Mixed methods study of engagement in behaviors to prevent type 2 diabetes among employees with pre-diabetes

Study sample

We conducted a mixed methods observational study of University of Michigan (U-M) employees who were found to have pre-diabetes during employer-sponsored screenings organized by MHealthy, U-M's campus-wide wellness program. The screenings were advertised through email and flyers and conducted across multiple days and times at numerous locations. Employees of U-M were offered a $100 incentive for completing the screening, which included an HRA and measurements of body mass index (BMI), blood pressure, blood glucose, and cholesterol. Participants were asked, but not required, to be fasting. Screenings were conducted between 1 January 2014 and 31 May 2014.

Screening test results were communicated in person by a registered nurse health coach from The StayWell Company, LLC (‘StayWell’). Prior to the communication of fasting blood glucose (FBG) test results, the health coach verified that the employee had been fasting. When an employee had an FBG measurement of 100 to 125 mg/dL (inclusive), the health coach informed the employee that their FBG was in the pre-diabetes range and provided brief counseling about this condition using talking points based on American Diabetes Association21 and National Diabetes Prevention Program9 guidelines (see online supplementary figure S1). Employees with pre-diabetes were also provided with a one-page handout that summarized these points and listed contact information for local DPPs.

Recruitment

Employees identified as having pre-diabetes during screenings were invited by the StayWell health coaches to be contacted by our research team to learn more about a study ‘designed to identify ways to help people prevent type 2 diabetes’. Employees who agreed to be contacted signed a form and provided their contact information. Research staff then contacted these individuals by phone or email. Those who were willing to participate were emailed a survey link to an online informed consent document.

Online surveys

Individuals who provided their free and informed consent were asked to complete two online surveys. A link to the first survey was emailed to participants as soon as possible after they provided informed consent. This first survey inquired about demographics and health literacy,22 ,23 and served as an initial contact with participants to facilitate subsequent data collection in the second survey. The first survey was completed in a median of 18.5 days after screenings (IQR 9–34). The response rate for the first survey was 100%.

Approximately 2 months after completion of the first survey, participants were emailed a link to a second survey. This survey contained questions about engagement in our key behaviors of interest such as, stage of change related to weight loss, discussing pharmacotherapy, and participating in a DPP;24 and physical activity in the past 7 days.25 This survey also included questions about potential mediators of engagement in our key behaviors of interest, such as knowledge of whether type 2 diabetes is preventable; perceptions of risk for type 2 diabetes;26 ,27 level and locus of motivation to prevent type 2 diabetes;28 and perceived competence to (1) attempt weight loss, (2) increase physical activity, (3) discuss with one's primary care provider pharmacotherapy for prevention of type 2 diabetes, and (4) participate in a DPP.29 We measured locus of motivation to prevent type 2 diabetes using the Treatment Self-Regulation Questionnaire (TSRQ). The TSRQ includes subscales that permit measurement of three different types of motivation: autonomous motivation (ie, motivation that comes from internal sources), controlled motivation (ie, motivation that comes from external sources), and amotivation (ie, the absence of motivation).28 These three motivational constructs are important because autonomous motivation is more likely to produce long-term, sustained healthy behaviors relative to either controlled motivation or amotivation.30 The second survey was completed in a median of 94 days after screenings (IQR 78–111). The response rate for the second survey was 96%.

For both surveys, participants received email reminders to encourage completion. After 10 attempts to contact, participants were categorized as being lost to follow-up. Participants received a $10 gift card for each survey completed. All surveys were completed between February and October 2014.

Semistructured telephone interviews

After the surveys, we conducted semistructured telephone interviews with participants to identify facilitators of and barriers to engagement in behaviors to prevent type 2 diabetes. For purposive sampling, we categorized participants based on a manual review of their second survey responses as either (1) attempting weight loss and having gotten at least 150 min of moderate physical activity in the past week, having discussed pharmacotherapy for prevention of type 2 diabetes with a primary care provider since the screening, or participating in a DPP; or (2) not having carried out any of these things recommended to them at the screening. We then randomly selected participants in each of these 2 groups and asked them via email to participate in a telephone interview. Interviews consisted of open-ended questions which aimed to elicit participant-identified facilitators of and barriers to engagement in recommended behaviors (ie, factors that participants felt made behavior change easier and harder, respectively), as well as emotional reactions to receiving a pre-diabetes diagnosis, understanding of pre-diabetes, and reasons for any behavior changes.

