Search tips
Search criteria 


Logo of acmMary Ann Liebert, Inc.Mary Ann Liebert, Inc.JournalsSearchAlerts
Journal of Alternative and Complementary Medicine
J Altern Complement Med. 2009 August; 15(8): 911–919.
PMCID: PMC3191375

A Multivariate Test of an Expanded Andersen Health Care Utilization Model for Complementary and Alternative Medicine (CAM) Use in African Americans

Carolyn Brown, Ph.D.,corresponding author1 Jamie Barner, Ph.D.,1 Tom Bohman, Ph.D.,2 and Kristin Richards, Ph.D.3



The objectives of this study were (1) to determine which Andersen Model variables [predisposing, enabling, and need (PEN)] are related to complementary and alternative medicine (CAM) use by African Americans in the past 12 months; and (2) to determine whether the addition of disease states to the Model will explain significant variation in CAM use in the past 12 months.


The 2002 National Health Interview Survey was used with 4256 African American adults (n = 23,828,268 weighted) selected as the study population. The dependent variable, CAM Past 12 Months, represented participants' use of at least 1 of 17 CAM modalities during the past 12 months. The Andersen Model variables [predisposing (e.g., age); enabling (e.g., insurance); and need (e.g., medical conditions)] and prevalent disease states (≥10%) comprised the independent variables. Logistic regression analyses, incorporating the sampling weights, were employed.


Among predisposing factors, CAM use was associated with middle-aged to older, more educated, and female African Americans. Region (Northeast less likely than South) was the only significant enabling factor. Need factors had the most frequent relationships, with more medical conditions, more physician visits, better health status, prescription and over-the-counter medication use, more frequent exercise, and having activities of daily living limitations being associated with CAM use. After adjusting for PEN factors, the disease states of pain/aching joints, recurring pain, and migraine were related to CAM use.


African American CAM users are middle-aged to older, female, educated, and have more medical conditions (especially pain-related). Users report higher utilization of “traditional” care (e.g., physician visits), indicating that CAM is likely a complement to conventional treatment in this population. Health care providers should use these factors as prompts for inquiring about CAM use in African American patients.


Recent national studies revealed that a majority (67.6%–71.3%) of African American adults use complementary and alternative medicine (CAM) in a typical year.1,2 The most common therapies include prayer specifically for health reasons, herbals, and relaxation. CAM use is prevalent among people of all races/ethnicities with chronic conditions,24 and this holds true for African Americans as well. The prevalence of common disease states in African Americans (e.g., hypertension) is significantly higher (range: 1.4–2.9 times) for CAM users compared to nonusers.1

CAM use varies in African Americans. Compared to other ethnic and racial groups, African Americans are less likely to use body-based techniques (e.g., chiropractic care),57 dietary strategies (e.g., nutritional supplements, herbals),7,8 and psychologic therapies.9 However, they are more likely to use home remedies6,10,11 and mind/body (e.g., relaxation) and spiritual healing practices.2,12,13 One (1) study found that 35.4% of African American adults used home remedies.14 Use of prayer in coping with illness is common among African Americans,15 and religiosity and prayer have been found to be significant in managing hypertension, along with prescribed medication.16

National estimates indicate that African American CAM users are more likely to be older, female, college-educated, and insured.1 Previous studies have shown similar characteristics of users except that lower education had traditionally been a predictor of CAM use.14,1719 Socioeconomic issues such as lower income, urbanicity, living with a grandparent as a child, comorbidities, influence of family/friends/social group, community group involvement, and less access to care have all predicted CAM use in African Americans.12,14,17 Although most factors were identified in convenience samples, some (e.g., community group involvement) are unique predictors in African Americans and may give some insight into culturally related variables that might impact CAM use. Finally, African Americans and need for care variables have rarely been examined in CAM studies. Two (2) studies found that significant predictors of CAM use included a higher number of prescription medications, use of a caregiver, and current cigarette smoker.8,20 How representative these findings are among African Americans is unknown.

Until recently, there was limited generalizable evidence that African Americans were substantial users of CAM. Therefore, more information is needed about their CAM use on a larger scale and about intragroup variations in use as well. Moreover, most CAM studies examining African Americans have been descriptive and atheoretical; thus, factors identified as affecting their CAM use should be examined in a multivariate, theoretically based framework to better identify the unique contribution of each predictor.

