- Split View
-
Views
-
Cite
Cite
Nina R. Sperber, Hayden B. Bosworth, Cynthia J. Coffman, Jennifer H. Lindquist, Eugene Z. Oddone, Morris Weinberger, Kelli D. Allen, Differences in osteoarthritis self-management support intervention outcomes according to race and health literacy, Health Education Research, Volume 28, Issue 3, June 2013, Pages 502–511, https://doi.org/10.1093/her/cyt043
- Share Icon Share
Abstract
We explored whether the effects of a telephone-based osteoarthritis (OA) self-management support intervention differed by race and health literacy. Participants included 515 veterans with hip and/or knee OA. Linear mixed models assessed differential effects of the intervention compared with health education (HE) and usual care (UC) on pain (Arthritis Impact Measurement Scales-2 [AIMS2] and Visual Analogue Scale), function (AIMS2 mobility and walking/bending), affect (AIMS2) and arthritis self-efficacy by: (i) race (white/non-white), (ii) health literacy (high/low) and (iii) race by health literacy. AIMS2 mobility improved more among non-whites than whites in the intervention compared with HE and UC (P = 0.02 and 0.008). AIMS2 pain improved more among participants with low than high literacy in the intervention compared with HE (P = 0.05). However, we found a differential effect of the intervention on AIMS2 pain compared with UC according to the combination of race and health literacy (P = 0.05); non-whites with low literacy in the intervention had the greatest improvement in pain. This telephone-based OA intervention may be particularly beneficial for patients with OA who are racial/ethnic minorities and have low health literacy. These results warrant further research designed specifically to assess whether this type of intervention can reduce OA disparities.
Introduction
Osteoarthritis (OA) is the most common cause of disability in the United States [1]; individuals from racial/ethnic minority groups have higher OA incidence and more severe symptoms [2–7]. Although not well studied in OA, disparities in other chronic disease outcomes and functional health status are also strongly associated with low health literacy, which is more prevalent among non-whites [8, 9]. Specifically, 24% of blacks and 41% of Hispanics have below basic health literacy levels compared with 9% of whites [10, 11]. The US Department of Health and Human Services has made it a priority to eliminate health disparities, and understanding how health literacy and race relate to intervention self-management support outcomes for people with OA could yield new information about ways to reduce OA disparities [12].
Self-management support programs have been associated with improvement in pain, function and other important outcomes among patients with OA; however, there is little research comparing effectiveness of these programs according to participant demographic characteristics, and the evidence base for success among racial and ethnic minorities and those with low literacy is particularly limited [13–15]. In previous studies of self-management support interventions for other chronic diseases, black participants have shown greater improvements than whites [16, 17]. In addition, there is emerging evidence in other chronic disease areas to support the effectiveness of self-management support for patients with low literacy [18–23].
We wanted to explore whether a 12-month telephone-based self-management support intervention for veterans with hip and/or knee OA yielded differences in outcomes according to participants’ race and health literacy. Specifically, we conducted a secondary analysis to examine whether the effects of the OA intervention on pain, function and psychosocial variables differed by race, health literacy or their interaction. This knowledge is important for determining whether dissemination of the program among racial and ethnic minority individuals, as well as those with lower literacy levels, could help to improve outcomes in these groups and reduce disparities in disease burden.
Methods
This study was reviewed and approved by the Institutional Review Board of the Department of Veterans Affairs Medical Center (VAMC), which reviews all issues related to human subjects protections and ethics (Clinical trials registration number: NCT00288912); informed consent was obtained from all participants. This is an exploratory secondary analysis of data from a randomized controlled trial of a 12-month telephone-based OA self-management support intervention. The trial resulted in moderate improvements in pain, the primary trial outcome, compared with a health education (HE) control group [21]. A detailed description of the trial has been published previously [24, 25].
Participants and procedure
Participants were enrolled in this clinical trial between 2006 and 2008 if they met the following eligibility criteria: enrolled in primary care at the Durham VAMC, had a physician diagnosis of hip and/or knee OA and had persistent, current self-reported joint symptoms [26]. Exclusion criteria were having other rheumatological conditions, psychoses, dementia or serious health conditions that would likely prevent participation in the study; being on a waiting list for arthroplasty; participating in another OA-related or lifestyle intervention study.
Participants were stratified by race and randomly assigned to the OA intervention (n = 172), HE (n = 172) or usual care (UC) arm (n = 171), following consent and baseline assessments. Random assignment was computer generated, and the study coordinator contacted participants by telephone to inform them of their group assignment, so that research assistants conducting the assessments would be blind to randomization group. The same health educator (a Masters-trained counselor with previous experience in delivering health-related interventions) administered both the OA self-management support and HE interventions for consistency. The health educator was not involved in outcome assessment.
The OA intervention arm received written and audio versions of self-management educational materials covering 10 topics (the basics of OA and self management, exercise, healthy eating and weight management, medications, joint injections and surgery, talking with your doctor, joint care, complementary and alternate therapies, stress management and sleep) and monthly phone calls from the health educator to review key points, develop weekly self-management goals and action plans and solve problems. The HE (attention control) arm received materials regarding common health problems, such as hypertension and high cholesterol, and related screening recommendations as well as monthly phone calls from the health educator to review these topics. We attempted to make both interventions accessible to those with low literacy by creating relatively simple educational materials, providing audio versions of those materials and having the health educator review key points of the materials verbally via phone. For those with lower than an eighth-grade reading level as indicated at baseline by the Rapid Estimate of Adult Literacy in Medicine (REALM) score, the health educator encouraged participants to use audio versions of the educational materials and to ask a family member or friend to assist with reading and/or writing tasks as needed. The UC arm, as well as the other two study arms, received usual treatments offered at the VA for OA, as prescribed or recommended by health care providers. All participants were reimbursed $10 after baseline and follow-up assessments ($20 total), and those in the HE and UC groups received the written OA self-management materials after completing follow-up assessments.
