Self-reported voice problems among teachers: prevalence and associated factors

OBJECTIVE: To estimate the prevalence of self-reported voice problems and to identify associated factors. METHODS: A cross-sectional study was carried out in a sample of 393 public elementary and middle school teachers in Florianópolis, Southern Brazil, in 2009. A self-administered questionnaire was used. A multivariable Poisson regression model was performed to estimate prevalence ratios and their related 95% confi dence intervals. RESULTS: The prevalence of voice problems was 47.6% (95%CI 42.6;52.5). In the fi nal adjusted analysis the following variables remained associated with a higher prevalence of voice problems: being female and the presence of rhinitis/sinusitis and pharyngitis. CONCLUSIONS: A high prevalence of self-reported voice problems was found among the teachers studied. DESCRIPTORS: Voice Disorders, epidemiology. Faculty. Occupational Health. Cross-Sectional Studies.


INTRODUCTION
Voice problems are common disorders among teachers who are professional voice users where the voice is a key instrument in their work. 9sphonia is any change in voice due to a functional and/or organic disorder of the vocal tract that prevents natural production of voice.It can manifest as several symptoms such as tiredness or strain while talking, hoarseness, persistent throat clearing or cough, sensation of chest tightness or heaviness in the throat, voice failures, among others. 2,9A 2006 study found an association between voice disorders and professional voice use.a A review of Brazilian and international studies on vocal disorders among teachers showed a prevalence ranging between 20% and 89%. 12gh vocal demand, sociodemographic factors and risks directly related to work organization and environment are all related to the occurrence of dysphonia.While working teachers are exposed to other harmful factors in the workplace that can affect their general and vocal health by creating competing sounds and leading to greater vocal strain and demand.Environments with internal and external background noise, classrooms with inadequate acoustics, excessive number of students in a classroom, exposure to dirt and chalk in the classroom a Medeiros AM.Disfonia e condições de trabalho das professoras da rede municipal de ensino de Belo Horizonte [Mater's dissertation].Belo Horizonte: Universidade federal de Minas Gerais; 2006.
are some of the harmful agents that can negatively affect teachers' vocal health. 3[11] The multifactorial nature of causes of dysphonia, its high prevalence and limited knowledge about vocal output have encouraged speech and language therapists (SLTs) to develop collective works with teachers.A sound knowledge of the professional use of voice as well as of the work environment and job characteristics can help planning specifi c prevention actions targeted to this professional group 2,3 such as Workplace Health and Wellness Program at local public schools in the city of Florianópolis, Southern Brazil.
This study aimed to estimate the prevalence of selfreported voice problems and to identify associated factors among teachers.

