|Home | About | Journals | Submit | Contact Us | Français|
We report the development and initial validation of patient–derived, health related quality of life (HRQOL) questionnaires for intermittent exotropia (IXT).
In a development phase, 27 children (aged 2–17 years) with IXT and one of their parents. In an initial validation phase, 33 children with IXT and 49 control children (aged 5–17 years) along with one parent for each child. Children in the control group had no strabismus or amblyopia.
Individual patient interviews generated 35 items for Child and Proxy (parental assessment of child’s HRQOL) questionnaires and 46 items for a Parent questionnaire. To reduce to a feasible number of items, questionnaires were administered to 5–17 year old children with IXT (n=15) and parents of 2–17 year old children with IXT (n=27). Responses were analyzed using standard item reduction methodology. Three final derived IXT questionnaires (IXTQ): Child, Proxy, and Parent (12, 12, and 17 items respectively) were administered to children with IXT and control children, and to parents of IXT and control children. Likert-type scales ranging from ‘never’ (score 100, best HRQOL) to ‘almost always’ (score 0, worst HRQOL) were used for responses.
Median scores for IXT and control groups, compared using Wilcoxon tests.
Median Child scores were significantly lower (worse HRQOL) in the IXT group compared with the control group: 85 (quartiles 73–92) versus 92 (79–96); P=0.04. Median Proxy IXTQ scores were significantly lower for IXT children than controls: 83 (75–94) versus 98 (92–100); P<0.0001. Median Parent IXTQ scores were also significantly lower in the IXT group compared with the control group: 68 (quartiles 56–79) versus 93 (87–99); P<0.0001.
We have developed and validated a new 3-part patient-derived HRQOL questionnaire for children with IXT and their parents, comprising Child, Proxy and Parent questionnaires. These questionnaires detect reduced HRQOL in children with IXT as reported by the child themselves and as perceived by their parents (proxy report). Childhood IXT also appears to affect the HRQOL of the parents. The IXTQ HRQOL questionnaires may prove useful in the clinical assessment of IXT and for clinical trials.
Evaluation of severity of childhood intermittent exotropia (IXT) relies principally on assessment of angle of exodeviation, ability to control, and stereoacuity. Nevertheless, recent studies have demonstrated a great deal of variability in measures of control and stereoacuity in IXT, even from moment to moment.1, 2 New approaches to assessing severity of IXT, such as measuring effects on health related quality of life (HRQOL) may therefore be helpful in determining indications for intervention. In a previous report3 we described the identification of specific HRQOL concerns for children with IXT and their parents. We now describe the use of these patient-derived concerns to develop condition-specific HRQOL questionnaires for children with IXT and their parents, and the initial validation of those questionnaires, by comparing scores in children with and without IXT.
Institutional Review Board approval was obtained. Each parent gave informed consent and children 8 years and older gave informed assent, before participating. All procedures and data collection were conducted in a manner compliant with the Health Insurance Portability and Accountability Act.
We report two phases comprising this present study: first the development of final IXT questionnaires and second the initial validation of the newly developed final IXT questionnaires. For both phases of the study, we enrolled children with IXT along with one of their parents. All children with IXT had either divergence excess or basic type IXT: no child had a near angle ≥10 prism diopters (pd) greater than distance (convergence insufficiency type exotropia). None of the children with IXT had amblyopia, defined as ≥ 0.2 LogMAR (Logarithm of the Minimum Angle of Resolution) inter-ocular difference and ≥0.3 LogMAR in one eye. In the second, testing phase of the study, three children in the control group had an interocular acuity difference of 0.2 LogMAR, attributable to refractive error. Patients with neurodevelopmental delay or coexisting ocular pathology were not included.
Questionnaire items were identified from individual interviews of children with IXT and one of their parents, as reported previously.3 Unique statements or phrases regarding the effects of IXT on HRQOL were converted into questions. Statements or phrases considered likely to discriminate between patients based on socio-economic, cultural, or educational status or that did not explicitly address HRQOL (but were more descriptive of symptoms) were not included. Three types of questionnaires were derived: Child (35 items addressing child’s own HRQOL), Proxy (35 items parallel to the Child questionnaire addressing parent’s perception of their child’s HRQOL) and Parent (46 items addressing the parent’s own HRQOL). Child and Proxy questionnaires contained identical items pertaining to the child’s HRQOL expressed in the first person for the Child and third person for the Proxy.4 These questionnaires were administered to a cohort of children with IXT and their parents to reduce the number of items in each questionnaire to a more feasible number addressing concerns.
