Changes in Diabetes Self-Care Behaviors Make a Difference in Glycemic Control: The Diabetes Stages of Change (DiSC) study

The complete methodology for this study is described elsewhere (9). Briefly, a total of 1,029 individuals with type 1 or type 2 diabetes were recruited for this 12-month study, approved by the Research Ethics Boards of the University of Toronto and Dalhousie University. Two thirds of the cohort were recruited from Southern Ontario, Canada and one third of the subjects were recruited from Nova Scotia. The aim was to recruit individuals from the general diabetes population who were not engaging in optimal self-care practices. Participants were considered as being in a pre-action stage if they performed SMBG fewer than four times per day if treated with insulin or fewer than two times per day if treated with oral antihyperglycemic agents alone, and/or if they had a BMI >27 kg/m², and/or if they smoked cigarettes. Patients were ineligible for participation in the study if they were on diet therapy alone, if they could not respond to the materials written in English, if they required more than usual care because of their health status (e.g., were pregnant or on dialysis), or if they had no telephone. Participants were recruited from posters in local family practices, diabetes education centers, newspaper ads, and Canadian Diabetes Association chapter activities.

This study was a randomized split-plot design with two randomized between-subject factors: treatment (PTC versus TAU) and strips (free strips versus no strips) and one within-subject factor (pre- versus post-study). Because treatment recommendations are different, participants were stratified according to whether or not they took insulin or oral agents alone and were then randomized into treatment or strips conditions.

Participants received a PTC intervention for each of the behaviors appropriate for them. The PTC intervention is fully described elsewhere (9). In summary, PTC is an integrated, multicomponent intervention program that provides monthly mail or telephone contact for 12 months. The components of the intervention are based on the individual’s current stage of change, as well as stage-matched processes and principles of change. A handbook providing general diabetes information and an introduction to the TTM was mailed to PTC participants after recruitment. Individual stage was assessed using quarterly assessment surveys. Survey responses were used to generate a personal report from a computer-based “expert system” (10–16), which provides personalized feedback for each self-care behavior. This report is stage-matched and individualized to provide participants with the strategies most useful in moving toward action and maintenance. Counselors conducted telephone calls 1 month after the personal report to answer questions, provide tips on behavior change, and facilitate personal goal-setting. Seven newsletters were mailed approximately every other month so that monthly contact was made for 12 months.

TAU involved regular family physician or endocrinology visits and/or diabetes education sessions as prescribed. Education and physician care is covered by provincial health plans. No attempt was made to influence medical management in either group.

All subjects with BMI >27 kg/m² (a measure of obesity) were enrolled in the healthy eating intervention, but only those reporting being in a pre-action stage for healthy eating (diet >30% fat, n = 445) were used for the analyses reported herein.

The protocol required participants to visit the diabetes center twice during the study, for baseline and end-of-study assessments. Participants were asked to use a memory blood glucose meter (LifeScan Profile; LifeScan, Milpitas, California) and phone modem to send blood glucose results to a data center (Enact, Mountain View, California) for the duration of the study. These were distributed free of charge to all participants. Patients’ blood glucose data were used solely for outcome assessment and not as an intervention.

Stage of change was assessed through questions addressing intention to change and current behavior, as described in Table 1. A specific action criterion was used to drive the staging algorithms. Participants were informed of this criterion before intention was assessed. TTM measures were developed using exploratory and confirmatory analyses, structural equation modeling, and split-half cross-validation (15).

HbA_1c was measured from a venous blood sample using an immunoturbidometric method. Height and weight were measured and BMI was calculated (kg/m²). Dietary intake was measured using a validated food frequency instrument, the NCI Block (17,18). For PTC participants, the staging measures were obtained at baseline and at 3, 6, 9, and 12 months. For TAU participants, there was no contact from the initial assessment until the 12-month end-of-study assessment, except for reminders to download the data from the blood glucose meter, if required. Health care utilization information (visits with family physician, medications used, diabetes education received) was also collected at baseline and at 12 months.

Participants who did not complete the study were coded as remaining in pre-action for the intention-to-treat (ITT) analyses. The main comparisons were between the proportion of participants in PTC versus TAU, and free strips versus no free strips for the SMBG intervention, across the stages (pre-action versus action and maintenance stages) at end of study. Differences in proportions were tested using χ² analyses. If main effects of both treatment and strips were significant for the self-testing risk, the interaction was analyzed by comparing the outcome for each of the four groups (TAU alone, TAU plus strips, PTC alone, and PTC plus strips) using χ² analysis.

