Self-management education for persons with arthritis: Managing comorbidity and eliminating health disparities


  • The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.



To compare short-term and long-term effectiveness of the Arthritis Self-Help Course (ASHC) and the Chronic Disease Self-Management Program (CDSMP) for persons with arthritis concerning health care use, health-related quality of life, health behaviors, and arthritis self-efficacy.


Forty-eight workshops were randomized to the ASHC (n = 26) or CDSMP (n = 22). A total of 416 individuals, including 365 African Americans, participated. The mean age for each group was 64 years, mean years of education was 11.7, mean number of chronic conditions was 4, and 75–80% of participants in each group were female. Multivariate statistical tests were used to assess effectiveness within and between programs for all workshop participants and African Americans.


At 4 months all ASHC participants including African Americans, had significant improvements (P ≤ 0.05) in self-efficacy, stretching and strengthening exercises, aerobic exercises, and general health. All CDSMP participants had statistically significant improvements in self-efficacy, disability, pain, and general health. African American CDSMP participants showed statistically significant improvements in general health. Trends toward improvement (P = 0.051–0.100) were shown in 5 variables among African American CDSMP participants and in 4 variables among all CDSMP participants. Statistically significant differences between the 2 programs at 4 months were seen in pain and disability in both groups. The CDSMP produced stronger results. Significant results at 1 year within and between programs were minimal for both groups.


When populations with arthritis and multiple comorbid conditions are targeted, the CDSMP may be most cost effective.


Two community-based self-management education interventions, the disease-specific Arthritis Self-Help Course (ASHC) and the generic Chronic Disease Self-Management Program (CDSMP), have consistently been found to be effective in improving quality of life and reducing health care costs (1–5). Studies comparing the effectiveness of these interventions for persons with arthritis have been limited, although arthritis is the leading cause of disability among US adults (6). If arthritis prevalence rates remain stable, the number of persons ages 65 years and older with arthritis and chronic joint symptoms will nearly double by 2030, and the number of persons with comorbid conditions will increase (7, 8).

The lack of evaluations of the comparative effectiveness of the ASHC and CDSMP for African Americans is also an important knowledge gap, especially because arthritis, as well as comorbidities, have been found to affect African Americans disproportionately (9). Age-adjusted estimates from the 2002 National Health Interview Survey indicate that non-Hispanic African Americans have significantly higher prevalence rates of severe pain, activity limitation, and work limitation attributable to arthritis than non-Hispanic whites (7). Findings from the Johnston County Study in North Carolina verify these conclusions (10, 11). However, the Arthritis Foundation reports that the classic evidence-based arthritis education program, the ASHC offered in a 6-week workshop format, has reached <1% of the US population with arthritis (12). Few African Americans have been included in effectiveness studies. None were included in the study recently reported by Lorig et al comparing the effectiveness of the ASHC and CDSMP among persons with arthritis (6).

An increase in the number of persons with arthritis who receive evidence-based arthritis education is an objective of Healthy People 2010 (13). Widespread dissemination of currently available evidence-based interventions such as the ASHC and CDSMP is an approach to improving the health of all Americans, reducing disparity among African Americans, and reducing the national impact of arthritis. If generic self-management education programs such as the CDSMP produce equal or greater benefits for persons with arthritis compared with the disease-specific ASHC, they may be more cost effective. The present study can provide information to guide decision making by health care policymakers, voluntary health organizations, health care providers, and persons with arthritis. We compared short-term (4 months) and long-term (1 year) effectiveness of the ASHC and CDSMP among persons with arthritis, including African Americans, on 4 dimensions: health care use, health-related quality of life, self-management behaviors, and self-efficacy for arthritis self-management.


Preliminary work.

We used a participatory action research approach. Working with African Americans in the southeastern US, we first evaluated the cultural acceptability of the ASHC and CDSMP. We then revised content and delivery methods and field tested the revisions. We conducted 5 focus groups with adult African Americans with self-identified arthritis and other chronic conditions (n = 69). Three focus groups were used to inform the initial workshop revisions (n = 46), and 2 were conducted to refine the modified courses (n = 23). Study investigators and community leaders co-led focus groups and analyzed the transcripts.

