Lam BL, Lee DJ, Gómez-Marín O, Zheng DD, Caban AJ. Concurrent Visual and Hearing Impairment and Risk of MortalityThe National Health Interview Survey. Arch Ophthalmol. 2006;124(1):95-101. doi:10.1001/archopht.124.1.95
LESLIEHYMANPhDAuthor Affiliations: Bascom Palmer Eye Institute (Dr Lam), Department of Epidemiology and Public Health (Drs Lee and Gómez-Marín, Ms Zheng, and Mr Caban), and Department of Pediatrics (Dr Gómez-Marín), University of Miami School of Medicine, Miami, Fla.
To determine the association between reported concurrent visual and hearing impairment and risk of mortality.
Design, Setting, and Participants
Annual cross-sectional multistage area probability surveys of the US civilian noninstitutionalized population living at addressed dwellings were conducted by the National Center for Health Statistics, Hyattsville, Md. Mortality linkage with the National Death Index of participants from 1986 to 1994 was performed through 1997. Complete reported visual and hearing impairment data and survival status were available for 116 796 adults aged 18 years and older. A total of 3620 participants reported visual impairment only, 12 330 reported hearing impairment only, and 1461 reported concurrent visual and hearing impairment.
Main Outcome Measure
Risk of mortality.
Mortality linkage identified 8949 deaths with an average follow-up of 7.0 years. After controlling for survey design, age, marital status, educational level, self-rated health, and number of nonocular and nonauditory conditions, white participants and “other-race” participants, but not African American participants, reporting concurrent visual and hearing impairment had significantly increased risk of mortality in comparison with their counterparts reporting no impairment (white participants: hazard ratio [HR] = 1.23, 95% confidence interval [CI], 1.04-1.46 for men and HR = 1.63, 95% CI, 1.37-1.93 for women; African American participants: HR = 1.50, 95% CI, 0.94-2.40 for men and HR = 0.92, 95% CI, 0.51-1.63 for women; participants of other races: HR = 2.47, 95% CI, 1.33-4.57 for men and HR = 2.23, 95% CI, 1.01-4.90 for women). Risk of mortality was generally greater for participants reporting concurrent impairment as compared with that for participants reporting either visual impairment alone or hearing impairment alone.
In the United States, white persons and those of other races, but not African American persons, reporting concurrent visual and hearing impairment have an increased risk of mortality. Reported concurrent impairment is an independent predictor of mortality among white persons and those of other races for both men and women.
Visual and hearing impairment increases with age and occurs commonly among older persons. Of persons aged 70 years or older, 18% report blindness in 1 or both eyes or some other trouble seeing, 33% report hearing problems, and 9% report both visual and hearing problems.1 Despite notable improvements in the treatment of ocular and auditory diseases in the last 2 decades, there is no evidence that visual or hearing impairment rates reported by adults are declining.2,3 With the aging of the US population, the number of persons with multiple sensory impairments is increasing, and the morbidity and mortality associated with concurrent visual and hearing impairment are likely to become an increasing public health concern. For example, combined declines in vision and hearing are associated with decreased quality of life,4 imbalance,5 and falls.6 Functional status, measured by activities of daily living and instrumental activities of daily living scores, is diminished in elderly persons with concurrent visual and hearing loss as compared with those with single sensory impairment.7- 9 Furthermore, an increased risk of mortality among adults with eye diseases such as glaucoma, retinopathy, and cataracts has been described.10- 18 Visual impairment has also been found to be independently associated with reduced survival after controlling for the presence of a disabling eye disease.11,13,19 Information on the association between hearing impairment and increased risk of mortality is also available but limited.8,20,21 To date, only 1 large-scale study8has examined the association between concurrent visual and hearing impairment and the risk of mortality. In this study of 5444 adults aged 55 to 74 years from the 1971 to 1975 National Health and Nutrition Examination Survey after adjustment for demographics and chronic conditions, compared with those without impairment, the risk of mortality with a mean follow-up of 10.1 years was similar for participants with measured concurrent visual and hearing impairment (relative risk [RR] = 1.10, 95% confidence interval [CI], 0.64-1.89), for those with measured visual impairment only (RR = 1.20, 95% CI, 1.02-1.42), and for those with measured hearing impairment only (RR = 1.12, 95% CI, 0.88-1.44). In the same analysis, the risk of mortality was also similar for participants with self-reported concurrent visual and hearing impairment (RR = 1.08, 95% CI, 0.69-1.69), for those with self-reported visual impairment only (RR = 1.10, 95% CI, 0.95-1.27), and for those with self-reported hearing impairment only (RR = 0.86, 95% CI, 0.65-1.12). Even with a relatively large cohort (n = 5444), the number of participants with concurrent visual and hearing impairment at baseline was small, with only 36 participants having measured concurrent impairment and 64 participants having self-reported concurrent impairment.8
The purposes of the present study are to examine the association between reported concurrent visual and hearing impairment and risk of mortality in a large representative sample of the US noninstitutionalized adult civilian population and to compare this risk with that of reduced survival associated with reported visual impairment only or reported hearing impairment only.
