Abstract
Objectives
Mindfulness meditation apps are used by millions of adults in the USA to improve mental health. However, many new app subscribers quickly abandon their use. The purpose of this study was to determine the behavioral, demographic, and socioeconomic factors associated with the abandonment of meditation apps during the COVID-19 pandemic.
Method
A survey was distributed to subscribers of a popular meditation app, Calm, at the start of the COVID-19 pandemic in March 2020 that assessed meditation app behavior and meditation habit strength, as well as demographic and socioeconomic information. App usage data were also collected from the start of each participant’s subscription until May 2021. A total of 3275 respondents were included in the analyses. Participants were divided into three cohorts according to their subscription start date: (1) long-term subscribers (> 1 year before pandemic start), (2) pre-pandemic subscribers (< 4 months before pandemic start), and (3) pandemic subscribers (joined during the pandemic).
Results
Meditating after an existing routine was associated with a lower risk of app abandonment for pre-pandemic subscribers (hazard ratio = 0.607, 95% CI: 0.422, 0.874; p = 0.007) and for pandemic subscribers (hazard ratio = 0.434, 95% CI: 0.285, 0.66; p < 0.001). Additionally, meditating “whenever I can” was associated with lower risk of abandonment among pandemic subscribers (hazard ratio = 0.437, 95% CI: 0.271, 0.706; p < 0.001), and no behavioral factors were significant predictors of app abandonment among the long-term subscribers.
Conclusions
These results show that combining meditation with an existing daily routine was a commonly utilized strategy for promoting persistent meditation app use during the COVID-19 pandemic for many subscribers. This finding supports existing evidence that pairing new behaviors with an existing routine is an effective method for establishing new health habits.
Preregistration
This study is not pre-registered.
Keywords: Mindfulness meditation, Mobile app, mHealth, App abandonment, Habit cues, Survival analysis
Mindfulness meditation is currently used by millions of adults in the USA to reduce psychological symptoms from stress, anxiety, and depression and to increase overall well-being (Bostock et al., 2019; Eberth & Sedlmeier, 2012; Edenfield & Saeed, 2012, 2012; Lacaille et al., 2018). The recent development of mindfulness meditation mobile phone apps has made meditation more accessible to the general population and contributed to the growing popularity and practice of meditation. Similar to other health-promoting daily behaviors though, the benefits of meditation are primarily experienced through persistent practice over time (Shen et al., 2020; Tang et al., 2012), and many new meditation app subscribers quickly stop (or “abandon”) their use. Dropout rates in meditation app interventions typically range from 21 to 54% (e.g., Goldberg et al., 2020; Huberty et al., 2019; Puzia et al., 2020), and in general, the daily use of health apps among paying subscribers is less than 4% (Kerst et al., 2020). Thus, there is a need to better understand the determinants of mobile meditation app abandonment in order to design new behavioral tools and interventions that can promote more persistent use of meditation apps.
One important behavioral determinant of meditation app abandonment is the formation of a strong daily habit, which has been shown to increase the persistent use of meditation apps and prevent app abandonment (Stecher, Sullivan, et al., 2021; Wood & Neal, 2016). Psychology research has defined habits as automatic or reflexive behavioral responses to environmental cues (Gollwitzer, 1999; Wood & Neal, 2007), where cues can be external, such as a visual reminder, or internal, such as physical sensations or the completion of the proceeding action in one’s daily routine (Champion & Skinner, 2008; McArthur et al., 2018). The use of environmental cues to trigger daily behaviors has been shown to support a wide range of healthy habits, such as hand washing, flossing, medication adherence, and meditation (e.g., Burner et al., 2014; Hussam et al., 2017; Judah et al., 2013; Lally & Gardner, 2013; Saghafi-Asl et al., 2020; Stecher, Mukasa, et al., 2021). One successful strategy for developing a new habit is to pair the desired behavior (e.g., meditation) with an existing daily routine, which serves as the consistent environmental cue to trigger the desired behavior (Wood & Neal, 2016). This strategy, called “anchoring,” has been used to help individuals establish strong habits in various health behavior contexts (e.g., Armitage & Arden, 2008a, 2008b; O'Carroll et al., 2013). In one recent study, meditation app subscribers who were instructed to anchor their meditation practice into an existing daily routine (i.e., environmental cue) were more likely to continue using the app over an 8-week follow-up period than those who did not use cues (Stecher, Sullivan et al., 2021). This behavioral intervention approach for establishing persistent meditation app use was not successful for many participants though, so the role of environmental cues in preventing meditation app abandonment is still unknown. It is also unclear what type of environmental cues, such as using an alarm or an existing daily routine, are the most supportive of meditation app habits and thus the most protective against meditation app abandonment.
Another behavioral determinant of persistent health habits is the time of day of behavioral performance, which in turn may have an important influence on meditation app abandonment. It has been shown that some health behaviors are more persistently performed if completed in the morning (Kouchaki & Smith, 2014; Pignatiello et al., 2020; Stecher, Sullivan, et al., 2021). In the recent study on meditation app habits by Stecher et al. (2021a, b, c), those who meditated in the morning were more likely to persistently meditate than those who meditated at other times of the day. However, another recent study found that those who meditated at the same time each day were more likely to persistently meditate regardless of what time of the day they meditated (Stecher, Berardi et al., 2021). Thus, it is unclear if and how the time of day of meditation app use is associated with meditation app abandonment.
An important additional determinant of meditation app abandonment was the COVID-19 pandemic, which significantly disrupted many individuals’ and households’ daily routines and health habits (Burner et al., 2014; Carden & Wood, 2018; Carroll et al., 2020; Lally & Gardner, 2013; Souza et al., 2021). Whether it was working from home, providing childcare, or taking care of sick family members, the disruption of daily routines due to the COVID-19 pandemic may have contributed to greater meditation app abandonment. Since habits are formed over time, the impact of the COVID-19 pandemic on meditation app abandonment may depend on how long subscribers had been using the app prior to the start of the pandemic. Therefore, we examined the impact of the pandemic on meditation app abandonment separately among three cohorts of app subscribers based on their length of app subscription. In addition to the COVID-19 pandemic and behavioral determinants of app abandonment, users’ demographics and socioeconomic status may be associated with meditation app abandonment. For example, existing research has observed the greatest meditation app use among White, educated women (e.g., Goldberg et al., 2020; Huberty et al., 2019; Stecher, Berardi, et al., 2021). Given the importance of these characteristics for understanding aggregate app use, this research will examine how demographic and socioeconomic factors may also be indicative of meditation app abandonment.
The purpose of this study was to determine the behavioral, demographic, and socioeconomic factors associated with meditation app abandonment during the COVID-19 pandemic. For this research, subscribers to a popular meditation app called “Calm” were divided into three cohorts based on their likelihood of having established a meditation habit at the onset of the COVID-19 pandemic: (1) those who started their subscription at least 1 year before the pandemic (before March 2019), (2) those who started their subscription just before the World Health Organization (WHO) declared COVID-19 a pandemic (March 11, 2020) (i.e., November 2019–February 2020), and (3) those who started their subscription during the pandemic (March–May 2020) in order to examine how the COVID-19 pandemic differentially impacted these three groups.