Theoretical Framework

The Andersen Healthcare Utilization Model,21 the theoretical framework for this study, has been used extensively to examine relationships between predisposing, enabling, and need for care (PEN) factors and health care utilization.2229 The model purports that health care utilization is dependent on individuals' propensity to use services (predisposing), their ability to access services (enabling) and their illness level (need). Our study extends the Andersen Model to include the most prevalent disease states among African Americans and aims to determine whether these variables add additional prediction to CAM use while adjusting for PEN variables. The conceptual framework is shown in Figure 1.

FIG. 1.
Study Model. CAM, complementary and alternative medicine.

Overall, need for care factors are the strongest predictors of health care utilization, followed by enabling and predisposing factors.25 Andersen's Model is a useful framework for studying CAM use, a form of health care utilization, and for grouping the factors shown to affect CAM use in African American adults. Two (2) studies have used a variation of the Andersen Model for studying CAM use in selected populations,30,31 and 1 found that 28% of the variance in CAM use in patients with cancer was explained by PEN factors.31

African Americans are substantial users of CAM, yet there is little generalizable evidence about which factors predict their use of CAM modalities, including those that are unique (e.g., use of a caregiver) to African Americans. Thus, further examination of CAM use in African Americans using a nationally representative sample is needed. Moreover, no known studies have (1) assessed their CAM use utilizing a theoretical model; or (2) included disease states as a predictor. We hypothesize that the PEN variables and disease states will be significant predictors of CAM use in African Americans and that need variables will be the strongest predictors. The presence of identified predictors could serve as prompts to health care professionals to inquire about CAM use, especially since most CAM users do not voluntarily reveal this information.4,32,33 Thus, uncovering factors associated with CAM use is important, particularly when used concurrently with conventional care and potential problems (e.g., drug–herbal interactions) could be avoided.


The objectives of this study are the following.

  1. Determine which Andersen Model variables (predisposing, enabling, and need [PEN]) are related to CAM use by African Americans in the past 12 months; and
  2. Determine whether the addition of disease states to the Andersen Model will explain significant variation in CAM use by African Americans in the past 12 months.

Materials and Methods

Data source

Data from the 2002 National Health Interview Survey (NHIS) were used for this study. The 2002 NHIS population comprised all civilian, noninstitutionalized persons who resided in the 50 states and the District of Columbia in 2002. Only African American adults (reported as black non-Hispanic), an oversampled group in the 2002 NHIS survey, were included. The sample included 4256 adults (≥18 years of age) with the weighted sample representing 23,828,268 African American adults nationwide. The Institutional Review Board of The University of Texas at Austin approved this study.

CAM utilization measures

The dependent variable, CAM Past 12 Months, represents whether they had used or seen a practitioner for 17 CAM modalities during the past 12 months. The 17 modalities included acupuncture, Ayurveda, biofeedback, chelation therapy, chiropractic care, energy healing therapy/Reiki, folk medicine, herbal products, homeopathic treatment, hypnosis, massage, megavitamin therapy, naturopathy, prayer, relaxation techniques, special diets, and yoga/t'ai chi/qigong. Additional analyses of CAM use excluding prayer were conducted.

Andersen Model factors

The independent predictors have been empirically shown to impact CAM use; however, some variables unique to African Americans were not available in the NHIS database. Predisposing factors increase a person's need for health care services. Enabling factors facilitate (or impede) use of health care services. Need factors are related to illness that requires use of health care services. Disease state factors include 13 illnesses with at least 10% prevalence in the NHIS sample. Table 1 includes a description of all the predictors.

Table 1.
Predictor Descriptions

Data analysis

The data analysis involved analyzing CAM use (with and without prayer) in the past 12 months. The analyses used a generalized linear model with a binomial distribution for the outcomes.34 The logistic regression analyses were conducted using Proc Surveylogistic in SAS Version 9.1.3 SP4. This procedure accounts for the complex, multistage cluster sampling design used in the NHIS.

