Influence of Personality on mHealth Use in Patients with Diabetes: Prospective Pilot Study

Background: Mobile technology for health (mHealth) interventions are increasingly being used to help improve self-management among patients with diabetes; however, these interventions have not been adopted by a large number of patients and often have high dropout rates. Patient personality characteristics may play a critical role in app adoption and active utilization, but few studies have focused on addressing this question. Objective: This study aims to address a gap in understanding of the relationship between personality traits and mHealth treatment for patients with diabetes. We tested the role of the five-factor model of personality traits (openness to experience, conscientiousness, extraversion, agreeableness, and neuroticism) in mHealth adoption preference and active utilization. Methods: We developed an mHealth app (DiaSocial) aimed to encourage diabetes self-management. We recruited 98 patients with diabetes—each patient freely chose whether to receive the standard care or the mHealth app intervention. Patient demographic information and patient personality characteristics were assessed at baseline. App usage data were collected to measure user utilization of the app. Patient health outcomes were assessed with lab measures of glycated hemoglobin (HbA1c level). Logistic regression models and linear regression were employed to explore factors predicting the relationship between mHealth use (adoption and active utilization) and changes in health outcome. Results: Of 98 study participants, 46 (47%) downloaded and used the app. Relatively younger patients with diabetes were 9% more likely to try and use the app (P=.02, odds ratio [OR] 0.91, 95% CI 0.85-0.98) than older patients with diabetes were. Extraversion was negatively associated with adoption of the mHealth app (P=.04, OR 0.71, 95% CI 0.51-0.98), and openness to experience was positively associated with adoption of the app (P=.03, OR 1.73, 95% CI 1.07-2.80). Gender (P=.43, OR 0.66, 95% CI 0.23-1.88), education (senior: P=.99, OR 1.00, 95% CI 0.32-3.11; higher: P=.21, OR 2.51, 95% CI 0.59-10.66), and baseline HbA1c level (P=.36, OR 0.79, 95% CI 0.47-1.31) were not associated with app adoption. Among those who adopted the app, a low education level (senior versus primary P=.003; higher versus primary P=.03) and a high level of openness to experience (P=.048, OR 2.01, 95% CI 1.01-4.00) were associated with active app utilization. Active users showed a significantly greater decrease in HbA1c level than other users (ΔHbA1c=−0.64, P=.05). Conclusions: This is one of the first studies to investigate how different personality traits influence the adoption and active utilization of an mHealth app among patients with diabetes. The research findings suggest that personality is a factor that should be considered when trying to identify patients who would benefit the most from apps for diabetes management. (JMIR Mhealth Uhealth 2020;8(8):e17709) doi: 10.2196/17709 JMIR Mhealth Uhealth 2020 | vol. 8 | iss. 8 | e17709 | p. 1 https://mhealth.jmir.org/2020/8/e17709 (page number not for citation purposes) Su et al JMIR MHEALTH AND UHEALTH


Diabetes Self-Management
Globally, type 2 diabetes is a common chronic disease, and its incidence is rapidly increasing in countries such as China [1,2].Treatment for type 2 diabetes is largely self-managed; patients are responsible for engaging in health-promoting behavior on a day-to-day basis [3].Health-promoting behaviors include dietary control, physical activity, and blood glucose monitoring, and these health-promoting behaviors are often incorporated as essential components of treatment programs in order to keep blood glucose levels within target ranges and to prevent long-term complications [4].Although long-term self-management and lifestyle behaviors are critical for controlling diabetes, these skills prove difficult for many patients to develop [3].