Method
Participants
Our study examined data collected as part of a larger longitudinal study assessing the mental health and health behaviors during the COVID-19 pandemic of a US-based sample of meditation app subscribers (Green et al., 2021). Subscribers included in the study had first subscribed to the Calm meditation app at least 90 days prior to study enrollment, but there was no limit placed on how long someone’s app subscription could have been prior to enrollment (e.g., subscription lengths varied from 90 to over 2000 days prior to the pandemic). In the larger longitudinal study, a series of six surveys were distributed from April 2020 to May 2021, with the final survey administered 12 months from the baseline survey. Participants were recruited on a rolling basis between April and May 2020. App usage data were collected for all survey respondents from the time of their subscription until the final 12-month survey. The data analyzed in this paper include all available app usage data and survey data obtained from the baseline survey (administered in April–May 2020). The baseline survey occurred, on average, 5 months (SD = 11.14) after respondents first subscribed to the Calm app.
Procedures
All study procedures were conducted online. Calm subscribers were emailed an invitation to participate in the study, which was advertised as the “COVID-19 Health and Well-being Survey.” The email contained information about the study timeline and a link to a 1-min eligibility survey to be completed in Qualtrics (2020). The Calm data team sent this recruitment email to subscribers if they had opened an email from Calm at least once in the last 90 days and used Calm at least once in the last 90 days. Calm subscribers were then determined eligible to participate if they indicated they (1) were at least 18 years old, (2) could read and understand English, and (3) lived in the USA. If eligible, subscribers were directed to an electronic informed consent form, and consenting subscribers then completed the baseline survey.
Participants’ subscription start dates were obtained from the Calm data team. We used these start dates to divide participants into three cohorts based on their likelihood of having established a meditation habit at the onset of the COVID-19 pandemic: (1) long-term subscribers (i.e., subscription state date on or before March 2019), (2) pre-pandemic subscribers (i.e., subscription start date between November 2019 and February 2020), and (3) pandemic subscribers (i.e., subscription start date between March and May 2020). In the long-term subscriber cohort, subscription start dates ranged from July 2014 to February 2019. Since Calm only offers annual subscriptions, the long-term subscribers had all renewed their annual meditation app subscription at least once before the start of the data collection period. Conversely, the annual subscription renewal dates among the pandemic subscriber cohort occurred towards the end of our data collection period, so app abandonment among our three cohorts represents different levels of payment(s) and commitment to the meditation app. The duration of meditation app subscription was analyzed through these stratified analyses because we did not hypothesize that app abandonment was continuously related to the length of participants’ app subscription.
App usage data were analyzed for all study participants from their start date until May 2021. All app usage data were obtained via the Calm data informatics team.
Measures
The baseline survey contained investigator-developed questions that assessed meditation app behavior as well as demographic and socioeconomic information. Specifically, respondents were asked to identify what type of environmental cue they used to trigger their meditation (e.g., reminder, alarm, existing routine behavior) by responding to the question, “Currently, what is the primary initiator of your meditation practice?” Respondents selected from a list of options: time of day, alarm, part of a daily routine, specific emotions, physical sensations, whenever I can, reminder, or other. Respondents were asked to identify what time of day they typically meditated using the app by responding to the question, “Currently, what time of day do you typically meditate?” Respondents selected from a list of options: morning (wake–11 am), afternoon (11 am–4 pm), or evening (4 pm–sleep). Demographic and socioeconomic status information collected included gender, age, race/ethnicity, education level, income, and employment status.
Data Analyses
Cox proportional hazards regression models were estimated to assess the relationships between app abandonment and (1) meditation practice environmental cues, (2) meditation practice time of day, and (3) demographic and socioeconomic variables. We estimated three regression specifications that increasingly incorporated each of these variables to better understand the sensitivity of the main results, where the final model (Model 3) included all available behavioral, demographic, and socioeconomic predictors. The reference variable in each set of predictors refers to the omitted variables from the regression analyses. The coefficients for the included groups are the difference in the outcome (i.e., app abandonment) between each included group and the reference group (e.g., time of day compared to reminder).
The date of app abandonment was defined as the last meditation session completed for each respondent, which was used to calculate the days until app abandonment. This measure was considered unobserved (or censored) if the date of app abandonment occurred on or after April 1, 2021, which was 30 days before the end of our data collection period. That is, even if participants did not record a meditation session after April 1, 2021, we did not consider them to have abandoned the app because they could have continued with their practice in June 2021 or later. Statistical significance was set at p < 0.05. All analyses were conducted in Stata and SPSS (IBM, 2020; StataCorp, 2021).
Results
A total of 8386 respondents completed the baseline survey. After removing any cases with missing app usage data or those who never completed a meditation session with the app, 3275 respondents were included in the analyses. Respondents’ subscription start dates ranged from July 27, 2014, to May 21, 2021. A total of 1468, 513, and 1294 respondents were included in the pre-pandemic subscriber, long-term subscriber, and pandemic subscriber cohorts, respectively. Respondents with subscriptions that did not occur within the dates that comprised the three cohorts (i.e., April–October 2019) or with missing start date data were not included in the analysis (n = 1646).
Table 1 shows descriptive statistics for the total sample by subscriber cohort. Among the pre-pandemic subscribers, respondents were predominantly female (75.4%), White (80%), earned more than $100,000 (53.4%), and were employed (67.6%). There were few statistically significant differences in these observable characteristics between the three subscriber cohorts, namely, meditation cued by specific emotions, meditating in the morning, Native Hawaiian or Pacific Islander ethnicity, earning more than $100,000, and being unemployed. Additionally, a total of 704 (47.9%) participants abandoned the app within the pre-pandemic subscriber cohort, where the median time to app abandonment was 15.0 months (95% CI: 14.50, 15.49). Among the pandemic subscribers, there were 273 (53.2%) who abandoned and the median time to app abandonment was 10.2 months (95% CI: 9.48, 10.50). Finally, among the long-term subscriber cohort, 464 (35.8%) abandoned the app and the median time to abandonment was 41.0 months (95% CI: 39.21, 42.79).
Table 1.
