2. Methods and Data (continued)

2.6 Selection of a Sample of 1.96 Million Medicare Beneficiaries from the EDB

We selected a nationally representative stratified, simple, random sample of traditional fee-for-service Medicare beneficiaries designed to over-sample minorities to create and validate an SES index (discussed in the next section of this report), prepare tabulations (discussed in the following section), and conduct multivariate analyses (discussed in the final section).This sample was initially drawn for analyses performed in the previous task order contract and reported on in the report titled Health Disparities: Measuring Health Care Use and Access for Racial/Ethnic Populations (2005).

We drew the sample from the full 10 segments of the mid-2003 unloaded EDB. To be eligible for inclusion in the sample, beneficiaries must have been enrolled in traditional fee-for-service (FFS) Medicare (Part A, Part B, or both) for the full 12 months of the 2002 calendar year and not have been enrolled in a Group Health Organization at all during the 2002 calendar year. In addition, beneficiaries must have been alive for the full 12 months of calendar year 2002. We set these criteria to allow the maximum opportunity (period of time) for beneficiaries to submit claims documenting their use of preventive and other Medicare covered services.

Table 2.12 presents a distribution of the beneficiaries included on the EDB eligible for the sample by their NEWRACE (their race/ethnicity code resulting from the use of the naming algorithm created by RTI) and by EDBRACE (their race/ethnicity code on the EDB before using the naming algorithm), respectively.

The primary sampling goal at the time this sample was selected was to have sufficient sample size to provide equally accurate and precise estimates of health care utilization for the different racial/ethnic groups. We therefore stratified by race/ethnicity and sampled such that, to the extent possible, the same number of Medicare beneficiaries would be included in the sample in each of the different racial/ethnic groups. This meant that the sample included disproportionately more minorities than their representation in Medicare. The sampling rates based on the NEWRACE code was 11 percent for Black Medicare beneficiaries, 1.2 percent for White, 26 percent for Hispanic, 71 percent for Asian/Pacific Islander, and 100 percent of American Indian/Alaska Native, Other, and Unknown.

Table 2.12 Distribution of Medicare fee-for-service beneficiaries believed eligible for the study sample by NEWRACE and EDBRACE

Race/Ethnicity Category	Frequency of NEWRACE	Frequency of EDBRACE
Non-Hispanic White	25,907,883	27,091,613
Non-Hispanic Black	3,025,397	3,087,034
Hispanic	2,081,123	730,147
Non-Hispanic Asian/Pacific Islander	592,010	453,950
Non-Hispanic American Indian/Alaska Native	121,024	122,156
Non-Hispanic Other	183,242	412,198
Unknown	61,567	75,148
Total	31,972,246	31,972,246

Source: EDBRACE is from the Medicare EDB from mid-2003 and NEWRACE is the result of running the algorithm on those same beneficiaries from the Medicare EDB from mid-2003.

These sampling rates produced the sample presented in Table 2.13 according to NEWRACE and EDBRACE.

Table 2.13 Distribution of Medicare fee-for-service beneficiaries selected for the study sample by NEWRACE and EDBRACE

Race/Ethnicity Category	Frequency of NEWRACE	Frequency of EDBRACE
Non-Hispanic White	333,334	658,279
Non-Hispanic Black	333,334	350,879
Hispanic	545,643	191,402
Non-Hispanic Asian/Pacific Islander	421,859	312,785
Non-Hispanic American Indian/Alaska Native	121,024	121,496
Non-Hispanic Other	183,242	299,015
Unknown	61,567	66,147
Total	2,000,003	2,000,003

Source: EDBRACE is from the Medicare EDB from mid-2003 and NEWRACE is the result of running the algorithm on those same beneficiaries from the Medicare EDB from mid-2003.

