Search

Objective: To compare measures of spatial access to care commonly used by policy makers and researchers with the more comprehensive enhanced two-step floating catchment area (E2SFCA) method., Study Setting: Fourteen southwestern Pennsylvania counties., Study Design: We estimated spatial access to buprenorphine-waivered prescribers using three commonly used measures-Euclidean travel distance to the closest prescriber, travel time to the closest provider, and provider-to-population ratios-and the E2SFCA. Unlike other measures, the E2SFCA captures provider capacity, potential patient volume, and travel time to prescribers., Data Collection/extraction Methods: We measured provider capacity as the number of buprenorphine prescribers listed at a given address in the Drug Enforcement Agency's 2020 Controlled Substances Act Registrants Database, and we measured potential patient volume as the number of nonelderly adults in a given census tract as reported by the 2018 American Community Survey. We estimated travel times between potential patients and prescribers with Bing Maps and Mapbox application programming interfaces. We then calculated each spatial access measure using the R programming language. We used each measure of spatial access to identify census tracts in the lowest quintile of spatial access to prescribers., Principal Findings: The Euclidean distance, travel time, and provider-to-population ratio measures identified 48.3%, 47.2%, and 69.9% of the census tracts that the E2SFCA measure identified as being in the lowest quintile of spatial access to care, meaning that these measures misclassify 30%-52% of study area census tracts as having sufficient spatial access to buprenorphine prescribers., Conclusions: Measures of spatial access commonly used by policy makers do not sufficiently accurately identify geographic areas with relatively low access to prescribers of buprenorphine. Using the E2SFCA in addition to the commonly used measures would allow policy makers to precisely target interventions to increase spatial access to opioid use disorder treatment and other types of health care services., (© 2021 Health Research and Educational Trust.)

Objective: To develop an automated, reproducible method for delineating hospital service areas (HSAs)., Data Sources/setting: Discharge data from all Swiss acute care hospitals for the years 2013 to 2016., Study Design: We derived HSAs and hospital referral regions for Switzerland using a newly developed flow-based, automated, objective, and reproducible method using all discharge data. We compared our method to the classical, partially subjective approach used to delineate the Swiss Health Care Atlas by delineating four sets of intervention-specific HSAs., Principal Findings: Based on 4 105 885 discharges, the fully automated method delineated 63 HSAs. Comparison with existing HSAs reveals good overlap and comparable measures of health utilization between the methods and shows that in the Swiss setting, our method outperforms a cluster-based approach to defining HSAs. While the classical method potentially takes an entire day to delineate the regions, our method took approximately 10 minutes., Conclusions: Hospital service areas are used to analyze differences in use of health care that may indicate underuse and overuse. Our new, fully automated, objective, and reproducible method provides a useful tool for hospital services researchers that will enable them to delineate and update patient-flow-based HSAs., (© Health Research and Educational Trust.)

Objective: To examine geographic access to community health centers (CHC accessibility) before and after Health Center Program expansion in three Southern states., Data Sources: Community health center data were from the Health Resources and Services Administration (1967-2016). Population estimates and sociodemographic characteristics were from the American Community Survey (2006-2015)., Study Design: We used the two-step floating catchment area method to calculate CHC accessibility for census tracts in 2008 and 2016. We mapped census tract-level variation and used spatial regression to assess to what extent indicators of potential CHC need were associated with change in accessibility from 2008 to 2016., Principal Findings: Community health center accessibility increased by 192 percent overall, and the proportion of tracts with no accessibility decreased by 65 percent. Indicators of potential need were not associated with greater gains in CHC accessibility from 2008 to 2016, but census tracts with less accessibility at baseline saw larger accessibility increases., Conclusions: Community health center accessibility substantially increased from 2008 to 2016, but increases did not differentially impact groups with greater potential need. This approach for measuring CHC accessibility offers significant improvement in granularity over traditional CHC accessibility measures., (© Health Research and Educational Trust.)

Objective: To demonstrate the use of multiple-membership multilevel models, which analytically structure patients in a weighted network of hospitals, for exploring between-hospital variation in preventable hospitalizations., Data Sources: Cohort of 267,014 people aged over 45 in NSW, Australia., Study Design: Patterns of patient flow were used to create weighted hospital service area networks (weighted-HSANs) to 79 large public hospitals of admission. Multiple-membership multilevel models on rates of preventable hospitalization, modeling participants structured within weighted-HSANs, were contrasted with models clustering on 72 hospital service areas (HSAs) that assigned participants to a discrete geographic region., Data Collection/extraction Methods: Linked survey and hospital admission data., Principal Findings: Between-hospital variation in rates of preventable hospitalization was more than two times greater when modeled using weighted-HSANs rather than HSAs. Use of weighted-HSANs permitted identification of small hospitals with particularly high rates of admission and influenced performance ranking of hospitals, particularly those with a broadly distributed patient base. There was no significant association with hospital bed occupancy., Conclusion: Multiple-membership multilevel models can analytically capture information lost on patient attribution when creating discrete health care catchments. Weighted-HSANs have broad potential application in health services research and can be used across methods for creating patient catchments., (© 2017 The Authors. Health Services Research published by Wiley Periodicals, Inc. on behalf of Health Research and Educational Trust.)

Objective: To investigate the association between older adults' potentially avoidable hospitalization rates and both a geographic measure of primary care physician (PCP) access and a standard bounded-area measure of PCP access., Data Sources: State physician licensure data from the Virginia Board of Medicine. Patient-level hospital discharge data from Virginia Health Information. Area-level data from the American Community Survey and the Area Health Resources Files. Virginia Information Technologies Agency road network data. US Census Bureau TIGER/Line boundary files., Study Design: We use enhanced two-step floating catchment area methods to calculate geographic PCP accessibility for each ZIP Code Tabulation Area in Virginia. We use spatial regression techniques to model potentially avoidable hospitalization rates., Data Collection/extraction: Geographic accessibility was calculated using ArcGIS. Physician locations were geocoded using TAMU GeoServices and ArcGIS., Principal Findings: Increased geographic access to PCPs is associated with lower rates of potentially avoidable hospitalization among older adults. This association is robust, allowing for spatial spillovers in spatial lag models., Conclusions: Compared to bounded-area density measures, unbounded geographic accessibility measures provide more robust evidence that avoidable hospitalization rates are lower in areas with more PCPs per person. Results from our spatial lag models reveal the presence of positive spatial spillovers., (© Health Research and Educational Trust.)

