Maintenance Notice

Due to necessary scheduled maintenance, the JMIR Publications website will be unavailable from Wednesday, July 01, 2020 at 8:00 PM to 10:00 PM EST. We apologize in advance for any inconvenience this may cause you.

Who will be affected?

Advertisement

Citing this Article

Right click to copy or hit: ctrl+c (cmd+c on mac)

Published on 02.09.20 in Vol 4, No 9 (2020): September

Preprints (earlier versions) of this paper are available at http://preprints.jmir.org/preprint/17785, first published Feb 10, 2020.

This paper is in the following e-collection/theme issue:

    Original Paper

    Guiding Glucose Management Discussions Among Adults With Type 2 Diabetes in General Practice: Development and Pretesting of a Clinical Decision Support Tool Prototype Embedded in an Electronic Medical Record

    1Department of General Practice, University of Melbourne, Melbourne, Victoria, Australia

    2School of Psychology, Deakin University, Geelong, Victoria, Australia

    3Australian Centre for Behavioural Research in Diabetes, Diabetes Victoria, Melbourne, Australia

    4Department of Medicine, St Vincent's Hospital, University of Melbourne, Melbourne, Australia

    Corresponding Author:

    John Furler, MBBS, PhD

    Department of General Practice

    University of Melbourne

    Melbourne, Victoria

    Australia

    Phone: 61 3 8344 7276

    Email: j.furler@unimelb.edu.au


    ABSTRACT

    Background: Managing type 2 diabetes (T2D) requires progressive lifestyle changes and, sometimes, pharmacological treatment intensification. General practitioners (GPs) are integral to this process but can find pharmacological treatment intensification challenging because of the complexity of continually emerging treatment options.

    Objective: This study aimed to use a co-design method to develop and pretest a clinical decision support (CDS) tool prototype (GlycASSIST) embedded within an electronic medical record, which uses evidence-based guidelines to provide GPs and people with T2D with recommendations for setting glycated hemoglobin (HbA1c) targets and intensifying treatment together in real time in consultations.

    Methods: The literature on T2D-related CDS tools informed the initial GlycASSIST design. A two-part co-design method was then used. Initial feedback was sought via interviews and focus groups with clinicians (4 GPs, 5 endocrinologists, and 3 diabetes educators) and 6 people with T2D. Following refinements, 8 GPs participated in mock consultations in which they had access to GlycASSIST. Six people with T2D viewed a similar mock consultation. Participants provided feedback on the functionality of GlycASSIST and its role in supporting shared decision making (SDM) and treatment intensification.

    Results: Clinicians and people with T2D believed that GlycASSIST could support SDM (although this was not always observed in the mock consultations) and individualized treatment intensification. They recommended that GlycASSIST includes less information while maintaining relevance and credibility and using graphs and colors to enhance visual appeal. Maintaining clinical autonomy was important to GPs, as they wanted the capacity to override GlycASSIST’s recommendations when appropriate. Clinicians requested easier screen navigation and greater prescribing guidance and capabilities.

    Conclusions: GlycASSIST was perceived to achieve its purpose of facilitating treatment intensification and was acceptable to people with T2D and GPs. The GlycASSIST prototype is being refined based on these findings to prepare for quantitative evaluation.

    JMIR Form Res 2020;4(9):e17785

    doi:10.2196/17785

    KEYWORDS



    Introduction

    Type 2 diabetes (T2D) affects more than 420 million people worldwide [1]. In Australia, T2D affects 1.2 million people, amounting to more than Aus $6 (US $4.3) billion annually in direct and indirect health care costs [2]. Early achievement and maintenance of glycated hemoglobin (HbA1c) in an appropriate target range reduces downstream complications and all-cause mortality [3]. Although the general target is HbA1c of ≤7% (53 mmol/mol), the Australian Diabetes Society and international guidelines suggest that targets need to be individualized based on several factors, including age, duration of diabetes, comorbidities, and risk of hypoglycemia [4-6].

    In Australia, as in many countries, most clinical care of people with T2D is based on general practice or primary care [7]. Treatment intensification by general practitioners (GPs) can help people with diabetes achieve glycemic targets [8,9]. However, more than 40% of Australian adults with T2D have an HbA1c level above the target range [10]. Treatment intensification is typically delayed for 8 to 10 years, while HbA1c remains out of target [11]. Barriers to GPs delivering evidence-based treatment intensification include lack of familiarity with clinical diabetes guidelines and the complex and rapidly changing treatment options for optimizing blood glucose levels (BGLs) [12-14]. Health system factors (eg, competing priorities in busy, reactive primary care settings [15]) and patient-related factors (eg, psychological resistance to insulin initiation [16]) can also play a role.

    Clinical decision support (CDS) tools used in real time can enable GPs and people with diabetes to navigate this complex environment and intensify treatment in an appropriate and timely manner to achieve personalized HbA1c targets. CDS tools and evidence-based electronic support can improve the process of care measures [17-19], the use of guidelines by GPs [20,21], and outcomes such as HbA1c [22-25], without substantially increasing health care expenditure [26]. They can also reduce consultation duration and increase screening (eg, lipids) for complications associated with T2D [27,28].

    Using CDS tools in real time during the consultation may also encourage treatment-specific conversations between GPs and people with diabetes, supporting shared decision making (SDM), an important aspect of quality care. A CDS tool embedded within an electronic medical record (EMR) has the added benefit of automatically extracting and using information present in the EMR to guide personalized clinical care. Until recently, most CDS tools have focused on helping people with diabetes achieve a standardized set of diabetes goals (eg, HbA1c target of <7%) that are not necessarily individualized or person-centered [29]. Recently, CDS tools that encourage individualized diabetes care and are integrated within the EMR have become available [30,31]. There is a plethora of self-management apps for people with T2D, many with a focus on displaying blood glucose data with the option to add data about medications, diet, and exercise. Many of them can be shared with health professionals. However, at the point of clinical care, CDS tool design and efficacy are often inconsistent [20,32-34]. Few CDS tools combine the capacity for automatic deployment within the EMR in real time in the consultation with the capacity to make management recommendations (beyond specialized closed-loop insulin systems).

