The results of the present study demonstrate that the psychometric properties of the Brazilian version of CHAQ were appropriate as a whole for the evaluation of HRQOL in children and adolescents with CP, with possible limitations related to the presence of a significant ceiling effect.
The rate of missing data was low, as also observed for the healthy Brazilian population and for subjects with juvenile idiopathic arthritis , indicating good acceptability and effort efforts by the informants in filling out the questionnaires.
The low frequency of the floor effect in the domains suggests that the instrument is able to evaluate and to discriminate patients with smaller motor incapacity. The floor effect was greater for the arising, walking and gripping domains only for the patients with the hemiparetic form of the disease, and only for the gripping domain for the patients with the diparetic form, i.e., this occurred for the tasks executed with less difficulty by these children/adolescents. In the visual analogue scales the floor effect was significant in all the clinical forms of the disease, a fact that may limit the evaluation of patients with less impairment and a lower frequency of pain as perceived by the parent/guardian.
The ceiling effect found in all domains suggests the possibility of the instrument being insensitive to verify differences in HRQOL among the patients with greater motor incapacity. Nevertheless, the instrument was as effective in detecting differences in HRQOL between groups, as GMFM, the instrument used as an external criterion for the evaluation of physical function.
The predominance of the ceiling effect in the quadriplegia and extrapyramidal group was expected since these patients have more motor limitations and the instrument used in the present study covers very specific functional abilities. The great heterogeneity of the population studied hinders the elaboration of an appropriate questionnaire for the whole spectrum of possible motor manifestations in this disease. The evaluation of HRQOL should be complemented with more specific instruments for the patient with greater motor difficulties caused by CP [11, 14].
The variability of the scores obtained with the instruments of HRQOL is an indicator of good sensitivity in detecting changes in health conditions. Because this was a cross-sectional study, one of its limitations was the impossibility to test the sensitivity and responsiveness of the instrument. Prospective studies are necessary to evaluate this property and to verify the influence of the floor and ceiling effects on the sensitivity and responsiveness of CHAQ in children and adolescents with CP over time or after interventions. For a future longitudinal study the necessity to include the quadriplegic group should be verified, as CHAQ is an instrument that focuses on daily activities, and we do not expect to have a significant modification with the treatment program in this dimension for this group (we should consider the very high CHAQ scores, in all domains, with many ceiling effects to reinforce this idea). Others instruments with others dimensions could be more useful to evaluate the outcome of the quadriplegic group. But in this cross-sectional study we believe that it was important to evaluate all motor forms of cerebral palsy because it shows us that from the caregiver perspective these patients are very different in the domains measured by this instrument.
In general, CHAQ has been used to evaluate patients with juvenile idiopathic arthritis and musculoskeletal diseases, populations in which the percentage of individuals with lower motor incapacity is high, generating a considerable floor effect and an insignificant ceiling effect [17, 19]. Modifications in the options of answers have already been proposed by Lam et al.  for the evaluation of patients with musculoskeletal diseases in order to improve the sensitivity of the instrument and its ability to distinguish between patients with milder motor difficulties and the control groups. For the specific population with CP, changes could be made in the questionnaire in order to reduce the ceiling effect and to improve the differentiation of more seriously affected individuals.
In spite of these considerations, the results of the present study demonstrated that the instrument was capable of detecting differences among all the types of CP for the disability index and for the arising domain. Most of the domains detected more difficulties in the quadriplegia group compared to the diparetic and hemiparetic groups, although they did not differentiate the latter groups from one another, except for the arising and walking domains. Limitations were observed in the visual analogue scales which are more generic and subjective.
Reliability was found to be appropriate for all domains and the variations found in the correlation coefficient between the items and the domain itself did not suggest redundancy in the questions. The validity was also shown to be generally appropriate for the aspects tested.
In the evaluation of the face validity the instrument was considered appropriate for the study population on the basis of the perception of the informant. The face validity is the extent to which a measure "looks like" what it is intended to measure . In other words, to verify this validity it is necessary to ask the respondent, during completion of the measure, whether the items and scales look reasonable at "face value".
The category of "not applicable" answers was introduced in the original elaboration of CHAQ as an option for younger children, although each domain presents at least one question that can be answered by children under nine years. However, we believe that further information can be obtained when analyzing the proportion of "not applicable" items, because this type of answer suggests inadequacy of the question which is not due only to the influence of the age factor but also to the motor limitation of the patient. Therefore the proportion of questionnaires with "not applicable" items for each domain was analyzed and shown to be useful in the evaluation of face validity in the present study. If the parents/guardians say that the item is "not applicable" we need to think about the value of this question for these patients. The opportunity to have this option in the original version of CHAQ and to use it to access the face validity was very important. It was the first time that this option was used for this purpose in the instrument but future studies should not miss the opportunity offered by the instrument.