Measures
Race and health literacy
Race was assessed from the medical record and then dichotomized into either a white or non-white category. Health literacy was measured with the REALM, a validated word recognition test and predictor of reading ability [27]. The REALM assesses an individual’s command of health-related vocabulary. Although it does not capture all aspects of literacy (e.g. reading ability), it has been shown to have predictive power and correlate highly with a widely used measure of reading fluency, the Test of Functional Health Literacy in Adults [28]. With this measure, participants were asked to read 66 commonly used medical lay terms and received 1 point for every word read correctly. Using the total scores, we categorized patients as either having low REALM (≤60), which indicates being unable to read most patient education materials or needing materials adjusted to meet their needs, or high REALM (>60), which indicates being able to read standard patient education materials [29].
Intervention outcomes
Arthritis outcomes examined in these analyses were the AIMS2 pain, mobility, walking and bending, and affect subscales [26], the Pain Visual Analogue Scale (VAS) [30] and the Arthritis Self-efficacy Scale [28], collected at baseline and the end of the trial (12 months). The AIMS2 subscale scores are measured on a 5-point Likert scale (‘all days’ to ‘no days’) and can range from 0 to 10, with higher scores indicating worse outcomes. AIMS2 pain consists of five items assessing typical pain, pain severity and pain during specific times of the day. AIMS2 mobility and walking and bending subscales are domains from the AIMS2 physical function scale that relate to lower extremity arthritis. These include a total of 10 items that ask about one’s ability to get around outside of the home. The AIMS2 affect subscale includes 10 items that encompass mood and tension. The Pain VAS consists of a 10-cm horizontal line on which individuals draw a vertical line indicating the severity of their pain during the past week, with anchors of ‘no pain’ and ‘pain as bad as can be’. The Arthritis Self-Efficacy Scale measures how certain patients are that they can perform eight specific activities or tasks related to arthritis. Items are scored on a Likert Scale (1 = very uncertain to 10 = very certain), and total scores range from 1 to 10, with higher scores indicating better self-efficacy.
Data analysis
We used linear mixed models for this exploratory analysis to assess whether the effect of the OA intervention on change in pain, function and psychosocial variables between baseline and 12 months differed by race (white/non-white) and health literacy (high/low REALM) independently, as well as for the combinations of race and literacy (i.e. interaction effect), compared with the UC and HE arms. A benefit of using a mixed-model framework for longitudinal analysis is that all available data are used. The estimation procedure used in this framework yields unbiased estimates of parameters when missing outcomes are assumed to be ignorable (related to either observed covariates or response variables but not to unobserved variables) [31]. All models were fit with a common baseline across treatment groups by race and/or literacy level, as is the standard practice in a randomized controlled trial: for example, baseline means for whites in the OA arm were assumed to be equivalent to means for whites in the HE and UC arms but not equivalent to means for non-whites [32–34]. For the model addressing the intervention effects by race for each outcome, the primary predictors included a race indicator variable, a 12-month follow-up indicator variable, treatment group by 12-month follow-up variables (OA intervention was the reference group), race by treatment group variables and then the three-way race by treatment group by 12-month follow-up variables. The model addressing the intervention effect by literacy was similarly parameterized for each outcome, substituting the race variable with the literacy variable; however, we also included the main effect of race in the literacy model as this was a stratification variable for randomization in our original study. For outcomes where we saw differential effects of the OA intervention by race and/or literacy, we examined whether there were also differences in outcomes according to interaction of race and literacy (i.e. four-way interaction).
Results
Of our analysis sample of 515 patients, 461 (90%) completed the study. Completion rates were similar for white and non-white participants (88 and 91%). At 12-month follow-up, we had more missing data for the Pain VAS (n = 92 versus n = 54 for other outcomes), because we completed some 12-month assessments over the phone due to the logistics of travelling to the Durham VAMC for a follow-up visit. The mean age of participants (n = 515) was 60 years, and most (93%) were male. Slightly more than half (54%) of participants were white, and most non-white participants were black (43% of the total sample). About 67% of participants had at least some college education and 70% had high literacy (>60 on the REALM); 37% of non-white participants and 24% of white participants had a low literacy (≤60 on REALM). Additional characteristics are provided in Table I.
Variable . | Mean SD or % . |
---|---|
n | 515 |
Age (mean ± SD) | 60 ± 10 |
Male | 93% |
Race | |
White | 54% |
Black | 43% |
At least some college | 67% |
High (>60) REALM | 70%a |
Self-reported perceived inadequate incomeb | 73%a |
Self-reported years with OA symptoms (mean ± SD) | 16.1 ± 12.2 |
Fair or poor health | 32% |
Body mass index, mean (SD) | 31.8 (6.6) |
Variable . | Mean SD or % . |
---|---|
n | 515 |
Age (mean ± SD) | 60 ± 10 |
Male | 93% |
Race | |
White | 54% |
Black | 43% |
At least some college | 67% |
High (>60) REALM | 70%a |
Self-reported perceived inadequate incomeb | 73%a |
Self-reported years with OA symptoms (mean ± SD) | 16.1 ± 12.2 |
Fair or poor health | 32% |
Body mass index, mean (SD) | 31.8 (6.6) |
aSix missing for REALM; seven missing for self-reported perceived income. b‘You have money to pay the bills, but only because you have to cut back on things’ or ‘You are having difficulty paying the bills, no matter what you do’.
Variable . | Mean SD or % . |
---|---|
n | 515 |
Age (mean ± SD) | 60 ± 10 |
Male | 93% |
Race | |
White | 54% |
Black | 43% |
At least some college | 67% |
High (>60) REALM | 70%a |
Self-reported perceived inadequate incomeb | 73%a |
Self-reported years with OA symptoms (mean ± SD) | 16.1 ± 12.2 |
Fair or poor health | 32% |
Body mass index, mean (SD) | 31.8 (6.6) |
Variable . | Mean SD or % . |
---|---|
n | 515 |
Age (mean ± SD) | 60 ± 10 |
Male | 93% |
Race | |
White | 54% |
Black | 43% |
At least some college | 67% |
High (>60) REALM | 70%a |
Self-reported perceived inadequate incomeb | 73%a |
Self-reported years with OA symptoms (mean ± SD) | 16.1 ± 12.2 |
Fair or poor health | 32% |
Body mass index, mean (SD) | 31.8 (6.6) |
aSix missing for REALM; seven missing for self-reported perceived income. b‘You have money to pay the bills, but only because you have to cut back on things’ or ‘You are having difficulty paying the bills, no matter what you do’.
