The national underutilization of eyecare services warrants granular analysis
There was no significant correlation between the usage of services and market saturation
Usage was significantly correlated with individual and contextual factors
Focus on increasing the density of providers may not increase realized access
To characterize usage of ophthalmologic services by Medicare Fee-For-Service (FFS) beneficiaries relative to geography-specific market saturation, demographics, and contextual factors
Data sets from Centers for Medicare & Medicaid Services, US Census Bureau, US Department of Agriculture, and Housing and Urban Development, were used to calculate county- and state-level ophthalmologic service usage, market saturation, and demographic characteristics. Negative binomial regression models were used to evaluate the association between results and demographic or population-specific variables.
Ophthalmologic service usage ranged from 58.2% to 15.2%, whereas saturation ranged from 21,763 to 91.4 FFS beneficiaries per registered ophthalmologist. Usage was significantly associated with demographic characteristics in each geography: lower proportion of African American ( P = .009), Hispanic ( P < .001), and other race beneficiaries ( P < .001), relative to white beneficiaries; a higher proportion of female ( P < .001) relative to male; a higher proportion of adults having completed an associate degree or some college ( P = .001), or holding a bachelor’s degree or higher ( P < .001), relative to a high school diploma; a lower proportion of adults in each geography experiencing poverty ( P = .009), geographies with lower Multidimensional Deprivation Index ( P < .001); a higher urban-influence code ( P < .001).
There was no significant correlation between the usage of ophthalmologic services and the geographic market saturation of ophthalmologists (Spearman rho, –0.030, P = .227).
Conclusions and Relevance
Ophthalmologic service usage is significantly influenced by population demographics; however, increased provider density alone appears insufficient to promote the usage of eye care services.
T here is evidence for underutilization of eye care services in the United States. Only 49.6% to 58.3% of individuals with visual impairment report using eye care in the past year (including all eye care providers). The majority of people with diabetes likely fail to receive necessary screening. Among Medicare beneficiaries with health insurance coverage, rates of eye examinations are inadequate, with approximately 73% of at-risk Medicare beneficiaries failing to receive regular eye examinations. A reasonable response to this problem would be increasing provider density in areas with suboptimal levels of eye care service utilization. However, that increased availability of ophthalmologists may not automatically result in increased usage if other barriers exist, including an understanding of the importance of routine eye care, as well as individual and contextual factors.
Local density of providers is a complex determinant of health care availability and access. The temporal geographic distribution of ophthalmologists and optometrists in the United States has been explored as early as the 1970s. More recent survey data from 1995 to 2017 have shown a decrease in the national density of ophthalmologists with a persistent rural/urban disparity in availability and an aging workforce. An aging workforce may impact the supply of providers if there exists a net decrease due to retirement. The density of providers is important considering prior work showing the availability of ophthalmologists plays a role in accessing eye care as well as visual health outcomes.
Although provider availability is a critical public health metric, there are a multitude of characteristics of patients and their environment that influence access and use of health care services. Demographic and socioeconomic differences in utilization of ophthalmology services can be related to race/ethnicity, income, and education. To account for these complex factors, Andersen and Davidson’s behavioral health model categorizes characteristics that influence health behaviors and outcomes as individual or contextual. Contextual characteristics are measured at some aggregate rather than individual level, such as at the neighborhood, city, county, or state level. Both categories can be subcategorized as predisposing, enabling, and need factors. For example, contextual predisposing characteristics relate to demographic composition of a community (eg, educational level, racial/ethnic composition); contextual enabling characteristics may relate to health policy, financing, or density of resources (eg, distribution of health services facilities and personnel, per capita community income, insurance coverage); and contextual need characteristics related to the physical environment (eg, quality of housing, water, air). These characteristics influence health care usage and outcomes. For example, the relationship between an individual’s race/ethnicity and her or his routine preventive service use is influenced by the county-level racial/ethnic composition. These findings extend to ophthalmology, where county-level contextual factors were found to influence eye care use independent of individual characteristics. These trends in potential access and utilization are critical to explore given persistent racial and socioeconomic disparities in visual impairment.
Better understanding provider density and patient barriers to access can require the use of databases of differing methodology to compare data on physician location, patient demographics, and usage. Deciphering the specific contribution of ophthalmologist density to access of vision services is critical to closing gaps in eye care coverage. Health policy attempting to increase the density of providers in certain locales may miss their target if this is not the root cause of access issues. Therefore, the present study sought to explore the factors associated with ophthalmologic service usage among Medicare beneficiaries, the geographic density of ophthalmologists serving those beneficiaries, and the potential correlation between usage and saturation.
This study did not qualify as human subject research and thus did not require institutional review board approval. This study adhered to the tenets of the Declaration of Helsinki.
Market Saturation Data Set
The Medicare market saturation data set was downloaded from the Centers for Medicare and Medicaid Services (CMS) and filtered for “ophthalmology” in calendar year 2018. From this data set, “Total Payments,” “Total Payment per user,” “Total Payment per beneficiary,” and “Users” were sourced. “Users” were defined as the subset of Fee-For-Service (FFS) beneficiaries who have a paid claim for an ophthalmology service. Ophthalmology services were defined by the CMS specialty code on the noninstitutional code (specialty code 18 or 41), with claim type 71 and 72 for Medicare Part B claims. “Usage” of services was calculated as the total number of health care users divided by the total number of FFS beneficiaries in a given geography. Geographic saturation of ophthalmology providers was reconstructed from an alternate database using provider addresses.
Specifically, similar to prior studies, ophthalmologist information was sourced from the Physician Compare National Downloadable File, which provides information about individual eligible professionals (EPs). This file was accessed August 12, 2020, and is updated twice monthly. The providers were filtered to include the primary specialty of ophthalmology. Because of incomplete data across data sets, providers in Puerto Rico/Guam/Virgin Islands were excluded from analysis. The zip codes of providers were converted to county-level Federal Information Processing Standard (FIPS) codes using Housing and Urban Development’s ZIP Code Crosswalk Files (HUD-USPS ZIP-COUNTY crosswalk for 2nd quarter 2020). Briefly, for ZIP codes corresponding to multiple counties, the county code was assigned to the county with the highest total ratio of addresses in that zip code to the total addresses in the entire county. Within each county code, a count of providers was calculated from the number of unique National Provider Identifiers (NPI).