Role of mHealth
Mobile technology for health (mHealth) interventions can benefit chronic disease management by delivering real-time monitoring and reminders to a large number of people, enabling the delivery of tailored support and providing low-cost, remote health care services [5].For diabetes, mHealth interventions are increasingly being used to assist patients with lifestyle changes and health-promoting behaviors [6].Some recent studies and reviews [7][8][9] have shown that mHealth smartphone interventions for diabetes self-management have reduced glycated hemoglobin (HbA 1c ) levels and have significantly facilitated self-management for patients; however, the effectiveness of mHealth in improving diabetes outcomes critically depends on voluntary patient engagement.Most studies, in contrast, rely upon randomized trials where patients are exogenously assigned to the mHealth interventions [10].In many studies that are designed as such, the patients who are required to use the mHealth interventions showed low utilization of the health apps and often had high dropout rates [5,11].Therefore, within the literature to date, there is a limited understanding of how willing diabetic patients are to adopt mHealth interventions.Within this question, there is also the matter of individual differences that may make patients more or less likely to engage with a given intervention.In order to make use of the full potential of mHealth, researchers, technology companies, and clinicians have been exploring ways of attracting users and increasing mHealth use [12], with the understanding that personality characteristics may be an important consideration when building or recommending different mHealth products [13].Based on this, we drew on insights from personality research to explore individual differences that could explain heterogeneity in the adoption and active utilization of mHealth to improve diabetes self-management.

Personality Traits and mHealth Intervention
The five-factor model of personality (also referred to as the big five personality traits) offers a comprehensive framework for examining distinct personal characteristics and their influences.Within the big five, agreeableness encompasses the traits of courtesy, cooperation, trust, and tolerance.Conscientiousness represents tendencies such as being self-disciplined, organized, and persistent in goal-directed behavior.Extraversion is frequently associated with being sociable, gregarious, and optimistic.Neuroticism (or emotional instability) is characterized by insecurity, anxiousness, and hostility.Openness to experience represents one's curiosity and willingness to explore new ideas [14].Taken together, the big five capture the essence of one's personality.
Past research has recognized that the personal characteristics of users are essential factors in predicting technology adoption and continued utilization of different kinds of apps [15][16][17].For example, previous studies have found that more conscientious individuals may be less likely to use leisure apps, but would prefer communication and business apps [18,19].Openness to experience can be used to predict greater smartphone use and may be positively correlated with the use of social apps [16,20].Extraverts are more likely to use social networking and instant messaging apps than they are to use apps related to books, references, and education [20].Individuals with high agreeableness are less likely to use apps related to communication, browser usage, productivity, and gaming [18,19].Neurotic individuals are likely to adopt apps in general, due to their fastidious and meticulous nature as well as their interest in creative activities [19].
Previous studies [21][22][23][24][25] have also examined the effects of the five personality traits on patient intent to use mHealth apps.Openness to experience had a positive influence on acceptance of an mHealth app for hypertension [21] and was correlated with better adherence to a self-care app for cancer [22].High conscientiousness was significantly associated with the total number of points earned on an mHealth weight loss app [23].Gender played a moderating role in the relationship between two specific personality traits, extraversion and emotional stability, and in the behavioral intention to use mHealth apps, in general [24].Personality traits have been found to influence self-care behavior and glycemic control in patients with type 2 diabetes [25]; however, little is known about how the big five personality traits influence adoption preference and active utilization of diabetic self-management apps.Our exploratory study was designed to help fill this knowledge gap.

Study Objectives
Using data from a 3-month period, this study explored three relationships: (1) the relationship between the big five personality traits and mHealth adoption, (2) the relationship between the big five personality traits and active utilization of the app, and (3) the effect of mHealth app usage on health outcomes in patients with diabetes.Findings related to the association between personal characteristics and mHealth use in the context of diabetes self-care will contribute to further understanding of the mechanisms underlying high dropout rates Patients who were interested in participating attended an orientation session, which included an educational presentation on diabetes by a physician and a detailed description of the study by the research team.In the latter portion of the session, staff explained that patients would choose which treatment they wanted to receive-either the app treatment or standard care-if they chose to participate.The staff then presented the app to explain its functions and why and how it could help the patients.Patients were informed that all participants in both groups would receive the same perks including free HbA 1c tests, a portioned dinner plate as a gift, an educational handbook about diabetes self-management, and physician guidance for their health management.They were also informed that, as an additional incentive, those in either group who decreased their HbA 1c level during the 3-month period would receive a gift worth 50 RMB (approximately US$ 7).At the end of the orientation session, the prospective participants were entirely free to choose which group to join, and if they chose to participate, they signed a statement to indicate that they were providing informed consent.
Participants attended a second session on the morning following the first session.In the second session, all participants underwent baseline measurement, including HbA 1c laboratory tests, and BMI measurements, and completed a questionnaire.Then, participants who chose the mHealth intervention were sent instructions on how to use the app.
Over the course of 1 month (November 2018), we conducted seven orientation sessions and recruited 98 patients.Participant exclusions and categories are summarized in Figure 1.The study duration was 6 months, but the analyses in this paper reflect midpoint findings (3 months).

