PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of hsresearchLink to Publisher's site
 
Health Serv Res. 2004 October; 39(5): 1403–1428.
PMCID: PMC1361076

Associations of Race, Education, and Patterns of Preventive Service Use with Stage of Cancer at Time of Diagnosis

Abstract

Objective

To go beyond the documentation of disparities by race and SES by analyzing health behaviors regarding preventive and cancer screening services and determining if these behaviors are associated with stage of cancer when first diagnosed.

Data

Stage of cancer for Medicare patients diagnosed in 1995 with breast, colorectal, uterine, ovarian, prostate, bladder, or stomach cancer; and use of influenza and pneumonia immunization, mammography, pap smear, colon cancer screening, and the prostate specific antigen test during the two years preceding diagnosis of cancer.

Study Design

Hypothesis tested: health behaviors regarding use of preventive and cancer screening services are associated with stage of cancer when first diagnosed.

Data Collection/Extraction Methods

Information was extracted from the database formed by the linkage of Surveillance, Epidemiology, and End Results (SEER) cancer registries with Medicare files.

Principal Findings

Black and white patients (of higher and lower SES) who used more of the preventive and cancer screening services were at a lower risk of having late stage cancer for six cancers studied (breast, colorectal [male and female], prostate, uterine, and male bladder cancer) than their counterparts who used fewer of these services.

Conclusions

The use of preventive and cancer screening services is a health behavior associated with better health outcomes for the elderly diagnosed with cancer. The lack of preventive service use can serve as a marker for identifying persons at risk of late stage cancer when first diagnosed. Strategies that encourage the use of preventive services by low users of these services are likely to reinforce a range of healthy behaviors that help to ameliorate disparities in health outcomes.

Keywords: Health behaviors, race, socioeconomic status, disparities, preventive service use, stage of cancer

The passage of Medicare in 1965 was accompanied by a confident expectation that covering most of the costs of physicians' and inpatient hospital services would remove the major barriers to health care for the elderly. A lingering concern was the lack of coverage for prevention and early detection of disease. Over time, Medicare was amended to cover pap smears (1990), routine screening mammography (1991), influenza immunizations (1993), colorectal cancer screening (1998), and prostate cancer screening, including the prostate specific antigen (PSA) test (2000).

During the past decade new concerns have been raised by a succession of studies indicating that race and socioeconomic status (SES) continue to influence the use of health care services by the elderly enrolled in Medicare; these studies show three distinct patterns: black beneficiaries and persons less economically and socially advantaged use (1) fewer preventive and cancer screening services, (2) fewer common surgical procedures to improve health and functioning, but (3) more procedures associated with poor outcomes of chronic diseases than white beneficiaries and beneficiaries in higher socioeconomic status (Ayanian et al. 1993; Gornick 2000; Health Care Financing Administration 1995a; Gornick et al. 1996). Disparities in the use of preventive, cancer screening, and other services to maintain health are especially troubling because blacks and disadvantaged persons experience higher rates of morbidity and mortality (National Center for Health Statistics 2002).

Why substantial disparities persist in the use of preventive and cancer screening services, especially among the elderly covered by Medicare, remains a fundamental question. Although it is true that barriers relating to coinsurance requirements for services such as mammography may explain some of the disparities by race and SES, the greatest disparities are found in the use of flu shots, which are “free.” Moreover, about 90 percent of the elderly when surveyed respond that they have a usual source of care (Health and Health Care of the Medicare Population 2001).

The Institute of Medicine convened two committees to consider ways of promoting health, particularly among vulnerable subgroups; committee reports were in accord that no single factor can explain disparities in health—given the multitude of complex social and behavioral factors that can influence health outcomes (Smedley and Syme 2000; Committee on Health and Behavior 2001). Smedley and Syme (2000) noted that “behavioral and social interventions … offer great promise to reduce disease morbidity and mortality, but as yet their potential to improve the public's health has been relatively poorly tapped.” The Committee on Health and Behavior (2001) recommended research that “integrates biological, psychological, behavioral, and social variables.”

