Effects of Gender, Disability, and Age in the Receipt of Preventive Services

Abstract

Purpose of the Study: We extend research to examine relations between gender, disability, and age in the receipt of preventive services. Design and Methods: We pool Medical Expenditure Panel Survey data for years 2001–2007. Using logit models, we examine the relations between gender, disability, and age and the receipt of preventive services. Results: For most services, both women and men with disabilities had higher probabilities of receiving preventive services relative to those without disabilities. There was a pattern of more significant differences for men relative to women. Predicted probabilities for receipt of services were significantly higher among older adults relative to younger adults. A usual source of care was a significant predictor across services. For example, we estimate that adults aged 18–64 with a place as a usual source of care received 59% of recommended services, whereas those with a person as a source of care received 63% of services relative to 47% for those without a usual source of care. Among older adults, the predicted percentage of preventive services received for no usual source of care was 52% and that for a place or a person as a usual source of care were 71% and 76%, respectively. Across gender, disability, and age, receipt of a range of clinical preventive services is suboptimal. Implications: Policy actions that may mitigate the differences we observed include mechanisms to support access to a usual source of care, financial incentives to enhance the receipt of preventive services, and implementation of community-based prevention services with attention to their linkage to clinical care.

One of four overarching goals of Healthy People 2020 is the elimination of health disparities (Department of Health and Human Services [DHHS], 2011), echoed as a strategic direction in the 2011 National Prevention Strategy (National Prevention Council, 2011). The Agency for Healthcare Research and Quality (AHRQ, 2008, p. 1) has been tasked “with identifying disparities in health care among racial, ethnic and socioeconomic groups in the general U.S. population and within specific priority populations as well as tracking these over time to monitor progress in reducing such disparities.” In April 2012, AHRQ released its ninth report, finding that although some progress has been made, many of the largest disparities in access and quality are unchanged (AHRQ, 2012).

Individuals with disabilities have been identified as a priority population in tracking and eliminating health disparities. An estimated one in five individuals in the United States experience a disability (Census Bureau, 2010). Disability prevalence varies notably by sociodemographic characteristics. Disability prevalence increases with age, with 5.8% of children aged 5–15 experiencing a disability relative to 41.9% of adults aged 65 and older (Waldrop & Stern, 2003). Blacks and Hispanics experience a higher rate of disability at 24.3% and 20.9%, respectively, as opposed to 18.5% of the white population (Waldrop & Stern, 2003). Disability prevalence declines with increasing education and income.

With regard to gender, at younger ages, disability prevalence is lower among females relative to males, 4.3% versus 7.2% for children aged 5–15. This pattern reverses with age; by age 65, 43.0% of women relative to 40.2% of men living in the community experience a disability (Waldrop & Stern, 2003). Using the National Health Interview Survey (NHIS), Altman and Bernstein (2008) found that a similar share of women and men aged 18 and older do not report a disability, 49.5% and 50.5%, respectively. Among adults with disabilities, women experience higher rates of limitations in basic (e.g., sensory: 58% vs. 42%) and complex (e.g., work: 57.2% vs. 42.8%) activity limitations than men. The higher prevalence of disability among women has been attributed to a higher incidence, a longer duration of disability among older women and selective mortality among older men (Leveille, Resnick, & Balfour, 2000, p. 111). Despite this higher prevalence, limited research has focused on disability, women, and women’s health care use until recently (Chevarley, Thierry, Gill, Ryerson, & Nosek, 2006; Parish & Ellison-Martin, 2007). Studies concerning women and disability found that women with disabilities experience poorer access to care (Chevarley et al., 2006), poorer health outcomes (National Council on Disability, 2009), and less satisfaction with their health care (Iezzoni, Davis, Soukup, & O’Day, 2003) than their counterparts who are not disabled.

One focus of research on women and disabilities has been their receipt of preventive services. Receipt of these services is essential to maintaining health across the life span (National Prevention Council, 2011). Preventive services include immunizations, screening tests for conditions such as hypertension and cancer, and counseling for behaviors such as smoking (Benson & Aldrich, 2012). In their 2012 report, AHRQ noted that disparities in the receipt of preventive services were an area in which improvement was necessary. AHRQ states that “Preventive services that avert the onset of disease, foster the adoption of healthy lifestyles, and help patients to avoid environmental health risks hold the greatest potential for maximizing population health” (AHRQ, 2012, p. 24).

