Printer Friendly

Factors affecting interstate use of inpatient care by Medicare beneficiaries.

This article examines the extent to which interstate inflow and outflow of patients affects their observed use of Medicare Part A inpatient care. Interstate patient flow can bias utilization rates and may be due to seasonal migration, interstate inpatient care market areas, or purposive seeking of specialized/high-quality care. Examination of state level patient flow data drawn from 1987 Medicare discharge indicate that most interstate patient flow occurs between adjacent states probably as an outgrowth of interstate markets. Regression analyses of patient flow data suggest that while seasonal migration is an important determinant of patient flow, its importance is secondary to that of indicators of the availability of specialized services. These findings suggest research questions that may be best answered in detailed analyses of inpatient utilization in interstate market areas and seasonal migration.

Medicare Inpatient Care

Use and Interstate

Patient Flow

As the elderly in the United States increase in number and become more mobile, Medicare beneficiaries are increasingly likely to obtain Medicare Part A covered inpatient care in a state other than their state of residence. The interstate flow of Medicare hospitalizations may be stimulated by the decreasing availability of inpatient care in rural areas, the localized supply of specialized services, geographical variation in the cost of inpatient care, and seasonal migration.

Interstate patient flow concerns health planners for two reasons. First, patient flow may be an indicator of inadequate services. Second, interstate patient flow, especially if sizable, should be accounted for in estimates of future service utilization. Often it is not, and bias in discharge rates may result.(1) This is especially important when using historical utilization and cost data to determine reimbursement rates for capitated care, since interstate patient flow can distort utilization rates in areas where interstate flows are frequent.

The impact of "snowbirds" on health care utilization rates for elderly persons is frequently of concern, although other factors may lead elderly persons to obtain care outside of their state of residence. Previous research using Medicare inpatient data found that hospitalizations for nonresident beneficiaries comprised over 10 percent of the Medicare inpatient caseload in several states and that many Medicare beneficiaries were receiving inpatient care in states that were not proximate to their state of residence (Lubitz, Deacon, and Walton 1979). That study, however, did not explore the determinants of interstate flow of Medicare inpatients.

Several recent studies have used geographic models to examine the movement of hospital inpatients (for the total population) across small areas (Harner and Slater 1980; Folland 1983; Mayer 1983; McGuirk and Porell 1984; Cohen and Lee 1985; Luft, Garnick, Mark, et al. 1990). Most of these studies have not assessed the influence of factors other than distance on utilization and either have not included out-of-state patients or hospitals in their analysis or have been unclear about whether interstate patient flows are or are not included in their studies.

Cohen and Lee (1985) augmented their gravity model with data on physician supply, bed supply, and hospital characteristics, and found that these factors were significant predictors of hospital utilization. Their analysis points out the limitations of gravity models and suggests that supply factors must be considered when explaining patient mobility.

Luft, Garnick, Mark, et al. (1990) employed a multinomial choice gravity model which also included data on patient characteristics to determine the factors influencing choice of hospital in three urban California market areas for 12 types of conditions. This study found that distance, charges, ownership, and quality of care proxies affected hospital choice. One of the quality proxies used was number of discharges for out-of-state patients. However, the flow of patients to out-of-state hospitals was not considered since (unlike Griffith et al. 1985) these discharges were not represented in the California state discharge data used in this study.

Patient flow is also an issue in studies of hospital competition. Both Morrisey, Sloan, and Valvona (1989) and Garnick et al. (1987) have employed criteria developed by Elzinga and Hogarty (1973) to define market areas as having little entering patient flow from outside (LIFO) and little patient flow out from inside the market area (LOFI). Both groups of authors suggest that the ideal percentage of patients entering or exiting the market area be 10 percent or less, but they relax this standard in their studies to 25 percent or, for some urban markets, 40 percent.

Morrisey, Sloan, and Valvona (1989) found significant interstate flow of Medicare discharges in three of the four inpatient care markets they examined. Possible explanations for interstate inpatient flow varied across market areas. These were: availability of care in rural areas (Omaha, Nebraska), a multistate metropolitan market (Philadelphia, Pennsylvania) and seasonal migration (Phoenix, Arizona). That study suggests that interstate patient flow may be due to three factors: availability and quality of services, interstate hospital market areas, and seasonal migration; and it points out the need to use multistate data to analyze interstate inflow and outflow of patients in order to avoid the shortcomings inherent in using discharge data for only one state (as evidenced in Garnick, Lichtenberg, Phibbs, et al. 1989; and Luft, Garnick, Mark, et al. 1990). Since interstate patient flow for seasonal migrants will generally occur across nonadjacent states, market area-based studies are often inadequate for enumerating these patient flows.