Differential effects by race
Estimated mean scores for most of the measures improved for both whites and non-whites, across study arms (Table II). The only outcome for which we found a differential treatment effect by race was the AIMS2 mobility subscale. Non-whites in the OA intervention had greater improvement in AIMS2 mobility between baseline and 12 months than whites, and this difference between race categories was greater and significant when compared with both the HE arm (P = 0.02) and UC arm (P = 0.01).
. | Baseline . | Change in outcome(12 months − baseline) . | . | |||
---|---|---|---|---|---|---|
Outcome/ randomization arm . | White . | Non-white . | Whitea . | Non-whitea . | Differenceb (white − non-white) . | Difference in change between white and non-white participants for HE or UC versus OA . |
AIMS2 mobilityc | ||||||
OA | 1.1 | 2.3 | 0.1 | −0.7 | 0.8 | — |
HE | 1.1 | 2.3 | −0.0 | −0.2 | 0.2 | −0.6 (−1.2 to −0.1); P = 0.02 |
UC | 1.1 | 2.3 | −0.2 | −0.3 | 0.0 | −0.8 (−1.3 to −0.2); P = 0.01 |
AIMS2 pain | ||||||
OA | 5.6 | 6.2 | −0.6 | −0.9 | 0.3 | — |
HE | 5.6 | 6.2 | 0.0 | −0.4 | 0.4 | 0.1 (−0.8 to 0.9); P = 0.84 |
UC | 5.6 | 6.2 | −0.3 | −0.5 | 0.2 | −0.1 (−1.0 to 0.7); P = 0.80 |
AIMS2 walking and bending | ||||||
OA | 6.5 | 6.3 | −0.4 | −0.5 | 0.1 | — |
HE | 6.5 | 6.3 | 0.2 | −0.1 | 0.3 | 0.2 (−0.9 to 1.2); P = 0.77 |
UC | 6.5 | 6.3 | −0.4 | 0.1 | −0.5 | −0.6 (−1.6 to 0.4); P = 0.23 |
AIMS2 affect | ||||||
OA | 6.4 | 6.5 | −0.5 | −0.4 | −0.1 | — |
HE | 6.4 | 6.5 | 0.1 | 0.1 | −0.1 | 0.0 (−0.9 to 1.0); P = 0.93 |
UC | 6.4 | 6.5 | −0.2 | −0.2 | −0.0 | 0.1 (−0.9 to 1.0); P = 0.90 |
Pain VASd | ||||||
OA | 5.2 | 6.6 | −1.0 | −1.1 | 0.1 | — |
HE | 5.2 | 6.6 | 0.4 | −0.4 | 0.8 | 0.8 (−0.3 to 1.8); P = 0.15 |
UC | 5.2 | 6.6 | 0.2 | −0.2 | 0.4 | 0.4 (−0.7 to 1.4); P = 0.51 |
Arthritis self-efficacye | ||||||
OA | 6.0 | 5.5 | 0.3 | 0.5 | −0.2 | — |
HE | 6.0 | 5.5 | 0.1 | −0.1 | 0.3 | 0.4 (−0.4 to 1.2); P = 0.29 |
UC | 6.0 | 5.5 | 0.2 | −0.1 | 0.3 | 0.5 (−0.3 to 1.2); P = 0.21 |
. | Baseline . | Change in outcome(12 months − baseline) . | . | |||
---|---|---|---|---|---|---|
Outcome/ randomization arm . | White . | Non-white . | Whitea . | Non-whitea . | Differenceb (white − non-white) . | Difference in change between white and non-white participants for HE or UC versus OA . |
AIMS2 mobilityc | ||||||
OA | 1.1 | 2.3 | 0.1 | −0.7 | 0.8 | — |
HE | 1.1 | 2.3 | −0.0 | −0.2 | 0.2 | −0.6 (−1.2 to −0.1); P = 0.02 |
UC | 1.1 | 2.3 | −0.2 | −0.3 | 0.0 | −0.8 (−1.3 to −0.2); P = 0.01 |
AIMS2 pain | ||||||
OA | 5.6 | 6.2 | −0.6 | −0.9 | 0.3 | — |
HE | 5.6 | 6.2 | 0.0 | −0.4 | 0.4 | 0.1 (−0.8 to 0.9); P = 0.84 |
UC | 5.6 | 6.2 | −0.3 | −0.5 | 0.2 | −0.1 (−1.0 to 0.7); P = 0.80 |
AIMS2 walking and bending | ||||||
OA | 6.5 | 6.3 | −0.4 | −0.5 | 0.1 | — |
HE | 6.5 | 6.3 | 0.2 | −0.1 | 0.3 | 0.2 (−0.9 to 1.2); P = 0.77 |
UC | 6.5 | 6.3 | −0.4 | 0.1 | −0.5 | −0.6 (−1.6 to 0.4); P = 0.23 |
AIMS2 affect | ||||||
OA | 6.4 | 6.5 | −0.5 | −0.4 | −0.1 | — |
HE | 6.4 | 6.5 | 0.1 | 0.1 | −0.1 | 0.0 (−0.9 to 1.0); P = 0.93 |
UC | 6.4 | 6.5 | −0.2 | −0.2 | −0.0 | 0.1 (−0.9 to 1.0); P = 0.90 |
Pain VASd | ||||||
OA | 5.2 | 6.6 | −1.0 | −1.1 | 0.1 | — |
HE | 5.2 | 6.6 | 0.4 | −0.4 | 0.8 | 0.8 (−0.3 to 1.8); P = 0.15 |
UC | 5.2 | 6.6 | 0.2 | −0.2 | 0.4 | 0.4 (−0.7 to 1.4); P = 0.51 |
Arthritis self-efficacye | ||||||
OA | 6.0 | 5.5 | 0.3 | 0.5 | −0.2 | — |
HE | 6.0 | 5.5 | 0.1 | −0.1 | 0.3 | 0.4 (−0.4 to 1.2); P = 0.29 |
UC | 6.0 | 5.5 | 0.2 | −0.1 | 0.3 | 0.5 (−0.3 to 1.2); P = 0.21 |
aFor AIMS2 and Pain VAS, changes in the negative direction indicate improvement (i.e. lower scores) at 12 months. bFor AIMS2 and Pain VAS, positive scores indicate greater improvement for non-whites. cThe potential range of AIMS2 measures is 0–10, with lower scores indicating better health outcomes. dThe potential range of Pain VAS is 0–10, with lower scores indicating less pain. eThe potential range of arthritis self-efficacy is 1–10, with higher scores indicating better self-efficacy.
