A multidisciplinary team was assembled, consisting of clinical endocrinologists (LAD, RZE), experts in medical education (PAG, MLSGJ, ICMC) and software developers/game designers (JBA, RMS). The endocrinologists compiled a list of main topics on insulin initiation and adjustment for treatment of adults with DM, in the context of a primary health care setting, outlining the minimum curriculum of the game (Textbox 1).

Periodic team meetings were scheduled during the design and development phases, to review each step, correct problems, and make decisions for the following stages, in an iterative process. The group agreed that the game should be developed in a way to satisfy 2 basic conditions: (1) to be educationally adequate, and (2) to be self-motivating and attractive to the final users.

The authors (LAD, RMS) developed the story of InsuOnline based on Joseph Campbell´s classical description of the myth cycle [26]. The story begins in a primary health care clinic, where an experienced medical doctor (the mentor) wants to go on vacation, but he cannot leave without getting a substitute physician to take care of his diabetic patients. So, he randomly chooses a younger doctor to replace him (ie, the player´s avatar). The player proceeds to see a series of diabetic patients who usually require insulin initiation or adjustment in order to obtain a better glycemic control. If the player successfully reaches the end of the game, the mentor character is finally able to go on vacation, and the player´s character is thanked by the mentor, obtains the respect of the nurse and the patients, and a brand new office. The main objective of the game is to have each player correctly initiate or adjust insulin for each patient case.

We created 16 diabetic patients, reflecting common clinical scenarios in actual primary health care clinical practice, each one corresponding to at least one of the topics of game´s minimum curriculum. Each game level has a new patient presented to the player. The patients/levels are disposed in order of increasing complexity. The player´s avatar must make decisions on the best therapeutic option to improve each patient´s glycemic control at every level. Right decisions lead to progression to the next level, and incorrect choices lead to a new attempt at the same level. The basic structure of each level is shown in Figure 1.

Several ways of interaction and game elements were included to create a pleasant experience and to maximize players´ intrinsic motivation. The player can read patients´ clinical charts, dialogue with them (be choosing among a few options of "talks"), see their lab tests and fingertip glucose readings, review their physical exam, answer quiz challenges, solve puzzles, get tips from the mentor and a nurse, and, prescribe insulin. Sound track, score, visual and sound effects, patients' mood (the patient gets angry if the player makes too many errors), and between-levels animation scenes (machinimas) were also included, to improve the gaming experience and players´ motivation.

Several pedagogical elements were included in the game, aiming for the best educational effects. These were based on the principles of adult learning and problem-based learning, including motivation, goal-orientation, relevancy-orientation, self-pacing, timely and appropriate feedback, reinforcement of learning, informal environment, contextualization, and practical (ie, hands-on) approach with active participation of the learner [27,28].

The game gives immediate feedback (presented by the mentor character), comparing player’s decisions with recommendations from clinical guidelines [1,29-31]. Correct decisions lead to gaining points and progression in the game. At every step, the game offers additional learning resources: help files (eg, algorithms, summaries, and clinical pearls), orientation from the mentor, and links to bibliographic references.

The next step will be the usability and playability assessments. For this, we will enroll 4 physicians who work in Londrina, Brazil (2 female and 2 male; 2 with gaming/computer previous experience and 2 without experience) and 4 undergraduate medical students from Universidade Estadual de Londrina (UEL) (2 female and 2 male, with and without gaming experience), who will play the game on a desktop computer, in a controlled environment, each participant alone in a single session. Usability data will be assessed using Web-based System Usability Scale (SUS), as previously described by Brooke [32], and playability will be assessed by Web-based Heuristic Evaluation for Playability, as described by Desurvire et al [33] The actions of the players will be recorded by the software for further analysis by the researchers (LAD, RMS). Further adjustments will be made in the game, according to responses obtained in this phase of the study.

After final adjustments in the game, guided by usability and playability assessments, the efficacy of InsuOnline as an educational tool will be assessed in an unblinded randomized controlled trial [34]. We will send a letter of invitation to all primary care physicians (PCPs) who work in Londrina to participate in the study. If those PCPs do not fulfill our sample size, we will also invite PCPs from other cities in the state of Paraná (Brazil), such as Maringá, Curitiba, or São José dos Pinhais. The PCPs who are willing to participate in the study will be included and randomized at study entry, using a random number generator, to 1 of 2 groups. We will exclude clinical endocrinologists or diabetologists.

Physicians enrolled in the Group A will be exposed to InsuOnline. They will be asked to play the game until its end, on the Web (with an individual login) in their own time and rhythm. Physicians enrolled in the Group B will undertake a traditional instructional session, during one afternoon, composed of a short lecture and a group discussion of clinical cases, which will be identical to the ones included in InsuOnline. This traditional instructional session will focus on the same teaching topics of the game, and it will be coordinated by a clinical endocrinologist not linked to the research team in order to avoid potential biases. This endocrinologist will be trained by the researcher endocrinologists, and he will use didactic material prepared by our team.

We will evaluate subjects´ knowledge on insulin therapy using a Web-based questionnaire, containing 10-20 multiple choice questions. The questions will be clinical vignettes of diabetic patients who require initiation and/or adjustment of insulin. For each question, the participant should choose the best option for achieving a better glycemic control, according to current guidelines [1,29-31. Questions regarding insulin therapy will be selected from the American Diabetes Association Self-Assessment Program, Module 2 (Pharmacological Treatment of Hyperglycemia) [35], translated to Portuguese, and adapted to be compatible with the game´s clinical scenarios and learning objectives. Participants´ attitudes regarding diabetes will also be assessed, by the application of Web-based Diabetes Attitudes Scale, version 3 (DAS-3) [36]. The questionnaire and the DAS-3 will be answered by the participants at 3 time points: before the intervention (pre-test), immediately after the intervention (after-test), and 6 months after the intervention (late-test). The average number of right answers will be presented and compared in the different time points within each group (intragroup) using analysis of variance (ANOVA), and between the 2 groups at each time point (intergroup) using Student´s t test. A bicaudal significance level of 0.05 will be adopted. Data analysis will be done by intention-to-treat, with last observation carried forward, which means that all randomized participants´ data will be included in the final analysis.

The participants will be asked a few Likert-scale and free text questions at the end of each intervention to assess their general impressions about the intervention (mostly to assess if it was pleasant or enjoyable). Software-recorded data on player´s usage of the game will be analyzed to assess the number of logins, time spent on the game, etc. For the participants in Group A that do not finish the game, the reasons for early withdrawal will be collected by an online questionnaire. We will contact (by phone or email) the participants who do not eventually fulfill the questionnaires at least twice.

In order to detect a minimum standard deviation of 0.5 on the average of right answers with 80% of statistical power at 5% of significance level, we will need to include 128 subjects in our study (64 in each group). We will enroll 160 subjects (80 in each group), in order to compensate for an expected 25% dropout rate.

Participation will be anonymous and voluntary, and all subjects will be asked to provide written informed consent, according to Brazilian Health Ministry´s regulations (Multimedia Appendix 1). The study protocol was approved by our local Research Ethics Committee (UEL, #051/2011 and #051/2012), and registered by UEL research board (Research Project #07471).