The target audience was identified through a brainswarming session [24] between the lead developer (MS) and IPC specialists (GC, TRN, VS, MP, SH, and MA). The game was developed at the request of these specialists, who were therefore included in the development process from the start.

Field observations by specialists from the Geneva University Hospitals infection control program were used to identify the acquisition objectives. We decided to use the term “acquisition objectives” rather than “learning objectives” as the main purpose of this serious game was to promote and stimulate desirable behaviors in HCWs rather than to teach them new procedures.

The theoretical bases relevant to the creation of gamified content in the context of the COVID-19 pandemic were already identified in the course of a previously published project [23]. Therefore, we decided to use these bases and to reassess the relevant references. Briefly, we had searched the medical literature via PubMed (using Medical Subject Headings [MeSH] and Boolean operators) and retrieved articles based on their titles and abstracts. References from the most relevant articles were also searched manually to identify sources that could have been missed. A total of 12 articles was included through this procedure. Further details are available in our previous publication [23].

We also performed a search for articles reporting information related to SARS-CoV-2 transmission. The MeSH keywords “covid-19,” “sars-cov-2,” and “transmission” were used in combination with the Boolean operators “AND” and “OR” to retrieve potentially relevant references from MEDLINE using the PubMed engine. To avoid inducing doubts or even mistrust in HCWs [25-27], the results retrieved through this search were anaylzed to determine whether they were in line with our institutional [28] and federal [29] guidelines. Whenever conflicting information was identified, a consensus was reached with the IPC specialists to determine and use the most accurate and least confusing messages.

The scientific grounds on which this serious game relies were decided during brainswarming sessions between the developers and IPC specialists from the Geneva University Hospitals. Each iteration of the serious game was reviewed by at least 2 such specialists to ensure that the intended message was correctly and adequately conveyed by the serious game.

The scientific foundations reviewed or established during the previous stage were used to create the general design of the serious game. We used Nicholson’s RECIPE mnemonic (reflection, engagement, choice, information, play, and exposition) for meaningful gamification to guide the development of the game mechanics [30].

Arnab et al’s [31] model of transforming learning mechanics into game mechanics was adapted to suit the development of this serious game. The original model states that learning mechanics should be decided according to learning objectives and that these learning mechanics should then be translated into game mechanics. As we had decided to use the term “acquisition objectives” rather than “learning objectives,” we adapted Arnab et al’s [31] model to develop a very similar “acquisition mechanics – game mechanics” model.

Design decisions were tailored to the target population and the COVID-19 context. Ease of access to the serious game was considered to be of paramount importance, as was the ability to rapidly capture the player’s attention and interest. True to the SERES framework, we decided to use an iterative approach to build the game. We therefore sought regular feedback from the IPC specialists as well as from potential end users and implemented the required adaptations accordingly.

All the data gathered through the scientific foundation and design foundation stages were used to choose a development platform and to proceed with the development of the serious game. Acquisition objectives were reassessed, and theoretical bases were used to decide upon acquisition mechanics and game mechanics.

In line with the theoretical bases, the game was tested on various platforms (smartphones, tablets, laptops, desktop computers) by different users at the end of each development loop and before its final release.

Early on in the development process, developers and IPC specialists agreed that the target audience should be composed of all the professionals who could be in direct contact with patients within the Geneva University Hospitals. The game was therefore designed to target many different kinds of health care professionals including physicians, nurses, physiotherapists, assistants, technicians, and therapists but also non–health care professionals. Indeed, some of these employees have close and regular contact with patients (stretcher-bearers, housekeepers, etc) and might therefore be exposed to patients or play a role in SARS-CoV-2 transmission.

