Professional Documents
Culture Documents
Creation and Validation of The Singing Voice Handicap Index
Creation and Validation of The Singing Voice Handicap Index
net/publication/6167566
CITATIONS READS
148 4,074
8 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Seth M Cohen on 04 June 2014.
Objectives: We developed and validated a disorder-specific health status instrument (Singing Voice Handicap Index;
SVHI) for use in patients with singing problems.
Methods: Prospective instrument validation was performed. Of 81 original items, those with poor statistical validity
were eliminated, resulting in 36 items. The ability to discriminate dysphonic from normal singers, test-retest reliability,
internal consistency, and construct validity were assessed.
Results: We included 112 dysphonic and 129 normal singers, professional and nonprofessional, of classical, country,
rock, choral, and gospel repertoire. Dysphonic singers had worse SVHI scores than normal singers (p <, .001, rank sum
test). Test-retest reliability was high (Spearman correlation, 0.92; p < .001). Internal consistency demonstrated a Cron-
bach's a of .97, and the correlation between the SVHI and self-rated singing voice impairment was .63 (p < .001, Spear-
man correlation).
Conclusions: The SVHI is a reliable and valid tool for assessing self-perceived handicap associated with singing prob-
lems.
Key Words: larynx, quality oflife, singing, voice.
402
Cohen et al. Singing Voice Handicap Index 403
emotional, social, and economic impact of singing determined by Cronbach's a. The item-total correla-
voice problems. Because we wanted to ensure that tions were calculated for all 81 items and, much as in
items in the final version were as clear as possible, prior studies, items with low item-total correlations
some of the preliminary items were reworded ver- of less than 0.6 were eliminated, as these items min-
sions of other items. Patients were asked to note the imally contributed to the internal reliability of the
frequency of their experience for each item of the overall questionnaire.'-^ Of redundant items, those
questionnaire. Each item was individually scored on with the lower item-total correlation were also elim-
a 5-point Likert scale anchored by "never" (score of inated. Associations between sex and item response
0) and "always" (score of 4). was assessed with y} statistics, to compare the fre-
The preliminary questionnaire was administered quency of a positive response (2 to 4) versus a nega-
to 86 consecutive new patients, all singers. Singing tive response (0 or 1) between men and women.
style and professional status were noted. The inter- The final version, the SVHI, had 36 questions
nal consistency of the preliminary questionnaire was (Table 1). The raw total scores ranged from 0 to 144,
404 Cohen et al, Singing Voice Handicap Index
with the higher number representing more self-per- teachers. This control group consisted of 129 sing-
ceived handicap. The raw scores were scaled from ers who anonymously completed the SVHI. To de-
0 to 100 by multiplying the raw total score by 100 termine the ability of the SVHI to discriminate be-
and then dividing by 144 to produce the final SVHI tween singers with voice problems and singers with-
score. To determine the number of factors or sub- out voice problems, a ^test was used to compare the
scales, we performed principal component analysis SVHI scores of the singing patients and the control
by the Quartimax and Varimax methods (SPSS Inc, group. Statistical analysis was performed with Sig-
Chicago, Illinois). maStat 2.03 (SPSS Inc).
The SVHI was administered to a second, consec- RESULTS
utive group of 112 new patients, all singers. Again,
the diagnosis, singing style, and professional status The initial questionnaire of 81 items was com-
were determined. This final version was completed pleted by 86 consecutive new singing patients. The
at the initial presentation, and a second copy was mean age was 34 years with a range of 14 to 66
mailed 1 week after the initial patient visit. This tim- years; 39.5% were male and 60.5% were female.
ing was chosen so that the patients' vocal health state The characteristics of this initial cohort are repre-
would not have changed and also so they would not sented in Table 2.
remember their initial answers. The test-retest reli- The Cronbach's a of the preliminary question-
ability was determined by a Spearman correlation. naire was .97. Two items with item-total correla-
The validity was determined in several ways. In tions of less than 0.6 were retained because of face
addition to the SVHI, the 112 patients also self-rat- validity; they concerned performance cancellation
ed the severity of their singing problems on a 10- and the economic impact of the singing problem.
cm horizontal visual analog scale (VAS) anchored Four items for which more than 50% of the patients
at 0 by "not a problem" and at 10 by "severe prob- answered "never" were eliminated. Three redundant
lem." Correlation between the SVHI and VAS was items were also removed. As a result, the final ver-
assessed with the Spearman correlation. The VAS sion ofthe SVHI contained 36 items. No association
scores were also divided into 3 severity categories between item response and sex was identified for
(group 1, scores of 0 to the 25th percentile; group 2, any of the retained items (p > .2, x^).
the 25th percentile to the 75th percentile; group 3, The final SVHI was completed by a second co-
the 75th percentile to the highest score). By an anal- hort of 112 consecutive new patients, all singers.
ysis of variance, the SVHI scores from each severity The mean age was 35.3 years with a range of 16
category were compared. to 67 years; 40.2% were male and 59.8% were fe-
Furthermore, singers who denied having any male. The patients had had problems with their sing-
problems with their voice and had not sought med- ing voice for a mean of 16.4 months (range, 1 week
ical care because of a voice problem were identi- to 120 months). Patient characteristics are shown in
fied from professionals, amateur singing students, Table 3, and diagnoses in Table 4.
choirs and amateur singing groups, and singing The principal component analysis found that the
Cohen et al. Singing Voice Handicap Index 405
TABLE 4. DIAGNOSES OF FINAL PATIENT COHORT control group's SVHI scores regardless of profes-
Diagnosis* Percent sional, income, or classical singing status (p < .001,
Benign vocal fold lesiont 43.7 rank sum test).