Due to the large number of predictors in the model, the effective sample size would have been reduced from an unweighted sample of 4256 to 3011 (29% loss of respondents) due to missing values on individual predictors. The analyses used multiple imputation to account for missing data across all of the predictors. SAS Proc MI was used to create 10 datasets with different imputed missing values. The logistic regression model using the survey weights was then run on each of the 10 imputed datasets and the results were integrated using Proc MIanalyze, which correctly averages the 10 estimates for each parameter and then computes an adjusted t-test statistic.

The predictors in Model 1 were based on the Andersen Model. Model 2 tests whether the addition of disease states adds incremental predictive power. The multivariate Wald χ2 test was used to evaluate the statistical significance of the set of coefficients associated with the disease states. Adjusted odds-ratios (AOR) are reported, which represent the unique, additional explanation provided by an individual predictor. A p-value less than 0.05 was considered statistically significant.


A total of 67.6% of African Americans used CAM in the past 12 months, when prayer for health reasons was included. When prayer for health reasons was not included, only 27.0% reported using CAM in the past 12 months. Table 2 shows descriptive statistics for each variable in the analysis. Table 3 shows the results of the tests of each Andersen Model predictor under Model 1. Of the predisposing predictors, three were related to CAM use: age, education, and gender. Middle-aged to older, more educated, and female African Americans were more likely to use CAM. In particular, the following were statistically significant: age group 35–44; age group 45–54; high school degree; some college experience; bachelor's degree; advanced degree; and female.

Table 2.
Frequencies/Means of Model Factors
Table 3.
Adjusted Odds Ratios and Associated p-Values for Two Models Predicting Complementary and Alternative Medicine Use in the Past 12 Months

Of the enabling predictors, only region had a unique relationship to CAM use. Northeast region respondents were less likely to use CAM compared to those in the south.

Of the need predictors, characteristics related to CAM use were the following: higher total medical conditions; health status change as better; higher number of physician visits; prescription medication use; OTC medication use; more vigorous activity <3 times per week; more vigorous activity ≥3 times per week; moderate activity <3 times per week; moderate activity ≥3 times per week; and an activities of daily living (ADL) limitation.

After adjusting for the predictors in Model 1, the disease states of pain/aching joints (30 days), recurring pain, and severe headache predicted CAM use. The multivariate Wald test showed a statistically significant effect for the set of predictors [Wald F(13) = 18.9, p < 0.001].

Analyses of CAM 12 months without prayer

Additional analyses were conducted excluding prayer from CAM use (i.e., CAM without prayer) and results indicate that characteristics of African American CAM users differ when prayer for health reasons is excluded. Regarding predisposing factors, age is no longer significant. Although education and gender remain significant, the relationship with education is much stronger, with more education related to CAM use: high school degree (AOR = 1.76, p < 0.000); some college or associate's degree (AOR = 2.36, p < 0.000); bachelor's degree (AOR = 3.16, p < 0.000); and advanced degree (AOR = 4.20, p < 0.000). Additionally, family size emerged as a significant predictor, with those with a smaller family size more likely to use CAM (without prayer) (AOR = 0.90, p < 0.016).

Additional enabling factors were significant in the CAM without prayer model. Respondents who delayed care were more likely to use CAM (without prayer) (AOR = 1.33, p = 0.038). Also, compared to the south, all regions were more likely to use CAM (without prayer): midwest (AOR = 1.26, p < 0.000), northeast (AOR = 1.49, p < 0.000), and west (AOR = 1.55, p = 0.006).

The majority of differences between CAM users with and without prayer occurred with need factors. The following were not significant when prayer was included but emerged as significant predictors once prayer was excluded: anxiety/depression (AOR = 1.41, p = 0.004); strength activity <3 times per week (AOR = 1.71, p < 0.000); and strength activity ≥3 times per week (AOR = 2.23, p < 0.000). The following factors were significant when prayer was included, but were no longer significant once prayer was excluded: physician visits, OTC medication use, and moderate activity <3 times per week. All other relationships were the same as those when prayer was included.

When disease states were added, delayed care and mid-west region were no longer significant, which were similar to the CAM with prayer model. However, when comparing disease states, there were differences in CAM with and without prayer. In CAM without prayer, pain/aching joints, recurring pain and severe headache were no longer significant, whereas insomnia (AOR = 1.50, p = 0.004) was the only disease state that emerged as significant.


This use of a nationally representative sample enables our findings to be generalized to all civilian, noninstitutionalized African Americans residing in the United States. A notable contribution of our study is the sensitivity analysis that predictors of CAM use differed when prayer for health reasons was included versus excluded.