The receipt of preventive services among people with disabilities is of importance for a variety of reasons. Individuals with disabilities can benefit from health promotion and prevention activities as can individuals in the general population (Altman & Bernstein, 2008; Wei, Findley, & Sambamoorthi, 2006). Additionally, for individuals with disabilities, preventive services can aid in alleviating the severity of disability and prevent the development of secondary conditions (Rimmer, 1999). Therefore, it would be pertinent to examine the extent to which disparities in receipt of these services exist among adults with disabilities, particularly women as they represent a significant share of the population with disabilities.

In a nationally representative sample of women aged 18–64, Parish and Huh (2006) found that women with a disability were less likely to have received a pap smear than women without a disability, with predicted probabilities of screening at .58 and .63, respectively. Among women aged 18 and older in the National Health Interview Survey (NHIS) and Healthy People 2000, Iezzoni, McCarthy, Davis, Harris-David, and O’Day (2001) found that women with lower extremity mobility difficulties that were significant and long term had lower odds of receiving mammograms and pap smears. However, receipt of a clinical breast examination did not differ between women with and without disabilities. Wei and colleagues (2006) examined receipt of preventive services among women aged 51–64 years. Using Medical Expenditure Panel Survey (MEPS) data, they found that women with disabilities had lower odds of receiving screening for cervical and breast cancer. However, they had significantly higher odds of being screened for colon cancer. Receipt of mammography varied by type of disability in Ahmed and colleagues’ analysis of the 2000 NHIS (Ahmed, Smith, Haber, & Belcon, 2009). Although women with a sociability limitation had significantly lower odds of receipt of mammography relative to women without this limitation, women with a severe physical limitation had significantly higher odds of screening than women without a physical limitation.

With regard to counseling related to smoking and exercise, Iezzoni and colleagues (2001) found evidence for both lower odds (e.g., receipt of counseling to quit smoking among women with major lower limb extremity difficulties) and higher odds (e.g., receipt of counseling related to exercise among women with some difficulties related to lower limb extremity difficulties) relative to women without these disabilities. The odds of receiving an influenza vaccination were higher among women with a disability in Wei and colleagues’ study (Wei et al., 2006).

Thus, the evidence to date on disparities related to preventive and screening services among women with disabilities is mixed, varying by disability definition and type of disability, ages included, services studied, and data source. We extend this work in the following ways. First, in contrast to previously reviewed research related to women and disabilities, limited work has examined use of preventive services between men with and without disabilities. Research in the general older adult population suggests that there may be differences by gender in receipt of preventive services. Between older adults with and without disabilities, data from the Behavioral Risk Factor Surveillance Survey (BRFSS) indicated that less than 42% of men aged 65 and older were up to date for all recommended preventive services and only 32% of women of the same age group were up to date for these services (Shenson, Adams, Bolen, & Anderson, 2011). Although a larger share of men received no core service relative to women, among those receiving preventive care, men had a higher receipt of all core preventive services. Importantly, as discussed in our conceptual framework, both disability and age may moderate gender differences in receipt of health services (Calasanti, 2010; Gibbs, 2008; O’Brien, Hunt, & Hart, 2005; Smith, Braunack-Mayer, Witter, & Warin, 2007). We extend existing literature to more fully explore these relations between gender, disability, age, and receipt of preventive services. Thus, we compare receipt of preventive services, testing differences across gender and disability status for adults aged 18–64 relative to adults aged 65 and older. After testing age-related differences in receipt of preventive care, we examine the role of a usual source of care, a policy-amenable variable, in addressing observed disparities. Our examination of preventive services includes receipt of seven specific measures and one summary measure. Prior work has examined fewer measures, but only the work of Shenson, Bolen, Adams, Seeff, and Blackman (2005) and Shenson and colleagues (2011) appears to examine a summary measure of receipt.

Methods

Behavioral Model of Health Services

We use Andersen’s behavioral model of health care as a conceptual framework (Andersen, 1995). Andersen’s initial model focused largely on factors associated with the individual and how those factors affected the use of health services. Now termed population factors, these include predisposing factors (e.g., demographics such as gender and age), enabling factors (e.g., income and health insurance), and need factors. The model has been expanded to include external environmental factors, use of services, and the health outcomes associated with use of services. Environmental factors are those factors associated with the health care system (e.g., public policy, health care organizations). Such factors have significant influence on the availability and accessibility of health care. Population factors form the second component of the model and include the previously described predisposing, enabling, and need factors. The third factor, use of services, is regarded to be influenced by both environmental and population factors. Health outcomes, in turn, are influenced by the external environment, population characteristics, and service use. Health outcomes include perceptions of health care by patients, valuation of health status by physicians and other health care professions, and overall satisfaction of care by patients. We focus on gender and age, both predisposing demographic characteristics, disability as a measure of need, and use of preventive services.