mHealth Intervention
Participants who indicated intention to use the app were given access to the health management app (DiaSocial) designed by our research team; screen captured images from the app are shown in Figure 2. The app was available in both iPhone and Android versions.App components consisted of educational resources, tracking features, and feedback.Participants had continuous access to the app throughout the intervention period.
Educational information included basic self-management strategies as well as guidance for app use.Participants who used the app were instructed to use it daily and record their progress towards diabetes self-care goals such as exercise, a healthy diet, managing glucose levels, and medication adherence.Physical activity was objectively measured using the participant's smartphone pedometer or was manually entered by participants into the app.Other records were manually entered.The app provided feedback through graphical displays of logged behavior.Two types of graphical feedback were provided.The first comprised separate graphs of glucose levels, sleep, medication adherence, diet, weight, and moderate to vigorous intensity physical activity.An information breakdown was provided in the form of daily, weekly, and monthly line charts.The second mode of graphical feedback was a red arrow displayed on the analysis graphs designed to alert participants ( 1) when blood glucose level prior to meals (breakfast, lunch, or dinner) was greater than 7 mmol/L, ( 2) when blood glucose level after meals (breakfast, lunch, or dinner) was greater than 10 mmol/L), (3) when BMI was greater than 25 kg/m 2 .
The app was also gamified such that participants earned scores within the app for use and for reaching target levels of glucose, exercise, nutrition, and medication adherence.The app ranked all the participants according to their daily total scores.At the same time, app users were also assigned to different treatment arms, such as team competition and individual competition, but these experimental conditions did not work out as planned and all analyses were collapsed across the various treatment arms.

HbA 1c Laboratory Test
Glycated hemoglobin (HbA 1c level) was chosen as our key clinical outcome because higher HbA 1c levels have been associated with more complications and poorer health outcomes [27].HbA 1c level provides a picture of average blood glucose level over a period of months.Patients had blood drawn within 7 days of the study's launch to assess baseline HbA 1c level and 90 days later to determine postintervention HbA 1c (post HbA 1c ) levels.Given the 1-month recruitment span, there was variability in the dates on which HbA 1c levels were measured.HbA 1c values of participants were measured at Daqing Hospital as part of the study, and researchers collected these lab values for analysis.

Questionnaire Design
The baseline questionnaire was prepared in two languages, English and Chinese, using a semantic translation technique.The content validity of this instrument was assessed by an endocrinologist and also by type 2 diabetic patients to ensure that the questions were appropriate and relevant in the current Chinese setting and culture.The questionnaire items included demographic information (age, gender, education, BMI, and time since diagnosis), measures of health behaviors and attitudes, and a personality inventory.
Personality was assessed with the previously developed Chinese version of the Ten-Item Personality Inventory [28], a widely used, brief measure of the big five personality traits; it has two items per dimension, with each item consisting of a pair of adjectives.Half of the items represent the positive pole of the given dimension and the other half represent the negative pole.Participants rated each item on a 7-point Likert scale, ranging from 1 (disagree strongly) to 7 (agree strongly).A total score for each dimension was generated by transforming the scores of the negative pole items (reverse-scored items), then computing the averaged dimension score.Higher scores indicated a higher level of that personality dimension.
We conducted binary logistic regression analyses to determine the association of personality traits with mHealth adoption preference and active utilization.To address our first research question, all patients who agreed to participate in the study were included.To address the second question, the subset of patients who actually used the mHealth app were used in the analysis.An analysis of variance was used for numerical variables and chi-square tests were used for categorical variables to compare the differences between the three types of participants after three months: active users, dropouts, and nonadopters.We also conducted multiple linear regression to explore the relationship between the amount of app usage (number of days per week) and ΔHbA 1c for adopters.Analyses included several covariates as well as the five personality traits of interest.