This study integrates into health services research some of the theories and findings from health psychology and the behavioral and social sciences. In this study we concentrate on a health behavior exemplified by a beneficiary's use of an array of covered preventive and cancer screening services (hereafter, described simply as “preventive services”). Health behaviors have been a subject of interest to health psychologists for a long time. Rosenstock (1969, p. 169) defined health behavior as “the activity undertaken by persons who believe themselves to be healthy, for the purpose of preventing or detecting disease in an asymptomatic stage.” It is generally believed that health behaviors reflect various beliefs and attitudes, such as an individual's conviction that illness is preventable or controllable—which may translate into behaviors relating to preventive service use. Studies, going as far back as the 1930's, have found a consistency in an individual's health behaviors, although the correlations were often only small or modest (Rosenstock 1969; Norris 1997).

For this study we analyzed health behavior to see if there was some consistency in the use of an array of services that included flu shots, pneumonia immunizations, mammography, pap smears, and prostate and colon cancer screening. It is important to emphasize that use is driven not only by the behavior of beneficiaries themselves, who can initiative services such as flu shots, but also—and very significantly—by the behavior of providers, who may (or may not) recommend patients for services such as mammograms (O'Malley et al. 2001), colonoscopy, and PSA tests.

The consistency of an individual's use of preventive services was illustrated by a study in Massachusetts reporting that women who had a mammogram and men who had a prostate specific antigen test (PSA) were more likely to have colorectal screening than their counterparts without a mammogram or a PSA test (Lemon et al. 2001). Further insight into the consistency of behaviors regarding preventive services was provided by a Medicare study that found women who had a mammogram were six times as likely to have a pap smear as women who did not have a mammogram, and men who had a flu shot were more likely to have a prostate screening test than men who did not have a flu shot; these patterns held across races and income groups (Gornick, Eggers, and Riley 2001).

Additional insight into health behaviors comes from studies showing correlations across a variety of health behaviors, such as the use of preventive services and use of seatbelts, smoking cessation, regular exercise, alcohol consumption, and dietary fat intake (Hofer and Katz 1996; Costakis, Dunnagan, and Haynes 1999; Boutelle et al. 2000; Gornick, Eggers, and Riley 2001). Further insight comes from studies showing a consistency between health behaviors and behaviors that are not exclusively related to health, such as precautionary behaviors regarding crime prevention and hazard preparedness (e.g., having batteries ready for potential hurricanes), albeit, again, some of the correlations were only small or modest (Norris 1997; Weinstein 1993).

We were interested in determining if health behaviors regarding preventive service use were associated with health outcomes—specifically, stage of cancer when first diagnosed. We hypothesized that Medicare beneficiaries who used more of an array of preventive services covered by Medicare would have a lower probability of late stage of cancer at time of diagnosis than others who used fewer of these services. Our rationale was: If the use of preventive services (a) is correlated—as the health psychology literature indicates—with other healthy behaviors (such as exercising, not smoking, controlling weight) and the belief that illness is preventable or controllable and (b) promotes opportunities to discuss signs and symptoms of illness with health care providers—then a greater use of preventive services will be associated with a lower probability of late stage cancer at the time a beneficiary is first diagnosed with cancer. In essence, our hypothesis was that the use of preventive services is consistent with a range of healthy behaviors that promote health and early detection of disease.

We studied eight frequently diagnosed cancers in 1995—breast, colorectal, uterine, and ovarian cancer in women and prostate, colorectal, bladder, and stomach cancer in men, and an array of preventive services: influenza and pneumonia immunization, colonoscopy, sigmoidoscopy, barium enema, mammography, pap smear, and the PSA test. The time period selected for studying the use of preventive services prior to the diagnosis of cancer was set at two years. However, as described in the Methods section, we used the 24 months ending in the third month prior to diagnosis as our two-year observation period. The health outcome was stage of cancer when first diagnosed.

Four of the eight types of cancers studied have a specific screening procedure associated with it. It is important to stress, however, that the objective of this research was not to analyze whether the use of any particular screening service, such as the PSA test or mammography, was related to stage of prostate or breast cancer when first diagnosed. Such studies require substantially different methods, data, and design than used in this study (Friedman et al. 1995). Rather, the objective was to determine if the health behavior characterized by using available preventive services was associated with stage of cancer at time of diagnosis. That finding would be of particular importance because data from the National Cancer Institute (Ries et al. 1997) show that elderly blacks are more likely to have late stage cancer when first diagnosed—and more likely to have lower five-year survival rates—than their white counterparts.

Two specific questions were studied: Are beneficiaries with a stronger history of use of preventive services less likely to have late stage cancer when first diagnosed? Is this true for blacks and whites and for groups with different educational attainment?