Although the findings of Morrisey et al. just reviewed suggest other reasons for interstate patient flow, conventional wisdom often attributes the interstate flow of elderly inpatients to seasonal migration. The impact of seasonal migration on the use of health services has not been examined in detail largely because of the scarcity of data sources that can be used to define a population of seasonal migrants. Most studies of seasonal migration have been based on convenient populations not selected systematically. Aggregate estimates of seasonal migrants from Census of Population data have only become available relatively recently and have been limited to state level estimates (Hogan 1987).

Elderly seasonal migrants, like elderly permanent migrants, are above average in income, education, and health status, and migrate to the same areas of the country. Most U.S. seasonal migrants travel to Florida or Arizona (Hogan 1987; Atchley 1991). The choice of migration destination appears to be affected by residence: Florida for those east of the Mississippi River and Arizona for those west of the Mississippi. Monahan and Greene (1982) found that elderly seasonal migrants in Tucson, Arizona were in slightly better health and had significantly lower use of inpatient, physician, and emergency room care than elderly permanent residents.

Data Sources

In order to account for all possible interstate patient flow patterns, a nationwide discharge database must be used. One of the few inpatient databases suitable for nationwide patient flow research is the Medicare Provider Analysis Review (MEDPAR) database maintained by the Health Care Financing Administration. MEDPAR data is composed of all inpatient discharges for Medicare beneficiaries (including those for which a bill was submitted but no payment was made).

Two of the dependent variables used in this analysis (percent of discharges for nonresident Medicare beneficiaries in the state Medicare inpatient caseload and percent of discharges for Medicare beneficiaries receiving Medicare inpatient care outside their state of residence) were obtained from a 20 percent sample of MEDPAR data for fiscal year 1987 (1,973,743 discharges). These data were aggregated from the patient level to the state level using locational data for beneficiary and provider on the MEDPAR record.(2)

These two variables measure inflow and outflow of Medicare inpatients, respectively. When subtracted from 100 percent, these two patient flow measures are similar, respectively, to the Commitment Index and the Relevance Index described in Dever (1980). A third dependent variable, net patient flow ratio, was constructed by taking the ratio of percentage inflow to percentage outflow for each state.

The following independent variables are examined: average hospital occupancy ratio, physicians per thousand population, average Medicare inpatient stay charge, state per capita income, state cost of living ratio, percent age 65 and over, percent of state population in rural areas, percent of hospitals with 500 or more beds, resident per bed ratio, percent of surgeries among all state inpatient admissions, percent of state hospitals with Council of Teaching Hospitals (COTH) membership, beds per capita, hospital personnel per capita, state Medicare beneficiaries per capita, outpatient visits per capita, percent of elderly seasonal migrants to Florida, percent of elderly seasonal migrants to Arizona, and percent of total elderly seasonal migrants.

Demographic and hospital data were obtained from the 1986 Statistical Abstract of the United States and the 1986 edition of Hospital Staitistics, published by the American Hospital Association. State cost-of-living data were obtained from a previous study, by Serow et al. (1986), on the effects of cost-of-living differences on elderly migration.

The percent of nonpermanent residents per elderly (65 or older) state resident was obtained from U.S. Census supplementary report data (as reported in Hogan 1987). The numerator in this percentage included nonpermanent state residents of all ages who were U.S. citizens. However, the trends in these data were overwhelmingly dominated by elderly migrants. The data are used here as a proxy for data on elderly seasonal migrants since they comprise the only source of nationwide migration data. One of their major limitations is that they are available only at the state level. This, coupled with the desire to keep the resultant patient flow matrix manageable and consistent in form with those used in previous studies of elderly migration (McLeod et al. 1984; Serow et al. 1986) dictated that the data be analyzed at the state level.