We found that self-management, particularly when addressed within a context of mutual help, was consistent with African American cultures. In the words of a hymn sung spontaneously by focus group participants, “If I can help somebody as I pass along, then my living shall not be in vain” (14). We also learned that content needed to be strengthened concerning 1) cultural dimensions of eating, 2) dietary fat and salt, 3) cross-cultural communication with professional health care providers, 4) faith and spirituality, and 5) mutual help and social support. We tailored 3 existing activities by adding a sentence to each that 1) recognized cultural barriers to healthy eating, 2) addressed dietary fat and salt, and 3) emphasized the 2-way nature of cross-cultural communication with health professionals. We added 1 cognitive symptom management activity, using one's spirituality, to the curricula of both programs and consistently affirmed the existing strengths of African American participants (faith, spirituality, and mutual support) throughout program curricula. The Patient Education Research Center of Stanford University, the developer of the ASHC and CDSMP, and our granting agency approved the inclusion of these modifications.

We also learned of the need for a third lay leader, known and respected by the community, to recruit participants and arrange group meetings at acceptable times and places. These individuals, termed community coordinators, linked the university with workshop leaders and the target populations. Workshop delivery methods, e.g., via community coordinators, are not part of the curricula of either the ASHC or the CDSMP and have frequently been modified by other disseminating institutions. The process of modifying both the curricula and the program delivery methods is described in detail in a forthcoming article.

Clinical trial design.

We used a group randomized controlled trial design to assess changes in outcome variables at 4 and 12 months postbaseline. After participants completed baseline questionnaires and community coordinators composed workshops, researchers randomized workshops to either the ASHC or CDSMP according to predetermined randomization lists. Workshops, not participants, were randomized to maintain the community-based focus of the interventions and to minimize diffusion of disease self-management information and skills (i.e., bias) among participants. The 4-month and 12-month posttest data collection intervals were those customarily used in outcome studies of the ASHC and CDSMP (2, 4). The study was approved by the Institutional Review Board of the University of North Carolina at Chapel Hill. All participants completed a mailed informed consent form.


The ASHC and the CDSMP are community-based disease self-management interventions. Both systematically use the efficacy-enhancing strategies of skill mastery, modeling, symptom reinterpretation, problem solving, and social persuasion. Both include content on exercise, anger/fear/frustration, depression, fatigue management, healthy eating, medication management, and working with health professionals. The ASHC does not include content on advance directives, better breathing, or communication strategies; the CDSMP does not include content on arthritis pain management, osteoporosis, energy conservation, sleep, or making nontraditional treatment decisions (6, 15, 16).

The classic versions of the ASHC and CDSMP are administered in a small group workshop format. The 6 weekly sessions are 2–2.5 hours in length. Both are led by trained lay persons with chronic disease who follow detailed protocols outlined in leader manuals. We cross-trained 81 lay leaders and assigned them to co-lead workshops of 10–12 participants and/or to recruit participants and arrange meeting places. Workshop leader training required 32 hours; community coordinators completed 8 hours of training.


Outcome measures evaluated health care use, health-related quality of life, self-management behaviors, and self-efficacy for arthritis self-management. Measures were chosen because they were valid and reliable, because they were sensitive to change in the range expected in this study, and because their previous use facilitated comparison of findings (6, 17).

Health care use.

We measured self-reported visits to physicians, number of hospitalizations, and number of nights in the hospital. Ritter and colleagues found that chart audit data correlated highly with self-report of outpatient visits (r = 0.70) and days in the hospital (r = 0.83) (18).

Health-related quality of life.

The ASHC and CDSMP aim to improve quality of life as well as reduce health care use. Consequently, we also selected measures reflecting health-related quality of life: self-reported health, health distress, activity limitation, pain, fatigue, and disability. The self-rated health item is used in the National Health Interview Survey and has been found to be a better predictor of future health status than physiologic measures (19–21). The Health Distress Scale was adapted from the Medical Outcome Study and was also used in an early CDSMP effectiveness study (17, 22). The Activities Limitation Scale measures the impact of disease on daily role activities such as recreation and chores and was developed for an early CDSMP effectiveness study (17). Pain and fatigue were measured using visual numeric scales developed by Stewart et al (23) and adapted for use in the studies of the ASHC and CDSMP by Lorig et al (17). Disability was measured with the Short Health Assessment Questionnaire (HAQ) (24). The HAQ is an established measure of arthritis-related disability and has been used in the National Health Interview Survey (25).

Self-management behaviors.

We measured 4 self-management behaviors: minutes per week of stretching, strengthening, and aerobic exercise and use of cognitive symptom management techniques. We used items developed for the original Arthritis Self-Management intervention studies and validated by the Stanford Patient Education Research Center for CDSMP studies (17).