Descriptive statistics for each subscription start date cohort
| Subscription start date: | p-value | ||||||
|---|---|---|---|---|---|---|---|
| < 4 months before COVID-19 | > 12 months before COVID-19 | During COVID-19 | |||||
| n | Percent | n | Percent | n | Percent | ||
| Meditation cue | |||||||
| Time of day | 301 | 20.5 | 98 | 19.1 | 246 | 19.0 | 0.995 |
| Alarm | 8 | 0.5 | 4 | 0.8 | 4 | 0.3 | 0.403 |
| Part of a daily routine | 557 | 37.9 | 183 | 35.7 | 537 | 41.5 | 0.004 |
| Specific emotions (e.g., boredom) | 197 | 13.4 | 94 | 18.3 | 197 | 14.6 | 0.005 |
| Specific physical sensations (e.g., fatigue) | 129 | 8.8 | 49 | 9.6 | 129 | 7.7 | 0.410 |
| Whenever I can | 204 | 13.9 | 63 | 12.3 | 204 | 12.8 | 0.931 |
| Reminder | 72 | 4.9 | 22 | 4.3 | 72 | 4.2 | 0.868 |
| Meditation time of day | |||||||
| Morning (wake–11 am) | 564 | 38.6 | 191 | 37.4 | 541 | 42.0 | 0.009 |
| Evening (4 pm–sleep) | 721 | 49.3 | 246 | 48.1 | 583 | 45.2 | 0.560 |
| Afternoon (11 am–4 pm) | 177 | 12.1 | 74 | 14.5 | 165 | 12.8 | 0.134 |
| Gender | |||||||
| Female | 1107 | 75.4 | 391 | 81.3 | 987 | 76.3 | 0.642 |
| Male | 264 | 18.0 | 87 | 18.0 | 234 | 0.9 | 0.543 |
| Other | 8 | 0.5 | 3 | 0.6 | 73 | 22.9 | 0.502 |
| Age | |||||||
| Under 25 | 47 | 3.6 | 14 | 3.1 | 27 | 2.3 | 0.589 |
| 25–34 | 233 | 17.7 | 73 | 16.0 | 229 | 19.3 | 0.118 |
| 35–44 | 294 | 22.3 | 118 | 25.8 | 269 | 22.6 | 0.289 |
| 45–54 | 334 | 25.4 | 104 | 22.8 | 257 | 21.6 | 0.389 |
| 55–64 | 225 | 17.1 | 95 | 20.8 | 228 | 19.2 | 0.325 |
| Over 65 | 183 | 13.9 | 53 | 11.6 | 179 | 15.1 | 0.110 |
| Race/ethnicity | |||||||
| Native American/Alaska Native | 6 | 0.4 | 2 | 0.4 | 4 | 0.3 | 0.564 |
| Asian | 36 | 2.5 | 16 | 3.1 | 22 | 1.7 | 0.175 |
| Black or African American | 36 | 2.5 | 12 | 2.3 | 33 | 2.6 | 0.674 |
| Native Hawaiian or Pacific Islander | 4 | 0.3 | none | none | 5 | 0.4 | 0.031 |
| White Non-Hispanic | 1175 | 80 | 408 | 79.5 | 1044 | 80.7 | 0.348 |
| Hispanic/Latinx | 88 | 6 | 23 | 4.5 | 82 | 6.3 | 0.683 |
| Other | 123 | 8.4 | 52 | 10.1 | 104 | 8.0 | 0.112 |
| Education | |||||||
| High school/GED or less | 23 | 1.7 | 9 | 1.9 | 32 | 2.6 | 0.063 |
| Some college | 147 | 10.7 | 53 | 11.0 | 114 | 9.2 | 0.143 |
| Two-year college | 514 | 37.3 | 160 | 33.2 | 478 | 38.8 | 0.056 |
| Bachelors degree | 69 | 5.0 | 32 | 6.6 | 60 | 4.9 | 0.269 |
| Graduate degree | 626 | 45.4 | 228 | 47.3 | 549 | 44.5 | 0.465 |
| Income | |||||||
| $20,000 or less | 43 | 3.3 | 13 | 2.8 | 37 | 2.9 | 0.919 |
| $21,000–$40,000 | 77 | 5.9 | 34 | 7.3 | 57 | 4.4 | 0.182 |
| $41,000–$60,000 | 122 | 9.4 | 56 | 12.1 | 106 | 8.2 | 0.082 |
| $61,000–$80,000 | 152 | 11.7 | 70 | 15.1 | 156 | 12.1 | 0.159 |
| $81,000–$100,000 | 214 | 16.4 | 83 | 17.9 | 176 | 13.6 | 0.422 |
| More than $100,000 | 696 | 53.4 | 207 | 44.7 | 636 | 49.1 | 0.002 |
| Employment status | |||||||
| Employed | 928 | 67.6 | 337 | 70.6 | 853 | 65.9 | 0.253 |
| Unemployed | 92 | 6.7 | 43 | 9.0 | 77 | 6.3 | 0.028 |
| Unable to work | 46 | 3.4 | 15 | 3.1 | 35 | 2.7 | 0.592 |
| Homemaker | 66 | 4.8 | 10 | 2.1 | 55 | 4.5 | 0.108 |
| Student | 36 | 2.6 | 15 | 3.1 | 26 | 2.1 | 0.545 |
| Retired | 204 | 14.9 | 57 | 11.9 | 182 | 14.8 | 0.264 |
| Observations | 1468 | 513 | 1294 | ||||
| Observed app abandonment | 704 | 47.9 | 273 | 53.2 | 464 | 35.8 | |
| Average time to abandonment | 15.51 months | 41.14 months | 10.24 months | ||||
Bold entries in this table are totals (n) for each group
Table 2 shows the results of Cox regression models estimated among the pre-pandemic subscribers who started < 4 months before the pandemic. The final model (Model 3) included all available behavioral, demographic, and socioeconomic predictors. From Model 3 in Table 2, we can see that using an existing routine to cue daily meditation was associated with a lower risk of app abandonment compared to those who used a different cue (hazard ratio = 0.607, 95% CI: 0.422, 0.874; p = 0.007). Using an existing routine to cue meditation, for example, could be using toothbrushing in the morning or washing dishes after dinner as a cue that it is time to meditate. To illustrate the difference in app abandonment between those who did and did not use an existing daily routine to cue meditation, Fig. 1a shows the probability of survival (i.e., not abandoning the app) within these groups over time. This figure shows that those who used an existing daily routine to cue their meditation showed delayed time until app abandonment compared to those who used another type of meditation cue over all months following their first subscription date. For example, subscribers who meditated after a daily routine were roughly 61% likely to continue using the meditation app 10 months after their subscription app compared to only a 39% chance of maintaining meditation app use among those not using an existing routine as a meditation cue.
Table 2.
