Adults (aged ≥18 years) living in Edmonton, Alberta, Canada, with T2D and currently enrolled in the Alberta’s Caring for Diabetes (ABCD) cohort study [29] constituted the target population for this study. The ABCD cohort study [29] was designed to explore different aspects of diabetes care and the development of complications among individuals with T2D in Alberta. The ABCD cohort enrolled 2040 participants between 2012 and 2013 from all over the province of Alberta at inception and provided a suitable eligible population from which to draw participants for this intervention. The characteristics of the ABCD cohort have been reported elsewhere [29].

The inclusion criteria were as follows: individuals aged ≥18 years identified as having T2D and currently enrolled in the ABCD cohort study; able to read, understand, and converse in English; and willing to provide informed consent. For practical considerations, we prescreened all cohort participants living in Edmonton for enrollment in the intervention. Based on postal codes, 745 cohort participants who participated in the year 1 ABCD survey lived in Edmonton. However, due to relocation and mortality-related attrition, we invited only 485 ABCD cohort participants living in Edmonton between July 2017 and October 2017 to participate. Those who responded were screened for eligibility and subsequent recruitment into the study between November 2017 and February 2018. Those taking exogenous insulin and having physiological or medical conditions that interfere with usual digestive functions were excluded.

We sent letters explaining details of the study to all eligible participants, asking them to contact the study staff if they were interested in participating in this study. We performed detailed prescreening over the phone to determine full eligibility once we received responses from those invited to participate. A maximum of 2 telephone contacts were made to remind eligible individuals who did not contact the research staff after 2 weeks of expressing their interest in participating in the study. We invited eligible participants to complete baseline anthropometry, biochemical, clinical, and dietary data collection at the Human Nutrition Research Unit located within the Alberta Diabetes Institute at the University of Alberta. A trained dietitian and registered nurse with data collection experience collected anthropometric data and relevant clinical measures using point-of-care instruments (DCA Vantage, Cholestech LDX, and BPTru).

The University of Alberta and Athabasca University Health Research Ethics Boards reviewed and approved the study protocol. In line with research ethics requirements, all participants received adequate information about the study and had the opportunity to ask questions. We obtained written informed consent from participants prior to obtaining any study measurements after providing them an explanation of (1) the purpose of the study; (2) the allocation process; (3) the use of data and the means of assuring confidentiality; (4) voluntary participation and the participants’ right to withdraw from the study at any time; and (5) any potential harm that could occur as a result of the intervention.

Study participants allocated to the usual care (control arm) received standard printed copies of Canada’s Food Guide and Diabetes Canada (formerly Canadian Diabetes Association) GI resources in line with current Diabetes Canada Clinical Practice Guidelines [31,32]. The control group did not receive extra support aimed at increasing knowledge or skills for daily consumption of low-GI foods.

Participants (n=33) allocated to the enhanced low-GI education intervention group received the same information as the control group in addition to vetted, evidence-based, learner-centered, low-cost, and actionable low-GI messages delivered through a Web-based platform with chat rooms, customized videos featuring a registered dietitian, and print material. Based on individual preference and needs, we provided additional support through email, SMS text messages, phone calls, or mail to reinforce participants’ learning. The GI concept and content of this intervention were in line with current Diabetes Canada Clinical Practice Guidelines as well as T2D patients’ suggested content and preferred modes of learning GI information [16,25]. The intervention website was managed by a trained research assistant who also moderated chat room discussions under the supervision of coinvestigators: KS, a registered dietitian and researcher, and STJ and JAJ, who also possess extensive research experience.

Participants received brief tutorials on website log-in, navigation, and usage during baseline data collection, which was reinforced at first log-in with a short video introduction to the program. The video emphasized the importance of low-GI eating and summarized the various aspects of the intervention. Those allocated to the intervention group covered a total of 6 modules over 12 weeks. These modules were aimed at enhancing knowledge and skills for improved GI dietary behavior change. Each module included customized videos featuring a registered dietitian, links to reliable websites, chat rooms for social support, and quizzes. Evidence-based GI content included GI values of foods; low-GI shopping, recipes, and cooking tips; and advice for eating out. Participants received a new module every 2 weeks. Specific topics covered in these modules included (1) general healthy eating for diabetes patients; (2) summary of the GI concept; (3) identifying, choosing, and shopping low-GI foods; (4) low-GI recipes, menus, and meal planning; (5) guidelines for eating out and snacking; and (6) GI concept and general diabetes self-management and healthy lifestyles. Participants were encouraged to outline and track personal, easily achievable goals that could enhance their GI knowledge and skills under each module. For example, in module 3, a participant could set a simple goal to learn how to identify low-GI versions of foods that he or she usually consumed. We delivered each module through the intervention website using user-friendly text, graphical displays, and module summary videos. All videos were developed in line with Canada’s Food Guide and Diabetes Canada Nutrition Therapy Clinical Practice Guidelines-based GI recommendations [31,32] to teach participants “hands-on” application of GI to daily meal planning.