    Utilization of CDS tools by health care professionals (HCPs) is low [32,35,36]. One way to increase uptake and utilization is through co-design that includes end users (eg, GPs and people with diabetes) [37,38]. The co-design theory suggests that technologies, services, and systems should be designed with the intended users, giving them the opportunity to inform development [39]. The focus is on engaging with end users to “jointly articulate ideas ... [and engage] ... with mock-ups and prototypes” [40].

    The aim of this study was to use co-design principles to develop an EMR-based CDS tool for real-time use in consultations to support GPs and people with diabetes to select individualized HbA1c targets and appropriate medication options together. In this paper, we describe the co-design cycles of feedback and tool refinement, drawing on first- and second-stage interviews and focus groups.


    Methods

    The University of Melbourne Human Research Ethics Committee provided ethical approval for both stages (stage 1: 1851169 and stage 2: 1851535).

    Development of the GlycASSIST

    We used an iterative two-stage co-design and refinement process to develop our CDS tool, GlycASSIST (Multimedia Appendix 1). This work was led by a multidisciplinary working group of academics and clinicians (with expertise in general practice, endocrinology, diabetes education, health psychology, behavioral science, and health informatics), who met monthly to critically review the design and refinement of GlycASSIST. A literature review identified T2D-related CDS tools currently available to GPs and summarized local guidelines [6,41] into evidence-based algorithms for personalizing HbA1c targets (Multimedia Appendix 2), informing the design of the initial prototype before the first co-design stage (Figure 1). Algorithms also reflected cost to the patient via the Pharmaceutical Benefits Scheme (PBS). The PBS is a national system for providing medications at an affordable price to patients who meet specific prescribing criteria relating to current glycemic levels, previous medication use, intolerance, and contraindications to other medications. Given the number of medications available, it can be complex and time consuming for clinicians to navigate.

    Figure 1. Initial prototype—page 1. ACR: albumin/creatinine ratio; BP: blood pressure; eGFR: estimated glomerular filtration rate; HbA1c: glycated hemoglobin; HDL: high-density lipoprotein; LDL: low-density lipoprotein; Total chol: total cholesterol.
    View this figure

    We used interviews and focus groups to engage clinicians and people with diabetes in the co-design process. Although each data collection method has particular strengths, we used both to accommodate timing and venue preferences for participants. Stage 1 and 2 participants were recruited through networks of the working group, advertisements on social media (eg, Twitter), and professional association electronic communications. GPs and people with diabetes participated in both stages. Endocrinologists and diabetes educators were engaged in stage 1. The study was conducted in 2018.

    Stage 1

    Stage 1 interviews and focus groups were completed by trained facilitators (HM, ET, and BK) either in person or via web-based video chat. Clinicians were eligible to participate if they were a GP, an endocrinologist (with a case mix consisting of ≥50% people with T2D), or a diabetes educator. People with diabetes were eligible to participate in a separate focus group if they were aged 18 years or older, had T2D diagnosed for at least one year, and were taking glycemic medications. In-depth semistructured interviews and focus groups were designed, first, to identify the experiences and expectations of discussions about HbA1c treatment targets and treatment decisions between clinicians and people with diabetes and, second, to gain feedback on the features and appearance of the initial GlycASSIST design (see Multimedia Appendices 3 and 4 for interview and focus group guides). Participants were shown the initial prototype (Figure 1 and Multimedia Appendix 5) as part of the interview and focus group and HCPs were also shown existing CDS tools [28,30,31]. The GlycASSIST prototype was further developed based on the feedback obtained.

    Stage 2

    GP-Simulated Consultations and Interviews

    Stage 2 involved simulated clinical consultations and think-aloud interviews to understand how GPs and people with diabetes interacted with the second prototype of GlycASSIST (Multimedia Appendices 6-11) and how it could be used to facilitate SDM and treatment intensification and to gain further feedback on the design. The prototype consisted of 2 main elements: an HbA1c calculator to inform personalized HbA1c targets and a medication intensification tool.

    Stage 2 sessions with GPs were all facilitated by a GP (HM) and software programmer (SL) in person in a mock/simulation clinical environment. GPs were eligible to participate if they had not participated in stage 1. The GPs used a desktop computer setup with EMR software that directly interacted with GlycASSIST, allowing GlycASSIST to automatically extract information from the EMR and use it to make recommendations. For GP participants in stage 2, an initial briefing on the features of GlycASSIST was followed by testing in a simulated consultation, with a diabetes educator playing the part of a person with diabetes (a simulated patient named Maureen). The GP was asked to interact with the simulated patient as they would any person with diabetes seeking care. The simulated patient had basic knowledge of T2D management and played Maureen, using a script (Textbox 1) to respond to the anticipated questions asked by the GP and to improvise in response to unexpected questions. The simulated patient’s characteristics, contained within the script as well as additional clinical history, were uploaded to the EMR before the testing session to enable the GP to access clinical data as they might do routinely in a consultation and for GlycASSIST to extract and present Maureen’s information to the GP. Following the simulated consultation, a semistructured interview was conducted (see Multimedia Appendix 12 for the guide) to follow up on comments made during the simulation and to identify feedback specific to the use of GlycASSIST and its role in treatment intensification and SDM. Clinicians were encouraged to verbalize their thoughts, including barriers and facilitators using GlycASSIST and overall usability issues [42]. GPs were asked to explain to the facilitator why and how they interacted with GlycASSIST if they decided to use it.


    Textbox 1. Summary of Maureen, a person with diabetes profiled in the clinical vignette used during the simulation of a clinical consultation in stage 2.
    View this box
    Focus Groups With People With T2D

    People with T2D were eligible to participate in 1 of the 2 focus groups based on the same criteria as stage 1 (some from stage 1 participated in stage 2). Each focus group (facilitated by HM and BK) involved participants viewing a video of a simulated consultation between a GP (HM) and Maureen, in which the GP used GlycASSIST with the person with diabetes to discuss an appropriate HbA1c target and treatment plan. Participants were asked to comment on GlycASSIST and how it was used during the simulated consultation, including what they liked/disliked and suggest improvements (see Multimedia Appendix 13 for the guide).

    Data Analysis

    All sessions were audio-recorded, transcribed verbatim, and checked for accuracy before importing to NVivo 10 for analysis (version 2012, QSR International Pty Ltd). Two investigators (BK and JF) completed an inductive thematic analysis to identify emerging themes across participant groups. Disagreements were resolved by consensus. Themes and illustrative quotes were reviewed by the entire study team.


