Quality of life (QoL) is an important public health issue for the policy development (Phillips, 2006). QoL refers to the perception of subjective health and well-being of an individual, which is a multidimensional concept covering several dimensions, including social relationships, physical and psychological health (The World Health Organization, 1995). Assessing the QoL more scientifically is an essential step to improve QoL of people's, and researchers have invested lots of effort into developing appropriate assessment tools for the general public or specific groups. For example, there are generic- and condition-specific instruments for evaluating QoL of children and adolescents. The former one is applicable to all population subgroups, whereas the latter one is useful to those with specific disability or illness (Fava et al., 2009; Davis et al., 2013; Bullinger et al., 2015). This study focused on the generic QoL instrument for children and adolescents.

Currently, the KIDSCREEN (The KIDSCREEN Group Europe, 2006), the Pediatric Quality of Life Inventory 4.0 (PedsQL 4.0) (Varni et al., 2006), the KINDL (Ravens-Sieberer and Bullinger, 1998), the Child Health Questionnaire (CHQ) (Raat et al., 2002), and Brief Multidimensional Students' Life Satisfaction Scale (BMSLSS) (Huebner, 1994) are well-known generic questionnaires that have been adopted to assess QoL in children and adolescents. These measures share some commonalities, but each one has its preference. The KIDSCREEN was designed to measure health-related quality of life (HRQoL), the PedsQL 4.0 included a wide definition of functioning, disability, and health (FDH), the KINDL, and the CHQ were appropriate to evaluate FDH with some HRQoL features, and the BMSLSS focused on life satisfaction (Seligson et al., 2003; Alamolhoda et al., 2021). To our knowledge, psychometric properties of the PedsQL 4.0, the KNIDL, the CHQ, and the BMSLSS have been validated in China (Ng et al., 2005; Lin et al., 2012, 2014; Ye et al., 2014), where has a population of 321 million children and adolescents under 20 years old (National Bureau of Statistics, 2011), whereas the KIDSCREEN has not been fully tested in this context.

KIDSCREEN instruments include self-report and proxy (parents) versions, and each of the two versions has three forms with 52, 27, and 10 items, respectively. The tools were initially and simultaneously developed in 13 European countries (The KIDSCREEN Group Europe, 2006). The KIDSCREEN-52 instrument with 52 items assesses 10 dimensions of HRQoL: physical well-being, psychological well-being, moods and emotions, self-perception, autonomy, parent relation and home life, financial resources, peers and social support, school environment, and bullying (Ravens-Sieberer et al., 2008; Zhu et al., 2019). The KIDSCREEN-27 is a short version of KIDSCREEN-52 using 27 items to measure five facets (i.e., physical well-being, psychological well-being, autonomy and parent relation, social support and peers, and school environment) merged from the 10 dimensions mentioned above (Ng et al., 2015). The KIDSCREEN-10 comprises 10 items derived from the 27-item version (The KIDSCREEN Group Europe, 2006). Evidence from prior studies indicated that KIDSCREEN-10 results in one global HRQoL score (Ravens-Sieberer et al., 2010; Haraldstad et al., 2011).

The KIDSCREEN instruments are deemed as cross-national HRQoL measures, and their psychometric properties have been studied in considerable research. Ravens-Sieberer et al. have conducted a cross-cultural survey in 13 European countries to assess the reliability and validity of KIDSCREEN indexes, and the result showed that all three versions of KIDSCREEN were reliable and valid (Robitail et al., 2007; Ravens-Sieberer et al., 2008, Ravens-Sieberer et al., 2010). Moreover, such supportive result for KIDSCREEN-52 has been found in investigations from China (Ng et al., 2015; Zhu et al., 2019), South Korea (Hong et al., 2007), South Africa (Taliep and Florence, 2012), Serbia (Stevanovic et al., 2013), Iran (Parizi et al., 2014), Japan (Nezu et al., 2015), Turkey (Baydur et al., 2016), and Colombia (Jaimes-Valencia et al., 2019, 52). In addition, studies from China (Ng et al., 2015), Serbia (Stevanovic et al., 2013), Turkey (Baydur et al., 2016), Norway (Andersen et al., 2016), Japan (Nezu et al., 2016), and Colombia (Vélez et al., 2016) have found a similar result for KIDSCREEN-27. Regarding KIDSCREEN-10, its validity and reliability can be supported in studies from Serbia (Stevanovic et al., 2013), Turkey (Baydur et al., 2016), Japan (Nezu et al., 2016), and Iran (Nik-Azin et al., 2014). Comparatively, a little evidence evaluated measurement properties of KIDSCREEN-10, especially no empirical endeavor has explored its reliability and validity in China. Given KIDSCREEN-10 was recommended for large epidemiological studies (The KIDSCREEN Group Europe, 2006), it is necessary to translate and test the applicability of KIDSCREEN-10 in more countries.

