Skip Navigation U.S. Department of Health and Human Services www.hhs.gov
Agency for Healthcare Research Quality www.ahrq.gov
Archive print banner

This information is for reference purposes only. It was current when produced and may now be outdated. Archive material is no longer maintained, and some links may not work. Persons with disabilities having difficulty accessing this information should contact us at: https://info.ahrq.gov. Let us know the nature of the problem, the Web address of what you want, and your contact information.

Please go to www.ahrq.gov for current information.

Chapter 7 - Outcomes and Safety Net Performance

Introduction

To date, most discussions of safety nets have been largely descriptive, usually providing details on the composition or structure of area providers and data on the need or demand for services. In this book, we provide a more comprehensive look at safety nets, including information on demand, support, structure, and community context. In this chapter, we link this information to a set of indicators on outcomes and performance of the safety net. As the attention of policymakers increasingly turns to efforts to strengthen local safety nets, understanding more about the relationship of these factors to outcomes and performance is likely to be useful in making decisions concerning the allocation of scarce resources.

Obtaining measures of outcomes and performance of the safety net is difficult, especially when attempting to provide uniform measures across multiple geographic areas. The number of potential outcome measures for vulnerable populations is relatively small to start, and this data book is limited by two additional factors: (1) the need to use readily available data (i.e., no new data were collected) and (2) the desire to show the same measures across all sites. Therefore, the analysis focuses on three types of measures:

  • Preventable/avoidable hospitalizations from hospital discharge data sets.
  • Birth outcomes from vital statistic records.
  • Self-reported access indicators from the National Health Interview Survey.

Preventable/avoidable hospitalizations involve ambulatory care sensitive (ACS) conditions, for which access to timely and effective ambulatory care can help prevent the need for inpatient care. There are three types of ACS conditions: (1) chronic conditions (such as diabetes, asthma, and congestive heart failure), for which effective management can prevent serious flareups; (2) acute conditions (such as ear infections, gastroenteritis, and cellulitis), for which early intervention can prevent more serious progression of the condition; and (3) preventable illnesses (such as pertussis, tetanus, and rheumatic fever), for which immunization can prevent the disease. The list of ACS conditions includes a broad range of diagnoses across age groups, body systems, and genders; the full list is shown in "Appendix A: Technical Information." Because access problems differ by age, analysis is provided separately for children ages 0-17, young adults ages 18-39, and older adults ages 40-64. Because the ACS condition list was developed for populations under age 65, this book provides no analysis for the elderly. The databases used for the ACS conditions include the patient's ZIP Code of residence, enabling analysis at all levels included in this book (Metropolitan Statistical Area [MSA], county, city, and county residual).

For these conditions, higher use rates for a population subgroup or geographic area can be an indication of access problems or concerns about performance of the safety net. It is important to note that these admissions are not necessarily "inappropriate" in the sense of being unnecessary or unwarranted. They are simply conditions that effective ambulatory care might have prevented from becoming so severe that admission is perceived to be necessary. And, of course, not all hospital admissions for ACS conditions are preventable or avoidable. In some cases, the best possible care cannot prevent progression of the condition to the stage that requires hospitalization. Vital statistic records provide data on both process measures and outcome-related measures for births, a critical concern for any health care system. Included in this book are (1) late/no prenatal care for deliveries (no prenatal care or prenatal care initiated in the third trimester); (2) low birth weight full-term births (newborns weighing less than 2,500 grams and with a gestation period of 37 weeks or longer); and (3) preterm births (deliveries before 37 weeks gestation). While these data are available at most geographic levels, they are suppressed at the county level and for many smaller cities to protect confidentiality.

Self-reported access measures from the National Health Interview Survey provide another important perspective on barriers to care in many communities. Included among the measures in this book are (1) having no usual source of care, (2) being unable to obtain needed care, (3) having no doctor visit in the last year, and (4) having no doctor visit in the last 2 years. These data are presented for populations with incomes below 200 percent of the Federal poverty line, emphasizing those Americans who are most likely to experience access to care problems. Because of small sample sizes, data for these measures are available only at the MSA level and only for a subset of 34 of the largest MSAs included in this book.