    Results

    Stage 1

    In total, 12 clinicians (4 GPs, 5 endocrinologists, and 3 diabetes educators) and 6 people with diabetes (3 males and 3 females; 4 people aged <65 years, and 4 people with diabetes duration >10 years) participated in stage 1 focus groups (range 96-112 min) or interviews (range 39-92 min). Six themes were identified around key characteristics participants valued in such a tool or would like to see changed or enhanced in the next iteration of the prototype (Table 1).

    Table 1. Themes identified in stage 1 with exemplar quotes from health professionals and people with diabetes.
    View this table
    Balancing Information Needs—Relevance Is Key

    Clinicians wanted less information on the screen to avoid overcrowding and to reduce the number of clicks needed to progress to different GlycASSIST screens. Information retained on the screen needs to be relevant for T2D management. In contrast, people with diabetes wanted GlycASSIST to display a greater range of information (eg, weight, current medications, and comorbidities) to provide the GP with as much information as possible.

    The presentation of all possible drug classes (n=8) was deemed important by clinicians and people with diabetes, but the inclusion of all side effects for each drug class was considered inappropriate by several clinicians, as some side effects are considered more common and/or serious than others. Instead, clinicians suggested listing only the most common and serious side effects. This was also supported by people with diabetes who felt it was important that medication information (both positive and negative) could be used by the GP during the discussion, where appropriate weighting could be given to each side effect.

    Both clinicians and people with diabetes reported that prompts to establish a chronic disease management plan or refer to an appropriate specialist (eg, endocrinologist, dietician) would facilitate holistic and individualized care.

    Credibility

    Clinicians considered evidence-based recommendations critical to building trust in GlycASSIST. It was also important that the evidence was clear and updated as evidence changes. People with diabetes wanted updated information about new therapies.

    Using GlycASSIST to Reduce Prescribing Complexity

    Clinicians reported that GlycASSIST could be a useful tool to reduce the complexity associated with choosing diabetes glycemic medications for people with diabetes. Clinicians, especially GPs, indicated that GlycASSIST would be more useful if it included guidance on how to prescribe a medication that is subsidized by the Australian Government via the PBS. Having access to these complex subsidy regulations within GlycASSIST would make it easier to choose a medication affordable for the person with diabetes while also enabling the person with diabetes to choose a non–PBS-listed medication (ie, as a private nonsubsidized prescription) if they preferred. People with diabetes also saw the need for the tool to address complexity in the range of medication options available and in choosing based on side effects and cost.

    The Clinician, Not GlycASSIST, Retaining Clinical Autonomy

    It was important for GPs to make GlycASSIST work for them. For some, this meant having GlycASSIST directly generate a medication prescription, rather than only making recommendations. Others wanted GlycASSIST to perform time-saving tasks, such as entering clinical notes into the EMR, whereas for others, the autonomy to ignore or override HbA1c targets or medications recommended by GlycASSIST was a priority. For people with diabetes, this was seen in how they wanted the tool to be something used by their doctor to personalize explanations and discussions, for example, in relation to medication contraindications or side effects. Several GPs indicated that it was important that GlycASSIST did not pop up on the screen, preferring a discrete icon that did not demand the GP’s attention and could be minimized if desired.

    Not Just About Medications

    Both clinicians and people with diabetes reported that including lifestyle modifications (eg, regular exercise) is important for holistic care of T2D. Therefore, several participants suggested having a lifestyle assessment prompt in GlycASSIST.

    GlycASSIST Prototype Refinement

    Not all the findings from stage 1 could be accommodated in the second iteration of the tool. Several features were maintained and strengthened in the refined second prototype (Multimedia Appendices 6-11). Evidence-based recommendations and associated algorithms were enhanced by adding more text demonstrating the latest evidence in hover over boxes and including an additional algorithm related to PBS prescribing rules. Formatting features, such as color coding and not having GlycASSIST pop up and consume too much space on the screen, were also maintained.

    Enabling clinicians to retain autonomy in their use of the tool was maintained by ensuring clinicians could continue to override recommendations, select medication classes as they deemed appropriate (ie, whether PBS is listed or not), and modify the auto-populated fields. However, it was not possible for GlycASSIST to have the capability to populate the EMR with clinical notes or generate prescriptions (rather than returning manually to the EMR and prescribing from there). These features would be possible through collaboration with EMR software vendors, but this depth of integration was not possible in this pretesting study.

    The main changes made to the GlycASSIST prototype focused on information presentation and trading broadness for specificity. All changes led to a more specific focus purely on glycemia and medications (Multimedia Appendices 6-11), with removal of extraneous information, such as reference to other cardiovascular disease (CVD) risk factors, referrals, and patient preferences. For example, questions that focused on the acceptability of weight gain and injectable treatments for people with diabetes and lifestyle concerns (eg, smoking history) were removed. Yes/no responses to these issues potentially oversimplified important topics for clinical discussion. Information about medication effects, side effects, and administration were retained within the medication summary boxes (drawing on local evidence-based resources [43] to facilitate discussion). Focusing only on glycemic medications also simplified the tool, which was valued by both clinicians and people with diabetes. Finally, some changes were made to enhance the visual appeal of the tool. Clinicians appreciated the use of color coding to indicate contrasting results and to draw their attention to important information. However, the size of the font used on the screen was too small in some places.

    GlycASSIST was given the ability to display the availability of PBS cost-subsidy for each recommended medication class and their associated medications. In addition, a link to the Australian Government’s PBS webpage was provided [44].

    Stage 2

    Eight GPs (4 females, 5 trained in Australia and 5 in practice for >10 years) participated in the stage 2 computer simulations (range 35-56 min, including consultations and follow-up interviews, but not including the initial briefing). Six people with diabetes (4 women, 3 aged >65 years and 2 with T2D >10 years) participated in 1 of the 2 focus groups (range 78-83 min). Four themes were identified during the simulated consultations with GPs, debriefing interviews with GPs, and focus groups with people with diabetes (Table 2).

    Table 2. Themes identified in stage 2 and exemplar quotes from general practitioners and people with diabetes.
    View this table
    Using GlycASSIST to Support SDM

    Clinicians reported that the second prototype of GlycASSIST could help them intensify treatment in collaboration with people with diabetes. They appreciated the advice about PBS availability for each recommended medication class and option. This helped GPs “comply with the [PBS] regulation” [GP1] while helping people with diabetes by ensuring that prescribed medications were affordable to them. It saved time by avoiding the need to read through the information-dense material on the government PBS website or within the EMR prescribing software.