Since this analysis is limited to inpatient discharges for Medicare beneficiaries, the inpatient flow data presented may not be similar to those that would have been obtained from inpatient data for the entire population. However, other patient origin studies have shown similarities in patterns of accessing inpatient care between Medicare and non-Medicare patients (Lubitz, Deacon, and Walton 1979). Since the MEDPAR data used is a 20 percent sample, some sampling error may be associated with the dependent variables.

Flow of Medicare

Inpatients

The 1987 Medicare discharge data, aggregated to the state level, show that 7.8 percent of the discharges comprising state Medicare inpatient caseloads are for nonresident beneficiaries. Similarly, 7.9 percent of Medicare hospitalizations are shown to occur outside the state of residence of the hospitalized beneficiary (Table 4). On average, the data suggest little patient flow across states.

While a comparatively small percentage of Medicare discharges are involved in aggregate interstate patient flow, patient flow data disaggregated by state (Tables 1-3) indicate substantial variation in interstate inpatient flow. Indeed, some states do not meet the 90 percent little in from outside" (LIFO) and "little out from inside" (LOFI) benchmarks used to determine self-contained market areas. In contrast, some other states exhibit little inflow or outflow of discharges for Medicare beneficiaries.

Several factors may account for the observed interstate variation in patient flow. Inpatient market areas may cross state boundaries. Residents.of areas lacking specialized or high-quality inpatient services (e.g., rural areas) often obtain inpatient care outside of local health care market areas (Moscovice 1989; Kane et al. 1978; Codman Research Group, Inc. 1990; General Accounting Office 1990). Medicare beneficiaries in these areas may cross state boundaries to obtain desired care especially if such care is closer in distance than other, in-state sources of care. Seasonal migration or extended travel, or both, may also explain out-of-state hospitalizations. Morrisey, Sloan, and Valvona (1989) found that the most important explanation for interstate flow varied across market area although multiple factors may have affected interstate patient flow significantly.

The data in Table 1 indicate some notable differences among states in the percentage of Medicare inpatients treated within their state of residence, in adjacent states, and in outlying states. In some states, over 97 percent of the inpatients treated under Medicare are state residents (e.g., California 97.4 percent, and Michigan 97.3 percent). In contrast, only 61.3 percent of Medicare discharges treated in the District of Columbia are for District residents. Other states treating relatively low percentages of discharges for in-state beneficiaries are Nevada (82.3 percent), North Dakota (85.4 percent), Minnesota (85.7 percent, and Vermont 86.0 percent.

[TABULAR DATA OMITTED]

In most states, Medicare discharges from out of state originate in adjacent states. Generally, the low percentage of residents treated in the states just mentioned reflects relatively high percentages of Medicare inpatient admissions from adjacent states. For example, 33.5 percent of Medicare inpatients in the District of Columbia are from adjacent states. Similarly, in Nevada and North Dakota, over 11 percent of Medicare inpatients are from adjacent states. Often this is because metropolitan statistical areas (MSAS), and related health care markets, span more than one state and the urban area closest to a particular rural area is in an adjacent state.

Some states have a relatively high percentage of Medicare inpatients from nonadjacent states. In Alaska and Florida, over 8 percent of Medicare inpatient admissions are from nonadjacent states. Other states with a high percentage of Medicare inpatient admissions from out of state are Arizona (7.9 percent), Hawaii (6.6 percent), Minnesota (6.2 percent), and Nevada (6.1 percent).

In some instances, this pattern of patient flow can be explained by the influx of seasonal migrants (Morrisey, Sloan, and Valvona 1989; Hogan 1987) and reflects trends in elderly retirement migration (Longino 1990; Atchley 1991; Long 1989). However, this patient flow pattern may also reflect the specialized inpatient resources available in states such as Minnesota, where some hospitals draw patients from a national market. Generally, hospitals in areas with high concentrations of specialized services and health care practitioners draw patients from areas that have less specialized services available locally (Luft, Robinson, Garnick, et al. 1986). Unfortunately, the reasons for patient flow across nonadjacent states are not directly observable from these data.