METHODS
A cross-sectional study including teachers actively working in 36 public elementary and middle schools (grades 1 to 9) was conducted in the city of Florianópolis.
Florianópolis had an estimated population of 416,269 inhabitants in 2007 and a literacy rate of 96.7%.There were 107 public schools, of which 36 elementary and middle schools and 71 preschools with 3,945 teachers, 31.8% of them permanent and 68.2% replacement/ contracted.Elementary/middle school teachers accounted for a third of all teachers working in public schools, of which 56% were permanent, 38% replacement and 6% relocated.b All teachers actively working in 2009 were eligible to participate in the study, regardless of the type of their employment contract, totaling 1,044.
Physical education, sign language, support rooms, and foreign language teachers and those performing administrative activities were excluded from the study because the characteristics of their work organization, physical workplace and vocal demand were different from those of the target population.
For sample size estimation, a universe of 1,044 teachers was used based on the lowest self-reported prevalence of voice problems in the literature 7 (17%), with a sampling error of three percentage points, 95% confidence level (α = 5%) and 10% added to compensate for losses and non-responses, totaling 420 teachers.For operational reasons, a new sample size was then estimated considering a sampling error of four percentage Systematic sampling was carried out.The sample fraction was calculated using the following formula: k = N/n = 1,044/420 = 2.48 ~ 3. The sample was drawn from a single list of eligible teachers in alphabetical order and every third name was selected.
Data was collected using a self-administered questionnaire, which was pre-tested in a sample of 20 teachers from a state public school with similar characteristics as those defi ned in the present study.Questionnaires from other similar studies were the basis for the development of the study instrument used. 3,6The questionnaire comprised questions on demographic and socioeconomic characteristics (gender, age, marital status, education level, family income in the month prior to the study, number of persons living in the respondent's household); information on the teacher's work organization at school (type of employment, years working as a teacher, number of classes, number of students per class, total hours worked per week); information about their work environment (noise levels, room acoustics, dirt and chalk dust, moisture, resting area, rest break, relationship with colleagues, performance monitoring, teacher-student relationship, school violence); information on health behaviors and self-reported morbidities (hydration level, hypertension, diabetes, rhinitis/ sinusitis, asthma, hearing loss, depression, pharyngitis, ulcers, gastritis, physical activity, alcohol use, smoking) and self-reported voice problems.
Data was collected from May to July 2009 and data collection was repeated in August to minimize losses.
The school management authorized the researchers to conduct the study prior to data collection.The questionnaires were delivered to school coordinators who handed them out and returned them 15 days later.
The study was conducted in schools where teachers worked and anonymity of institutions and respondents was assured.The questionnaires were coded and placed into sealed envelopes with instructions for questionnaire completion and a consent form enclosed.
The dependent variable was self-reported voice problem (yes/no) "in the last four weeks" obtained by asking the following question: "Do you have any voice problem?".
Data were analyzed using Stata version 9.0.
First, it was carried out the descriptive analysis of all variables using frequency distributions for categorical variables and measures of central tendency and dispersion for continuous variables, later converted into categorical variables for analysis.All categories with few responses were combined (classroom noise levels -"negligible" + "acceptable;" outside school noise levels -"unacceptable" + "high;" and moisture in the classroom -"always" + "often;" good relationship with colleagues -"no" + "sometimes;" water intake during the day -"none" + "less than 1 L;" and alcohol use -"daily" + "weekends").The outcome prevalence was estimated based on the independent variables by testing differences between proportions using Pearson's chi-square test and the chi-square for linear trend if applicable.
For the analysis of factors associated with self-reported voice problems, Poisson regression with robust variance was performed and the reference category was no voice problems.The magnitude of the association of each factor with voice problems was assessed by crude and adjusted prevalence ratios (PR) and their related 95% confi dence intervals (95% CI) and p-value (Wald test).The analyses followed a theoretical hierarchical model where the variables were arranged in four blocks.The fi rst block, more distal, included demographic and socioeconomic variables, which conditioned the variables in block 2, work organization, and block 3, environmental (physical and psychosocial) factors, which in turn had an effect on the variables in block 4 (health-related behaviors and self-reported morbidities) and the study outcome.Variables with p < 0.20 in the bivariate analysis were selected to be included in the multiple analysis, and those with p ≤ 0.05 in the hierarchical blocks remained in the fi nal model.
The study was approved by the Human Research Ethics Committee of the Pro-Reitoria de Pesquisa e Extensão da Universidade Federal de Santa Catarina (protocol no.237/2008).All participants signed a free consent form.

RESULTS
Of 420 teachers randomly selected, 393 (93.6%) completed the questionnaire.Of the 27 (6.4%)teachers who did not participate in the study, four were on sick leave, fi ve were on maternity leave, three were on leave due to voice problems and 15 refused to participate.
The prevalence of self-reported voice problems was 47.6% (95%CI 42.6;52.5).Table 1 shows that most of the study sample were female (86.8%), 57.1% were married, and over half had specialization.The mean age was 40.2 years (SD 7.8) and average per capita income was 1765.08 reais (SD 1,038.43).The prevalence of self-reported voice problems was higher among women (p = 0.021) and those in the fi rst tertile of per capita income (p = 0.013).
Regarding the characteristics of work organization (Table 2), 63.4% of the respondents worked 21 to 40 hours weekly, 64.9% were permanent teachers, and 53.9% had 28 or fewer students per class.No variable in this group was statistically associated with self-reported voice problems.
Table 3 shows that more than half of the respondents rated the noise level in the classroom and in the school area as high or unacceptable.Over 60% reported frequent or continuous exposure to dirt and chalk dust in the classroom.The following variables were associated with the study outcome: unacceptable noise levels in the classroom (p = 0.019), unacceptable noise levels in the school area (p = 0.004), constant exposure to dust in the classroom (p = 0.007), no break between classes (p = 0.002), constant monitoring of their performance (p = 0.036), strained teacher-student relationship (p < 0.001) and school violence (p = 0.003).
As for health-related behaviors and self-reported morbidities (Table 4), 80.0% of the teachers reported drinking water during classes, but 62.0% reported drinking less than one liter of water a day.The most frequently reported comorbidities were rhinitis/sinusitis (47.0%) and depression (27.6%).The presence of morbidities such as rhinitis/sinusitis (p < 0.001), depression (p = 0.001) and pharyngitis (p < 0.001) and physical inactivity (p = 0.043) were associated with higher prevalence of self-reported voice problems.
In the fi nal adjusted model (

