|Home | About | Journals | Submit | Contact Us | Français|
To examine whether the usual source of preventive care, (having a usual place for care only or the combination of a usual place and provider compared with no usual source of preventive care) is associated with adults receiving recommended screening and prevention services.
Using cross-sectional survey data for 24,138 adults (ages 18–64) from the 1999 National Health Interview Survey (NHIS), we estimated adjusted odds ratios using separate logistic regression models for receipt of five preventive services: influenza vaccine, Pap smear, mammogram, clinical breast exam, and prostate specific antigen.
Having both a usual place and a usual provider was consistently associated with increased odds for receiving preventive care/screening services compared to having a place only or neither. Adults ages 50–64 with a usual place/provider had 2.8 times greater odds of receiving a past year flu shot compared with those who had neither. Men ages 50–64 with a usual place/provider had nearly 10 times higher odds of receiving a PSA test compared with men who had neither. Having a usual place/provider compared with having neither was associated with 3.9 times higher odds of clinical breast exam among women ages 20–64, 4.1 times higher odds of Pap testing among women ages 21–64, and 4.8 times higher odds of mammogram among women ages 40–64.
Having both a usual place and usual provider is a key variable in determining whether adults receive recommended screening and prevention services and should be considered a fundamental component of any medical home model for adults.
The medical home construct has evolved over the past 40 years as a model for the delivery of care and includes the “early identification of special health care needs, provides ongoing primary care, and coordinates with a broad range of specialty, ancillary, and related services.”1 There is a growing body of research indicating that having a usual source of care improves timely access to medical care, improves quality of care received, and results in improved health status.2,3,4,5 The American College of Physicians released a policy monograph that called for “a comprehensive public policy initiative that would fundamentally change the way that primary care and principal care (whether provided by primary care or specialty care physicians) are delivered to patients by linking patients to a personal physician in a practice that qualifies as an advanced medical home.”6 This advanced medical home includes patients working directly with a physician, and the physician’s care team, to promote a “continuous healing relationship” focused on patient-centered care where physicians are directly accountable to each patient. Several state health reform proposals also promote the use of a medical home construct as part of their reform initiatives to focus health care on the individual and preventive primary care.
Fundamental to the development of the medical home concept will be to define and assess the required-or recommended-attributes of the medical home. Some of the key attributes of the “advanced” medical home include: (1) use of evidenced-based medicine, (2) use of the Chronic Care Model, (3) development of medical care plan in partnership with patients, (4) enhanced communication and access through telephone and e-mail, (5) use of health information technology, and (6) performance measurement.6 It is not clear how many physician clinics currently meet these standards or if there is any empirical evidence to support the association between key attributes and improved health status or improved patient care outcomes.
Research related to the medical home concept has focused only on a few key dimensions, primarily related to the usual site of care and/or usual provider. A study by Xu, using data from the national 1996 Medical Expenditure Panel Survey (MEPS), found that, for adults, having a usual doctor was more important than having a usual site of care for receipt of certain preventive services such as blood pressure and cholesterol level checks.5 However, having a usual doctor had no more effect than having a usual site of care on the use of flu shots, pap smears, and mammograms.5 Devoe et al., using the same data, found that receipt of preventive services was strongly associated with having insurance and having a usual provider for adults 18 years or older.7 They concluded that having health insurance and a usual provider were key to access to preventive services. However, Ettner found that it was the specific qualities of the patient-provider relationship that were associated with improved use of prevention services and that individual patients’ lifestyle choices were much more salient determinants of health status than the delivery of medical care services.8
In addition to the difficulty in identifying the attributes of a medical home, most research in this area is based on data from questions included on national surveys. However, these questions are not consistently available or comparable over time or across surveys. Table 1 shows salient questions in the most recent public-use data available for four national health surveys. The Behavioral Risk Factor Surveillance System (BRFSS) asks about a usual provider; the National Health Interview Survey (NHIS) asks about a usual place; MEPS asks about both usual place and usual provider, but has smaller sample size and lacks the preventive services utilization variables; the State and Local Area Integrated Telephone Survey (SLAITS) asks a series of 19 questions intended to measure “medical home,” but this survey monitors children and is fielded periodically rather than annually. Even when a usual provider is identified, there are limited data on whether that provider meets the characteristics of primary care often associated with a medical home: namely, first-contact care, long-term person focused, comprehensive care, coordination of specialty care, and other needed services.2
We suggest in this paper that a foundational element of a medical home is that it should include both a usual place and a usual provider. We use the NHIS to test this assumption. We define a continuum of usual source of care based on available data that suggests that a combination of usual place and usual provider represents a more coordinated approach to providing preventive care, leading to increased access and better health outcomes, and that having a place for care, but no usual provider, or no usual place, would be less advantageous. Unique to our approach is the ability to assess both dimensions (place and provider) of usual source of care, resulting in three distinct types of the usual source of care construct.