Twenty-seven children with IXT (aged 2–17, median 5 years) were recruited with one of their parents. Two (7%) of 27 had undergone previous surgery but had recurrent IXT. Median angle of deviation at distance by prism and alternate cover test (PACT) was 20 pd (range 10 to 30 pd). Four (15%) children wore refractive correction. Fifty-two percent were female and for 93% race was self-reported as ‘White.’ Fifteen (56%) were aged 5 to 17 years (n = 6 aged 5 to 7 years and n = 9 aged 8 to 17 years) and completed the Child questionnaires, which were designed to be completed by 5- to 17-year-old children. Proxy questionnaires were completed by parents of the 5- to 17 year children (i.e. 15 of 27 parents). In 13 (87%) of 15 cases, the mother of the child completed the Proxy questionnaire. Parent questionnaires were completed by all 27 parents (i.e. parents of 2- to 17-year-old children). In 19 (70%) of 27 cases, the mother of the child completed the Parent questionnaire. For 2 (7%) of 27 children, both the father and mother of the child circled responses on the questionnaires: where there were differing responses, one parent was selected at random.
We developed the Child questionnaire in two formats, similar to the design of the generic PedsQL™ questionnaires.5 The format for 5- to 7-year-olds used simple language and a 3-point Likert type scale for responses (‘Not at all’, ‘Sometimes’, ‘A lot’) as well as the option to respond ‘I don’t know.’ A matching card with face symbols representing the responses could also be used, depending on the child. Questionnaires were read, verbatim, by a trained observer, in a neutral tone of voice. If the child did not respond, the question was repeated verbatim without any explanation or elaboration. Inability to respond despite repeating the question was recorded as ‘I don’t know.’ The format for 8- to 17-year-olds used a 5 point Likert type scale for responses (‘Never’, Almost never’, ‘Sometimes’, ‘Often’, ‘Almost always’) as well as the option to respond ‘I don’t know.’ Questionnaires were self-administered following simple verbal and written instructions, with supervision as necessary (depending on the child) by a trained observer. If the child did not understand a question, it was read verbatim by a trained observer, without any explanation or elaboration. Inability to respond despite repeating the question was recorded as ‘I don’t know.’ Parents were instructed not to interfere with their child’s responses and communication (verbal and non-verbal) between child and parent was limited by positioning the child with their back to their parent whenever possible. Any comments made by the child regarding the wording or clarity of a question were recorded.
For Proxy and Parent questionnaires a 5-point Likert type scale was used for responses (‘Never’, Almost never’, ‘Sometimes’, ‘Often’, ‘Almost always’) as well as the option to rate the question ‘Not applicable.’ Questionnaires were self-administered, following simple verbal and written instructions including an explanation to use the ‘Not applicable’ option only if the item could not be understood or if it was considered to be not at all relevant. Any comments regarding the wording or clarity of a question were recorded.
Child and Proxy responses were analyzed together to enable retention of items that were pertinent from both child and parent perspectives and the development of a proxy questionnaire that paralleled the child self-report.6 For 8- to 17-year-old Child questionnaires, and for Proxy questionnaires, the 5-point scale used for responses was collapsed to the 3 categories used for the Child 5- to 7-year-old questionnaire: ‘Never’ and ‘Almost never’ were converted to ‘Not at all’; ‘Often’ and ‘Almost always’ were converted to ‘A lot.’ Questionnaire items were excluded using the following four pre-defined criteria: 1) removed items where ≥ 20% of either Child or Proxy responses were ‘Not applicable’; 2) removed items where ≥80% of either Child or Proxy responses were ‘Not at all’; 3) removed items where there were any comments regarding the meaning of the item; and 4) removed items that were unlikely to be responsive to intervention or that would become irrelevant after treatment. To further reduce items in parallel for the child and proxy questionnaires, we plotted the correlation of the individual item scores with the total scores (item-total correlation using Cronbach’s alpha) against number of items retained in item-total correlation rank order. To determine if there were any systematic differences between the scores of younger and older children and between children and parents, two separate graphs were plotted showing 1) Child 5–7 and Child 8–17 data and 2) Child 5–17 and Proxy 5–17 data. The inflection point on the graph was used to determine the threshold for retaining items, i.e. points on the steep portion of the plot would indicate high item-total correlation, and items that should be retained. Common items with item-total correlation values greater than the inflection threshold on each graph were retained. Factor analysis was performed, replacing any remaining ‘Not applicable’ responses with a ‘Not at all’ value.
