The analyses described below are based on measures of the Medicare population and policies, commercial plan characteristics, and county attributes (Table 1). These are developed from three main data sources.
Medicare. The Health Care Financing Administration(now known as Centers for Medicare and Medicaid Services(CMS)) (HCFA(now known as CMS)) reports the number of Medicare beneficiaries, risk plan enrollment, service area, and plan participation data on monthly or quarterly bases through its web page. Only risk plans are included in the analysis, so that a plan that switches from a cost contract in 1998 to a risk contract in 1999 is considered a new plan. In general, numbers are used from late in 1998 (typically September) and early in 1999 (typically April) because of the project's focus on service area changes between 1998 and 1999. Although plans can describe their Medicare+Choice service areas to include only parts of counties based on zip code, most available data report numbers of eligible and enrolled beneficiaries by county, not zip code. As a result, county-level data are used, which may overstate the effect of plan service area reductions and new plan offerings.
HCFA(now known as CMS) also reports payment rates by county. Because of price differences by county, some analyses adjust these payment rates by a 1997 wage index. This is an imperfect measure of geographic price differences, but the resulting adjusted "real" payment rate is more comparable across areas than the nominal rate.
HCFA(now known as CMS) data were also used to create a county-level demographic index. Based on 1995 beneficiary data and demographic risk adjustment factors, this index describes differences in beneficiary populations with regard to age, sex, institutional status, working status, and participation in Medicaid.
Commercial Plans. Information about health maintenance organizations' (HMO) commercial enrollment and service areas is available from InterStudy's Competitive Edge. The July 1998 version was used in the analyses below, presumably describing HMOs at about the time they were deciding whether or not to continue (or join) Medicare managed care. Problems with data on commercial HMO products and enrollment have been documented by others (PPRC, 1997). Despite this, the InterStudy data have been widely used by health sector analysts and are considered the best source of standardized information about plans nationwide.
The InterStudy data report a number of plan characteristics. The analyses below use InterStudy as the source for plan enrollment level and growth, tax status, age, affiliation, participation in Medicaid and the Federal Employees Health Benefits Plan (FEHBP), numbers of physicians and hospitals, and service area. Changes in service area are based on comparisons between the InterStudy data from July 1998 and those from January 1995.
Counties. Geographic and demographic data about counties has been used from the Bureau of Health Professions' Area Resource File (ARF). The ARF has been used to describe local health markets (admission rates, beds, physicians) and local economic conditions (unemployment and median incomes (again adjusted for geographic variation in wages)).
Medicare+Choice Participation. The focus on describing plans by whether or not they participated in Medicare+Choice requires merging data from HCFA(now known as CMS) and InterStudy. Some Medicare+Choice plans could not be matched with a commercial plan described by InterStudy. For the purposes of some of the analyses, plans were considered to be in Medicare+Choice only if a match could be made between the HCA and InterStudy data (Figure 1). Not all commercial plans that matched a Medicare+Choice plan reported having Medicare enrollees, nor did all plans that reported Medicare enrollees match a Medicare+Choice plan. The lack of reported Medicare enrollment by some plans that were labeled as Medicare+Choice does not raise serious concerns of erroneously calling plans Medicare+Choice participants when they are not, because the match with HCFA(now known as CMS) data on Medicare+Choice plans was made based on compelling correspondence between plan name, address, and other identifying information. Conversely, since some Medicare+Choice plans could not be matched to InterStudy plans, some plans are incorrectly marked as not being in Medicare+Choice when in fact they are. This will presumably downward bias measured differences between the two groups. This understatement of the number of Medicare+Choice plans and the counties they serve complicates comparisons with information reported by HCFA(now known as CMS).
Similarly, for plan-county level analyses, a county was only considered in Medicare+Choice if it was listed in the commercial service area by InterStudy and the Medicare+Choice service area by HCFA(now known as CMS). In many cases, the Medicare+Choice service area included counties not listed as part of plans' commercial areas, but for these analyses these counties were not included. The plan-county pairs are therefore based solely on the commercial service area, with individual plan-county observations considered to be in Medicare+Choice only if the plan matched a Medicare+Choice plan and the county was in the Medicare+Choice service area, as described by HCFA(now known as CMS). This mismatch between commercial and Medicare+Choice service areas merits additional analysis.