In her discussion of gender relations and aging research, Calasanti (2010, p. 721–722) extends the concept of gender as presented in Andersen’s model, suggesting that “men as well as women have gender, a set of ideals created in relations of inequality and in concrete institutions, motivating behaviors that shape experiences of old age.” Calasanti explores this in the context of health care and aging, finding, for example, that although older men and women perceive a healthy diet to be important to healthy aging, women take more responsibility for such a diet (Wu, Goins, Laditka, Ignatenko, & Goedereis, 2009). Although men use fewer health care services over their lifecourse (Kaye, Crittenden, & Charland, 2008), Calasanti argues that “gender interacts with age relations in shaping health behaviors and experiences of health conditions” (p. 730). Relatively recent research on masculinity offers a theoretical perspective from which to examine these relations. As discussed by Canham (2009, p. 2), masculinity has been conceptualized as

a cultural construct that may be defined at its core by certain physical features and an inner sense of being male. It is also characterized by certain behavioral and affective traits such as toughness, power, control, independence . . . among other characteristics.

Focusing on the trait of independence, Smith and colleagues (2007) suggest that “key aspects of independence change for men as they age” (p. 333). Although independence at younger ages may be “health damaging” through men’s reluctance to seek health care, older men, seeking to preserve independence in the context of healthy aging, may use health services to maintain health. For example, relative to younger men with arthritis, age served as a protective factor among older men with arthritis, appearing to reduce expectations of strength and independence associated with their social roles (Gibbs, 2008). O’Brien and colleagues (2005) found older men more willing to seek medical care than their younger counterparts, seeming to balance concerns with “waiting it out” with increased risks of harm associated with waiting as they aged. With regard to preventive health service use, this suggests that use varies by gender, but the impact of gender changes as individuals age.

Much as gender and age interact, gender and disability may interact. O’Brien and colleagues (2005) found that among men who had survived a serious illness, there was a greater openness to seeking preventive care. Interpretation of a disease’s symptoms in the context of social roles may also influence health care use (Calasanti, 2010; Solimeo, 2008), including that of preventive care. Age of onset, type of disability, and other characteristics of disability likely further shape the relations between gender and disability (Shuttleworth, Wedgworth, & Wilson, 2012). Thus, the impact of disability may differentially influence men and women in their use of preventive services and in the context of maintaining health.

Hypotheses:

• H₁: We hypothesize that the relations between disability, gender, and receipt of preventive services will vary by age.

• H₂: We hypothesize that receipt of preventive services will be higher among older relative to nonelderly adults.

Data Source and Sample

This study uses the MEPS. The MEPS is a nationally representative survey of noninstitutionalized adults and children in the United States. The MEPS uses the NHIS as its sampling frame. MEPS data are collected through in-house, in-person interviews, following a multistage probability design. Subpopulations, including blacks and Hispanics, are oversampled to yield greater precision of estimates for those groups. In addition to sociodemographic and health insurance information, data are collected on health status, functional limitations, and health care access and utilization. We pool five MEPS panels for years 2001–2007, so as to attain an adequate sample size of adults with and without disabilities (N = 46,145).

Variables

Dependent Variables.

We examine the receipt of seven preventive and screening services. Five of these services are U.S. Preventive Services Task Force (USPSTF) “A”- or “B”-rated services: (a) counseling to quit smoking, (b) blood pressure screening, (c) screening for cervical cancer for women aged 21–64 every 3 years (the USPSTF does not recommend screening for cervical cancer in women aged 65 and older who have adequate screening at earlier ages), (d) screening for breast cancer among women aged 40–75 by means of mammography every 2 years (although this was recently revised upward to age 50, age 40 was the recommended age during the years of our analyses), and (e) screening for colon cancer among adults aged 50–75 by either sigmoidoscopy every 5 years or colonoscopy every 10 years. (The USPSTF recommends screening for adults aged 50–75; as the MEPS question asks whether the respondent “ever received” a sigmoidoscopy or colonoscopy, we do not set an upper age limit, consistent with AHRQ [2011]. The MEPS question for receipt of a fecal occult blood exam is asked “if ever received” rather then within the previous 2 years, the USPSTF measure, and is not included as a screening measure in our study.) Intensive counseling and behavior modification for adults who are obese is a “B”-rated recommendation. We use the MEPS measure related to counseling for exercise among adults who are obese as an initial examination of this prevention activity among adults. This service is also tracked in the AHRQ National Healthcare Disparities Report (AHRQ, 2012), as is our seventh measure, cholesterol screening. (The USPSTF recommendations for cholesterol screening vary by gender, age, and risk; we do not have information specific to risk factors in the MEPS.) We examine receipt of each of these services as a discrete service.

