the delay discounting subscale.
However, DDT performance was not related to C-QDQ subscale scores, which may be partly because the DDT does not include an actual waiting period, and delay aversion may not be as obvious if the delay is only imagined. Imagined delay aversion and discounting of the delayed reward may be more easily compensated by the adult’s past experience of successfully enduring the delay and receiving the rewards. In addition, previous studies comparing laboratory and self-assessment measures of executive function, which are based on independent but related mental structures [
57‐
60], found that the C-QDQ score is associated with TCIP performance but not with DDT performance, which may also indicate that the structure of the C-QDQ overlaps more with that of the TCIP. Although the current results cannot indicate superiority or inferiority of the self-rating scale versus the laboratory task, the combination of the two in future studies may address concerns that the delay-related difficulties are not fully captured using the laboratory task alone [
14,
61].
By comparing the difference of C-QDQ scores between adult ADHD patients and healthy controls, we explored the ecological characteristics of reward-delay impulsivity of adults with ADHD: The level of reward-delay impulsivity in ADHD group and all subgroups was higher than healthy controls. The level of delay aversion in ADHD-C group was higher than ADHD-I group. The level of delay discounting in female ADHD was higher than male ADHD. Specifically, adults with ADHD scored significantly higher on both two QDQ subscales than healthy controls, excluding possible effects of gender, age, and the level of education.
This study found that there was an interactive effect between diagnosis and gender. Only in the ADHD group, gender has a significant impact on the C-QDQ total score and delay discounting subscale score. The delay discounting subscale score of female ADHD was significantly higher than male ADHD, which was consistent with the results of most previous studies [
25,
27,
28]. Patros et al.’s study found that girls and boys with ADHD have different patterns of cognitive control and delay discounting, girls with ADHD (but not boys) show an increase in delay discounting in real delay discounting tasks, they were more likely to choose immediate small rewards and showed greater reward-delay impulsivity [
28]. However, previous evidence for gender differences in delay discounting levels comes mainly from children, so the current study adds to the evidence for gender differences in delay discounting in adults with ADHD. This study did not find gender differences in the level of delay aversion within the ADHD group, suggesting that both male and female ADHD experience the same level of negative emotions when faced with delay. There have been no previous studies on this, it needs to be further verified in future studies.
In addition, the study also analysed the differences in C-QDQ scores between different subgroups of ADHD patients and healthy controls. The study found that compared with the ADHD-I group, the ADHD-C group had more obvious negative emotions towards delay, but the degree of discounting for future rewards was basically the same as the ADHD-I group. However, both two ADHD subgroups were more averse to delay than the healthy controls, and the degree of discounting for future rewards was greater. This is different from the results of Scheres et al., who showed that compared with ADHD-I and healthy children and adolescents, children and adolescents with ADHD-C showed greater delay discounting, and there was no inter-group difference in the subjective score of waiting difficulty [
62]. The results of this study may be different from those of Scheres et al. due to the following differences: (1) The subjects of Scheres et al. ‘s study were children but not adults; (2) Delay discounting was measured in their study using a delay discounting task, while the possibility that the scale is different from the paradigm has been discussed above; (3) The difficulty in waiting scale, which mainly evaluates the difficulty of experiencing a wait during a delay discounting task, may be different from the focus of delay-related negative emotions reflected in the C-QDQ Delay aversion subscale. However, because there are few studies on the differences in reward-delay impulsivity between different subtypes of adults with ADHD, further verification is needed in future studies.
In view of the large proportion of ADHD with comorbidity, we further compared the C-QDQ scores of the ADHD with or without comorbidity and the healthy controls to analyse the influence of comorbidity on the level of reward-delay impulsivity. Compared with ADHD without comorbidity, ADHD with comorbidity mainly showed greater discounting of future rewards but not more negative emotions about delay. In fact, except for adults with ADHD, some studies have shown that people with schizophrenia and those with depression or more severe depressive symptoms also exhibit higher rates of delay discounting than healthy controls [
63‐
66]. Most of the participants with ADHD in this study had comorbid major depressive disorder, so the current result may mean that ADHD comorbid with other mental disorders that may cause greater delay discounting, which will aggravate the depreciation of future rewards. Based on the comparison of the differences in reward-delay impulsivity levels between adults with ADHD and healthy controls and between different subgroups, the results of this part generally support the clinical sensitivity of the C-QDQ.
In summary, compared with the laboratory tasks of measuring delay-related difficulties, the QDQ is more more convenient to conduct. Adult ADHD has been understudied, in part because of a lack of relevant research instruments. In this study, the C-QDQ was found to have good reliability and validity. It can distinguish adults with ADHD who exhibit severe delay-related difficulties from healthy controls. In addition, the present study confirmed the existence of reward-delay impulsivity in adults with ADHD in a relatively large sample, which highlights the importance of devoting attention to delay-related behaviours of adults with ADHD. The C-QDQ is expected to promote the research on delay-related behaviours in Chinese adults with ADHD.
Limitations and future directions
The present study has several limitations. First, ADHD is a neurodevelopmental disorder, and the symptoms of some adults with ADHD continue to old age [
13]. The young age of the samples in this study have affected the representativeness of the samples, as did high level of education and the absence of the ADHD-Hyperactivity/Impulsivity subtype. Second, due to the limitations of available tasks, the two tasks in this study may not have captured only the negative emotions related to delay but rather those mixed with a variety of other factors, it may lead to a decrease in the persuasiveness of the result of the convergent validity of the delay aversion subscale. In the future, a scale assessing impatience due to waiting can be added, similar to the setting of Scheres et al. [
62]. Mood changes due to delays before and after the experiment were collected separately to supplement the laboratory task and further explore the convergent validity of the delay aversion subscale. Third, due to the relatively small sample size of healthy controls, this study included all samples in the calculation of construct validity, which still may affect the use of the scale in specific populations. In the future, the sample size will be expanded to cover adults of a variety of ages and verify the reliability and validity of the scale in healthy controls and ADHD populations separately, making the research results more representative and persuasive.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.