. | Baseline . | Change in outcome(12 months − baseline) . | . | |||
---|---|---|---|---|---|---|
Outcome/ randomization arm . | White . | Non-white . | Whitea . | Non-whitea . | Differenceb (white − non-white) . | Difference in change between white and non-white participants for HE or UC versus OA . |
AIMS2 mobilityc | ||||||
OA | 1.1 | 2.3 | 0.1 | −0.7 | 0.8 | — |
HE | 1.1 | 2.3 | −0.0 | −0.2 | 0.2 | −0.6 (−1.2 to −0.1); P = 0.02 |
UC | 1.1 | 2.3 | −0.2 | −0.3 | 0.0 | −0.8 (−1.3 to −0.2); P = 0.01 |
AIMS2 pain | ||||||
OA | 5.6 | 6.2 | −0.6 | −0.9 | 0.3 | — |
HE | 5.6 | 6.2 | 0.0 | −0.4 | 0.4 | 0.1 (−0.8 to 0.9); P = 0.84 |
UC | 5.6 | 6.2 | −0.3 | −0.5 | 0.2 | −0.1 (−1.0 to 0.7); P = 0.80 |
AIMS2 walking and bending | ||||||
OA | 6.5 | 6.3 | −0.4 | −0.5 | 0.1 | — |
HE | 6.5 | 6.3 | 0.2 | −0.1 | 0.3 | 0.2 (−0.9 to 1.2); P = 0.77 |
UC | 6.5 | 6.3 | −0.4 | 0.1 | −0.5 | −0.6 (−1.6 to 0.4); P = 0.23 |
AIMS2 affect | ||||||
OA | 6.4 | 6.5 | −0.5 | −0.4 | −0.1 | — |
HE | 6.4 | 6.5 | 0.1 | 0.1 | −0.1 | 0.0 (−0.9 to 1.0); P = 0.93 |
UC | 6.4 | 6.5 | −0.2 | −0.2 | −0.0 | 0.1 (−0.9 to 1.0); P = 0.90 |
Pain VASd | ||||||
OA | 5.2 | 6.6 | −1.0 | −1.1 | 0.1 | — |
HE | 5.2 | 6.6 | 0.4 | −0.4 | 0.8 | 0.8 (−0.3 to 1.8); P = 0.15 |
UC | 5.2 | 6.6 | 0.2 | −0.2 | 0.4 | 0.4 (−0.7 to 1.4); P = 0.51 |
Arthritis self-efficacye | ||||||
OA | 6.0 | 5.5 | 0.3 | 0.5 | −0.2 | — |
HE | 6.0 | 5.5 | 0.1 | −0.1 | 0.3 | 0.4 (−0.4 to 1.2); P = 0.29 |
UC | 6.0 | 5.5 | 0.2 | −0.1 | 0.3 | 0.5 (−0.3 to 1.2); P = 0.21 |
. | Baseline . | Change in outcome(12 months − baseline) . | . | |||
---|---|---|---|---|---|---|
Outcome/ randomization arm . | White . | Non-white . | Whitea . | Non-whitea . | Differenceb (white − non-white) . | Difference in change between white and non-white participants for HE or UC versus OA . |
AIMS2 mobilityc | ||||||
OA | 1.1 | 2.3 | 0.1 | −0.7 | 0.8 | — |
HE | 1.1 | 2.3 | −0.0 | −0.2 | 0.2 | −0.6 (−1.2 to −0.1); P = 0.02 |
UC | 1.1 | 2.3 | −0.2 | −0.3 | 0.0 | −0.8 (−1.3 to −0.2); P = 0.01 |
AIMS2 pain | ||||||
OA | 5.6 | 6.2 | −0.6 | −0.9 | 0.3 | — |
HE | 5.6 | 6.2 | 0.0 | −0.4 | 0.4 | 0.1 (−0.8 to 0.9); P = 0.84 |
UC | 5.6 | 6.2 | −0.3 | −0.5 | 0.2 | −0.1 (−1.0 to 0.7); P = 0.80 |
AIMS2 walking and bending | ||||||
OA | 6.5 | 6.3 | −0.4 | −0.5 | 0.1 | — |
HE | 6.5 | 6.3 | 0.2 | −0.1 | 0.3 | 0.2 (−0.9 to 1.2); P = 0.77 |
UC | 6.5 | 6.3 | −0.4 | 0.1 | −0.5 | −0.6 (−1.6 to 0.4); P = 0.23 |
AIMS2 affect | ||||||
OA | 6.4 | 6.5 | −0.5 | −0.4 | −0.1 | — |
HE | 6.4 | 6.5 | 0.1 | 0.1 | −0.1 | 0.0 (−0.9 to 1.0); P = 0.93 |
UC | 6.4 | 6.5 | −0.2 | −0.2 | −0.0 | 0.1 (−0.9 to 1.0); P = 0.90 |
Pain VASd | ||||||
OA | 5.2 | 6.6 | −1.0 | −1.1 | 0.1 | — |
HE | 5.2 | 6.6 | 0.4 | −0.4 | 0.8 | 0.8 (−0.3 to 1.8); P = 0.15 |
UC | 5.2 | 6.6 | 0.2 | −0.2 | 0.4 | 0.4 (−0.7 to 1.4); P = 0.51 |
Arthritis self-efficacye | ||||||
OA | 6.0 | 5.5 | 0.3 | 0.5 | −0.2 | — |
HE | 6.0 | 5.5 | 0.1 | −0.1 | 0.3 | 0.4 (−0.4 to 1.2); P = 0.29 |
UC | 6.0 | 5.5 | 0.2 | −0.1 | 0.3 | 0.5 (−0.3 to 1.2); P = 0.21 |
aFor AIMS2 and Pain VAS, changes in the negative direction indicate improvement (i.e. lower scores) at 12 months. bFor AIMS2 and Pain VAS, positive scores indicate greater improvement for non-whites. cThe potential range of AIMS2 measures is 0–10, with lower scores indicating better health outcomes. dThe potential range of Pain VAS is 0–10, with lower scores indicating less pain. eThe potential range of arthritis self-efficacy is 1–10, with higher scores indicating better self-efficacy.
