Games and Diabetes

Abstract

Here we review 18 articles that describe the design and evaluation of 1 or more games for diabetes from technical, methodological, and theoretical perspectives. We undertook searches covering the period 2010 to May 2015 in the ACM, IEEE, Journal of Medical Internet Research, Studies in Health Technology and Informatics, and Google Scholar online databases using the keywords “children,” “computer games,” “diabetes,” “games,” “type 1,” and “type 2” in various Boolean combinations. The review sets out to establish, for future research, an understanding of the current landscape of digital games designed for children with diabetes. We briefly explored the use and impact of well-established learning theories in such games. The most frequently mentioned theoretical frameworks were social cognitive theory and social constructivism. Due to the limitations of the reported evaluation methodologies, little evidence was found to support the strong promise of games for diabetes. Furthermore, we could not establish a relation between design features and the game outcomes. We argue that an in-depth discussion about the extent to which learning theories could and should be manifested in the design decisions is required.

Childhood diabetes, which presents when the child’s body either produces insufficient insulin to convert the glucose in the body into usable energy or the insulin that is produced is ineffective for the task, is a significant problem. More than 79 000 children under 15 years are estimated to develop type 1 diabetes annually worldwide, while the total worldwide incidence is estimated to be almost half a million.¹ The number of children with type 2 diabetes is more difficult to gauge. However, the global rise of childhood obesity² suggests that there might be a hidden worldwide epidemic of type 2 diabetes among children and young people.³

Self-management is the cornerstone to maintaining an optimal immediate and longer-term health status for diabetics. While, in many countries, young people with diabetes are accustomed to frequent visits to hospital where they undergo a range of repeated assessments (including monitoring blood glucose levels, weight and height, and eye and foot health), increasingly they also participate in educational sessions, where they learn about their condition and how best to manage it. In adolescence, the young person begins to adopt more responsibility for the management of their condition⁴ and thus present additional support requirements.

In the field of education, computer games have been suggested as a medium to deliver effective pedagogy,⁵ one that exploits the interests and experiences of the young people to whom they are targeted and thus leads to better learning, especially for children with additional needs.⁶ Educational games aim to leverage the affordances of entertainment games, many of which encourage persistence, risk taking, attention to detail, and problem solving skills; and well-designed educational games can enable players to actively construct understanding at their own pace. Computer games designed to support learning, it is argued, have many advantages,⁷ and could be especially important for children with diabetes because diabetes, if not well controlled, can be life-threatening.

For children with diabetes, well-designed computer games⁸ could offer a risk-free space in which to test and explore different scenarios for food consumption and insulin production. They could also offer an alternative fun and engaging way to teach newly diagnosed patients about necessary routines or to motivate them, by means of game techniques such as awarding points, to maintain those routines and thus achieve a healthy life style. In any event, developing games for children with diabetes is an interdisciplinary task that encompasses knowledge from learning research, game design, human-computer interaction, medicine, health education, and nutrition.

Recent research efforts reviewed games for diabetes.^9,10 Our review adds to the existing body of knowledge, as it covers the period from 2010 to 2015 and is not limited to diabetes self-management. It sets out to establish, for future research, an understanding of the current landscape of digital games designed for children with diabetes. It draws on the techniques of (but makes no claims to the status of) systematic reviews¹¹ and investigates, in particular, the use and impact of well-established learning theories in such games. Eighteen games were identified and reviewed from a technological, methodological, and theoretical perspective. Our findings highlight some common themes and gaps in the design and the methodological approaches, and we propose potential opportunities for using learning theories to inform future designs.

The article is organized as follows. First, we introduce our data collection and analysis methods. Second, we present our results. And third, we propose design opportunities that are informed by different learning paradigms.

Method of data collection and analysis

We undertook searches covering the period January 2010 to May 2015 in the ACM, IEEE, Journal of Medical Internet Research, Studies in Health Technology and Informatics, and Google Scholar online databases using the keywords “children,” “computer games,” “diabetes,” “games,” “type 1,” and “type 2” in various Boolean combinations. This process identified 35 articles. Articles were included in the review if they addressed the design and/or the evaluation of 1 or more games for children with diabetes.^12-26 Twenty articles were excluded as they focused on a technology (such as a mobile device) and lacked the gaming component. We undertook further snowball tracing, using references given in the included articles. This led to the inclusion of 2 additional articles from 2004^27,28 and 1 from 1997,²⁹ making a total of 18 articles identified for review.

Two independent reviewers coded the 18 articles according to the following dimensions; target age, diabetes type, game goal, game play, technology type, theoretical framework used, evaluation methodology (if used), and results (if any) (see Table 1). Disagreements between the reviewers were discussed and resolved.

Table 1.

Synopsis of Collected Data.