Return to Contents

Variation in Outcomes and Safety Net Performance

Given the enormous variation documented in the demand, support, structure, and context of local safety nets, it is not surprising that wide differences exist in these outcome and performance measures. These differences are particularly pronounced for ACS conditions. Overall, potentially preventable hospitalization rates for all age groups are substantially higher in central city areas than in suburban counties (Go to Table 7-1). Rates for central city areas are 39 percent higher than those for suburban areas for children and 55 percent higher for older adults. The somewhat more equal geographic distribution of preventable hospitalizations for children suggests that access-related problems are more severe for adults than for children, perhaps reflecting positive results from the investment in support for children's health care services through the State Children's Health Insurance Program (SCHIP) and Medicaid.

Table 7-1: Preventable Hospitalizations by Area Type and Region, 1999

Area Average Preventable Hospitalization Rate
Ages 0-17 Ages 18-39 Ages 40-64
MSA9.886.7318.83
Suburban County8.665.8315.78
City12.038.2024.44
Northeast11.536.9720.12
South10.667.6520.14
Midwest10.157.4920.97
West7.705.1815.58
All Areas9.836.6618.83

There are large differences in physician practice style across the country, with generally lower utilization levels on the West Coast. While our rates attempt to account in part for these differences by adjustment based on admissions for a set of conditions with wide variation in admission rates but no differences based on income, some of the differences observed may relate to practice style (go to Appendix A: Technical Information). This may explain the somewhat lower rates for areas in the West, and all data should be examined carefully to consider how practice style might influence underlying rates.

For ACS conditions, there are also large differences within area types and regions, as shown in Table 7-2. For example, even among suburban counties, the practice style-adjusted potentially preventable hospitalization rates for children ranged from 3.1 per 1,000 for counties in suburban Portland, OR, and Seattle, WA, to 25.6 per 1,000 in suburban Richmond, VA. Similar large differences exist for the central cities, ranging from 5.6 per 1,000 in Portland, OR, to 28.3 per 1,000 in Jersey City, NJ, and 29.2 per 1,000 in Newark, NJ. At the MSA level, differences among areas are somewhat diluted by the mix of central cities and suburbs in the data, but large differences remain for both children and adults, as illustrated in Figures 7-1 and 7-2.

Table 7-2: Preventable Hospitalizations per 1,000 Children Ages 0-17 by Area Type and Region, 1999

Area Range of Variation Average
Index* High Low High/Low Rate
MSA0.30525.344.965.119.88
Suburban County0.32925.613.108.278.66
City0.38729.214.336.7412.03
Northeast0.44229.213.578.1811.53
South0.31825.613.706.9210.66
Midwest0.36019.813.415.8210.15
West0.25217.543.105.667.70
All Areas0.40029.213.109.439.83

* Coefficient of variation-an index that measures the amount of variation (higher = more variation).

Figure 7-1: Preventable/Avoidable Hospitalizations Per 1,000 Children Ages 0-17, Metropolitan Areas, 1999

Figure 7-1: Preventable/Avoidable Hospitalizations Per 1,000 Children Ages 0-17, Metropolitan Areas, 1999
[D] Select for Text Description.

Figure 7-2: Preventable/Avoidable Hospitalizations Per 1,000 Adults Ages 40-64, Metropolitan Areas, 1999

Figure 7-2: Preventable/Avoidable Hospitalizations Per 1,000 Adults Ages 40-64, Metropolitan Areas, 1999
[D] Select for Text Description.

As shown in Table 7-3, there were only small regional differences in birth indicators. However, central cities have 76 percent higher rates for late/no prenatal care than suburban counties, 32 percent higher rates of low birth weight full-term births, and 13 percent higher rates of preterm births. Again, large differences exist within regions. Figure 7-3 illustrates this variation at the MSA level. Some of the highest levels of late/no prenatal care exist in the Tucson, AZ (8.3 percent), and Las Vegas, NV-AZ (7.7 percent), MSAs, with the lowest rates observed in the Portland, ME (1.3 percent), Providence-Fall River-Warwick, RI-MA (1.8 percent), and Ventura, CA (1.6 percent), MSAs.

Table 7-3: Birth Outcomes by Area Type and Region, 1999

Area Average for Area
Late or No Prenatal Care (%) Low Birth Weight
Full-Term Births (%)
Preterm Births (%)
MSA3.72.411.6
Suburban County2.92.211.0
City5.12.912.4
Northeast4.42.611.3
South3.22.712.8
Midwest3.72.512.0
West3.82.110.6
All Areas3.82.511.6

Figure 7-3: Percent of Total Births with Late or No Prenatal Care, Metropolitan Areas, 1999

Figure 7-3: Percent of Total Births with Late or No Prenatal Care, Metropolitan Areas, 1999
[D] Select for Text Description.

