ARTICLE IN PRESS Prevalence and Occupation of Patients Presenting With Dysphonia in the United States *Michael S. Benninger, †Chantal E. Holy, *Paul C. Bryson, and *Claudio F. Milstein, *Cleveland, Ohio, and †Menlo Park, California Summary: Objective. Voice disorders are common conditions that may have a significant impact on patient quality of life, yet their prevalence and epidemiology are poorly documented. In this study, we estimated the prevalence, demographics, and occupation of patients with dysphonia. Methods. Using the Commercial and Medicare MarketScan databases of 146.7 million lives (2008–2012), the prevalence of dysphonia was estimated. Patient demographics and industry occupation were evaluated. Prevalence estimates overall and by industry were made using Medical Expenditure Panel Survey. Industry estimates were compared with US government employment statistics to assess differences between dysphonia and the general population. Results. A gradual increase in the diagnosis of dysphonia was noted from 1.3% to 1.7% of the population from 2008 to 2012, with an associated increase in the diagnosis of acute laryngitis, the largest diagnostic category. A strong correlation was present between diagnosis and age, with acute laryngitis more common in the younger populations and malignancies in older ages. Benign neoplasms were more prevalent in the service industry, with 2.6 times increased likelihood compared with the general population, and malignancies were more prevalent in the manufacturing industry, with 1.4 times increased likelihood. Almost 3 million laryngoscopies and stroboscopies were performed with $900 million in costs. Conclusion. Prevalence rates of the diagnosis of dysphonia are increasing and are associated with large healthcare costs. Prevalence rates also differ somewhat between industries, and there appears to be a higher percentage of malignant neoplasms in the manufacturing industry and benign neoplasms in the service industry. Key Words: Dysphonia–Prevalence–Etiology–Industry–Occupation. INTRODUCTION Voice disorders are relatively common and can have a significant impact on patient quality of life and healthcare costs1,2 Although there has been significant work within the voice care community to evaluate quality of life and outcomes, there has been a paucity of data related to general prevalence rates and overall costs.2–6 With an estimated prevalence of approximately 1% in the US population, dysphonia’s most prevalent manifestation is acute laryngitis, with a relatively small proportion of patients experiencing far more debilitating conditions such as laryngeal malignancies.2 A telephone survey of 1326 patients in Iowa and Utah has shown that 29.9% of the population has a lifetime history of dysphonia, with 6.6% of patients having a current voice disorder.3 These prevalence estimates are not noted in larger claims-based diagnostic prevalence rates. Utilizing the 2012 National Health Interview Survey, an estimated 17.9 million adult patients reported a voice problem affecting approximately 1 in 13 US adults. Despite this, only 10% sought care from a healthcare worker. Proper diagnosis and evaluation of the underlying causes of dysphonia often requires diagnosis and Accepted for publication January 12, 2017. Presented at the Annual Meeting of the Voice Foundation: Care of the Professional Voice, Philadelphia, 2015. Source of funding: License to access the MarketScan database was funded by Johnson & Johnson. Financial disclosures: CEH is a paid employee of Acclarent Inc. From the *Head and Neck Institute, The Cleveland Clinic, Cleveland, Ohio; and the †Johnson & Johnson, Menlo Park, California. Address correspondence and reprint requests to Michael S. Benninger, Head and Neck Institute, Lerner College of Medicine, The Cleveland Clinic, 9500 Euclid Avenue, A-71, Cleveland, OH 44195. E-mail: [email protected] Journal of Voice, Vol. ■■, No. ■■, pp. ■■-■■ 0892-1997 © 2017 Published by Elsevier Inc. on behalf of The Voice Foundation. http://dx.doi.org/10.1016/j.jvoice.2017.01.011 management by an otolaryngologist. Recent studies have shown that patient diagnoses often change as patients transition from primary to subspecialty care, and increased time in nonspecialist care results in greater healthcare cost, potentially due to delays in targeted treatments.7,8 Direct yearly expenditures for a voice problem were $300 million.5 Despite these efforts to assess prevalence, impact, and cost, there are notable limitations in the information available related to dysphonia, in part due to limitations in the data and potentially in the methodology of the evaluations. In addition, there has not been good data developed in relationship to different voice disorder diagnoses or variations