Dimension		Number of articles	References
Target ages	Young children	8	13,19,22,23,25,27-29
	Teenagers	11	12-16,19-22,28,29
	Adults	6	14,17,18,24-26
	Children and adults	2	25,28
Diabetes type	Type 1	12	12,13,15,16,19-24,27,29
	Type 2	2	17,26
	Type 1 and type 2	4	14,18,25,28
Game aim	Self-management	8	12-15,21,24,25,27
	Blood glucose checking	4	16,19,25,28
	Diabetes education	7	17,18,20,22,23,25,29
Game play	Individuals	8	12,15,16,18,21,24,27,28
	Individuals and family	3	22,23,25
	Groups	7	13,14,17,19,20,25,29
Technology type	PC	8	13,15,17,21,22,24,27,29
	Mobile	3	18-20
	PC and mobile	1	25
	Mobile and wearable	2	16,28
	Gym machine	1	14
	Website	2	12,23
Theoretical framework	Rewards/incentives	7	13,15-17,19,20,28
	Social cognitive theory	4	12,17,25,29
	Social learning	2	13,15
	Others	10	12,13,16-18,21,22,25,26,29
	Unspecified	4	14,23,24,27
Evaluation methodologies	No evaluation	2	13,14
	Usability	8	18-22,24,26,27
	Psychological	1	15
	Clinical and psychological	4	12,16,17,29
	Clinical and usability	1	28
	Usability and psychological	2	23,25
Results	Usability-satisfaction	9	16,18,20,22,23,25-28
	Clinical: improvement	1	17
	Clinical: no improvement	1	12
	Clinical: mixed results	3	16,29,28
	Psychological, social, or emotional	6	12,15,17,23,25,29
	No reported results	5	13,14,19,21,24

Note: The numbers given for each value do not necessarily add up to the total number of included articles. This is because any particular article might have included references to 1, none, or many of the given values.

Findings

In this findings section, first we discuss 3 very different examples of the games described in the articles included in this review. Then we discuss the articles in terms of their game content, their stated theoretical framework, and the methods used to evaluate their game.

Game Examples

Packy and Marlon²⁹ is one of the earliest games designed for young people who have diabetes identified in the search. It was designed for children between 8 and 16 years of age. Players take the role of Packy or Marlon, who are adolescent elephants, and who must rescue food and diabetes supplies from a diabetes camp. While the aim of the game is simple (to rescue food and diabetes supplies), players must also monitor their character’s blood glucose, take correct insulin amounts, review a diabetes logbook, and find foods containing the correct amount of food exchanges. While playing, they also answer questions relating to self-care and diabetes-related social situations. The elephant characters have a prescribed meal plan and a set of food exchanges and insulin doses. If they follow their prescription, their blood glucose level will be in the green area: low-green before meals or high-green after meals. The players have multiple options in terms of how insulin doses are administrated either as fixed or adjustable doses, while the blood glucose meter reflects the changes in insulin level and the difference between the prescribed and consumed food.

DAILY (Daily Automated Intensive Log for Youth)²⁸ is a game that focuses on helping children aged between 8 and 18 years old who have diabetes to manage their blood glucose behavior. The players are encouraged to check their blood glucose levels 4 times a day, and to upload their blood glucose readings, their insulin dose and their intake of carbohydrates into the game. After 3 sets of readings had been input, the player is prompted to predict their next blood glucose reading. The games use a behaviorist gamification approach (“the use of game design elements in non-game contexts”),³⁰ with points being awarded for playing the game and for accuracy of predictions.

Power Defense²¹ is a game that claims to apply concepts of experiential learning (an approach that stresses the importance of direct experience and reflective observation).³¹ The young person with diabetes is challenged to maintain a balanced blood sugar level, as represented by a power station output level, and is provided with a range of tools and materials, such as a coolant that represents insulin. The player also has to survive “attack waves” which represent food, and to try and maintain an optimal energy (blood sugar) level. While the players have to answer diabetes-specific questions, it was not clear from the article how the game encourages the player to reflect on their experience, or to conceptualize or build on it.

Game Content

Twelve of the games^{12-16,19,21,24-25,27,28} discussed in the articles focused on self-management aspects of diabetes care. These included barriers to self-management, and the relationship between food, insulin, physical exercise and blood glucose level. Out of these 12, 4 games^16,19,25,28 focused on encouraging the player to increase their blood glucose checking behavior. Seven other games^{17,18,20,22,23,25,29} focused on educational topics related to diabetes: diabetes symptoms, the biology and mechanics of insulin injections, estimating the carbohydrates intake, and healthy life-styles.

Theoretical Frameworks

Seven of the games^{13,15-17,19,20,28} discussed in the articles relied on rewards, incentives, and leader-board-style game scenarios, which is to say they applied, although they did not discuss, a behaviorist gamification approach (“the use of game design elements in non-game contexts”).³⁰ Although frequently dismissed as “chocolate-coated broccoli,”³² this approach is still very common in educational computer games (cf Kahn and Reed³³).