Utility of the Andersen Model in predicting CAM use

Predisposing factors

Consistent with previous CAM research in the general population,1 our results show that African American CAM users compared to nonusers are middle-aged to older, more educated, and female. Older age as a predictor of CAM use is consistent with prior research in African Americans (using convenience samples and examining bivariate relationships) and supports the substantial evidence in the religion and health literature, which shows high levels of religiosity in older African Americans. A recent study showed that older African Americans were more heavily involved in prayer life across several dimensions of prayer compared to whites.35 On the other hand, education was also positively and significantly related to CAM use—a finding that is inconsistent with prior research in African Americans. The reversal of education effects points to the value of examining predictors in a multivariate framework as well as across a broad spectrum of CAM modalities. This difference in education effects also could be explained by our use of a nationally representative sample.

Enabling factors

Only regional differences were significantly related to CAM use, indicating that users are more likely to reside in the south compared to other regions. Regional variations in prayer are well established,36 and our findings further support this relationship, particularly in light of the pervasive use of prayer in the southern population. Future research should examine reasons for regional differences such as cultural variation among African Americans across regions or regional variations in the use of specific CAM modalities. Although past research has shown bivariate relationships with income and insurance and CAM use,1 the current research shows no relationships between traditional access indicators and CAM use. Again, this is not surprising given that prayer was such a dominating CAM modality used by African Americans. Although CAM use may stem from a lack of access to conventional care, this was not supported in our study (although there was a large enough effect for the delay care variable [AOR = 1.44] that future studies may examine this issue more carefully). Including prayer may affect these relationships since some people may wait for prayer to “work” before using alternatives. Whether or not employment of CAM contributes to that delay should be studied.

Need factors

As expected, the need predictors (e.g., higher number of medical conditions) were the most predominant predictors of African Americans' CAM use, even when adjusting for enabling and predisposing factors. CAM users were higher health care utilizers and engaged in more physical activities compared to nonusers. Our results seem to indicate higher levels of health focus in CAM users, which may be linked to their tendency to report better health status compared to the previous year. Taken together, these findings indicate that African Americans use CAM as a supplement to conventional medicine and reflect a somewhat health-conscious orientation in CAM users.

Disease factors

Only pain-related diseases predicted CAM use. People who suffered with pain/aching joints, recurring pain and severe headaches (including migraines) were more likely CAM users compared to those without these conditions.

Use of CAM excluding prayer

When examining CAM use, studies may or may not incorporate prayer. Our sensitivity analysis highlighted changes in relationships, which may help in the understanding of the impact of prayer.

When prayer was excluded, the relationship between age and CAM use no longer existed, further validating the extensive use of prayer in middle-aged to older African Americans.35 The strength of the association between education and CAM use increased substantially, which suggests that this group was more educated than those who pray. Perhaps this finding reflects a more self-care orientation among CAM users (without prayer), especially since the other more prevalent CAMs (e.g., herbals and relaxation techniques) tend to be chosen by proactive consumer types.

Regional differences indicate that CAM users more likely reside in the northeast and west as opposed to the south, implying that prayer is more prevalent in the south. In addition, CAM users (without prayer) tended to delay care and had less contact with the health care system (e.g., medical visits) compared to CAM users with prayer, although anxiety and depression were more prevalent among those who did not use prayer. Whether or not this connotes substitution of conventional care with CAM should be studied further.

Finally, CAM users who do not pray for health reasons were more likely to have insomnia as opposed to the increased likelihood of pain-related conditions found in CAM users who pray. A possible connection between pain and use of prayer in African Americans is an interesting one and should be studied.

In summary, CAM use is substantial among U.S. adults and African Americans. Previous literature portrayed African American CAM users as uneducated, sick, and poor. However, our findings reveal that African American CAM users have quite a different profile—a health-conscious lifestyle profile that is adopted due to greater attention to disease prevention and control. Generally, health care professionals should routinely inquire about CAM use among all patients. However, according to our study, they should suspect CAM use among African Americans who are middle-aged to older, more educated, female, and Southern. CAM users also tend to be big health care utilizers and have a greater number of chronic conditions, pain-related diseases, and ADL limitations, but they also report better health status compared to the previous year and greater physical activity levels. Health care providers should use these factors as prompts for inquiring about CAM use in African American patients.