Participant Classification and Summary
Based on app usage (average days per week) for each participant (see Multimedia Appendix 1), we defined participant categories as shown in Table 1.Of the patients who agreed to participate (N=98), 58% (57/98) initially indicated their willingness to download and use the app; however, 11 patients never used the app.Nonadopters, therefore, included patients who were not willing to use the app and patients who never used the app even if they initially indicated a willingness to do so.Adopters included dropouts (those who initially used the app, but then became inactive later on) and active users.As defined in Table 1, if an app user did not continue to use the app past the first two months, we regarded that user as a dropout.Users who kept using the app until the third month, were defined as active users.Active users were further divided into 2 groups: low-frequency users (whose weekly average app usage was less than 3.5 days) and high-frequency users (whose weekly average app usage was greater than 3.5 days).Detailed demographics for participant subcategories are shown in Multimedia Appendix 2. Patients who used the app, but did not continue use the app in the third month.

App Adoption Preference Among Diabetic Patients and Association With Personality Traits
Table 2 presented the results of the binary logistic regression analyses associating app adoption with sociodemographic and personality traits among diabetic patients.Relatively younger patients with diabetes (mean 56.07, SD 9.06 years) had 9% higher odds of trying the app (P=.02, odds ratio [OR] 0.91, 95% CI 0.85-0.98)compared to the odds of older patients with diabetes (mean 62.04, SD 8.04 years).Diabetic patients who were less extraverted had 29% higher odds of trying the app (P=.04,OR 0.71, 95% CI 0.51-0.98)compared to the odds of those who were more extraverted.At the same time, diabetic patients who were more open to experience had 73% higher odds of adopting use of the app (P=.03,OR 1.73, 95% CI 1.07-2.80)compared to the odds of those who were less openness.Gender (P=.43,OR 0.66, 95% CI 0.23-1.88),education (senior: P=.99, OR 1.00, 95% CI 0.32-3.11;higher: P=.21, OR 2.51, 95% CI 0.59-10.66),and baseline HbA 1c level (P=.36,OR 0.79, 95% CI 0.47-1.31)were not associated with app adoption.Ordered logistic regression for all participants, which had 5 ordered levels ranging from no intention to use to high-frequency users, also provided similar results and can be found in Multimedia Appendix 3.

Relationship Between Personality Traits and App Usage Among Diabetic Patients
Table 3 presented the relationship between personality traits and app usage.The results showed that a lower education level-specifically, participants who completed primary school compared to those who had a vocational (senior) or university (higher) education-was associated with greater odds of active app utilization (senior versus primary P=.003 and higher versus primary P=.03).Higher openness (P=.048,OR 2.01, 95% CI 1.01-4.00)was also significantly associated with greater active utilization.We found similar results by using the amount of app usage (number of days per week) as an outcome variable to conduct ordinary least squares linear regression in Multimedia Appendix 4, and by ordered logistic regression in Multimedia Appendix 5.

mHealth App Usage and Patient Health Outcomes
Because some patients did not have their final HbA 1c level measured, Table 4 reports descriptive statistics and differences among categories for retained participants after 3 months.Active users showed a greater decrease in HbA 1c level (ΔHbA 1c =-0.64,P=.05) than those shown by users in the other groups.We also observed an unexpected decrease in HbA 1c level for nonadopters and increase in HbA 1c level for dropouts.We observe similar connections between extraversion and openness and health outcomes as we did for overall app adoption-less extraverted participants and those who were more open to experience were more likely to experience a decrease in HbA 1c level.
Next, we used linear regression to test the relationship between the amount of app usage (number of days per week) and ΔHbA 1c among adopters (Table 5).We found that if app users increased their usage by ten days, their HbA 1c level decreased 0.20 points (β=-0.02,P=.02) under controlled covariables.More detailed comparisons among all categories are shown in Multimedia Appendix 6.