Determinants of app abandonment among subscribers who started just before (< 4 months) the pandemic
| Model 1 | Model 2 | Model 3 | |||||
|---|---|---|---|---|---|---|---|
| Hazard ratio [95% CI] |
p-value | Hazard ratio [95% CI] |
p-value | Hazard ratio [95% CI] |
p-value | ||
| Meditation cue | |||||||
| Time of day |
0.705 [0.503, 0.989] |
0.043 |
0.684 [0.487, 0.96] |
0.028 |
0.797 [0.546, 1.162] |
0.238 | |
| Alarm |
1.236 [0.527, 2.901] |
0.626 |
1.407 [0.598, 3.311] |
0.434 |
1.659 [0.566, 4.862] |
0.356 | |
| After a daily routine |
0.538 [0.389, 0.745] |
< 0.001 |
0.535 [0.386, 0.742] |
< 0.001 |
0.607 [0.422, 0.874] |
0.007 | |
| Specific emotions |
0.976 [0.689, 1.381] |
0.890 |
0.88 [0.62, 1.249] |
0.475 |
0.919 [0.619, 1.363] |
0.674 | |
| Specific physical sensations |
0.862 [0.592, 1.254] |
0.437 |
0.751 [0.514, 1.099] |
0.140 |
0.938 [0.614, 1.433] |
0.767 | |
| Whenever I can |
0.819 [0.577, 1.163] |
0.265 |
0.777 [0.547, 1.104] |
0.159 |
0.965 [0.652, 1.428] |
0.858 | |
| Reminder | Reference | Reference | Reference | ||||
| Meditation time of day | |||||||
| Morning (wake–11 am) |
0.845 [0.652, 1.095] |
0.202 |
0.854 [0.642, 1.138] |
0.282 | |||
| Evening (4 pm–sleep) |
1.254 [0.99, 1.59] |
0.061 |
1.162 [0.896, 1.507] |
0.259 | |||
| Afternoon (11 am–4 pm) | Reference | Reference | |||||
| Gender | |||||||
| Woman |
1.196 [0.959, 1.49] |
0.112 | |||||
| Other |
1.986 [0.767, 5.147] |
0.158 | |||||
| Man | Reference | ||||||
| Age | |||||||
| Under 25 |
2.472 [1.419, 4.304] |
0.001 | |||||
| 25–34 |
1.856 [1.222, 2.821] |
0.004 | |||||
| 35–44 |
1.723 [1.15, 2.581] |
0.008 | |||||
| 45–54 |
1.438 [0.964, 2.145] |
0.075 | |||||
| 55–64 |
1.314 [0.895, 1.929] |
0.163 | |||||
| Over 65 | Reference | ||||||
| Race/ethnicity | |||||||
| Native American/Alaska Native |
0.768 [0.216, 2.724] |
0.682 | |||||
| Asian |
0.708 [0.349, 1.434] |
0.337 | |||||
| Black or African American |
0.997 [0.501, 1.982] |
0.993 | |||||
| Native Hawaiian or Pacific Islander |
0.696 [0.158, 3.063] |
0.631 | |||||
| White |
0.752 [0.451, 1.252] |
0.273 | |||||
| Hispanic/Latinx |
0.746 [0.413, 1.346] |
0.330 | |||||
| Other | Reference | ||||||
| Education | |||||||
| High school/GED or less |
1.325 [0.706, 2.486] |
0.381 | |||||
| Some college |
1.617 [1.233, 2.12] |
0.001 | |||||
| Bachelors degree |
1.118 [0.928, 1.347] |
0.240 | |||||
| Two-year degree |
1.319 [0.909, 1.914] |
0.146 | |||||
| Graduate degree | Reference | ||||||
| Income | |||||||
| $20,000 or less |
1.149 [0.716, 1.846] |
0.565 | |||||
| $21,000–$40,000 |
1.211 [0.844, 1.737] |
0.298 | |||||
| $41,000–$60,000 |
0.778 [0.57, 1.061] |
0.113 | |||||
| $61,000–$80,000 |
1.132 [0.873, 1.468] |
0.350 | |||||
| $81,000–$100,000 |
1.045 [0.827, 1.319] |
0.713 | |||||
| More than $100,00 | Reference | ||||||
| Employment status | |||||||
| Employed |
0.905 [0.63, 1.3] |
0.590 | |||||
| Unemployed |
0.966 [0.611, 1.527] |
0.882 | |||||
| Unable to work |
0.735 [0.422, 1.28] |
0.277 | |||||
| Homemaker |
0.607 [0.353, 1.043] |
0.070 | |||||
| Student |
0.648 [0.336, 1.248] |
0.194 | |||||
| Retired | Reference | ||||||
| Observations | |||||||
| Total | 1468 | 1462 | 1244 | ||||
| Percent who abandoned | 48.0% | 47.8% | 47.7% | ||||
Fig. 1.
Probability of survival by daily routine as meditation cue. b Probability of survival by age
Additional predictors of an increased risk of app abandonment included being younger in age and having completed less education. Those under 25 (hazard ratio = 2.472, 95% CI: 1.419, 4.304; p = 0.001), between 25 and 34 (hazard ratio = 1.856, 95% CI: 1.222, 2.821; p = 0.004), and between 34 and 44 (hazard ratio = 1.723, 95% CI: 1.15, 2.581; p = 0.008) had a significantly higher risk of abandonment compared to those over 65. Those who completed some college (hazard ratio = 1.617, 95% CI: 1.233, 2.120; p = 0.001) had a higher risk of abandonment compared to those with a graduate degree. To illustrate the difference in app abandonment between age groups, Fig. 1b shows the probability of survival (i.e., not abandoning the app) among the under 25 and over 65 age groups. For example, subscribers who were over 65 had a roughly 57% chance of maintaining their app use 15 months after first subscribing, while subscribers under 25 had only a 30% chance of maintaining their meditation app use after 15 months.
To assess the robustness of the findings in Model 3, Models 1 and 2 describe the association between app abandonment and meditation cues or meditation time of day without controlling for the demographic or socioeconomic variables. The results in Table 2 show that meditating after a daily routine remained significant in Model 1 (hazard ratio = 0.538, 95% CI: 0.389, 0.745; p < 0.001) and Model 2 (hazard ratio = 0.535, 95% CI: 0.386, 0.742; p < 0.001). Those who meditated using an existing routine as a cue had a lower risk of abandoning the app in Models 1 and 2 compared to those who used a different meditation cue. Using time of day as a meditation cue was significantly associated with a lower risk of app abandonment in Model 1 (hazard ratio = 0.705, 95% CI: 0.503, 0.989; p = 0.043) and a lower risk in Model 2 (hazard ratio = 0.684, 95% CI: 0.487, 0.96; p = 0.028). However, this variable did not remain significant in Model 3.
Table 3 displays the results of Cox regression models estimated among study participants who subscribed at least 1 year before the pandemic. The only significant determinant of app abandonment was earning between $61,000 and $80,000 per year (hazard ratio = 0.624, 95% CI: 0.393, 0.988; p = 0.044) compared to those who earned more than $100,000. There were no significant behavioral predictors in any of the models (Models 1–3).
Table 3.