Outcomes for glycemic control (HbA_1c), weight, meter memory data, and eating behavior were analyzed using repeated-measures ANOVA. The SMBG frequency data per memory meter was coded to obtain an average weekly blood glucose level for weeks 5–8 and 49–52 of the study. Data from weeks 1–4 were not used to minimize any temporary increase in SMBG frequency associated with starting the study.

Participants’ baseline characteristics are shown in Table 2. There were no significant differences between groups. The 1,029 participants received one or more interventions, depending on their current stage for each behavior. A total of 860 participants were enrolled for SMBG; 77.7% completed the entire year of the study. A total of 445 participants were in pre-action for healthy eating, and 168 participants were in pre-action for smoking; completion rates were 81.2 and 80.4%, respectively. Participants who did not complete the entire 12 months of the study did not have different baseline demographic characteristics from those who did complete the study (data not shown). A total of 58.2% of participants were enrolled for two behaviors (89% of these for both SMBG and healthy eating) and 8.2% of participants were enrolled for three behaviors.

We hypothesized that significantly more participants in the PTC intervention than the TAU intervention would shift from pre-action to an action stage (either action or maintenance). In view of this directional hypothesis and the fact that all participants were in a pre-action stage at baseline and could not regress further, a one-tailed statistical test was used. ITT analyses comparing treatment groups for each self-care behavior are presented in Table 3.

There was a significant treatment effect for PTC versus TAU (χ² = 45.42, P < 0.001) and for whether or not free testing strips were provided (χ² = 13.13, P < 0.002). Table 3 shows the data examining the four treatment groups (TAU alone, TAU plus strips, PTC alone, PTC plus strips); 10.9% of the TAU group moved to action and 7.5% moved to maintenance, whereas of those receiving testing strips alone, 17.3% moved to action and 9.7% moved to maintenance. Of those receiving PTC alone, 12.1% moved to action and 18.4% moved to maintenance. When PTC and strips were combined, 10.2% of participants moved to action and 33.2% moved to maintenance (P < 0.001).

The χ² analysis based on the ITT sample (445 in pre-action at baseline) demonstrated that PTC was associated with more movement to action/maintenance than TAU (P < 0.001). The superiority of PTC can be seen in the larger proportion of participants who are in maintenance at end of study (11.4%) compared with TAU (0%) (χ² = 26.52, P < 0.001).

There were significantly more PTC than TAU patients who quit smoking (17.8% in action and 5.5% in maintenance vs. 11.6% in action and 0% in maintenance, respectively) in the ITT analysis for the 168 enrolled smokers (χ² = 7.01, P = 0.02).

Differences in favor of PTC are seen for SMBG frequency in those providing end-of-study data (P = 0.002). Only PTC participants significantly increased SMBG frequency (from 1.4 to 1.6 tests per day) as measured by downloaded meter memory data, compared with a slight decrease in the TAU group (1.46 to 1.41 tests per day, P < 0.002). Likewise, the results for the NCI Block Food Frequency Questionnaire indicated that PTC participants, relative to TAU participants, reported lower percentage of calories from fat (35.24 vs. 36.1%, P < 0.004) and higher daily vegetable (2.24 vs. 2.06, P < 0.011) and fruit servings (1.89 vs. 1.68, P < 0.01).

Participants’ visits to family physicians remained consistent: 4.61 visits for the TAU group and 4.75 visits for the PTC group during the study year; 4.87 and 4.74 visits, respectively, for the prior year (NS). Participants in both groups changed the dose or frequency of oral agent therapy equally during the study year (84.6% of the PTC group versus 83.3% of the TAU group). Although 22% of PTC insulin-using participants changed to an insulin analog, as compared with 13% of the TAU group, this did not reach significance (P = 0.082). Both PTC and TAU participants increased the frequency of injections (from 2.45 to 2.76 per day in the PTC group, from 2.32 to 2.54 in the TAU group; pre- versus post-F = 32.36, P < 0.001; interaction F = 0.96, NS). Finally, more patients in the PTC group participated in diabetes education programs than patients in the TAU group (47.7 and 37.7%, respectively; P = 0.002).