The data in Table 2 show differences among states in the percentage of Medicare discharges treated in the beneficiary's state of residence, in adjacent states, and in outlying states. In California, Hawaii, Louisiana, and Texas, at least 97 percent of inpatient care for state Medicare beneficiaries occurs in-state. In contrast, only 79.8 percent of inpatient discharges for Alaska Medicare beneficiaries are treated in-state. Other states treating relatively few discharges for state beneficiaries in-state are Wyoming (82.3 percent), Idaho (83.0 percent), Vermont (83.8 percent), and New Hampshire 85.5 percent .

[TABULAR DATA OMITTED]

Most of the Medicare inpatient discharges treated outside of the state of residence are treated in adjacent states, as observed for Wyoming, Idaho, Vermont, West Virginia, New Hampshire, and Delaware. Again, this is often due to MSAS that span more than one state and to rural areas where the closest urban areas are in adjacent states.

The unusually high percentage of discharges for Alaska Medicare beneficiaries treated in nonadjacent states is due in part to Alasks'a geographic isolation from the contiguous 48 states. Hospitals in Alaska are usually small, and patients seeking specialized care that may be unavailable in these hospitals necessarily do so in a nonadjacent state. Also, many elderly Alaskans may be seasonal migrants.

Nearly 7 percent of discharges for Florida Medicare beneficiaries obtain inpatient care in nonadjacent states. This may reflect seasonal migration for those beneficiaries who have changed their residence to Florida but still may receive care in their former state of residence. Such hospitalizations may be a prelude to what Longino (1990) labels "reverse migration," where elderly persons who have previously migrated return to their previous state of residence often as a result of declining health and increased dependency.

Table 3 shows the variation among states in net patient flow. Some states (e.g., Tennessee, Hawaii, Oregon, and Utah) and the District of Columbia have their net patient flow clearly dominated by patient inflow. The reasons for relatively high inflow vary across states. Tennessee and the District of Columbia show the effects of interstate inpatient markets. Other states (e.g., Wyoming, Oklahoma, Alaska, Illinois, South Carolina) have their net patient flow dominated by outflow. In Wyoming, Oklahoma, and Alaska, states where there are many small rural hospitals, the need to find specialized inpatient care doubtlessly stimulates patient outflow.
Table 3: Net Patient Flow Ratio by State
 Inflow Outflow Ratio
Alabama 4.7% 5.5% 0.85%
Alaska 8.3 20.2 0.41
Arizona 10.8 7.0 1.54
Arkansas 7.5 8.1 0.93
California 2.6 2.1 1.24
Colorado 7.7 4.9 1.57
Connecticut 5.1 6.5 0.78
Delaware 8.0 12.1 0.66
District of Columbia 38.7 11.7 3.31
Florida 9.1 7.7 1.18
Georgia 6.1 4.3 1.42
Hawaii 6.6 3.0 2.20
Idaho 7.7 17.0 0.45
Illinois 3.2 7.9 0.41
Indiana 8.4 6.4 1.23
Iowa 6.8 9.4 0.72
Kansas 5.4 11.3 0.48
Kentucky 6.0 7.3 0.82
Louisiana 3.9 2.9 1.17
Maine 4.3 5.9 0.73
Maryland 7.4 9.7 0.76
Massachusetts 5.9 3.2 1.84
Michigan 2.7 5.9 0.46
Minnesota 14.3 8.2 1.74
Mississippi 4.8 9.4 0.51
Missouri 9.9 5.4 1.83
Montana 5.4 6.0 0.90
Nebraska 7.9 7.7 1.03
Nevada 17.7 11.9 1.49
New Hampshire 12.9 14.5 0.89
New Jersey 4.1 8.4 0.49
New Mexico 6.9 10.2 0.68
New York 3.8 4.3 0.88
North Carolina 4.5 4.5 1.00
North Dakota 14.6 7.7 1.90
Ohio 4.5 4.9 0.92
Oklahoma 3.4 9.0 0.38
Oregon 11.9 5.7 2.09
Pennsylvania 4.5 3.7 1.26
Rhode Island 6.9 7.0 0.99
South Carolina 3.6 8.3 0.43
South Dakota 10.2 9.8 1.04
Tennessee 11.2 3.2 3.50
Texas 4.6 2.3 1.78
Utah 9.5 4.1 2.06
Vermont 14.0 16.2 0.86
Virginia 6.5 7.5 0.87
Washington 5.0 5.9 0.85
West Virginia 11.3 13.7 0.82
Wisconsin 5.0 6.1 0.82
Wyoming 5.6 17.7 0.32


The extremity of variation in net flow across states is evidenced by the ninefold difference between the largest and smallest net flow ratios. Indeed, it appears that the number of states with very high or very low net patient flow ratios is substantial.