This study has several limitations. For several variables in the NHIS, not all questions were asked in a way to precisely measure each variable. For example, one item in the NHIS asks if any household member is covered by health care insurance. This item was used as a proxy for individual coverage. In addition, variables found to be unique predictors of African American CAM use (e.g., living with a grandparent) were not included in the database. The validity of some questions may have been problematic and not worded in language commonly used by African Americans. For example, many African Americans use garlic as a home remedy, but “home remedy,” a familiar term among many African Americans, was not offered as a choice in the survey. Thus, the prevalence of CAM use in African Americans may have been underestimated. Finally, CAM use was a binary variable that could not distinguish regular CAM users from periodic users or those who tried CAM on a limited basis.


African Americans are substantial users of CAM, and their use is predicted by predisposing, enabling, need, and disease (PEND) factors. Supporting our hypothesis, need variables were the strongest predictors of CAM use. Because of the differences in PEND factors between CAM users who pray and those who do not pray, health care professionals may consider these issues when conducting patient interviews and obtaining medical histories. Such information may be critical to creating optimal therapeutic plans.


This work was supported by the National Institutes of Health/National Center for Complementary and Alternative Medicine (1 R21 AT002858-01A1).

Disclosure Statement

No competing financial interests exist.


1. Brown CM. Barner JC. Bohman T. Richards K. Patterns of complementary and alternative medicine (CAM) use in African Americans. J Altern Complement Med. 2007;13:751–758. [PubMed]
2. Barnes PM. Powell-Griner E. McFann K. Nahin RL. Advance Data from Vital and Health Statistics. Hyattsville, MD: National Center for Health Statistics; 2004. Complementary and Alternative Medicine Use Among Adults: United States, 2002. No. 343. [PubMed]
3. Eisenberg DM. Kessler RC. Foster C, et al. Unconventional medicine in the United States: Prevalence, costs, and patterns of use. NEJM. 1993;328:246–252. [PubMed]
4. Eisenberg DM. Davis RB. Ettner SL, et al. Trends in alternative medicine use in the United States, 1990–1997. JAMA. 1998;280:1569–1575. [PubMed]
5. Ni H. Simile C. Hardy AM. Utilization of complementary and alternative medicine by United States adults. Med Care. 2002;40:353–358. [PubMed]
6. Mackenzie ER. Taylor L. Bloom BS, et al. Ethnic minority use of complementary and alternative medicine (CAM): A national probability survey of cam utilizers. Altern Ther. 2003;9:50–56. [PubMed]
7. Egede LE. Zheng D. Ye X. Silverstein MD. The prevalence and pattern of complementary and alternative medicine use in individuals with diabetes. Diabetes Care. 2002;25:324–329. [PubMed]
8. Lee MM. Chang JS. Jacobs B. Wrensch MR. Complementary and alternative medicine use among men with prostate cancer in 4 ethnic populations. Am J Public Health. 2002;92:1606–1607. [PubMed]
9. Bair YA. Gold EB. Greendale GA, et al. Ethnic differences in use of complementary and alternative medicine at midlife: Longitudinal results from SWAN participants. Am J Public Health. 2002;92:1832–1840. [PubMed]
10. Arcury TA. Quandt SA. Bell RA. Vitolins MZ. Complementary and alternative medicine use among rural older adults. Complement Health Pract Rev. 2002;7:167–186.
11. Brown CM. Segal R. The effects of health and treatment perceptions on the use of prescribed medication and home remedies among African American and white American hypertensives. Soc Sci Med. 1996;43:903–917. [PubMed]
12. Lee MM. Lin SS. Wrensch MR, et al. Alternative therapies used by women with breast cancer in four ethnic populations. J Natl Cancer Inst. 2000;92:42–47. [PubMed]
13. Cushman LF. Wade C. Factor-Litvak P, et al. Use of complementary and alternative medicine among African-American and Hispanic women in New York City: A pilot study. JAMWA. 1999;54:193–195. [PubMed]