Determinants of app abandonment among subscribers who started at least 1 year before the pandemic
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Hazard ratio [95% CI] |
p-value | Hazard ratio [95% CI] |
p-value | Hazard ratio [95% CI] |
p-value | |
| Meditation cue | ||||||
| Time of day |
1.052 [0.573, 1.93] |
0.871 |
1.012 [0.545, 1.88] |
0.969 |
1.203 [0.557, 2.597] |
0.638 |
| Alarm |
0.722 [0.163, 3.202] |
0.668 |
0.683 [0.153, 3.049] |
0.617 |
0.747 [0.152, 3.675] |
0.719 |
| After a daily routine |
1.007 [0.562, 1.806] |
0.981 |
0.996 [0.551, 1.802] |
0.991 |
1.415 [0.677, 2.959] |
0.356 |
| Specific emotions |
1.184 [0.643, 2.183] |
0.587 |
1.072 [0.579, 1.984] |
0.825 |
1.543 [0.718, 3.317] |
0.267 |
| Specific physical sensations |
1.385 [0.728, 2.634] |
0.321 |
1.171 [0.608, 2.255] |
0.637 |
1.384 [0.622, 3.079] |
0.425 |
| Whenever I can |
0.976 [0.512, 1.861] |
0.940 |
0.907 [0.475, 1.734] |
0.768 |
1.293 [0.578, 2.893] |
0.531 |
| Reminder | Reference | Reference | Reference | |||
| Meditation time of day | ||||||
| Morning (wake–11 am) |
0.833 [0.558, 1.245] |
0.373 |
0.747 [0.466, 1.198] |
0.226 | ||
| Evening (4 pm–sleep) |
1.327 [0.92, 1.914] |
0.131 |
1.411 [0.909, 2.19] |
0.125 | ||
| Afternoon (11 am–4 pm) | Reference | Reference | ||||
| Gender | ||||||
| Woman |
1.302 [0.872, 1.944] |
0.197 | ||||
| Other |
2.045 [0.4, 10.447] |
0.390 | ||||
| Man | Reference | |||||
| Age | ||||||
| Under 25 |
1.212 [0.434, 3.379] |
0.714 | ||||
| 25–34 |
1.058 [0.541, 2.07] |
0.869 | ||||
| 35–44 |
0.816 [0.431, 1.544] |
0.531 | ||||
| 45–54 |
0.85 [0.442, 1.636] |
0.627 | ||||
| 55–64 |
1.289 [0.709, 2.347] |
0.405 | ||||
| Over 65 | Reference | |||||
| Race/ethnicity | ||||||
| Native American/Alaska Native |
0 [0, 2.50839] |
0.947 | ||||
| Asian |
1.003 [0.349, 2.885] |
0.996 | ||||
| Black or African American |
2.339 [0.797, 6.863] |
0.122 | ||||
| Native Hawaiian or Pacific Islander | None | |||||
| White |
0.921 [0.426, 1.99] |
0.834 | ||||
| Hispanic/Latinx |
0.658 [0.243, 1.78] |
0.410 | ||||
| Other | Reference | |||||
| Education | ||||||
| High school/GED or less |
1.168 [0.399, 3.415] |
0.777 | ||||
| Some college |
0.677 [0.395, 1.161] |
0.156 | ||||
| Bachelors degree |
0.909 [0.654, 1.263] |
0.570 | ||||
| Two-year degree |
1.383 [0.77, 2.484] |
0.278 | ||||
| Graduate degree | Reference | |||||
| Income | ||||||
| $20,000 or less |
1.745 [0.803, 3.79] |
0.159 | ||||
| $21,000–$40,000 |
0.724 [0.405, 1.295] |
0.276 | ||||
| $41,000–$60,000 |
0.918 [0.573, 1.47] |
0.721 | ||||
| $61,000–$80,000 |
0.624 [0.393, 0.988] |
0.044 | ||||
| $81,000–$100,000 |
0.844 [0.564, 1.262] |
0.409 | ||||
| More than $100,000 | Reference | |||||
| Employment status | ||||||
| Employed |
0.817 [0.449, 1.486] |
0.508 | ||||
| Unemployed |
0.654 [0.315, 1.36] |
0.256 | ||||
| Unable to work |
0.746 [0.285, 1.953] |
0.550 | ||||
| Homemaker |
0.903 [0.275, 2.967] |
0.867 | ||||
| Student |
0.859 [0.347, 2.129] |
0.743 | ||||
| Retired | Reference | |||||
| Observations | ||||||
| Total | 513 | 511 | 441 | |||
| Percent who abandoned | 53.2% | 53.2% | 50.8% | |||
Finally, Table 4 shows the results of Cox regression models estimated among the subscribers who started during the pandemic. Two types of meditation cues were significantly associated with a lower risk of app abandonment: meditating after an existing routine (hazard ratio = 0.434, 95% CI: 0.285, 0.66; p < 0.001) and meditating “whenever I can” (hazard ratio = 0.437, 95% CI: 0.271, 0.706; p < 0.001) both had a lower risk of app abandonment compared to meditating using reminders. Being unemployed was associated with a higher risk of app abandonment (hazard ratio = 1.733, 95% confidence interval: 1.839, 3.111; p = 0.023) compared to those who were retired.
Table 4.
Determinants of app abandonment among subscribers who started during the pandemic
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Hazard ratio[95% CI] | p-value | Hazard ratio[95% CI] | p-value | Hazard ratio[95% CI] | p-value | |
| Meditation cue | ||||||
| Time of day |
0.570 [0.383, 0.85] |
0.006 |
0.573 [0.380, 0.863] |
0.008 |
0.697 [0.445, 1.091] |
0.114 |
| Alarm |
0.143 [0.019, 1.046] |
0.055 |
0.16 [0.38, 0.863] |
0.072 |
0.453 [0.06, 3.405] |
0.442 |
| After a daily routine |
0.382 [0.262, 0.555] |
< 0.001 |
0.393 [0.267, 0.579] |
< 0.001 |
0.434 [0.285, 0.66] |
< 0.001 |
| Specific emotions |
0.642 [0.429, 0.961] |
0.031 |
0.647 [0.431, 0.971] |
0.036 |
0.732 [0.468, 1.145] |
0.172 |
| Specific physical sensations |
0.668 [0.426, 1.048] |
0.079 |
0.646 [0.409, 1.02] |
0.061 |
0.743 [0.455, 1.214] |
0.236 |
| Whenever I can |
0.380 [0.244, 0.592] |
< 0.001 |
0.373 [0.238, 0.585] |
< 0.001 |
0.437 [0.271, 0.706] |
< 0.00 |
| Reminder | Reference | Reference | Reference | |||
| Meditation time of day | ||||||
| Morning (wake–11 am) |
0.892 [0.661, 1.205] |
0.457 |
0.890 [0.644, 1.23] |
0.482 | ||
| Evening (4 pm–sleep) |
1.086 [0.818, 1.443] |
0.568 |
1.093 [0.804, 1.485] |
0.569 | ||
| Afternoon (11 am–4 pm) | Reference | Reference | ||||
| Gender | ||||||
| Woman |
1.109 [0.851, 1.446] |
0.444 | ||||
| Other |
0.969 [0.296, 3.174] |
0.959 | ||||
| Man | Reference | |||||
| Age | ||||||
| Under 25 |
1.149 [0.522, 2.529] |
0.73 | ||||
| 25–34 |
0.676 [0.434, 1.051] |
0.082 | ||||
| 35–44 |
0.896 [0.59, 1.362] |
0.609 | ||||
| 45–54 |
0.731 [0.483, 1.106] |
0.138 | ||||
| 55–64 |
0.799 [0.532, 1.198] |
0.277 | ||||
| Over 65 | Reference | |||||
| Race/ethnicity | ||||||
| Native American/Alaska Native |
0.72 [0.089, 5.835] |
0.758 | ||||
| Asian |
1.154 [0.467, 2.851] |
0.756 | ||||
| Black or African American |
1.835 [0.833, 4.039] |
0.132 | ||||
| Native Hawaiian or Pacific Islander |
0 [0, 9.62] |
0.937 | ||||
| White |
1.092 [0.584, 2.043] |
0.783 | ||||
| Hispanic/Latinx |
1.384 [0.685, 2.798] |
0.365 | ||||
| Other | Reference | |||||
| Education | ||||||
| High school/GED or less |
0.854 [0.405, 1.8] |
0.677 | ||||
| Some college |
1.261 [0.869, 1.83] |
0.222 | ||||
| Bachelors degree |
1.124 [0.9, 1.404] |
0.303 | ||||
| Two-year degree |
1.219 [0.785, 1.894] |
0.378 | ||||
| Graduate degree | Reference | |||||
| Income | ||||||
| $20,000 or less |
1.356 [0.775, 2.375] |
0.286 | ||||
| $21,000–$40,000 |
1.251 [0.751, 2.083] |
0.389 | ||||
| $41,000–$60,000 |
0.764 [0.52, 1.124] |
0.172 | ||||
| $61,000–$80,000 |
1.305 [0.969, 1.759] |
0.080 | ||||
| $81,000–$100,000 |
1.09 [0.824, 1.443] |
0.545 | ||||
| More than $100,000 | Reference | |||||
| Employment status | ||||||
| Employed |
1.366 [0.917, 2.035] |
0.125 | ||||
| Unemployed |
1.839 [1.087, 3.111] |
0.023 | ||||
| Unable to work |
1.845 [0.968, 3.518] |
0.063 | ||||
| Homemaker |
1.382 [0.766, 2.494] |
0.283 | ||||
| Student |
0.801 [0.328, 1.954] |
0.625 | ||||
| Retired | Reference | |||||
| Observations | ||||||
| Total | 1294 | 1289 | 1120 | |||
| Percent who abandoned | 35.9% | 49.9% | 36.8% | |||
