2 Method
2.1 Participants
Baseline data from a longitudinal study examining predictors of changes in PQOL for a random sample of 439 youth with chronic conditions, ages 11–17 years, were used to assess psychometric properties of youth and parent reports of the SLSS and the BMSLSS. For each youth enrolled in the study, one parent (most often the mother) also participated. A prospective cohort design (Bijleveld et al.
1998) is being used for the PQOL study, with a 3-year follow-up of each participant.
This study focuses on youth with the kinds of physical, developmental, and communicative chronic conditions for which rehabilitation services are typically provided in children’s treatment centers. Youth were recruited from eight children’s treatment centers. Study youth had any one or more of the following chronic conditions: cerebral palsy, spina bifida, autism spectrum disorder, acquired brain injury, developmental delay, cleft lip and/or palate, Down syndrome, arthritis, communication disorder, amputation, or any other non-progressive muscular or central nervous system disorder. Youth with any progressive health condition, where life expectancy is not anticipated to extend beyond young adulthood (e.g., muscular dystrophy) were not included in the study. A significantly decreased life span may have a profound effect on youths’ PQOL.
To be eligible for the study, youth also needed to be cognitively able to independently understand and answer the study questionnaires with the guidance of a study interviewer (i.e., a trained health professional). Parents assessed youths’ cognitive ability prior to entrance into the study. The parent was provided with a series of sample questionnaire items and then made the determination whether their child would be able to respond accurately to them. Study interviewers also made a clinical judgement of the youths’ cognitive ability to participate in the study following the baseline interview. If a youth was deemed not able to participate, they were excluded from the study, and their data was not used.
2.2 Measures
Questionnaires measuring youths’ PQOL and the hypothesized predictors of changes in quality of life are administered to youth and their parents shortly at study entry and every 12 months over the course of 3 years. At the time of writing this paper, the baseline interviews have been completed, and investigators are continuing to collect data for the additional time points.
The youth and parent report versions of the SLSS contain seven context-free items each that are rated using a six-point scale: (6 = strongly agree; 5 = moderately agree; 4 = mildly agree; 3 = mildly disagree; 2 = moderately disagree; 1 = strongly agree). Five items are positively worded (e.g., my life is going well). Two items are negatively worded (e.g., I would like to change things in my life) and require reverse coding in the scoring.
The BMSLSS youth and parent report versions contain five items each that inquire about satisfaction with various life domains (i.e., family, friends, school, self, living environment). Items are rated using a seven-point scale: (7 = delighted; 6 = pleased; 5 = mostly satisfied; 4 = mixed; 3 = mostly dissatisfied; 2 = unhappy; 1 = terrible).
2.3 Procedures
Baseline data collection occurred either in a private office at the youth’s treatment centre or in the privacy of the youth’s home. The parent and youth made the choice of location. Interviewers obtained written informed consent in person from youth and parents just before conducting the baseline assessment. After a standardized introduction to the questionnaire, each youth took part in a face-to-face interview (30–60 min) with additional support as required from the study interviewer (e.g., further explanation of questions). The parent questionnaire (each 30–60 min) was self-completed and was done independently in a separate room at the same time as the youth interview. The youth and parent were not given access to each other’s responses (see McDougall et al.
2011 for greater detail about the original study).
2.4 Data Analyses
Descriptive analyses of the SLSS and BMSLSS youth report were conducted. Independent t tests or analyses of variance were also carried out to determine mean differences in SLSS and BMSLSS total scores for the youth report in relation to gender, age, and chronic condition groups. Cronbach’s alpha was computed to examine the internal consistency of the youth and parent report versions of the SLSS and the BMSLSS. Pearson correlation coefficients were estimated between SLSS scores and BMSLSS scores within the youth and parent report versions to assess convergent validity of the two measures. Exploratory principal components factor analyses with varimax rotation were used to examine the structure of the youth and parent report versions for both measures. Intraclass correlation coefficients (ICC) (two way random effects models) were estimated to evaluate youth-parent (inter-rater) agreement, and paired t tests were conducted to assess mean differences between the youth and parent reports on the SLSS and BMSLSS total scores and individual item scores. To adjust for multiple testing, p values were set at 0.01. SPSS Version 19 was used for all data analyses.
4 Discussion
This paper is the first to assess the psychometric properties of both the SLSS and the BMSLSS using a sample of youth with chronic conditions. Moreover, it is the first to compare the ratings of these youth report measures of PQOL with parent proxy versions.
This research provides evidence of the internal consistency and factor structure of the BMSLSS youth and parent reports when used with youth with chronic conditions. In addition, the one-factor solution found for the BMSLSS reflects the findings of other studies based on the general population (Funk et al.
2006; Seligson et al.
2005). Evidence of internal consistency was also indicated for the youth and parent reports of the SLSS. The two-factor structure of the SLSS youth and parent reports, however, differed from the one-factor structure that has been found consistently with normative samples of children and youth (Dew and Huebner
1994; Gilman and Huebner
1997). The first factor of the SLSS was comprised of five positively worded items, whereas the second factor included two negatively worded items.
Similar results have been found when assessing the factor structure of other measures of life satisfaction when used with youth with emotional difficulties or cognitive limitations (Griffin and Huebner
2000; Sawatzky et al.