The National Health Interview Survey (NHIS) is conducted annually by the National Center for Health Statistics (NCHS), Hyattsville, Md.22 The NHIS uses a complex sample survey design to obtain, in a cost-effective manner, population samples that are representative of the US civilian noninstitutionalized population. Households are selected annually from 192 of the approximately 1900 primary sampling units (primarily counties or a small number of contiguous counties) that cover the 50 states and the District of Columbia. Small clusters of households are drawn at random annually from these primary sampling units. A stratified approach is used to ensure an oversampling of African American persons and, in 1992, Hispanic persons.
Each year, nearly 50 000 households are selected to participate in the NHIS (a half sample of households was drawn in 1986 owing to budgetary restrictions). Information is obtained annually from approximately 120 000 family members using a primary household respondent, with all of the adults at home participating in the interview. The primary household respondent provides information for children residing in the household and for adults residing in the household who are not present at the interview. Throughout the survey years of 1986 to 1994, the response rate ranged between 94% and 98%.23
During the period from 1986 to 1994, information on visual impairment, hearing impairment, and selected ocular and auditory diseases was obtained annually in a randomly derived, one-sixth sample of households already selected for the NHIS. Participants were asked to name conditions and impairments that were related to the following: (1) activity limitation in the previous 2 weeks and the previous 12 months; (2) health care visits in the previous 2 weeks; and (3) hospital stays in the previous 12 months. A series of standardized questions was used to detail the name, characteristics, cause, onset, and effects of each described condition and impairment.24,25 Trained medical coders used this information to generate an International Classification of Diseases, 9th Revision (ICD-9) code for each condition.24,26 Except in 1986, the visual and hearing questions were administered to approximately 14 000 to 15 000 adults each year. In most cases, the participants themselves answered all of the questions (61%-63% across survey years), and for the remaining participants, the responses were obtained from their relatives or other proxies. For simplicity in the present study, self- or proxy-reported data are referred to as “reported.” Complete visual and hearing reported data and survival status were available for 116 796 adult participants of the NHIS in 1986 to 1994 who were aged 18 years or older at the time of the household interview. The average number of adults per household participating in the survey was 1.85.
Concurrent visual and hearing impairment was defined as reporting both visual impairment and hearing impairment. For visual impairment, participants were asked to indicate whether they or any family members had “blindness in 1 or both eyes” and “any other trouble seeing with 1 or both eyes even when wearing glasses.” Modified ICD-9 codes were used by the NCHS to classify visual impairment as follows: (1) blind in both eyes; (2) visual impairment in both eyes; (3) blind in one eye and visually impaired in the other; and (4) blind or visually impaired in one eye only, with good vision or vision not mentioned in the other. In the present analyses, participants falling into categories 1 through 4 were grouped into a single reported visual impairment category. For hearing impairment, participants were asked to indicate whether they or any of their family members were “deaf in 1 or both ears” or “had any other trouble hearing with 1 or both ears.” Modified ICD-9 codes were used by the NCHS to classify hearing impairment as follows: (1) deaf in both ears; (2) hearing impairment in 1 or both ears; (3) deaf in one ear and hearing impairment in the other; and (4) deaf or hearing impairment in 1 ear only (with good hearing or hearing not mentioned in the other ear).26,27 In the present analyses, participants falling into categories 1 through 4 were grouped into a single reported hearing impairment category.