    Clinicians and people with diabetes perceived the HbA1c calculator as useful for prompting collaborative conversations around appropriate HbA1c targets, although several GPs were surprised by the lower recommended HbA1c target of 6.5% and overrode it, independently suggesting that 7% was a more appropriate target. One GP cited concerns about hypoglycemia and clinician autonomy when overriding the target.

    Some GPs appreciated how several individualized drug class recommendations were made by GlycASSIST, providing people with diabetes with a choice. However, some GPs wanted specific recommendations about the single most appropriate class for the person with diabetes, rather than being presented with a variety of clinically appropriate possibilities.

    Several GPs suggested that although GlycASSIST could facilitate SDM around medication choice, they would be less inclined to discuss all options if they had already decided what medication they would prescribe. Rarely did the GP ask the simulated patient if they wanted to consider changes to their current medication management before presenting options using GlycASSIST. Furthermore, GPs rarely discussed all recommended medication classes with the person. Instead, most chose 1 or 2 classes that they deemed appropriate and presented these to the person to choose the final option based on their own preferences.

    Experienced GPs suggested that GlycASSIST would be more useful for less-experienced GPs (eg, GP registrars) when choosing appropriate medication options for people with diabetes, despite stating that they themselves are not always aware of new medications and can forget existing ones.

    People with diabetes mentioned that they would like to see and discuss the GP’s screen as they used GlycASSIST, potentially facilitating SDM. However, in the computer simulation, several GPs did not turn their screen to show the recommendations or information provided by GlycASSIST to the simulated patient. Some GPs and people with diabetes reported that the information presented by GlycASSIST might overwhelm the person (eg, by using medical jargon), and 1 GP mentioned that it would be physically impossible for them to show their screen to the person because of their office layout. One GP suggested that SDM was not something that they routinely engaged in, preferring to make clinical decisions themselves, without involving the person with diabetes.

    GlycASSIST Features

    Overall, the second prototype of the tool was mostly well received. Clinicians often reported that GlycASSIST was easy to use, having the potential to shorten consultations and avoiding the need to seek out guidelines midconsultation. However, more features were requested.

    Several GPs and people with diabetes suggested adding features to the screen presenting recommended medication classes. GPs liked the tabular display of medication classes, but they suggested that more information was needed, such as more detailed descriptions of each class (eg, listing both the generic and brand names for medications belonging to each class, their dose, frequency, and delivery mechanism). Some GPs requested side effects to be explicitly categorized into rare and common.

    Clinicians often recommended to the simulated patient that they may benefit from seeing a diabetes educator and other allied health professionals as well as adopting and maintaining lifestyle changes. This suggests that using GlycASSIST did not dominate the consultation and still allowed the GP to address other important elements of T2D management. Some GPs requested additional features to further support them to deliver this care, for example, an integrated link to a people with diabetes friendly handout, including information appropriate for people with low health literacy. Several GPs suggested that GlycASSIST could present CVD risk factors (eg, lipid profile) and allow for CVD risk calculation. Finally, some GPs recommended having prompts to refer to allied health professionals. This was also suggested by people with diabetes as well as a prompt for the GP to talk with them about lifestyle modifications as part of their T2D management.

    Visual Appeal and Information Presentation

    Several GPs and people with diabetes reported that they liked how information was presented on the GlycASSIST screens. Most of the positive feedback was related to the graphical representation of HbA1c results, enabling a trend to appear, and the tabular display of medication class choices. Overall, this layout made it easy to obtain and understand information. The use of color was also reported as appealing, with 1 GP and some people with diabetes also recommending the use of images, such as needles and syringes, when referring to injectable medications.

    Clinicians and people with diabetes indicated that less information on the screen made it easier to interact with GlycASSIST. However, they were aware that a compromise must be met where relevant information remained accessible. Some GPs suggested that additional hover over boxes be used, enabling the GP to see important information (eg, updated evidence for suggestions and considerations for assessing the risk of hypoglycemia).

    Some available information was not clear to some GPs. Some did not notice that the person’s current medications and estimated glomerular filtration rate (a measure of renal function, extracted from the pathology results section of the EMR) were presented on the screen, so they sought that information manually from the EMR. Clinicians and people with diabetes also reported that the text size needed to be larger.

    One GP mentioned that the subtle but obvious appearance of the GlycASSIST icon in the corner of the desktop increased the likelihood that she would click on it, suggesting that easy access and less dominating nature of the tool may be valued.

    The presentation of relevant medication information, such as expected HbA1c reduction, weight loss, and side effects such as nausea and genitourinary infections, was deemed important for SDM by GPs and appeared to be reflective of recommendations made in clinical guidelines.

    Workflow and Navigation

    Most GPs began the simulated consultation by asking the simulated patient general questions (eg, “How’s work going?”), assessing lifestyle factors (eg, “Are you walking every day?”), and asking about their ability to manage their diabetes (eg, “Are you taking your tablets as prescribed?”) before opening GlycASSIST.

    GlycASSIST was usually opened once the GP wanted to talk about the simulated patient’s HbA1c, although several GPs accessed HbA1c data from the investigation panel in the EMR first and then used GlycASSIST to demonstrate the trend in results over time. Clinicians progressed smoothly to the screen presenting the recommended medication classes once they had suggested to the person that their HbA1c could be lowered. Once on this screen, some GPs discussed the classes with which they were most familiar and disregarded those that were unfamiliar.

    Some clinicians progressed from the screen presenting the recommended medication classes to choose a specific medication to prescribe, but others found it difficult to identify a tab to facilitate this. Occasionally, GPs wanted to leave GlycASSIST to use the EMR (eg, to calculate CVD risk) or go back to the previous screen. Most GPs found these switching tasks difficult.