To assess the robustness of the findings in Model 3 of Table 4, we found that meditating after an existing routine was also significant in Model 1 (hazard ratio = 0.382 95% CI: 0.262, 0.555; p < 0.001) and Model 2 (hazard ratio = 0.393 95% CI: 0.267, 0.579; p < 0.001) compared to using meditation reminders. Subscribers who meditated after an existing routine were less likely to abandon the app in Model 1 and less likely to abandon the app in Model 2. Meditating “whenever I can” was also significantly associated with a lower risk of app abandonment in Model 1 (hazard ratio = 0.380, 95% CI: 0.244, 0.592; p < 0.001) and a lower risk in Model 2 (hazard ratio = 0.373, 95% CI: 0.238, 0.585; p < 0.001). Cueing meditation based on the time of day was also associated with a lower risk of meditation app abandonment in Model 1 (hazard ratio = 0.570, 95% CI: 0.383, 0.850; p = 0.006) and a lower risk in Model 2 (hazard ratio = 0.573, 95% CI: 0.380, 0.863; p = 0.008) compared to using reminders; however, this variable did not remain significant in Model 3.
To directly compare these statistical relationships between cohorts, Cox proportional hazards regression models were estimated including all cohorts and interactions between each cohort and (1) meditation cue type and (2) meditation time of day (see Appendix A). Consistent with the findings presented in Table 4, pandemic subscribers (hazard ratio = 0.288, 95% CI: 0.127, 0.654; p = 0.003) and pre-pandemic subscribers (hazard ratio = 0.446, 95% CI: 0.202, 0.983; p = 0.045) who meditated as part of a daily routine were less likely to abandon the app than the long-term subscribers. Additionally, pandemic subscribers who meditated “whenever I can” were less likely than the other cohorts to abandon the app (hazard ratio = 0.310, 95% CI: 0.126;0.762; p = 0.011).
Discussion
The purpose of this study was to determine the behavioral, demographic, and socioeconomic factors that were associated with meditation app abandonment during the COVID-19 pandemic. Approximately 48%, 53%, and 36% of respondents abandoned the app in the pre-pandemic subscriber, long-term subscriber, and pandemic subscriber cohorts, respectively. The most robust predictor of persistent meditation app use, and thus lower app abandonment, was meditating after an existing daily routine. Specifically, anchoring meditation onto an existing routine to cue meditation was significantly associated with a lower risk of abandonment among the pre-pandemic subscribers and pandemic subscribers, which suggests that anchoring meditation to an existing routine was a beneficial strategy for maintaining a meditation practice for many subscribers during the COVID-19 pandemic. This behavioral approach for establishing a persistent meditation practice was significantly more effective than using reminders or alarms, which helps to guide the design of future interventions that aim to maintain app engagement and prevent abandonment.
Additionally, meditating according to a specific time of day was weakly associated with a lower risk of meditation app abandonment for those in the pre-pandemic subscriber and pandemic subscriber cohorts. However, there were no significant differences in app abandonment between those who chose to meditate in the mornings versus the afternoons or evenings. This suggests that consistency in the time of day of meditation is an important determinant of persistent meditation app use, but that the optimal time of day for meditation varies between individuals. Future meditation app interventions should ask participants to identify a time of day for meditation that would be most appropriate for their schedules, and then encourage them to consistently meditate at their chosen time of day. Our findings suggest that this strategy may be more effective for preventing app abandonment than requiring all participants to meditate at the same time of day.
A few demographic and socioeconomic variables were also significantly associated with meditation app abandonment, which help us to identify the app users most in need of additional meditation supports. In the pre-pandemic subscriber cohort, those who were younger and had completed less formal education were associated with a higher risk of app abandonment. In the long-term subscriber and pandemic subscriber cohorts, having a lower income was associated with a higher risk of app abandonment, and in the pandemic subscriber cohort, being unemployed was associated with a higher risk of app abandonment. Taken together, these findings suggest that a lower socioeconomic status was associated with faster app abandonment during the COVID-19 pandemic. Previous research has shown that lower socioeconomic status is typically underrepresented in meditation app studies and among meditation app subscribers, which highlights the need to investigate reasons for app abandonment, and low app use in general, among this population. Also, future meditation interventions should consider targeting these groups for additional behavioral supports to help establish more persistent meditation habits that can help mitigate the burden of the COVID-19 pandemic or other potential adverse events.
There were several important differences in our findings between subscriber cohorts that provide a deeper understanding of the value of meditation routines and using environmental meditation cues. For the pre-pandemic subscribers, using an existing daily routine to cue meditation was associated with a lower risk of app abandonment. Since this cohort began using the app a few months before the pandemic, our findings suggest that using an existing routine to trigger meditation may have enabled many of these subscribers to establish a meditation habit that was more resilient to lifestyle or personal impacts of the pandemic. For those who subscribed to the app during the pandemic, the rate of app abandonment was the smallest among the three subscriber cohorts and we found that both meditating “whenever I can” and meditating with an existing routine were associated with a lower risk of app abandonment. Thus, the pandemic subscribers likely had a higher level of initial motivation for using the meditation app, and both using a routine and allowing for some flexibility in their meditation practice appear to have supported persistent meditation among these more motivated users. Finally, we found that none of the behavioral factors was significantly associated with app abandonment among the long-term app subscribers. These long-term subscribers were also the most likely to abandon the app during the pandemic, which suggests that the pandemic disrupted many long-standing meditation habits more than it affected habits that were newly created among the pre-pandemic and pandemic subscriber cohorts. Thus, future research is needed to better understand the strategies used by the individuals who maintained their engagement with the app over many years, which appear to be different and less flexible strategies than those used to establish meditation habits during the pandemic.