White and Black beneficiaries have the same size sample, while American Indian/Alaska Native, Other, and Unknown have fewer, and Hispanics and Asians/Pacific Islanders have slightly more. As already mentioned above, our sampling goals were to have as close to equal sample for each race/ethnicity as possible in order to provide equally accurate and precise estimates of utilization for each of the racial/ethnic groups. Thus, we sampled 100 percent of the American Indian/Alaska Native, Other, and Unknown beneficiaries and therefore could not get them any closer to the White and Black numbers. In the case of Hispanic and Asian/Pacific Islanders beneficiaries, however, we increased their allocation in the sample slightly because we wanted the estimates to be as close to equally precise across the race/ethnicity categories when the NEWRACE code was used rather than the EDBRACE. (The number of Asian/Pacific Islanders and Hispanics would have dropped whenever the EDBRACE code was used instead of the NEWRACE and so would the precision of the estimates; therefore, we decided to sample more Asian/Pacific Islanders and Hispanics to maintain the precision of the estimates).

After the sample was selected, we cross-referenced it with CMS's denominator file as a final check on eligibility and discovered that slightly less than two percent of the selected sample, spread proportionately across the racial/ethnic groups, did not meet all of our desired sample eligibility criteria – alive and enrolled in fee-for-service Medicare for the entire year. We identified and discarded the ineligibles from the sample and recalculated the weights of the remaining sample to correctly represent the intended population of eligible Medicare fee-for-service beneficiaries. Table 2.14 presents the final number of sampled beneficiaries by NEWRACE and EDBRACE.

Table 2.14 Distribution of final study sample of selected Medicare fee-for-service beneficiaries by NEWRACE and EDBRACE

Race/Ethnicity Category	Frequency of NEWRACE	Frequency of EDBRACE
Non-Hispanic White	329,954	647,653
Non-Hispanic Black	328,246	345,559
Hispanic	534,196	187,920
Non-Hispanic Asian/Pacific Islander	415,190	308,890
Non-Hispanic American Indian/Alaska Native	120,557	121,025
Non-Hispanic Other	171,032	283,603
Unknown	60,946	65,471
Total	1,960,121	1,960,121

Source: EDBRACE is from the Medicare EDB from mid-2003 and NEWRACE is the result of running the algorithm on those same beneficiaries from the Medicare EDB from mid-2003.

Table 2.15 contains the distribution of the weighted number of Medicare beneficiaries contained in the EDB who are represented in the sample distributed by the NEWRACE and EDBRACE variables.

Table 2.15 Distribution of weighted study sample of selected Medicare beneficiaries by NEWRACE and EDBRACE

Race/Ethnicity Category	Frequency of NEWRACE	Frequency of EDBRACE
Non-Hispanic White	25,645,178	26,779,400
Non-Hispanic Black	2,979,217	3,053,618
Hispanic	2,037,463	720,664
Non-Hispanic Asian/Pacific Islander	582,651	449,914
Non-Hispanic American Indian/Alaska Native	120,557	121,818
Non-Hispanic Other	171,032	397,030
Unknown	60,946	74,600
Total	31,597,044	31,597,044

Source: EDBRACE is from the Medicare EDB from mid-2003 and NEWRACE is the result of running the algorithm on those same beneficiaries from the Medicare EDB from mid-2003.

Return to Contents

2.7 Identifying the Top 10 Metropolitan Statistical Areas (MSAs) for Asian/Pacific Islander and Hispanic Elderly Population

In addition to the tabular investigations to be performed of racial and ethnic disparities in health care utilization by Medicare beneficiaries at the national level, this task was intended to provide additional insight into racial and ethnic disparities in smaller areas with high concentrations of elderly Asians and Hispanics. Rather than target states with large Asian and Hispanic residents, the task called for us to identify metropolitan statistical areas (MSAs) in which there would potentially be large numbers and high concentrations of Asian/Pacific Islander and Hispanic Medicare beneficiaries. The purpose for looking at utilization in selected MSAs was to investigate whether, in areas where members of these minorities were a significant segment of the health care market, there were fewer disparities in services used and greater similarities in health care use to that of non-Hispanic Whites. It is important to keep in mind that in such analyses, the number of Medicare beneficiaries in the sample and the frequency with which the services were used for the MSAs studied had to be adequately large to achieve statistically reliable estimates.

To investigate this issue, we identified the ten MSAs, or the Primary MSAs within Consolidated MSAs, from the 2000 US Census with the largest number of persons 65 years of age and older who identified themselves in the Census as Asian, Native Hawaiian, or Pacific Islanders. The MSAs we identified are listed in Table 2.16 for Asians/Native Hawaiians /Pacific Islanders. We did the same for persons who were identified as Hispanic or Latino. Those MSAs are identified and listed in Table 2.17 for Hispanics/Latinos. The MSAs are listed according to the number of elderly Asian/Native Hawaiians/Pacific Islander and Hispanic/Latino population in each.