We used previously identified resources to establish the theoretical background [20,31-37]. These resources were extracted from Verschueren et al [22], which described the development of a serious game designed to reduce perioperative anxiety in children, and from our recent study describing the development of a gamified e-learning module that was used to teach the adequate use of PPE to prehospital personnel in the context of the COVID-19 pandemic [23]. Moreover, we sought to include all 6 elements of Nicholson’s meaningful gamification [30]. These elements are listed below and follow Nicholson’s RECIPE mnemonic:

Audio elements were not considered necessary as many hospital employees will play the serious game on institutional computers and will not be able to activate the sound. Attractive graphics adapted to the target population were however deemed essential to increase engagement [47,48].

The IPC specialists defined 10 acquisition objectives according to their field observations. Table 1 summarizes these objectives, the RECIPE element to which they are linked, and gives implementation examples.

To keep track of the player’s choices and to strengthen IPC messages, a virus counter was created. Whenever the player selects an answer that can lead to a possible contamination, a red virus image appears, and the virus count rises incrementally. Conversely, every time the player selects an answer that matches a desirable behavior, a positive token in the form of a “thumbs-up” image is awarded. If the player reaches a total of 5 viruses, a “game over” screen (Figure 18) appears. The user can then choose to decrease the virus count by exchanging “thumbs-up” tokens against viruses or to restart the level. We decided that the players should only restart the level, and not the entire game, in order to provide a more positive feedback and to avoid generating frustration. Moreover, a positive message is displayed at the end of each of the 4 levels. These messages are designed to strengthen the player’s motivation to comply with IPC guidelines and are not influenced by the virus count.

Feedback mechanisms were used extensively and systematically in this serious game [32]. After a simple-choice interaction, the expected answer is immediately provided to the player along with an explanatory message (Figures 8 and 9). When multiple choice interactions are involved, players are allowed to adapt their answer if they were initially wrong. To help them find the correct answers, the number of wrong answers is provided along with the number of correct answers missing from their initial choice (Figure 14).

We first created a storyboard using PowerPoint 16 (Microsoft Corp). This storyboard, split into 4 game levels, was refined through multiple iterations until we deemed it advanced enough to begin the actual game development. A common and simple terminology was used to suit the broad target audience. We chose to develop the actual game under Articulate Storyline 3 (Articulate Global). This software is easy to use, embeds many elements that are essential for creating interactive content and serious games, and allows developers to export their creations to many different platforms using HTML5.

The first version of the game was developed using stick figures created by PresenterMedia (Eclipse Digital Imaging Inc). Figure 1 shows the welcome screen of this first version, and Figure 3 gives an example of a simple-choice interaction.

Although developers and IPC specialists found this first version to be both adequate and usable after having been refined through multiple iterations, they felt that it resembled a gamified e-learning module rather than a serious game. We therefore hired a well-known Swiss graphic designer to help us enhance the graphical aspects of the game and create a unique atmosphere. First, sketches in line with the multiple iterations concept of the SERES framework were produced by the graphic designer. These sketches were revised, adapted, and finally validated. They were then transformed into actual images that were embedded in the serious game. Figure 2 shows the updated (and final) version of the welcome screen, and Figure 4 displays the updated version a simple-choice interaction.

Each iteration of the game was tested by at least 3 authors whose feedback was systematically assessed and, when relevant and feasible, integrated into the next iteration of the game. The final version was validated by all the authors and by IPC specialists, some of whom act as consultants for the World Health Organization. Apart from the authors, all of whom are part of the medical or nursing staff, the game was tested by nonclinical staff from the Geneva University Hospitals training center before its release. These testing sessions showed that the game takes 15 minutes to complete on average.

The heuristic evaluation procedure described by Davids et al [49] was used to screen for usability issues. This led to the addition of blinking images in the welcome screen to facilitate identification of interactive elements.

This serious game can be freely accessed on the internet [50]. The corresponding author can be contacted to obtain a SCORM (shareable content object reference model) package of the game. We decided to make this package available to facilitate results and completion tracking on most current learning management system. This will allow for the tracking of completion and, on recent platforms, to gather detailed results down to the level of individual questions.