The aim of the international pilot study analyses was to reduce the item pool resulting from the previous project work which constituted the preliminary KIDSCREEN version for the pilot test
In the pilot questionnaire 159 KIDSCREEN items were included for children. Adolescents were asked 8 additional items. Furthermore, a general health item was applied. These items had been previously clustered into the following 7 dimensions which constituted the theoretical KIDSCREEN Quality of Life model: Physical (mobility, energy and relaxation, health and complaints), Self-Perception (body-image, self-assurance and self-esteem), Psychological Well-being (positive and negative emotions, worries and stress, life-satisfaction and optimism), Social Functioning (autonomy, opportunity to finance and to participate in activities), Social Relations (home life and parents, friends, group participation and social support), Cognitive and School Functioning (concentration and learning, achievement and teachers), Personal Environment (physical and home conditions).
The KIDSCREEN pilot questionnaire was completed by items covering socio-demographic information and health, selected on the base of international studies, recommendations and existing questionnaires. Adolescents also completed the Youth Quality of Life Instrument, YQOL (Patrick et al., 2002) and health care system utilisation questions. It was agreed to administer the sections of the questionnaire in the same sequence in all countries.
The pilot study was carried out in a convenience sample of schools with the exception of the Netherlands. Children and adolescents between 8 to 18 years of age were included in the study, as were their parents. The pilot study included respondents from different regions (urban vs. rural) as well as different socio-economic environments. In the Netherlands a representative sample was drawn from the database of the Regional Health Authorities in Leiden (GGD). Questionnaires were sent to a child and its parents asking them to fill in the questionnaire independently from each other.
Either the mother or father completed the questionnaire at home and returned it cost-free or through the schools to the respective centre. At each participating centre the child-parent set of questionnaires was matched by special codes.
The multinational KIDSCREEN pilot study sample used for the item reduction analysis included the data of 3977 children and adolescents. Between 48.1% and 89.3% of the contacted families (informed by an information sheet together with a written consent form) agreed to participate in the study. The sample comprised 1437 (36.1%) children between 8 and 11 years and 2469 (62.1%) adolescents aged between 12 and 18 years. The average age of all children and adolescents was 12.7 years with a standard deviation of 2.6 years. 1942 (48.8%) of the children and adolescents were female, 2019 (50.8%) were male. Distributions by age and gender were fairly good and comparable across countries. Despite the similar procedures of data collection in countries the response rates for children and adolescents were quite different and ranged between 48.1% (Germany) and 89.3% (Switzerland). The overall response rate was 56.7%. Related to the responding children and adolescents the response rates of parents are even more different and range between 23.6% (United Kingdom) and 99.4% (Netherlands).
For item reduction, the statistical analysis using the pilot study data was divided into 4 major steps:
- Item reduction with methods based on Item Response Theory (IRT);
- Item reduction with methods rooted in Classical Test Theory (CTT);
- Comparison of the results of both methods taking theoretical considerations into account. Thereafter, item reduction of the combined version using methods of Item Response Theory;
- Improvement of the scales’ predictive validity by examining item functioning across countries, age groups and gender.
This reduction process led to the construction of a multidimensional research version of the KIDSCREEN-52 questionnaire consisting of 52 items within 10 dimensions.