Participants who were invited to participate in an interview received email reminders to set up an interview appointment, and after 10 unsuccessful attempts to contact were categorized as being lost to follow-up. Participants who completed an interview received a $10 gift card. In each of the two groups, we stopped conducting interviews once we reached thematic saturation. Eighty-three per cent of participants who were invited to be interviewed completed an interview. All interviews were conducted between August and November 2014.

Statistical analysis

We used data from both surveys to compare demographic characteristics and potential behavioral mediators of participants who either (1) were attempting weight loss and had gotten at least 150 min of moderate physical activity in the past week, had discussed pharmacotherapy for prevention of type 2 diabetes with a primary care provider since the screening, or were participating in a community DPP; or (2) had not carried out any of these things recommended to them at the screening. We compared continuous variables using Wilcoxon rank-sum tests. For categorical variables we used χ² tests or Fisher's exact tests. We used Stata V.13 (Stata Corp, College Station, Texas, USA) to conduct all analyses.

Qualitative analysis

All interviews were audio recorded and transcribed verbatim. Four members of the research team independently reviewed a subset of transcripts using modified grounded theory to identify salient themes.31 They then met to discuss the themes, refine them, and achieve consensus on codes and their definitions. Once the coding scheme was established, two members of the research team independently coded all transcripts in Dedoose software. They then met to discuss their coding and resolve differences by consensus. After all transcripts were double-coded, the research team met to discuss the code summaries and memos, and to identify key themes with a focus on facilitators of and barriers to engagement in recommended behaviors.32

The study protocol was approved by the University of Michigan Medical School Institutional Review Board. Data were analyzed in 2015.

Sample characteristics

Of the 17 310 U-M employees who attended the screenings, 5279 provided an FBG and 523 had an FBG in the pre-diabetes range. Eighty-five of these 523 pre-diabetic individuals consented to study participation and 82 completed both surveys (see online supplementary figure S2). Most (72.0%) of the participants were women, white (73.2%), and had at least a college education (59.8%). The median age was 50.5 (IQR 40 to 56.5) and the median household income was $75 000 (IQR $55 000 to $125 000). Approximately two-thirds (65.9%) had never been told they have pre-diabetes (see online supplementary table S1). The age, gender, and race of the 82 employees who consented to and completed both surveys were not significantly different from the 523 employees who in the screening had an FBG of 100 to 125 mg/dL.

Quantitative analyses

Approximately 3 months after the screening, 53.7% of participants were engaged in at least one strategy to prevent type 2 diabetes that had been recommended to them by a health coach at the screening (figure 1). Most participants (79.3%) were trying to lose weight, and most (57.3%) reported they had gotten at least 150 min of at least moderate physical activity in the last week. Fewer than half (45.1%) reported they were trying to both lose weight and had gotten at least 150 min of at least moderate physical activity in the last week. Nearly 1 in 5 (18.3%) had talked with a primary care provider about whether metformin for prevention of type 2 diabetes would be right for them. Few (3.7%) reported having participated in a DPP in the past 30 days.

Request permission

Compared with participants who were not engaged in at least one recommended strategy to prevent type 2 diabetes (ie, ‘non-engagers’), there were no differences in the demographic characteristics of participants who were engaged in at least one recommended strategy (ie, ‘engagers’) (table 1). Compared with non-engagers, engagers reported higher median levels of motivation to prevent type 2 diabetes (9 vs 7 (on a scale of 1–10), p=<0.001), higher median levels of personal importance of avoiding type 2 diabetes (10 vs 9 (on a scale of 1–10), p=0.01), and lower median estimations of their risk for developing type 2 diabetes in the next 3 years (40% vs 60%, p=0.02). Engagers also had higher median levels of autonomous motivation (6.3 vs 5.8 (on a scale of 1–7), p=0.02) and controlled motivation (4.4 vs 3.7 (on a scale of 1–7), p=0.02) to prevent type 2 diabetes, as well as higher median patient activation scores (70.2 vs 63.1 (on a scale of 0–100), p=0.002). There were no other statistically significant differences between the engagers and non-engagers in potential mediators of engagement in behaviors to prevent type 2 diabetes such as a previous diagnosis of pre-diabetes, knowledge that type 2 diabetes is preventable, or perceived competence to engage in each recommended preventive behavior.