Objective: To analyze the general practitioners (GPs) with regard to the degree of urbanization, social deprivation, general health, and disability., Data Sources: Small area population data and GP practice data in England., Study Design: We used a floating catchment area method to measure spatial GP accessibility with regard to the degree of urbanization, social deprivation, general health, and disability., Data Collection: Data were collected from the Office for National Statistics and the general practice census and analyzed using a geographic information system., Principal Findings: In all, 25.8 percent of the population in England lived in areas with a significant low GP accessibility (mean z-score: -4.2); 27.6 percent lived in areas with a significant high GP accessibility (mean z-score: 7.7); 97.8 percent of high GP accessibility areas represented urban areas, and 31.1 percent of low GP accessibility areas represented rural areas (correlation of accessibility and urbanity: r = 0.59; p<.001). Furthermore, a minor negative correlation with social deprivation was present (r = -0.19; p<.001). Results were confirmed by a multivariate analysis., Conclusion: This study showed substantially differing GP accessibility throughout England. However, socially deprived areas did not have poorer spatial access to GPs., (© Health Research and Educational Trust.)

Objective: To develop an automated, data-driven, and scale-flexible method to delineate hospital service areas (HSAs) and hospital referral regions (HRRs) that are up-to-date, representative of all patients, and have the optimal localization of hospital visits., Data Sources: The 2011 state inpatient database in Florida from the Healthcare Cost and Utilization Project., Study Design: A network optimization method was used to redefine HSAs and HRRs by maximizing patient-to-hospital flows within each HSA/HRR while minimizing flows between them. We first constructed as many HSAs/HRRs as existing Dartmouth units in Florida, and then compared the two by various metrics. Next, we sought to derive the optimal numbers and configurations of HSAs/HRRs that best reflect the modularity of hospitalization patterns in Florida., Principal Findings: The HSAs/HRRs by our method are favored over the Dartmouth units in balance of region size and market structure, shape, and most important, local hospitalization., Conclusions: The new method is automated, scale-flexible, and effective in capturing the natural structure of the health care system. It has great potential for applications in delineating other health care service areas or in larger geographic regions., (© Health Research and Educational Trust.)

Objective: To develop a method of hospital market area identification using multivariate data, and compare it with existing standard methods., Data Sources: Hospital Episode Statistics, a secondary dataset of admissions data from all hospitals in England, between April 2005 and March 2006., Study Design: Seven criteria for catchment area definition were proposed. K-means clustering was used on several variables describing the relationship between hospitals and local authority districts (LADs) to enable the placement of every LAD into or out of the catchment area for every hospital. Principal component analysis confirmed the statistical robustness of the method, and the method was compared against existing methods using the seven criteria., Principal Findings: Existing methods for identifying catchment areas do not capture desirable properties of a hospital market area. Catchment areas identified using K-means clustering are superior to those identified using existing Marginal methods against these criteria and are also statistically robust., Conclusions: K-means clustering uses multivariate data on the relationship between hospitals and geographical units to define catchment areas that are both statistically robust and more informative than those obtained from existing methods.

Purpose: To assess the extent of bypass for inpatient care among patients living in Critical Access Hospital (CAH) service areas, and to determine factors associated with bypass, the reasons for bypass, and what CAHs can do to retain patients locally., Methods: Six hundred and forty-seven subjects, aged 18 years and older, who had been admitted to a hospital for inpatient care in the past 12 months and lived within 15-20 miles of 25 randomly selected CAHs were surveyed by phone during the period from early February through late July 2005. Survey questions included demographic characteristics, general health status, travel time/distance to health care, questions on satisfaction with local health services, bypass behavior, and solicited suggestions on how local hospitals could retain patients locally., Findings: About 60% of surveyed patients bypassed their local CAHs for inpatient care including 16% who were referred to another facility by the local CAH/health care providers and would use the local hospital if needed services were available. Bypass rates ranged from 16% to 70% across the sampled CAHs. Factors associated with bypass included age, income, satisfaction with the local hospital, and traveling distance/time. Lack of specialty care, limited services, and the quality/reputation of local services/doctors were most frequently mentioned as reasons why patients bypass local CAHs., Conclusions: The bypass rate for sampled CAHs is considerably higher than the 20%-50% bypass rates documented in the literature for all hospitals in general using discharge/administrative data. The sizeable variation in bypass rates across CAHs suggests that the appropriate response/fix should come from the facility/community levels.

Context: Access to transportation to transverse the large distances between residences and health services in rural settings is a necessity. However, little research has examined directly access to transportation in analyses of rural health care utilization., Purpose: This analysis addresses the association of transportation and health care utilization in a rural region., Methods: Using survey data from a sample of 1,059 households located in 12 western North Carolina counties, this analysis tests the relationship of different transportation measures to health care utilization while adjusting for the effects of personal characteristics, health characteristics, and distance., Findings: Those who had a driver's license had 2.29 times more health care visits for chronic care and 1.92 times more visits for regular checkup care than those who did not. Respondents who had family or friends who could provide transportation had 1.58 times more visits for chronic care than those who did not. While not significant in the multivariate analysis, the small number who used public transportation had 4 more chronic care visits per year than those who did not. Age and lower health status were also associated with increased health care visits. The transportation variables that were significantly associated with health care visits suggest that the underlying conceptual frameworks, the Health Behavior Model and Hagerstrand's time geography, are useful for understanding transportation behavior., Conclusions: Further research must address the transportation behavior related to health care and the factors that influence this behavior. This information will inform policy alternatives to address geographic barriers to health care in rural communities.

Objective: To describe patterns in ambulatory care sensitive (ACS) admissions at the zip code level based on zip code demographic and other characteristics. These patterns include trends over time, persistence within zip codes over time, and variation between and within socioeconomic strata., Data Sources: New York State hospital discharge data 1990-1998, U.S. census data 1990, and New York State birth records 1990., Study Design: Age- and sex-adjusted rates and volumes of ACS admissions are calculated at the zip code level. Descriptive statistics are analyzed cross-sectionally and over time. Kernel density functions are estimated across income strata. Ordinary and quantile regression techniques are used to determine the impact of socioeconomic variables on average and extreme values of the distribution of ACS admission rates., Principal Findings: Ambulatory care sensitive admissions rates declined during the study period but in conjunction with a greater decline in overall admission rates. Thus, as a percentage of total admissions, they actually rose by 4 percent. Ambulatory care sensitive admissions are geographically concentrated and rates are highly persistent within zip codes over time. Even on a log scale ACS admissions are typically greater and exhibit more variability among low-income zip codes. Other variables positively associated with ACS admissions are total population, births to unwed mothers (a proxy for family structure), black population, Hispanic population, and the number of non-ACS admissions. Births to immigrant mothers (a proxy for immigrant population) are negatively associated with ACS admissions., Conclusions: The concentration and persistence of ACS admissions point to a chronic, geographically limited deficiency of primary ambulatory care in the most underserved neighborhoods. Much of the difference in preventable hospitalization levels between high- and low-income areas is driven by very high volumes in the low-income areas unrelated to population density. New York data suggest that most costs from preventable hospitalizations could be saved by focusing on targeted neighborhoods. Socioeconomic and area utilization variables play a role in both average and extreme values of the rate of preventable hospitalizations at the zip code level. Since variables that affect the average volume of preventable hospitalizations can change the distribution of that volume, analysis based on averages alone may be inadequate. The findings on area demographics and non-ACS admissions point to the need to better understand social and cultural issues as well as local admitting practice patterns to encourage appropriate and efficient use of the health care delivery system.