Muscle tension dysphonia 37.5
DISCUSSION
Chronic laryngitis 12.5
Acute laryngitis 8.9 Singers have demanding voice needs and are at
Vocal fold edema 8.9 risk for developing voice problems that lead to med-
Laryngopharyngeal reflux 5.4 ical evaluation and treatment. Measuring the im-
Spasmodic dysphonia 1.8 pact of singing voice problems from the patient's
Superior laryngeal nerve palsy 1.8 perspective will facilitate the evaluation and man-
Recurrent laryngeal nerve palsy 1.8 agement of patients who sing vocationally or avo-
*Patients may have more than one diagnosis. cationally. Hence, this study's objective was to cre-
tincludes polyps, cysts, pseudocysts, nodules, and sulcus. ate and validate a health status instrument for use in
singers.
items loaded on a single factor that explained 51.5%
Current knowledge of the impact of singing prob-
of the variance. Hence, the SVHI was scored as a
lems is limited by the available health status in-
single scale. The test-retest reliability was assessed
struments. Rosen and Murry" and Behrman et al'^
by having the patients complete the SVHI at 2 points
found that singers had lower Voice Handicap Index
in time. Half of the patients completed the second
(VHI) scores than nonsingers. Several hypotheses
SVHI a mean of 17.3 days (SD, 9.5 days) after the
may explain this finding. Singers' voice problems
first SVHI. The test-retest reliability was .92 (p <
may differ from those of nonsingers, and singers
.001, Spearman correlation). The critical difference
may be more sensitive to voice changes and thus
score (95% confidence interval for the difference of
present earlier." However, the VHI does not ad-
the mean SVHI scores at the two time points) was
dress limitations that result from singing voice prob-
14. Cronbach's a for the final version of the SVHI
lems. Because the VHI may not be sensitive enough
was .97.
for use in singers, important handicaps experienced
The construct validity was evaluated by compar- by singers may go unmeasured and unnoticed. Ad-
ing the SVHI scores to patients' self-rated severity ditionally, the singing voice might be more suscep-
of singing problems on a VAS. The correlation be- tible to various medical disorders such as reflux or
tween the VAS and SVHI was 0.63 (p < .001, Spear- allergies, but their influence on the singing voice
man correlation). Additionally, the SVHI scores may not be adequately measured on questionnaires
from the 3 severity categories (group 1, mean [±SD] that primarily focus on the speaking voice.''•'•^ Thus,
25.7 ± 14.7; group 2, 40.8 ± 17.3; group 3, 64.8 ± having a voice-related instrument that assesses sing-
20.5) based upon the VAS were statistically signifi- ing voice impairment and the resultant handicap is
cantly different (p < .001, analysis of variance; p < a necessary component of diagnosis for singers with
:.O5, Bonferroni f-test; for all group comparisons). vocal difficulties.
To determine the discriminant validity, we had
The SVHI can be a valuable tool for measuring the
129 singers who denied having a problem with their
handicap resulting from singing voice impairment.
singing voice and had not sought medical care for
Morsomme et al'^ recognized the lack of an instru-
a voice problem complete the SVHI. Three fourths
ment for assessing the singing voice and adapted the
of this control group were classical singers, and the
VHI to singers. Their questionnaire was developed
rest were country, choral, pop, rock, or gospel sing-
and evaluated in 37 dysphonic classical singers and
ers. One third said singing was either a primary or
has only been published in French. In contrast, our
secondary income source, with one fifth being pro-
study used a larger cohort of patients. Moreover, the
fessional singers and the remainder students, sing-
SVHI was validated across a diverse group of sing-
ing teachers, or amateurs. Compared to the 112 sing-
ers with various causes of handicap, singing styles,
ing patients, the control group of singers had lower
and professional status. Hence, the SVHI has wide
SVHI scores: a median of 22 versus a median of 61
application to singers with different kinds of train-
(p < .001, rank sum test). Because the singing pa-
ing, styles, and voice disorders.