DISCUSSION
The prevalence of self-reported voice problems was 47.6%, very close to that found in similar studies carried out in cities of the state of Rio Grande do Sul, c Bahia, Northeastern Brazil, d and Pará, Northern Brazil, 8 and much higher than 17% 7 used in the estimate of the study sample, though it should be noted that different defi nitions of voice problems and methods were used. 3,6,16espite these diffi culties in comparing results, the high prevalence of vocal problems in teachers is a consensus in Brazil and in other countries. 2,9,10male teachers more often reported more voice problems than their male colleagues, 10,14 which is partly explained by gender-related differences of the larynx.For example, hyaluronic acid, a protein that increases water fl ow to the lamina propria of vocal folds leading to fewer traumas during emission, is more abundant in males.The glottic proportion in the female larynx is smaller, which may hinder phonic adaptations for heavy voice use. 1 Social and cultural factors, expressed by specifi c characteristics and the potentially demanding female social role may also contribute to voice problems among women. 15In addition, it should be noted the small number of men included in the study sample.
There was found a signifi cant association between selfreported voice problems and rhinitis/sinusitis and pharyngitis.Other studies also showed a positive association between dysphonia and respiratory problems such as allergic rhinitis 13 and pharyngitis. 4Exposure to dirt and chalk dust in the classroom increases the likelihood of developing upper airway conditions.These respiratory problems are associated with environmental conditions and affected by individual predisposing factors. 4e present study did not fi nd an association between self-reported vocal problems and work organization (years working as a teacher and workload), which is consistent with other studies. 2,4This fi nding may be explained by the healthy worker effect, since the development of severe vocal problems may lead to a career change, or even early retirement. 4,16It may also be that during the course of their careers, teachers tend to develop compensatory strategies or techniques to minimize the diffi culty in voice production, which may have an effect on their perception of voice quality. 13,16esides, this fi nding may have resulted from an homogeneous distribution of the population according to the variables workload and years working as a teacher.
The prevalence of self-reported voice problems was associated with low per capita income in the crude analysis.Income is a determinant of the health-disease process, and those with lower income tend to get sick more often, are more susceptible to diseases and are more exposed to several risk factors. 5 the crude analysis, the prevalence of self-reported voice problems was signifi cantly higher among teachers who reported exposure to dirt and chalk dust and unacceptable noise levels in the classroom and in the school area.These environmental conditions of dust and noise negatively interfere with learning activities, and similar data were reported to have a signifi cant association in other studies.Teachers strain their voice in noisy environments for long periods and do not have enough time to rest or recover it and are often exposed to stressful situations. 9Although the association lost signifi cance in the adjusted model, there were few reports of rest breaks between classes, and bearing in mind the school day-today and total workload of teachers, it can be assumed that teachers work two to three shifts daily with vocal overload. 3study found an increased prevalence of vocal symptoms between 1998 and 2001, and concluded that the deterioration of working conditions during the 1990s with increasing inappropriate behaviors and noise levels in the classroom can be all be stressors among teachers. 11In the present study, even though on borderline statistical signifi cance, persistent strained teacherstudent relationship was associated with the outcome.
Recent studies have showed an association between strained relationship with students and lower voicerelated quality of life and reported voice symptoms. 6,7study has reported a positive association with water intake for maintaining and improving vocal quality, 17 however our study did not corroborate this fi nding.Given the cross-sectional design of this study a cause and effect relationship cannot be established.It is believed that people with voice problems drink more water to alleviate their voice symptoms.However, A major limitation of this study related to the exclusion of teachers on medical leave, which probably underestimated the actual prevalence of the outcome studied.Also, studies based on self-administered questionnaires are prone to the existence of self-report bias.
We used the time frame "in the last four weeks" for self-report of voice problems that allows adequate recall by the teachers and can improve the quality of information provided.
The study results showed that voice problems in teachers can be associated with gender-related factors and self-reported morbidities and, on borderline statistical signifi cance, with psychosocial and work environment aspects as well.These fi ndings reinforce the need for actions for promoting voice health among teachers so that there is a good interaction between the teachers, their working conditions and their general and voice health.
These actions should involve structural changes in the workplace as well as increased awareness about voice care among teachers through workshops to develop Municipal Administration of Florianópolis.Perfi l de Florianópolis.Florianópolis.[cited 2009 Nov 05].Available from: http:/portal.pmf.sc.gov.br/arquivosl/arquivos/pdf/0511 2009 13.00.43.d53d27cbe464ff1805a76dbb9631cf6c.pdf/ points, which meant reducing the precision of the prevalence estimate. b

Table 1 .
Prevalence of voice problems according to individual characteristics among teachers of local public schools.Florianópolis, Southern Brazil, 2009.
a Pearson's chi-square test b Chi-square for linear trend

Table 2 .
Prevalence of voice problems according to job characteristics among teachers of local public schools.Florianópolis, Southern Brazil, 2009.
a Pearson's chi-square test b Chi-square for linear trend

Table 3 .
Prevalence of voice problems according to work environment characteristics among teachers of local public schools.Florianópolis, Southern Brazil, 2009.
a Pearson's chi-square test b Chi-square for linear trend

Table 4 .
Prevalence of voice problems according to healthrelated behaviors and self-reported morbidities among teachers of local public schools.Florianópolis, Southern Brazil, 2009.
a Pearson's chi-square test b Chi-square for linear trend