Covariates included age, sex, race, marital status, educational status, self-rated health, and the number of nonvision- and nonhearing-related health conditions. Race was coded as African American, white, and other races (which included Aleut, Eskimo or American Indian, Asian or Pacific Islander, multiple races, and unknown or not specified). Marital status was coded as either married or other. Educational status was coded as less than the 12th grade, completion of high school (or General Education Diploma equivalent), and more than a 12th-grade education. Health status indicators included a general question assessing health status as excellent, very good, good, fair, or poor. In the NHIS, a condition count was generated for any acute condition, chronic condition, or impairment. In the present analysis, ICD-9 and impairment codes were used to exclude from these counts conditions associated with the visual and auditory systems (ICD-9 codes 360-368, 370-379, 380-387, 388.0, 388.1, 388.3-388.8, X00-X09). A 3-category nonvision and nonhearing health condition variable was then created as follows: (1) no conditions; (2) 1 condition; and (3) 2 or more conditions.
Information was collected by the NCHS to perform a mortality follow-up through linkage with the National Death Index. Information was obtained on the date and underlying cause of death. The latter was recoded and reported using ICD-9.26 The mortality linkage for participants of the 1986 to 1994 NHIS survey years is complete through December 31, 1997, with an average follow-up of 7.0 years.28 We used the binary variable (assumed alive, assumed dead) provided by NCHS to classify participants’ survival status. This classification was based on the matching algorithm developed by the NCHS that assigns a numerical value indicating the probability of a “true” match based on information recorded on the death certificate and collected during the survey (eg, social security number, full name, sex, birth month, birth year). This algorithm was developed and tested in previous NCHS mortality linkage studies; the NCHS estimates that use of the recommended cutoff values for the mortality linkage will correctly classify more than 97% of NHIS decedents and more than 99% of NHIS nondecedents.
All of the analyses were completed using the Software for the Statistical Analysis of Correlated Data package to account for the complex sample survey design of the NHIS.29 Software for the Statistical Analysis of Correlated Data takes into account that data from individuals participating in such survey designs are not independent since both clusters of households and clusters of participants are selected into the sample. The NCHS generated sample weights for each NHIS participant for use in all of the design-adjusted analyses. These weights take into account the probability of selection into the sample, household nonresponse to the interview, and poststratification adjustments designed to make the sample more representative of the US population. To account for the aggregation of data over multiple survey years, these sample weights were modified in the present analysis by dividing the original weight by 9, the number of years combined.30 To compare proportions and to assess associations between nominal variables, χ2 analyses were used. Cox regression analyses were performed using the Proc Survival procedure in Software for the Statistical Analysis of Correlated Data. Hazard ratio (HR) estimates and the corresponding 95% confidence intervals (CIs) are reported.
Table 1 provides frequency and prevalence rates of reported hearing, visual, and concurrent visual and hearing impairment among participants stratified by sociodemographic characteristics. Of the 116 796 adult participants, 3620 (3.1%) reported visual impairment only, 12 330 (10.6%) reported hearing impairment only, and 1461 (1.3%) reported concurrent visual and hearing impairment. Male, older, and less-educated participants were more likely to report impairment. Among racial groups, participants in the other-races group reported the least visual, hearing, and concurrent impairment, and African American participants reported less hearing impairment as compared with white participants.
Table 2 presents the sociodemographic characteristics, reported health status, reported number of nonocular and nonauditory conditions, and reported impairment status of study participants with and without available survival status information. Owing in part to the large sample size, all of the χ2 tests comparing the distribution of these measures by available survival status were statistically significant (Table 2). Participants who were older, not married, and in the other-races group were more likely to lack survival status information. The age and sex distributions were similar in the 2 survival status categories, as were the distributions of level of education and age-adjusted reported health status. Compared with participants without available survival status, participants with available survival status had slightly higher percentages of reported visual, hearing, and concurrent impairment.
Table 3 shows the mortality HRs and the corresponding 95% CIs stratified by race and sex subgroups for all of the self- and proxy-reported study participants. These race- and sex-specific analyses were necessary because prior analyses of risk of mortality revealed significant interactions of concurrent reported visual and hearing impairment with race as well as sex. After controlling for the complex sample survey design, age, marital status, educational level, self-rated health, and the number of nonocular and nonauditory conditions, the HRs for participants with concurrent reported visual and hearing impairment in comparison with those with no reported impairment remained significant for all of the white, other-race, and sex subgroups but not for African American men or women. The HRs for those reporting concurrent impairment were higher than for those with reported visual impairment only or hearing impairment only for men of all of the race subgroups and for white women.