    Discussion

    Principal Findings

    The aim of this study was to co-design and refine a prototype EMR-based CDS tool (GlycASSIST) for real time, in consultation use to support treatment intensification and SDM between GPs and people with diabetes when choosing HbA1c targets and diabetes medications. The co-design process supported an iterative refinement of the tool based on early user experience. The findings from stage 1 interviews and focus groups with clinicians and people with diabetes indicated that GlycASSIST was perceived as useful for T2D management but required modifications to improve information clarity and functionality. Modifications were made to reduce the amount of information on the screens and to add recommendations regarding PBS availability, making the information presented more specific to glycemic medications. Stage 2 findings suggested that these changes were well received by GPs and people with diabetes, but more changes were required. Participants perceived GlycASSIST as useful for SDM and presented well. However, the participants recommended additional features. GPs wanted more information describing medications, added functionality (eg, being able to prescribe medications using GlycASSIST), and better navigation (eg, being able to return easily to a previous screen). People with diabetes wanted additional prompts for discussion of lifestyle, suggesting that such a feature in the tool may help overcome barriers that people with diabetes have raised on these topics.

    Unlike existing CDS tools, GlycASSIST is unique in its specific focus on individualization of both HbA1c targets and medication recommendations, resulting in recommendations tailored to the individual and based on both clinical and personal factors. Few existing CDS tools are EMR-integrated [28,30], and none calculate both an individualized HbA1c target and make individualized medication recommendations on evidence-based, person-centered factors. Integration with the EMR is the strength of GlycASSIST, as it provides enough information and drug-specific advice to facilitate a conversation between the GP and person with diabetes about the most appropriate medication for them.

    GlycASSIST was considered to make treatment intensification easier for both experienced GPs and registrars. This was based on the functionality of GlycASSIST (eg, using evidence-based algorithms to inform the presentation of appropriate medication classes and prescribing rules [45]) and its content (eg, including all T2D medication classes). Both clinicians and people with diabetes also felt that GlycASSIST has the potential to facilitate SDM, prompting discussion around medication side effects and outcome preferences. However, GPs in our study often overrode the HbA1c target suggested by GlycASSIST and primarily discussed medication classes with which they were most familiar. Our findings suggest that although GlycASSIST might provide helpful recommendations, these recommendations do not necessarily inform clinical decisions or collaborative shared approaches to treatment choices. In a study in the United States, although >60% of physicians opened the Diabetes Wizard during a randomized controlled trial (RCT), few were still using it 1 year after incentives (and the intervention period) ended, despite most physicians being satisfied with the tool [30]. Some clinicians found it challenging to make explicit and share the conversation about medication choice that they would usually have in their head. Furthermore, the use of GlycASSIST in practice did not automatically lead to engagement of people with diabetes in SDM. Simply having the information and tools available to GPs may not necessarily lead to SDM if this is not the usual consulting style of a given practitioner. It was beyond the remit of our small pretesting study to include specific training or intervention to enhance SDM. Although GlycASSIST has the potential to save time in consultations and aid clinical recommendations, additional behavior change strategies may be required to facilitate uptake and reduce clinical inertia when GlycASSIST is tested experimentally [46,47].

    GlycASSIST was designed specifically for (and explicitly limited to) 2 tasks: assisting GPs and people with diabetes to choose individualized HbA1c targets and T2D medications. However, this explicit focus exists in tension with the ideal of holistic and integrated diabetes care. Holistic care for people with T2D includes the important role that lifestyle interventions can play in achieving glycemic targets. It also involves linking T2D to the broader clinical issue of CVD. GlycASSIST could, for example, have the capability for automated calculation and display of individualized CVD risk and highlight modifiable risk factors and lifestyle interventions for CVD. It has been recommended that T2D CDS tools include all domains of T2D care (eg, smoking frequency and other CVD risk factors) to facilitate increased and ongoing use [36]. This may generate a practical one-stop shop for GPs to access multimorbidity CDS. Thus, a targeted and focused tool such as GlycASSIST may benefit from integration into a more comprehensive chronic disease management tool, protecting its specificity while being embedded within a holistic approach to care by partnering GlycASSIST with related chronic disease management tools.

    The evidence-based nature of recommendations generated by GlycASSIST was mentioned frequently as beneficial and necessary to facilitate treatment intensification, assuming that GlycASSIST remains up-to-date. The features that allow GlycASSIST to make treatment intensification easier are time, funding, and resource intensive [45]. A tool such as GlycASSIST needs continual updating as new evidence and clinical guidelines become available. Further development and testing of GlycASSIST in an RCT to establish efficacy at improving patient outcomes, SDM, and overcoming treatment inertia is necessary. At this stage, no plans exist for commercialization.

    Strengths and Limitations

    A strength of our study is that it is specific to the local context of the anticipated users (eg, by incorporating PBS rules and only medications approved for use in Australia) and provides person-centered individualized support for both establishing HbA1c targets and choosing appropriate medications with people with diabetes, something not found in existing CDS tools specific to T2D management. With the multiplicity of EMR software providers in Australia and worldwide, integrated CDS tools must be developed locally to be clinically useful, incorporating their own prescribing rules. Co-design was used to enhance the design of GlycASSIST, ensuring that it met the needs of end users: both clinicians and people with diabetes. This process enabled the testing of GlycASSIST with end users to explore their interest in and interaction with GlycASSIST, which is critical for identifying both barriers to and enablers of adoption in real-world clinical settings [38,45].

    One limitation, as discussed earlier, is that GlycASSIST focuses on HbA1c only, although this was a compromise made explicitly and as part of our co-design process. Within that focus on glucose levels, GlycASSIST addressed pharmacological treatment intensification only, not lifestyle interventions. This limitation was a compromise, made consciously based on feedback about avoiding too much information in the tool. Another limitation of the tool is that it does not allow for uploading finger-prick BGL data. Many industry-based and other apps allow for the entry and display of individual BGL data, which could be used alongside GlycASSIST. Our tool focused on automated data extraction from the EMR for use in a consultation. Integration with the many BGL display apps could be an area for future development. Our tool did not include recommendations for deintensifying therapy. Again, this could be an area for future development. In this early formative study, GlycASSIST was developed to integrate with only 1 commercially available EMR software (Best Practice). There are multiple EMR vendors in Australia, and any future widespread implementation of GlycASSIST would require collaboration with the vendors. Our study was a formative evaluation and co-design of a prototype tool, so we did not test GlycASSIST in actual clinical practice. The GPs participating in the stage 2 computer simulations were provided with a comprehensive vignette of the simulated patient (Maureen). The details within the vignette might have inadvertently prompted the GP to make certain assessments, search for investigations, or make certain treatment suggestions (eg, based on CVD history). It is unclear if, and to what extent, this detail influenced the GPs’ interactions with GlycASSIST or their comments about it, especially regarding suitability for clinical practice. Furthermore, the presence of the researcher and the simulated nature of the pilot testing likely impacted the way the GP interacted with the tool. Finally, GPs indicated that they participated because they were interested in GlycASSIST and clearly saw a clinical need for the tool, suggesting that their views may not be representative of the broader clinician community.