Qualitative analyses

We conducted semistructured telephone interviews with 22 engagers and 18 non-engagers. From these interviews, key themes emerged about participant-identified facilitators of engagement in behaviors to prevent type 2 diabetes (box 1) and barriers to engagement (box 2).

Among the 22 engagers, the two predominant facilitators of engagement in behaviors to prevent type 2 diabetes, each cited by 14 participants, were the support from a member of their social network (eg, family or friends) and use of external resources or tools, such as rewards and activity trackers. For example, one participant who cited the support of social network members said, “It was helpful to me that I have family members who have experienced the same thing… having both family members that I could talk to that have gone through this…” A participant who used an activity tracker said, “I have a Fitbit that makes it easier cause I like to challenge myself to make sure I get my steps in every day. So, lots of times, I’ll get home in the evening and I’ll see them at 9000 steps and I’ll like, go out and walk up and down the driveway.” Motivation to avoid getting type 2 diabetes was cited by 13 engagers, one of whom noted, “If it hadn't been for the prediabetes I probably would have left it where I was at. So that definitely was a big motivator.” In addition, engagers described barriers they faced while engaging in behaviors to prevent type 2 diabetes: 17 cited competing demands, 11 mentioned periods of low motivation, and 11 identified people in their social network who impeded their efforts.

Among the 18 non-engagers, the predominant barrier to engagement in behaviors to prevent type 2 diabetes, cited by 16 participants, was competing demands such as work or family responsibilities. One participant explained, “The kids get busy in school and activities and…stuff for myself seems to be the first thing that gets cut out.” The next most common barrier, cited by 12 participants, was insufficient motivation to engage in behavior change. One participant who faced this challenge said, “The answer is I need to put myself on a program and stick to it. If I do that, it'll work. I just need to be motivated and take the time and effort…the only problem is me.” Ten participants described lack of resources, such as affordable healthy foods or exercise facilities to support healthy choices, as a key barrier. For example, one participant stated, “I don't understand why it just seems so much harder to buy the healthier foods, the salads…I incorporate it in my diet every day but, you know after a while you have to stretch your money and…sometimes all you can get is lunch meat and bread.” Non-engagers also described facilitators of attempts to engage in behaviors to prevent type 2 diabetes: eight cited external resources or tools (eg, rewards and activity trackers), seven identified supportive people in their social network, and seven mentioned access to key resources like exercise facilities or healthy foods.

Limitations

Our data rely on participant self-report and focus only on engagement in preventive strategies ∼3 months after a workplace screening. Our sample may not be representative of other populations, particularly those with lower incomes or less education. Although the StayWell health coaches invited study participation from all employees found to have pre-diabetes, and key demographic characteristics of study participants and all employees found to have pre-diabetes in the screening were similar, employees who were already engaged in behaviors to prevent type 2 diabetes could have been more likely to participate in the study. Further, because of our study design we were unable to determine whether postscreening engagement in recommended behaviors was a direct result of the screening. We did not inquire about participants' BMI and were unable to measure how successful individuals who were trying to lose weight had been (ie, how much progress they had made towards losing 7% of their body weight as recommended). We measured physical activity through a widely used survey scale, which may be less valid and reliable than objective measures of physical activity. Finally, we were unable to link our survey and interview data to other biometric screening and HRA data that had been collected during the workplace screening.

In conclusion, most employees with pre-diabetes who we surveyed had engaged in at least one recommended strategy to prevent type 2 diabetes ∼3 months after they had been found to have pre-diabetes during a workplace screening. Further, we identified key facilitators of and barriers to engagement in recommended preventive behaviors. More research is needed to understand employees' reactions to and understanding of a pre-diabetes diagnosis, measure longer-term engagement in preventive behaviors among employees with pre-diabetes, and test promising strategies to optimize their ongoing engagement in strategies to delay or prevent the onset of type 2 diabetes.