The role of health apps in glycemic control and dietary-related outcomes in the clinical management of adults with type 2 diabetes

More than 100 million, or 9.4%, of American adults have diabetes, and 95% of these cases are type 2 diabetes (T2DM). Another 84.1 million have prediabetes, which can lead to T2DM within five years of onset (CDC, 2017). Most Americans (81%) own smartphones (Pew 2019), and 58% have downloaded some form of mHealth app (Krebs & Duncan 2015). There are thousands of available apps that are directed towards diabetes, which can assist patients with a myriad of diabetes self-management tasks including tracking blood glucose, calorie and carbohydrate counting, monitoring body weight, reminders for taking medications and health appointments (Martinez-Perez et al., 2013; Chomutare et al., 2011; Rao et al., 2010; Tran et al., 2012). Previous studies have established that the use of diabetes apps in the management of T2DM is associated with significant improvements in glycemic control (Aminuddin et al., 2018; Hou et al., 2018; Kim et al., 2018; Wu et al., 2018). The objectives of this research were to: develop and administer a valid instrument to determine factors associated with app use by clinicians treating those with T2DM; establish the reliability of a questionnaire for use in adults with T2DM; evaluate the viability of conducting an app feasibility study at the Riverside Diabetes Wellness Center (RDWC); test feasibility of a future app intervention in patients with T2DM in a 12-month prospective observational study at the RDWC; and assess glycemic control and dietary-related outcomes of app users vs. nonusers at baseline of the study. Objective one was met through the development, face and content validation, and administration of the Clinician Apps Survey (CAS). The first draft of the survey contained 35 questions, 12 of which involved demographics. Fourteen technology and smartphone-related questions were modified and adopted (Lieffers et al., 2014; Jospe et al., 2015), and to assess behavior change taxonomy, one additional question was added (Michie et al. 2013). The research team developed eight original questions about types of instructional and social media used during patient counseling sessions (n=3), personal use of electronic devices for dietary tracking (n=2), assessment of patients' dietary intake during counseling sessions (n=1), possible reasons for recommending smartphone apps to clients (n=1) and preferred smartphone app features (n= 1). Face and content validity were tested by an expert panel (n = 11), which resulted in 49 changes to the survey. The revised CAS survey contained 37 questions with two open-ended and 35 multiple choice questions and an option of other that allowed participants to type a response; it was transferred into an online survey. Clinicians (n=719) were recruited to complete the CAS through electronic mailing lists within professional dietetics and diabetes membership organizations. There were more app enthusiasts (53%; n=380) compared to non-app users (20%; n=145), another 15% (n=179) that use them personally, but not professionally, and only 2% (n=15) that they use them for work but not personally. Apps enthusiasts were those that use apps both personally and professionally. Whereas 68% recommended pen/paper methods for diet and physical activity monitoring, 62% recommended apps. Most agreed that apps were superior to traditional methods for patients to track dietary intake (62%) and physical activity (58%), make better food choices (34%), lose weight (45%), and track blood glucose (43%). App enthusiasts used the American Association of Diabetes Educators self-care guidelines (p<.001) and advanced counseling techniques (e.g., motivational interviewing) more often than did nonapp users (p<.004). Apps most frequently recommended to clients were MyFitnessPal (n=425), CalorieKing (n=356), and Fitbit (n=312). Objective two was achieved through the validation and reliability of a questionnaire for use in adult patients with T2DM for use in the future Real People with Diabetes (RPWD) longitudinal study. The majority of questions were obtained from the CAS (Karduck and Chapman-Novakofski 2018). First, the questions were transformed for patient use and underwent cognitive interview testing with the research team (n=4). Based on feedback from the research team, questions were reworded to decrease the length and improve the flow of the script. Next, the survey underwent cognitive interviewing testing using the concurrent think-aloud method (Willis 2005) with a convenience sample of adults with diabetes (n=9) from the local community. Interviews were audio transcribed and analyzed using qualitative methodology, resulting in further modifications to the survey resulting in 18 questions. Reliability testing was conducted with another convenience sample of adults (n=20) with T2DM from the community, and participants completed the RPWD survey twice, two weeks apart. All questions within the RPWD questionnaire were found to be consistent and reliable using the test-retest methodology. All paired reliability statistics for all survey questions (paired sample correlations, Kendall’s Tau, Spearman’s Rho, and Intraclass Correlation Coefficients) were significant with a (p<.001). Test-retest reliability was moderate to high (r = 0.92-.99) for all questions (p<.001) with