Differential effects by REALM
Estimated mean scores for most of the measures improved for participants in both high and low REALM score categories, across study arms (Table III). The only outcome for which we found a differential treatment effect by REALM category was the AIMS2 pain subscale. We found that participants with low REALM scores in the OA intervention had greater improvement in AIMS2 pain between baseline and 12 months than those with high REALM scores when compared with the HE arm (P = 0.05).
. | Baseline . | Change in outcome (12 months – baseline) . | . | |||
---|---|---|---|---|---|---|
Outcome/ randomization arm . | High REALM . | Low REALM . | High REALMa . | Low REALMa . | Differenceb (high − low) . | Difference in change between high and low REALM categories for HE or UC versus OA . |
AIMS2 mobilityc | ||||||
OA | 1.5 | 1.8 | −0.2 | −0.4 | 0.2 | — |
HE | 1.5 | 1.8 | −0.1 | −0.2 | 0.1 | −0.1 (−0.7 to 0.5); P = 0.79 |
UC | 1.5 | 1.8 | −0.1 | −0.5 | 0.3 | 0.1 (−0.5 to 0.7); P = 0.68 |
AIMS2 pain | ||||||
OA | 5.9 | 6.0 | −0.5 | −1.1 | 0.5 | — |
HE | 5.9 | 6.0 | −0.3 | 0.1 | −0.4 | −0.9 (−1.9, −0.0); P = 0.05 |
UC | 5.9 | 6.0 | −0.3 | −0.6 | 0.4 | −0.2 (−1.1 to 0.7); P = 0.68 |
AIMS2 walking and bending | ||||||
OA | 6.5 | 6.3 | −0.4 | −0.5 | 0.1 | — |
HE | 6.5 | 6.3 | 0.2 | −0.1 | 0.3 | 0.2 (−0.9 to 1.2); P = 0.77 |
UC | 6.5 | 6.3 | −0.4 | 0.1 | −0.5 | −0.6 (−1.6 to 0.4); P = 0.23 |
AIMS2 affect | ||||||
OA | 3.5 | 3.6 | −0.1 | −0.3 | 0.2 | — |
HE | 3.5 | 3.6 | −0.3 | 0.1 | −0.4 | −0.7 (−1.4 to 0.0); P = 0.07 |
UC | 3.5 | 3.6 | −0.1 | −0.3 | 0.2 | −0.0 (−0.7 to 0.7); P = 0.92 |
Pain VASd | ||||||
OA | 5.7 | 6.0 | −0.9 | −1.3 | 0.4 | — |
HE | 5.7 | 6.0 | 0.0 | −0.1 | 0.1 | −0.3 (−1.4 to 0.8); P = 0.62 |
UC | 5.7 | 6.0 | −0.0 | 0.1 | −0.1 | −0.5 (−1.6 to 0.6); P = 0.37 |
Arthritis self-efficacye | ||||||
OA | 5.7 | 5.9 | 0.3 | 0.6 | −0.2 | — |
HE | 5.7 | 5.9 | 0.1 | −0.1 | 0.2 | 0.4 (−0.4 to 1.3); P = 0.31 |
UC | 5.7 | 5.9 | 0.1 | −0.0 | 0.1 | 0.4 (−0.4 to 1.2); P = 0.37 |
. | Baseline . | Change in outcome (12 months – baseline) . | . | |||
---|---|---|---|---|---|---|
Outcome/ randomization arm . | High REALM . | Low REALM . | High REALMa . | Low REALMa . | Differenceb (high − low) . | Difference in change between high and low REALM categories for HE or UC versus OA . |
AIMS2 mobilityc | ||||||
OA | 1.5 | 1.8 | −0.2 | −0.4 | 0.2 | — |
HE | 1.5 | 1.8 | −0.1 | −0.2 | 0.1 | −0.1 (−0.7 to 0.5); P = 0.79 |
UC | 1.5 | 1.8 | −0.1 | −0.5 | 0.3 | 0.1 (−0.5 to 0.7); P = 0.68 |
AIMS2 pain | ||||||
OA | 5.9 | 6.0 | −0.5 | −1.1 | 0.5 | — |
HE | 5.9 | 6.0 | −0.3 | 0.1 | −0.4 | −0.9 (−1.9, −0.0); P = 0.05 |
UC | 5.9 | 6.0 | −0.3 | −0.6 | 0.4 | −0.2 (−1.1 to 0.7); P = 0.68 |
AIMS2 walking and bending | ||||||
OA | 6.5 | 6.3 | −0.4 | −0.5 | 0.1 | — |
HE | 6.5 | 6.3 | 0.2 | −0.1 | 0.3 | 0.2 (−0.9 to 1.2); P = 0.77 |
UC | 6.5 | 6.3 | −0.4 | 0.1 | −0.5 | −0.6 (−1.6 to 0.4); P = 0.23 |
AIMS2 affect | ||||||
OA | 3.5 | 3.6 | −0.1 | −0.3 | 0.2 | — |
HE | 3.5 | 3.6 | −0.3 | 0.1 | −0.4 | −0.7 (−1.4 to 0.0); P = 0.07 |
UC | 3.5 | 3.6 | −0.1 | −0.3 | 0.2 | −0.0 (−0.7 to 0.7); P = 0.92 |
Pain VASd | ||||||
OA | 5.7 | 6.0 | −0.9 | −1.3 | 0.4 | — |
HE | 5.7 | 6.0 | 0.0 | −0.1 | 0.1 | −0.3 (−1.4 to 0.8); P = 0.62 |
UC | 5.7 | 6.0 | −0.0 | 0.1 | −0.1 | −0.5 (−1.6 to 0.6); P = 0.37 |
Arthritis self-efficacye | ||||||
OA | 5.7 | 5.9 | 0.3 | 0.6 | −0.2 | — |
HE | 5.7 | 5.9 | 0.1 | −0.1 | 0.2 | 0.4 (−0.4 to 1.3); P = 0.31 |
UC | 5.7 | 5.9 | 0.1 | −0.0 | 0.1 | 0.4 (−0.4 to 1.2); P = 0.37 |
aFor AIMS2 and Pain VAS, changes in the negative direction indicate improvement (i.e. lower scores) at 12 months. bFor AIMS2 and Pain VAS, positive scores indicate greater improvement for participants with low REALM. cThe potential range of AIMS2 measures is 0–10, with lower scores indicating better health outcomes. dThe potential range of Pain VAS is 0–10, with lower scores indicating less pain. eThe potential range of arthritis self-efficacy is 1–10, with higher scores indicating better self-efficacy.