The current study aims to validate the cross-cultural adaption of the Mandarin Chinese self-report questionnaire of KIDSCREEN-10 using the Rasch model. Although the KIDSCREEN-52 and the KIDSCREEN-27 have been tested in the context of China (Ng et al., 2015; Zhu et al., 2019), we cannot infer that the KIDSCREEN-10 originally developed them also has good psychometric properties. For example, the KIDSCREEN-52 and the KIDSCREEN-27 are multidimensional scales, whereas the KIDSCREEN-10 is deemed as an unidimensional measure (Ravens-Sieberer et al., 2014). Therefore, based on the results of Zhu et al. (2019) and Ng et al. (2015), we still do not know whether the KIDSCREEN-10 is unidimensional. Moreover, the Rasch model is an approach exploring the performance of each item rather than the total test score, as in the classical test theory (CTT) (da Rocha et al., 2013). The Rasch model provides a detailed analysis of how items work within scales (Tennant et al., 2004), and thus it has many potential advantages over CTT methods in evaluating self-reported health outcomes (Hays et al., 2000). Currently, it has been increasingly applied in the psychology and health fields (Rocha et al., 2012; Ng et al., 2015; Vélez et al., 2016). Previous studies have demonstrated that this approach is appropriate, and actually more accurate to discover the psychometric properties of KIDSCREEN-10 (Erhart et al., 2009; Ravens-Sieberer et al., 2014). Therefore, the Rasch model is adopted to examine the properties of the KIDSCREEN-10 instrument for measuring the HRQoL among Chinese children and adolescents.

This was the first study to examine the psychometric properties of KIDSCREEN-10 in the context of Chinese society. First, results found that the internal consistency of the questionnaire was acceptable, which were consistent with prior studies on KIDSCREEN-10 (Erhart et al., 2009; Ravens-Sieberer et al., 2010; Stevanovic et al., 2013; Nezu et al., 2016). Second, known-group validity was verified. The known-group validity, also named as construct validity, reflects that a test can discriminate between two groups known to vary on the variables of interest (Langevin, 2009; Hendriks et al., 2017). In this study, the HRQoL level measured by the KIDSCREEN-10 questionnaire demonstrated significant differences among different socioeconomic statuses/health statuses/academic performance groups. Similar differences across socioeconomic status were also found in studies from Turkey (Baydur et al., 2016) and European countries (Erhart et al., 2009; Ravens-Sieberer et al., 2010). Moreover, KIDSCREEN-10 has validity in predicting HRQoL levels of children and adolescents.

Seven items (i.e., items 1, 2, 6, 7, 8, 9, and 10) exhibited ceiling effects. Although previous studies did not exhibit such effects, these studies examined the overall ceiling effect of the questionnaire rather than each item (Nik-Azin et al., 2014; Baydur et al., 2016). The ceiling effects indicate that most of the items were easy for the respondents, and thus the KIDSCREEN-10 cannot differentiate well the respondents with high degrees of HRQoL. However, these items do not demonstrate floor effects. This is meaningful from a public health and clinical perspective because it is more important to differentiate well between respondents with low-QoL as these are of risks for various health problems and represent the target population for interventions.

The original KIDSCREEN-10 questionnaire measured by a 5-point Likert scale (i.e., response categories include “never,” “seldom,” “sometimes,” “usually,” and “always”) demonstrated disordered response categories. It is because response options with similar meaning may make it difficult for respondents to distinguish the differences of options when answering, and thereby result in disordered categories (Zhong et al., 2014). After collapsing the categories of “seldom” and “sometimes,” the response categories ordered appropriately. Therefore, the KIDSCREEN-10 questionnaire was better to be measured by the four response categories in China.

