Pediatric Voice Outcome Survey.
Hartnick CJ, Volk M, Cunningham M. Establishing Normative Voice-Related Quality of Life Scores Within the Pediatric Otolaryngology Population. Arch Otolaryngol Head Neck Surg. 2003;129(10):1090-1093. doi:10.1001/archotol.129.10.1090
To establish normative values for voice-related quality of life across a broad pediatric otolaryngology population using the Pediatric Voice Outcome Survey (PVOS).
Multiphysician outpatient pediatric otolaryngology practice.
The PVOS was completed by 385 parents of children and adolescents aged 2 to 18 years. Of the 385 parents, 75 were readministered the instrument 2 weeks after no intervention had been provided. Data were collected regarding the patients' age, main diagnosis, and operative intervention.
The PVOS demonstrated robust internal consistency with an overall Cronbach α value of .70. Test-retest reliability demonstrated a weighted κ value of 0.89 (95% confidence interval, 0.84-0.95) The mean ± SD converted score (based on a 0-100 scale) for the overall population was 80.5 ± 19.9. Converted PVOS scores are provided for each subpopulation according to main diagnosis. The PVOS scores varied significantly according to age (P<.05) and preoperative and postoperative status following adenoidectomy (P<.05).
The PVOS represents a valid and reliable instrument to measure voice-related quality of life in a broad pediatric otolaryngology population.
THE FIELD of pediatric laryngology is in its nascency; novel strategies are being sought to assess pediatric vocal disorders as well as to measure the various treatment strategies that are presently available and that are currently being developed. With regard to diagnostic strategies, evaluation of a child's voice requires dynamic evaluation of the anatomy coupled with acoustic and aerodynamic measures. The functional impact of the child's vocal disorder as it relates to his or her quality of life is also important to determine. Health-related quality of life (HRQOL) can be defined in this context as "the subjective and objective impact of dysfunction associated with illness or injury, medical treatment, and health care policy."1 Voice-related quality of life (VRQOL) measures a specific domain or aspect of HRQOL as it pertains directly to how one's impression of his or her voice informs the broader concept of HRQOL.
Whereas meaningful data have been previously published with regard to strategies of obtaining voice samples and for performing voice analysis in children,2,3 there has been little work exploring VRQOL in the pediatric population. For young children, parent proxy instruments are important because (1) parents have the ability to comprehend the scope of the problem and (2) parental concern is an important driving force in the child's presentation to the specialist for consultation. Those instruments validated to explore issues related to VRQOL in adults4- 6 have not been validated for children and, specifically, parent proxy application. Recently, Hartnick et al7 modified the Voice Outcome Survey (VOS) as developed and validated by Gliklich et al5 to allow for parent proxy application. The Pediatric VOS (PVOS) was subsequently validated by examining a specific population of children with and without tracheotomies. The PVOS was shown to be internally consistent and "able to support a proposed interpretation of scores based upon theoretical implications within the constructs" (a concept known as construct validity).8 To broaden the applicability of the PVOS to children with a broad spectrum of vocal disorders, normative scores are necessary for cross comparisons. The present study was initiated to provide normative data regarding VRQOL for a variety of children with voice-related and non–voice-related concerns within the scope of a routine pediatric otolaryngology practice.
The PVOS (Figure 1) is a 4-item instrument designed for parent proxy application to measure VRQOL. The raw scores are assigned a value on a 0 to 100 scale for ease of interpretation. Low scores describe a relatively "poor" VRQOL; high scores describe a "better" VRQOL.
Over a 5-month period (February-June 2002), the PVOS was administered to all parents of children and adolescents between the ages of 2 and 18 years presenting to the outpatient office practice of 3 pediatric otolaryngologists (C.J.H., M.V., and M.C.). The administration of the instrument was performed in accordance with Massachusetts Eye and Ear Infirmary (Boston) institutional review board approval. Information regarding the children's age and presenting diagnoses was collected as well as information regarding previous therapeutic interventions that had been performed. To analyze test-retest reliability, the first 75 parents who completed the instrument were readministered the instrument 2 weeks after initial application. All initial instruments were completed according to standard paper application at the pediatric otolaryngologist office. The 75 parents who completed a second questionnaire did so at their home and returned the completed questionnaire by standard mail within a preaddressed envelope they were given. The completed forms were entered onto a Microsoft Access database (Microsoft Inc, Redmond, Wash), and analysis was performed by linking the database to the SAS 8.1 (SAS Institute Inc, Cary, NC) software package.