    Conclusions

    Our prototype has both face and content validity as well as acceptability and feasibility. Co-design incorporating local context and end user views led to a targeted tool that retained the capacity for integration with broader chronic disease management support. GlycASSIST requires additional refinement and evaluation, possibly as part of a suite of chronic disease management CDS tools, to establish its efficacy and broader acceptability for use in Australian general practice.

    Acknowledgments

    The authors wish to thank the study participants for volunteering their time and insights into this study and to the working group (which included the authors and Associate Professor Mark Kennedy, Professor Sophia Zoungas, and Mr Ken Clarke) who also volunteered their time and expert insights that informed the overall design of the study and GlycASSIST. This work was funded by the Royal Australian College of General Practitioners—Diabetes Australia and by the Networked Society Institute (University of Melbourne).

    Conflicts of Interest

    ET and JS were supported by core funding from the Australian Centre for Behavioural Research in Diabetes (ACBRD), derived from the collaboration between Diabetes Victoria and Deakin University. JN reports grants and travel support from MSD, speaker fees and research support from Sanofi, nonfinancial support from Medtronic, nonfinancial support from Abbott, and research funding from Eli Lilly/Boehringer Ingelheim, outside the submitted work. JS has served on the advisory boards of Janssen, Medtronic, Roche Diabetes Care, and Sanofi Diabetes. Her research group (ACBRD) has received honoraria in respect of this and also received unrestricted educational grants/in-kind support from Abbott Diabetes Care, AstraZeneca, Medtronic, Roche Diabetes Care, and Sanofi Diabetes; sponsorship to attend educational meetings from Medtronic, Roche Diabetes Care, and Sanofi Diabetes; consultancy income and/or speaker fees from Abbott Diabetes Care, AstraZeneca, Medtronic, Novo Nordisk, Roche Diabetes Care, and Sanofi Diabetes. ET has undertaken research funded by an unrestricted educational grant from Abbott Diabetes Care, AstraZeneca, and Sanofi Diabetes; has received speaker fees from Novo Nordisk to ACBRD; and has served on an AstraZeneca advisory board. RA has served as a member of several advisory boards and has conducted paid presentations for pharmaceutical companies, such as AstraZeneca, Sanofi, and Novartis. JF has received unrestricted educational grants for research support from Roche, Sanofi, and Medtronic. DN is on advisory boards to Abbott Diabetes Care and Novo Nordisk and has had various financial relationships with pharmaceutical industries outside the submitted work, including consultancies, grants, lectures, educational activities, and travel. GK has served as a member of several advisory boards and has conducted paid presentations for pharmaceutical companies, including AstraZeneca, Sanofi, Lilly, and Novo Nordisk. DB, SL, and BK have no competing financial interests. The intellectual property in GlycASSIST is retained by the University of Melbourne.

    Multimedia Appendix 1

    Codesign phases.

    PNG File , 754 KB

    Multimedia Appendix 2

    HbA1c algorithm.

    PNG File , 39 KB

    Multimedia Appendix 3

    Stage 1 Interview guide.

    PDF File (Adobe PDF File), 92 KB

    Multimedia Appendix 4

    Stage 1 focus group guide.

    PDF File (Adobe PDF File), 145 KB

    Multimedia Appendix 5

    Initial prototype treatment intensification page.

    PNG File , 44 KB

    Multimedia Appendix 6

    Second prototype - button within the EMR.

    PNG File , 241 KB

    Multimedia Appendix 7

    Second prototype - opening page.

    PNG File , 976 KB

    Multimedia Appendix 8

    Second prototype - treatment options page.

    PNG File , 1070 KB

    Multimedia Appendix 9

    Second prototype - expanded treatment options.

    PNG File , 1276 KB

    Multimedia Appendix 10

    Second prototype - prescribing page.

    PNG File , 177 KB

    Multimedia Appendix 11

    Prototype 2 additional screens 6.

    PNG File , 198 KB

    Multimedia Appendix 12

    Stage 2 GP Interview guide.

    PDF File (Adobe PDF File), 390 KB

    Multimedia Appendix 13

    Stage 2 focus group guide.