For Parent questionnaires, items were excluded using the same 4 pre-defined criteria described for Child and Proxy questionnaires described above. Once steps 1–4 were completed, any remaining ‘Not applicable’ responses were replaced with a ‘Never’ value for factor analysis. To further reduce the number of items, loading thresholds were analyzed (a measure of the correlation between an item and the underlying factor) and because we had a large number of items, we retained only items loading above a threshold of 0.70 on the strong factors.
As a first step in validating final IXT questionnaires (IXTQ), 5- to 17 year-old children were recruited to complete questionnaires, along with one parent for each child. Questionnaires were administered as described for the development phase with the 3-point Likert type scale used for 5-to 7-year olds (‘Not at all’ =score 100, ‘Sometimes’ =score 50, ‘A lot’ =score 0) and the 5-point Likert type scale for 8–17 year olds (‘Never’ =score 100, Almost never’ =score 75, ‘Sometimes’ =score 50, ‘Often’ =score 25, ‘Almost always’ =score 0). Proxy and Parent questionnaires were self-administered and the 5-point Likert type scale used for responses (‘Never’ =score 100, Almost never’ =score 75, ‘Sometimes’ =score 50, ‘Often’ =score 25, ‘Almost always’ =score 0).
Thirty-three children (median age 8 years, range 5–15 years) with IXT were recruited with one of their parents for this phase of the study. Fifteen (45%) of 33 were aged 5- to 7 years and 18 (55%) were aged 8- to 15 years. One had undergone previous surgery nearly 10 years prior and had recurrent IXT. Median angle of deviation at distance by prism and alternate cover test (PACT) was 25 (range 10 to 40 pd). Eleven (33%) children wore a habitual refractive correction. Twenty-two (67%) were female and for 28 (85%) race was self-reported as ‘White.’ All 33 children completed the 12-item final Child questionnaire. Thirty-one of 33 parents completed the 12-item Proxy questionnaire (two overlooked) and all 33 parents completed the 17-item Parent questionnaire. The same parent completed both Proxy and Parent questionnaires: in 27 (82%) of 33 cases, this was the mother of the child.
Forty-nine control children (median age 8 years, range 5–13 years) were recruited with one of their parents for this phase of the study. 19 (39%) of 49 were aged 5- to 7 years and 30 (61%) were aged 8- to 13 years. All children were orthotropic with no more than 10 pd of horizontal and no vertical heterophoria by PACT. Seventeen (35%) of 49 children wore a habitual refractive correction. Thirty-three (67%) were female and for 46 (94%) race was self-reported as ‘White.’ All 49 children completed the 12-item Child final questionnaire. All 49 Parents completed the 12-item Proxy questionnaire and 17-item Parent questionnaire. The same parent completed both Proxy and Parent questionnaires: in 37 (76%) of 49 cases, this was the mother of the child. IXT and control groups were therefore comparable for age, gender, race and which parent completed the questionnaires.
Median Child, Proxy and Parent questionnaire scores were calculated, ranging from 100 (best HRQOL) to 0 (worst HRQOL) and compared between IXT and control groups using Wilcoxon tests. Scores on individual questions and subscales were also compared between groups. For Parent questionnaires, subscale scores, also ranging from 100 to 0, were calculated and compared between IXT and control groups. Scores were calculated as an average of all answered questions. Where more than one answer was circled, the response indicating better HRQOL (higher score) was used. All statistical analyses were done using SAS statistical software version 9.1.3.
