Sound Scientific Foundations for CliniPup were established at the earliest stage of development to ensure that the game was relevant, theoretically driven, and evidence based, in line with the governing research methodologies. This stage assessed, based on objective criteria, whether there was a relevant unmet medical need for a clearly defined target audience that can be addressed with a serious game. In line with this aim, an exploratory review of the research literature on perioperative anxiety and pain was performed. To this end, the PubMed database was search electronically from March to April 2015 with a search strategy (terms used in duplicate: pain, anxiety, surgery, perioperative, children, and pediatric). The following criteria were applied for inclusion: English language and review. Similarly, the following criteria were applied for exclusion: published before 2000 and nonpediatric population.

This stage assessed what the objectives of a serious game should be in the context of the target audience defined in the first step. Evidence collected in the exploratory literature search was reviewed again with the aim of defining the outcome objectives.

Formulating a hypothesis of how a serious game might achieve the intended outcome objectives is a vital step toward purposeful design and the evaluation and validation of its causal effect on the outcome [47]. In particular, modifiable barriers and drivers (outcome determinants) contributing to pain, anxiety, and control in children were collected from the scientific literature. To this end, a critical review of the scientific literature was performed. From September to October 2015, the PubMed database was searched electronically with the following search strategy: pain OR children OR preoperative, postoperative OR anxiety OR induction OR perioperative OR education OR pharmacological treatment OR non-pharmacological treatment OR determinants. The following criteria were applied for inclusion: English language, review, case study, retrospective study, and clinical trial. The criteria for exclusion were published before 2000, nonpediatric population, and pharmacological intervention only.

Barriers and drivers (determinants) were extracted directly from the literature and classified into categories, thereby arriving at a manageable amount of key factors. The relative importance of each determinant was also defined based on qualitative descriptions in the selected articles. Moreover, various experts in the field such as nurses, pediatricians, and surgeons were consulted to evaluate the resulting determinant categories.

Consistent with the SERES Framework Methodology, learning objectives were defined for each determinant and subsequently mapped within a theoretical framework. The Information-Motivation-Behavioral skills (IMB) model of behavior change was selected as the theoretical framework because it has been used in the design of health promotion interventions and was aligned with educational strategies to prepare children for surgery in the literature [8,44]. These learning objectives, in turn, guided the Design Foundations.

It also became evident that there were a substantial number of clinical practice documents available, which were not considered primary research. Therefore, the secondary literature on educational programs for children in the perioperative setting was also reviewed. These findings played an important role in informing the learning objectives and various game design aspects, which are described in the Results section.

Both the literature collected via the exploratory search and the critical review were analyzed to understand how to validate a serious game in the context of a nonpharmacological intervention aimed at addressing perioperative anxiety and pain in children. The findings were integrated into the game design process and the subsequent clinical evaluation.

Game mechanics (GMs) are rules or methods that define the interactions and flow of a game session. They describe interactions, game conditions, and triggers in an abstract manner. In 2015, Arnab proposed a model for translating learning objectives into learning mechanics (LM) and mapping these to relevant GMs [48]. This so-called LM-GM model guides developers in the development of more effective, pedagogy-driven serious games, as it ensures that game mechanics are selected based on their ability to contribute toward the intended outcomes [48]. The LM-GM model was utilized to map GMs to the learning objectives identified in the Scientific Foundations stage. For example, one such learning objective was to recall the sequence of events for the upcoming procedure (knowing what to expect and do). This involves Blooms’ Ordered Thinking Skills “understanding” and “retention.” Based on the LM-GM model, several LMs address these thinking skills: “exploration,” “repetition,” and “planning” [48]. Each of these LMs was, in turn, mapped to one or more GMs such as “story,” “cascading information,” and “strategy/planning” [48]. As such, the scientific foundations established in Stage 1 were explicitly and systematically translated into the game construct. The implementation of selected GMs in the resulting serious game was then documented to retrospectively assess the translation to the game development stage and ensure that the selected GMs were indeed applied.