Overall, parents completed the PVOS instrument for 385 children and adolescents, 204 (53%) of whom were between the ages of 2 and 5 years, 126 (33%) between 6 and 10 years, and 52 (14%) between 11 and 18 years (3 children did not have their age reported). The main diagnoses for these children at time of initial presentation are presented below, with adenotonsillar hypertrophy being the most common (n = 119 [31%]).Table 1
Of the 385 children, 222 had undergone surgery in the 6 months prior to the completion of the PVOS. The types of procedures performed are presented below, with myringotomy and tympanostomy tube placement being the most common (n = 88; 40%).Table 1
Seventy-five parents completed the PVOS within a 2-week period in which no intervention had been performed. The test-retest weighted κ value was 0.89 (95% confidence interval, 0.84-0.95).
The PVOS scores were assigned values on a 0 to 100 scale. The mean ± SD converted score for the 385 patients enrolled was 80.5 ± 19.9. Table 1 describes the mean converted scores for the PVOS according to the presenting diagnoses of the children. Table 1 also depicts the PVOS mean converted scores according to age, sex, and operation-specific preoperative and postoperative status. Two-tailed t tests were used to analyze for significant differences in VRQOL between subpopulations. The PVOS scores varied significantly when the first 2 subgroups (age groups 2-5 and 6-10 years) were compared (P<.05). The PVOS scores also varied significantly for children before and after adenoidectomy (P<.05).
There is a paucity of literature published to date with regard to establishing standards for (1) creating and validating a new instrument to assess HRQOL or health status or (2) validating a previously validated disease-specific instrument for use with other distinct populations. With regard to choosing previously validated disease-specific instruments, several important issues arise, namely, which instrument to choose that bests suits the population to be studied and what statistical measures are necessary to demonstrate the validity and reliability of this instrument to measure what it purports to measure. It is important to keep in mind the essential points that an instrument must demonstrate. The instrument must be shown to measure what it purports to measure, and it must do so consistently as well as show that it measures changes in measurements where changes in fact exist. Finally, it should not be onerous to administer or to complete.
There have been several instruments designed with the purpose of measuring HRQOL specifically for vocal issues. The Voice Handicap Index has been validated as a 10-item instrument designed to measure the emotional, physical, and functional aspects of adults with voice disorders and record the patient's own perception of a vocal handicap.4,9 The VRQOL measure is another 10-item instrument that was designed and validated as a self-administered instrument for adult populations with voice disorders to measure social-emotional and physical functioning together in a generalized construct.6,9 Finally, the VOS was designed and validated as a 5-item instrument to measure clinical change in adults with unilateral vocal paralysis who would be undergoing surgery.5
All 3 of these instruments have undergone rigorous clinometric analysis for the purpose of initial validation. With regard to choosing an instrument to measure HRQOL for vocal issues in the pediatric population, none of these 3 instruments have been validated for this purpose. Moreover, to administer any adult instrument to a population in which the children are primarily under the age at which self-administered forms would be appropriate, each instrument would have to be altered to accommodate parent proxy application. It is not clear that measuring physical, emotional, or functional aspects of voice disorders in the adult population is the same as measuring these domains within a pediatric population. Therefore, it is the initial task of any pediatric application of an instrument designed to probe issues related to VRQOL to demonstrate validity and reliability specifically for this population. The initial studies of the PVOS addressed some of these issues from the context of a population of children with and without tracheotomies. The VOS had been altered for parent proxy administration. Whereas focus groups are often used in the original design phase of an instrument, for revalidation purposes to another population, factor analysis was used to demonstrate the cohesiveness of the internal structure of the PVOS with regard to its ability to record a global construct. This is a statistical method that can lead to item reduction, much as a focus group might say that one item did not fit well in terms of measuring what the instrument should be designed to measure. The initial factor analysis supported the results obtained from reliability analysis in which Cronbach α values had been recorded and supported the removal of 1 item from the initial VOS that had focused on swallowing issues. The PVOS in its revised and final form was able to discriminate between 2 pediatric populations in which differences in HRQOL regarding vocal issues were hypothesized to exist. It has advantages as a brief instrument with low administration and application burden with the potential to discriminate between different pediatric populations who will be undergoing speech or surgical therapy to address specific vocal disorders.