Self-efficacy for arthritis self-management.

We measured participants' confidence in controlling their arthritis symptoms with an 8-item scale developed originally for the Arthritis Self-Management intervention and subsequently used in many studies (26, 27).

Data collection procedures.

Baseline and postbaseline data were collected via self-report questionnaire. Baseline data were collected prior to workshop randomization. Baseline and posttest questionnaires were mailed with a self-addressed and stamped business reply envelope. Questionnaires that were not returned 2 weeks after mailing were followed by a telephone interview conducted by a research team member, as were questionnaires that were returned with missing data. Participants received a $10.00 check for completing each questionnaire.


Participants were recruited from 12 counties in eastern North Carolina with African American populations exceeding the state average of 21.6% (28). Community leaders and community coordinators gave talks at churches, community and senior centers, public libraries, clubs, and service organizations. They also distributed fliers to churches, health care providers' offices, senior centers, and pharmacies; arranged for public service announcements on radio and television stations; and placed advertisements in local African American newspapers. ASHC and CDSMP participants were not recruited separately because most could not prioritize their interest in the management of a single condition such as arthritis.

Sampling criteria included the following: self-reported physician diagnosis of any type of arthritis or persistent chronic joint pain as defined by the Centers for Disease Control and Prevention (CDC), 18 years of age or older, and no previous participation in an arthritis self-management program (9).

A total of 499 persons, including 442 African Americans, completed consent forms and baseline questionnaires. Of these, 416, including 365 African Americans, participated in workshops and were included in the analysis. These numbers of participants provided at least 80% power to detect an effect size of 0.30 SD at the 2-sided 0.05 significance level for all effectiveness measures without cluster adjustment. The ASHC and CDSMP have been shown to result in effect size changes in the range of 0.25 to 0.35 (4, 26). Because we randomized at the workshop level and not the participant level, we used generalized estimating equations (GEE) to account for any intracluster (i.e., intraworkshop) correlations and to report percentage change scores as well as significance levels.

Statistical analyses.

Comparisons of participants' baseline scores on descriptor and dependent variables were made using cluster-adjusted chi-square and 2-sample t-tests. The success of each intervention was determined by comparing 4-month and 12-month scores with baseline scores using cluster-adjusted paired t-tests and the percentage change from baseline. All tests were conducted at the 2-sided 0.05 significance level. Within- and between-program differences ranging between 0.051 and 0.099 significance levels were considered as informative trends. The 2 programs were compared using cluster-adjusted analyses of covariance. Least squares means were generated and compared after adjusting for a prespecified list of covariates: ethnicity, rurality of residence, number of concomitant chronic diseases, and the baseline value of the outcome in question. Participants who attended ≥1 workshop session (n = 416) were included in the analysis. Cluster adjustment was performed through GEE methodology using the Genmod procedure of the SAS computer package, version 9 (29). Findings are reported for all participants and for African Americans. Similarities and differences are noted.


Randomization and retention.

A total of 499 individuals enrolled in the study, including 442 African Americans (Figure 1). Of these, 416 (83%), including 365 (82%) African Americans, participated in the randomly assigned workshops. A total of 231 individuals, including 208 African Americans, participated in 26 ASHC workshops, and 185 individuals, including 157 African Americans, participated in 22 CDSMP workshops. The decrease in participants, from 499 to 416, between enrollment and workshop assignment to the beginning of the workshops was largely due to timing. Workshops were held within a 6-month period during the summer and fall when many persons were unavailable.

Figure 1.

Flow chart of study participants: all participants (African American participants). ASHC = Arthritis Self-Help Course; CDSMP = Chronic Disease Self-Management Program.

All participants in the ASHC workshops completed a mean ± SD of 4.34 ± 2.02 sessions, while the CDSMP participants completed a mean of 4.47 ± 1.80 sessions. African American completion figures were similar: 4.33 ± 2.02 sessions for the ASHC workshops and 4.36 ± 1.83 sessions for the CDSMP workshops. The frequency of workshop session attendance at the ASHC and CDSMP, for all participants and African Americans, was similar and is displayed in Table 1. At 4 months the retention rate for all participants was 77% (n = 320) and for African Americans was 78% (n = 283).

