According to technology adoption theorists, determinants of user perception shaped by prior experience with mobile health apps can predict likelihood of technology adoption and intention to use that technology (Davis, 1989; Guner & Acarturk, 2020; Ma et al., 2021; Zhao et al., 2018). This theoretical perspective, also known as Technology Acceptance Model (TAM) (Davis, 1989), suggests that perceived ease of use and perceived usefulness of mHealth apps are likely to increase favorable attitude toward and willingness to use them. Since the introduction of TAM (Davis, 1989), numerous scholars have reported evidence of its key theoretical premises: the easier the user experience with technology is, the more useful the technology is to the user and the more favorable attitude toward the technology is; the more useful the technology is to the user and the more favorable attitude toward the technology is, the greater intention to use the technology is. This key principle of technology adoption varies based on other individual and social conditions (e.g., physical, psychological, and social capacities) (Binyamin & Zafar, 2021; Cakmak & Mantoglu, 2021; Hirose & Tabe, 2016). A fundamental question arising from this observation is whether the technology itself, rather than individual experience with the technology, might shape or facilitate exercise behavior.

According to motivational technology theorists, on the other hand, determinants of user perception facilitated by technological affordances (i.e., user action possibilities afforded by technology such as a photo upload feature on the app that supports customization) (Evans et al., 2017; Sundar, 2008), prompt users to take action and achieve goals while generating feelings of control and confidence (Sundar et al., 2012). Mobile health apps are popular examples of motivational technologies. Scholars have examined the effectiveness of motivational technology in driving users to exercise regularly, to eat healthy food, to take medicine as prescribed, to schedule medical check-ups in a timely manner, and to engage in various other activities that promote health and fitness (e.g., Doherty et al., 2018; Molina & Sundar, 2020; Wu et al., 2019). According to the motivational technology model (MTM), the design of technology, along with the information transmitted, leads people to use the technology and establish healthy routines (Sundar et al., 2012). Unlike the technology acceptance model, which explains technology adoption based on perceived usefulness and perceived ease of use (Davis, 1989; Talukder et al., 2019), the primary theoretical drive of MTM is persuasive technology (Fogg, 1998, 2009), an “interactive system designed to aid and motivate people to adopt behaviors” (Orji & Moffatt, 2018, p. 66). In this way, the technological features of mobile health apps can shape behavior by extending the actions that users can take, delivering content in engaging ways, and enabling interaction with other users (Fogg, 1998).

The mechanism behind motivational technology is user engagement, a process of media interaction involving cognitive assessment and psychological and behavioral reaction as user experiences accumulate (J. Oh et al., 2018). According to J. Oh et al. (2018), user engagement with technology consists of four dimensions. The first dimension is physical interaction, which enables users to engage with communication technology. This dimension of user engagement is distinct from engagement with media content. While people tend to consume media content passively, engagement with communication technology requires active initiation of interaction with an interface. Depending on how likely postpartum women are to interact voluntarily with a mobile health app for diet and fitness, the degree to which they achieve a desired goal (e.g., postpartum weight loss) is likely to differ. The second and third dimensions of user engagement are interface assessment (i.e., the initial evaluation of the interface as natural, intuitive, and easy to use in a heuristic rather than elaborative manner) and absorption (i.e., a psychological state of being immersed in the environment without a sense of time) in the technology interface, which naturally occur during physical interaction with the technology device (Oh et al., 2018, p. 739). In other words, the more postpartum women find the apps easy, fun, and interesting to use after user-initiated interaction, the more likely they are to continue using them. Finally, user engagement with a technology interface involves digital outreach, which is a behavioral intention shaped by technological features associated with relatedness (Sundar et al., 2012). Given the social interaction and personal record keeping readily afforded by mobile communication devices (Lee & Cho, 2017), mobile diet and fitness app users can share their experience with an app with other users (Talukder et al., 2019) when the first three stages of app engagement have been positive experiences (J. Oh & Kang, 2021).