by age and sex. Because there is an impact on work activity,3 a better understanding of prevalence and specific diagnosis based on the industry in which the patient works would be expected to be valuable. The at-risk population for vocal disorders may significantly vary based on etiology. This preliminary study was thus designed to evaluate changes in prevalence of dysphonia from 2008 to 2012, by etiology, and evaluate demographic as well as occupational profiles of patients presenting with various conditions, to start understanding the impact of various dysphonia diseases on patients. Because of the large prevalence of disease in the >65-year-old population, both the Commercial and Medicare MarketScan databases were queried, and a projection methodology was used to estimate nationwide counts of disease on a yearly basis. METHODS The study is a retrospective analysis of a large administrative database designed to collect and pay for medical and drug claims (aka claims-based analysis), with no identification of individual ARTICLE IN PRESS 2 Journal of Voice, Vol. ■■, No. ■■, 2017 patients and therefore was exempt from the need to obtain an institutional review board approval. Databases The Truven Health MarketScan Commercial Claims and Encounter Database (Truven Health Analytics, Ann Arbor, MI) as well as the MarketScan Medicare database were queried. These very large databases (in 2013: Commercial: 36.7 million; Medicare: 3.2 million) track patients longitudinally across all sites of care over multiple years. Recently, Truven Health researchers developed a projection methodology to allow estimates of US-wide frequencies of diagnoses or procedures from those obtained directly from the databases, making these databases invaluable tools to estimate prevalence and incidence of disease and procedures. These databases have recently been used for a number of healthcare cost and epidemiology studies for dysphonia and other general otolaryngological conditions.2,8,9 Dysphonia disease identification All patients with a primary or secondary diagnosis of dysphonia, in both outpatient and inpatient care, were identified from 2008 to 2012. Projection methodology Nationwide projections were obtained using a proprietary methodology developed by MarketScan. Briefly, the MarketScan population was subdivided into 72 demographic cells linked to similar subpopulations from the Medical Expenditure Panel Survey (MEPS).10 Each cell within the MEPS survey includes a number of people with similar characteristics. A weight factor was assigned to each cell, such that the weight would be equal to the ratio of the MEPS-estimated population for that particular cell, divided by the MarketScan sample size for that cell. In our analyses, patients with 12 months of consecutive enrollment in MarketScan were included to avoid the potential of statistical noise. Each calendar year was analyzed separately. Unique patient diagnostic categories were counted only once in a given calendar year, such that a patient with, for example, two diagnoses of acute laryngitis only accounted for one count of prevalence for acute laryngitis. Compounded annual growth rates for each dysphonia disease type were estimated from 2008 to 2012. Using the MEPS methodology allowed for projection of patient counts to the entire population covered via Commercial or Medicare supplemental insurance. To further project to nationwide estimates, all prevalence counts were further multiplied by the respective ratios that accounted for all other patients. (For younger than 65 years old, the additional ratio was equivalent to: (total population under 65) / (MarketScan projected commercial population). For those older than 65 years old, the additional ratio was equivalent to: (total population over 65) / (MarketScan projected Medicare population). Cohort analyses All patients identified in the prevalence analysis were further characterized by age, work status (employed vs not employed), and industry affiliation. The ratio of patients with dysphonia by work status and industry affiliation was compared with those obtained from the overall projected MarketScan population. Statistical analyses Continuous variables were shown as means and standard deviations. Proportions were calculated for all categorical variables and presented as ratios with confidence intervals. To evaluate statistical differences between proportions, t tests for unpaired cohorts were conducted (α = 0.05). All statistical analyses were performed using SAS EG 4.3 (SAS Institute, Inc., Cary, NC). RESULTS The total volume of covered lives in the MarketScan Commercial and Medicare databases increased from 36.8 million in 2008 to 56.1 million in 2012. Of these populations, 343,364 patients in 2008 and 661,578 