Table 4 shows the mortality HRs and the corresponding 95% CIs stratified by race and sex subgroups including only the 63% of NHIS participants who were interviewed directly, excluding the proxy-reported participants. The HR estimates for the self-reported study participants only were similar and, generally, slightly lower when compared with estimates for all of the participants presented in Table 3. The differences between the HRs for all of the participants vs self-reported participants only were all less than 0.27, except for African American men. For example, the HRs for white men and white women with reported concurrent visual and hearing impairment interviewed directly were 1.16 and 1.55, respectively, and the HRs that included both self- and proxy-reported responses were 1.23 and 1.62, respectively (Table 3). The HR for African American men with reported concurrent visual and hearing impairment interviewed directly was 0.96, compared with an HR of 1.50 when including both self and proxy responses (Table 3); the 95% CIs of neither of these HRs were significant.
The results of this study indicate that reported concurrent visual and hearing impairment is an independent predictor of mortality for both white men and white women. This finding was present but not significant in African American men, and it does not hold for African American women. For white men, white women, and African American men, but not for African American women, the risk of mortality related to reported concurrent visual and hearing impairment is higher than the mortality risk associated with either reported visual impairment only or hearing impairment only. In comparison with white persons and African American men, African American women had the highest HR for mortality risk associated with reported visual impairment only and the lowest HR associated with reported hearing impairment only, but these HRs were not significantly different from those of white persons and African American men. The reasons for these findings are not clear. A stronger association between reported visual impairment and increased risk of all-cause mortality and cardiovascular disease mortality exists for women as compared with men.13 However, sex-specific mortality risk associated with reported hearing impairment has not been previously examined in a large-scale study to our knowledge. In the current study, compared with those with no reported visual or hearing impairment, white women with reported hearing impairment only had a slightly higher mortality risk than their male counterparts. In contrast, among African American persons, this sex difference was reversed: African American women with reported hearing impairment only had a slightly lower mortality risk than their male counterparts. Further research is needed to clarify the relationship between mortality risk and dual or single sensory impairment.
There are several advantages to the use of the NHIS. The NHIS is designed to be representative of the US population; only institutionalized and military groups have been omitted from direct sampling. Survey response rates were excellent (ie, 95%-98%), and determination of survival status was available for more than 96% of the participants. Aggregating data over 9 survey years allowed for the identification of more than 3700 participants with reported visual impairment, more than 12 800 with reported hearing impairment, and nearly 1500 with reported concurrent visual and hearing impairment, making this analysis the largest mortality study to date, to our knowledge, to examine associations between reported dual impairment and mortality.
Few previous studies have examined the associations between risk of mortality and concurrent visual and hearing impairment owing in part to the relatively large number of participants required for meaningful results. In the aforementioned National Health and Nutrition Examination Survey study of 5444 adults aged 55 to 74 years from 1971 to 1975, only 36 participants had clinical evidence of concurrent visual and hearing impairment.8 In another study of 1408 participants aged 65 years and older, the risk of mortality after 1 year, in comparison with those without impairment, was similar for participants with concurrent visual and hearing impairment (RR = 1.18, 95% CI, 0.54-2.60), for those with visual impairment only (RR = 1.34, 95% CI, 0.69-2.61), and for those with hearing impairment only (RR = 1.30, 95% CI, 0.56-3.03).31 This study also had low statistical power given that only 15 participants had concurrent impairment. Because of the lack of statistical power, sex- and race-specific analyses were not possible in these previous studies.
In the present analyses, the greatest risk of mortality associated with reported concurrent visual and hearing impairment was found in participants of the other-race subgroup. This mixed group consisted of Asian or Pacific Islander (49%), American Indian (15%), and multiple and other (36%) races. Men and women of other races with reported concurrent impairment had significantly higher risk of mortality as compared with those with no impairment (men: HR = 2.47, 95% CI, 1.33-4.57; women: HR = 2.23, 95% CI, 1.01-4.90). These findings suggest that further research in multiple sensory impairment should include races other than white and African American to determine why persons of other races with concurrent impairment have higher mortality risk.
Several mechanisms have been postulated to account for associations between sensory impairment and mortality. These mechanisms include (1) exposure to factors known to increase the risk of sensory impairment and major cause-specific deaths (eg, cardiovascular disease), (2) aging, and (3) impaired psychosocial functioning.
Numerous studies have found eye disease to be associated with reduced survival, even after controlling for cardiovascular and other risk factors.11,15- 19,32- 36 In the present analyses, significant reduction in survival is associated with reported concurrent visual and hearing impairment in white persons, even after controlling for all of the covariates, including the number of nonocular and nonauditory conditions. Taken together, these findings suggest that exposure to factors known to increase the risk of sensory impairment and cause-specific mortality cannot fully account for the associations between concurrent visual and hearing impairment and reduced survival.