Methods

Information was derived from the database formed by linking the Surveillance, Epidemiology, and End Results (SEER) program data (sponsored by the National Cancer Institute) with Medicare administrative data (maintained by the Centers for Medicare and Medicaid Services). Methods establishing the linked database have been described elsewhere (Warren et al. 2002).

SEER Data

SEER contains information from 11 cancer registries covering the states of Connecticut, Hawaii, Iowa, New Mexico, Utah; the metropolitan areas of Detroit, San Francisco–Oakland, Atlanta, Seattle-Puget Sound, San Jose–Monterey, and Los Angeles county. The registries cover about 14 percent of the U.S. population. Two of the SEER registries, Los Angeles and San Jose, were excluded because of incomplete information from the U.S. Census on education at the zip code or census tract level.

Medicare Data

Medicare enrollment files were used for demographic information, except that race came from SEER. Utilization information came from claims data.

Census Data

The 1990 U.S. census (zip code or census tract level) was used to create an education variable for each study patient. A higher education area was defined as having less than one-third non-high school graduates, and a lower education area as having at least one-third non-high school graduates. Linkage of the SEER-Medicare database with data from the U.S. census has been described elsewhere (Bach et al. 2002).

Study Patients

Patients in the database first diagnosed with cancer in 1995 who met the following criteria were selected: (a) the patient was at least 67 years of age and continuously enrolled in Medicare Parts A and B at time of diagnosis and during the 27 months prior to month of diagnosis; (b) had no HMO enrollment for the same period (about 85 percent of all beneficiaries in 1995), and (c) no prior diagnosis for this cancer exists on the individual's record. Patients selected into the study under these criteria consisted of eight frequently diagnosed types of cancer: breast (5,242), colorectal (2,946), uterine (1,032), and ovarian (645) cancer in women, and prostate (7,286), colorectal (2,219), bladder (1,405), and stomach (465) cancer in men.

Stage of Cancer at Diagnosis. SEER identifies stage of cancer as “in situ,”“localized,”“regional,”“distant,” except that for prostate cancer, stages used are “localized/regional” and “distant.” Stage is assigned based on information about whether the neoplasm is confined to the organ of origin, has extended to surrounding tissue or lymph nodes, or has spread to other parts of the body; a case is designated as “unstaged” if information in the record is insufficient to assign a stage (SEER 1997).

Preventive Services. Services included were: influenza immunization, pneumonia immunization, colon cancer test (either colonoscopy, sigmoidoscopy, or barium enema, omitting fecal occult blood testing because of data limitations), mammography, pap smear, and PSA test.

A number of other services that are important in preventing disease and promoting health, such as blood pressure and cholesterol screening, were not included because of data limitations.

Algorithm Used to Include Routine Screening Services. Medicare has separate payment codes for diagnostic and screening mammography. However, an analysis of the use of these codes indicated that they are sometimes used interchangeably (Health Care Financing Administration 1995b), and therefore codes cannot be depended upon to determine whether a service was performed for routine screening or for diagnostic purposes. For this study, we wanted to exclude mammograms that were likely to be associated with the diagnosis of breast cancer. To devise an algorithm that would primarily include routine screening and exclude mammograms ordered for diagnostic purposes, month-by-month frequency counts of mammograms were tallied for 27-months preceding the diagnosis of breast cancer. The number of mammograms was fairly stable each month during this period until the third month before cancer was diagnosed, when the number began to increase. Based on this analysis, preventive services were counted if used in the 24-month period ending in the third month before diagnosis of cancer. A preventive service was counted only once, even if the service (e.g., flu shots) was used more than once. For services such as pneumonia immunization, which are believed to be effective for many years, using an equal “window” of observation (24 months) would be expected to produce unbiased results.

Validation of Algorithm. The validity of counting services used 24-months ending in the third month prior to the cancer diagnosis was tested by comparing, for example, mammography use among breast cancer patients and among other cancer patients during the same period. The comparisons showed that mammography was used in that period by 42 percent of breast cancer patients, 48 percent of uterine cancer patients, and 39 percent of women with ovarian cancer.