Determinants of Interstate

Patient Flow

The bivariate correlations between the dependent variables and independent variables (listed in Table 4) are displayed in Table 5. The following variables are significantly correlated (.05 level) with the percent of discharges for nonresident beneficiaries in the state Medicare inpatient caseload: physicians per thousand population, average Medicare inpatient stay charge, resident per bed ratio, beds per capita, percent of state hospitals with COTH membership, hospital personnel per capita, and outpatient visits per capita. In each instance, the correlation observed is positive, indicating that increases in these variables are associated with an increasing percentage of Medicare inpatients from out of state.

[TABULAR DATA OMITTED]

[TABULAR DATA OMITTED]

The percent of state population in rural areas, the percent of seasonal migrants, and the percent of seasonal migrants to Arizona were the only variables that were significantly correlated with the percent of discharges for Medicare beneficiaries receiving care out of state. The correlation between these variables and the outflow of Medicare inpatients is positive, indicating that rural states and states from which seasonal migration is high have a greater percentage of discharges for Medicare beneficiaries who received inpatient care out of state.

Several variables were significantly correlated with the net patient flow ratio: physicians per thousand population, average Medicare inpatient stay charge, resident per bed ratio, beds per capita, hospital personnel per capita, outpatient visits per capita, and percent of seasonal migrants. Each of these correlations is positive except the correlation with seasonal migration, which is negative.

These findings suggest the likelihood of available inpatient capacity within-state, but a tendency on the part of beneficiaries to pass up one source of available care for inpatient care elsewhere. Some of this may represent bypassing nearby rural hospitals for the nearest urban hospitals, which sometimes are located in an adjacent state (Morrisey, Sloan, and Valvona 1989; General Accounting Office 1990; Moscovice 1989). But it may also be due to the lack of physician specialists in these states - as American Medical Association (AMA) data suggest - and it may lead to beneficiaries seeking specialized physician services in areas with high levels of such services (Luft, Robinson, Garnick, et al. 1986).

Regression Analysis

While the classification of interstate patient flow presented in Tables 1, 2, and 3 appears to isolate patient flow brought about by location in interstate market areas or by seasonal migration, the effects of actively seeking specialized/quality care relative to other factors must be statistically inferred, in this instance by regression analysis. The independent variables listed in Table 5 were included in stepwise multiple regression analyses (Tables 6-8) for each of the three dependent variables listed in Table 5. Some independent variables were excluded after preliminary analysis because of multicollinearity. Variables were retained if significant at the .05 level.

In the regression analysis for net patient flow (Table 6), only two independent variables, physicians per thousand population and percent seasonal migrants, are statistically significant. The strength of the physician supply variable indicates that it is the most important factor affecting net patient flow. Also, physicians per thousand population represents a proxy for supply of other medical personnel and is correlated with residents per bed, COTH membership, and percent of surgeries. The latter correlation may well be expected given the tendency for areas with high concentrations of physicians to have higher surgery rates (Cromwell and Mitchell 1986).

[TABULAR DATA OMITTED]

The negative slope for percent of seasonal migrants indicates the effect on net patient flow of the outflow of "snowbirds" to Florida and Arizona. Snowbirds clearly reduce net patient flow through increasing the outflow of discharges for Medicare beneficiaries. Although physician supply is clearly the most important variable in this regression, seasonal migration represents an important factor influencing net patient flow.

The [R.sup.2] for this regression was .280. The states with the largest (absolute) residuals (Tennessee and North Dakota) had unusually high patient inflow from adjacent states due to the presence of the major city of a multistate Metropolitan Statistical Area.(3) Also, it is possible that the effects of inflow and outflow of discharges for Medicare beneficiaries interact to diminish the predictive power of several independent variables. Effects of the independent variables listed in Table 5 on the inflow and outflow of Medicare inpatient discharges are considered in the two regressions discussed next.