Previous findings document wide use of mobile health apps among women during (perinatal) and after pregnancy (postpartum) (Fernandez et al., 2017). Results from an online survey with 509 women in the perinatal period showed that mobile devices were most popular for the Internet access. A majority of the respondents (72.3%) downloaded mobile apps, and more than half of the apps downloaded were health related (Osma et al., 2016). Despite the availability and popularity of mobile health apps among postpartum women, however, whether mobile health app use has contributed to actual health improvement among postpartum women remains unknown. For example, a systematic review of 36 articles published between 2011 and 2016 revealed that few scholars had effectively and accurately measured the quality and usability of health apps although many mobile health apps targeted a wide range of public health concerns (e.g., general health fitness, physical activity, smoking) and chronic and serious illnesses (e.g., cancer, heart disease, asthma) (McKay et al., 2018). Similarly, a content analysis of 87 mobile apps designed for weight management during pregnancy revealed that most apps designed for pregnant women offered users a simple weight tracking feature, and only a small percentage of the reviewed apps offered guidelines for weight management (n = 19), calorie recommendation (n = 8), or exercise recommendation (n = 7) (Dahl et al., 2018).

Findings from a few previous studies imply the potential of weight-loss intervention using mobile health apps. For example, through a 14-week randomized weight-loss trial for low-income postpartum women that offered (a) individually tailored recommendations and guidelines for healthy weight management and physical activity via text messages, (b) biweekly counselling calls from a health coach, and (c) access to Facebook support groups, Herring et al. (2014) found that the intervention yielded greater weight loss than traditional postpartum care. Similarly, a 16-week weight-loss intervention for low-income postpartum women delivered through Facebook was effective for postpartum women with limited access to traditional weight-loss programs. The intervention retained participants at 89% (pilot test 1), 83% (pilot test 2), and 86% (pilot test 3). The average rates of active engagement in the program ranged from 55% (pilot test 2) to 67% (pilot test 3) (Silfee et al., 2018). Another Facebook intervention for weight loss (i.e., 19 postpartum women over 12 weeks) involved the Diabetes Prevention Program. The program, delivered via a private Facebook group, provided the participants a variety of informational content: nutritional information for breastfeeding women, recipes for kids, and links to exercise videos, self-care, and solutions for lifestyle challenges common to postpartum women. The intervention resulted in the loss of 4.8% of the baseline weight, and 58% of participants lost more than 5% of the baseline weight. The findings suggest the efficacy of weight loss intervention through Facebook for postpartum women (Waring et al., 2018).

While previous findings suggest the effectiveness of alternative postpartum weight management, particularly through social media, they neither explain how the apps worked nor reflect real app use outside the study setting. These limitations call for a theory-driven research model of the relationship between diet and fitness app use and exercise behavior and weight management among real app users during the postpartum period.

Initially, Oh and colleagues (2018) tested the model of user engagement with communication technology using website users in a controlled experiment setting. They found that both physical interaction and interface assessment predicted user absorption and digital outreach (i.e., continuum model of user engagement) given the sequence of engagement with a website: from initial interaction with the website and simultaneous assessment of the website interface to subsequent immersion in the website and connection with other users (p. 745). The four dimensions of user engagement generated independent effects on outcome variables (e.g., attitude toward the website interface, attitude toward the learned content, and content recall) rather than linear cumulative effects. Interface assessment, absorption, and digital outreach led to favorable attitude toward the website interface, interface assessment and digital outreach led to favorable attitude toward the learned content, and physical interaction facilitated content recall (J. Oh et al., 2018). Later, Oh and Kang (2020) made the model more applicable to the interfaces of motivational technology (i.e., smartwatches and smart fitness trackers) by confirming a more linear process of user engagement. However, their process model of user engagement did not include behavioral outcomes of motivational technology use.