Charlson Comorbidity Score. The Charlson score, a measure of the health status of each patient in the study, was computed based upon the patient's diagnoses for all Medicare hospitalizations in the 24 months prior to the cancer diagnosis. It ranges from 0 to 9. Of the patients included in this study, the Charlson score was zero for 77 percent or more of the cases. Unstaged cases were omitted in the multivariate analyses of factors associated with late stage cancer; however, the average Charlson score for patients whose cancer was unstaged tended to be greater than for patients whose cancer was staged, suggesting that omitted cases were sicker and likely to be suffering from a more-advanced stage of cancer than staged cases. The Charlson score has been described elsewhere (Charlson et al. 1987).

Statistical Analysis

A previous study based on SEER data dichotomized stages of cancer into early and late stage (Riley et al. 1999). Examination of the distributions of stages for the cancers in this study indicated that, without losing predictive power, we could dichotomize as early stage (“in situ” or “localized”) and late stage (“regional” or “distant”), except for prostate cancer where early stage was “in situ” or “localized” or “regional” and late stage was “distant.”

Cross-tabulations

Patients with cancers that have a specific screening procedure were stratified according to whether they used the preventive service specific to their cancer—mammography for breast cancer, a colon cancer test for colorectal cancer, and the PSA test for prostate cancer. Chi-square tests were performed, excluding unstaged cases, to determine if differences found in stage of cancer by preventive service use were statistically significant.

Multivariate Analyses

For each of the cancers, a step-wise logistic regression model was developed. The dependent variable was stage of cancer dichotomized as early stage=0 and late stage=1; patients with unstaged cancer were excluded. For the set of cancers that have a specific screening procedure, independent variables in the models included use of the preventive service specific to that cancer (e.g., mammography for breast cancer), receipt of one other preventive service, and receipt of two or more other services; age (five groups: 67–74; 75–79; 80–84; 85–89; 90+), race (white, black, other), education (two groups: person living in a zip code or census tract with [a] one-third or more non-high school graduates and [b] less than one-third non-high school graduates), a variable indicating if Medicaid covered out-of-pocket costs (buy-in status), dummy variables for each of the nine SEER registries in the study, a dummy variable for living in a metropolitan and nonmetropolitan area, and a variable (0, 1, or 2 or more) for the Charlson comorbidity score. The model used for cancers without a specific screening procedure was similar except that the independent variable for “use of a specific screening procedure” was omitted.

Results

Table 1 shows stage of cancer at time of diagnosis for the eight types of cancer patients, by race and by education. Among prostate cancer patients, the proportion with late stage cancer was significantly higher among black men (8.9 percent) than white men (6.7 percent); among uterine cancer patients, the proportion with late stage cancer was significantly higher among black women (38.6 percent) than white women (22.2 percent). Unstaged cases were more frequent among black patients than white patients, except for ovarian cancer. To a large extent, disparities in stage of cancer according to educational attainment mirror those found by race (Table 1). The proportion with late stage prostate cancer was significantly higher among men living in areas with lower educational attainment (8.6 percent) than among men living in areas with higher educational attainment (6.3 percent). Among women with uterine cancer, corresponding figures were 28.6 percent and 21.5 percent. As shown, the proportion of unstaged cases was nearly always greater among patients living in areas with lower educational attainment.

Table 1
Medicare Beneficiaries Diagnosed with Cancer in 1995 by Stage of Cancer, by Race and Education, Ages 67 and Older

For the majority of services, utilization of preventive services was significantly lower for blacks than whites and in areas with lower educational attainment than higher educational attainment (Table 2). In particular, the influenza immunization rate among black patients with each of the eight types of cancer was less than 70 percent of the rate for white patients. The one notable exception was in the use of colon cancer tests: among patients diagnosed with colorectal or ovarian cancer, a significantly higher proportion of black patients than white patients had used a colon cancer test.

Table 2
Use of Preventive Services by Medicare Beneficiaries Diagnosed with Cancer In 1995, by Race and Education, Ages 67 and Older

Across all eight types of cancer, compared to whites, a higher proportion of blacks used not a single preventive service in the 24-month period prior to diagnosis of cancer while a smaller proportion of blacks used two or more preventive services (Table 2). Disparities in the use of preventive services were particularly striking among men diagnosed with stomach cancer—51.1 percent of black men were without a single preventive service compared to 25.3 percent of their white counterparts. With the exception of a colon cancer test, patients living in areas with lower educational attainment had lower rates of use of each preventive service (Table 2). For most cancers in the study, a higher proportion of patients in areas with lower educational attainment had used not a single preventive service while a smaller proportion used two or more services compared to patients in areas with higher educational attainment.