We also create a summary measure that reflects the percentage of services received, as appropriate for age, gender, and health risk. For example, a woman aged 45 who smokes should receive screening for blood pressure, cholesterol, cervical, and breast cancer, as well as counseling to quit smoking. We calculate the percentage of the five recommended services she receives. A man aged 70 who does not smoke but is obese should receive screening for blood pressure, cholesterol, and colon cancer, as well as counseling related to exercise; we calculate the percentage received for four services.

Independent Variables.

Our analyses focus on the relations between gender, disability, and age among adults aged 18 and older. Gender and age are both predisposing characteristics in Andersen’s framework. We analyze the effects of age by two subgroups: adults aged 18–64 and adults aged 65 and older. Disability represents a measure of health care need in Andersen’s model. Our measures of disability begin with six disability types, drawn from the International Classification of Functioning, and include sensory, physical, cognitive, functional, mental health-, and work-related disabilities (World Health Organization, 2001). Following Altman and Bernstein (2008), we combine these variables to create measures of basic activity (i.e., sensory, physical, cognitive, and mental health) and complex activity (i.e., functional and work) limitations. These variables are intended to capture different aspects of the disablement process (Altman and Bernstein, 2008, p. 10). To facilitate analyses, we create mutually exclusive measures of basic only, complex only, or both basic and complex activity limitations relative to individuals without disabilities.

Additional predisposing variables include race and ethnicity, marital status, and education. A usual source of care can be characterized as an enabling variable in Andersen’s framework. We categorize individuals as having a usual source of care that is a provider or a place relative to reporting no usual source of care. Income is included as a second enabling variable, with income measured as a ratio of family income to the Federal poverty threshold, adjusted for family size. Self-reported health status captures a second dimension of “need” and includes excellent, very good, good, or fair/poor self-rated health. To capture contextual influences, we include whether the individual resides in a metropolitan statistical area (MSA) and region.

Health insurance is an enabling variable (McWilliams, 2009). However, health insurance is difficult to incorporate in an analysis spanning working-age and older adults. Adults aged 18–64 are covered privately, through Medicare if they have significant disabilities and receive Social Security Disability Insurance, after a 2-year waiting period or through Medicaid if they are categorically eligible as a low-income youth or a low-income pregnant woman, have significant disabilities, are eligible for Supplemental Security Income or have low income, with some exceptions such as 209b states, or are uninsured. Some adults aged 18–64 will also be dually insured for Medicare and Medicaid. Approximately 96% of adults aged 65 and older are insured through Medicare; Medicaid and private insurance serve to supplement Medicare among older adults (Medicare Payment Advisory Commission, 2010). Thus, for example, private insurance among those aged 18–64 would largely be primary, whereas it would be supplemental among older adults. In our comparison of working-age and older adults, we do not include health insurance. We separately conducted analyses for working-age adults with health insurance included and examined whether this changes comparisons between younger and older adults.

Data Analyses

The MEPS uses a complex sample design, with stratification, clustering, and multistage sampling. Stata version 12 is used in our analyses. We employ the class of Stata survey procedures designed for use with complex surveys to account for the sample design. The MEPS includes sample weights that account for probabilities of selection, nonresponse, and poststratification. The MEPS provides the design variables needed to obtain population-weighted estimates of counts and proportions and correct standard errors for estimates.

We begin by descriptively comparing younger and older adults by gender, disability, and covariates and then compare their receipt of preventive services. To conduct our multivariate analyses, we use logit models to model the probability of receipt of a given preventive service as a function of gender; disability; and the predisposing, enabling, need, and contextual covariates described previously. We use an ordinary least squares regression for our summary measure. Given that we are interested in how gender and disability status might interact with each other to modify the probability of receiving a service, we use these discrete variables to create dummy variables for each of the eight possible gender–disability combinations (e.g., female without a disability, male with a complex disability). We enter all of these dummy variables into the regression except for men with no disabilities, which serves as the reference gender–disability category. We similarly use women with no disabilities as the reference category in our models of pap smear and mammography. To test differences among women by disability status in models that include both men and women, we use postestimation tests of significance. Use of gender–disability dummy variables is analytically similar to including interaction terms; in that, it permits effects to differ among different groups (Wooldridge, 2009). Incorporating gender/limitation combinations in the model simplifies the task of assessing the statistical significance of differences between various gender–disability groups and avoids the complexities associated with interpreting interaction terms in nonlinear models (Ai & Norton, 2003).