In fact, the direct impact of user engagement with diet and fitness apps on exercise behavior and subsequent weight management among postpartum women remains unclear. For other populations, scholars have reported weight-related outcomes of user engagement with apps. A review of 24 studies about the relationship between engagement with mobile health apps and weight management revealed inconsistent results. Findings from more than half of the studies show a positive relationship between user engagement and fat loss while others show no such significant association (Spaulding et al., 2021). The positive associations depended on additional sources of weight management, such as professional human coaching (Bennett et al., 2018; Johnston et al., 2013; Patel et al., 2019), and no associations emerged for participants with health conditions such as obesity or diabetes (Godino et al., 2016; Koot et al., 2019; van Beurden et al., 2019). In addition, overall weight loss did not differ between the intervention group (multiple technology access in addition to human coaching) and the control group in a 24-month trial, though weight loss was greater in the intervention group through the first 12 months (Godino et al., 2016). van Beurden et al. (2019) also found greater weight loss in the intervention condition (i.e., ImpulsPal - app-based coaching with no human interaction) than the control condition only during the first month, not after the 3-month trial. A similar pattern emerged in a large-scale study of three different countries in Europe: Switzerland, the United Kingdom, and Germany (N = 19,211); higher engagement with an app designed for obese patients led to more weight loss during the first three months, but this effect did not persist after the full six months (Lehmann et al., 2024).

Furthermore, despite some evidence that app engagement promoted weight loss (e.g., Bennett et al., 2018), the role of exercise in this relationship among postpartum women remains unclear. For general adult app users, some findings suggest the benefits of using mobile health apps to self-monitor exercise behavior in combination with personalized feedback about workout progress. After the intervention, time spent exercising and frequency of self-monitoring were higher among participants who used the app (Voth et al., 2016). Serrano et al. (2017) also found that higher engagement with the app (i.e., Lose it!) related to a higher number of custom dietary and exercise logs.

Building on previous findings, we proposed a model to test user engagement (J. Oh et al., 2018; Oh & Kang, 2021) with mobile health apps to address how that engagement might motivate postpartum women to exercise and manage their weight. As previous findings suggest, if postpartum women engage with a mobile diet/fitness app because the design features motivate them to move through the stages of user engagement (i.e., initiating interaction with the technology and assessing the app interface [H1], absorption in the app [H2], and digitally outreach to other users [H3]) (Oh & Kang, 2021), they are more likely to engage in exercise (H4) (Serrano et al., 2017; Voth et al., 2016). Finally, previous findings show a positive association between exercise and weight management among postpartum women (Hanley et al., 2022; Vernon et al., 2010), some idicating the related but distinct roles of exercise amount and exercise intensity (Mascarenhas et al., 2018). For example, longitudinal data about obstetric outcomes at hospital clinics located in the Midwestern (1,003 women over 6 weeks of pregnancy and postpartum care) revealed that vigorous exercise predicted favorable postpartum adaptation in the form of psychological well-being (Sampselle et al., 1999). Another systematic review of 76 studies about the effects of physical activity on maternal health during pregnancy and postpartum revealed that moderate-intensity physical activity lowered the risk of excessive gestational weight gain while little evidence emerged for the role of simple physical activity in postpartum weight loss regardless of exercise amount (DiPietro et al., 2019). Given the uncertain impact of exercise quantity and quality on weight outcomes, we proposed research questions to address the relationship between exercise amount (RQ1) and exercise intensity (RQ2) on the final weight status of postpartum app users.