. | Baseline . | Change in outcome (12 months – baseline) . | . | |||
---|---|---|---|---|---|---|
Outcome/ randomization arm . | High REALM . | Low REALM . | High REALMa . | Low REALMa . | Differenceb (high − low) . | Difference in change between high and low REALM categories for HE or UC versus OA . |
AIMS2 mobilityc | ||||||
OA | 1.5 | 1.8 | −0.2 | −0.4 | 0.2 | — |
HE | 1.5 | 1.8 | −0.1 | −0.2 | 0.1 | −0.1 (−0.7 to 0.5); P = 0.79 |
UC | 1.5 | 1.8 | −0.1 | −0.5 | 0.3 | 0.1 (−0.5 to 0.7); P = 0.68 |
AIMS2 pain | ||||||
OA | 5.9 | 6.0 | −0.5 | −1.1 | 0.5 | — |
HE | 5.9 | 6.0 | −0.3 | 0.1 | −0.4 | −0.9 (−1.9, −0.0); P = 0.05 |
UC | 5.9 | 6.0 | −0.3 | −0.6 | 0.4 | −0.2 (−1.1 to 0.7); P = 0.68 |
AIMS2 walking and bending | ||||||
OA | 6.5 | 6.3 | −0.4 | −0.5 | 0.1 | — |
HE | 6.5 | 6.3 | 0.2 | −0.1 | 0.3 | 0.2 (−0.9 to 1.2); P = 0.77 |
UC | 6.5 | 6.3 | −0.4 | 0.1 | −0.5 | −0.6 (−1.6 to 0.4); P = 0.23 |
AIMS2 affect | ||||||
OA | 3.5 | 3.6 | −0.1 | −0.3 | 0.2 | — |
HE | 3.5 | 3.6 | −0.3 | 0.1 | −0.4 | −0.7 (−1.4 to 0.0); P = 0.07 |
UC | 3.5 | 3.6 | −0.1 | −0.3 | 0.2 | −0.0 (−0.7 to 0.7); P = 0.92 |
Pain VASd | ||||||
OA | 5.7 | 6.0 | −0.9 | −1.3 | 0.4 | — |
HE | 5.7 | 6.0 | 0.0 | −0.1 | 0.1 | −0.3 (−1.4 to 0.8); P = 0.62 |
UC | 5.7 | 6.0 | −0.0 | 0.1 | −0.1 | −0.5 (−1.6 to 0.6); P = 0.37 |
Arthritis self-efficacye | ||||||
OA | 5.7 | 5.9 | 0.3 | 0.6 | −0.2 | — |
HE | 5.7 | 5.9 | 0.1 | −0.1 | 0.2 | 0.4 (−0.4 to 1.3); P = 0.31 |
UC | 5.7 | 5.9 | 0.1 | −0.0 | 0.1 | 0.4 (−0.4 to 1.2); P = 0.37 |
. | Baseline . | Change in outcome (12 months – baseline) . | . | |||
---|---|---|---|---|---|---|
Outcome/ randomization arm . | High REALM . | Low REALM . | High REALMa . | Low REALMa . | Differenceb (high − low) . | Difference in change between high and low REALM categories for HE or UC versus OA . |
AIMS2 mobilityc | ||||||
OA | 1.5 | 1.8 | −0.2 | −0.4 | 0.2 | — |
HE | 1.5 | 1.8 | −0.1 | −0.2 | 0.1 | −0.1 (−0.7 to 0.5); P = 0.79 |
UC | 1.5 | 1.8 | −0.1 | −0.5 | 0.3 | 0.1 (−0.5 to 0.7); P = 0.68 |
AIMS2 pain | ||||||
OA | 5.9 | 6.0 | −0.5 | −1.1 | 0.5 | — |
HE | 5.9 | 6.0 | −0.3 | 0.1 | −0.4 | −0.9 (−1.9, −0.0); P = 0.05 |
UC | 5.9 | 6.0 | −0.3 | −0.6 | 0.4 | −0.2 (−1.1 to 0.7); P = 0.68 |
AIMS2 walking and bending | ||||||
OA | 6.5 | 6.3 | −0.4 | −0.5 | 0.1 | — |
HE | 6.5 | 6.3 | 0.2 | −0.1 | 0.3 | 0.2 (−0.9 to 1.2); P = 0.77 |
UC | 6.5 | 6.3 | −0.4 | 0.1 | −0.5 | −0.6 (−1.6 to 0.4); P = 0.23 |
AIMS2 affect | ||||||
OA | 3.5 | 3.6 | −0.1 | −0.3 | 0.2 | — |
HE | 3.5 | 3.6 | −0.3 | 0.1 | −0.4 | −0.7 (−1.4 to 0.0); P = 0.07 |
UC | 3.5 | 3.6 | −0.1 | −0.3 | 0.2 | −0.0 (−0.7 to 0.7); P = 0.92 |
Pain VASd | ||||||
OA | 5.7 | 6.0 | −0.9 | −1.3 | 0.4 | — |
HE | 5.7 | 6.0 | 0.0 | −0.1 | 0.1 | −0.3 (−1.4 to 0.8); P = 0.62 |
UC | 5.7 | 6.0 | −0.0 | 0.1 | −0.1 | −0.5 (−1.6 to 0.6); P = 0.37 |
Arthritis self-efficacye | ||||||
OA | 5.7 | 5.9 | 0.3 | 0.6 | −0.2 | — |
HE | 5.7 | 5.9 | 0.1 | −0.1 | 0.2 | 0.4 (−0.4 to 1.3); P = 0.31 |
UC | 5.7 | 5.9 | 0.1 | −0.0 | 0.1 | 0.4 (−0.4 to 1.2); P = 0.37 |
aFor AIMS2 and Pain VAS, changes in the negative direction indicate improvement (i.e. lower scores) at 12 months. bFor AIMS2 and Pain VAS, positive scores indicate greater improvement for participants with low REALM. cThe potential range of AIMS2 measures is 0–10, with lower scores indicating better health outcomes. dThe potential range of Pain VAS is 0–10, with lower scores indicating less pain. eThe potential range of arthritis self-efficacy is 1–10, with higher scores indicating better self-efficacy.