Principal Results
This study suggested a relationship between individual characteristics and mHealth app use (adoption and active utilization) in patients with diabetes.We found that patients with diabetes who were relatively younger, less extraverted, and more open to experience were more likely to adopt use of the mHealth self-management app.In addition, education level and openness to experience were associated with active utilization of the app.Finally, active users were also associated with better clinical outcomes than those of dropouts.

Comparison With Prior Work
Our finding that relatively younger patients were more likely than older patients to try to adopt the app was in agreement with previous studies [29,30] which found that older adults were less likely to adopt new technology.Older people reported that they do not go online for various reasons including cost, lack of skills, lack of interest, and concerns about information security; however, once older adults did get online, they were just as enthusiastic as younger users [31].This was also consistent with our own findings, as age was negatively associated with initial adoption, but not with actual use once adopted.In contrast to the outcomes suggested by prior research [32,33], we unexpectedly found that gender, education, and baseline HbA 1c were not associated with app adoption.
We also found that mHealth app users with lower education levels may have been less likely to drop out.Low education levels have been associated with inadequate health literacy [34,35], which, in turn, has been associated with lower diabetes-related knowledge and lower engagement in mobileand web-delivered self-care interventions [36].This finding may therefore seem surprising; however, Paasche-Orlow et al [37] argued that patients with low health literacy may have difficulty acquiring self-management skills, but once these skills are acquired, they may follow directions more readily than those with higher literacy [37].In line with this reasoning, our mHealth app provided diabetes self-management educational resources, which may have enabled patients with low education to learn more about diabetes management skills and may have encouraged app usage more than it did for those who already possessed diabetes self-management knowledge [38].
Our results also suggested that extraverted patients with diabetes were less likely to adopt the app, despite the fact that it incorporated and emphasized social features.This may possibly be explained by the fact that the study focused on motivating long-term and continuous self-management, which may not fulfill the social desires of extraverts.Compared to extraverts who prefer to meet friends or participate in social activities to get health support, introverted people may be more likely to use the mHealth app instead of a social network app for this purpose [19].In fact, a previous study [39] found a similar pattern in which higher extraversion in students was associated with a preference for face-to-face mental health services over eHealth services.
The positive association between openness to experience and patient engagement in mHealth was consistent with the results of previous research [21] on technology adoption.High openness to experience is reflected in curiosity and novelty-seeking-open XSL • FO RenderX individuals are willing to try new and different things, are willing to actively seek out new and varied experiences, and value change [15].At the same time, some research has also found that openness plays an important role in promoting healthy behavior and lowering mortality for patients with chronic disease [40,41].While mechanisms underlying these findings are not well understood, they are consistent with the notion that openness should be associated with the adoption of novel self-management tools [21] and associated with better health behavior adherence [22].This means that people more open to experiences may be more likely to adopt and actively use an mHealth app for diabetes management.
The analysis of our findings showed that conscientiousness, agreeableness, and emotional stability did not have a significant relationship with acceptance of the mHealth app.Of particular note, conscientiousness, which reflects self-discipline, did not have a significant role in active app use, which was inconsistent with results of a previous study focusing on a weight loss app [23].This unexpected finding underscored the need for future research, but differences could be attributed to different population characteristics.Perhaps, our study's sample of older participants had less prior experience with mHealth apps; therefore, open-mindedness was an especially important predictor of app adoption and active use.There have also been studies [42] suggesting that conscientious patients were more likely to fulfill tasks in accordance with existing strategies, whereas using new technologies could be viewed as more time-consuming and complex than the traditional methods they already use.
Lastly, our pilot study offered support that mHealth improved clinical outcomes of patients with diabetes.This reduction was on the magnitude of a 0.6-point reduction in HbA 1c level, which was sizable considering the relatively short intervention period; however, we also noticed that the HbA 1c level of participants who dropped out increased while the HbA 1c level of nonadopters who never used the app decreased, even though there was no significant difference between the two groups.This may be related to other differences between dropouts and nonadopters; patients with good daily self-management behavior may be less interested in adopting the mHealth app.Meanwhile, patients who had poor self-management behavior may have started using the app to improve their health, yet may not have continuously used the app or followed its guidance, and their HbA 1c level may have increased or remained constant.Because this study was not randomized, readers are cautioned against interpreting the results as causal.