We obtained ethical approval from the Institutional Review Board of the first author’s university and obtained informed consent prior to study participation. Participants were recruited through the Qualtrics panel service (Qualtrics, 2021) . After consenting to participation, only those who met the following criteria were eligible to complete the survey: currently residing in the United States, identifying as female, being 18 years or older, having a child aged no more than 5 years, and being a current user of at least one mobile diet and fitness app. Eligible participants answered a series of questions regarding their use of mobile diet and fitness apps, their experiences and engagement with these apps, duration and intensity of exercise, and height and weight information at the time of the study. We also measured dietary restraint, length of breastfeeding, length of postpartum, and history of app use as control variables and collected demographic data (i.e., race/ethnicity, income, education, marital and employment status, and current smoking status). Responses from those who failed the attention check questions during the survey (e.g., failed to answer the name of an object in an image shown or the name of the capital of the United States) were automatically removed to ensure the quality of the data. A total of 289 responses were retained for main data analysis after excluding 13 participant responses that contained invalid or suspicious height and current weight entries (e.g., less than 3 feet with more than 170 pounds).

The average age was 31.4 years old (SD = 6.64), and most participants were White (73%), followed by Black (13.8%), Hispanic (6.6%), and Asian (5.9%). More than 70% of the participants reported that that the age of their youngest child age was 3 years old or younger. About 80% of the participants breastfed their youngest child, and the average length of breastfeeding was 11.5 months (SD = 8.56). The majority of the participants (76.1%) had been using at least one mobile diet/fitness app for more than 6 months, while 23.9% of the participants had been using one for fewer than 3 months (23.9%). Table 1 summarizes detailed information about the participants’ demographics and mobile app usage.

We examined univariate normality of all measured variables before the main analysis. We found two variables that were problematic due to abnormality of data distribution based on skewness and kurtosis less than −2.0 or greater than 2.0: physical interaction and total hours of exercise per week (Tabachnick & Fidell, 2006). First, we excluded 36 participants with physical interaction that exceeded the normal range in the data (i.e., 11.82 [mean] +13.48 [standard deviation] = 25.3 physical interactions per day). We also excluded six additional cases of less than one physical interaction. We excluded one participant based on a suspicious response to total hours of exercise per week (i.e., 41 hours). Thus, 245 cases comprised the final data set for the main analysis. Because the variable matrices were not identical (ratio or interval), we used standardized Z-scores of all univariates, including the covariate (i.e., retrained eating tendency) (Tabachnick & Fidell, 2006).

We performed confirmatory factor analysis to establish construct validity of the user engagement measure using AMOS 22.0 (J. Oh et al., 2018; J. Oh & Kang, 2021). The data fit the original user engagement with four factors, χ2 = 112.06, df = 68, p < .01; CFI = .98; TLI = .97; RMSEA = .05, 90% CI:.03 −.07 (Kline, 2005). The standardized loadings for all indicators were higher than the absolute value of .70, ranging from .72 to .87, except for two items (physical interaction: “How many times do you actually tap things on the diet/fitness app(s) per day?, α = .67; absorption: “Interacting with the diet/fitness app(s) makes me curious,” α = .66). See Table 2 for all indicators of the four factors in the user engagement scale, factor loadings, Cronbach’s alphas, and means and standard deviations.

Due to the insignificant path that emerged between physical interaction and interface assessment in the process model of user engagement, we modified the proposed model according to previous findings. Because the process model of user engagement tested by J. Oh and Kang (2020) involved data about wrist-worn device users (e.g., Apple Watch, Fitbit), we speculated that those wearable devices had characteristics that might influence user experience. In fact, smartwatches can operate with stand-alone apps on the device or companion apps on a smartphone (Chen et al., 2021). This fundamental difference between apps on smartwatches and apps on smartphones might force smartwatch users to engage in physical interaction to assess the interface. However, smartphone app users might not need physical taps on mobile apps to assess the interface due to prior experience with other apps. Because participants were asked to report their experience with mobile apps retrospectively, the sequence of actions required to assess the app interface might differ from the sequence required for a smartwatch interface. Thus, we modified the proposed model to account for the original continuum model of user engagement (J. Oh et al., 2018).