patients in 2012 presented with dysphonia. Using the projection methodology described above, the overall prevalence for each dysphonia diagnosis was projected for the US-wide population (304.3 million in 2008 and 313.8 million in 2012). The results are shown in Table 1. Briefly, acute laryngitis represented approximately 54% of all cases of dysphonia, followed by general “dysphonia” diagnoses (22%) and chronic laryngitis (6%). All other diseases were fairly rare, accounting for less than 200,000 patients per year and thus falling under the “orphan disease” definition (according to US criteria, a disease that affects fewer than 200,000 people). TABLE 1. Prevalence Estimates and Compound Annual Growth Rates (CAGR) for All Conditions From 2008 to 2012 Acute laryngitis Dysphonia Other Chronic laryngitis Malignant neoplasm of larynx Edema of larynx Vocal fold polyps Paralysis 2008 2009 2010 2011 2012 CAGR (%) 2,206,593 760,265 341,549 248,846 104,009 78,142 74,177 69,390 2,604,168 951,775 389,834 293,151 97,814 92,839 77,137 76,452 2,530,288 1,068,905 437,986 310,972 103,084 113,531 83,030 81,167 2,650,974 1,134,977 489,226 325,397 105,131 123,482 85,104 88,945 2,754,056 1,247,471 565,490 342,601 108,209 130,184 88,694 94,418 5.7 13.2 13.4 8.3 1.0 13.6 4.6 8.0 The CAGR of the US population over the same time period was estimated at 0.8%. ARTICLE IN PRESS Michael S. Benninger, et al Prevalence and Occupation of Patients With Dysphonia 3 FIGURE 1. Prevalence of dysphonia in the United States, 2008–2012. There was a gradual increase in prevalence rates of the diagnosis of dysphonia from 2008 (3.8 million) to 2012 (5.3 million), with an associated increase of percentage of the population from 1.3% to 1.7% (Figure 1). The average age of patients in each diagnostic category was compared. Except for acute laryngitis, all dysphonia diseases affected a population with an average age of >50, and for malignant disease the population was >60 years old, as shown in Table 2. The distribution of the specific diagnoses is noted in Figure 2, and the growth in the prevalence of various diagnoses is shown in Figure 3. There was a strong correlation of specific diagnoses by age, with acute laryngitis, laryngospasm, and benign lesions being more common in the younger age groups, and malignancies, paralysis, and paresis being more common in the older age groups (Figure 4). The procedures used to make or confirm the diagnosis are presented in Figure 5, along with projections for 2018. Cost estimates are also shown. A majority of patients had a flexible laryngoscopy performed, accounting for 2.2 million of the 2.7 million procedures TABLE 2. Average Age of Patients Presenting With Dysphonia Diagnoses Diagnosis Malignant neoplasm of larynx Paralysis Paresis Chronic laryngitis Dysphonia Benign neoplasm of larynx Edema of larynx Vocal fold polyps Other diseases of vocal cords Laryngeal spasm Acute laryngitis Average Age Standard Deviation 66.16 58.97 58.58 54.92 54.82 53.86 53.10 52.38 50.05 42.58 19.53 11.72 19.41 18.58 17.96 20.06 17.78 18.87 17.13 21.44 24.68 23.63 performed. The overall estimated cost of diagnosis is $986 million. Cost estimates were not obtained for cost of treatment (Figure 5). Industry representation in the MarketScan database is shown in Table 3 and Figure 6. Manufacturing (35%) and service (24%) industries represent 59% of the patients. Significant differences were identified between specific voice diagnoses and the industry of the patient in comparison with the overall MarketScan population (Table 3 and Figure 7), with acute and chronic laryngitis and benign neoplasms of the larynx higher in the service industry (1.3 times more likely), and malignant neoplasms of the larynx higher in the manufacturing industry (1.4 times more likely). DISCUSSION In this study, the total prevalence and compounded annual growth rates of all dysphonia conditions were evaluated from 2008 to 2012. Our analysis suggested that approximately 1% of the overall US population suffered from dysphonia on a yearly basis. Acute laryngitis accounted for slightly more than half of all cases, whereas more severe conditions were relatively rare (less than 200,000 per year). However, the growth of all conditions, except malignant neoplasm of the larynx, was significantly larger than that of the overall US population. The growth of all conditions except acute laryngitis can be due to overall population trends, such as aging. Growth in acute laryngitis, however, is unexplained and may be due to changes in the diagnostic practices within the medical community. Another consideration is that there is a growing recognition by both primary care and specialty providers that there may be a role for laryngopharyngeal reflux in both acute and