We use the logit models to estimate the predicted probability of receipt of the service for each gender–disability combination. To do this, we estimate the model and then systematically change the value of the gender–disability combination while leaving all other variables at their original values. So, for example, we set everyone to be female with no disability, then use the postestimation predict function to calculate the predicted probability of receipt of a given service. We report predicted probabilities for the various gender–disability combinations and use standard t tests to evaluate whether differences between groups are statistically significant. We leave these as predicted probabilities because these estimates provide information regarding the level of the effect (e.g., a predicted probability of .64 for women with no disabilities suggests that the average predicted probability for women with no disabilities is .64, with all other covariates evaluated at their original values in the sample) and because they lend themselves to looking across the full set of gender–disability combinations. Finally, to examine age differences, we estimate the above logit models separately for adults aged 18–64 and for adults aged 65 and older (with different age cuts for mammography and colonoscopy as discussed previously). We then tested the joint significance of the difference in coefficients for the gender and disability coefficients between working-aged and older adults on each dependent variable adjusted for covariates.

Findings

Table 1 presents descriptive information on the study sample. Significantly more older adults were female relative to male. A higher share of older adults were non-Hispanic White, whereas a higher share of nonelderly adults were black or Hispanic. More older adults lacked a high school degree, whereas more nonelderly adults held college degrees. Although a higher share of older adults were widowed, a higher share of working-age adults were never married. Almost 90% (89.06%) of older adults had a usual source of care who was a person compared with 62.8% among adults aged 18–64. A quarter (25.43%) of working-age adults had no usual source of care compared with 6.23% among older adults. Older adults self-reported poorer health, with 25.8% reporting fair or poor health relative to 12.1% among working-age adults. Although the share of both younger and older adults with complex activity limitations was similar, older adults had a higher share with basic activity or both basic and complex activity limitations. Finally, receipt of preventive services was higher among older adults across all services. Using our summary measure, on average, older adults received 75.7% of recommended preventive services relative to 60% among working-age adults.

Table 2 presents predicted probabilities for each service and the summary measure for each gender and disability combination by age. In the following discussion, we discuss only those findings that are statistically significant, as reported in the table.

Differences in Receipt of Services Between Women With and Without Disabilities

Aged 18–64.

Among working-age women, those with basic activity limitations (pp = .86) and both basic and complex activity limitations (pp = .92) had higher predicted probabilities for blood pressure screening than did women with no disabilities (pp = .83). Women with complex activity limitations (pp = .61) and both basic and complex activity limitations (pp = .64) had a higher probability of cholesterol screening than did women with no disabilities (pp = .54). Screening for cervical cancer was lower among women with basic activity (pp = .82) and both basic and complex activity limitations (pp = .80) relative to women with no limitations (pp = .83). Screening for breast cancer by mammography was significantly higher among women with complex activity limitations relative to women with no limitations (pp = .77 vs .72, respectively). Similarly, the probability of a colonoscopy/sigmoidoscopy was higher for women with basic activity (pp = .44), complex activity (pp = .50), and both basic and complex activity (pp = .48) limitations than for women with no limitations (pp = .42). Among women aged 18–64 who were obese, receipt of counseling related to exercise was higher for women with basic activity (pp = .66), complex activity (pp = .65), and both basic and complex activity limitations (pp = .64) relative to women with no limitations (pp = .59). For women who smoked, receipt of counseling to quit smoking was higher for those with basic activity (pp = .55) and both basic and complex activity (pp = .58) limitations compared with women with no limitations (pp = .52). Among women aged 18–64, we did not observe any significant differences across the type of disability with regard to the percentage of preventive services received.

Aged 65 and Older.

Although similar patterns were observed among older women, there were fewer significant differences. Older women with basic activity (pp = .96) or both basic and complex activity limitations (pp = .98) had a higher probability of receipt of blood pressure screening than do older women without a disability (pp = .94). No statistically significant differences by disability were observed for screening for cholesterol or breast cancer. Older women with a basic activity limitation had a higher probability of receipt of a colonoscopy/sigmoidoscopy relative to older women without a limitation, pp = .59 versus pp = .56, respectively. Older women who were obese and had a basic activity limitation had a higher probability of receipt of counseling related to exercise (pp = .69) relative to older women who were obese without a disability (pp = .64). No significant differences related to receipt of counseling for older women who smoked were observed by disability status. With regard to the summary variable, among older women, women with complex activity limitations had a significantly lower predicted percentage of services received relative to older women with no limitations, 69% versus 75%, respectively.