Several survey-reported measures of access show substantial regional differences, as displayed in Table 7-4. For the six northeastern MSAs with sufficient data for analysis, 12.7 percent of the low-income population report that they have no usual source of care, compared with 25.5 percent for the 12 MSAs with data in the western United States. As illustrated in Figure 7-4, substantial differences exist within regions, with a more than threefold difference between the highest and lowest areas in each region. Among the MSAs for which we have data, the highest rates of lacking a usual source of care occur among low-income populations in West Palm Beach-Boca Raton, FL (38.8 percent), and Ventura, CA (39.3 percent), while the lowest rates occur in Philadelphia, PA-NJ (7.8 percent), and Boston, MA-NH (6.5 percent).

Table 7-4: Survey Measures by Area Type and Region, 1999-2000a

Area Average for Low-Income Populations in Area
No Usual Source of Care (%) No Doctor Visit in Last Year (%)
Northeast12.712.4
South22.520.4
Midwest20.719.9
West25.525.4
All Areas20.820.6

a Survey data are available for only 34 MSAs.

Figure 7-4: Percent of Population with No Usual Source of Care, Households with Incomes Below 200 Percent of Poverty 34 Metropolitan Areas, 1999-2000

Figure 7-4: Percent of Population with No Usual Source of Care, Households with Incomes Below 200 Percent of Poverty 34 Metropolitan Areas, 1999-2000
[D] Select for Text Description.

Return to Contents

How the Outcome Measures Are Related to One Another

As discussed in Chapter 1, this book includes multiple outcome and performance measures because of the complex and interrelated nature of the various factors contributing to access problems and socioeconomic disparities in health outcomes. Each indicator potentially measures a different aspect of the problem. For example, access barriers for children are fundamentally different from those for adults, given differences in eligibility for publicly supported programs and the focus of direct service initiatives for vulnerable populations. Similarly, pregnant women face access barriers potentially different from those faced by women who are not pregnant, again because of the nature of Medicaid eligibility and outreach initiatives in many communities to improve birth outcomes.

Not surprisingly, then, geographic areas may perform better on some indicators than on others. For example, as illustrated in Figure 7-5, there is only moderate association between preventable hospitalization rates for young adults and the rates of late/no prenatal care at the city and county level (R2 = .290). These data suggest that the barriers to timely and effective routine ambulatory care for young adults differ from those to obtaining prenatal care. While the association between potentially preventable hospitalization rates for children and adults at the city and county level is considerably stronger (R2 = .613), some areas have better outcomes for children and worse outcomes for adults (or vice versa). Again, this finding suggests that access problems and the performance of the safety net differ for adults and children. The association in preventable hospitalization rates for young adults and older adults is quite strong (R2 = .858), suggesting that both groups have similar access problems and that older adults' experience in navigating the health care system is not sufficient to obviate serious barriers to care.

Figure 7-5: Preventable/Avoidable Hospitalizations (Adults Ages 18-39) and Percent of Births with Late or No Prenatal Care Cities, County Residuals, and Suburban Counties, 1999

Figure 7-5: Preventable/Avoidable Hospitalizations (Adults Ages 18-39) and Percent of Births with Late or No Prenatal Care Cities, County Residuals, and Suburban Counties, 1999
[D] Select for Text Description.

Interestingly, little or no association exists between either preventable hospitalization rates or birth measures and the self-reported indicators of access to care from the National Health Interview Survey. While some of the lack of association may be attributable to the small number of MSAs for which the survey data are available and to the lack of geographic resolution of any analysis at the MSA level (go to Chapter 2), these indicators may be measuring different aspects of the access to care problem. Preventable hospitalizations and birth outcomes are quasi-outcome measures that may be affected by a complex array of factors, including insurance status, care-seeking behavior, and the performance of the health care delivery system. Survey measures such as having a usual source of care may be more sensitive to "front door access" (are services available?) and less influenced by how well these services perform or by the care-seeking behavior of patients.

Return to Contents

How Outcome Measures Are Related to Demand, Support, Structure, and Context Measures

As noted in previous sections, there are some very strong associations between many of the outcome measures and some of the individual demand, support, structure, and contextual indicators. For example, in Chapter 3, there was a very strong association between preventable hospitalization rates for older adults and area poverty rates (R2 = .531), with more moderate associations observed for preventable hospitalization rates for children (R2 = .341). Figures 7-6 and 7-7 illustrate these relationships for children and older adults, respectively. Similar strong associations exist between poverty levels and birth outcomes as well as between race/ethnicity and both potentially preventable hospitalization rates and birth outcomes.