The regression for the percent of nonresident Medicare inpatients (Table 7) indicates that supply of physicians and beds has a major effect on the inflow of Medicare discharges from other states. As in the regression for net patient flow, physicians per thousand population is the most important predictor of inflow of Medicare inpatients. This, again, reflects not only the importance of physicians as gatekeepers to inpatient care but also the fact that physician supply is highly correlated with the supply of other health care personnel and the availability of surgical specialists and specialized procedures.

[TABULAR DATA OMITTED]

Another supply variable, inpatient beds per capita, also stimulates the inflow of patients but it is barely statistically significant. Thus, much of the inflow of hospitalizations for nonresident Medicare beneficiaries appears to be a function of supply variables.

Surprisingly, occupancy rate is negatively associated with nonresident inflow, although the sign of the bivariate correlation between these two variables is positive. This implies, counterintuitively, that low occupancy rates are associated with drawing inpatients from other states. However, the result observed may be due to a suppression effect.

Suppression effects occur when a variable that appears to be uncorrelated with the dependent variable in a bivariate analysis becomes significant in the presence of other predictors that control for an irrelevant shared variation that had reduced the magnitude of the observed bivariate relationship (Pedhazur 1982; Cohen and Cohen 1983). Once the effects of physicians and bed supply are controlled for, high occupancy rates given high supply indicate less of a loss of inpatients to other states.

The [R.sup.2] for this regression is .541. The highest residual value was observed for Nevada, which had a high inflow of discharges from both adjacent and nonadjacent states; the lowest residual value was observed for California, which had a very low inflow of discharges from out of state. Inflow, therefore, appears to be much more predictable than total patient flow and appears to be unaffected by seasonal migration but strongly affected by the availability of services.

The regression for the percent of discharges for Medicare beneficiaries receiving inpatient care outside their state of residence (Table 8) indicates that location, supply of beds, and specialized services are the most important factors influencing outflow of Medicare inpatients. Seasonal migration (to Arizona only) is also a significant, but minor predictor of outflow.

[TABULAR DATA OMITTED]

The most important predictor in this analysis, percent of rural residents, suggests an important link to the supply of physicians and services. Rural states lose Medicare inpatients to other states for inpatient care (presumably, to more urbanized states with more developed medical care facilities). This is consistent with anecdotal statements indicating that many rural residents are obtaining hospital care in urban areas rather than in nearby rural areas (General Accounting Office 1990; Moscovice 1989).

This effect is reinforced by the negative slope for inpatient beds per capita. Thus, outflow of Medicare inpatients appears directly related to shortage of beds and other health personnel.

The interpretation of the effect of the percent of COTH hospitals is less clear-cut due to suppression effects. One possible interpretation is that access per se is at issue rather than access to specialized services. Also, available in-state care may be more distant than out-of-state care.

While seasonal migration is a significant determinant of the outflow of Medicare inpatients, apparently only seasonal migration to Arizona (rather than total seasonal migration) is significant here. This may be because seasonal migrants to Arizona come from a larger number of states than do Florida's seasonal migrants.

The [R.sup.2] for this regression is .413. Alaska and Florida had the largest residual scores. The Alaska residual value suggests that additional service supply/quality variables may be needed in this regression. The Florida residual may be capturing the effects of seasonal migration to northern states during the summer.(4) While Arizona's seasonal migration affected the outflow of Medicare hospitalizations, the migration effect was small compared to the impact of location and supply factors.

In these regressions, the independent variables represent absolute levels. However, it may be possible that while two states may have similar rates of physicians or beds per capita, inflow or outflow may occur because these rates are higher or lower than in neighboring states. While this would complicate the regression model, it is probable that relative differences among neighboring states may significantly affect interstate patient flow.

Summary

Interstate flow of inpatient discharges for Medicare beneficiaries has a small but significant effect on the size of state Medicare inpatient caseloads. While seasonal migration accounts for an important percentage of the interstate flow of Medicare inpatients, most interstate flow occurs either because inpatient markets cross state boundaries or because appropriate specialized care is most readily obtained in another state. Usually, this involves patient flow to an adjacent rather than a distant state.

The statistical analyses in this article suggest that the interstate flow of Medicare inpatients is primarily due to variation across states in the availability of physicians, inpatient beds, specialists, and specialized services. Some of these variations are also a function of urban/rural location and have been cited as reasons why rural residents seek treatment in urban areas rather than locally (Moscovice 1989). This effect was apparent whether patient flow was measured in terms of inflow, outflow, or the ratio of inflow to outflow.