Associations between sensory impairment and reduced survival as a reflection of poor health or accelerated aging have been postulated.10,33,35,37,38 Self-rated health,10 which is itself a predictor of reduced survival in numerous studies, including the NHIS,39,40 was included in the present multivariate statistical models as a partial control for overall health status. However, sensory impairment also contributes to the self-rated health assessment, and whether sensory impairment serves as a marker of generalized cellular aging is not certain.41
Associations between sensory impairment and mortality could be mediated by changes in psychosocial functioning. Visual impairment has been shown to increase the risk of impaired activities of daily living,36,42,43 depression,43,44 and social isolation.21 Hearing impairment is associated with increased social isolation45 and depression.46,47 These functional status indicators are themselves associated with either reduced overall survival or increased risk of cause-specific mortality.48- 50
The present study has several limitations. The NHIS does not annually assess important risk factors, including smoking status, which is associated with disabling eye conditions such as cataract. The NCHS has never published a study that validated reported chronic conditions against standardized, physician-confirmed diagnoses. Owing to the self- or proxy-reported nature of ascertainment of all of the ocular and auditory conditions in the NHIS, there is likely some misclassification of reported visual and hearing impairment. In a study of 570 persons aged 50 years and older, questions related to visual acuity were found to have a sensitivity of approximately 82% and a specificity ranging from 87% to 89%.51 However, visual acuity, contrast and glare sensitivity, stereoacuity, and visual fields are significant independent risk factors for self-reported visual disability in the elderly, and visual acuity alone is not the only dimension of subjective visual disability.52 Agreement between self-reported hearing loss and pure-tone air-conduction audiometry has a sensitivity ranging from 76% to 93% and a specificity ranging from 56% to 82%.53- 56
Differences between reported impairment and clinical measures may be owing in part to interpersonal variations in feelings and adaptations to physical disabilities. This interpersonal variation is taken into account in our analysis by controlling for self-rated health in our model because self-rated health itself is related to risk of mortality and reflects the physical and mental state of the person.39
Only a few standardized visual and hearing impairment questions were administered in the NHIS. However, respondents who reported impairment were then asked a series of questions to determine whether they had discussed the condition with their health care provider, whether their health care provider assigned a diagnosis or named the condition, and whether the impairment affected the left, right, or both sides. These questions were also asked whenever sensory conditions or impairment were mentioned earlier in the interview. Medical coders used this information to determine the level of impairment. Therefore, determination of impairment in the NHIS is unique when compared with other self-reported indices of sensory function in which single items or a standardized series of items are used to determine the presence and extent of impairment.
Validation studies conducted by the NCHS and others suggest that proxy reports lead to slightly lower prevalence estimates of chronic conditions when compared with reports obtained directly from respondents.57,58 In this study, the prevalence rates of concurrent visual and hearing impairment for proxy-reported participants were consistently lower than those for self-reported participants, with differences ranging from 0.1% to 1.5%. To address this potential limitation of misclassification of impairment, we repeated our mortality analyses including only the 63% of NHIS participants who were interviewed directly (Table 4). In general, the exclusion of proxy-reported participants decreased the mortality HRs slightly but did not alter the conclusion of this study. Further, the exclusion of proxy observations generally led to larger 95% CIs for the HR estimates and resulted in a sample less reflective of the US community-residing adult population than the NHIS was designed to obtain. These findings nevertheless suggest that the inclusion of proxy reports in the mortality analyses did not lead to systematic bias of the HR estimates for all of the NHIS participants (Table 3).
To summarize, after controlling for survey design, age, marital status, educational level, reported health status, and number of nonocular and nonauditory conditions, reported concurrent visual and hearing impairment is an independent predictor of mortality for both white men and white women and for men and women in the other-races category, but not for African American men or African American women. Furthermore, for white men and white women, the risk of mortality related to reported concurrent impairment is higher than the mortality risk associated with either reported visual impairment or hearing impairment only.
Correspondence: Byron L. Lam, MD, Bascom Palmer Eye Institute, 900 NW 17th St, Miami, FL 33136 (firstname.lastname@example.org).
Submitted for Publication: July 16, 2004; final revision received March 11, 2005; accepted April 29, 2005.
Financial Disclosure: None.
Funding/Support: This study was supported by grant AG021627 from the National Institute on Aging, Bethesda, Md.