Follow-up path analysis of the modified model revealed slightly better fit to the data (χ2 = 42.25, df = 23, p < .01; CFI = .97; TLI = .92; RMSEA = .06, 90% CI: .03-.09, N = 245) (see Figure 2). The paths between physical interaction and absorption (b = .13, SE = .04, p < .001) and between interface assessment and absorption (b = .78, SE = .04, p < .001) were significant, along with the rest of the paths found to be significant in the initial model (absorption → digital outreach: b = .88, SE = .06, p < .001; digital outreach → hours of exercise per week: b = .13, SE = .06, p < .05; digital outreach → workout intensity: b = .19, SE = .06, p < .01; workout intensity →current BMI: b = −.26, SE = .06, p < .001), except for the path between total hours of exercise per week and current BMI.

Table 4 reports the zero-order correlations among the variables included in the path models.

Our findings reaffirm the promising role of user engagement with diet and fitness apps in facilitating exercise behavior and assisting women in losing gestational weight during the postpartum period. The findings offer evidence that links the theoretical framework of user engagement to postpartum women who need technology to help them set exercise routines and maintain a healthy weight. More importantly, we found that the dynamics among the user engagement dimensions differed across technological devices. The original linear process model of user engagement suggested by J. Oh and Kang (2020) shows that the four dimensions of user experience with wearable fitness tracking devices (e.g., Apple Watch and Fitbit) proceeded in a single linear chain of user experience. However, the findings from our study show that the linear process of user engagement might not occur during interaction with mobile apps on smartphones, for the two interface-level user engagement dimensions (i.e., physical interaction and interface assessment) showed no correlation (see Figure 1).

Differences in purpose of physical interaction with technology devices might explain the different results of our extended user engagement model from the model in previous study (J. Oh & Kang, 2020). The types of action that prompt physical interaction with smartwatches (i.e., checking, tapping, and updating) are mandatory and frequent, given the small screen, for interface assessment (Chen et al., 2021). However, physical interaction might not be always required for interface assessment when using companion apps or stand-alone diet and fitness apps on a smartphone, which was the primary device type in our study. Checking, tapping on, and updating diet and fitness apps might be entirely independent of app interface assessment because these physical actions while using a smartphone fall into the category of “a checking habit” (Oulasvirta et al., 2012, p. 105). Smartphone users develop a checking habit when (and because) they feel a compulsion to check for updates (e.g., push notifications). As habitual checking on mobile apps grows, users learn how to allocate cognitive resources efficiently, except in the case of smartphone addiction; hence, they are likely to optimize their capacity to obtain information instead of spending cognitive resources on assessing an interface with which they are already familiar (Oulasvirta et al., 2012).

Likewise, the fitness app users might have already done an interface assessment based on myriad everyday interactions with various apps. App users might perceive an app interface as natural, intuitive, and easy to use because they have been using apps for a while. Moreover, habitual checking on smartphones oftentimes occurs without any purposeful or intended outcomes as smartphone use increases (Lukoff et al., 2018). In this regard, J. Oh and Kang (2020) did not measure familiarity with smartwatch interfaces beyond the length of time users had been using their smartwatches. Moreover, they included data from participants whose history of smartwatch and smart tracking device use was at least one month. Therefore, whether the participants had owned and interacted with their smartwatches long enough to feel as familiar with the interface as they might with a smartphone is uncertain.

While the paths among all four factors of user engagement were significant in the modified model, relatively weak coefficients from physical interaction to absorption and from digital outreach to the first behavioral outcome (i.e., total hours of exercise per week) suggest a need for app designs that might improve the effects of diet and fitness apps on postpartum weight management. User experience with apps related to absorption and digital outreach might be akin to elements of the app interface that maximize hedonic and social motivations (J. Oh & Kang, 2021; Putri et al., 2019). For example, Edney et al. (2019) developed a mobile app for physical activity to test the effect of gamification features on app use, attrition rate, and amount of physical activity. They found that gamification features facilitated user engagement better than the basic app design, producing a correlation between the frequency of app use (higher engagement) and frequency of physical activity. We also strongly recommend mobile app designs that improve social connectedness in order to motivate digital outreach among app users. Social support is an influential factor in the quality of postpartum experience (Baker & Yang, 2018) and level of physical activity (DeLuca & Bustad, 2017). In particular, participation in an online weight loss program that was predominantly popular among women (average age of 37 years old) increased when social support cues (e.g., encouraging remarks made by other users) were highly available (Hwang et al., 2014).