Table 3 (all races combined) provides additional evidence of the strong behavioral patterns associated with the use of preventive services. Consistently, patients who had the preventive service specific to their cancer were more frequent users of each of the other preventive services than patients who did not have the specific service. As examples, among women with breast cancer who had a screening mammogram, 20.9 percent also had a colon cancer test while among breast cancer patients without a mammogram, only 9.8 percent had a colon cancer test; and, among men with prostate cancer who had a PSA test, 22.7 percent had a colon cancer test while among men with prostate cancer without a PSA test, only 8.7 percent had a colon cancer test.

Table 3
Percent of Breast, Colorectal, and Prostate Cancer Patients Using Each Preventive Service in the 24-Month Period before Cancer Diagnosis

In general, there was an inverse relationship between the proportion of patients with late stage cancer and the number of preventive services used (Table 4). For example, among breast cancer patients with a mammogram, of those patients with only one preventive service 20 percent had late stage breast cancer, while 17 percent of those patients with two to five preventive services had late stage cancer. Among breast cancer patients without a mammogram, those patients with zero preventive services 36 percent had late stage breast cancer, while 24 percent of those with two to five preventive services had late stage cancer. Similarly, among female colorectal patients with or without a colon cancer test, there was an inverse relationship between the proportion with late stage cancer and the number of preventive services used. Among prostate cancer patients (Table 4) who did not have a PSA test, there was an inverse relationship between the proportion with late stage cancer and the number of preventive services used. Among male colorectal patients there was no statistically significant association between the number of preventive services used and the proportion with late stage cancer.

Table 4
Distribution of Patients by Stage of Cancer When First Diagnosed, by Type of Cancer and Number of Preventive Services Used

Among women with uterine cancer and men with bladder cancer (Table 4), the proportion with late stage cancer was inversely related to the number of preventive services used. For women with ovarian cancer and men with stomach cancer, there was no steady relationship between the number of services used and stage of cancer. The inverse relationships between the number of preventive services used and the proportion of cases with late stage cancer generally held true by race and educational attainment (data not shown).

Multivariate Analyses

Table 5 shows the odds ratios of late stage cancer from the step-wise regression models for the independent variables that were statistically significant. The only variables that remained in the final models were those relating to use of preventive services and the Charlson score. The use of the preventive service specific to a cancer was strongly (inversely) associated with late stage cancer for each of the four cancers. And, as hypothesized, the use of additional preventive services had independent and significant (inverse) relationships with late stage cancer. Female colorectal cancer patients who had a Charlson comorbidity score of 2 or more were less likely to have late stage cancer at diagnosis while prostate cancer patients who had Charlson comorbidity scores of 2 or more were more likely to have late stage cancer at time of diagnosis than their counterparts with comorbidity scores of zero.

Table 5
Multivariate Analyses: Odds Ratios of Late-Stage Cancer (95% confidence limits) from Regression Models

For the four cancers that do not have a specific screening procedure, the use of two or more preventive services was statistically significant for uterine cancer and male bladder cancer. Male stomach cancer patients who had a Charlson comorbidity score of 2 or more were less likely to have late stage cancer than their counterparts with comorbidity scores of zero.

Discussion

A major strength of this study was the availability of the SEER-Medicare database, linking clinical information on stage of cancer and Medicare enrollee and utilization information about services received before the diagnosis of cancer. One limitation of the SEER-Medicare database is that it does not contain individual SES measures such as education or income. As a proxy for individual data, we used area-wide data from the U. S. Census, an approach that was validated by Krieger (1992), Geronimus, Bound, and Niedert (1993), and Gornick et al. (1996), with the caveat that data from the census reflect both characteristics of the individual and the area where the individual resides. Another limitation of the SEER-Medicare database is that it does not contain information about other beneficiary behaviors that affect health (such as smoking, diet, exercise), their beliefs and attitudes about prevention, or whether the characteristics of providers (e.g., race, gender, prevention orientation) who treat blacks and disadvantaged patients differ from those of providers who treat white and advantaged patients.

The study shows clearly, nonetheless, that before their cancer diagnosis, blacks and patients in lower SES used fewer preventive services (with the exception of colon cancer tests) than whites and patients in higher SES. Particularly noteworthy were disparities in the use of flu shots—only 25.9 percent of black men compared to 54.6 percent of white men had an influenza immunization in the two-year period before colorectal cancer was diagnosed.