To sustain enthusiasm, participants were granted access to subsequent modules at the end of the preceding module on the first day of each 2-week cycle; this was accompanied with electronic reminders delivered through email or SMS text message based on participants’ choosing. Participants responded to short quizzes meant to bolster key GI principles and lessons learned. Chat rooms were activated for each module to enable participants to share experiences in the form of success stories, challenges, and tips that enhance a sense of community for social support among participants. The chat room forum was monitored, and timely responses were provided to questions and concerns of participants. In addition, we provided weblinks to the international tables of GI and GL values of foods [33] and additional evidence-based information on GI concept as well as general diabetes self-management and healthy lifestyles. A review of similar Web-based studies has shown that interventions that provide interactive elements such as the following have been effective at generating and sustaining participants’ interest and exposure to Web-based interventions: (1) quizzes, searchable database, audio or video; (2) counselor support through counselor-led chat sessions, email, or phone contacts; (3) peer support through Web-based discussion forums or chats; and (4) regular updates of information on intervention websites [34].

Participants in the enhanced GI education arm also received copies of “The Shopper’s Guide to GI Values: the Authoritative Source of Glycemic Index Values for More Than 1,200 Foods” [35] in line with T2D patients’ preference for print-based material as a source GI information [25]. Briefly, the Shopper’s Guide, which was recommended to participants seeking to know more about GI in a previous study [36], is a lightweight, handy book coauthored by expert GI research scientists. It contains GI values of over 1200 foods arranged by categories to help identify healthier low-GI carbohydrate alternatives using handy household measures. The Shopper’s Guide is updated regularly and has comprehensive data on carbohydrates per serving and GL, a shopping list of low-GI essentials, ideas for gluten-free meals, facts about sugar and sweeteners, and tips for everyday meals and dining out. Furthermore, the Shopper’s Guide provides links to supplementary resources with reliable, evidence-based GI information [35].

In addition to the website and the Shoppers Guide [35], participants in the intervention arm were offered periodic emails, SMS text messages or telephone calls, or postal mail prompts to visit the website and/or use the print materials to acquire more GI knowledge as per individual preference. Participants were encouraged to use these mediums to seek assistance regarding specific personal dietary issues, which they may not want to post in the chat room discussion section of the website.

Our primary outcome measure was GI-related dietary behavior change and intake, measured using a 3-day food record.

Daily dietary intake was assessed for all participants at baseline and at 3 months using a 3-day food record. The 3-day food records are valid and reliable for capturing dietary behavior change by asking participants to record their food consumption as they eat [37]. All participants were asked to record, in as much detail as possible, descriptions of foods and beverages consumed over a 3-day period (ie, 2 weekdays and 1 weekend day). Participants were given further instructions on how to fill out the 3-day food record. Color photographs were provided to assist participants with estimating and recording appropriate portion sizes of foods and beverages they consumed in the 3-day food record logbooks. Pictures showing sample portions sizes of foods measured against items including a finger, palm of a hand, and a hockey puck were included and participants were encouraged to choose photographs that best represented their portion sizes or specify whether they consumed more or less. Mean daily food consumption and nutrient intake were estimated using the Food Processor Diet Analysis and Fitness Software (ESHA Research) at baseline and 3 months using the Canadian nutrient file.

All carbohydrate-containing foods identified from the 3-day food records were assigned GI values corresponding to the best geographic and botanical matches in the published International Table of Glycemic Index and Glycemic Load Values [33,38] or the updated University of Sydney Web-based database [39,40]. GI values were averaged for foods having more than one GI value from very similar matches. As the International Table and the University of Sydney Web-based databases do not provide an exhaustive entry of glycemic data for every food, the instances where foods could not be matched directly to those in the International Tables or Web-based database, GI values were calculated from the estimated GL [39] or matched to listed foods with similar characteristics (ingredients, composition, and physical properties) based on all information available to HMA, a trained dietitian, and from his subjective experience and knowledge of foods [41-43]. As recommended [41,43,44], daily average GI and GL was calculated for all carbohydrate-containing foods identified from the 3-day food records using published international GI tables [33,38] (Figure 1).

Secondary outcome measures, including GI knowledge and skills, self-efficacy, body mass index (BMI; weight and height), waist circumference, clinical measures (glycated hemoglobin [HbA_1c], systolic blood pressure, total cholesterol, and high-density lipoprotein), and computer proficiency were assessed at the baseline and at 3 months after completing the intervention. In addition, demographic data were collected.

Pre- and postintervention GI concept knowledge and self-efficacy were assessed and quantified using the Glycemic Index Foods Quiz from a previous study [45]. Dietary data from a previous intervention within the same population showed that, out of 196 participants, 16% were not familiar with low-GI eating and 28% did not include low-GI foods in their diets [46]. About 35% (70/199) indicated that they did not know about GI, and of those who claimed the knowledge of GI, only 34% reported choosing low-GI foods for >6 months in another study [47]. These corroborate previous findings in which only 38% of people with diabetes received nutrition therapy across Canada [28] with <40% of dietitians including the GI concept in T2D dietary self-care counseling [26,27]. Overall dietary self-care behavior was assessed using dietary items in the validated and widely used Summary of Diabetes Self-care Activities measure [48]. The Glycemic Index Foods Quiz, therefore, enabled assessment of the net change in GI knowledge and self-efficacy due to the intervention.