While we found a significant relationship between user engagement and exercise quantity and quality, only exercise quality (i.e., workout intensity) affected postpartum weight. This finding aligns with previous findings about the importance of vigorous or moderate intensity of exercise to postpartum weight loss (DiPietro et al., 2019; Sampselle et al., 1999). Therefore, fitness apps that promote at least medium-intensity workout programs should help postpartum women more than simple nudges to exercise a certain amount of time every day. Apps that couple intense workout routines with hedonic and social elements (e.g., competition with other users or challenge badges to earn every week) can optimize the role of app engagement in postpartum weight management.

The limitations of the current study open pathways to future study. First, the data we used consisted of self-reported user recall of frequency of interaction with apps per day and hours of exercise per week. Thus, self-reported data might introduce social desirability bias or recall errors (Adams et al., 2005). Because of this measurement limitation, exclusion of more than 40 cases from the initial data was necessary to avoid inaccurate analysis. A live log of data entries collected through a prototype mobile app might be a methodological solution that ensures complete protection of user privacy (e.g., Toro-Ramos et al., 2021). Related to self-report responses, online panel sampling raises concerns about data reliability (Belliveau & Yakovenko, 2022), despite multiple layers of eligibility screening to ensure quality data. In future studies, scholars should consider direct recruitment from postpartum care clinics or related organizations, such as the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC), emulating previous randomized controlled trials (Herring et al., 2014; Phelan et al., 2017; Sampselle et al., 1999).

Second, our data included relatively larger percentages of overweight (32.9%) and obese (24.6%) participants. Previous findings suggest that weight management and weight loss are likely to differ significantly between postpartum women within a healthy weight range and people within a range considered overweight or obese. For example, the gestational weight gained during pregnancy played a significant role in the stress of postpartum women, increasing the likelihood of postpartum weight retention among overweight women (Leonard et al., 2021). In another study, users within a healthy range of BMI reported a higher level of app use than those who were overweight or obese, regardless of app features designed to promote app use (e.g., game elements) (Edney et al., 2019). Therefore, our findings might not accurately reflect the use of diet and fitness apps for exercise routines and postpartum weight management across different weight groups - differences which often coincides with disparities in mobile health app access and usage among postpartum women from underrepresented populations, such as those from low socioeconomic backgrounds or minority racial groups (e.g., Napolitano et al., 2021; Phelan et al., 2017), calling for future research.

We also acknowledge the limitation of using cross-sectional data to confirm causal links between user engagement, exercise behavior, and weight status. Although we controlled for history of app use and length of postpartum to ensure the validity of causal inferences in our model, as guided by a theoretical model (Stage et al., 2004), path analysis has its own statistical limitations in demonstrating causality (Bollen & Pearl, 2013). Additionally, cross-sectional data cannot provide evidence of weight loss or BMI change from gestational to postpartum period, a change that might be the ultimate goal of postpartum weight management. Scholars should aim to collect longitudinal data to identify causality between app engagement and exercise and weight management behaviors among postpartum women.

In closing, physical activity on a regular basis and proper weight management are crucial to women’s health, especially during the postpartum period (Harrison et al., 2016; Leonard et al., 2021). Showing a positive relationship among engagement with diet and fitness apps, exercise behavior, and favorable weight management in postpartum women, our findings can help current postpartum care programs enhance their effectiveness. Clinics and community postpartum care programs would benefit from timely implementation of mobile apps in their existing programs and encouragement of postpartum women to continue using apps for postpartum weight management after participation in care programs.