The purpose of our study, however, was to move beyond the documentation of disparities in utilization and analyze behaviors relating to the use of an array of services. Preventive and cancer screening services have traditionally been analyzed from the perspective of efficacy of a particular service. This study, however, takes a different approach—analyzing the use of preventive and cancer screening services from the perspective of health behaviors and health outcomes.

The fact that this study includes services such as flu shots and PSA tests indicates that our findings reflect both beneficiary and provider behaviors. It was striking to find such a high proportion of patients who had used none of the preventive services included in this study during the two years before cancer was diagnosed. As noted, 51.1 percent of black men and 25.3 percent of white men used not a single preventive service in the two-year period before their stomach cancer was diagnosed. At the same time, there were a substantial number of patients who had used two to five of the preventive services studied.

Our objective was to test the hypothesis that a greater use of preventive services would be associated with a lower probability of late stage cancer when first diagnosed. The study included cancers with and without a specific screening procedure associated with it. Our conjecture was that a greater use of preventive services would (a) be consistent with other healthy behaviors (a line of reasoning with underpinnings in the field health psychology), including knowledge of and responding to signs and symptoms of cancer, and (b) offer opportunities to meet with physicians and other providers to call attention to unusual symptoms that foretell of cancer.

Our hypothesis was supported in analyses of the four cancers with a specific screening procedure as well as for uterine and male bladder cancer. In particular, among women with uterine cancer who had used none of the preventive services, 30 percent had late stage cancer, and among those who had used two to five preventive services, 20 percent had late stage cancer, while among men with bladder cancer who had used none of the preventive services, 27 percent had late stage cancer, and among those who had used two to four services, 18 percent had late stage cancer.

Although this study fully supports earlier findings that race is associated with stage of cancer, our multivariate analyses—designed to test the odds of late stage cancer—included variables for the use of preventive services as well as race, SES, and other independent variables. Using step-wise regression, in the final models variables for preventive service use were the only variables that were consistently significant in predicting the odds of late stage cancer.

The most important finding was that for six of the eight cancers studied Medicare beneficiaries—black or white, of lower or higher SES—who used more of the available preventive services were less likely to have cancer diagnosed at a late stage than their counterparts who used fewer of these services. This finding provides a cautionary note regarding the more than 40 million persons under age 65 who are uninsured. For them, the costs of health care are likely to present significant barriers that result in lowering the use of preventive services in favor of acute care needs. Forgoing the use of preventive services is likely to have an impact not only on the health of the uninsured today but also in the future—by weakening the reinforcement of healthy behaviors that develop over a lifetime. This study argues for the following conclusions:

One, beneficiaries and providers need to take a more active role in promoting a range of healthy behaviors, including attitudes about disease prevention. The nonuse of a particular preventive service, such as flu shots, could serve as a marker to identify beneficiaries who are likely to be nonusers of other preventive services. In the year 2000, the rate of influenza immunization among black beneficiaries was still less than half the rate for white beneficiaries (Centers for Medicare and Medicaid Services 2003). Efforts made to promote the use of one preventive service—such as flu shots—may have a multiplicative effect by raising health consciousness about the need for and availability of other Medicare preventive services. Consistently, white beneficiaries and those enjoying higher socioeconomic status—and their providers—take greater advantage of available services that can promote health and prevent disease. This is true not only in the Medicare fee-for-service sector but also in the managed care sector (Virnig et al. 2002), and in programs under the supervision of the Veterans Administration (Peterson et al. 1994).

Two, more knowledge is needed about the beliefs and attitudes that influence beneficiary and provider behaviors regarding disease prevention. It has been found, for example, that women who receive care from female physicians have higher rates of mammography and pap smears, and that differences in beliefs of male and female physicians and patient preferences for female physicians contribute to that finding (Lurie et al. 1993, Lurie et al. 1997). Survey questions that asked respondents not only if they had a flu shot (or mammogram, pap smear, etc.) but also why they did not—probing for the respondent's attitudes and beliefs as well as whether or not a flu shot (mammogram, pap smear, etc.) had been recommended by their provider—could help to explain some of the disparities in health care.