The results from this study build on several findings from the existing literature such as the importance of a daily routine for establishing persistent health habits. Specifically, daily routines have been identified as a strong determinant of persistent health behaviors such as taking medication and daily exercise (e.g., Argent, 2018; Phillips et al., 2016; Thorneloe et al., 2018). Additionally, anchoring (or pairing) a new habit to an existing routine has been shown to be an effective intervention for promoting physical activity (Prestwich et al., 2003), dieting (Achtziger et al., 2008), and smoking cessation (Armitage & Arden, 2008a, 2008b). These findings support the conclusion that incorporating meditation into one’s existing routine can be a successful way to build a persistent meditation practice and prevent abandonment. Future research is still needed to investigate the types of routines that are most supportive of a new meditation app habit.
Meditating at a consistent time of day was also associated with a lower risk of abandoning the app, which is supported by literature on the temporal consistency of healthy habits. For example, two recent studies found that temporal consistency was a strong predictor of persistent physical activity (Kaushal & Rhodes, 2015; Kaushal et al., 2017), and prior research has hypothesized that temporal consistency can help to create a protected time in the day for the targeted behavior (Rhodes & De Bruijn, 2010). Another recent study also found that meditating at roughly the same time each day was associated with greater meditation app persistence than those who meditated at different times in the day (Stecher, Berardi, et al., 2021). Future research should investigate the mechanisms that underlie the role of temporal consistency on habit formation, and additional research is needed to identify the causal effect of temporal consistency on behavioral persistence.
Limitations and Future Research
Our study used a large sample of meditation app users to investigate the determinants of app abandonment during the COVID-19 pandemic; however, these findings should be considered in light of the following limitations. First, we used app usage data collected during the COVID-19 pandemic, so our results have unknown generalizability to app abandonment patterns either pre- or post-pandemic. The generalizability of our findings may also be limited by the relatively homogenous sample demographics and our focus on a single meditation app, Calm. Specifically, participants of this study were mostly White, educated women who earned over $100,000 per year. While this is generally representative of current meditation app users (e.g., Goldberg et al., 2020; Stecher, Berardi, et al., 2021), our results may not apply to the app abandonment of other health apps or to the meditation app abandonment among specific clinical study populations, e.g., meditation app use among cancer patients. However, our findings are still able to demonstrate that those with lower socioeconomic status are more likely to abandon a meditation app and therefore may need additional support for maintaining their meditation app use. Additionally, the Calm app was one of the top two most popular meditation apps at the time of this study, so our study sample is drawn from a large population of meditation app users and our findings are likely to generalize to other meditation apps that have well-developed user experiences (e.g., Goldberg et al., 2020; Huberty et al., 2019; Stecher, Berardi, et al., 2021).
This study examined the behavioral, demographic, and socioeconomic variables that were associated with meditation app abandonment among a real-world sample of meditation app users. We found that meditating after an existing daily routine was associated with a lower risk of app abandonment for many meditation app subscribers. This suggests that combining meditation with an existing daily routine is a promising strategy for maintaining meditation app habits. Given the significant mental health benefits that are associated with persistent long-term meditation, these findings demonstrate how combining meditation with an existing routine can be used to improve mental health outcomes.
Appendix
Table 5
Table 5.
Determinants of app abandonment among all subscribers; interactions by cohort
| Model 1 | Model 2 | Model 3 | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Hazard ratio[95% CI] | Cohen’s d | p-value | Hazard ratio[95% CI] | Cohen’s d | p-value | Hazard ratio[95% CI] | Cohen’s d | p-value | |
| Meditation cue | 0.722 | 0.958 | 0.969 | ||||||
| Time of day |
1.052 [0.573, 1.931] |
3.394 | 0.87 |
1.013 [0.546, 1.881] |
3.206 | 0.967 |
1.351 [0.647, 2.822] |
3.593 | 0.423 |
| Alarm |
0.722 [0.163, 3.204] |
0.950 | 0.668 |
0.683 [0.153, 3.051] |
0.895 | 0.618 |
0.843 [0.18, 3.952] |
1.068 | 0.828 |
| Part of a daily routine |
1.008 [0.562, 1.806] |
3.383 | 0.98 |
0.997 [0.551, 1.804] |
3.301 | 0.993 |
1.324 [0.653, 2.687] |
3.668 | 0.437 |
| Specific emotions |
1.183 [0.642, 2.181] |
3.792 | 0.59 |
1.071 [0.579, 1.982] |
3.411 | 0.827 |
1.42 [0.683, 2.954] |
3.797 | 0.348 |
| Physical sensations |
1.386 [0.729, 2.636] |
4.226 | 0.32 |
1.172 [0.609, 2.257] |
3.509 | 0.634 |
1.343 [0.619, 2.917] |
3.391 | 0.456 |
| Whenever I can |
0.976 [0.512, 1.861] |
2.967 | 0.941 |
0.908 [0.475, 1.735] |
2.752 | 0.77 |
1.271 [0.59, 2.738] |
3.251 | 0.54 |
| Reminder (reference) | |||||||||
| Cohort | < 0.001 | < 0.001 | < 0.001 | ||||||
| Pandemic subscribers |
184.27 [81.199, 418.175] |