2009). Research has indicated that some children have problems responding to negatively worded items (Marsh
1986). Huebner et al. (
2012) suggest that these findings may reflect cognitive-developmental limitations in youth. However, since both the youth and parent versions of the SLSS had a two-factor structure, one consisting of positive items and one of negative items, another possible explanation is that this phenomenon might be due to method effects (Marsh
1996).
That being said, since a one factor structure has been found consistently when the SLSS has been tested on normative youth samples, there might possibly be something different occurring with this sample of youth with chronic conditions. One speculation might be that the second factor that consists of one negatively worded question about desire for a different kind of life and another about desire to make changes in life may be reflective of a real construct capturing youths’ desire for life change related to their chronic conditions, as well as their parents’ awareness of that desire. The desire for life change may be much more salient for youth with chronic conditions than those without chronic conditions, especially during adolescence, as these youth become more aware of friendship issues, condition-related responsibilities, desire for autonomy from parents, attitudes of others, and barriers to school and community participation (Blum
1992; Groce
2004).
The Multidimensional Students’ Life Satisfaction Scale (MSLSS) (Huebner
1994), the long version of the BMSLSS has been tested on a sample of 844 middle school students when the negatively-worded items and other problematic items were removed, and acceptable reliability coefficients and fit across several indices have been found (Huebner et al.
2012). Likewise, in the current study, when the negatively worded items in the SLSS were removed, Cronbach’s alpha improved for both the youth and parent versions and a one-factor structure was indicated for both versions. When using the SLSS with youth with chronic conditions and their parents, one composite measure that consist of the five positively worded items could be used to capture the construct of overall life satisfaction, and a separate summed index could be created using the two negatively worded items to capture the construct of desire for life change. However, further testing of the SLSS on other samples of youth with chronic conditions is advisable. Confirmatory factor analysis would be a useful approach to use in future work to determine whether the two-factor solution of the SLSS found in this study is meaningful or an artifact (Marsh
1996).
This research provides initial evidence that the two PQOL measures are suitable to use with youth who have chronic conditions. Means scores for both the SLSS and the BMSLSS are in line with what has been found in the general population of similar aged youth (Funk et al.
2006; Gilman and Huebner
1997). The associations between youth life satisfaction reports using the SLSS and the BMSLSS and age and gender did not reach statistical significance These findings are also reflective of past studies using normative samples that have found demographic variables to have little relationship to PQOL (Dew and Huebner
1994; Funk et al.
2006; Gilman and Huebner
1997; Huebner and Alderman
1993).
Past studies have found that youth with chronic conditions often report their QOL to be similar to youth without chronic conditions (e.g., Majnemer et al.
2007), supporting the assertion that PQOL is representative of more than health status. Indeed, the associations between youth life satisfaction reports using the SLSS and the BMSLSS and chronic condition were not statistically significant. However, differences in SLSS reports between youth with an amputated limb and youth with autism spectrum disorder did approach statistical significance. It may therefore be of benefit for future research to look more closely at these associations using larger samples of each type of condition, and at the factors that might lead groups of children with physical conditions to report different levels of PQOL compared to youth with mental and behavioral conditions.
This study sample was composed of youth with the type of chronic conditions for which rehabilitation services are provided. This work provides an initial indication that, although the SLSS and the BMSLSS were originally created for use with children in the general population, they may also be used reliably and validly (with some modification to the SLSS) in an interview format with youth with cerebral palsy, acquired brain injury, and other chronic conditions, such as spina bifida or communication disorders given that youth have the cognitive ability to understand the questions and response options.
Findings indicated moderate inter-rater agreement between youth and parents for the total scores on the SLSS and the BMSLSS. At the same time, significant mean differences were found between youth and parent reports for the two measures, indicating that youth view their PQOL differently than their parents. This finding reflects other studies that youth with chronic conditions report higher scores with respect to their HQOL/QOL (Morrow et al.
2012; Upton et al.
2008; White-Koning et al.
2007).
Several possibilities may contribute to youth reporting a higher PQOL: (a) youth may have more accurate view of their life quality; (b) youth may be more satisfied than their parents realize; or (c) response shift bias (i.e., youth could be idealizing their life quality to the interviewer). Parents complete the study questionnaire independently, whereas the youth completes the questionnaire as part of an interview. Youth may be less likely to tell another person that they are dissatisfied with life. Efforts were made to minimize response shift bias in this study by training interviewers to conduct the interviews using a standardized protocol.
Regardless of differences between reports, each reflect the individual’s beliefs about the youth’s well being (Upton et al.
2008). Moreover, both the youth and the parent report can provide its own contribution to an understanding of a youths’ PQOL (Parsons et al.
1999). The SLSS and BMSLSS youth and parent versions appear to be measuring PQOL similarly enough to warrant use of the youth report alone when youth perspectives are of primary interest, parent report when youth reports are not possible (i.e., due to a youth’s cognitive limitations), or parallel use when nuances between parent and youth viewpoints are important to understand.
Ceiling effects were found for SLSS and BMSLSS in only 7 and 11 % of individuals, respectively, indicating that this is not an issue with these measures when used with youth with chronic conditions. Convergent validity of both the youth and parent versions of the SLSS and the BMSLSS was also indicated with this sample.