Reliability and validity of the KIDSCREEN-52 health-related quality of life questionnaire in a Greek adolescent population

The study was conducted during the year 2003 in Greece within the framework of the European project 'Screening and Promotion for HRQoL in Children and Adolescents - A European Public Health Perspective' [4]. The sampling was random, multistaged and based on the age and sex distribution of school-age children living in the 54 geographical sectors of the country, according to data from the National Census of 2001. A sample of 1,900 adolescents (11 to 18 year olds) was recruited. Adolescents filled in the questionnaire at school. A total of 1,194 (that is, 63% response rate) of self-reported questionnaires (40.07% boys) were returned. Inclusion criteria for the adolescents were to be between 11 and 18 years old, to be able to read and complete the questionnaires themselves, and to consent to be involved in the study. Ethical approval was attained from the National Ministry of Education. Previous research on the representativeness of the present sample has reported that non-responder interviews showed no significant differences between responders and non-responders with regard to the general perceived health of adolescents and parents, the marital status of parents, highest educational level, and type of residence, indicating that a selection bias is unlikely [13].

The KIDSCREEN-52 questionnaire consists of 52 items and is based on a multidimensional HRQoL construct. The questionnaire was translated from English into Greek according to international guidelines using a backward-forward translation technique and international harmonization sessions [14, 15]. The instrument assesses either the frequency of behavior/feelings or intensity of an attitude using a five-point Likert scale (1 = never, 2 = seldom, 3 = sometimes, 4 = often, 5 = always) and the recall period is 1 week. The questionnaire is categorized into ten dimensions: physical well-being, psychological well-being, moods and emotions, self-perception, autonomy, parent relation and home life, peers and social support, school environment, social acceptance (bullying) and financial resources. The score for each dimension was transformed to a 0 to 100 point scale with higher scores indicate better HRQoL.

The SDQ questionnaire contains 25 items, categorized into 5 scales of 5 items each: hyperactivity/inattention, emotional symptoms, conduct problems, peer problems and prosocial behavior [12]. Responses to each of the 25 items consisted of 3 options: not true, somewhat true, or certainly true. The version for youths was used in the present study, which was found to have satisfactory psychometric properties in Greek adolescents [16].

To assess familial socioeconomic status, the Family Affluence Scale (FAS) was used [17], addressing issues of family car ownership, children having their own unshared room, the number of computers at home and time the adolescents spent on holiday in the past 12 months. The FAS was collected from adolescents in seven categories (from 0 the lowest, to 7 the highest FAS category) and was recoded into three groups in the analysis (low FAS level (0 to 3), intermediate (4 to 5) and high FAS level (6 to 7)). The psychometric properties of the FAS are acceptable and support its use as a self-reported measure for adolescents [18, 19]. Low FAS was expected to be associated with lower scores on KIDSCREEN-52 dimensions, but especially for financial resources.

To assess special chronic healthcare needs, the Children with Special Health Care Needs Screener (CSHCN) was included in the questionnaire, as a measure of physical and emotional chronic health status in children [20, 21]. The CSHCN contains five question sequences: each question is followed by two additional questions, asking about the presence and duration of any health conditions. The five questions address the use or need of prescription medication; the use or need of medical, mental health or educational services; functional limitations; use and need of specialized therapies (occupational therapy, physiotherapy, speech therapy, and so on); and treatment or counseling for emotional or developmental problems, all associated with a health problem that has lasted or is expected to last 12 months or longer. It was expected that adolescents with a chronic health condition would have lower scores of HRQoL, especially for the dimension of Physical Well-Being [3].