When we prepared the tabulations for this report, it became apparent that the two sets of 10 MSAs reduced to a single set of 16, because four MSAs were common to both lists. The MSAs common to both lists are Los Angeles, New York, Chicago, and San Diego.

Table 2.16 Top 10 MSAs for Number of Asian/Native Hawaiian/Pacific Islander Population 65 Years of Age and Older: 2000

Metropolitan Statistical Area	Total population: Total	Total population: Asian/NH/PI	Percent of total population who are Asian/NH/PI	Total population 65+	Asian/NH/PI population, 65+	Percent of all 65+ persons who are Asian/NH/PI
Los Angeles--Long Beach, CA PMSA	9,519,338	1,164,553	12.2%	926,673	120,811	13.0%
Honolulu, HI MSA	876,156	481,051	54.9%	117,737	85,429	72.6%
New York, NY PMSA	9,314,235	851,460	9.1%	1,109,821	62,887	5.7%
San Francisco, CA PMSA	1,731,183	406,087	23.5%	227,628	53,814	23.6%
Oakland, CA PMSA	2,392,557	411,819	17.2%	254,863	36,265	14.2%
San Jose, CA PMSA	1,682,585	435,868	25.9%	160,527	31,926	19.9%
Orange County, CA PMSA	2,846,289	395,723	13.9%	280,763	30,181	10.7%
Chicago, IL PMSA	8,272,768	384,932	4.7%	888,505	24,964	2.8%
San Diego, CA MSA	2,813,833	263,363	9.4%	313,750	22,593	7.2%
Washington, DC--MD--VA--WV PMSA	4,923,153	332,919	6.8%	446,288	20,703	4.6%
Total in 10 MSAs	44,372,097	5,127,775	11.6%	4,726,555	489,573	10.4%

Source: U.S. Census Bureau, Census 2000 Summary File 1 (SF 1) 100-Percent Data, Table P12. Available online at: http://factfinder.census.gov/servlet/DTGeoSearchByListServlet?ds_name=DEC_2000_SF1_U&_lang=en&_ts=164553925612

Table 2.17 Top 10 MSAs for Number of Hispanic/Latino Population 65 Years of Age and Older: 2000

Metropolitan Statistical Area	Total population: Total	Total population: Hispanic or Latino	Percent of total population who are Hispanic or Latino	Total population 65+	Asian/NH/PI population, 65+	Percent of all 65+ population who are Hispanic or Latino
Los Angeles--Long Beach, CA PMSA	9,519,338	4,242,213	44.6%	926,673	187,447	20.2%
Miami, FL PMSA	2,253,362	1,291,737	57.3%	300,552	184,625	61.4%
New York, NY PMSA	9,314,235	2,339,836	25.1%	1,109,821	146,219	13.2%
San Antonio, TX MSA	1,592,383	816,037	51.2%	169,748	59,486	35.0%
Riverside--San Bernardino, CA PMSA	3,254,821	1,228,962	37.8%	342,423	49,499	14.5%
Chicago, IL PMSA	8,272,768	1,416,584	17.1%	888,505	44,620	5.0%
El Paso, TX PMSA	679,622	531,654	78.2%	66,073	43,210	65.4%
Houston, TX, PMSA	4,177,646	1,248,586	29.9%	311,213	37,968	12.2%
San Diego, CA PMSA	2,813,833	750,965	26.7%	313,750	34,149	10.9%
McAllen-Edinburg-Mission, TX MSA	2,813,833	750,965	26.7%	313,750	34,149	10.9%
Total	42,447,471	14,369,674	33.9%	4,484,032	820,070	18.3%

Source: U.S. Census Bureau, Census 2000 Summary File 1 (SF 1) 100-Percent Data, Table P12. Available online at: http://factfinder.census.gov/servlet/DTGeoSearchByListServlet?ds_name=DEC_2000_SF1_U&_lang=en&_ts=164553925612

Return to Contents
Proceed to Next Section