Abstract
Purpose: :
To estimate the association between self–reported visual impairment or blindness and annual medical care expenditures and health utility.
Methods: :
Data from Medical Expenditure Panel Survey are available from 1996–2002. Each subject was in the rotating panel study for two years. The data include (1) self–reported blindness, and visual impairment (difficulty reading a newspaper, seeing faces, or both); (2) annual medical care expenditures; and (3) health utility data (US algorithm applied to EuroQOL responses from 2000–2002). All analyses focused on adults aged 40 or older. Analyses used all years available as a series of cross–sectional data sets. Medical care expenditures were inflation–adjusted to 2002. A simple linear regression with annual expenditures as the dependent variable included only visual impairment and blindness; more complex regressions controlled for age, self–reported health status, race, education, income and insurance. Health utility data were analyzed similarly.
Analyses were run using survey commands in SAS 9 that estimate standard errors accounting the for the fact that an observation may be in the data for two years, the complex sampling design, and the weights.
Results: :
In simple analyses, the average blind individual had medical expenditures $5208 (p<0.001) higher than an individual reporting no visual impairment. The increment for individuals with visual impairment who are not blind was $3110 (p<0.001). When controlling for confounders the increases were attenuated but significant ($2567 and $1275 respectively, p unchanged). In simple analyses, the average blind individual had a decrement of 0.21 (p<0.001) health utility units compared with individuals with no self–reported visual impairments; individuals who were impaired but not blind had a decrement of 0.16 (p<0.001). In the regression controlling for confounders, the decrements were 0.11 and 0.08 respectively (p unchanged).
Conclusions: :
Self–reported blindness and visual impairment are associated with substantial increments in medical care expenditures and decrements in health utility, even when controlling for confounders. Utilities based on societal–preference algorithms applied to functional states reported by blind individuals lead to relatively small health utility differences between those with and without visual impairments.
Keywords: quality of life • clinical (human) or epidemiologic studies: health care delivery/economics/manpower