    PDF File (Adobe PDF File), 110 KB

    References

    1. World Health Organization. Global Report on Diabetes. Geneva, Switzerland: World Health Organisation; 2016.
    2. Lee CM, Colagiuri R, Magliano DJ, Cameron AJ, Shaw J, Zimmet P, et al. The cost of diabetes in adults in Australia. Diabetes Res Clin Pract 2013 Mar;99(3):385-390. [CrossRef] [Medline]
    3. Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HA. 10-year follow-up of intensive glucose control in type 2 diabetes. N Engl J Med 2008 Oct 9;359(15):1577-1589. [CrossRef]
    4. Inzucchi S, Bergenstal RM, Buse JB, Diamant M, Ferrannini E, Nauck M, et al. Management of hyperglycemia in type 2 diabetes, 2015: a patient-centered approach: update to a position statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 2015 Jan;38(1):140-149. [CrossRef] [Medline]
    5. Davies MJ, D'Alessio DA, Fradkin J, Kernan WN, Mathieu C, Mingrone G, et al. Management of hyperglycemia in type 2 diabetes, 2018. A consensus report by the American diabetes association (ADA) and the European association for the study of diabetes (EASD). Diabetes Care 2018 Dec;41(12):2669-2701 [FREE Full text] [CrossRef] [Medline]
    6. Gunton JE, Cheung NW, Davis TM, Zoungas S, Colagiuri S, Australian Diabetes Society. A new blood glucose management algorithm for type 2 diabetes: a position statement of the Australian Diabetes Society. Med J Aust 2014 Dec 11;201(11):650-653. [CrossRef] [Medline]
    7. Britt H, Miller G, Bayram C. A decade of Australian general practice activity 2006-07 to 2015-16. In: Sydney University Press; 2016:16.
    8. Canivell S, Mata-Cases M, Real J, Franch-Nadal J, Vlacho B, Khunti K, et al. Glycaemic control after treatment intensification in patients with type 2 diabetes uncontrolled on two or more non-insulin antidiabetic drugs in a real-world setting. Diabetes Obes Metab 2019 Jun;21(6):1373-1380. [CrossRef] [Medline]
    9. Manski-Nankervis J, Furler J, O'Neal D, Ginnivan L, Thuraisingam S, Blackberry I. Overcoming clinical inertia in insulin initiation in primary care for patients with type 2 diabetes: 24-month follow-up of the Stepping Up cluster randomised controlled trial. Prim Care Diabetes 2017 Oct;11(5):474-481. [CrossRef] [Medline]
    10. Wan Q, Harris MF, Jayasinghe UW, Flack J, Georgiou A, Penn DL, et al. Quality of diabetes care and coronary heart disease absolute risk in patients with type 2 diabetes mellitus in Australian general practice. Qual Saf Health Care 2006 Apr;15(2):131-135 [FREE Full text] [CrossRef] [Medline]
    11. Khunti K, Gomes MB, Pocock S, Shestakova MV, Pintat S, Fenici P, et al. Therapeutic inertia in the treatment of hyperglycaemia in patients with type 2 diabetes: a systematic review. Diabetes Obes Metab 2018 Feb;20(2):427-437 [FREE Full text] [CrossRef] [Medline]
    12. Furler J, Spitzer O, Young D, Best J. Insulin in general practice - barriers and enablers for timely initiation. Aust Fam Physician 2011 Aug;40(8):617-621 [FREE Full text] [Medline]
    13. Davies MJ, D'Alessio DA, Fradkin J, Kernan WN, Mathieu C, Mingrone G, et al. Management of hyperglycaemia in type 2 diabetes, 2018. A consensus report by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD). Diabetologia 2018 Dec;61(12):2461-2498. [CrossRef] [Medline]
    14. Okemah J, Peng J, Quiñones M. Addressing clinical inertia in type 2 diabetes mellitus: a review. Adv Ther 2018 Nov;35(11):1735-1745 [FREE Full text] [CrossRef] [Medline]
    15. Parchman ML, Pugh JA, Romero RL, Bowers KW. Competing demands or clinical inertia: the case of elevated glycosylated hemoglobin. Ann Fam Med 2007;5(3):196-201 [FREE Full text] [CrossRef] [Medline]
    16. Holmes-Truscott E, Blackberry I, O'Neal DN, Furler JS, Speight J. Willingness to initiate insulin among adults with type 2 diabetes in Australian primary care: results from the Stepping Up study. Diabetes Res Clin Pract 2016 Apr;114:126-135. [CrossRef] [Medline]
    17. Cleveringa FG, Gorter KJ, van den Donk M, van Gijsel J, Rutten GE. Computerized decision support systems in primary care for type 2 diabetes patients only improve patients' outcomes when combined with feedback on performance and case management: a systematic review. Diabetes Technol Ther 2013 Feb;15(2):180-192. [CrossRef] [Medline]
    18. de Belvis A, Pelone F, Biasco A, Ricciardi W, Volpe M. Can primary care professionals' adherence to evidence based medicine tools improve quality of care in type 2 diabetes mellitus? A systematic review. Diabetes Res Clin Pract 2009 Aug;85(2):119-131. [CrossRef] [Medline]
    19. Holbrook A, Thabane L, Keshavjee K, Dolovich L, Bernstein B, Chan D, COMPETE II Investigators. Individualized electronic decision support and reminders to improve diabetes care in the community: COMPETE II randomized trial. Can Med Assoc J 2009 Jul 7;181(1-2):37-44 [FREE Full text] [CrossRef] [Medline]
    20. Dowding D, Randell R, Gardner P, Fitzpatrick G, Dykes P, Favela J, et al. Dashboards for improving patient care: review of the literature. Int J Med Inform 2015 Feb;84(2):87-100. [CrossRef] [Medline]
    21. Nirantharakumar K, Chen Y, Marshall T, Webber J, Coleman J. Clinical decision support systems in the care of inpatients with diabetes in non-critical care setting: systematic review. Diabet Med 2012 Jun;29(6):698-708. [CrossRef] [Medline]
    22. Jeffery R, Iserman E, Haynes RB, CDSS Systematic Review Team. Can computerized clinical decision support systems improve diabetes management? A systematic review and meta-analysis. Diabet Med 2013 Jun;30(6):739-745. [CrossRef] [Medline]
    23. Gunathilake W, Gunawardena S, Fernando R, Thomson G, Fernando D. The impact of a decision support tool linked to an electronic medical record on glycemic control in people with type 2 diabetes. J Diabetes Sci Technol 2013 May 1;7(3):653-659 [FREE Full text] [CrossRef] [Medline]
    24. Alharbi NS, Alsubki N, Jones S, Khunti K, Munro N, de Lusignan S. Impact of information technology-based interventions for type 2 diabetes mellitus on glycemic control: a systematic review and meta-analysis. J Med Internet Res 2016 Nov 25;18(11):e310 [FREE Full text] [CrossRef] [Medline]
    25. Riazi H, Larijani B, Langarizadeh M, Shahmoradi L. Managing diabetes mellitus using information technology: a systematic review. J Diabetes Metab Disord 2015;14:49 [FREE Full text] [CrossRef] [Medline]