Differential effects by race and REALM (interaction)
We found that there was a differential effect according to the combination of race and REALM score category on change in AIMS2 pain, for the OA intervention compared with the UC arm. The results are presented in Table IV, and the differential effect, or interaction, is illustrated in Fig. 1. Non-whites with low REALM score in the OA intervention had the greatest improvement in mean AIMS2 pain scores across all arms; in the OA intervention, there was a greater difference in AIMS2 pain change between high and low REALM categories for non-whites compared with whites. In contrast, in the UC arm, the health literacy effect by race was in the opposite direction of what was found in the OA intervention, and there was virtually no difference between REALM categories in the HE arm. We found no differences in treatment effect by the combination of race and REALM category for the AIMS2 mobility subscale.
. | Change in outcome (12 months – baseline) . | . | |||||
---|---|---|---|---|---|---|---|
. | White . | Non-white . | . | ||||
Outcome/randomization arm . | High REALMa . | Low REALMa . | Difference (high − low)b . | High REALMa . | Low REALMa . | Difference (high − low)b . | Difference in change between combinations of race and REALM for HE or UC versus OA . |
AIMS2 painc | |||||||
OA | −0.7 | −0.4 | −0.2 | −0.3 | −1.5 | 1.2 | — |
HE | −0.1 | 0.4 | −0.5 | −0.5 | −0.1 | −0.4 | 1.3 (−0.6 to 3.2); P = 0.18 |
UC | −0.1 | −0.7 | 0.6 | −0.5 | −0.6 | 0.1 | 1.8 (−0.0 to 3.7); P = 0.05 |
AIMS2 mobility | |||||||
OA | 0.2 | −0.1 | 0.3 | −0.7 | −0.6 | −0.1 | — |
HE | −0.0 | 0.1 | −0.1 | −0.1 | −0.4 | 0.3 | −0.8 (−2.1 to 0.4); P = 0.18 |
UC | −0.2 | −0.3 | 0.1 | −0.0 | −0.5 | 0.5 | −0.7 (−1.9 to 0.5); P = 0.24 |
. | Change in outcome (12 months – baseline) . | . | |||||
---|---|---|---|---|---|---|---|
. | White . | Non-white . | . | ||||
Outcome/randomization arm . | High REALMa . | Low REALMa . | Difference (high − low)b . | High REALMa . | Low REALMa . | Difference (high − low)b . | Difference in change between combinations of race and REALM for HE or UC versus OA . |
AIMS2 painc | |||||||
OA | −0.7 | −0.4 | −0.2 | −0.3 | −1.5 | 1.2 | — |
HE | −0.1 | 0.4 | −0.5 | −0.5 | −0.1 | −0.4 | 1.3 (−0.6 to 3.2); P = 0.18 |
UC | −0.1 | −0.7 | 0.6 | −0.5 | −0.6 | 0.1 | 1.8 (−0.0 to 3.7); P = 0.05 |
AIMS2 mobility | |||||||
OA | 0.2 | −0.1 | 0.3 | −0.7 | −0.6 | −0.1 | — |
HE | −0.0 | 0.1 | −0.1 | −0.1 | −0.4 | 0.3 | −0.8 (−2.1 to 0.4); P = 0.18 |
UC | −0.2 | −0.3 | 0.1 | −0.0 | −0.5 | 0.5 | −0.7 (−1.9 to 0.5); P = 0.24 |
aChanges in the negative direction indicate improvement (i.e. lower scores) at 12 months. bPositive scores indicate greater improvement for participants with low REALM. cThe potential range of AIMS2 measures is 0–10, with lower scores indicating better health outcomes.