Objective: To develop and characterize utilization-based service areas for the United States which reflect the travel of Medicare beneficiaries to primary care clinicians., Data Source/study Setting: The 1996-1997 Part B and 1996 Outpatient File primary care claims for fee-for-service Medicare beneficiaries aged 65 and older. The 1995 Medicaid claims from six states (1995) and commercial claims from Blue Cross Blue Shield of Michigan (1996)., Study Design: A patient origin study was conducted to assign 1999 U.S. zip codes to Primary Care Service Areas on the basis of the plurality of beneficiaries' preference for primary care clinicians. Adjustments were made to establish geographic contiguity and minimum population and service localization. Generality of areas to younger populations was tested with Medicaid and commercial claims., Data Collection/extraction Methods: Part B primary care claims were selected on the basis of provider specialty, place of service, and CPT code. Selection of Outpatient File claims used provider number, type of facility/service, and revenue center codes., Principal Findings: The study delineated 6,102 Primary Care Service Areas with a median population of 17,276 (range 1,005-1,253,240). Overall, 63 percent of the Medicare beneficiaries sought the plurality of their primary care from within area clinicians. Service localization compared to Medicaid (six states) and commercial primary care utilization (Michigan) was comparable but not identical., Conclusions: Primary Care Service Areas are a new tool for the measurement of primary care resources, utilization, and associated outcomes. Policymakers at all jurisdictional levels as well as researchers will have a standardized system of geographical units through which to assess access to, supply, use, organization, and financing of primary care services.

Rural patients who are admitted to hospitals outside their residence county or who travel great distances for hospitalization deprive local rural hospitals of revenue. To provide more information about such rural residents, we studied their characteristics compared to those admitted in the same county. Characteristics studied included illness severity, demographics and county resources. To validate the findings and to provide a different analytic approach, characteristics of residents who travel long distances for admission were also studied. We studied admissions for ambulatory care sensitive conditions, as they might be most responsive to policy changes such as increasing recruitment of local primary-care physicians. Hospital discharges during 1994 for 248,656 New York State residents were studied. We constructed multivariate models using logistic regression and ordinary least squares methods. The models were applied to residents in three types of geographic location along an urban-rural continuum. Outside admissions were associated with younger age, higher illness severity and fewer county hospital resources. Same county admissions were associated with nonwhite race, and lack of insurance. Surprisingly, in rural counties, outside admissions were directly associated with increased primary-care providers. Results from the distance model generally supported findings from the outside admission model.

Objective: To use Medicaid claims data to construct patient origin-based market areas for dental services and compare constructed market areas with those based on the practice county., Data Sources: North Carolina Medicaid claims, eligibility, and provider files, the Cooperative Health Information Systems' dentist licensure files, and the Log Into North Carolina data., Study Design: A visit-level file was created from the Medicaid claims data and aggregated by provider practice county and patient county of residence. Using the aggregated file and an algorithm based on the Elzinga-Hogarty approach, patient travel patterns were used to construct mutually exclusive patient origin market areas., Data Analysis: Market area characteristics were compared across definitions using Pearson correlation coefficients. In addition, estimations of provider participation were performed using market area characteristics as control variables. The beta coefficients associated with market area characteristics were compared across market area definitions., Principal Findings: Medicaid claims data, when combined with provider licensure files, can be used to construct market areas based on patient origin data. However, measures of market area characteristics are correlated highly between the two types of market areas studied. Furthermore, beta coefficients on market area variables in models of provider participation are similar in sign, significance, and magnitude across market definitions., Conclusions: Compared with market areas constructed using patient origin data, county-based market areas adequately proxy for dental markets. Using the county as the market area also avoids the time and computational costs associated with using a patient origin-based approach and facilitates the use of widely available data.

Objective: To draw together insights from three perspectives (health economics, organizational ecology, and institutional theory) in order to clarify the factors that influence entries of providers into healthcare markets. A model centered on the concept of an organizational field is advanced as the level of analysis best suited to examining the assortment and interdependence of organizational populations and the institutional forces that shape this co-evolution. In particular, the model argues that: (1) different populations of healthcare providers partition fiscal, geographic, and demographic resource environments in order to ameliorate competition and introduce service complementarities; and (2) competitive barriers to entry within populations of providers vary systematically with regulatory regimens., Data Sources: County-level entries of hospitals and home health agencies in the San Francisco Bay Area using data from the American Hospital Association (1945-1991) and California's Office of Statewide Health Planning and Development (1976-1991). Characteristics of the resource environment are derived from the Area Resource File (ARF) and selected government censuses., Methods of Analysis: A comparative design is applied to contrast influences on hospital and home health agency entries during the post-World War II period. Empirical estimates are obtained using Poisson and negative binomial regression models., Results: Hospital and HHA markets are partitioned primarily by the age and education of consumers and, to a lesser extent, by urbanization levels and public funding expenditures. Such resource partitioning allows independent HHAs to exist comfortably in concentrated hospital markets. For both hospitals and HHAs, the barriers to entry once generated by oligopolistic concentration have declined noticeably with the market-oriented reforms of the past 15 years., Conclusion: A field-level perspective demonstrates that characteristics of local resource environments interact with interdependencies of provider populations and broader regulatory regimes to affect significantly the types of provider organizations likely to enter a given healthcare market.

Classification of U.S. counties into metropolitan and nonmetropolitan masks a high degree of variation in population size, density, and relationship between urban and rural areas. This article reviews the urban influence county coding system and illustrates how it applies to U.S. counties. The urban influence codes can be useful for analysis of rural health policy.