    26. Gilmer TP, O'Connor PJ, Sperl-Hillen JM, Rush WA, Johnson PE, Amundson GH, et al. Cost-effectiveness of an electronic medical record based clinical decision support system. Health Serv Res 2012 Dec;47(6):2137-2158 [FREE Full text] [CrossRef] [Medline]
    27. Dagliati A, Sacchi L, Tibollo V, Cogni G, Teliti M, Martinez-Millana A, et al. A dashboard-based system for supporting diabetes care. J Am Med Inform Assoc 2018 May 1;25(5):538-547. [CrossRef] [Medline]
    28. Koopman RJ, Kochendorfer KM, Moore JL, Mehr DR, Wakefield DS, Yadamsuren B, et al. A diabetes dashboard and physician efficiency and accuracy in accessing data needed for high-quality diabetes care. Ann Fam Med 2011 Sep 12;9(5):398-405 [FREE Full text] [CrossRef] [Medline]
    29. Wilkinson MJ, Nathan AG, Huang ES. Personalized decision support in type 2 diabetes mellitus: current evidence and future directions. Curr Diab Rep 2013 Apr;13(2):205-212 [FREE Full text] [CrossRef] [Medline]
    30. O'Connor PJ, Sperl-Hillen JM, Rush WA, Johnson PE, Amundson GH, Asche SE, et al. Impact of electronic health record clinical decision support on diabetes care: a randomized trial. Ann Fam Med 2011;9(1):12-21 [FREE Full text] [CrossRef] [Medline]
    31. Ampudia-Blasco FJ, Benhamou PY, Charpentier G, Consoli A, Diamant M, Gallwitz B, et al. A decision support tool for appropriate glucose-lowering therapy in patients with type 2 diabetes. Diabetes Technol Ther 2015 Mar;17(3):194-202 [FREE Full text] [CrossRef] [Medline]
    32. Ali SM, Giordano R, Lakhani S, Walker DM. A review of randomized controlled trials of medical record powered clinical decision support system to improve quality of diabetes care. Int J Med Inform 2016 Mar;87:91-100. [CrossRef] [Medline]
    33. Crosson JC, Ohman-Strickland PA, Cohen DJ, Clark EC, Crabtree BF. Typical electronic health record use in primary care practices and the quality of diabetes care. Ann Fam Med 2012;10(3):221-227 [FREE Full text] [CrossRef] [Medline]
    34. Moja L, Kwag KH, Lytras T, Bertizzolo L, Brandt L, Pecoraro V, et al. Effectiveness of computerized decision support systems linked to electronic health records: a systematic review and meta-analysis. Am J Public Health 2014 Dec;104(12):e12-e22. [CrossRef] [Medline]
    35. Meigs JB, Cagliero E, Dubey A, Murphy-Sheehy P, Gildesgame C, Chueh H, et al. A controlled trial of web-based diabetes disease management: the MGH diabetes primary care improvement project. Diabetes Care 2003 Mar;26(3):750-757. [CrossRef] [Medline]
    36. O'Connor PJ, Sperl-Hillen JM, Fazio CJ, Averbeck BM, Rank BH, Margolis KL. Outpatient diabetes clinical decision support: current status and future directions. Diabet Med 2016 Jun;33(6):734-741 [FREE Full text] [CrossRef] [Medline]
    37. Tomlin E. Patients at the Centre of Design to Improve the Quality of Care; Exploring the Experience-based Co-Design Approach Within the NHS. White Rose University Consortium. 2018.   URL: http://etheses.whiterose.ac.uk/22180/1/Tomlin_EM_Psychology_PhD_2018.PDF [accessed 2020-08-12]
    38. Garg AX, Adhikari NK, McDonald H, Rosas-Arellano MP, Devereaux PJ, Beyene J, et al. Effects of computerized clinical decision support systems on practitioner performance and patient outcomes: a systematic review. J Am Med Assoc 2005 Mar 9;293(10):1223-1238. [CrossRef] [Medline]
    39. Mulvale G, Moll S, Miatello A, Murray-Leung L, Rogerson K, Sassi RB. Co-designing services for youth with mental health issues: novel elicitation approaches. Int J Qual Methods 2019 Feb 21;18:160940691881624. [CrossRef]
    40. Steen M, Kuijt-Evers L, Klok J. Early user involvement in research and design projects? A review of methods and practices. Eur Group Organ Stud 2007 Jan:2007.
    41. General Practice Management Of Type 2 Diabetes: 2016–18. The Royal Australian College of General Practitioners. 2016.   URL: https:/​/www.​racgp.org.au/​FSDEDEV/​media/​documents/​Clinical%20Resources/​Guidelines/​Diabetes/​General-practice-management-of-type-2-diabetes_1.​pdf [accessed 2020-08-12]
    42. Rose AF, Schnipper JL, Park ER, Poon EG, Li Q, Middleton B. Using qualitative studies to improve the usability of an EMR. J Biomed Inform 2005 Feb;38(1):51-60 [FREE Full text] [CrossRef] [Medline]
    43. Australian Medicines Handbook. 2019.   URL: https://amhonline.amh.net.au/ [accessed 2020-08-12]
    44. Pharmaceutical Benefits Scheme (PBS). 2019.   URL: http://www.pbs.gov.au/pbs/home [accessed 2019-02-04]
    45. Cheerkoot-Jalim S, Khedo K, Jodheea-Jutton A. An exploratory study of evidence-based clinical decision support systems. In: Information Systems Design and Intelligent Applications. New York, USA: Springer; 2019:207-218.
    46. Bodenheimer T, Wagner EH, Grumbach K. Improving primary care for patients with chronic illness. J Am Med Assoc 2002 Oct 9;288(14):1775-1779. [CrossRef] [Medline]
    47. Bodenheimer T, Wagner EH, Grumbach K. Improving primary care for patients with chronic illness: the chronic care model, part 2. J Am Med Assoc 2002 Oct 16;288(15):1909-1914. [CrossRef] [Medline]


    Abbreviations

    ACBRD: Australian Centre for Behavioural Research in Diabetes
    BGL: blood glucose level
    CDS: clinical decision support
    CVD: cardiovascular disease
    EMR: electronic medical record
    GP: general practitioner
    HbA1c: glycated hemoglobin
    HCP: health care professional
    PBS: Pharmaceutical Benefits Scheme
    RCT: randomized controlled trial
    SDM: shared decision making
    T2D: type 2 diabetes


    Edited by G Eysenbach; submitted 10.02.20; peer-reviewed by S Seidu, C Mitchell, T Skonnord; comments to author 12.06.20; revised version received 20.06.20; accepted 26.07.20; published 02.09.20

    ©Breanne E Kunstler, John Furler, Elizabeth Holmes-Truscott, Hamish McLachlan, Douglas Boyle, Sean Lo, Jane Speight, David O'Neal, Ralph Audehm, Gary Kilov, Jo-Anne Manski-Nankervis. Originally published in JMIR Formative Research (http://formative.jmir.org), 02.09.2020.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR Formative Research, is properly cited. The complete bibliographic information, a link to the original publication on http://formative.jmir.org, as well as this copyright and license information must be included.