Table 1. Frequency of workshop session attendance (ASHC and CDSMP) for all participants and for African Americans*
Number of sessions attendedAll participantsAfrican Americans
ASHC (n = 231)CDSMP (n = 185)ASHC (n = 208)CDSMP (n = 157)
  • *

    Values are the number (percentage). ASHC = Arthritis Self-Help Course; CDSMP = Chronic Disease Self-Management Program.

024 (10.4)12 (6.5)23 (11.1)12 (7.6)
112 (5.2)8 (4.3)9 (4.3)6 (3.8)
27 (3.0)10 (5.4)7 (3.4)9 (5.7)
315 (6.5)8 (4.3)12 (5.8)7 (4.5)
428 (12.1)29 (15.7)27 (13.0)28 (17.8)
551 (22.1)49 (26.5)47 (22.6)42 (26.8)
694 (40.7)69 (37.3)83 (39.9)53 (33.8)

At 12 months we collected outcome data from the 248 participants, including 220 African Americans, enrolled in the first wave of 28 workshops. Only the 167 participants (67%), including 145 (66%) African Americans, who enrolled in the first wave of workshops completed 12 months of followup before the study period ended. Participants enrolling in the first and second waves did not differ on demographic or disease variables.

At baseline there were no statistically significant differences between all ASHC and CDSMP participants in demographic and disease characteristics when accounting for clustering. Furthermore, when comparing all participants in the 2 programs at baseline, none of the outcome variables revealed a P value <0.180. Consequently, we did not add any covariates to control for random imbalances. Among African Americans, we found 1 statistically significant random imbalance between the 2 programs at baseline. African American participants in the ASHC had a mean ± SD of 3.9 ± 1.7 chronic diseases, including arthritis, whereas participants in the CDSMP had a mean of 4.4 ± 2.0 chronic diseases (P = 0.004). Because the number of chronic diseases was prespecified as an adjustment factor, no additional covariates were included in analyses of African Americans.


We analyzed data from 416 individuals, including 365 African Americans, who participated in workshops during 2003 and 2004. The mean ± SD age for each group was 64 ± 12.78 years, and the mean number of years of education was similar for both groups (11.7 ± 3.14 for all participants and 11.7 ± 3.18 for African Americans). Approximately 82% of all participants were women, and 81% of African Americans were women. A total of 42% of all participants were married, as were 42.6% of African Americans. Participants reported a mean ± SD of 4.2 ± 1.88 chronic diseases, including arthritis; African Americans reported a mean of 4.1 ± 1.86. The most common comorbid conditions for both groups were hypertension, overweight, and diabetes.

Four-month outcomes.

We found 4 statistically significant improvements for all ASHC participants at 4 months (Table 2). Self-efficacy increased 13% (P = 0.004), self-reported general health improved 6% (P = 0.016), and minutes per week of both stretching and strengthening exercises and of aerobic exercise increased 39% (P = 0.023) and 38% (P = 0.016), respectively. The CDSMP produced 4 statistically significant improvements for all participants. Self-efficacy increased 14% (P = 0.038), disability decreased 18% (P = 0.032), self-reported general health improved 7% (P = 0.015), and pain decreased 14% (P = 0.050). A 13.0% decrease in health distress (P = 0.073), a 54% increase in cognitive symptom management (P = 0.054), an 8% decrease in fatigue (P = 0.065), and a 34.4% increase in minutes per week of stretching and strengthening exercises (P = 0.078) approached statistical significance.

Table 2. Four-month within-program changes from baseline for all participants (176 ASHC and 144 CDSMP) and African American participants (160 ASHC and 123 CDSMP)*
% changeP% changeP
  • *

    ASHC = Arthritis Self-Help Course; CDSMP = Chronic Disease Self-Management Program; ↑ = higher value desirable; ↓ = lower value desirable.

  • Statistically significant finding.