Objective: To develop and validate a technique for defining a practice population of discrete individuals based on multiyear family practice fee-for-service billings data., Data Sources/study Setting: Nineteen family physicians in Ontario, Canada who converted from fee-for-service to capitation payment. Data sources were fee-for-service billings data for the three-year period prior to the conversion from fee-for-service to capitation payment and the rosters of enrolled patients for the first and third years after the change to capitation payment., Study Design: The billings-based definition of the physician's practice population was compared against the Year 1 roster. We also compared the billings-based practice population and the Year 1 roster to the physician's Year 3 roster to identify patients who might have been missed during the roster development process. Our principal analyses were an assessment of the sensitivity of the billings-based definition of the practice population (EPP), the positive predictive value of EPP, and the agreement between EPP and the rostered patient population (RPP). We also examined the ratio between EPP and RPP to determine EPP's accuracy in estimating the practice denominator., Data Collection/extraction Methods: The practice population for each physician at the time of conversion from fee-for-service to capitation payment was defined as (a) all persons for whom the physician billed the provincial health insurance plan for at least one visit during the year immediately prior to joining the capitation-funded program; and (b) all additional patients for whom the physician billed the plan for at least one service in each of the two preceding years. Data extraction was carried out within the Ministry of Health in order to preserve the anonymity of patients and physicians. Data were provided to the investigators stripped of patient and physician identifiers., Principal Findings: The mean sensitivity and positive predictive value of EPP were 95.3 percent and 87.4 percent, respectively. The level of agreement between EPP and RPP averaged 84.4 percent. The mean ratio of EPP to RPP was 1.21 (95 percent C.I. 1.030-1.213). Correction for roster false-negatives increased the sensitivity, positive predictive value, and agreement between EPP and the practice population, and reduced the mean ratio of EPP to the practice population to 1.068 (95 percent C.I. 1.010-1.127)., Conclusions: The practice population can usefully be defined in fee-for-service family practice on the basis of multiyear fee-for-service billings data. Further research examining alternative encounter-based practice population definitions would be valuable.

Objective: We assess the effect of variations in the supply and specialty distribution of physicians on admission rates for ambulatory care-sensitive conditions (ACS) and for all causes, and on mortality rates among Medicare beneficiaries of various health care service areas (HCSA)., Data Sources: For the Medicare beneficiaries, sources were the Health Care Financing Administration's 1992 enrollment and impatient (Part A) files for a 5 percent sample of that population; for the overall populations and for the medical resources of the HCSAs, the Area Resource File., Study Design: This observational, cross-sectional study employed multiple linear regression to assess the influence of population characteristics and of the supply of physicians on hospital admissions, and Poisson regression in the analysis of the factors that affect mortality., Principal Findings: Physician supply levels vary nearly fourfold or more when comparing the top and bottom deciles of the HCSAs, Medicare admissions for ACS conditions vary about threefold, and admission rates for all causes and mortality rates vary about 1.5-fold. Physician supply levels and distributions have very little influence on ACS admission rates, and even less on the admissions for all causes and on mortality, except in HCSAs with very low physician supply levels (one-fourth the national average or less). However, these HCSAs account for only about 1 percent of the U.S. population., Conclusions: Physician supply levels and the proportions of specialists and generalists have negligible effects on health status as measured by mortality rates and by rates of admission for all causes and for conditions presumed to be sensitive to the adequacy of ambulatory care. Reductions in admissions for such conditions are not likely to be achieved through broadening of insurance to levels that exist under Medicare, nor through increases in the supply of physicians, nor, conversely, through a reduction in any presumed oversupply of physicians.

Objective: This study investigates the determinants of primary care office visit rates., Data Sources: Blue Cross and Blue Shield of Iowa subscriber information was sorted by residence into geographic health service areas. Cost-sharing information was also obtained from Blue Cross. Physician supply data were obtained from The University of Iowa, Office of Community-Based Programs. Hospital data were reported by the Iowa Hospital Association., Study Design: Cases were classified into ambulatory care groups (ACGs). Use rates were computed for each group in each service area. Ordinary least squares regression models were developed to model geographic variation in each ACG-specific primary care visit rate., Principal Findings: Regression models were not significant for five out of eleven ACGs studied. Out-of-pocket expense significantly affected utilization in three out of six. The number of primary care practices per capita had a significant effect on utilization in two ACGs. The supply of hospital outpatient services was significant in one ACG., Conclusions: Study findings reveal that some ACGs are price-sensitive and some are not. Policies aimed at changing levels of primary care use should taken into account whether varying cost-sharing will influence consumer behavior in the desired direction.

In the late 1980s several published articles predicted a crisis in the availability of obstetric care due to declining numbers of rural obstetrical providers. Several state and national studies documented the adverse impact of malpractice and time demands on both urban and rural physicians. But only limited information is available to document current trends in rural obstetrical practice and assess whether or not the predicted crisis occurred. This study sought to provide that updated information for rural Minnesota. A telephone survey of all rural Minnesota obstetrical providers was used to document the number, location, and specialty of rural obstetrical providers, their practice limitations, and plans for future practice. This data was combined with state perinatal statistics for each county to further assess obstetrical care availability and perinatal outcomes. All rural Minnesota obstetricians and certified nurse midwives provide obstetrical care as did 69 percent of all rural family physicians. Only 27 percent of rural obstetrical providers put any type of restrictions on their obstetrical practices. During the past year, 67 currently practicing rural physicians have stopped providing obstetrical care while 55 new obstetrical providers have begun rural practice. Two to 3 percent of current rural providers plan to retire or discontinue obstetrical services during the next five years. The provider demographics from the survey identified eight counties with no prenatal providers, and 12 additional communities of decreased provider availability. However, only two of the counties with no prenatal providers and five of the counties with areas of limited providers had increased percentages of adverse prenatal outcomes such as low birthweight or late prenatal care. This study concluded that Minnesota does not have a serious statewide problem with availability of rural obstetrical providers. However, a few isolated regions of the state have limited provider availability, including limited availability of local high-risk services and consultants.

Objective: Our purpose is a descriptive analysis of variations in hospital use among small areas of Maryland., Data Source: The data are Maryland patient discharge records from acute care hospitals for 1985-1987 and small area population estimates by age, gender, race, and income., Findings: The common finding was excess geographic variability among Maryland's 115 areas. The hypothesis of uniform rates was rejected for most DRGs, including low-variation mastectomy and hernia repair. Clustering of high-use rates occurred in neighboring areas for orthopedic, vascular, and elective procedures. Admission rates for most nondiscretionary procedures and medical DRGs were reduced in affluent areas while discretionary surgery increased with income level. Elective procedures had extreme variation and were related to income. Coronary artery disease rates declined with income while coronary artery procedure rates increased, indicating that access and patient selection were factors in the use of coronary bypass and angioplasty., Conclusions: The issue is not the ubiquitous variation among small areas but its extent and identification of geographic patterns. Hospital use is related to demography, morbidity, medical resources, access, selection for care, and physician practice patterns. Heterogeneity of these factors ensures that uniform delivery of health care rarely holds. There is little evidence that incidence of surgical disease is the main source of variation in use of discretionary surgery. Rather, variations reflect differing medical opinion on appropriate use. Without evaluation, excessive use cannot be distinguished from underservice. Morbidity explains the variability of nondiscretionary surgery and conditions related to lifestyle. Access plays an important role for discretionary surgery. Geographic analysis can identify variation and relate incidence to socioeconomic and specific local effects. Hospital data do not permit direct assessment of appropriate care. Understanding the reasons for variation requires information beyond incidence data. The challenge is to identify and explain small area variations or to fix them.

