Study populations and data collection
General population sample
In 2002, by means of the SPSS random number generator, a random sample of 500 out of 9022 children aged 2 months-4 years in the general population of six municipalities allocated to the service area of ‘Carinova Salland’ (single regional provider of Well-Child Care for the 0–4 year olds) were mailed a questionnaire. The parents themselves decided if either the father or the mother should complete the questionnaire. Up to two reminders were sent; no incentives applied. After two weeks, the same questionnaire was mailed again to assess test-retest reliability in a random subgroup of 158 parents who had returned the first questionnaire, by applying random numbers generated by SPSS.
Respiratory illness sample
January 2000 to July 2001, at Erasmus University Medical Center Rotterdam and HAGA Hospital, The Hague, the Netherlands, patients were retrieved by diagnosis asthma or other disease of trachea/bronchus (ICD-9 coding system 493 and 519.1, respectively) or the reason for encounter ‘wheezing/cough’ as registered by the prospective problem oriented patient classification system [25]. Eligible patients were maximally 5 years old, visiting the pediatric outpatient or emergency department with recurrent lower respiratory complaints during at least 3 months within the past year and being treated with bronchodilators or corticosteroids as documented in the patient record [26]. Parents of all eligible patients were asked to participate (n = 230), and 217 agreed and the questionnaire was sent. After 10 days and 2 months, reminding letters were sent, the third reminder was by telephone. After 2 weeks, all parents who returned the questionnaire were mailed the same questionnaire again to assess test-retest reliability.
Infant and Toddler Quality of Life Questionnaire
The ITQOL consists of 103 items (10 multi-item scales and 2 single-item scales; see Table 1) that generally refer to the situation during the past 4 weeks. It was translated into Dutch according to international guidelines, including three independent forward and two backward translations [13, 27]. Per scale, the items that have 4, 5 or 6 response options, were summed up with equal weight per item (some recoded and/or recalibrated) and transformed into a 0 (worst possible score) to 100 (best possible score) scale [9, 10, 13, 28]. ITQOL-scales General behavior and Getting along, and Change in health are only relevant for children aged one year and older [9].
Table 1 ITQOL-scales, number of items per scale and score interpretationa
TNO-AZL Pre-school Children Quality of Life Questionnaire (TAPQOL)
The TAPQOL, which is in Dutch originally, consists of 43 items divided over 12 multi-item scales that cover physical, social, cognitive and emotional functioning domains (see Table 5) [24]. TAPQOL-scales Social functioning, Motor functioning and Communication are only relevant for children aged 1.5 years and older [24].
Table 5 Concurrent validity of the ITQOL assessed by Pearson-r correlation coefficients between ITQOL scale scores and scale scores of related (‘parallel’) vs. unrelated (‘non-parallel’) TAPQOL scales in the general population sample (n = 257–410) and in the clinical sample of children with respiratory illness (n = 106–138)a
Other data
In addition, the questionnaires consisted of items on standard socio-demographic variables, the presence of parent-reported current chronic conditions, and presence of wheezing and/or dyspnea and use of asthma medication during the preceding four weeks as defined in the ISAAC epidemiological measurement instrument [29, 30], and number of visits to the family physician during the past 12 months related to health problems of the child. Furthermore the questionnaire consisted of an item on the time needed to complete the ITQOL questionnaire and an item on the perceived difficulty of the ITQOL questionnaire.
Analysis
Only questionnaires concerning children, of whom at least one of the parents was born in a Dutch speaking country, were considered eligible for analysis; in other cases it is questionable whether the respondents had adequate mastery of the Dutch language (questionnaires were in Dutch).
Feasibility
We evaluated the response rates, ITQOL-questionnaire completion times, and perceived difficulty by the parents, and presence of missing and/or non-unique answers. We assessed mean scale scores and score distributions and presence of floor and ceiling effects (> 25% of the respondents have the minimal and/or maximal score). Additionally, mean scores per gender/age subgroup in the general population sample were evaluated.
Reliability
In both samples, overall and in gender/age subgroups, Cronbach's α was used to evaluate the internal consistency of scales; ≥0.70 is considered adequate [31]. We assessed whether (on average) Pearson-r correlation coefficients between the items and their own scale score (without the item under consideration) were higher than the correlation coefficients between these items and any other scale, to evaluate whether the ITQOL-multiitem scales represent separate domains; the average Pearson-r correlation coefficients were calculated by applying Fisher's z transformations [32]. Additionally, in both samples, we assessed scaling success in terms of the percentage of (corrected) item-total correlations with the own scale being higher than the corresponding item-other scale correlations (not including the single-item scale Change in health) [13]. In both samples, test-retest reliability of the ITQOL-scales was, at the individual level, assessed by test-retest Intraclass Correlation Coefficients (ICCs) [33]; ≥0.70 is considered adequate [34]. At the group level, test-retest reliability was assessed by two-sided paired-samples t tests, and by effect sizes: d = meant2 − meant1)/SDt1; 0:20 ≤ d<0.50 is considered small, 0:50 ≤ d<0.80 moderate, and d ≥ 0.80 large [35].
Concurrent validity
In both the general population sample and in the clinical sample, we evaluated whether specific ITQOL-scales correlated better with their assumed ‘parallel’ TAPQOL scales (see below) than with any other scale, as measured by Pearson-r correlation coefficients; scales are assumed to be ‘parallel’ if they pertain to domains that are considered identical. We hypothesized relatively high correlation coefficients between the following (‘parallel’) ITQOL-scale/TAPQOL-scale (in italics) pairs: Physical functioning-Motor functioning; Temperament/moods-Problem behavior/Positive mood/Anxiety; General behavior-Problem behavior; Getting along-Problem behavior/Social functioning.
Discriminative validity
In both the general population sample and in the clinical sample separately, we evaluated the ability of the ITQOL to discriminate between subgroups of children with no parent-reported chronic conditions (excluding asthma in the clinical sample) and subgroups with ≥2 parent-reported chronic conditions. Similarly, in the general population sample (respectively clinical sample), the ITQOL-scores in the subgroup with 0 (respectively ≤3) physician-visits during the past 12 months were compared with those in the subgroup with ≥4 (respectively ≥8) visits (Table 6).
Table 6 Discriminative ability of the ITQOL between subgroups differing in number of reported chronic conditions and physician visits within the general population sample and within the clinical sample; and between a subgroup of the clinical sample (with asthma) and a gender/age matched subgroup of the general population sample (no asthma)
Additionally, we compared the ITQOL-scores in a subgroup of the clinical sample (n = 94; only children of whom the parents confirmed the presence of asthma) with ITQOL-scores in a gender/age-matched subgroup of the general population sample (n = 188; only children of whom the parents denied the presence of asthma); each clinical subgroup-child was matched to two general population-children with the same gender/age (6 months classes).
If the ITQOL has adequate discriminative validity, we hypothesize that relatively low ITQOL-scores will occur in subgroups with relatively many conditions and/or visits. Differences were evaluated by independent-samples t tests and by effect sizes (d) that were defined as d = [Mean (no conditions) − Mean (with condition)_=SD in the conditions-subgroup [35].
All analyses were done in SPSS, Version 11.0. The Medical Ethical Review Board of Erasmus MC — University Medical Center Rotterdam approved this study.