Health distress ↓
 All participants−1.30.777−13.00.073
 African Americans−0.690.834−13.40.077
Self-efficacy ↑
 All participants12.90.00414.10.038
 African Americans12.50.00511.20.102
Activity limitation ↓
 All participants−2.30.729−9.80.227
 African Americans−1.90.783−8.00.571
Disability ↓
 All participants4.80.543−17.60.032
 African Americans4.00.638−20.00.073
Self-reported general health ↓
 All participants6.40.0167.20.015
 African Americans5.40.0178.40.011
Cognitive symptom management ↑
 All participants23.90.24154.10.054
 African Americans23.70.29753.80.062
Fatigue ↓
 All participants2.20.686−7.80.065
 African Americans2.20.709−5.80.188
Pain ↓
 All participants1.50.789−13.70.050
 African Americans2.30.624−13.30.142
Stretching/strengthening exercise (minutes/week) ↑
 All participants39.20.02334.40.078
 African Americans33.00.04760.30.056
Aerobic exercise (minutes/week)↑
 All participants37.70.01617.50.180
 African Americans42.90.02331.90.919
Physician visits (past 4 months) ↓
 All participants6.10.771−17.30.149
 African Americans10.70.745−27.60.098
Hospitalizations ↓
 All participants12.50.77641.50.471
 African Americans40.00.66988.00.496
Nights in hospital ↓
 All participants43.60.570−36.40.557
 African Americans221.50.441−119.30.587

An examination of the within-program 4-month differences from baseline for African Americans is also reported in Table 2. The ASHC produced significant changes in the same 4 variables as seen in the analysis of all participants. Self-efficacy increased 13% (P = 0.005), self-reported general health improved 5% (P = 0.017), and minutes per week spent performing stretching and strengthening exercises and aerobic exercises increased 33% (P = 0.047) and 43% (P = 0.023), respectively. African American CDSMP participants showed a statistically significant 8% increase in self-reported general health (P = 0.011). The 13% improvement in health distress (P = 0.077), 20% improvement in disability (P = 0.073), 54% improvement in cognitive symptom management (P = 0.062), and 60% improvement in time spent in stretching and strengthening exercises (P = 0.056) among African Americans represent noteworthy trends.

As seen in Table 3, comparison of the ASHC and CDSMP at 4 months revealed statistically significant differences among all participants in disability (mean change 0.043 versus −0.070; P = 0.009) and pain (mean change −0.005 versus −0.931; P = 0.002). Both outcomes indicate that the CDSMP produced greater improvements than the ASHC. The analysis of the African American subset produced findings consistent with those observed among all participants: similar and significant differences between the ASHC and CDSMP adjusted means for disability (0.020 versus −0.095; P = 0.018) and pain (0.149 versus −0.867; P = 0.001). A trend toward improvement in the number of physician visits (P = 0.07) was also seen.

Table 3. Comparison of 4-month mean changes between programs for all participants (176 ASHC and 144 CDSMP) and African American participants (160 ASHC and 123 CDSMP)*
 ASHCCDSMPP of difference
  • *

    See Table 2 for definitions.

  • Statistically significant finding.

Health distress ↓   
 All participants−0.177−0.1990.559
 African Americans−0.063−0.1800.445
Self-efficacy ↑   
 All participants0.7720.6110.601
 African Americans0.8300.6040.515
Activity limitation ↓   
 All participants−0.025−0.1180.414
 African Americans−0.022−0.0880.614
Disability ↓   
 All participants0.043−0.0700.009
 African Americans0.020−0.0950.018
Self-reported general health ↓   
 All participants−0.221−0.2420.797
 African Americans−0.204−0.2670.424
Cognitive stress management ↑   
 All participants2.9623.6350.790
 African Americans2.9523.9210.728
Fatigue ↓   
 All participants0.005−0.3280.259
 African Americans0.043−0.1940.462
Pain ↓   
 All participants−0.005−0.9310.002
 African Americans0.149−0.8670.001
Stretching/strengthening exercise (minutes/week) ↑   
 All participants18.87913.0880.289
 African Americans14.0059.3730.425
Aerobic exercise (minutes/week) ↑   
 All participants26.6937.6130.082
 African American28.16515.0820.248
Physician visits (past 4 months) ↓   
 All participants0.910−0.1960.170
 African Americans0.366−1.0250.074
Hospitalizations ↓   
 All participants−0.0020.0540.603
 African Americans0.0470.0750.812
Nights in hospital ↓   
 All participants−0.131−0.4120.615
 African American0.255−0.1650.501

Twelve-month outcomes.

There were fewer significant results at 1 year. Among all ASHC participants (n = 83), only the 4-month improvement in arthritis self-efficacy was maintained for 12 months (14.9%; P = 0.036). Trends toward significant improvements were seen in self-reported general health, pain, and hospitalization. Among African American ASHC participants (n = 76), we found modest, although not statistically significant, maintenance in self-efficacy improvements from baseline (P = 0.069).