This research uses a nationally representative sample of 1,425 low-income elders from the 1987 National Medical Expenditure Survey (NMES) to assess the independent and interaction effects of health insurance coverage and residence on the use of seven health care services: doctor visits, visits to other personnel, telephone contact, emergency room visits, overnight hospital stays, outpatient visits, and prescription medicines. It is hypothesized that (a) elders without insurance to supplement Medicare; (b) those who lived in nonmetropolitan areas use fewer health services; and (c) insurance status and residence interact to influence use of health services, with nonmetropolitan elders using fewer services than those living in metropolitan areas, regardless of insurance type. Using multivariate statistical analyses with both main effects and interaction term models, the data indicate that the type of insurance that low-income elders have is associated with their use of health services, and that residence has only modest effect. Individuals who have Medicaid, and to a lesser extent private supplements, use services more frequently than do those without supplements. Finally, despite the hypothesis, residence does not interact with insurance status in influencing use of services; the relationship between insurance and use of services does not vary across area of residence. The data suggest that the ability to pay, rather than supply constraints associated with nonmetropolitan areas, are of primary importance in determining health care use among low-income elders.

This study performed a surveillance of a defined population in central Wisconsin during a two-year period that included six months of follow-up. The study included those who worked and lived on farms and those who only worked on farms. The injury rate for farm residents was 3.15 injuries per 100 persons per year. However, 195 of the 510 injuries (38.2%) involved persons who were not farm residents. For nonfarm residents, 88.7 percent of the victims were male, whereas for farm residents 79 percent of the victims were male. More than one half of the farm residents who were injured were the owners and an additional 21 percent were juvenile children. Of those aged 18 to 29 years, 51.3 percent were nonfarm resident victims and 15.2 percent were farm resident victims. Of nonfarm residents, 41.1 percent were married; of farm residents, 65.3 percent were married. More than one half of all injured nonfarm residents were seen within two hours of injury while one quarter of the injured farm residents were seen more than 24 hours after the injury. Only 4 percent of the injured farm residents were self-insured, but 15 percent of the nonfarm residents were self-insured. However, the time of injury, actual injuries, diagnoses rendered, and services received were similar for injured farm residents and injured nonfarm residents. Effective attempts to intervene on agricultural injuries will need to consider different risk factors for injuries for nonfarm residents as compared to farm residents.

Objective: This article uses geographic information systems and their related tools to empirically measure and display the geographic accessibility of the aged population to hospital facilities within Illinois., Data Sources and Study Setting: Geographic accessibility of Illinois' aged population is measured from each of the state's 10,796 census block groups to the state's 214 hospital facilities. Block group demographic compositions and centroids are obtained from 1990 census files. Hospital coordinates are obtained by the authors., Study Design: Of five alternative measures of accessibility considered, empirical estimates are obtained for two: choice set and minimum distance. Access to both general hospitals and the subset having specialized geriatric facilities is measured with special attention to differences in accessibility between the aged within metropolitan statistical areas (MSAs) and those outside MSAs. Cumulative accessibility distributions and their summary statistics provide a basis of comparison among subgroups. DATA COLLECTION AND EXTRACTION: Geographic information systems (GIS) and their related tools are used as a means of efficiently capturing, organizing, storing, and retrieving the required data. Hospitals and census block groups are geocoded to specific locations in the database, and aspatial attributes are assigned to the hospitals and block groups. The GIS database is queried to produce shaded isarithm and point distribution maps that show the location of hospitals relative to surrounding aged populations., Conclusion: The vast majority of Illinois' aged population is within close proximity to hospital facilities. Eighty percent (1,147,504 persons) of the aged in Illinois are within 4.8 miles (7.7 km) of a hospital and 11.6 miles (18.7 km) of two hospitals. However, geographic accessibility differences between the aged living in MSAs and those living outside MSAs to hospitals offering geriatric services are substantial; but there is no evidence that the aged's geographical accessibility to hospitals is less favorable than that of the general population. Detailed accessibility measures permitted by geographic information system technology call into question the continued use of crude empirical accessibility measures.

This study examines the effect of 13 strategic management activities on the financial performance of a national sample of 797 U.S. rural hospitals during the period of 1983-1988. Controlled for environment-market, geographic-region, and hospital-related variables, the results show almost no measurable effect of strategic adoption on rural hospital profitability and liquidity. Where statistically significant relationships existed, they were more often negative than positive. These findings were not expected; it was hypothesized that positive effects across a broad range of strategies would emerge, other things being equal. Discussed are possible explanations for these findings as well as their implication for a rural health policy relying on individual rural hospital strategic adaptation to environmental change.

Using the records of 2,171 rural residents of Illinois who received inpatient treatment for mental illness or substance abuse, this paper examines factors that influence the tendency to seek service from a distant rather than a local hospital. Results indicate that the age and insurance coverage of the individual, the per capita income of the community area, surrogates for the service orientation of the local hospital and the proximity of the patient's residence to an urban center are significant influences. With the exceptions of drug abuse requiring detoxification or other symptomatic treatment, drug abuse accompanied by comorbidity and psychosocial disorders, psychosis, and childhood disorders, the primary diagnosis of the individual failed to have a significant effect on the propensity to bypass local sources of inpatient treatment.