tients were more likely to be professional, more of-
ten used singing as a primary or secondary source The SVHI has important psychometric proper-
of income, and were less likely to be classical sing- ties. To ensure the content validity, we selected the
ers, the SVHI scores were compared within the sub- individual items according to patient report and the
groups of professional, income, and classical sing- expertise of laryngologists, speech pathoiogists, and
ing status. The patients' SVHI scores were twice the a vocal pedagogue experienced in caring for sing-
406 Cohen et al. Singing Voice Handicap Index
ers. Items were chosen to conceptually evaluate the theless, only 50% ofthe patients completed the sec-
physical, emotional, economic, and social impact ond SVHI. Whether this second SVHI was lost in
of singing voice problems. The SVHI demonstrat- the mail, misplaced by the patient, or not complet-
ed high internal consistency and high test-retest re- ed because of the effort and time needed to com-
liability. Hence, the SVHFs inherent, random vari- plete the SVHI a second time is not known. Also,
ability is low, enhancing its ability to detect changes although our test-retest reliability was high, patients
in singing voice-specific health status across treat- with more variability in their answers may not have
ment. completed the second SVHI, so our test-retest reli-
ability may have been inflated. Last, the responsive-
Additionally, the SVHI serves as a valid, disor-
ness ofthe SVHI to treatment-related changes in pa-
der-specific health status instrument. As evidenced
tients' singing voices was not evaluated and will be
by the correlation between self-rated singing voice
the subject of future investigations. Correlations be-
problems on the VAS and SVHI and the increased
tween the VHI and SVHI and associations between
SVHI scores across the 3 VAS severity categories,
singing styles, comorbidities, and the SVHI are im-
the SVHI does differentiate between different levels
portant avenues of study. Further analysis may also
of patient-perceived singing voice impairment. Last,
allow f'urther reduction of the number of items.
as evidenced by higher SVHI scores in the singing
patients compared to the control group, the SVHI CONCLUSIONS
discriminates between singers with vocal dysfunc-
tion and singers without voice problems; this find- The SVHI is a reliable and valid instrument for
ing further supports its validity. This difference per- measuring the patient-perceived impact of singing
sisted regardless of professional, income, or classi- voice problems. Consequently, the SVHI can aid in
cal singing status. the assessment of dysphonic singers. Factors influ-
encing the patient's perception of singing voice im-
Certain methodological issues must be addressed. pairment, as evidenced by the SVHI, are worthy of
Despite the 36 questions, the SVHI's burden ap- investigation. The SVHI may also serve as a valu-
peared minimal. Of all the patients approached, only able tool for the evaluation of treatment outcomes
1 patient refused to complete the SVHI. Most com- among singers and for comparing the effectiveness
pleted the questionnaire in 5 to 10 minutes. None- of different treatments.
REFERENCES
1. Jacobson BH, Johnson A, Grywalski CSilbergleit A, Ja- pact of non-neoplastic voice disorders: a meta-analysis. Ann
cobson G, Benninger MS.The Voice Handicap Index (VHI): de- Otol Rhinol Laryngol 2006; 115:128-34.
velopment and validation. Am J Speech Lang Pathol I997;6:66-
70. 8. Benninger MS, Ahuja AS, Gardner G, Grywalski C. As-
sessing outcomes for dysphonic patients. J Voice 1998;12:540-
2. Rosen CA, Lee AS, Osborne J, Zullo T, Murry T. Devel- 50.
opment and validation ofthe Voice Handicap Index-IO. Laryn-
goscope 2004; 114:1549-56. 9. Roy N, Merrill RM,Gray SD, Smith EM. Voice disorders
in the general population: prevalence, risk factors, and occupa-
3. Deary IJ, Vk'ilson JA, Carding PN, MacKenzie K. VoiSS:
tional impact. Laryngoscope 2005;! 15:1988-95.
a patient-derived Voice Symptom Scale. J Psychosom Res 2003;
54:483-9. 10. Phyland DJ, Oates J, Greenwood KM. Self-reported
4. Gliklich RE, Glovsky TM, Montgomery WW. Validation voice problems among three groups of professional singers. J
of a voice outcome survey for unilateral vocal cord paralysis. Voicel999;13:602-ll.
Otolaryngol Head Neck Surg 1999; 120:153-8. 11. Rosen CA, Murry T. Voice Handicap Index in singers. J
5. Hogikyan ND, Sethuraman G. Validation of an instru- Voice 2000;14:370-7.
ment to measure voice-related quality of life (V-RQOL). J Voice 12. Behrman A, Sulica L, He T. Factors predicting patient
1999; 13:557-69. perception of dysphonia caused by benign vocal fold lesions.
6. World Health Organization. International Classification Laryngoscope 2004; 114:1693-700.
of Impairments, Disabilities, and Handicaps: a manual of clas-
13. Morsomme D,Simon C, Jamart J, Remade M, Verduyckt
sification relating to the consequences of disease. Geneva, Swit-
1. A proposal to adapt the Voice Handicap Index to the sing-
zerland: World Health Organization, 1980:25-43.
ing voice [in French]. Rev Laryngol Otol Rhinol (Bord) 2005;
7. Cohen SM, Dupont WD,Courey MS. Quality-of-life im- 126:305-13.
View publication stats