Based on findings from the biomedical literature and through interaction with key stakeholders (nurses, pediatricians, parents, etc), a time-motion exercise was carried out to map anxiety and pain throughout a child’s (and their parents) surgical journey. Additionally, when and where (which setting) each of the outcome determinants was most relevant was identified. In turn, this information helped to define various aspects of the serious game such as which characters (stakeholders) should be included, the relevant settings to incorporate, key time points to address, and the most applicable visuals and content elements.

Based on review of the literature, the target audience was defined as children aged 6-10 years with a high unmet medical need at the individual and population level. Although other age segments also showed unmet needs, children aged 6-10 years were selected due to the substantial physical, psychological, social, and economic burdens associated with inadequate preparation and management [5,8,19,28,29]. Moreover, management for children in this segment was considered especially challenging because of several unique characteristics: need for support and information, need for control, capacity to understand age-appropriate information, lack of communication and coping skills, and limited understanding of the surgical purpose and journey [8,19,30,41,49]. In addition, this segment undergoes a robust amount of ambulatory procedures each year [1,3].

Due to the relationship between parental factors (knowledge, skills, behavior, etc) and children’s pain and anxiety, parents are also an important target audience for a nonpharmacological intervention [8,9,16,28]. Therefore, parental unmet needs must be addressed simultaneously. Parental needs are mainly related to coping and communication skills, support, and information on the surgery process and journey [8,19,25,29].

There is a particular need to educate children and their parents about the concepts of pain and anxiety, allow them to explore the surgical pathway, and change their behavior with respect to communication and coping [8,19]. These needs are intended to indirectly, or directly, address the overarching goals, which are to reduce child anxiety and pain in the perioperative setting.

The literature search, described in the Methods, resulted in the selection of nine articles for inclusion in the study [8,17,18,25,29,30,49-51].

After review, it was clear that pain and anxiety have sensory, emotional, cognitive, and behavioral components that are interrelated with environmental, developmental, sociocultural, and contextual factors [8,12,52]. A number of determinants of pain and anxiety were considered fixed (eg, risk factors) or nonmodifiable and could not be addressed with an intervention (eg, child demographics, parent demographics, and type of procedure) [17,30]. Seven key modifiable determinants were identified: medical fears, current state, sense of control, ability to communicate, understanding (information/knowledge), focus on pain, and parental fears. These seven determinants were generated through categorization of approximately 100 relevant underlying parameters, which were directly defined in the literature. The determinants were also closely aligned with clinical experience suggesting that the following psychological factors are known to influence pain and therefore should be addressed by nonpharmacological interventions: vigilance to pain, avoidance, anger, involvement of the patient, making sense of the pain, and consistency [53]. In turn, the learning objectives were defined for each determinant, and a theoretical mechanism was defined as information, motivation, or behavioral skills, consistent with the IMB model (Multimedia Appendix 1).

As the focus of the development was on a serious game directed at children, it was determined to be infeasible (and impractical) to address the parental learning objectives alongside the child’s learning objectives in a single serious game. However, the parental learning objectives remain critical factors that should be addressed through a complementary or supplementary mechanism or medium.

The learning objectives were further categorized together to yield 11 identified objectives (Table 1).

Common approaches in the field to evaluate interventions leveraged structured clinical trials, and therefore, the researchers determined that the optimal approach to evaluate CliniPup was with a randomized controlled clinical trial [35,37,38]. With this in mind, a positive first step is to validate CliniPup in a pilot study before considering a larger, pivotal trial. Through this process, key validated outcome measures were identified for pediatric anxiety and pain and parental anxiety. These measures were examined using the modified Yale preoperative anxiety scale, the Wong-Baker Faces Pain Rating Scale (WBFPRS), and the state-trait anxiety inventory [10,54,55]. In addition to the clinical measures, an evaluation of user satisfaction and experience could also be performed in a pilot trial. Data on these aspects could be collected using structured questionnaires and Likert scales.