Utilization of surgical services by rural citizens is poorly understood, and few data are available about rural hospitals' surgical market shares and their financial implications. Understanding these issues is particularly important in an era of financially stressed rural hospitals. In this study information about rural surgical providers and services was obtained through telephone interviews with administrators at Washington state's 42 rural hospitals. The Washington State Department of Health's Commission Hospital Abstract Recording System (CHARS) data were used to measure market shares and billed charges for rural surgical services. ZIP codes were used to assign rural residents to a hospital service area (HSA) of the nearest hospital, providing the geographic basis for market share calculations. "Total hospital expenses" from the American Hospital Association Guide were used as a proxy for hospital budget, and the surgical financial contribution was expressed as a ratio of billed surgical charges to total hospital expense. For rural hospitals as a whole, 21 percent of admissions and 43 percent of billed inpatient charges resulted from surgical services. In 1989, 27,202 rural Washington residents were hospitalized for surgery. Overall, 42 percent went to the closest rural hospital, 14 percent went to other rural hospitals, and 44 percent went to urban hospitals. The presence of surgical providers markedly increased local market shares, but a substantial proportion of basic surgical procedures bypassed available local services in favor of urban hospitals. For example, about one-third of patients needing cholecystectomies, a basic general surgery of low complexity, bypassed local hospitals with staff surgeons.(ABSTRACT TRUNCATED AT 250 WORDS)

Objective: We test the hypothesis that hospital costs, after adjusting for DRG mix, are higher in distant patients than in local patients., Data Sources and Study Setting: Data were obtained from the Washington State Commission Hospital Abstract Reporting System (CHARS) and included all patients discharged from 15 metropolitan hospitals in the state of Washington during fiscal year 1987 (N = 181,072)., Study Design: Distant patients were initially defined as those patients residing outside a 15-mile radius of the hospital from which they were discharged; all other patients were considered local. Distance was determined using the patient's residence zip code. Hospital charge, calculated for all patients regardless of payer, served as a proxy for cost and was adjusted using the DRG weight., Principal Findings: Average charge (adjusted for DRG weight) was higher for distant patients in all but two hospitals. Overall adjusted charge for distant patients was 15 percent higher (p < .001). This finding persisted when different distances were used to dichotomize distant and local patients. When the 20 most common DRGs were examined individually, little charge difference was found in surgical DRGs that require tertiary center services (tertiary DRGs) and in those DRGs with both moderate and predictable resource use (routine DRGs); the charge difference seemed most prominent in those DRGs with a wide array of possible resource use (heterogeneous DRGs)., Conclusions: Results suggest that patients traveling long distances use more resources and incur higher hospital charges than local patients. This is not accounted for in prospective payment. We postulate that distance might serve in part as a proxy for severity-of-illness.

Objective: This study explores the impact of multiple-site practices on the distribution of physician services within a medical service region., Data Sources and Study Setting: A questionnaire was mailed to all urologists (100 percent response rate) practicing in north central Connecticut (the Hartford medical service area) and adjacent communities in September 1990. Data on community characteristics were obtained from the 1990 U.S. census and state government documents., Study Design: Descriptive statistics and maps were used to summarize the attributes of single- and multiple-site practices and the communities where they were located. Key practice and community variables were analyzed., Data Collection/extraction Methods: The questionnaires were coded and entered into a digital database with the tabulated community data. Responses of individual physicians were grouped by practice., Principal Findings: Multiple-site practices were common. Second-order sites accounted for 23 percent of total appointment capacity and were located in communities with higher than average elderly populations and incomes and lower than average minority populations., Conclusions: Analysis of multiple-site practices is important for the accurate assessment of medical service availability. Further research is needed to document the functioning of multiple-site practices across other specialties and geographic areas.

Objective: To provide a radius measure of the structure of local hospital markets that varies with hospital characteristics and is available for all hospitals in the United States., Data Sources: 1982 American Hospital Association (AHA) Survey of Hospitals, 1982 Area Resource File (ARF), and 1983 California Office of Statewide Health Planning and Development (OSHPD) discharge abstracts., Study Design: The OSHPD data were used to measure the radii necessary to capture 75 percent and 90 percent of each hospital's admissions. These radii were used as the dependent variables in regression models in which the independent variables were from the AHA and ARF. To estimate predicted market radii, the estimated parameters from the California models were applied to all nonfederal, short-term, general hospitals in the continental United States. These radii were used to define each hospital's service area, and all other hospitals within the calculated radii were considered potential competitors. Using this definition, we calculated two measures of local market structure: the number of other hospitals within the radius and a Herfindahl-Hirschman Index based on the distribution of hospital bed shares in the market., Data Extraction Methods: These measures were calculated for all nonfederal, short-term, acute care hospitals in the continental United States for whom complete data were available (N = 4,884)., Conclusions: These measures are available from the authors on computer-readable diskette, matched to hospital identifiers.

Objective: This article conceptually and empirically evaluates alternative index measures that have been used to distinguish among hospital caseloads. It introduces two new measures., Data Sources/study Setting: The study relies on 1987 Medpar data, which provide a 100 percent sample of Medicare Part A claims for the calendar year., Study Design: Descriptive statistics indicate the sensitivity of alternative caseload measures to hospital bed size, region, and urban/rural location. Multiple regression analysis then examines the ability of the caseload measures to distinguish among hospitals based on hospital- and area-specific characteristics., Data Collection/extraction Methods: A provider level file containing the number of cases treated by each provider in each DRG was constructed from the Medpar data and merged with data from the American Hospital Association and the Area Resource File., Principal Findings: Different indexes purporting to measure hospital specialization are often evaluating very different aspects of the hospitals' caseloads. Prior work has indicated a specialization among hospitals during the period from 1980 to 1985. Replication of this work using other indexes could verify the increase in specialization and might provide a clearer picture of market or hospital characteristics associated with changing caseloads.

This study examines the issues associated with enhancing the representation of rural areas in the National Medical Expenditure Survey (NMES). It reviews alternative definitions of rural areas and describes the classification schemes considered for future cycles of the NMES. A review of the NMES sample representation in rural areas and an examination of the analytical capabilities and limitations of the current design for deriving health care estimates of the rural population is provided. The options presented in this article would extend the analytical capacity of the NMES to permit estimates of rural areas of the national classified by census region, by gradations of rural classification, and for individuals residing in rural frontier counties.

Objective: The purpose of this study is to examine the geographic scope of rural hospital markets., Data Sources: The study uses 1988 Medicare patient discharge records (MedPAR) and hospital financial information (HCRIS) for all rural hospitals participating in the Medicare Program., Study Design: Hospital-specific market areas are compared to county-based market areas using a series of geographic and socioeconomic-demographic dimensions as well as indicators of market competitiveness. The potential impact of alternative market configurations on health services research is explored by estimating a model of rural hospital closure., Data Collection/extraction Methods: Hospital-specific market areas were defined using the zip code of patient origin. Zip code-level data were subsequently aggregated to the market level., Findings: Using the county as the hospital market area results not only in the inclusion of areas from which the hospital does not draw patients but also in the exclusion of areas from which it does draw patients. The empirical estimation of a model of rural hospital closure shows that the definition of a hospital market area does not jeopardize the ability to identify major risk factors for closure., Conclusions: Market area definition may be key to identifying and monitoring populations at risk from rural hospital decisions to downsize or close their facilities. Further research into the market areas of rural hospitals that have closed would help to develop alternative, and perhaps more relevant, definitions of the population at risk.