Three, health policy and research experts need to expand the theoretical framework traditionally used to study the causal pathways that lead to disparities in health and health care. Health insurance and many other traditional indicators of access to care, such as having a regular source of care, are clearly necessary but not sufficient conditions to eliminate disparities in access and utilization of health care. As this study indicates, ameliorating disparities in health outcomes will require gaining a better understanding of beneficiary and provider behaviors and the factors that influence these behaviors.

In conclusion, this study supports the premise raised in the Institute of Medicine reports that integrating the theoretical framework of health psychology and the behavioral and social sciences into new research has the potential for improving the public's health. Our study results indicate a need for a greater understanding of health behaviors to meet the Healthy People 2010 goals (Department of Health and Human Services 2002).

We recommend that an organization interested in promoting health bring together experts to identify and illuminate specific social and behavioral questions, with the goal of stimulating a new phase of health services research relating to understanding and improving patient and provider health behaviors. It seems clear that if health behaviors of vulnerable beneficiaries and their providers were similar to those of more advantaged beneficiaries and their providers, then gaps in health outcomes between races and SES groups would very likely begin to close.

Acknowledgments

The authors appreciate the computer programming work performed by Harold Cooper, CHD Research Associates. The opinions expressed in this paper are those of the authors and do not necessarily reflect those of the Centers for Medicare and Medicaid Services or the National Institutes of Health.