Differences in Receipt of Services Between Men With and Without Disabilities

Aged 18–64.

Among men aged 18–64, blood pressure screening for males at all levels of disability was higher than for males without, with predicted probabilities of .75 (basic activity), .87 (complex activity), and .88 (both basic and complex activity limitations) relative to .69 for men with no limitations. Cholesterol screening among males was higher for those with basic activity (pp = .49), complex activity (pp = .62), and both basic and complex activity limitations (pp = .64) relative to men with no limitations (pp = .46). Similarly, men aged 50–64 with basic activity (pp = .45), complex activity (pp = .54), and both basic and complex activity limitations (pp = .47) had higher probabilities of receipt of a colonoscopy/sigmoidoscopy relative to men without a disability (pp = .40). Among working-age men who were obese, those with basic activity (pp = .59), complex activity (pp = .61), and both basic and complex activity limitations (pp = .61) had higher probabilities of receipt of counseling than did men without disabilities (pp = .51). For men aged 18–64 who smoked, predicted probabilities for receipt of counseling were also significantly higher among those with basic activity (pp = .46), complex activity (pp = .56), and both basic and complex activity limitations (pp = .57) relative to men without disabilities (pp = .41). In contrast to working-age women, men with basic activity (55%), complex activity (66%), and both basic and complex activity limitations (65%) had significantly higher percentage of preventive services received than those men with no disabilities (49%).

Aged 65 and Older.

Among older men, those with basic activity (pp = .95) and both basic and complex activity limitations (pp = .97) had higher predicted probabilities of blood pressure screening relative to their counterparts without disabilities (pp = .93). There were no significant differences by disability status for cholesterol screening among men aged 65 and older. Although older men with basic activity limitations had a higher predicted probability of screening for colon cancer (pp = .60), older men with complex activity limitations had a lower predicted probability (pp = .44) relative to older men without a disability (pp = .56). For older men who were obese, those with basic activity limitations had a higher predicted probability of receipt of counseling relative to those older men without a disability, pp = .72 versus pp = .62. Older men with both basic and complex activity limitations had a higher probability of receiving counseling to quit smoking (pp = .72) than did their counterparts without a disability (pp = .61). There were no significant differences across disability among older men with regard to the percentage of recommended services received.

Gender and Disability Differences by Age

We tested the previously discussed age differences with a test of the joint significance of the gender and disability coefficients by age for each screening measure. All were significantly different by age, with older adults having higher predicted probabilities of receipt of preventive services. Additionally, older adults across gender and disability received a higher share of the recommended preventive services than their younger counterparts.

Effects of Covariates

The most consistent covariate across services and age groups was a usual source of care. Table 3 provides the predicted probabilities by service and age for individuals with no usual source of care, a usual source of care that is a place (e.g., outpatient clinic), and a usual source of care that is a person. In all cases, having a usual source of care significantly increased the predicted probability of receipt of a given service, with the effect of a person as a usual source of care having the largest effect. For example, among adults aged 18–64, having a usual source of care that is a place had a predicted probability of receiving 59% of recommended services, whereas a person was associated with a predicted 63% of services relative to 47% for those without a usual source of care. Among older adults, the predicted percentage of preventive services received for no usual source of care was 52% and that for a place or a person as a usual source of care were 71% and 76%, respectively. Moreover, having a usual source of care is more critical to older adults and their receipt of preventive services, as the test of the joint significance for this variable between younger and older adults for all measures was significant.

Finally, to examine whether the observed higher receipt of services among older adults was related to almost universal health insurance through Medicare, we added health insurance as a covariate to the models for working-age adults and assessed whether the predicted probabilities changed (data not shown). For the most part they did not, and when they did change, the change was never more than 2 probability points. The exception was counseling among men with complex activity limitations who were smokers. Here, adding health insurance reduced the predicted probability by 4 points, making it virtually the same point estimate for working-age and older adults. Across services, there were no significant effects of public insurance on receipt of services relative to individuals with private insurance. Individuals who were uninsured had a significantly lower predicted probability of receipt of preventive services across all measures.