Research was undertaken to determine the effects of hospital ownership, location, and Medicare's prospective payment system (PPS) on inpatient uncompensated care. A nonequivalent group design was used with repeated measures of uncompensated care (UNCC) on 137 system hospitals taken pre- and post-PPS. Investor-owned system hospitals demonstrated the largest increase in UNCC (37 percent) under the PPS. Results suggest that not-for-profit and investor-owned system hospitals are becoming more similar in levels of uncompensated care provided and that the PPS has had a negative effect on rural hospital profitability.

This set of six manuscripts describes the content and impact of the WAMI Rural Hospital Project (RHP), a research and development effort supported by the W.K. Kellogg Foundation, designed to improve the delivery of health services in six rural communities in the Pacific Northwest and Alaska. The major objective of the RHP--an activity which spanned a four-year period from 1985 through 1988--was to assist the project communities in improving the financial stability and quality of care of their local health care systems. Special attention was directed at helping the communities determine and implement an appropriate scope of health services, improve management and governance of the local health care enterprise, recruit and retain additional health personnel, and increase the extent to which community residents used local health services. In this first section we discuss the historical antecedents and conceptual underpinnings of the RHP and describe the five principal phases of the project. These include: (1) selection of communities for participation in the RHP, (2) comprehensive analysis of the health care system in each community, (3) community health services planning, including the development of comprehensive strategic plans, (4) implementation of techniques to improve local health services, and (5) project dissemination and evaluation. A pre-test, post-test model was employed to assess qualitative and quantitative changes in a variety of key measures of health system performance, including organization and management, scope of services, fiscal viability of the rural hospital, and utilization and patient satisfaction with health services in each community. The results of this evaluation constitute the balance of this report.

A major goal of the Rural Hospital Project (RHP) was to assist communities in defining an optimal scope of hospital and community health services. It was hypothesized that a rational basis for service planning would result in an expansion of locally provided health services, increased local hospital and physician market share, improved hospital workload performance, and higher levels of consumer satisfaction with community-based services. However, given the recent decline in performance of many small rural hospitals in general and in RHP hospitals in particular, at a minimum, stabilization of these troubled facilities could be considered a successful outcome. Data were collected from the six rural communities participating in the RHP both before and after the intervention (1985 and 1989) to assess changes in community scope of health services and utilization patterns. Comparative data were also compiled from peer group hospitals when available. Results generally demonstrated stabilization or expansion in: (1) the range of community and hospital services, (2) the availability of community physicians and visiting specialties, and (3) physician and hospital market share. While findings were mixed for patient days, average daily census, and number of births, substantial increases were documented for the number of surgical procedures, emergency room visits, and x-rays over the study period. RHP hospitals generally outperformed peer group hospitals on market share measures.

Rural health care facilities commonly employ community health care surveys as marketing research instruments to assess consumer utilization of and satisfaction with local services. However, there is little information on the use of survey findings as a way to design interventions to enhance consumer satisfaction and hospital viability. A community survey was administered in six Northwest rural communities as part of the Rural Hospital Project (RHP) to identify weaknesses in local health care services, guide remedial activities, act as a catalyst for change, and assess changes in community perceptions following project interventions. Descriptive findings revealed problems typically observed in small rural communities, including relatively low hospital and physician market share, outmigration for certain types of health care not available locally, and dissatisfaction with some aspects of hospital and physician services. Satisfaction with various aspects of care tended to be lower among males, the uninsured, and younger respondents. Comparisons of survey responses before (1985) and after (1989) the RHP generally demonstrated stability in hospital and physician market share, with project hospitals performing well in 1989 in comparison to other rural hospitals of similar size. The percentage of respondents who rated overall quality of local hospital and physician care positively generally increased or remained stable over the study period. Substantial decreases in satisfaction levels were found for access to care. Importantly, gains were made in those areas and services which received particular emphasis in the project.

Uniform hospital discharge abstract data from Maryland were used to examine the homogeneity of trauma-related DRGs with respect to a well-established measure of injury severity, the Injury Severity Score (ISS). Thirty DRGs were identified as including trauma cases with a wide range of severity; for each of these DRGs, ISS explains a significant amount of variation in length of stay. By applying statistical techniques similar to those used to create the original DRG groupings, these 30 DRGs were subdivided by severity and age categories to create a new set of severity-modified DRGs. The potential effects of using DRGs and modified DRGs to pay for inpatient care within the Maryland state regionalized system of trauma care were examined. Payments based on regional averages per DRG and per modified DRG were compared to actual hospital charges regulated by the state's Health Services Cost Review Commission. Using average charges per DRG as a basis of payment, approximately !1.4 million (11 percent of total hospital charges) would be shifted from trauma centers to nontrauma centers. This shift represents an 18 percent loss in revenues to trauma centers and a 30 percent gain in revenues to nontrauma centers. Using a payment system based on severity-modified DRGs, trauma centers would still experience a net loss in revenues and the nontrauma centers a net gain, but the total amount of the shift would be reduced from $11.4 million to $9.8 million. The results argue for the need to explore alternative payment systems not strictly based on current DRGs. Because of DRGs do not adequately reflect severity differences, using them to pay hospitals will create financial incentives that discourage regionalization of trauma care.

As part of a larger study of hospital choice, the travel patterns of more than 12,000 Medicare beneficiaries residing in three overlapping rural areas were examined. During 1986 these Medicare beneficiaries were admitted to one of 53 hospitals in an area that encompassed parts of Minnesota, North Dakota, and South Dakota. Information on ZIP code of residence, closest hospital, and hospital of admission were used to analyze hospital choices of the Medicare rural elderly residing in this area. To summarize their travel patterns, the admitting hospital was categorized based on whether it was urban or rural, its size and whether or not it was the closest facility. Findings indicated that 60 percent of these rural Medicare beneficiaries used hospital services at their closest rural hospital, regardless of its size. However, 79 percent of those whose closest hospital was larger than 75 beds used it, while only 54 percent of those whose closest rural hospital was fewer than 75 beds obtained services there. Overall, 30 percent of those residing in this rural market area went to an urban hospital. These patterns appeared to reflect an evaluation by the physician and/or individual of the relative attractiveness of the local hospital versus alternatives available, as well as the individual's characteristics. Travel patterns varied by the beneficiary's age as well as his or her relative complexity of illness, as measured by a Disease Staging methodology. Findings have implications for the provision and financing of hospital services in rural areas.

The Graduate Medical Education National Advisory Committee (GMENAC) projected the need for and supply of physicians and other providers, recommended time and access standards for health care services, and developed guidelines for the geographic distribution of physicians. Since this study, analysts have given scant attention to national problems of physician geographic distribution. The issue deserves additional scrutiny in light of the current continuing problems of underservice in rural areas. The emergence of geographic information systems offers a unique opportunity to acquire data on provider distribution and provide a framework for developing and testing redistribution policy.