14. Boyd EL. Taylor SD. Shimp LA. Semler CR. An assessment of home remedy use by African Americans. J Natl Med Assoc. 2000;92:341–353. [PMC free article] [PubMed]
15. Ellison C. Taylor RJ. Turning to prayer: Social and situational antecedents of religious coping among African Americans. Rev Religious Res. 1996;38:111–131.
16. Brown CM. Exploring the role of subjective religiosity in hypertension management among African Americans. J Health Care Poor Underserved. 2000;11:19–32. [PubMed]
17. Barnett MC. Cotroneo M. Purnell J, et al. Use of CAM in local African-American communities: Community-partnered research. J Natl Med Assoc. 2003;95:943–950. [PMC free article] [PubMed]
18. Cuellar N. Aycock T. Cahill B. Ford J. Complementary and alternative medicine (CAM) use by African American (AA) and Caucasian American (CA) older adults in a rural setting: A descriptive, comparative study. BMC Complementary and Alternative Med. 2003;3:8. [PMC free article] [PubMed]
19. Sternberg SA. Chandran A. Sikka M. Alternative therapy use by elderly African Americans attending a community clinic. JAGS. 2003;51:1768–1772. [PubMed]
20. Van Sickle D. Morgan F. Wright AL. Qualitative study of the use of traditional healing by asthmatic Navajo families. Am Indian Alsk Native Ment/Health Res. 2003;11:1–18. [PubMed]
21. Andersen R. Newman J. Societal and individual determinants of medical care utilization in the United States. Milbank Mem Fund Q. 1973;51:95–124. [PubMed]
22. Wolinsky FD. Coe RM. Miller DK, et al. Health services utilization among the noninstitutionalized elderly. J Health Soc Behav. 1983;24:325–337. [PubMed]
23. Mutran E. Ferraro K. Medical need and use of services among older men and women. J Gerontol. 1988;43:162–171. [PubMed]
24. Wolinsky F. Stump T. Johnson R. Hospital utilization profiles among older adults over time: Consistency and volume among survivors and descendents. J Gerontol B Psychol Sci Soc Sci. 1995;50B:S88–S100. [PubMed]
25. de Boer A. Wijker W. de Haes H. Predictors of health care utilization in the chronically ill: A review of the literature. Health Policy. 1997;42:101–115. [PubMed]
26. Nichol M. Margolies J. Gill M. Factors associated with antihypertensive prescribing. Ann Pharmacother. 1997;31:154–159. [PubMed]
27. Smith S. Kirking D. Access and use of medications in HIV disease. Health Serv Res. 1999;34(1 Pt 1):123–144. [PMC free article] [PubMed]
28. Smith S. Boyd E. Kirking D. Nonprescription and alternative medicine use by individuals with HIV disease. Ann Pharmacother. 1999;33:294–300. [PubMed]
29. Aparasu R. Mort J. Brandt H. Psychotropic prescription use by community-dwelling elderly in the United States. JAGS. 2003;51:671–677. [PubMed]
30. Wister AV. Chittenden M. McCoy B, et al. Using alternative therapies to manage chronic illness among older adults: An examination of the health context, predisposing and enabling processes. Can J Aging. 2002;21:47–62.
31. Shumay DM. Maskarinec G. Gotay CC, et al. Determinants of the degree of complementary and alternative medicine use among patients with cancer. J Altern Complement Med. 2002;8:661–671. [PubMed]
32. Bennett G. Brown CM. Use of herbal remedies by patients in a health maintenance organization. J Am Pharmacists Assoc. 2000;40:353–358. [PubMed]
33. Sleath B. Rubin RH. Campbell W, et al. Ethnicity and physician-older patient communication about alternative therapies. J Altern Complement Med. 2001;7:329–335. [PubMed]
34. McCullagh P. Nelder JA. Generalized Linear Models. 2nd. New York: Chapman and Hall; 1989.
35. Krause N. Chatters L. Exploring race differences in a multidimensional battery of prayer measures among older adults. Sociol Religion. 2005;66:23–43.
36. Taylor R. Chatters L. Jackson J. Religious and spiritual involvement among older African Americans, Caribbean Blacks, and non-Hispanic Whites: Findings from the National Survey of American Life. J Gerontol B Psychol Sci Soc Sci. 2007;62:1079–5014. [PubMed]

Articles from Journal of Alternative and Complementary Medicine are provided here courtesy of Mary Ann Liebert, Inc.