The present study was designed to obtain data regarding the scoring of the PVOS for a larger and broader pediatric population. Factor analysis supported the internal structure of the PVOS for this population. The instrument was shown to be reliable by means of evaluating Cronbach α values (Cronbach α values >.65 are believed to be sufficiently robust to lend support to the instrument design).10 In this administration, test-retest analysis was also used to further bolster reliability reporting. The PVOS has demonstrated test-retest reliability with a weighted κ value of 0.9 (since an ordinal scale was used, weighted κ values are presented). According to the criteria defined by Landis and Koch,11 this value implies a significant degree of test-retest consistency.
The purpose of this study was to derive normative data regarding pediatric VRQOL for a broad pediatric population. This research was not designed with a primary goal to achieve statistical power to analyze construct validity hypotheses; previous work supports the validity of the PVOS.7 However, it is a common clinical observation that a child's voice changes after an adenoidectomy; it therefore can be hypothesized that the mean PVOS scores would vary significantly before and after an adenoidectomy was performed. This is demonstrated by the data and, as such, is another measure of construct validity where the PVOS was able to discriminate between populations in which differences were hypothesized to exist. These data are not presented to document the change in voice and VRQOL after adenoidectomy; rather, it is presented to show that the PVOS allows for differences in subpopulations to be identified. With regard to the effect on VRQOL following adenotonsillectomy, there was a trend toward a difference, but the numbers were insufficient to provide conclusive information. There were no differences in VRQOL after tympanostomy tube placement, but, again, the numbers were too small to demonstrate statistical significances.
Analysis of how the PVOS was affected by the age of the child demonstrated significant differences in the age groups between 2 and 5 and 6 and 10 years. The exact reason for this statistical difference remains unclear. It may be that parental perception of or expectation that their children's voice and vocal ability changes over time. Analysis of the age group between 11 and 18 years was not included for several reasons. First, only 13% of the subjects were represented within this age group. Moreover, previous work demonstrated that the internal consistency of the PVOS, although still significant, diminished when this older age group was included with the other younger groups.7 This raises the hypothesis that self-administration of the PVOS to this age group may be more appropriate, and further research is ongoing in the attempt to clarify this issue. Because much of the work involving the pediatric airway (eg, surgery for velopharyngeal insufficiency and correction of subglottic stenosis) occurs before the age of 11 years, a parent proxy instrument remains in our opinion a reasonable design.
The PVOS is a brief, simple HRQOL instrument designed to analyze the VRQOL aspects of HRQOL. It is easily administered and can be incorporated into the scope of a busy clinician's practice to track changes after therapeutic strategies have been implemented. By design, it is not exhaustive in identifying all of the facets that incorporate vocal perception. Future work is needed to correlate changes in VRQOL with specific measures of vocal function including videoendoscopy with stroboscopy, acoustics, and aerodynamics. The PVOS is also limited in its availability (English only), and future work validating it in translation would broaden its application to a larger pediatric base.
Pediatric laryngology is a growing field in which diagnostic strategies are evolving and novel therapies need to be formally assessed. The PVOS represents a meaningful tool to collect data on these children and track their vocal changes over time.
Corresponding author and reprints: Christopher J. Hartnick, MD, Department of Otolaryngology, Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114 (e-mail: firstname.lastname@example.org).
Submitted for publication August 6, 2002; final revision received January 21, 2003; accepted February 5, 2003.