As physicians and other providers of health care services see their traditional markets erode, an increasingly important element of any provider location decision is the determination of a population base or "critical mass" that can professionally and financially support a given set of health care services. While the size of a local population is not the sole determinant of success, ultimately an adequate population base to support a given spectrum of services must be defined, and providers increasingly need tools for evaluating opportunities in the new economic market. This is especially true in rural areas. An earlier supply and demand model for estimating the critical mass of population needed to support a physician in any one of 25 specialties and subspecialties in urban and suburban areas is adapted to the rural market. The assumptions inherent in the earlier model are examined and the issue of "critical mass" is examined from a rural health care perspective in this paper.

Measurement of the availability of health care providers in a geographic area is a useful component in assessing access to health care. One of the problems associated with the county provider-to-population ratio as a measure of availability is that patients frequently travel outside their counties of residence for health care, especially those residing in nonmetropolitan counties. Thus, in measuring the number of providers per capita, it is important that the geographic unit of analysis be a health service area. We have defined health care service areas for the coterminous United States, based on 1988 Medicare data on travel patterns between counties for routine hospital care. We used hierarchical cluster analysis to group counties into 802 service areas. More than one half of the service areas include only nonmetropolitan counties. The service areas vary substantially in the availability of health care resources as measured by physicians and hospital beds per 100,000 population. For almost all of the service areas, the majority of hospital stays by area residents occur within the service area. In contrast, for 39 percent of counties, the majority of hospital stays by county residents occur outside the county. Thus, the service areas are a more appropriate geographic unit than the county for measuring the availability of health care.

Many recent studies have demonstrated that hospital utilization rates vary widely across small geographic areas. The variation is often attributed to the style of practice of the provider. This study demonstrates that hospital utilization varies widely between "micro" areas within individual hospital market areas. Further, the study demonstrates that hospital utilization rates within a hospital market area are more similar to each other than to rates in "micro" areas within other hospital market areas. After adjustment for available demographic, socioeconomic, and epidemiological factors, the utilization rates within "micro" areas are highly related to the group of hospitals that dominates the market area. After simultaneously adjusting for age and poverty, the market share-dominant group explains 35 percent of the variance in surgical use rates of "micro" areas and 39 percent of the variance in medical use rates.

A small-area analysis (SAA) in health services research often calculates surgery rates for several small areas, compares the largest rate to the smallest, notes that the difference is large, and attempts to explain this discrepancy as a function of service availability, physician practice styles, or other factors. SAAs are often difficult to interpret because there is little theoretical basis for determining how much variation would be expected under the null hypothesis that all of the small areas have similar underlying surgery rates and that the observed variation is due to chance. We developed a computer program to simulate the distribution of several commonly used descriptive statistics under the null hypothesis, and used it to examine the variability in rates among the counties of the state of Washington. The expected variability when the null hypothesis is true is surprisingly large, and becomes worse for procedures with low incidence, for smaller populations, when there is variability among the populations of the counties, and when readmissions are possible. The characteristics of four descriptive statistics were studied and compared. None was uniformly good, but the chi-square statistic had better performance than the others. When we reanalyzed five journal articles that presented sufficient data, the results were usually statistically significant. Since SAA research today is tending to deal with low-incidence events, smaller populations, and measures where readmissions are possible, more research is needed on the distribution of small-area statistics under the null hypothesis. New standards are proposed for the presentation of SAA results.

The purpose of this study was to explore how the primary care (PC) physician-to-population ratio changed from 1975 to 1985 in counties that were served by an Area Health Education Center (AHEC) in contrast to those counties that were not. The investigation attempted to determine whether any observed changes in this ratio were dependent upon either degree of urbanization or contiguity to a Standard Metropolitan Statistical Area (SMSA). The data source for this study was the Area Resource File. Results indicated that: (a) irrespective of AHEC status, increased degree of urbanization across counties was associated with relatively more PC physicians in both 1975 and 1985; (b) PC physician-to-population ratios increased from 1975 to 1985 for all American Medical Association (AMA) county code categories, regardless of AHEC versus non-AHEC designation; (c) AHEC counties demonstrated greater (or equivalent) absolute improvement for all AMA county code categories except for several categories which represented more urbanized counties; (d) for the least urbanized counties in the AMA code, the percentage improvement in PC physician-to-population ratios for AHEC counties ranged from 3 to 5 percent higher than corresponding percentage improvement in ratios for non-AHEC counties; and (e) AHEC counties showed greater absolute and percentage improvement in PC physician-to-population ratios than did non-AHEC counties for counties not contiguous to SMSAs.

This article reviews the current small-area variation analysis (SAVA) approach to population-based rates of surgery, and describes a new method for ascertaining variance based on the beta-binomial probability distribution of small-area rates. The critical review of the current SAVA approach focuses (1) on how incidence rates are calculated, and (2) on how the significance of the observed magnitude between the largest and smallest rates (i.e., the external quotient) is ascertained. While reducing the problems of calculating rates by considering only certain operative procedures, the new method addresses the current inadequacies of ascertaining significant differences among small areas. Not only does it correctly assess likelihood of an extermal quotient, it also can determine the particular area's rate, producing an unlikely extermal quotient. The method evaluates the probability that the observed magnitude of the extremal quotient is due solely to chance and study design effects, and tables of these probabilities are available for the method's application. A mathematical model, based on a combination of the binomial and beta distributions, uses (1) the sample size, (2) the average of the areas' rates, (3) the variance among the rates, and (4) a specific quotient level to determine the probability of observing the quotient by chance. After computerizing this calculation, probability tables for reasonable values of these four parameters are generated. In addition to looking at just one quotient for each sample, the probability tables facilitate the easy examination of intermediate quotients when the extremal quotient is unlikely due to chance. By alternatively ignoring the highest and lowest rates, two new quotients can be produced and tested. Given that one of these two quotients is likely due to chance, the excluded rate (i.e., producing the unlikely extremal quotient) can be classified as an outliner, and the associated small area should be the focus of more detailed investigation. The probability tables reveal that the external quotient is not the appropriate statistic to be applied in studies where many small areas are to be included. The probability of seeing even a "large" extremal quotient simply by chance rapidly approaches one as the sample size increases. However, an extremal quotient modeled from a beta-binomial distribution can be useful for studies with small sample sizes (e.g., six counties). The use of this beta-binomial model for small-area rates provides a new method of designing and evaluating small-area studies where costs or domain limit the number of areas under consideration.(ABSTRACT TRUNCATED AT 400 WORDS)