Association of Health Literacy with Complementary and Alternative Medicine Use: A Cross-Sectional Study in Adult Primary Care Patients

351 patients at an adult primary care clinic of an academic medical center in the southeastern United States were recruited for the study. The clinic assesses adults aged eighteen and older with the patient volume remaining the same across most months. Study patients were recruited over a 10-week period in the summer months. The time frame for recruitment was chosen in line with a research course for undergraduate students undertaking research projects. Our sample size estimation was based on determining a difference between health literacy status (adequate vs. inadequate) and CAM use (type 1 error rate of 0.05). Consenting patients completed validated surveys to assess health literacy, CAM use, and socio-demographic information. Approval for the study was obtained from our Institutional Review Board.

Age was assessed as a categorical variable: 16-34 years, 35-49 years, 50-64 years, and 65 or greater years old. Race/ethnicity was categorized as white, black and Hispanic or other racial origin. Marital status was categorized as currently married and not married. Four levels of education were created: less than high school graduate, equal to high school graduate/no college, less than college graduate, and greater than college graduate. Employment status was categorized as currently employed and unemployed. Personal income was categorized as ≤ $10,000, ≤ $25,000, ≤ $50,000, < $75,000 and ≥ $75,000. Four categories of insurance status were created: private, Medicaid, Medicare, and uninsured. Health status was based on the individual's current opinion of his or her health in comparison to 12 months previously (better, same, or worse than 12 months ago). For this analysis, perceived health status was categorized as better, same and worse.

We assessed health literacy by using the Revised Rapid Estimate of Adult Literacy in Medicine (REALM-R), an eight-item instrument designed to rapidly screen patients for potential health literacy problems. The REALM-R has been previously correlated with the Wide Range Achievement Test Revised (WRAT-R) (0.64) and a score of 6 on the REALM-R identified 26 of 30 persons scoring more than 1 standard deviation below the mean on the WRAT-R, corresponding to a sixth grade reading level" [38]. The REALM-R also has a shorter response burden than the WRAT-R and demonstrated a Cronbach's α of 0.91 [38]. The REALM-R is scored on a scale of 0 to 8 and asks patients to read a series of eight medical words, and a correct response is given for each correct pronunciation. These words include: osteporosis, allergic, jaundice, anemia, fatigue, directed, colitis and contipated. Scores of 6 or less correspond to a grade 6 reading level and identify patients at risk for poor health literacy [38]. For this analysis, health literacy was categorized as adequate health literacy (REALM-R > 6) and inadequate literacy (REALM-R ≤ 6).

We assessed CAM use using the International Questionnaire to Measure Use of Complementary and Alternative Medicine (I-CAM-Q) developed and tested by the National Research Center in Complementary and Alternative Medicine (NAFKAM) in Norway [39]. The I-CAM-Q asked patients if they had used herbs/herbal medicine, vitamins/minerals, and homeopathic remedies in the last 12 months. It also asked patients if they had used any of the following self-help practices in the last 12 months: meditation; yoga; qigong; tai chi; relaxation techniques; visualization, attending a traditional healing ceremony; or prayer for own health. For each type of CAM use, respondents were asked to indicate if they currently use it, main reason for last use, and to evaluate how helpful they found it [39]. For this analysis, CAM use was defined as using any of the above modalities or self-help practices.

We performed three sets of analysis. First, we ran a confirmatory factor analysis to determine the factor structure of the REALM-R and calculated Cronbach's α for REALM-R to establish its internal consistency in our study population. Second, we compared sample demographics by health literacy status (adequate vs. inadequate), and compared overall CAM use and individual CAM categories by health literacy status using chi-squared statistics. Third, we tested and found a race by literacy interaction in our analysis. We subsequently ran sequential logistic regression models stratified by race to test the association between health literacy and overall CAM use (Model 1), health literacy and overall CAM use controlling for education (Model 2), and health literacy and overall CAM use controlling for both education and other demographic characteristics (Model 3). We only report significant findings for both whites and African Americans separately because of the interaction. For each regression model, overall CAM use (meditation, yoga, tai chi, relaxation, visualization, traditional healing, prayer, herbal medicine, vitamins, and homeopathy) was the dependent variable, and health literacy was the primary independent variable with education and other demographic variables (age, sex, marital status, household income, and health status) sequentially added in the model as covariates. All analyses were performed with STATA version 10 (Stata Corp. LP, College Station, TX), and a two-tailed α of 0.05 was used to assess for significance. Variables were included in the model based on clinical relevance.