None of the statistically significant changes seen among all CDSMP participants (n = 84) at 4 months remained at 12 months. An increase in practice of cognitive stress management, however, emerged as statistically significant (81%; P = 0.031). The improvement in stretching and strengthening exercise approached statistical significance (83.3%; P = 0.055).

A comparison of the adjusted means between all participants in the ASHC and CDSMP at 12 months revealed no statistically significant differences. The stronger impact on disability and pain produced by the CDSMP program at 4 months postintervention was no longer present at 12 months. A decrease in number of physician visits approached statistical significance (P = 0.064), with changes among CDSMP participants accounting for the improvement.

The only significant change seen in African American CDSMP participants (n = 69) at 12 months was in minutes per week spent in stretching and strengthening exercise (increase of 83%; P = 0.045). We also saw marginal improvements from baseline in self-reported general health (6.5%; P = 0.098) and the practice of cognitive stress management techniques (86%; P = 0.054). Comparison of the adjusted means demonstrated that the program differences seen in disability and pain among African Americans at 4 months were not maintained at 12 months.


In summary, we found statistically significant improvements in all ASHC and CDSMP participants at 4 months in 2 of the 4 outcome variable categories, self-efficacy and health-related quality of life. All ASHC participants also reported statistically significant improvements in both stretching and strengthening exercise and aerobic exercise. All CDSMP participants also showed improvements in disability and pain. African American participants in both the ASHC and CDSMP showed statistically significant improvements in self-reported general health. When comparing the 4-month outcomes of all ASHC and CDSMP participants and African Americans, we found statistically significant improvements in pain and disability attributable to the CDSMP (see Table 2).

This study demonstrates positive benefits for persons with arthritis from both the disease-specific ASHC and the more generic CDSMP. The study is the first to document the positive benefits of these programs for African Americans and for persons with multiple comorbid conditions. The greater improvements in pain and disability among CDSMP participants may, in fact, be due to the high rate of comorbidity in the sample.

Unlike effectiveness studies reported earlier, the number of positive outcomes at 12 months was limited (3, 26). We speculate that these differences may be related to differences in sample composition. The sample in the study recently reported by Lorig et al, for instance, included 1) no African Americans; 2) participants with a mean of 15.4 years of education, compared with 11.7 years in this study; and 3) participants with an unspecified number of comorbid conditions, compared with an average of 4 chronic conditions in this sample (6).

We also speculate that the positive effects seen at 4 months are not sufficiently powerful to be maintained for 12 months because of the relatively low priority given to health by our study participants. Contextual and social address factors, such as family and church communities, place of residence, and socioeconomic status, could be more powerful predictors of disease self-management in the long term than a 6-week workshop. This is even more likely among individuals from the target populations who did not self-select as study participants. Although this finding is not consistent with those obtained by Lorig and colleagues (6), it is plausible. Reinforcement may be warranted in disease self-management education with some populations (30). The lack of maintenance of positive effects may also be due to lessened statistical power as a result of the sharply decreased sample size at 12 months.

Based on the results of this study, health care practitioners can confidently refer their patients with arthritis to either the ASHC or CDSMP. The CDSMP may be most feasibly offered in areas with small numbers of potential ASHC participants, an aged population with multiple diseases and limited finances, or a population with limited programmatic resources. Persons with several comorbid conditions may see the CDSMP as being most relevant to their lives because of its generic focus. Persons specifically wishing to improve their skills in arthritis self-management may choose the ASHC.

In summary, population density, community resources, participant levels of comorbidity, and professional and participant preferences are likely to influence which disease self-management program is most appropriate in any given situation. The CDC Arthritis Program has added the CDSMP to the list of previously approved community-based interventions, including the ASHC, that CDC-funded state arthritis programs can use federal dollars to support.


Dr. Goeppinger had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Goeppinger, Schwartz, Brady.

Acquisition of data. Goeppinger, Ensley.

Analysis and interpretation of data. Goeppinger, Armstrong, Schwartz, Brady.

Manuscript preparation. Goeppinger, Armstrong, Schwartz, Brady.

Statistical analysis. Armstrong, Schwartz.


The authors gratefully acknowledge the able and energetic assistance of our community partners, Sydney Barnwell, MD, Stephanie Fisher, BSN, Jerome Garner, MPH, Rose Haddock, MEd, and Theresa Taylor, BS, and of our university-based research managers and field coordinator, Dennis Zaenger, MPH, Alexis Moore, MPH, and Donna Harris, BSN.