We used cross-sectional survey data for adults from the 1999 NHIS to assess the impact of usual source of care on use of preventive services. We used this earlier survey because it is the most recent data to include both usual place and usual provider information together, along with several measures of preventive care services utilization. While complete information on both usual place and usual provider was asked in the 1993 to 1996 Access to Care Supplements of the NHIS, later NHIS questionnaires dropped the usual provider question in all but 2 years (1999 and 2001). Thus, change over time in this more comprehensive measure of usual source of care is impossible to assess with NHIS data.
We obtained data for 24,553 sample adults between the ages of 18 and 64 years who had complete usual source of care data. Of those, 415 (1.7 percent) had some missing covariate data and were thus excluded from further analysis. This resulted in an unweighted sample size of 24,138 sample adults.
The outcomes of interest were defined as receipt of selected preventive care services: influenza vaccine in the past year for adults ages 50 to 64 years, up-to-date (UTD) pap smear for women ages 21 to 64 years (excluding those who reported having had a hysterectomy), UTD mammogram for women ages 40 to 64 years, UTD clinical breast exam (CBE) for women ages 20 to 64 years, and prostate-specific antigen (PSA) in the past year for men ages 50 to 64 years. UTD status for cancer screenings was defined according to American Cancer Society (ACS) recommendations for average risk, asymptomatic people.9 See Table 2 for more details about the original survey questions, the operationalization of these measures, and the analytic samples for each.
We used the following continuum of usual source of care: having a ‘usual place and provider,’ having a ‘usual place only,’ or having ‘no usual place for preventive care’ (comprising 69.8 percent, 16.2 percent, and 14 percent of our weighted sample, respectively). Those who reported that they have a usual place where they receive preventive care and that they have a particular provider they see when they receive health care were classified as having both a usual place and provider. Those who have a usual place for preventive care, but no particular provider were classified as having a usual place only. Those who reported no usual place for preventive care were coded as having no place for preventive care even if they reported that they have a particular provider. Figure Figure11 shows how the NHIS survey questions about place and provider were used to construct our usual source of care classification.
Additional covariates were defined as follows. Age was classified in groups representing <= 29 years, 30–39 years, 40–49 years, 50–59 years, and 60–64 years. Race/ethnicity was a set of indicator variables constructed from self-reported race and Hispanic origin. This combination produced six categories representing five non-Hispanic groups (African-American, American Indian, Asian, Other/multiple race, and White), and a Hispanic group including all Hispanics regardless of race. Employment status was classified as currently employed or not. Educational attainment was classified as less than high school, a high school diploma, some college, and a college degree. Marital status was defined as currently married or not. Total family income was obtained from the imputed income files provided by the National Center for Health Statistics. These files are supplements to the NHIS data where values for income are imputed for each missing case. We merged these data with the sample adult data and collapsed the income values into six categories representing income levels of <$9,999, $10,000–19,999, $20,000–34,999, $35,000–54,999, $55,000–74,999, and >=$75,000. Financial barriers to care were classified as past year delays in medical care due to cost. Self-reported health status was measured on a five-point scale that we dichotomized into “poor” (fair and poor) and “good” (excellent, very good, and good). Insurance status was classified as private, public, and uninsured.
First, we assessed the extent to which the three usual source of care groups differed in background characteristics that are potentially associated with receipt of preventive care services using cross tabulations and design-based F-tests to account for the complex sample design. We then examined bivariate relationships between usual source of care and each preventive service using simple logistic regression models. Adjusted models were built and evaluated incrementally. Finally, we estimated adjusted odds ratios with separate logistic regression models for receipt of each of the five preventive services adjusted for all covariates listed in Table 3.
Analyses were conducted using Stata statistical software (SE version 9.2), which produces unbiased estimates from data collected through complex sampling designs.10,11 The survey commands in Stata properly account for the unequal probabilities of selection and the stratified/clustered sampling design of the NHIS data. Variance estimates were produced using Taylor series linearization.
Table 3 presents the background characteristics of sample adults ages 18 to 64 by usual source of care. These characteristics are significantly differentially distributed across the three source of care groups. As expected, the distributions reflect the general social gradient across the continuum. Respondents with a usual place and a usual provider were more likely to be White, female, and privately insured, and have higher educational attainment and higher incomes than other groups. Additionally, delays in medical care due to cost were less prevalent in those who indicated that they had a usual place and usual provider.
Table 4 presents the adjusted odds ratios (OR) for receipt of each preventive service by usual source of care. Compared to having no usual place or provider, having a ‘usual place only’ or ‘a usual place and provider’ was consistently associated with significantly increased odds of having received each of the preventive care/screening services.
Among adults ages 50 to 64 years, receipt of a flu shot in the past year was significantly associated with source of care after adjusting for sociodemographics, financial barriers to care, health status, and insurance coverage. Those who reported having a usual place had 1.6 times greater odds, while those who had both a usual place and a usual provider had 2.8 times greater odds of receiving a past year flu shot compared with those who reported no usual place for preventive care. Among men ages 50 to 64 years, having a usual place was associated with nearly five times higher odds (OR=4.6, 95% CI 2.5–8.4) of receiving a PSA test compared with those who had no place for preventive care. Men with both a place and provider had nearly ten times higher odds of receiving a PSA test compared with men who had no place for preventive care (OR=9.6, 95% CI 5.8–15.9).