Of the 35 original questionnaire items, 10 were removed because ≥20% of either Child or Proxy responses were ‘not applicable’ and an additional 3 items were removed because ≥80% of either Child or Proxy responses were ‘Not at all.’ There were no negative comments regarding the clarity or wording of questions and no items were considered potentially unresponsive to, or inappropriate after treatment, therefore no items were excluded for these reasons. In total, 13 (37%) of 35 items were removed, leaving 22 items. A ‘Not at all’ response was imputed for the 58 remaining ‘Not applicable’ responses (24 child responses and 34 proxy responses of a total 660 responses). Using Cronbach’s alpha, the correlation of the item score to total score (item-total correlation) was plotted against the number of items retained in item-total correlation rank order. The item-total vs. correlation rank plot for the Child questionnaire in children ages 5 to 7 was compared to the plot in children ages 8 to 17 (Figure 1A available online only at http://aaojournal.org) to rule out any gross differences in item functioning between these two age ranges. No meaningful differences were found and children ages 5 to 17 were combined for comparison of the Child questionnaire to the Proxy questionnaire. The inflection point of the item-total vs. correlation rank plot for the Child (ages 5 to 17) and Proxy questionnaires was determined to be 0.3 (Figure 1B available online only at http://aaojournal.org), which was then used as a threshold to further reduce the number of questionnaire items. Items with item-total correlation > 0.3 were retained, yielding 16 items on Child questionnaires and 20 on Proxy questionnaires, with 12 items common to both Child (ages 5 to 7, 8 to 17, and 5 to 17) and Proxy questionnaires. Factor analysis was performed on these 12 remaining common items and review of the scree plots indicated 1 prominent factor, accounting for 4.8% of the variance for the Child questionnaire (Figure 2A available online only at http://aaojournal.org) and 8.8% of the variance for the Proxy questionnaire (Figure 2B available online only at http://aaojournal.org). These analyses therefore resulted in final Child and Proxy questionnaires containing 12 parallel items (Table 1; also available on line in administrable format with user instructions: http://public.pedig.jaeb.org, accessed 15th June 2009). Cronbach’s alpha reliability coefficient was 0.93 for the final Child questionnaire and 0.97 for the final Proxy questionnaire.
Of the 46 original questionnaire items, 1 item was removed because ≥20% of responses were ‘Not applicable,’ 1 was removed because ≥80% of responses were ‘Never,’ 2 items were removed due to negative comments regarding wording or clarity and 1 item was removed as potentially irrelevant after treatment. In total, 5 (11%) of 46 items were removed, leaving 41 items for factor analysis. A ‘Never’ response was imputed for the 38 remaining ‘Not applicable’ responses, of a total 1107 responses. Factor analysis showed 3 strong factors on the scree plot (Figure 2C available on line only at http://aaojournal.org), accounting for 16%, 4% and 4% of the overall variance respectively. Principal component analysis also revealed three similar principal components (data not shown). Three factors were forced for remaining analyses. Loading thresholds were reviewed and 0.70 was applied, reducing the total number of items to 17. Repeat factor analysis confirmed the 3-factor structure. Items within the 3 factors addressed parental worry for their child in 3 broad areas which we proposed as subscales: 1) psychosocial effects of IXT (n=7 items), 2) function (n=8 items) and 3) surgery for IXT (n=2 items). (Table 2; also available online in administrable format with user instructions: http://public.pedig.jaeb.org, accessed 15th June 2009). Cronbach’s alpha reliability coefficient for the final Parent questionnaire was 0.92 overall, 0.94 for the psychosocial subscale, 0.94 for the function subscale and 0.91 for the surgery subscale.
Median Child IXTQ scores were significantly lower (worse HRQOL) in the IXT group compared with the control group: 85 (quartiles 73–92, range 54–100) versus 92 (quartiles 79–96, range 50–100); P=0.04 (Figure 3). At the individual question level the greatest difference between IXT and control groups was: ‘Does it bother you that you have to shut one eye when it is sunny?’ (P=0.007; question 7, Table 1).
Median Proxy IXTQ scores were significantly lower for IXT children than controls: 83 (quartiles 75–94, range 54–100) versus 98 (quartiles 92–100, range 65–100); P<0.0001 (Figure 3). At the individual question level most questions showed a significant difference between IXT and control groups, particularly: ‘It bothers my child because he/she has to shut one eye when it is sunny’ (P<0.0001; question 7, Table 1).
Median Parent IXTQ scores were also significantly lower in the IXT group compared with the control group: 68 (quartiles 56–79, range 24–100) versus 93 (quartiles 87–99, range 49–100); P<0.0001 (Figure 3). On each of the 3 subscales, median Parent scores were significantly lower in the IXT group compared with the control group: Psychosocial subscale 79 (quartiles 57–89, range 11–100) versus 100 (quartiles 93–100, range 50–100); P <0.0001, Function subscale 66 (quartiles 50–78, range 34–100) versus 88 (quartiles 78–97, range 41–100); P <0.0001, Surgery subscale 50 (quartiles 50–75, range 25–100) versus 100 (quartiles 100–100, range 38–100); P <0.0001.