The most frequently mentioned theoretical frameworks were social cognitive theory (a theory that holds that knowledge acquisition is directly related to observing others within the context of social interactions)³⁴ (4 times^12,17,25,29), which includes self-regulation and judgment of one’s behavior, and social constructivism (knowledge, it is argued, is actively constructed through engagement with other people)³⁵ (2 times^13,15), which includes using technology as a substitute for a “more capable peer.” Other models and theories were mentioned 1 time each: health belief model (the idea that people’s beliefs about health problems explain their engagement in health-promoting behavior),¹⁶ health communication theory (the study and practice of communicating promotional health information),¹⁸ experiential learning (an approach that stresses the importance of direct experience and reflective observation),²¹ Bandura’s theory of observation (the theory that people learn from one another by observation, imitation, and modeling),²² self-determination theory (using natural tendencies to promote healthy behavior),¹² behavior change theory (the importance of environmental, personal, and behavioral characteristics for determining behaviors),¹⁷ progressive mastery (an approach that involves individualized pacing, immediate feedback, and frequent evaluation),²⁵ and a so-called transtheoretical model.²⁵ Four articles did not mention any theoretical framework.^14,23,24,27

Evaluations

As noted in Table 1, while 2 articles^13,14 did not evaluate the games, and 3 others^19,21,24 reported only formative evaluation results, 16 articles^12,15-29 evaluated the games using 1 of 3 broad metrics—usability, psychological, or clinical—or by a combination of these approaches.

Eleven of the articles^18-28 evaluated their games in terms of the game’s usability or its user interface—although there was very limited analysis of how the children actually played the game (the play strategy and experience). Instead, these articles evaluated the games by means of self-report questionnaires that addressed human-computer interaction issues such as ease of use and motivation. Only 1 article²⁴ raised the issue of the variability in play strategies (known as balancing), that was included to avoid situations where the players master how to win the game while paying too little attention to the educational messages. Similarly, little was discussed about the interrelation between the affordances of the technology and its impact on its use. For instance, mobile devices, unlike tablets or PCs, might only allow shorter interactions to preserve battery life.¹⁸ Some articles did discuss issues related to the availability of an Internet data connection needed to upload meter readings.²⁸

Seven of the articles^{12,15-17,23,25,29} used psychological measures, however details of the instruments used were not always provided.^22,23

Only 5 of the articles^{12,16,17,28,29} used some form of clinical measure as a way of evaluating their game. Those that did use the number of blood glucose checks, the number of hypoglycemia or hyperglycemia episodes, or the outcomes of glycated hemoglobin test (HbA1c or A1c) for measuring the average blood glucose level over 3 months. This small number of clinical-based evaluations might be explained in various ways. Some of the games did not have a direct clinical intention (for example, 1 focused on helping children estimate carbohydrates count).¹⁸ Other reported games were prototypes or preliminary designs (eg, Stone¹⁹) and as yet unsuitable for clinical application. And, for a final group, the study duration was notably shorter than and thus incompatible with the typical 3-month diabetes testing cycle.¹⁵

The quality of evaluation evidence was somewhat unsatisfactory in many articles, either very limited in scope or lacking the appropriate details. This, together with the variation in evaluation methodologies, makes it impossible to compare the games effectively, to draw general conclusions about their efficacy, or to establish a relationship between the design features and the game outcomes.

Learning Theories: Opportunities for Design

As mentioned earlier, most educational games still tend to be behaviorist in design (grounded in principles of repetition and rewards), although most educational game researchers argue for a constructivist or social-constructivist approach (cf Barab et al³⁶). Children are understood to actively construct their own knowledge through engagement with aspects of the world around them, and with other people, which involves an internal reorganization of previously held ideas to accommodate newly encountered ideas.³⁷ The knowledge so gained is thought to be more useful and better learned than anything rote learned. However, while games designed to support children with diabetes might benefit from being more constructivist (or social constructivist, ie, also involving other people) in nature, giving children safe yet authentic and guided opportunities to explore and construct ideas about how best to self-manage and live with diabetes, it is also important to leverage the benefits of deliberate practice,³⁸ which is nothing more than a reinterpretation of some fundamental behaviorist principles.

Conclusions and Future Work

Albeit including only a small number of articles, this review has identified a wide range of computer games designed to support children who have diabetes, and a wide range of evaluation methodologies. One thing that is also worth noting is that the articles also frequently demonstrated a sincere effort to involve the end users, the children and their parents, at various stages in the design process.

The focus on many of the games was self-management (around checking blood glucose levels, estimating carbohydrate count, or balancing food intake with insulin and exercise), which is perceived as being a set of highly challenging but nevertheless essential tasks. The literature is not so informative in terms of clinical results since most of the published results are related to the game usability and psychological results. The clinical improvements or lack thereof, its alignment with the children, caregivers, and medical team expectations should be further explored.

In general, many of the challenges previously posed in the literature³⁹ (such as structuring the video game to promote lasting change, and understanding the multiplayer experience) are not addressed in the reviewed articles.

We are aware of the limitations of this brief review. First, despite an exhaustive search, the number of articles and games included is small. Second, as it was not possible to access the games directly, most of which were prototypes with limited availability, we reviewed the articles and not the games themselves. Accordingly, our judgments are based entirely on what the authors mentioned in their articles. Furthermore, the purpose of this review was to establish, for future research, an understanding of the current landscape of digital games designed for children with diabetes—the approaches, methodologies, and development strategies used. In the future, we plan to expand our review to include articles from other years and will explore methods that allow the evaluation of games, play experiences, and clinical effectiveness, to facilitate the development of more effective games grounded in learning theory to support children with diabetes.