total reliability equal to 0.99 (p< .001). Objective three was fulfilled by evaluation of the viability of conducting a future 12-month mobile health app feasibility study by conducting qualitative interviews of administrators (n=4) and practicing clinicians (n=5) at the RDWC. Two similar interview scripts were developed, one for administrators for individual phone interviews and one for the clinicians for an in-person semi-structured discussion group. The questions for the interview scripts and semi-structured discussion groups were developed based on the Unified Theory of Acceptance and Use of Technology 2 (UTAUT 2) model (Venkatesh, Morris, Davis, & Davis 2003) and our previous work with the CAS (Karduck & Chapman-Novakofski 2018). The scripts were first pilot-tested with an expert panel of participants from a similar outpatient healthcare organization that were known to the research team. Two administrators and two registered dietitian clinicians participated in one-on-one interviews with the researcher from the research team to allow the researcher to build rapport and confidence with the interview script. Based on the feedback from the expert panel, seven changes were made to the administrator’s script, and nine changes were made to the clinician’s script. Interviews were conducted individually with administrators (n=4) and a semi-structured discussion group (n=5) at the RDWC. The discussion groups and interviews were audio-recorded, transcribed verbatim, and a codebook was developed based upon constructs and mechanisms from UTAUT2. Both administrators and clinicians positively regarded health apps with more positive codes (n=2 for Administrators and n=3 for Clinicians) compared to negative ones. The primary constructs from both administrators and clinicians were Performance Expectancy, Effort Expectancy, and Facilitating Conditions. The primary motivators for administrators and clinicians to use health apps in the patient care setting were accessibility, availability, ease of use, convenience, and access to nutrition information. The primary deterrents to using health apps were the age of clients (administrators and clinicians perceive that clients older than 65 would not use health apps), cost of the apps, education about apps by both clinicians and patients, and security of the apps. Objective four was executed by measurement of the feasibility of a future app intervention in patients with T2DM at the RDWC using a mixed-methods approach, including qualitative interviews, clinic, and survey data called the RPWD study. Seven areas of feasibility from the National Institute of Health (NIH) were analyzed, including acceptability, demand, implementation, practicality, adaptation, integration, and limited efficacy testing (Bowen et al., 2009). Adults with T2DM were recruited from the RDWC for the RPWD study, and data were collected at baseline, 3, 6, 9, and 12 months. Acceptability was found to be high; adaptation, integration, and limited efficacy testing had a medium level; but demand, implementation, and practicality had a low level of feasibility. Overall results indicated that at baseline, a future app intervention at the RDWC had a low to medium level of feasibility. Objective five was accomplished in the assessment of glycemic control and dietary-related outcomes of app use at baseline in the RPWD study. Although 140 participants were enrolled in the RWPD study, 121 were included in the baseline analysis, 19 were excluded due to missing data. Most of the participants were smartphone users (72%, n=87). More than half (52.9%, n = 64) tracked some aspect of health in some way (via pen and paper, online, smartphone app, etc.). Blood glucose (69%, n=83) and diet/food intake (51%, n=62) were the most cited aspects of health tracked. Although most participants were smartphone users, using pen and paper was still the most popular method (51%) compared to using a smartphone app (18%) for tracking diet and food intake. A health care provider was the most commonly cited way to learn about a health-related app. Physical activity monitoring food or dietary tracking with nutrition analysis and calorie/carbohydrate calculator, including recommendations, were the most preferred app features. MyFitness Pal™, Fitbit™, and Weight Watchers™ were the most downloaded apps for the small number of app users, although these apps are not designed for diabetes self-management. The mean hemoglobin A1c (A1c) was 7.4%, which is above the recommended goal for adults of ≤ 7% (ADA 2020). Insulin use was the only significant predictor of the variance in A1c. The variation in body mass index (BMI) was explained by age, as age increased BMI decreased. There were no significant relationships discovered between smartphone use, tracking health behaviors, and A1c levels. While smartphone use is prevalent across the US, the actual use of health apps by patients with T2DM for diabetes self-management varies. Low use of health apps was found in the RPWD study. More research is needed to understand better how to educate both clinicians and patients on the available diabetes apps and how using these apps may improve glycemic control. Only then can interventions be developed to measure how apps affect glycemic control and dietary-related behaviors comprehensively.U of I OnlyAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD syste