Clinical outcome measures included HbA_1c, systolic blood pressure, and lipid profile. Capillary blood samples (35 μL) were collected from participants to assess HbA_1c value using previously validated point-of-care testing device for HbA_1c (DCA Vantage) [49] and lipid profile (Cholestech LDX). Systolic blood pressure was measured according to the standard protocols using (BPTru) [50].

In addition, weight, height, and waist circumference were assessed according to the Canadian Physical Activity, Fitness and Lifestyle Appraisal procedures [51]. Body weight in kilograms (kg) and height in meters (m) were measured for each subject in light clothing and with no shoes on. Body weight was measured to the nearest 0.1 kg with a portable digital scale (Tanita BWB-800S, Arlington Heights, IL, USA), and height was measured using a portable stadiometer (Tanita HR-100). Waist circumference was measured to 1 mm at the top of the iliac crest using a spring-loaded Gulick anthropometric tape (FitSystems Inc, Calgary, AB, Canada). Regular, monthly quality assurance checks were conducted on the point-of-care devices and scale.

Self-reported physical activity was assessed using the Godin Leisure Time Exercise Questionnaire [52]; the validity of this questionnaire is well established [53], and data suggest that self-reported physical activity estimates function as a suitable predictor of future behavior [54]. Participants were asked to report the frequency and duration of light-, moderate-, and vigorous-intensity leisure-time physical activity that lasted at least 10 minutes over a typical week during the past month. The number of weekly minutes for each intensity level was calculated by multiplying the frequency of activity by the duration in minutes. The sum of weekly minutes of moderate-to-vigorous physical activity (MVPA) gave the total MVPA minutes per week.

Participants’ computer proficiency was measured using the Computer Proficiency Questionnaire (CPQ) at baseline and 3-month postintervention. The CPQ was developed for evaluating the competencies of seniors with regards to use of computers and associated applications such as the internet [55]. The CPQ assesses competence across 6 different subscales: computer basics, printing, communication, internet, scheduling software (calendar), and multimedia use (entertainment) for gauging an individual’s specific and overall computer proficiency.

Demographic information including age, sex, marital status, education, employment status, income, and personal history of cardiovascular disease risk factors (eg, smoking) and time since T2D diagnosis were collected at baseline using a questionnaire.

Preference and usefulness of the Web-based, print [35], email, SMS text messages or telephone call, and postal mail were assessed by asking participants how many times they visited the webpage, read and made references to Shopper’s Guide [35], and how much time they spent on the website or reading the book. In addition, participants were asked which medium they found most helpful and whether the information about the GI concept was informative and helped increase their knowledge and self-efficacy for consuming low-GI foods. Website data measurement programs were built into the website design to compile data points as connections occur with the target audience [56]. Regularly collecting, tracking, and using measurement data makes it possible to understand participants’ characteristics and helps keep the intervention appealing and relevant for achieving the greatest effect [56].

Change in the mean daily GI of dietary carbohydrates from baseline to 3 months will be used as our primary effectiveness measure for improved low-GI knowledge and application. Descriptive statistics will be computed for all variables to determine the nature of the data and to test for normality assumptions. Changes in outcomes will be assessed using repeated-measures 2-way analysis of covariance. Potential sociodemographic and clinical factors associated with enhanced GI learning will be evaluated using generalized linear mixed-model analysis. Treatment condition, baseline scores, participant characteristics (eg, sex, education, and income), and computer proficiency that may be significantly related to outcomes will be controlled for. All data will be analyzed using Stata SE 12.1 (StataCorp).

Based on previous studies regarding the efficacy of GI-based nutritional education and glycemic control [14,45] and meta-analysis of studies on low-GI diets and diabetes management [8], an estimated effect size of d=1 was set for this intervention. Previous data [46] suggest an SD of GI intake of 4-5 units; thus, an effect size d=1 could be achieved with an absolute mean difference of 5 units of GI intake. Given the estimated effect size, 42 participants (21 per arm) would be sufficient for detecting an absolute mean difference of 5 units on the GI intake scale between means with an error of alpha=.05 (2-sided) and beta=.1 (power 1−β=.90; Multimedia Appendix 1); this difference is considered to have significant health benefits from a previous study in which a change of 15 GI units yielded a corresponding HbA_1c change of −1.5% [45] and another study in which a change of 4.6 GI units yielded an HbA_1c change of −0.25 (95% CI −0.50 to −0.004) [14]. With an estimated attrition rate of 30% (based on the ABCD cohort year 2 participation rate), eligible participants were oversampled (N=67) during recruitment to account for possible loss to follow-up during randomization and intervention periods. This sample size was feasible in view of the dietary assessment (3-day food record) method used to assess food intakes and change in dietary GI, the cost of biochemical and clinical measurements, and the duration of the study.