Based on the biomedical literature review and interviews with clinicians, it was established that the characters portrayed in the serious game needed to be realistic, yet age-appropriate, especially when visualizing medical professionals [8,41]. Parents should be visualized as kind, gentle, and comforting [8,19,44]. In addition, the literature indicated that the game’s protagonist should be authoritative; likable; relatable; and neutral in gender, age, and ethnicity [8]. Aligned with these points, the use of a nonhuman protagonist, such as an animal, was considered. Further research indicated that animals may offer a positive, comforting role model for children, and therefore, a dog, called CliniPup, was developed to represent the serious game’s protagonist [56,57].

The design requirements for the serious game’s settings were informed mainly by the ethnographic research performed (eg, time-motion exercise and interaction with medical stakeholders). This research suggested that the settings should be linked to the surgical journey and therefore cover the following environments: at home before the surgery, at the hospital before the surgery, at the hospital in the operating room, at the hospital after the surgery, and at home after the surgery [8,19]. The relevance of each of the seven determinants to the different settings was mapped using the time-motion exercise.

Due to the unique pedagogical requirements (eg, short attention span and limited language capabilities) of children aged 6-10 years, the serious game needed to be highly engaging, interesting, and fun [19]. In particular, age-appropriate visuals and language were recommended for educational tools used in health care settings targeted at children in order to enhance engagement [8,17,19]. This suggested that the use of dynamic animations, sounds, voice-overs, and limited text would be optimal. These choices are aligned with the Gestalt theory, which recommends that designers should limit cognitive efforts for optimal communication, particularly in health care settings [58,59]. Additionally, the use of multimedia techniques increases the likelihood of addressing heterogeneous learning styles, facilitating education in a diverse target audience [8,59,60].

With the pilot study design, we expected children to play the serious game CliniPup in their home setting and believed that all outcome data would be collected outside of the game itself. Therefore, minimal design requirements were necessary to accommodate the trial. At-home play was facilitated by designing the serious game to be accessible online and functional in both mobile and desktop environments. In addition, the serious game was linked to a database to allow for the collection and storage of demographic and user interaction data (eg, number of times played).

CliniPup was developed based on evidence generated from the Scientific Foundations and Design Foundations. A general narrative that explored the ambulatory surgical journey was presented, and a user interface was created to realize the narrative, as described in the Design Foundations (Figures 3-7). The use of this narrative was intended to directly address learning objectives 4, 5, and 6 and to indirectly address all other learning objectives listed in Table 1. The user interface was 2D and cartoon-like, aligned with target audience expectations. The protagonist, CliniPup, was visualized as confident, fun, and authoritative (Figure 2).

Consistent with the design requirements, minimal text was presented and voice-overs and sounds were implemented throughout. In particular, the narrative followed the five key settings of the surgical journey defined in the Design Requirements (Figures 3-7). The tool was accessible online and could be played on a personal computer or tablet.

“Anxiety monsters” would appear at key moments (defined by the time-motion exercise), in which users needed to help the protagonists chase them away (by clicking on the monsters; Figure 2). The purpose of this mini-game was to achieve learning objectives 8, 9, 10, and 11 (Table 1), and it served the purpose of collecting points for the user, which represented the overarching in-game objective (Figure 15).

At the end of the narrative (end of the surgical journey), users were presented with a number of challenges (questions/answers, puzzles, etc) to reinforce the key messages associated with the learning objectives (Table 1). For example, users review the objects (Figure 9) and the people that are present in the surgical room (Figure 10). Moreover, users are asked about the objects in the surgical room to address learning objectives 1 and 2 (Figure 11). In addition, users are requested to identify the correct sequence of events on the day of surgery in accordance with learning objective 4 (Figure 12). Users are also tasked with choosing the correct score on the WBFPRS with various examples to address learning objectives 8 and 9 (Figure 13). Additionally, users must choose actions that are aligned with control to address learning objective 11. In each mini-game, users receive points for completing the challenges correctly (Figure 15).

Based on the number of points collected, users would enter a final “anxiety-monster” mini-game (Figure 8) after which their final score would be presented (Figure 15).

The estimated gameplay time was 20 minutes. Additional information related to the game description is documented in Multimedia Appendix 2.