In an ITT analysis, a significant reduction in HbA_1c resulted only for those participants who reached action or maintenance stages over the course of the study. End-of-study HbA_1c for the SMBG intervention was 7.78% for those who reached an action stage, compared with 8.30% in those who stayed in pre-action (P < 0.003). In the healthy eating group, HbA_1c was 7.83% for those in action but remained at 8.38% for those in pre-action (P < 0.001). Increased SMBG accounted for 3% of the variance in end-of-study HbA_1c level in the non–insulin-treated group (P < 0.01) and 7% in the insulin-treated group (P < 0.01).

In the healthy eating group, weight reduction was greater for the PTC participants who were in an action stage at end-of-study as compared with those in pre-action, but this did not reach significance. Weight loss for those remaining in pre-action was 0.65 kg, compared with 1.38 kg in patients moving to action (NS). However, weight loss for those enrolled in both SMBG and healthy eating interventions, and who increased SMBG frequency as recommended, was significantly greater than those who remained in a pre-action stage (1.78- versus 0.26-kg loss, P < 0.01).

These data suggest that the PTC intervention is significantly better than TAU in helping individuals move into action stages of critical diabetes self-care behaviors. The PTC intervention was successful at helping more people engage in SMBG, make healthy low-fat food choices, and stop smoking. Patients who changed SMBG and healthy eating behavior had a significant improvement in metabolic control, whether they used insulin or oral agents. Results for smoking cessation suggest that PTC is superior to TAU, but the confidence in these findings is limited by small sample size.

The importance of self-care behavior change interventions for improved glycemic control is supported by recent studies.

For example, an intensive individualized behavioral change program for individuals with impaired glucose tolerance was successful in reducing the incidence of diabetes (19). However, individual interventions are labor-intensive and time-consuming to deliver to the large numbers of patients with diabetes. In contrast to face-to-face diabetes education efforts, the PTC interventions are efficient and have a broad reach. Intervention was accomplished by mailing individualized written material, with follow-up telephone counseling but no clinic visits. This program provides stage-matched as well as individually tailored feedback to participants’ personal survey responses. The system provides individualized normative and/or ipsative feedback, which allows even small positive steps to be reinforced. As implemented, this intervention may positively impact the health of broad populations of individuals, not just those who attend diabetes clinics, and augment the impact of diabetes care/education services.

Although SMBG is accepted as critical for patients using insulin, its role in patients with type 2 diabetes is less clear. In a cross-sectional sample of patients with type 2 diabetes from the National Health and Nutrition Examination Survey (NHANES), the frequency of SMBG was not found to be related to glycemic control as measured by HbA_1c (20). In our study, patients with type 2 diabetes taking oral agents who performed SMBG more frequently did demonstrate a significant reduction in HbA_1c, which supports other recent findings (21). The DiSC study data are unique because of the prospective nature of collection. Furthermore, the participants of this study were encouraged to learn more about using blood glucose information. Those that tested more had a significant reduction of HbA_1c regardless of treatment used.

This study also demonstrated the potentiating effects of access to free testing strips on both PTC and TAU interventions. Access to supplies does impact frequency of SMBG, and testing frequency improves HbA_1c. These findings support long-term benefit of financial coverage of testing supplies by third-party payers.

Future research is required to isolate the impact of each component of the PTC. Because the comparison group in this study was TAU, we are unable to state definitely whether the theory-specific aspects of PTC, or general aspects of behavioral intervention, were responsible for the outcomes. Future research should compare the efficacy of a stage approach with a nonstaged frequent contact approach in the context of complex multiple and concurrent self-care behaviors of chronic illness.

The PTC intervention has the capacity to deliver effective interventions to a large number of people. By helping people through the precontemplation and contemplation stages in a timely fashion, we may be able to redirect our efforts to diabetes care and education strategies that will have a positive impact on the prevention of long-term complications of diabetes, reducing the subsequent human and health care costs. Clearly, this study supports a new, innovative approach to this major health problem.

This research was supported by an unrestricted grant from LifeScan, a Johnson and Johnson Company. Pathways to Change (PTC) is a registered trademark of LifeScan, a Johnson and Johnson Company.

We thank Wanda Firth, Celia Fredericks, Lewinda Knowles, Lea Lehman, Hilary Ross, and Mildred Thompson for their valuable contributions to the DiSC study.