Inconsistent with prior finding that all items of KIDSCREEN-10 in 15 European countries exhibited a good fit to the Rasch partial credit model (Erhart et al., 2009), this study found that item 3 (“Have you felt sad?”) and item 4 (“Have you felt lonely?”) demonstrated misfit to the model. No matter before or after merging the response categories, both infit and outfit MNSQ values of items 3 and 4 were higher than the acceptable cut-off value of 1.4 (Zhong et al., 2014). Items misfit indicated that the response of these items was inconsistent with the overall response pattern (Liu et al., 2016). This might indicate that Chinese school-aged children's cognition of mental health was different from other aspects (such as activities participation, peer relationships, school performance, etc.) measured by KIDSCREEN-10 since these two items measured the mental health of a child (i.e., depressive moods and emotions and stressful feelings) (Ravens-Sieberer et al., 2010). Meanwhile, this result might explain why the eigenvalue of the first contrast of the original KIDSCREEN-10 was higher than 2 thus violating the assumption of unidimensionality. Moreover, after removing these two items and reconducting the Rasch analysis, we found item 5 (“Have enough time for self?”) was also a misfit item. Finally, the seven-item questionnaire assessed by a four-point Likert scale fitted the model well. It should be pointed out that, unlike deleting items 3 and 4 would delete a dimension (i.e., mental health) of the questionnaire, although item 5 was removed as well, the dimensions of the seven-item version did not decrease further. Because the seven-item questionnaire includes item 6, which reflects the same dimension (i.e., autonomy) as item 5 (Ravens-Sieberer et al., 2008, Ravens-Sieberer et al., 2014; Nezu et al., 2016).

Although the previous Rasch analyses showed the KIDSCREEN-10 was valid and reliable (Erhart et al., 2009; Ravens-Sieberer et al., 2014), they did not perform the DIF analysis. In the current study, both item 3 and item 4 exhibited DIF when comparing respondents aged from 8 to 11 years and 12 to 20 years. It can be implied that the use of the KIDSCREEN-10 in China was difficult because these items (items 3 and 4) cannot measure the HRQoL level of children and adolescents independently of age. Therefore, this study suggested that, to avoid the questionable conclusion, the DIF analysis should be conducted when the Rasch method is employed.

In summary, in contrast with previous studies that found the original KIDSCREEN-10 had acceptable validity and reliability in European countries by using the Rasch measurement model (Erhart et al., 2009; Ravens-Sieberer et al., 2014), our results showed that the psychometric features of the original KIDSCREEN-10 were deficiencies. This may be due to cultural differences. This study was conducted in the Chinese Mainland. China has a collectivist culture, which is different from the individualist culture of European countries. The psychometric properties of the seven-item version measured with four response categories (“never,” “seldom/sometimes,” “usually,” and “always”) performed better than the original KIDSCREEN-10, but it should be noted that the seven-item version does not contain items (i.e., items 3 and 4) on psychological well-being (Robitail et al., 2007), and thus it is not appropriate to measure the general HRQoL using the seven-item questionnaire. Accordingly, instead of advocating the seven-item version, this paper suggests that the psychological properties of the KIDSCREEN-10 should be further tested by using CTT. Previous studies demonstrated that KIDSCREEN-10 provided a CTT reliable and valuable assessment of general HRQoL (Erhart et al., 2009; Stevanovic et al., 2013; Nik-Azin et al., 2014; Baydur et al., 2016; Nezu et al., 2016). Moreover, the KIDSCREEN-52, the parent version of the KIDSCREEN-10, has been validated by CTT in China (Ng et al., 2015; Zhu et al., 2019); thus, the KIDSCREEN-10 may not permit a Rasch-based measurement of general HRQoL, but fulfill the requirements of CTT.

This study has some limitations. First, the values of infits and outfits calculated by using the WINSTEPS software may lead to a high type I error rate because the software only calculates unconditional outfit and infit statistics, the results may become unreliable for sample sizes above 250 (Müller, 2020). Second, our sample, investigated from an autonomous county for ethnic minorities where the population of minorities was more than the national majority, may result in selection bias. Third, the concurrent validity of KIDSCREEN was examined by using the Pearson correlation coefficient. This method may under- or overestimate the predictive power, since it cannot adjust for confounders. Moreover, this study cannot provide the test–retest reliability due to the cross-sectional investigation.

The Mandarin version of the KIDSCREEN-10 did not perform good psychometrical properties in China by using the Rasch analysis. The KIDSCREEN-10 demonstrated disordered response categories, item misfit, unidimensionality, and DIF. After adjusting the response categories and removing the three misfit items, the seven-item questionnaire measured by the four response categories perform good measurement characteristics. However, the seven-item version was not appropriate to measure general HRQOL because it did not contain items on psychological well-being of HRQoL. Therefore, instead of advocating the seven-item version, this paper suggested that the psychological properties of the KIDSCREEN-10 should be further tested by using CTT.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional review board of Sichuan University (K2019067). Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

ZG and YL proposed and designed this study. ZH collected the data. ZG and ZH analyzed the data and wrote the first draft. YL and JX made a revision. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.