For the study population, the presence of "not applicable" questions was expected considering the age range evaluated and the motor limitation of the patient. Although this type of answer was frequent in the study population as a whole, the proportion of questionnaires with more than 20% of "non-applicable" items was low and the value was a little higher only in the activities domain. Since the frequency of "not applicable" items was low, when considering the questionnaire as a whole, the correlations of this type of answer with age, clinical type and physical function determined by GMFM were not significant. The values obtained demonstrate that CHAQ is adequate for the evaluation of the functional capacity of children and adolescents with CP as a whole, according to the perception of the parents/guardians.
In the evaluation of the discriminant validity of the items the success rate in the dressing and activities domains was below the ideal level. Since this is a specific instrument, different from multidimensional questionnaires, it is understood that some items may correlate with more than one domain. For the Brazilian population with juvenile idiopathic arthritis and for healthy controls, the discriminant validity of the items failed in the dressing, walking and reaching domains . These data may suggest the need to review some items and to rearrange them into more homogeneous domains according to the concepts involved, but this does not represent a limitation of the use of the instrument.
From the discriminant validity it was expected that the instrument could discriminate different constructs. Actually, the analysis showed that the visual analogue scales really evaluate concepts that differ from the domains and the disability index, with non-significant correlations between them. Moderate and significant correlations among the domains were expected because a specific instrument only involving the physical dimension in the evaluation of functional capacity was used. These concepts were again confirmed when correlating GMFM, the specific instrument for the evaluation of physical function, with the CHAQ domains which corresponded to appropriate convergent validity. The absence of correlation of GMFM with CHAQ scales confirmed the different natures of the measured constructs and demonstrated appropriate divergent validity.
Moreover, GMFM served as an external criterion to verify differences among the clinical types of CP. CHAQ proved to be capable of detecting these differences in all domains, but mainly for the disability index and for the arising domain. The visual analogue scales were not as useful as the GMFM in the evaluation of the clinical types of CP. This result was expected because GMFM was not considered an external criterion for these scales since they deal with different domains.
The hypothesis raised for construct validity was satisfied, because CHAQ proved to be useful to discriminate the performance of the healthy population and the patients with CP as a whole in all the domains and scales and the disability index.
The high but not perfect correlation between disability index and GMFCS levels in the present study indicates that CHAQ has a strong correlation with the gross motor function, but it is built to measure others aspects of the physical construct, as hypothesized.
It is essential to examine the measuring properties of the instruments used in the evaluation of health status or HRQOL for the interpretation of the results and for the best applicability of these instruments in clinical practice.
The present study should be interpreted by considering possible inherent methodological limitations. Although CHAQ can be answered by the patient, in this study only the information provided by the parent/guardian was considered. Most of the studies of this nature generally resort to a relative to obtain information. Few studies have obtained the perception of the patient with cerebral palsy and they did not involve representatives of the total population suffering from this disease [15, 30, 31]. When working with children with developmental disorders, frequently not only physical but various other levels of communication delay, cognitive deficit, learning disability make the presence of a representative essential [2, 32]. Due to these limitations, the presence of a representative of the child or of the patients with developmental disorders has the advantage of providing further information about the health conditions and well-being of the patients in addition to the perspective of the health team, even if that implies a potential risk of increasing subjectivity.
Future studies should be conducted to determine the possibility of applying CHAQ directly to the patients with CP, although patients with cognitive limitations should be excluded. The psychometric properties should also be analyzed again for each population group studied.
Others instruments more frequently used in patients with CP to measure the child's performance by parent report like the Pediatric Evaluation of Disability Inventory (PEDI) and the Functional Independence Measure for children (WeeFIM) include a self-care scale [5, 16] and they also show a high correlation with GMFM and GMFCS. The Pediatric Quality of Life Inventory (PedsQOL) – Cerebral Palsy Module, a HRQOL specific instrument, has adequate reliability and validity but only includes few questions about ADL . So, these instruments do not provide information about abilities for activities of daily living they are only available in English. CHAQ is a more specific instrument and it is available in at least 32 countries . It would be useful to apply it in association with a generic HRQOL instrument.