Figure 7-6: Preventable/Avoidable Hospitalizations (Children Ages 0-17) and Percent of the Population Below Poverty Cities, County Residuals, and Suburban Counties, 1999

Figure 7-6: Preventable/Avoidable Hospitalizations (Children Ages 0-17) and Percent of the Population Below Poverty Cities, County Residuals, and Suburban Counties, 1999
[D] Select for Text Description.

Figure 7-7: Preventable/Avoidable Hospitalizations (Adults Ages 40-64) and Percent of the Population Below Poverty Cities, County Residuals, and Suburban Counties, 1999

Figure 7-7: Preventable/Avoidable Hospitalizations (Adults Ages 40-64) and Percent of the Population Below Poverty Cities, County Residuals, and Suburban Counties, 1999
[D] Select for Text Description.

Other demand, support, structure, and contextual indicators show varying levels of association with the different outcome and performance measures. However, these other indicators are often strongly associated with each other, and each has a different potential causal relationship with the outcome and performance measures in this book. Accordingly, interpreting the results of these associations can be difficult. Multivariate analysis permits the assessment of the impact of multiple factors together to try to sort out how they relate to one another and to understand the extent to which they individually contribute to "explaining" differences in some outcomes.

Analysis using multivariate techniques yields interesting results that have potentially important implications for policymakers, analysts, and researchers. In conducting the analysis, we grouped several related individual indicators together (using factor analysis techniques), because many of our measures are closely related to one another--for example, area unemployment and area poverty level. As a result, several of the indicators relating to personal circumstances (poverty, unemployment, disability, high school or less education level, single-parent households, and living alone) were combined into a single variable referred to as "personal distress." Similarly, several indicators relating to the community context (crime rates, housing vacancy rates, age of housing, and home ownership) were combined to form "community distress."

The results of this multivariate analysis for preventable hospitalization rates and birth outcomes at the city, county, and county residual level are contained in Tables 7-5 and 7-6. Some important conclusions are:

Federal and State financing of the safety net helps.

Medicaid programs with a greater extent of coverage and higher disproportionate share hospital payments are generally associated with lower preventable hospitalization rates and better birth outcomes.

Public facilities matter.

For adults, a greater presence of public hospitals is associated with lower preventable hospitalization rates. A greater public hospital presence is also associated with lower rates of preterm births.

More providers is not always the answer.

While having more pediatricians is associated with lower preventable hospitalization rates for children, greater availability of adult primary care physicians has no association with preventable hospitalization rates for adults, and having more obstetrician/gynecologists has no impact on birth outcomes.8

Levels of personal distress are a concern.

Across all age groups, higher levels of poverty, unemployment, disability, low education, and social isolation are associated with higher levels of preventable hospitalizations and worse birth outcomes.

Race/ethnicity is a factor.

Across all age groups, larger black and Asian populations are associated with higher preventable hospitalization rates and worse birth outcomes. For older adults, larger Hispanic populations are also associated with higher preventable hospitalization rates.

The multivariate analysis also produces some surprising results. Despite concerns about the impact of managed care on the safety net, areas with higher managed care penetration actually experience generally lower preventable hospitalization rates and better birth outcomes. Fully interpreting this finding will require more analysis, preferably using information on the extent of Medicaid managed care rather than general managed care penetration rates. However, it may suggest that the competitive pressures of managed care may actually improve safety net performance, perhaps by encouraging safety net providers to be more responsive to patient demands in the face of potential loss of market share.

There is significant concern about the impact of immigration on local safety nets, since large, often uninsured, immigrant populations put a substantial strain on safety net resources. In general, however, higher levels of foreign-born populations either are associated with better outcomes or have no association with outcomes, perhaps due to the better health status of these populations. The exception is areas with larger immigrant populations that have higher children's preventable hospitalization rates, since learning how to navigate the U.S. health care system may influence care-seeking behavior of foreign-born parents.