. | Change in outcome (12 months – baseline) . | . | |||||
---|---|---|---|---|---|---|---|
. | White . | Non-white . | . | ||||
Outcome/randomization arm . | High REALMa . | Low REALMa . | Difference (high − low)b . | High REALMa . | Low REALMa . | Difference (high − low)b . | Difference in change between combinations of race and REALM for HE or UC versus OA . |
AIMS2 painc | |||||||
OA | −0.7 | −0.4 | −0.2 | −0.3 | −1.5 | 1.2 | — |
HE | −0.1 | 0.4 | −0.5 | −0.5 | −0.1 | −0.4 | 1.3 (−0.6 to 3.2); P = 0.18 |
UC | −0.1 | −0.7 | 0.6 | −0.5 | −0.6 | 0.1 | 1.8 (−0.0 to 3.7); P = 0.05 |
AIMS2 mobility | |||||||
OA | 0.2 | −0.1 | 0.3 | −0.7 | −0.6 | −0.1 | — |
HE | −0.0 | 0.1 | −0.1 | −0.1 | −0.4 | 0.3 | −0.8 (−2.1 to 0.4); P = 0.18 |
UC | −0.2 | −0.3 | 0.1 | −0.0 | −0.5 | 0.5 | −0.7 (−1.9 to 0.5); P = 0.24 |
. | Change in outcome (12 months – baseline) . | . | |||||
---|---|---|---|---|---|---|---|
. | White . | Non-white . | . | ||||
Outcome/randomization arm . | High REALMa . | Low REALMa . | Difference (high − low)b . | High REALMa . | Low REALMa . | Difference (high − low)b . | Difference in change between combinations of race and REALM for HE or UC versus OA . |
AIMS2 painc | |||||||
OA | −0.7 | −0.4 | −0.2 | −0.3 | −1.5 | 1.2 | — |
HE | −0.1 | 0.4 | −0.5 | −0.5 | −0.1 | −0.4 | 1.3 (−0.6 to 3.2); P = 0.18 |
UC | −0.1 | −0.7 | 0.6 | −0.5 | −0.6 | 0.1 | 1.8 (−0.0 to 3.7); P = 0.05 |
AIMS2 mobility | |||||||
OA | 0.2 | −0.1 | 0.3 | −0.7 | −0.6 | −0.1 | — |
HE | −0.0 | 0.1 | −0.1 | −0.1 | −0.4 | 0.3 | −0.8 (−2.1 to 0.4); P = 0.18 |
UC | −0.2 | −0.3 | 0.1 | −0.0 | −0.5 | 0.5 | −0.7 (−1.9 to 0.5); P = 0.24 |
aChanges in the negative direction indicate improvement (i.e. lower scores) at 12 months. bPositive scores indicate greater improvement for participants with low REALM. cThe potential range of AIMS2 measures is 0–10, with lower scores indicating better health outcomes.
Discussion
This study adds to existing literature by demonstrating that a telephone-based OA self-management support intervention could potentially help mitigate race-related disparities in OA outcomes. Because this secondary analysis is an exploratory study, we focus here on describing patterns from the data and highlight areas that might warrant further investigation. Specifically, we found that non-white participants improved more than whites with respect to AIMS2 mobility scores. The baseline difference in AIMS2 mobility between races was 1.2 points, on a scale of 1 to 10 (with a score of 2.3 for non-whites and 1.1 for whites), and the absolute difference in AIMS2 mobility change between non-white and white OA intervention participants was 0.8 points (with non-whites improving and whites slightly worsening). These are relatively modest differences in terms of clinical significance. However, the overall effects of the self-management support intervention were also modest [35], and it is possible that an even more powerful intervention could result in larger racial differences in outcomes.
Of the few studies that have examined racial differences in effects of chronic disease self-management support interventions, including this one, there is mounting evidence that racial minorities can benefit even more than white participants [16, 17]. This may be partly because non-white patients (particularly black patients, who comprised the majority of our non-white participants) tend to start out with worse symptoms and OA-related outcomes than whites [2, 36]. Another underlying factor could be that black individuals place greater value on non-medical approaches, such as those emphasized in this and other self-management support interventions. It is also possible that racial minority patients may have inadequate exposure to or support for practicing these non-medical options prior to participating in these interventions [37].
Our findings also suggest that the OA self-management support intervention yielded a significant difference in AIMS 2 pain (the primary clinical trial outcome) according to REALM score category. Specifically there were somewhat greater improvements in the AIMS2 pain subscale for OA intervention participants with low REALM scores compared with those with high REALM scores. However, this overall difference between REALM categories was driven mainly by the difference among non-whites in the OA intervention: OA participants who were non-white and had low REALM scores improved the most. It is possible that the format of the OA intervention, which included audio recordings, easy-to-read written materials and monthly phone calls from a health educator, was particularly acceptable to non-white participants with low health literacy. Current modes of communicating with patients who have low health literacy about self-management of OA may not be adequate, as there has been little attention to developing evidence-based arthritis self-management support interventions directed at OA patients with low literacy [14]. Patients who had low health literacy and were non-white might have not only had less information about their disease but also been less actively involved in their care, before the intervention, than whites with low health literacy. Although this notion was not testable with our data, other studies have shown that there is generally less information exchange about arthritis between health care providers and non-white patients, as well as less active participation by non-white patients in their clinical visits [37]. In addition, because participants who had low health literacy had worse AIMS2 pain scores at baseline, they likely had more unmet needs related to managing their arthritis. These participants perhaps benefited more than those with high health literacy from the personalized, consistent attention to their arthritis needs.
There are limitations to consider when interpreting these results. This study was conducted at one VAMC, and VA health care users have poorer overall health and more severe symptoms than the general population [38–40]. This population was also largely male, which is characteristic of the VA patient population. These factors may limit generalizability. In addition, the literacy measure that we used, the REALM, is a test of word recognition and not reading comprehension; however, it is predictive of health outcomes [41–43]. An advantage of the REALM is that it requires less time to administer than comprehension tests, and thus, it is appropriate for use in clinical settings with lower level readers as an indicator of potential trouble with printed materials and patient-provider communication [44]. Finally, although we observed differences in two study outcomes (AIMS2 mobility and AIMS2 pain) according to race and literacy, there were no significant differences for the other study outcomes. The specific reason for why these effects were observed for some outcomes and not others is not clear, and these differing results highlight the need to further evaluate whether different behavioral interventions hold strong promise for reducing disparities in OA outcomes.
Conclusion
This telephone-based OA intervention appears to be a promising approach to help close the gap in OA outcomes between white and non-white patients. The program components, characterized by multimodal delivery of information, are also particularly appropriate for patients with low health literacy. This feature could have been a driving factor in the particularly beneficial effect for non-white participants, as reflected in the greater change in AIMS2 pain for non-white participants with low health literacy compared with those who were white or with high health literacy. Additional research is needed to further examine the potential of this or other self-management and behavioral support programs to mitigate racial disparities in OA outcomes. In particular, this study and most other studies have examined racial and ethnic differences in program effectiveness in an exploratory manner; an important next step is to design and power studies to specifically test whether a program can reduce baseline disparities in OA symptoms and functional limitations, as well as to elucidate underlying reasons for these effects.
Funding
This work was supported by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service (IIR 04-016). Drs Bosworth and Weinberger are funded by Career Scientist Award nos. 08-027 and 91-408, respectively. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Department of Veterans Affairs.
Conflict of interest statement
None declared.