chronic laryngeal disorders and in the edema of the larynx, which may play some role in the increases in these diagnoses. Industry representation was analyzed to further understand the burden of disease on patient and society overall. Approximately 40% of patients with acute laryngitis and 32% of patients with dysphonia were recorded as being full- or part-time employed. ARTICLE IN PRESS 4 Journal of Voice, Vol. ■■, No. ■■, 2017 FIGURE 2. Distribution of specific diagnoses. Nearly 30% of all patients with dysphonia were employed in the service industry, which may place higher demands on voice.11 The direct economic health impact of a medical condition is best estimated via claims data that incorporate all of the direct costs of evaluation and treatment. Although the purpose of this study was not to evaluate costs, and we did not evaluate all of the subsequent treatment costs, the direct healthcare diagnostic cost implications of dysphonia are substantial, nearing $1 billion a year, well above prior estimates utilizing the same database.5 The indirect cost impact of dysphonia would also be expected to be substantial, particularly if the more generous prevalence estimates from surveys are taken into account. This is the first in-depth evaluation of the industry distribution of patients with dysphonia and interesting relationships were noted. It is not unexpected that the service industry would have higher rates than the population as a whole, as service workers will frequently be in occupations that require significant voice use. Our study also identified a relatively high proportion of patients with malignant neoplasms of the larynx from the manufacturing industry. This may in part be due to higher ex- posures to carcinogenic chemicals12,13 and higher smoking rates in manufacturing than other industries.14 Our study shows that 3% of patients presented with benign neoplasms of the larynx. Laryngeal papillomas account for approximately 85% of benign laryngeal tumors, and although not directly recorded there is no reason to expect that this distribution would not be the case in our study. Study estimates of the prevalence of laryngeal papilloma have been largely restricted to pediatric patients. Utilizing the Optum Clinformatics and Truven MarketScan Medicaid databases, juvenile papilloma has been estimated to be 0.51 per 100,000 in the Optum database and 1.03 per 100,000 in the MarketScan Medicaid database.15 In this study, benign tumors are 1.3 times more likely in the service industry than the entire cohort. Because specifics related to the type of manufacturing or specific service industry are not identifiable in this database, and given the large number of patients represented, the US government labor specific data were used to estimate the possible relationships between diagnoses and specific occupations. Healthcare workers comprise 11.7% of the overall US labor market (http://www.bls.gov/ces/, FIGURE 3. 2008–2012 growth of prevalence of patients with dysphonia diagnoses. ARTICLE IN PRESS Michael S. Benninger, et al Prevalence and Occupation of Patients With Dysphonia 5 FIGURE 4. Average age in relationship to specific diagnosis. accessed December 14, 2015). If that distribution held true in our population, then 49% of the service occupation subjects in our trial would be in the healthcare sector. Based on this distribution, it could be estimated therefore that the risk of papilloma would be even greater in the healthcare industry, possibly as high as 2.65 times more likely than in the general worker population. Estimating prevalence rates for dysphonia is complicated and there is no perfect method to do so. Typically, prevalence rates for the global population rely on large claims databases, which provide information on a large enough group of patients to provide assumptions related to the larger population. One caveat in the use of claims data estimates of prevalence is that it relies on appropriate coding of the disorder. With improvements in coding over time and as more groups try to access these databases, it would be expected that data mining will provide better information. Another notable limitation is that claims databases only give information related to people who actually seek out healthcare for their problem. Other methods utilized to estimate prevalence of disease is through surveys, with larger surveys having more validity than smaller samples. Surveys will typically reveal higher prevalence rates, in that positive responses will occur in people who do not feel that they require a visit to the healthcare system. These are evident in the differences in prevalence rates of dysphonia identified in this study (1.7%) in comparison with surveys that have shown rates between 6% and 