The principal component analysis yielded an 8-item Realm-R with one factor having a measured eigenvalue of 5.02 that accounting for 63% of the variance. The eigenvalues provide the variance explained by each factor. Factor loadings ranged from 0.77 to 0.82. All individual items correlated at greater than 0.75 with this factor [40].

An item analysis for the 8-item Realm-R revealed a Cronbach's α of 0.91. The item-test correlation of the 8-item Realm-R ranged from 0.79 to 0.84 and the item analysis showed that alpha would not be meaningfully improved by dropping any one item from the scale.

A total of 351 men and women were invited and recruited. 347 (98.8%) participants completed the surveys. 142 respondents (41%) were African American. Table 1 shows the demographics of the study by health literacy status. In the study, 75% of the patients had adequate health literacy and 25% had inadequate health literacy. Participants with adequate literacy were more likely to be over 50 years old(50+; 55.3%), white (68.8%), female (69%), married (55%), college graduate or higher (51.2%), employed (61.6%), have a household income > $75,000 (31.2%), have private insurance (67.1%), and reported their health status as better or same as last year (84.2%).

In our sample, there were significant differences in health literacy level by demographic characteristics. Adequate health literacy was higher among white and Hispanic/other race compared to black race (93% white, 82% Hispanic/other versus 49% black; p < 0.001), and also higher among married (84% married versus 67% unmarried; p < 0.001), employed (81% employed versus 67% unemployed; p = 0.003), and the ones who had health insurance (85% privately insured versus 46% noninsured; p < 0.001). Adequate health literacy was higher among those with higher educational attainment (95% > college graduate versus 51% < high school graduate; p < 0.001), and higher household income (98% > $75,000 versus 52% < $10,000; p < 0.001), Gender and perceived health status were not statistically different by health literacy level.

Health literacy status (adequate vs. inadequate) varied by category of CAM use. Table 2 displays health literacy status and the different types of CAM use. In our overall sample, 80% of the participants used CAM. Categorized by health literacy levels, 82% of those with adequate health literacy used CAM, while 74% of those with inadequate health literacy used CAM and this difference did not attain statistical significance. By CAM use category: 17% used meditation (14% adequate vs. 24% inadequate; p = 0.03) and 56% used vitamins (61% adequate vs. 42% inadequate; p = 0.003) which were both statistically significant.

Fifty-two percent used prayer (49% adequate vs. 59% inadequate), 18% used relaxation (18% adequate vs. 19% inadequate), 14% used herbal medications (13% adequate vs. 17% inadequate), 6% of the sample used yoga (7% adequate vs. 4% inadequate), 5% used visualization (7% adequate vs. 2% inadequate), 4% used homeopathy (4% adequate vs. 6% inadequate) < 1% used tai chi (0.4% adequate vs. 2% inadequate), < 1% attended traditional healing (0.4% adequate vs. 2% inadequate), and which were not statistically significant. No subjects used qigong as a CAM modality and subsequently this modality was excluded from our analysis.

Table 3 shows sequentially built logistic models of CAM use. Our analysis did show a significant race literacy interaction and subsequent testing of our separate models by racial category (Table 3) revealed that adequate health literacy among whites was associated with increased CAM use (Model 1, OR 7.68; 95% CI 2.35-25.1, p = 0.001). Adequate health literacy among whites remained significantly associated with increased CAM use after controlling for education (Model 2, OR 7.70; 95% CI 2.14-27.7, p = 0.002) and other demographics (Model 3, OR 9.42; 95% CI 1.66-53.5; p = 0.01). Adequate health literacy among African Americans was not significantly associated with CAM use (Model 1, OR 1.15; 95% CI 0.50-2.64; p = 0.74 and remained not significantly associated with CAM use after controlling for education (Model 2, OR 0.86; 95% CI 0.33-2.23; p = 0.76) and socio-demographics (Model 3, OR 0.97 95% CI 0.27-3.48; p = 0.96).