Reply

Abstract

We sincerely thank Dr Gibson for her comments on our recent article concerning access to eye care providers.1Lee C.S. Morris A. Van Gelder R.N. et al.Evaluating access to eye care in the contiguous United States by calculated driving time in the United States Medicare population.Ophthalmology. 2016; 123: 2456-2461Abstract Full Text Full Text PDF PubMed Scopus (31) Google Scholar Dr Gibson notes that we have used a novel metric—shortest driving distance—as a proxy for access to care, and that comparisons of this metric with existing methodologies, such as county-level provider data studied by her own group, are difficult. We concur with this assessment, but wish to highlight ways in which we feel the current metric may differ from county-level provider data. We note that studies in many other fields of medicine have used driving distance as a measure of access to care.2Raykar N.P. Bowder A.N. Liu C. et al.Geospatial mapping to estimate timely access to surgical care in nine low-income and middle-income countries.Lancet. 2015; : S16Abstract Full Text Full Text PDF PubMed Google Scholar, 3Strauss K. MacLean C. Troy A. et al.Driving distance as a barrier to glycemic control in diabetes.J Gen Intern Med. 2006; 21: 378-380Crossref PubMed Scopus (73) Google Scholar As noted in our paper, driving distance estimated by network road analysis is superior to straight line and centroids of polygons.4Boscoe F.P. Henry K.A. Zdeb M.S. A nationwide comparison of driving distance versus straight-line distance to hospitals.Prof Geogr. 2012; 64Crossref PubMed Scopus (179) Google Scholar From a policy perspective, commercial insurance providers such as Blue Cross Blue Shield and the US Veteran's Administration Choice Program use driving distance to health care providers for defining their access to care standards. Use of county-level and driving distance metrics are likely complementary. We note that, in Figure 1 of Dr Gibson's group's work on this subject,5Gibson D.M. The geographic distribution of eye care providers in the United States: implications for a national strategy to improve vision health.Prev Med. 2015; 73: 30-36Crossref PubMed Scopus (38) Google Scholar there are many large areas in the United States with the lowest quartile availability of ophthalmologists immediately bordering an area with highest quartile availability, specifically in the rural regions such as Idaho, Montana, Wyoming, North Dakota, and South Dakota. Because their analysis compared the absolute percentages of eye provider availability per county, the results would underestimate the access to ophthalmologists in these rural areas. In contrast, driving distance weighted by the number of people living in a particular location would not be subject to these limitations. We agree with Dr Gibson's assertion that the census block groups can be large in rural regions. Although this is true, the US Census defined block groups as 2 levels smaller than a county and to be statistical divisions of census tracts to contain between 600 and 3000 people. Because our analysis is weighted by the number of people living in each geographical unit, the bias introduced is minimized. Compared to the counties, there are 2 more orders of magnitude of census block groups, and census block groups provide a much more detailed geospatial analysis. Although Dr Gibson notes in her letter that there are census block groups as large as 50 square miles, the largest single county in the United States spans >20 000 square miles and contains >2 million people according to the US Census. The average US county size is 996 land square miles compared with 16 square miles for a US census block group. It is also worth noting that a square census block group of 50 square miles has a longest transverse distance of 10 miles. In addition, we agree that driving time analysis of Medicare's beneficiaries' addresses relative to the nearest eye care provider would be superior; however, no complete dataset exists as released by the Centers for Medicare and Medicaid Services and so this analysis cannot be completed as suggested. A future possible analysis would be to repeat our analysis with census blocks instead of census block groups in rural regions of the United States. Furthermore, Dr Gibson suggested that we could have used our data to “create a measure of eye care provider capacity by calculating for each US Medicare beneficiary how many other Medicare beneficiaries had the same closest provider.” We regret that the suggested analyses cannot be completed given that our dataset does not contain patient-level data. The Centers for Medicare and Medicaid Services does not provide a complete dataset at the patient level, and with the available 5% or 20% patient-level Medicare data, certain rural areas may be represented incompletely in the database. Our analysis, although having limitations, is as comprehensive as possible given the available data. In summary, we agree that there is no single best measure of access to care, and that each current method has caveats and limitations. For the reasons stated in our paper as well as those enumerated herein, we believe our findings using driving time as an access metric provide a valid perspective in understanding access to eye care in the United States. Re: Lee et al.: Evaluating access to eye care in the contiguous United States by calculated driving time in the United States Medicare population (Ophthalmology. 2016;123:2456-2461)OphthalmologyVol. 124Issue 8PreviewLee et al1 state, “Our findings contrast with the results of Gibson,2 who analyzed 3143 counties in the United States and found that 24.1% of the counties were in the lower 2 quartiles of ophthalmologist availability but in the upper 2 quartiles of optometrist availability.” Gibson2 created population-weighted quartiles of the county-level number of ophthalmologists per capita and the county-level number of optometrists per capita. Lee et al1 maintain that “this method of analysis grossly underestimates availability.” This letter argues that Lee et al1 did not support these assertions and notes that it has not been established that driving time to the closest eye care provider, the measure of provider availability used by Lee et al,1 is a better predictor of vision health outcomes than other measures of the availability of eye care providers. Full-Text PDF