Discussion

We examined the receipt of seven preventive services among adults aged 18 and older, examining relations between gender, disability, and age. For most services, both women and men with disabilities had higher probabilities of receiving these services relative to those without disabilities. There were more significant differences by disability among adults aged 18–64 relative to adults aged 65 and older, supporting our first hypothesis that the effects of gender and disability would differ by age.

Although both women and men with disabilities had higher probabilities of screening than their counterparts without disabilities, there was a pattern of more significant differences across types of disability for men relative to women, particularly among adults aged 18–64. For screening for blood pressure, cholesterol, and colon cancer, as well as counseling related to smoking and exercise, predicted probabilities for men aged 18–64 with basic, complex, or both basic and complex activity limitations were significantly higher than for men with no disabilities. This was observed among women aged 18–64, only for screening for colon cancer and counseling for exercise among women who were obese. These findings are consistent with the work of O’Brien and colleagues (2005), finding gender effects to differ by illness experiences, with the experience of a life threatening illness related to men being more open to preventive care. The experience of disability may also influence men to be more willing to seek preventive care to maintain health. The pattern of more significant differences by disability among men was observed among older adults but was not as pronounced. It may be that age-related gender differences, with age appearing to be protective for men in their use of medical care (Gibbs, 2008; O’Brien, et al., 2005; Smith et al., 2007), attenuates the role of disability in seeking health care among older men.

There were few exceptions to the pattern of higher probabilities of service use for those with disabilities. For women aged 21–64, predicted probabilities for screening for cervical cancer among women with basic and both basic and complex activity limitations were significantly lower than for women without disabilities. The differences by disability in screening for cervical cancer are consistent with previous studies (Iezzoni et al., 2001; Wei et al., 2006); although statistically significant, the substantive differences were modest, one or three probability points. With regard to colonoscopy/sigmoidoscopy, older men with complex activity limitations had a significantly lower predicted probability of screening, 12 probability points. Counseling related to smoking was lower among older men with complex activity limitations relative to older men without a limitation, with a difference of 10 probability points. Previous work has found some differences by type of disability (Ahmed et al., 2009; Iezzoni et al., 2001), although findings are mixed, as previously discussed. Further understanding the relations between gender, type and duration of disability, and receipt of preventive services is important, particularly given the substantive differences we observe for some services.

Receipt of clinical preventive services also was higher among older adults relative to working-age adults. This supports our second hypothesis. Our general finding of higher receipt among older adults is consistent with Shenson and colleagues (2005), who also report that a higher percentage of adults aged 65 and older received a set of core clinical services than adults aged 50–64, using BRFSS data. Our finding holds after adjusting for health insurance among adults aged 18–64 and having a usual source of care for both younger and older adults. In part, our finding of higher receipt among older adults may relate to a greater exposure to the health care system, with more opportunities to provide basic screenings such as blood pressure. Further examining the relative importance of factors behind differential levels of preventive care is important, so as to improve receipt of these services across age groups.

Having a usual source of care is a substantively important predictor, with a significantly greater effect for older adults (Table 3). The importance of a usual source of care is consistent with previous literature (Blewett, Johnson, Lee, & Scal, 2008; Doescher, Saver, Fiscella, & Franks, 2004; Shenson et al., 2011). A quarter of adults aged 18–64 in our study had no reported usual source of care. Improving their access to a usual source of care could substantially increase their receipt of preventive services. For example, the predicted probability of receipt of mammography among women aged 40–64 would increase from .52 for women without a usual source of care to .77 for women with a usual source of care who was a provider. Among older adults, 6.23% reported no usual source of care and 4.71% reported a usual source of care that was a place. Increasing their access to a provider as a usual source of care would also significantly improve receipt of preventive services. For example, older women with no usual source of care had a predicted probability of .47 for receipt of mammography. This increases to .69 for a usual source of care that is a place and .78 for a usual source of care that is a person. Inclusion of higher reimbursement in the Affordable Care Act (ACA) for primary care providers (CDC, 2012; Kaiser Family Foundation, 2010) may facilitate access to a usual source of care among older adults. The addition of Medicare’s new Annual Wellness Visit may also facilitate discussion of and receipt of preventive services (Benson & Aldrich, 2012).

Given the observed importance of a usual source of care, understanding factors that facilitate or impede access to a usual source of care continues to be important. For example, data from the AHRQ National Disparities report (2009) indicate that black and Hispanic adults have lower odds of having a usual primary care provider. For black adults, having a usual primary care provider has worsened over time relative to their white counterparts. Although some work has found accounting for differences in socioeconomic status and insurance explain much of the observed ethnic disparities in having a usual source of care, these factors do little to alter racial disparities (Weinick, Zuvekas, & Cohen, 2000). The changing policy environment associated with the implementation of the ACA (e.g., support of medical homes that are constructed around a usual source of care; Fishman et al., 2012) further supports the continued examination of factors related to a usual source of care, with a focus on policy-amenable factors such as health insurance, provider reimbursement, and new models of medical care.