Seasonal migration is a significant, but comparatively weak, predictor for outflow and net flow only. While its effects are important, seasonal migration is clearly secondary to availability of inpatient beds and physicians as a reason why Medicare inpatients are treated outside their state of residence. Thus, seasonal migrants account for a far smaller percentage of out-of-state discharges than do beneficiaries seeking care in interstate market areas. It may be preferable also to determine capitated rates for market areas rather than for states or other arbitrary geographic units.

Implications for Future

Research

As noted earlier, interstate patient flow may be due to movement within market areas that cross state boundaries, transitional movement as a result of permanent migration, regular seasonal migration, vacation travel, or purposive care seeking. Unfortunately, classification of discharges by adjacency of state and regression analysis using proxy variables for seasonal migration and supply/quality of services provide only an imperfect accounting of type of interstate inpatient flow. Nevertheless, the data presented do indicate the relative importance of seasonal migration versus interstate markets in accounting for interstate patient flow.

Patient flow in interstate markets can be examined using methodology similar to that used by Luft, Garnick, Mark, et al. (1990) and a multistate discharge database. Specification of a multinomial logistic function incorporating independent variables to measure supply of specialized services and quality of care can allow one to evaluate whether people choose hospitals for their service offerings, their quality, or their closeness to home. This model may also be disaggregated by diagnosis-related group (DRG) to determine if certain conditions are more likely to be treated outside a beneficiary's usual inpatient market area, as Codman Research Group, Inc. (1990) found for rural residents.

Seasonal migrants and seasonal migration effects are much more difficult to isolate. One approach to this problem is to identify Medicare beneficiaries who represent patient inflow or outflow from migration centers (Florida, Arizona) and to obtain survey data concerning reasons for obtaining inpatient care out of state. These data can then be analyzed to evaluate quality of care and patient satisfaction and to describe seasonal migrants' "dual network' of health care providers.

Notes

1. In the description of the discharge data used in their study, Griffith et al. (1985) show the extent to which their discharge rates would have been biased if data on out-of-state discharges had not been obtained. 2. The beneficiary residence data on the MEDPAR record is abstracted from the residence data on the Medicare beneficiary file, which is obtained from Social Security eligibility data (Lubitz et al. 1980). This is not necessarily the address given to the hospital upon admission. Since the address recorded on MEDPAR does not change for seasonal migrants, MEDPAR data cannot be used to identify "snowbirds." Further, administrative delay in changing a beneficiary's address after moving may falsely indicate that care was obtained outside of the state of residence. 3. Four Tennessee metropolitan statistical areas (Memphis, TN-AR-MS; Johnson City-Kingsport-Bristol, TN-VA; Clarksville-Hopkinsville, TN-KY; and Chattanooga, TN-GA) are interstate MSAs where the major city is located in Tennessee. 4. Generally, seasonal migrants are portrayed as residents of northern states who migrate south for the winter. However, many residents of Florida and Arizona seasonally migrate to northern states in the summer or "return migrate" when their health declines (Longino 1990).