8%.3,8 Limitations of the present study include those inherent to all database analyses, such as reliance on proper physician coding. Additional limitations specific to this study include the fact that the projection methodology developed by MarketScan does not include uncertainty, and as such the results are presented as absolute values without error ranges. From an industry and FIGURE 5. Diagnostic procedures performed. ARTICLE IN PRESS 6 Journal of Voice, Vol. ■■, No. ■■, 2017 TABLE 3. Industry Affiliation for Patients With Industry Data MarketScan All dysphonia Acute laryngitis Benign neoplasm of larynx Chronic laryngitis Dysphonia Edema of larynx Laryngeal spasm Malignant neoplasm of larynx Manufacturing (%) Services (%) Transportation, Communications and Utilities (%) Finance, Insurance, Real Estate (%) Retail Trade (%) Other (%) 35 31 29 38 33 35 32 30 50 24 27 29 24 28 27 30 30 14 18 19 17 21 21 21 20 20 25 14 16 19 13 14 13 14 16 9 5 4 5 3 3 3 3 3 2 4 3 2 1 1 1 1 1 1 occupational standpoint, a key limitation is the fact that patients disclosed occupational or industry affiliation only in approximately 50%–60% of cases. However, despite the fact that not all patients provided this information, based on the signif- icant size of patients included in the study, industry and occupational information on 50%–60% patients can provide meaningful information in terms of overall burden of disease and potential associations. FIGURE 6. Industry representation in the MarketScan database. FIGURE 7. Differences in specific voice-related diagnoses in comparison with overall MarketScan population. ARTICLE IN PRESS Michael S. Benninger, et al Prevalence and Occupation of Patients With Dysphonia CONCLUSIONS The diagnosis of dysphonia is increasing with prevalence rates and current rates based on insurance claims evaluations being approximately 1.7% of the population. There is a difference in prevalence rates by age, with acute laryngitis being higher in younger age groups and malignant neoplasms in older ages. Prevalence rates also differ somewhat between industries, and there appears to be a higher percentage of malignant neoplasms in the manufacturing industry and benign neoplasms in the service industry. Overall prevalence rates are likely much higher, as many patients with voice problems do not seek medical care. REFERENCES 1. Benninger MS, Ahuja AS, Gardner G, et al. Assessing outcomes for dysphonic patients. J Voice. 1998;12:540–550. 2. Cohen SM, Kim J, Roy N, et al. Prevalence and causes of dysphonia in a large treatment-seeking population. Laryngoscope. 2012;122: 343–348. 3. Roy N, Merrill RM, Gray SD, et al. Voice disorders in the general population: prevalence, risk factors and occupational impact. Laryngoscope. 2005;115:1988–1995. 4. Cohen SM, Kim J, Roy N, et al. The impact of laryngeal disorders on work-related dysfunction. Laryngoscope. 2012;122:1589–1594. 5. Cohen SM, Kim J, Roy N, et al. Direct health care costs of laryngeal diseases and disorders. Laryngoscope. 2012;122:1582–1588. 7 6. Cohen SM, Kim J, Roy N, et al. Change in diagnosis and treatment following specialty voice evaluation: a national database analysis. Laryngoscope. 2015;125:1660–1666. 7. Cohen SM, Kim J, Roy N, et al. Delayed otolaryngology referral for voice disorders increases health care costs. Am J Med. 2015;128:426. 8. Bhattacharyya N, Orlandi RR, Grebner J, et al. Cost of burden of chronic rhinosinusitis: a claims-based study. Otolaryngol Head Neck Surg. 2011;144:440–445. 9. Benninger MS, Holy CE. The impact of endoscopic sinus surgery on health care use in patients with respiratory comorbidities. Otolaryngol Head Neck Surg. 2014;151:508–515. 10. The Agency for Healthcare research and Quality. Medical Expenditure Panel Survey, a US-based set of large-scale surveys of families funded by the US Department of Health and Human Services through the Agency for Healthcare Research and Quality (AHRQ). Available at: http://meps.ahrq .gov/mepsweb/. 11. Verdolini K, Ramig LO. Review: occupational risks for voice problems. Logoped Phoniatr Vocol. 2001;26:37–46. 12. Goldberg P, Leclerc A, Luce D, et al. Laryngeal and hypopharyngeal cancer and occupation: results of a case-control study. Occup Environ Med. 1997;54:477–482. 13. Rushton L, Bagga S, Bevan R, et al. Occupation and cancer in Britain. Br J Cancer. 2010;102:1428–1437. 14. Centers for Disease Control and Prevention. Current cigarette smoking prevalence among working adults—United States, 2004–2010. MMWR Morb Mortal Wkly Rep. 2011;60:1305–1309. 15. Marsico M, Mehta V, Chastek B, et al. Estimating the incidence and prevalence of juvenile-onset respiratory papillomatosis in publically and privately insured claims databases in the United States. Sex Transm Dis. 2014;41:300–305.