Using specific concerns identified directly from children with IXT and their parents, we have developed a 3-part questionnaire (Child, Proxy and Parent components) for assessing HRQOL in IXT. Children with IXT showed worse HRQOL (lower scores) than control children when measured using the Child IXTQ self report and when measured using the parental Proxy IXTQ report. Parents of children with IXT also showed lower scores when reporting their own HRQOL using the Parent IXTQ.
There are limited data on quality of life in children with IXT. McKeon et al,7 mailed two self-report questionnaires to children and adults (aged 8 to 46 years), and found both instruments differentiated between subjects with and without IXT. Nevertheless, the scales were not used in children less than 8 years of age, thereby excluding a large proportion of the population affected by IXT. The generic PedsQL instrument, allowing child self-report from 5 years of age and proxy reporting from 2 years of age, was administered by Powell et al to children with IXT and their parents (Quality of Life in Intermittent Exotropia. Powell CJ et al. Poster presentation, AAPOS 2007); they found mean scores were significantly lower (worse HRQOL) in their IXT cohort compared to published mean scores for normal populations. Nevertheless, generic instruments are typically less sensitive to change in a condition than condition specific instruments.8–10
Using the Child self-report IXTQ, we found children with IXT scored lower, indicating worse HRQOL, compared to a control group of children without strabismus and with normal visual acuity. Certain questions appeared to discriminate particularly well between children with IXT and children in the control group. Children with IXT were bothered by having to close one eye in sunlight, a symptom reported in a previous study, to affect up to 76% of children with IXT.11 Despite the well-recognized association between monocular eye closure and IXT,11–13 it is unclear whether troublesome monocular eye closure should be considered a criterion for intervention or whether it can be improved,13 although interestingly, some treatment studies stipulate resolution of monocular eye closure as necessary before defining IXT as ‘cured.’14 Prospective analysis of HRQOL using the IXTQ before and after treatment will allow evaluation of change in such effects of IXT.
We developed a 3rd IXT questionnaire to measure any effects of the child’s IXT on the HRQOL of the parents themselves. Previous studies have shown that mothers of chronically ill children are more likely to develop anxiety and depression15 and that strabismus in particular may affect the parent’s HRQOL:16 Akay et al found that mothers of children with strabismus demonstrated higher depression scores, were more nervous and distressed and had more problems with family functions than mothers in a control group.16 In our study, using a patient-derived, condition-specific questionnaire, parents of children with IXT also showed worse HRQOL than parents of children in a control group.
It is noteworthy that all items in the Parent IXTQ relate to different types of worry, with sub-scales assessing worry regarding the child’s everyday functioning, psychosocial interactions and possible surgery for IXT (Table 2). It is possible that improving parental education regarding the nature of IXT may reduce parental worry and the Parent IXTQ may be a useful tool for measuring the effectiveness of such interventions. We need further advances in our understanding of the natural history of IXT to better advise and reassure parents.
There are several limitations to this present study. Our study sample was racially homogeneous, despite recruiting consecutive patients, and this may have influenced the selection of final questionnaire items. We attempted to overcome this potential bias by excluding at the outset questions that may be discriminatory based on cultural, social, economic, or educational factors. We used what were considered to be the most appropriate methods for deriving final questionnaire items, but alternative item-reduction methods may have resulted in different final questionnaires. In addition, we did not have sufficient numbers to analyze data for any affect of age or severity of IXT on HRQOL during the validation phase and we have thus far only validated the questionnaires on children aged 5–17 years. Further validity testing of the IXTQ is planned.
We have developed and initially validated a new 3-part patient-derived HRQOL questionnaire for children with IXT and their parents. The IXTQ detects reduced HRQOL in children with IXT both as reported by the child themselves and as perceived by their parents. Childhood IXT also affects the HRQOL of the parents by causing worry about various aspects of the condition and its management. The IXTQ HRQOL questionnaires may prove useful in the clinical assessment of IXT and for clinical trials.
Financial support: Supported by National Institutes of Health Grants EY015799 (JMH), EY018810 (JMH), EY013844 (EAB), Research to Prevent Blindness, New York, NY (JMH as Olga Keith Weiss Scholar and an unrestricted grant to the Department of Ophthalmology, Mayo Clinic), and Mayo Foundation, Rochester, MN.
Meeting Presentation: This work was presented in part at the Annual Meeting of the Association for Vision and Research in Ophthalmology, held in Fort Lauderdale, FL May 3-7, 2009.
None of the sponsors or funding organizations had a role in the design or conduct of this research. No authors have any financial/conflicting interests to disclose.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.