References

Atchley, R. Social Forces and Aging. Belmont, CA: Wadsworth, 1991. Codman Research Group, Inc. The Relationship between Declining Use of Rural Hospitals and Access to Inpatient Services for Medicare Beneficiaries in Rural Areas. Lyme, NH: Codman Research Group, Inc., 1990. Cohen, J., and P. Cohen. Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence Erlbaum, 1983. Cohen, M., and H. Lee. "The Determinants of Spatial Distribution of Hospital Utilization in a Region." Medical Care 23, no. 1 (1985): 27-38. Cromwell, J., and J. Mitchell. "Physician-Induced Demand for Surgery." Journal of Health Economics 5, no. 4 (1986): 293-313. Dever, G. Community Health Analysis: A Holistic Approach. Germantown, MD: Aspen, 1980. Elzinga, K., and T. Hogarty. "The Problem of Geographic Market Delineation in Antimerger Suits." Antitrust Bulletin 18, no. 1 (1973): 45-81. Folland, S. "Predicting Hospital Market Shares." Inquiry 20, no. 1 (1983): 34-44. Garnick, D., H. Luft, J. Robinson, and J. Tetreault. "Appropriate Measures of Hospital Market Areas." Health Services Research 22, no. 1 (1987): 69-89. Garnick, D., E. Lichtenberg, C. Phibbs, H. Luft, D. Peltzman, and S. McPhee. "The Sensitivity of Conditional Choice Models for Hospital Care to Estimation Technique." Journal of Health Economics 8, no. 4 (1989): 377-97. General Accounting Office. Rural Hospitals: Federal Leadership and Targeted Programs Needed. Publication no. GAO/HRD-90-67. Washington, DC: U.S. General Accounting Office, 1990. Griffith, J., P. Wilson, R. Wolfe, and D. Bischak. "Clinical Profiles of Discharge Rates in Local Communities." Health Services Research 20, no. 2 (1985): 131-51. Harner, E., and P. B. Slater. "Identifying Medical Regions Using Hierarchical Clustering." Social Science & Medicine 14D, no. 1 (1980): 3-10. Hogan, T. "Determinants of the Seasonal Migration of the Elderly to Sunbelt Cities." Research on Aging 9, no. 1 (1987): 115-33. Kane, R., D. Olson, J. Newman, and J. Manson. "Giving and Getting Surgery in Utah: An Urban-Rural Comparison." Surgery 83, no. 4 (1978): 375-81. Long, L. Migration and Residential Mobility in the United States. New York: Russell Sage, 1989. Longino, C. "Geographical Distribution and Migration." In Handbook of Aging and the Social Sciences. 3rd ed. Edited by E. Shanas and R. Binstock. New York: Academic Press, 1990. Lubitz, J., R. Deacon, and C. Walton. "Medicare Hospital Patient Origin and Destination Data for Health Planning." Public Health Conference on Records and Statistics. Hyattsville, MD: National Center for Health Statistics, 1979. Lubitz, J., C. Walton, M. Newton, and S. Hadley. Hospital Patient Origin and Destination Data for Medicare Patients 65 Years and Over. Baltimore, MD: Health Care Financing Administration, 1980. Luft, H., J. Robinson, D. Garnick, S. Maerki, and S. McPhee. "The Role of Specialized Clinical Services in Competition among Hospitals." Inquiry 23, no. 1 (1986): 83-94. Luft, H., D. Garnick, D. Mark, D. Peltzman, C. Phibbs, E. Lichtenberg, and S. McPhee. "Does Quality Influence Choice of Hospital?" Journal of the American Medical Association 263, no. 21 (1990): 2899-2906. Mayer, J. "The Distance Behavior of Hospital Patients: A Disaggregated Analysis." Social Science & Medicine 17, no. 12 (1983): 819-27. McGuirk, M., and F. Porell. "Spatial Patterns of Hospital Utilization: The Impact of Distance and Time." Inquiry 21, no. 2 (1984): 84-95. McLeod, K., J. Parker, W. Serow, and N. Rives. "Determinants of State to State Flows of Elderly Migrants." Research on Aging 6, no. 3 (1984): 372-83. Monahan, D., and V. Greene. "The Impact of Seasonal Population Fluctuations on Service Delivery." Gerontologist 22, no. 2 (1982): 160-63. Morrisey, M., F. Sloan, and J. Valvona. "Defining Geographic Markets for Hospital Care." Law and Contemporary Problems 51, no. 2 (1989): 165-94. Moscovice, 1. "Rural Hospitals: A Literature Synthesis and Health Services Research Agenda." Health Services Research 23, no. 6 (1989): 891-930. Pedhazur, E. Multiple Regression in Behavioral Research. 2d ed. New York: Holt, Rinehart, and Winston, 1982. Serow, W., D. Charity, G. Fournier, and D. Rasmussen. "Cost of Living Differentials and Elderly Interstate Migration." Research on Aging 8, no. 2 (1986): 317-27.
COPYRIGHT 1992 Health Research and Educational Trust
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Buczko, William
Publication:Health Services Research
Date:Aug 1, 1992
Words:5916
Previous Article:Testing the null hypothesis in small area analysis.
Next Article:Evaluation of the HCFA model for the analysis of mortality following hospitalization.
Topics:

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters