This study was a comparison of one intervention arm, the CDSMP, and the control arm from an open-label, 4-arm randomized controlled trial that was designed to evaluate the effectiveness of two different type 2 diabetes mellitus (T2DM) self-care interventions (implemented singly and in combination) on glycemic control. Designed with the acknowledgment that both patients and researchers would be aware of the random assignment, the study protocol consisted of screening potential subjects for eligibility, randomizing them to one of four study arms, and following them over a 24-mo period. However, the primary end-point was change in HbA1c from randomization/baseline to 12 mo of follow-up. The current study reported here focuses on participants in two of the four original study arms.

The study protocol was approved by the Institutional Review Boards (IRB) of Scott and White Healthcare System and Texas A and M Health Science Center. All qualified participants accepted the conditions of the study and gave informed written consent at enrollment/orientation. Enrollment occurred between January 2009 and June 2011 and data collection was completed in July 2012. We adhered to the CONSORT protocol[30] and registered the trial with clinicaltrials.gov (NCT01221090).

Participants represent a subset of subjects that were recruited from seven participating clinics of a large integrated healthcare system, a university-affiliated, multi-specialty group practice associated with a 250000-member Health Maintenance Organization in central Texas. Potential participants were identified through electronic medical records if they: (1) had a diagnosis of T2DM; (2) were ≥ 18 years; (3) had a lab assessed HbA1c value ≥ 7.5% (≥ 58 mmol/mol) within the last six months; and (4) were able to communicate in English. Subjects were excluded if they: (1) had documented reports of alcoholism or drug abuse; (2) were pregnant or planning to become pregnant within 12 mo; or (3) were unwilling to sign an informed consent. Recruitment was solicited by physicians within the seven clinics who agreed to invite their patients to participate in the study.

Physicians were provided with IRB approved invitation-to-participate letters and a list of their T2DM patients meeting the threshold HbA1c level at their last visit. Contact was initiated with potential subjects through physician-sent letters, describing the study and requesting a completed screening enrollment card if interested. Subjects who returned a screening enrollment card were contacted by project coordinators, who provided additional information and screened them to determine eligibility. To verify the inclusion and exclusion criteria, subject permission was obtained to review their medical records. Other recruitment strategies included oral referrals by physicians and patient educators and posting messages in waiting areas of study clinics.

Lab assessments were continuously monitored at each phase of the study recruitment to ensure that enrolled participants had HbA1c values ≥ 7.5% (≥ 58 mmol/mol) within the last six months since individuals who previously met this criterion may no longer fulfill that requirement at orientation. A follow-up telephone interview was conducted to determine participation interest. Lab results were screened to ensure that the participant met qualifying HbA1c and if needed, tests were scheduled.

Participants randomized to the CDSMP arm were invited to attend a 6-wk, classroom-based program for diabetes self-management. The effectiveness of the CDSMP has been described elsewhere[10]. With the goal of increasing self-efficacy to ultimately decrease chronic disease related symptoms and avoidable healthcare utilization, the CDSMP teaches participants techniques to facilitate enhanced decision making, action planning, and effective communication. CDSMP workshops were hosted in clinical environments and community-based settings. While fidelity to the individual classes was not monitored, CDSMP license requires that lay leaders use pre-scripted materials and that experienced master trainers/lay leaders (who attend a required four-day training program) lead the workshops.

Participants randomized to the control arm did not receive any treatment other than their usual clinical diabetes care, along with some publicly available Texas Diabetes Council patient education materials.

Study measures were obtained at orientation/baseline, 6 mo, and 12 mo of follow-up. Participants received monetary compensation in the form of a gift card for travel expenses and time, consisting of $20 at orientation and at the 12-mo follow-up visit.

At orientation, a questionnaire was administered to obtain several pieces of information including: (1) demographics such as age, gender, and race/ethnicity; (2) diabetes self-care activity monitoring (number of days, 0-7, that any specific self-care activity was performed in the past week) as measured by the Summary of Diabetes Self-Care Activities instrument; (3) self-reported health-related quality of life (HRQOL) measures (e.g., number of days physical/mental health was not good); and (4) pain and fatigue measures (on a scale of 1-10, 1 indicating none and 10 severe). Questionnaires were administered every 6 mo. However, as our primary end point was 12 mo, analyses were only conducted for this time period.

Anthropometric data were obtained at orientation and at subsequent follow-up visits. Height in inches was measured without shoes. Weight was measured in pounds on a balance beam scale or an electronic scale without shoes. Body mass index (BMI) was computed from height and weight measurements. Blood pressures were recorded with either a mercury sphygmomanometer or a validated automated device. Participants who were unable to come in for their follow-up appointments had their height, weight, and blood pressure data abstracted from electronic health records (EHRs). Measures recorded fell within the range of 10 d prior to and 45 d after participants’ scheduled follow-up dates. This was done to obtain participant visits as close to their target dates as possible, but also allow for enough time after the target date to accommodate for scheduling errors (i.e., missed appointments, rescheduling).

Measures of HbA1c were collected from EHRs dating back 6 mo prior to orientation to the last day of study participation (45 d after the 12-mo follow-up period). If a participant did not have any HbA1c value within the EHR for any particular follow-up visit, a lab test was scheduled to obtain a measure. Of the HbA1c collected 6 mo prior to orientation, the value measured closest to the orientation date was considered as the baseline HbA1c value. HbA1c values that were measured on dates preceding the baseline HbA1c were not included; i.e., HbA1c values included in the analysis were those collected since the baseline HbA1c and until the last day of study participation.

A participant was considered to have completed a follow-up if there was an available HbA1c within the designated follow-up period, i.e., within the cut-off dates, defined as within 45 d after the scheduled follow-up dates. For the 6-mo follow-up measure, if at least one HbA1c was available after baseline and before the 6-mo cut-off, the participant was considered to have completed a follow-up. For the 12-mo follow-up measure, the designated range was between the 6-mo cut-off date and the 12-mo cut-off date. Participants who were unable to complete an assessment at one time period were not excluded from future assessments. For instance, if a participant did not have any HbA1c measured within the specified time period for their 6-mo follow-up but had one available for their 12-mo follow-up, he/she was considered to have completed the 12-mo follow-up, but not the 6-mo.

The primary study outcome measure was change in HbA1c from randomization to 12 mo of follow-up. Secondary outcome measures included BMI and blood pressure, along with several self-management behavioral measures (e.g., foot care) from randomization to 12 mo of follow-up.

Analysis was based on intent-to-treat. Descriptive statistics were used to describe baseline demographic, anthropometric, and clinical characteristics by study arm. Analysis of variance as used to compare average changes in self-management behaviors between study arms. To determine whether the treatment had an effect on the rate of change in HbA1c level over time, we used linear mixed models that included time as a continuous variable. A spatial power covariance structure with time as the distance measure accounted for the time-series correlation among repeated measurements on each subject. Forward selection was utilized, in which powers of time were added one at a time to the base model including treatment group effects only. Time and treatment effects were then added gradually and evaluated with likelihood ratio tests to assess any effect modification. The final mixed model included time, time squared, treatment group, and the interaction between time and treatment group as fixed effects. HbA1c values included in the analysis were those falling within the time frame of 6 mo prior to orientation until the 12-mo follow-up cut-off point.

The flow diagram of participant enrollment and disposition in the trial has been described elsewhere[31]. Of the subjects randomized, 101 entered the CDSMP arm and 95 entered the control arm. Of the participants assigned to the CDSMP, 75.6% attended 4 of 6 sessions required for successful completion.

Demographic and baseline clinical characteristics were generally comparable between the two groups (Table 1). The mean age of participants was 57.6 ± 10.9 years. Slightly more than a third (36.4%) was of minority status, self-reporting as either African American or Hispanic. The majority of participants had received post-secondary education; 40% had attended some college or vocational school, 20% were college graduates, and 13% had completed higher forms of education. Approximately one-third reported annual incomes greater than $50000, while almost 40% reported annual incomes between $25000 and $49999.

An overwhelming majority (92.9%) of the participants were either overweight or obese, with a mean BMI of 34.3 ± 7.4 kg/m². While measures of systolic blood pressure were comparable between study arms, with a mean of 134.8 ± 19.3 mmHg, measures of diastolic blood pressure were significantly different (P < 0.002). The mean baseline HbA1c for participants was 9.3% ± 1.6% and did not differ significantly between the two groups.

Table 2 summarizes participants’ diabetes self-care activity (DSCA) monitoring, HRQOL measures, and pain and fatigue measures at baseline. Participants in the control arm reported checking their feet more frequently than those in the CDSMP arm (P = 0.04). Although participants in the control group reported inspecting the inside of their shoes more frequently and also tended to report fewer unhealthy physical days and experience less limited days due to physical and mental health, these did not reach statistical significance (P≥ 0.05).

There were modest but statistically significant reductions in HbA1c from baseline to 12 mo of follow-up. The results of the linear mixed model are presented in Table 3. The adjusted reductions in HbA1c over the 12 mo of follow-up for the CDSMP and control groups were 0.559% and 0.576%, respectively (P < 0.0001). However, the interaction term of the treatment group and time was not statistically significant (P = 0.885), implying no significant difference in HbA1c reductions by treatment assignment.

The mean difference in the number of days (within the last 7 d), from baseline to 12 mo of follow-up, that participants reported using specific diabetes self-care activity features were compared between the CDSMP and control arms (table not shown). While there were no differences on 12 of the 14 self-care indicators, participants in the control arm had a higher rate of change in checking their feet than those in the CDSMP arm (increase of 0.28 d/mo vs 0.20 d/mo; P = 0.02). Similarly, participants in the control arm reported an increase of 0.15 d/mo eating 5 or more servings of fruits and vegetables compared to an increase of 0.01 d/mo reported by those in the CDSMP arm (P = 0.02).

The Stanford Chronic Disease Self-Management Program (CDSMP) represents one of the most studied evidence-based behavioral or self-care programs for chronic diseases including diabetes.

The CDSMP has been found to be highly effective in improving the general health of people with several chronic conditions such as heart disease and arthritis. Recent evidence indicates that the CDSMP is a useful and appropriate program for lowering glycated hemoglobin A1c (HbA1c) among people with type 2 diabetes who are out of control.

This study demonstrated that the CDSMP may not lower HbA1c among people with type 2 diabetes any better than good routine care in an integrated healthcare system.

Findings from this study show that people with type 2 diabetes managed with good routine care in an integrated healthcare system can also have good glycemic control. Nonetheless more research is needed to understand the benefits of self-care programs in primary care.

The study by Forjuoh et al aimed to assess the effectiveness of the CDSMP on the metabolic control. This is an interesting investigation from a practical point of view.