Across gender, disability, and age, receipt of a range of clinical preventive services is suboptimal. For example, the predicted probability for screening for colon cancer ranged from .40 among working-age men and .42 among working-age women with no disabilities to .60 among older men with basic activity limitations and .59 among older women with basic activity limitations. Receipt of services more routine to an office visit (e.g., blood pressure screening) tended to be high, particularly among older adults. Men aged 18–64 without disabilities were an important exception to this pattern, suggesting the need to design health promotion programs with the influence of gender in health care seeking in mind (Smith et al., 2007). For certain services that we examined (e.g., screening for breast cancer), the effect of a usual source of care brought receipt of preventive services substantially closer to state-level benchmarks (AHRQ, 2012). To the extent financial costs pose a barrier, the ACA authorized coverage for USPSTF “A” and “B” recommendations in Medicare with no copayments, included financial incentives for their provision in state Medicaid programs, and also expanded availability in private insurance plans, again with no cost sharing (Kaiser Family Foundation, 2010). These provisions should enhance receipt of preventive services. Provider factors, such as knowledge, time constraints, adequate reimbursement, and institutional support, also contribute to lower receipt of certain services such as screening and counseling related to smoking (Fiore et al., 2008). In its recently released report, CDC (2012) recommends the implementation of evidence-based practices targeted both to patients and clinicians as one mechanism to increase provision.

Related, one of four strategic directions of the National Prevention Strategy is Clinical and Community Preventive Services, intended to “Ensure that prevention focused health care and community prevention efforts are available, integrated and mutually reinforcing” (National Prevention Council, 2011, p. 11). Action steps the Council identifies include the previously discussed financial mechanisms to encourage provision. Additionally, the Council emphasizes reducing barriers to access, implementing community-based preventive services, and enhancing their linkage to clinical care. Model programs, such as CDC’s Sickness Prevention Achieved through Regional Collaboration (SPARC), have demonstrated success in increasing use of prevention services among adults aged 50 and older through such partnerships (Shenson, Benson, & Harris, 2008). Further expanding community-based programs will be important to increasing the receipt of clinical services and reducing observed disparities.

It is important to note some study limitations. Our use of MEPS did not allow us to explore certain dimensions of Andersen’s framework. For example, we did not have measures of potentially important enabling community factors, such as physician supply. Related, certain factors that may influence receipt of preventive services, such as beliefs about the efficacy of preventive care or cultural influences on health care seeking (Alexander, Miller, Cotch, & Janiszewski, 2008) are not included in the MEPS survey. Our data may have affected findings related to certain services, as the questions posed do not directly track with USPSTF recommendations. For example, we may underestimate screening for colon cancer, due to the way survey participants are queried about receipt of screening, particularly through the fecal occult blood test. Data from the BRFSS for the year 2012 suggest that this accounts for approximately 10% of screening for colon cancer (Joseph, King, Miller, & Richardson, 2012). The question related to counseling regarding exercise among individuals who are obese is only a rough approximation to the USPSTF measure, although the question is tracked in national disparities reports (AHRQ, 2012). It is also possible that selective mortality among men (Leveille et al., 2000) may influence study findings, particularly with regard to older men.

In summary, older women experience higher rates of disability that could be remedied in part through greater receipt of preventive care at younger ages. We extend previous working by examining receipt of preventive services to consider relations between gender, disability, and age. We find disability to be an important predictor of greater receipt of services, with the effect observed more consistently with men, particularly those aged 18–64. Receipt of preventive services was also significantly higher among adults aged 65 and older relative to adults aged 18–64. Enhancing access to a usual source of care, particularly among adults aged 18–64, would substantially increase their probability of receipt of preventive services. Designing health promotion programs that reflect gender and age differences in health seeking would likely increase use as well. More effectively, linking clinical care with community-based interventions, combined with the use of evidenced-based practices for receipt of preventive care, would further increase receipt of services across the life span. Continued exploration of the relations between gender, disability, and age in receipt of preventive services is warranted, given the substantive effects we observe.

Funding

This study was supported by a grant from the National Institute on Disability and Rehabilitation Research, U.S. Department of Education .

References

Author notes