Table 7-5: Multivariate Analysis of Community and Safety Net Characteristics on Patient Outcomes and Performance of the Safety Net: Preventable Hospitalizations in Cities, Suburban Counties, and County Residuals

Characteristics Preventable/Avoidable (ACS) Hospitalizations
Children Ages 0-17 Adults Ages 18-39 Adults Ages 40-64
Characteristics associated with
lower rates/ better outcomes
Greater extent of Medicaid coverage
More hospital outpatient capacity/use
Higher managed care penetration
More pediatricians
Greater concentration of low-income residents
Western U.S. residence
Higher level of disproportionate share hospital (DSH) payments
Greater extent of Medicaid coverage
More hospital outpatient capacity/use
Higher public hospital presence
Higher managed care penetration
Higher foreign-born population
Western U.S. residence
Eastern U.S. residence
Higher level of DSH payments
Greater extent of Medicaid coverage
Higher public hospital presence
Higher foreign-born population
Greater concentration of low-income residents
Western U.S. residence
Characteristics associated with higher rates/worse outcomes Greater levels of personal distress
Higher black population
Higher Asian population
Higher foreign-born population
Higher teaching hospital presence
Eastern U.S. residence
Greater levels of personal distress
Higher black population
Higher Asian population
Greater concentration of non-white residents
Greater levels of personal distress
Higher black population
Higher Asian population
Higher Hispanic population
Higher teaching hospital presence
Characteristics having no
association with outcomes
More community distress Higher level of DSH payments
Higher investor-owned hospital presence
Higher public hospital presence
Higher Hispanic population
Greater concentration of non-white residents
More community distress
Higher investor-owned hospital presence
Higher teaching hospital presence
Higher Hispanic population
Greater concentration of low-income residents
More primary care physicians
More community distress
More hospital outpatient capacity/use
Higher investor-owned hospital presence
Higher managed care penetration
Greater concentration of non-white residents
Eastern U.S. residence
More primary care physicians

Table 7-6: Multivariate Analysis of Community and Safety Net Characteristics on Patient Outcomes and Performance of the Safety Net: Birth Outcomes in Cities, Suburban Counties, and County Residuals

Characteristics Birth Indicators
Late/No Prenatal Care Low Birth Weight Full Term Preterm Births
Characteristics associated with lower rates/better outcomes Higher level of disproportionate share hospital (DSH) payments
Greater extent of Medicaid coverage
Higher managed care penetration
Higher foreign-born population
Greater extent of Medicaid coverage
Higher managed care penetration
Western U.S. residence
Higher level of DSH payments
Greater extent of Medicaid coverage
More hospital outpatient capacity/use
Higher public hospital presence
Higher managed care penetration
Higher foreign-born population
Eastern U.S. residence
Western U.S. residence
Characteristics associated with higher rates/worse outcomes Greater levels of personal distress
Higher teaching hospital presence
Higher black population
Eastern U.S. residence
Western U.S. residence
Greater concentration of low-income residents
Greater levels of personal distress
Higher investor-owned hospital presence
Higher teaching hospital presence
Higher black population
Higher Asian population
Greater levels of personal distress
Higher investor-owned hospital presence
Higher black population
Higher Asian population
Higher Hispanic population
Greater concentration of non-white residents
Characteristics having no association with outcomes More community distress
More hospital outpatient capacity/use
Higher investor-owned hospital presence
Higher public hospital presence
Higher Asian population
Higher Hispanic population
Greater concentration of non-white residents
More obstetrician/gynecologists
More community distress
Higher level of DSH payments
More hospital outpatient capacity/use
Higher public hospital presence
Higher Hispanic population
Higher foreign-born population
Eastern U.S. residence
Greater concentration of low-income residents
Greater concentration of non-white residents
More obstetrician/gynecologists
More community distress
Higher teaching hospital presence
Greater concentration of low-income residents
More obstetrician/gynecologists

Another unexpected finding relates to the impact of levels of community distress. Although the number of indicators included in this book is limited, the combined impact of crime rates, housing stock, housing vacancy rates, and home ownership has no association with preventable hospitalization rates or birth outcomes. More refined measures of community distress and more granulated analysis (e.g., at the neighborhood level) may be needed to clarify the impact of community distress on the safety net.

Finally, while the impact of investor-owned hospitals on the viability of local safety nets is a concern in many communities, higher investor-owned hospital presence has no association with preventable hospitalization rates or levels of late/no prenatal care. There is an association between a greater investor-owned hospital presence and higher levels of low birth weight and preterm births, suggesting that additional analysis will be required to understand the impact of these hospitals on the safety net.

8 The relationship between provider supply and preventable hospitalizations may vary by region. Refer to, for example, an analysis of New York State in Basu J, Friedman B, Burstin H. Primary care, HMO enrollment, and hospitalization for ambulatory care sensitive conditions: A new approach. Med Care 2002 Dec; 40(12):1260-9.

Return to Data Book Contents
Continue to Chapter 8

 

The information on this page is archived and provided for reference purposes only.

 

AHRQ Advancing Excellence in Health Care