References

  • Ayanian J Z, Udvarhelyi I S, Gatsonis C A, Pashos C L, Epstein A M. “Racial Differences in the Use of Revascularization Procedures after Coronary Angioplasty.” Journal of the American Medical Association. 1993;269(20):2642–6. [PubMed]
  • Bach P B, Guadagnoli E, Schrag D, Schussler N, Warren J. “Patient Demographic and Socioeconomic Characteristics in the SEER-Medicare Database: Applications and Limitations.” Medical Care. 2002;40(8, supplement):IV-19–25. [PubMed]
  • Boutelle K N, Murray D M, Jeffery R W, Hennrikus D J, Lando H A. “Associations between Exercise and Health Behaviors in a Community Sample of Working Adults.” Preventive Medicine. 2000;30(3):217–24. [PubMed]
  • Centers for Medicare and Medicaid Services. “Preventive Services” [accessed 2003]. Available at http://www.CMS.HHS.Gov/preventiveservices.
  • Charlson M E, Pompei P, Ales K L, MacKenzie C R. “A New Method of Classifying Prognostic Comorbidity in Longitudinal Studies: Development and Validation.” Journal of Chronic Diseases. 1987;40(5):373–83. [PubMed]
  • Committee on Health and Behavior. Health and Behavior: The Interplay of Biological, Behavioral, and Societal Influences. Washington, DC: National Academy Press; 2001.
  • Costakis C E, Dunnagan T, Haynes G. “The Relationship between the Stages of Exercise Adoption and Other Health Behaviors.” American Journal of Health Promotion. 1999;14(1):22–30. [PubMed]
  • Department of Health and Human Services. Healthy People 2010. Washington, DC: Department of Health and Human Services; 2002.
  • Friedman G D, Hiatt R A, Quesenberry C P, Jr, Selby J V, Weiss N S. “Problems in Assessing Screening Experience in Observational Studies of Screening Efficacy: Example of Urinalysis for Bladder Cancer.” Journal of Medical Screening. 1995;2(4):219–23. [PubMed]
  • Geronimus A T, Bound J, Niedert L. On the Validity of Using Census Geocode Characteristics to Proxy Economic Status. Ann Arbor, MI: Population Studies Center, University of Michigan; 1993.
  • Gornick M E. Vulnerable Populations and Medicare Services: Why Do Disparities Exist? New York: Century Foundation Press; 2000.
  • Gornick M E, Eggers P W, Riley G F. “Understanding Disparities in the Use of Medicare Services.” Yale Journal of Health Policy, Law, and Ethics. 2001;1(1):133–58. [PubMed]
  • Gornick M E, Eggers P W, Reilly T W, Mentnech R M, Fitterman L K, Kucken L E, Vladeck B C. “Effects of Race and Income on Mortality and Use of Services among Medicare Beneficiaries.” New England Journal of Medicine. 1996;335(11):791–9. [PubMed]
  • Health Care Financing Administration. Report to Congress: Monitoring the Impact of Medicare Physician Payment Reform on Utilization and Access. Baltimore, MD: Health Care Financing Administration; 1995a.
  • Health Care Financing Administration. 1992–93 Mammography Services Paid by Medicare: State and County Rates. Baltimore, MD: Health Care Financing Administration; 1995b.
  • Health and Health Care of the Medicare Popoulation: Data from the 1997 Medicare Current Beneficiary Survey. Rockville, MD: Westat; 2001.
  • Hofer T P, Katz S J. “Healthy Behaviors among Women in the United States and Ontario: The Effect on Use of Preventive Care.” American Journal of Public Health. 1996;86(12):1755–9. [PubMed]
  • Krieger N. “Overcoming the Absence of Socioeconomic Data in Medical Records: Validation and Application of a Census-based Methodology.” American Journal of Public Health. 1992;82(5):703–10. [PubMed]
  • Lemon S, Zapka J, Puleo E, Luckmann R, Chasan-Taber L. “Colorectal Cancer Screening Participation: Comparison with Mammography and Prostate-Specific Antigen Screening.” American Journal of Public Health. 2001;91(8):1264–72. [PubMed]
  • Lurie N, Margolis K L, McGovern P G, Mink P J, Slater J S. “Why Do Patients of Female Physicians Have Higher Rates of Breast and Cervical Cancer Screening?.” Journal of General Internal Medicine. 1997;12(1):34–43. [PMC free article] [PubMed]
  • Lurie N, Slater J, McGovern P, Ekstrum J, Quam L, Margolis K. “Preventive Care for Women: Does the Sex of the Physician Matter?.” New England Journal of Medicine. 1993;329(7):478–82. [PubMed]
  • National Center for Health Statistics. Health, United States, 2002, with Chartbook on Trends in the Health of Americans. Hyattsville, MD: National Center for Health Statistics; 2002.
  • Norris F H. “Frequency and Structure of Precautionary Behavior in the Domains of Hazard Preparedness, Crime Prevention, Vehicular Safety, and Health Maintenance.” Health Psychology. 1997;16(6):566–75. [PubMed]
  • O'Malley M S, Earp J A, Hawley S T, Shell M J, Mathews H F, Mitchell J. “The Association of Race/Ethnicity, Socioeconomic Status, and Physician Recommendation for Mammography: Who Gets the Message about Breast Cancer Screening?.” American Journal of Public Health. 2001;91(1):49–54. [PubMed]
  • Peterson E D, Wright S M, Daley J, Thibault G E. “Racial Variation in Cardiac Procedure Use and Survival following Acute Myocardial Infarction in the Department of Veterans Affairs.” Journal of the American Medical Association. 1994;271(15):1175–80. [PubMed]
  • Riley G F, Potosky A L, Klabunde C N, Warren J L, Ballard-Barbash R. “Stage at Diagnosis and Treatment Patterns among Older Women with Breast Cancer: An HMO and Fee-for-Service Comparison.” Journal of the American Medical Association. 1999;281(8):720–6. [PubMed]
  • Rosenstock I M. “Prevention of Illness and Maintenance of Health.” In: Kosa J, Antonovsky A, Zola I K, editors. Poverty and Health: A Sociological Analysis. Commonwealth Fund Book; 1969. pp. 168–90.
  • Ries L A G, Kosary C L, Hankey B F, Miller B A, Harras A, Edwards B K, editors. SEER Cancer Statistics Review, 1973–1994. Bethesda, MD: National Cancer Institute, National Institutes of Health; 1997.
  • Smedley B D, Syme S L, editors. Promoting Health: Intervention Strategies from Social and Behavioral Research. Washington, DC: National Academy Press; 2000. [PubMed]
  • Virnig B A, Lurie N, Huang Z, Musgrave D, McBean A M, Dowd B. “Racial Variation in Quality of Care among Medicare+Choice Enrollees.” Health Affairs. 2002;21(6):224–30. [PubMed]
  • Warren J L, Klabunde C N, Schrag D, Bach P B, Riley G F. “Overview of the SEER-Medicare Data: Content, Research Applications, and Generalizability to the United States Elderly Population.” Medical Care. 2002;40(8, supplement):IV-3–18. [PubMed]
  • Weinstein N D. “Testing Four Competing Theories of Health-Protective Behavior.” Health Psychology. 1993;12(4):324–33. [PubMed]

Articles from Health Services Research are provided here courtesy of Health Research & Educational Trust