A confirmatory factor analysis (CFA) with maximum likelihood procedure was conducted in order to test how well the ten-dimension KIDSCREEN-52 model fits the data. The variance of the latent constructs was fixed at 1 during parameter estimation and the factors were allowed to be correlated. The fit of the CFA model was assessed using the χ² test, the comparative fit index (CFI), the goodness of fit index (GFI) and the root mean square error of approximation (RMSEA) [22]. For the CFI and GFI indices, values close to or greater than 0.95 are taken to reflect a good fit to the data [23]. RMSEA values of less than 0.05 indicate a good fit and values as high as 0.08 indicate a reasonable fit [23]. Also, a non-significant χ² statistic indicates a good fit, but χ² is usually sensitive to sample sizes and is usually significant for large sample sizes such as ours [22]. Scale internal consistency reliability was determined by calculation of the Cronbach α coefficient. Scales with reliabilities equal to or greater than 0.70 were considered acceptable. Validity was further examined with the correlation (Pearson's r) of KIDSCREEN-52 dimensions with SDQ scales. Correlation coefficients between 0.1 and 0.3 are considered low, between 0.31 and 0.5 moderate and over 0.5 high. Multivariate analysis of variance (MANOVA) was used in order to compare the scores of adolescents with chronic diseases with those of adolescents with no chronic diseases. Differences on KIDSCREEN-52 scales between the three socioeconomic status groups (low, middle and high) were also determined by the use of MANOVA. In order to control for multiple testing, Bonferroni correction was used. The P values reported are two-tailed. The statistical significance level was set at 0.05 and analysis was conducted using SPSS and AMOS (SPSS, Chicago, IL, USA) statistical software.

The total sample included 715 adolescent girls and 478 boys (N = 1,193). The total sample mean age was 13.6 (SD = 1.7). A total of 50.3% of participants belonged to the aged 11 to 14 group and 49.7% belonged to the aged 15 to 17 group. A total of 37.6% were identified in the low FAS group, 45% in the medium FAS group and 17.4% in the high FAS group. According to the CSHCN screener, 3.4% of the adolescents were suffering from a chronic health condition.

Cronbach α and mean scale scores are presented in Table 1. All the scales of KIDSCREEN-52, exceeded the minimum reliability standard of 0.70. Cronbach α values ranged from 0.73 (bullying) to 0.90 (moods and emotions, social support and peers).

As defined from the results of CFA the ten-dimensional model fitted the data well. The RMSEA, CFI and GFI values were 0.048, 0.971 and 0.965, respectively. None of the item cross loadings exceeded the item loadings on the intended latent construct. Estimates of the ten-dimensional model are shown in Table 2. Factor loadings were high and ranged from 0.61 to 0.88 indicating a strong association between the latent factors and their respective items (Table 2). The χ² test of the model was significant (χ² = 2,508.2, df = 1,299, P < 0.05).

The associations of KIDSCREEN-52 dimensions with SDQ scales are shown in Table 3. Correlation analysis showed moderate to high coefficients for the expected relationships. The SDQ emotional symptoms subscale, as well as the total difficulties score correlated highest with the KIDSCREEN-52 moods and emotions dimension. Moderate correlations of the SDQ emotional subscale were also observed with the KIDSCREEN-52 dimensions of self-perception, psychological well-being, parent relation and home life, physical well-being and peers and social support. The SDQ hyperactivity subscale correlated highest with the KIDSCREEN-52 dimensions of school environment, moods and emotions and parent relation and home life. The SDQ peer problems subscale correlated highest with the KIDSCREEN-52 dimensions of peers and social support, moods and emotions and social acceptance (bullying).

Scores of KIDSCREEN-52 dimensions by FAS categories are shown in Table 4. There were no significant differences in scores between middle and high FAS groups except for physical well-being and financial resources. All scales except for social acceptance (bullying) and autonomy differed across FAS groups and lower scores were found for those belonging to low FAS category compared to those belonging to middle or high FAS category. Scores on self-perception were significantly different only between low and high FAS categories.

Table 5 presents the mean scores of KIDSCREEN-52 dimensions for healthy adolescents and adolescents with chronic health condition. Healthy adolescents scored significantly higher on physical well-being, psychological well-being, school environment and social acceptance (bullying) dimensions.

The main strength of this study is its large and representative sample size. Also, the strength of our findings is that they are based on several goodness of fit criteria and associations of the measurement model.

However, there are several limitations to this study. Firstly, the identification of the group with chronic health problems was performed with a screening instrument, and no clinical information was available for physical and mental health problems. Thus, future studies should be provided in groups so that the severity of clinical conditions will be available.

Furthermore, because of the cross-sectional design of the study it was not possible to test the sensitivity of the KIDSCREEN-52 instrument to change. Changes over time should be evaluated with a longitudinal study design in future research. Pilot test in Spanish adolescents [26] has shown that KIDSCREEN follow up instrumentation seems adequate for collecting factors with potential influence on HRQoL.