Implications
To the best of our knowledge, this is the first empirical paper to investigate how different personality traits influence the adoption and the active utilization of an mHealth app intervention for patients with diabetes.Our research extends the literature on the adoption of mHealth apps and active utilization to improve diabetes management by exploring the role of personality traits.This work has implications for app designers and practitioners who can leverage this knowledge to target individuals who are most likely to succeed.For instance, patients who are less extraverted and more open to experience may find an app-based intervention the most appealing and effective.With the development of machine learning models, then, an app designer may be able to predict mHealth app user personality traits using easily accessible data and in real time [43,44].Consequently, designers and practitioners can create or administer personalized mHealth services in a way that enhances patient engagement with helpful apps and which ultimately improves their lifestyle and health.

Limitations
Caution must be shown in generalizing the findings of this work because it has several limitations, but these limitations also provide opportunities for future research.First, the study used a nonrandomized experimental design in which only those who indicated their willingness to use the mHealth app were placed in the intervention group.While this allowed us to answer research questions about mHealth adoption preference, it made assessment of app efficacy more challenging.In future, a hybrid preference-randomized controlled trial [45] would enable examination of both questions.Second, although all participants, no matter which intervention they chose, were told that those who decrease their HbA 1c level during the 3-month period would receive a gift as an incentive, we cannot exclude the possibility that this incentive may have influenced the efficacy of the mHealth app intervention.Building on the encouraging findings from this pilot, future studies may use more rigorous research designs to address this potential impact.Third, as a pilot study, the sample was relatively small in size and came from one hospital in one city.Increasing and diversifying the sample size would provide more confidence that these results are generalizable.

Conclusions
Although there has been some research about how personality traits impact the use of new technology, there is relatively little that focuses on understanding the impact of individual characteristics, including personality traits, on mHealth app adoption and active utilization among diabetic patients.Our pilot study has made a strong start in addressing this gap by extending the mHealth literature.The study revealed that diabetic patients who are relatively young, introverted, and open to experience were interested in and willing to use the app.Moreover, active use of the app was associated with greater improvements in blood glucose level control.These research findings may have practical effects on the future development of mobile health apps for patients with diabetes.

XSL • FO
RenderX information, a link to the original publication on http://mhealth.jmir.org/,as well as this copyright and license information must be included.

Figure 2 .
Figure 2. Screen capture images from the DiaSocial app.

23 Dropouts
Patients who continued to use the app into the third month.23 Active users Patients whose average weekly app usage was less than 3.5 days 14 Low-frequency Patients whose average weekly app usage was more than 3.5 days 9 High-frequency a Includes Nonadopters: With intention but never used, and Adopters.

Table 1 .
Definitions of participant categories.Patients who were not willing to use the app at the beginning of the project.41 No intention to use Patients who indicated willingness to use the app, but then did not use it at all.

Table 2 .
Logistic regression for personality traits associated with adoption preference.
a OR: odds ratio.bN/A: not applicable.

Table 3 .
Logistic regression for personality traits associated with active utilization.
a OR: odds ratio.bN/A: not applicable.

Table 4 .
Descriptive statistics after 3 months and the differences among categories (N=66).