Among women, receipt of clinical breast exams, Pap smears, and mammograms were all strongly and consistently associated with usual source of care even after covariate adjustment. Women ages 20 to 64 years with a usual place for preventive care had twice the odds (OR=2.1, 95% CI 1.7–2.5), and those with both a usual place and a usual provider had nearly four times higher odds (OR=3.9, 95% CI 3.3–4.6) of receiving a clinical breast exam compared with those who had no place for preventive care. A similar pattern is noted for Pap smear testing. Women with a usual place had nearly twice the odds (OR=1.8, 95% CI 1.5–2.3) and women with both a usual place and provider had over four times the odds (OR=4.1, 95% CI 3.4–4.9) of being up to date with Pap smear testing compared with those who had no place for preventive care. Finally, women ages 40 to 64 years with a usual place only had 2.5 times higher odds (95% CI 1.8–3.4), and women with both a place and provider had nearly five times higher odds (OR=4.8, 95% CI 3.7–6.4) of having had a mammogram in the past year compared with women who had no usual place for preventive care.
Our findings suggest that having both a usual place and a usual provider are key to determining whether adults receive recommended screening and prevention services. The more comprehensively defined source of care appears to be associated with substantially improved likelihood of receiving recommended preventive services, which may ultimately contribute to improved health. Just having a usual place for preventive care is less effective.
Our findings are consistent with previous studies that examined the impact of place and/or provider on receipt of preventive services and recommended screenings. In a study of preventive services for women and children, Ettner found that having a usual source of care (place) was strongly associated with receipt of Pap smears, breast exams, and mammograms.12 Doescher et al. also found that continuity of both place and provider was significantly associated with increased receipt of flu shots and mammograms, above and beyond continuity of place alone.13 Again, our contribution is that our data source included a large adult sample with both survey questions that allowed us to test the effects of having only a usual place for preventive care or having both a usual place and a usual provider.
We also found that while existing surveys of health and health care utilization provide some limited information on the medical home construct, there should be a concerted effort to determine the key factors associated with having a usual place and a usual provider that increase the likelihood of use of preventive services and, in turn, increased health status. Yet some of these characteristics may be more personal and therefore difficult to assess. For example, there is evidence that the quality of the relationship, the ability of the provider to communicate, or the amount of time the provider spends with a patient influences use of preventive services.3,8 It will be critical to understand the relationship of these attributes to improvements in patient care and health status if the medical home construct is to take hold for adult populations.
Our study findings should be considered in light of several limitations. First, our usual source of care classification was based on a series of questions pertaining to usual place and usual provider. While the questions about place specifically distinguish a place for preventive care versus other health care, the provider question does not. It is possible that some of our subjects classified as having a usual place and provider for preventive care did not have a usual provider for preventive care. Second, some subjects may have been misclassified in the receipt of preventive care measures due to limitations of data availability in the NHIS. For example, the ACS guidelines for Pap smear testing take into account previous Pap smear results. NHIS does not contain data on previous Pap test results, so our up-to-date Pap test measure is based only on years since last Pap test. Finally, although we found strong statistically significant associations between source of care and receipt of preventive care services, it is possible that this observed relation is due to some other unobserved confounder or due to reverse causality. In other words, those who report a usual place and a usual provider may be systematically different from those who do not. These systematic differences may not have been fully accounted for by our available covariates, which resulted in significantly higher odds of service receipt even after covariate adjustment. However, given the magnitude of our effect estimates, an omitted covariate would have to be extraordinarily strong to attenuate these findings. It is also possible that those who have specifically sought preventive cares services are more likely to report having usual place and a usual provider because of their recent use of services.
We would also like to place our findings in the context of the benefits of primary care and prevention. A recent report on preventive care demonstrates the under-use of effective preventive care in the US resulting in the loss of life, preventable poor health status, and inefficient use of medical care.14 Specific to our measures tested, the study found that 12,000 lives would be saved each year if the portion of adults age 50 and older and others at risk were immunized against influenza, and nearly 4,000 lives would be saved each year if more women age 40 and older were screened for breast cancer.
We have demonstrated that having a usual place and usual provider are associated with an increased likelihood of receiving preventive services and recommended screenings compared with having no usual source of care. However, we know very little about the characteristics of those places and the providers that are seen. If, as the American College of Physicians6 and others advocate, all US citizens are to have a medical home, then we need to know much more about what constitutes a medical home and how to measure those attributes that contribute to success. We believe the place to start is for a patient to be linked with a provider who is seen routinely at the same place every time.
This project was funded, in part, by a grant to the University of Minnesota School of Public Health from the National Institutes of Health, National Institute of Child Health and Human Development to establish the “Integrated Health Interview Series (IHIS).” NICHD Contract #1-R01-HD046697–01.