440.837 | < 0.001 |
192.641 [80.926, 458.571] |
434.856 | < 0.001 |
233.342 [86.383, 630.316] |
460.241 | < 0.001 |
| Pre-pandemic subscribers |
40.586 [18.315, 89.94] |
99.966 | < 0.001 |
41.365 [17.685, 96.75] |
95.311 | < 0.001 |
48.638 [18.266, 129.512] |
97.276 | < 0.001 |
| Long-term subscribers (reference) | |||||||||
| Cohort*Meditation cue | 0.02 | 0.033 | |||||||
| Pandemic subscribers*Time of day |
0.505 [0.244, 1.044] |
1.361 | 0.065 |
0.527 [0.251, 1.108] |
1.391 | 0.091 |
0.46 [0.195, 1.082] |
1.053 | 0.075 |
| Pre-pandemic subscribers*Time of day |
0.665 [0.332, 1.332] |
1.873 | 0.249 |
0.67 [0.331, 1.357] |
1.861 | 0.266 |
0.555 [0.244, 1.266] |
1.321 | 0.162 |
| Pandemic subscribers*Alarm |
0.163 [0.014, 1.965] |
0.128 | 0.153 |
0.194 [0.016, 2.351] |
0.152 | 0.198 |
0.422 [0.034, 5.309] |
0.327 | 0.505 |
| Pre-pandemic subscribers*Alarm |
1.721 [0.309, 9.583] |
1.965 | 0.535 |
2.063 [0.368, 11.561] |
2.347 | 0.41 |
1.85 [0.284, 12.05] |
1.935 | 0.52 |
| Pandemic subscribers*Part of a daily routine |
0.347 [0.174, 0.695] |
0.980 | 0.003 |
0.362 [0.178, 0.735] |
1.003 | 0.005 |
0.288 [0.127, 0.654] |
0.689 | 0.003 |
| Pre-pandemic subscribers*Part of a daily routine |
0.531 [0.272, 1.036] |
1.557 | 0.063 |
0.534 [0.271, 1.05] |
1.548 | 0.069 |
0.446 [0.202, 0.983] |
1.107 | 0.045 |
| Pandemic subscribers*Specific emotions |
0.499 [0.24, 1.039] |
1.334 | 0.063 |
0.559 [0.268, 1.169] |
1.487 | 0.122 |
0.426 [0.182, 0.999] |
0.979 | 0.05 |
| Pre-pandemic subscribers*Specific emotions |
0.818 [0.405, 1.653] |
2.279 | 0.575 |
0.815 [0.402, 1.655] |
2.258 | 0.572 |
0.668 [0.292, 1.526] |
1.583 | 0.339 |
| Pandemic subscribers*Physical sensations |
0.447 [0.204, 0.98] |
1.118 | 0.044 |
0.514 [0.231, 1.142] |
1.260 | 0.102 |
0.491 [0.197, 1.224] |
1.054 | 0.127 |
| Pre-pandemic subscribers*Physical sensations |
0.618 [0.294, 1.302] |
1.626 | 0.206 |
0.637 [0.299, 1.359] |
1.650 | 0.244 |
0.68 [0.283, 1.635] |
1.518 | 0.389 |
| Pandemic subscribers*Whenever I can |
0.358 [0.163, 0.783] |
0.895 | 0.01 |
0.378 [0.172, 0.832] |
0.940 | 0.016 |
0.31 [0.126, 0.762] |
0.674 | 0.011 |
| Pre-pandemic subscribers*Whenever I can |
0.836 [0.401, 1.742] |
2.229 | 0.632 |
0.851 [0.407, 1.779] |
2.263 | 0.669 |
0.726 [0.308, 1.713] |
1.658 | 0.465 |
| Meditation time of day | 0.004 | 0.011 | |||||||
| Morning |
0.833 [0.558, 1.245] |
4.063 | 0.372 |
0.782 [0.498, 1.227] |
3.400 | 0.285 | |||
| Evening |
1.326 [0.919, 1.912] |
7.091 | 0.131 |
1.255 [0.828, 1.902] |
5.920 | 0.284 | |||
| Afternoon (reference) | |||||||||
| Cohort*Meditation time of day | 0.489 | 0.542 | |||||||
| Pandemic subscribers*Morning |
1.061 [0.643, 1.752] |
4.145 | 0.816 |
1.208 [0.694, 2.102] |
4.269 | 0.504 | |||
| Pre-pandemic subscribers*Morning |
1.013 [0.628, 1.634] |
4.152 | 0.957 |
1.106 [0.65, 1.882] |
4.081 | 0.709 | |||
| Pandemic subscribers*Evening |
0.81 [0.509, 1.287] |
3.418 | 0.372 |
0.873 [0.522, 1.459] |
3.332 | 0.604 | |||
| Pre-pandemic subscribers*Evening |
0.945 [0.611, 1.461] |
4.238 | 0.798 |
0.955 [0.587, 1.555] |
3.851 | 0.854 | |||
| Gender | 0.03 | ||||||||
| Female |
1.191 [1.022, 1.388] |
15.269 | 0.025 | ||||||
| Other |
1.733 [0.974, 3.081] |
5.895 | 0.061 | ||||||
| Male (reference) | |||||||||
| Age | 0.134 | ||||||||
| Under 25 |
1.559 [1.049, 2.319] |
7.718 | 0.028 | ||||||
| 25–34 |
1.162 [0.885, 1.525] |
8.360 | 0.28 | ||||||
| 35–44 |
1.148 [0.884, 1.491] |
8.567 | 0.301 | ||||||
| 45–54 |
0.998 [0.771, 1.293] |
7.561 | 0.988 | ||||||
| 55–64 |
1.056 [0.823, 1.354] |
8.315 | 0.67 | ||||||
| Over 65 (reference) | |||||||||
| Race/ethnicity | 0.153 | ||||||||
| Native American/Alaska Native |
0.47 [0.164, 1.345] |
0.875 | 0.159 | ||||||
| Asian |
0.903 [0.56, 1.459] |
3.701 | 0.678 | ||||||
| Black or African American |
1.379 [0.874, 2.174] |
5.944 | 0.167 | ||||||
| Native Hawaiian or Pacific Islander |
0.525 [0.126, 2.198] |
0.719 | 0.378 | ||||||
| White |
0.918 [0.65, 1.296] |
5.216 | 0.627 | ||||||
| Hispanic/Latinx |
0.933 [0.624, 1.393] |
4.551 | 0.733 | ||||||
| Other (reference) | |||||||||
| Education | 0.04 | ||||||||
| High school/GED or less |
1.207 [0.784, 1.857] |
5.486 | 0.393 | ||||||
| Some college |
1.284 [1.053, 1.567] |
12.713 | 0.013 | ||||||
| Bachelors degree |
1.077 [0.947, 1.226] |
16.318 | 0.257 | ||||||
| Two-year degree |
1.357 [1.059, 1.74] |
10.685 | 0.016 | ||||||
| Graduate degree (reference) | |||||||||
| Income | 0.124 | ||||||||
| $20,000 or less |
1.281 [0.931, 1.76] |
7.907 | 0.128 | ||||||
| $21,000–$40,000 |
1.074 [0.829, 1.391] |
8.136 | 0.591 | ||||||
| $41,000–$60,000 |
0.812 [0.657, 1.003] |
7.519 | 0.053 | ||||||
| $61,000–$80,000 |
1.077 [0.902, 1.286] |
11.835 | 0.414 | ||||||
| $81,000–$100,000 |
1.031 [0.877, 1.211] |
12.573 | 0.714 | ||||||
| Over $100,000 (reference) | |||||||||
| Employment status | |||||||||
| Employed |
1.081 [0.85, 1.375] |
8.789 | 0.526 | ||||||
| Unemployed |
1.182 [0.868, 1.608] |
7.529 | 0.288 | ||||||
| Unable to work |
1.011 [0.692, 1.477] |
5.238 | 0.955 | ||||||
| Homemaker |
0.86 [0.595, 1.244] |
4.574 | 0.424 | ||||||
| Student |
0.86 [0.555, 1.331] |
3.857 | 0.498 | ||||||
| Retired (reference) | |||||||||
| Observations | |||||||||
| Total | 3275 | 3262 | 2805 | ||||||
| Percent who abandoned | 32.6% | 32.50% | 1229 | ||||||
Author Contribution
All authors read and approved the final manuscript. Study conception and design were performed by Sullivan, Stecher, and Huberty. Interpretation of the data was performed by Sullivan, Stecher, and Chung. The first draft was written by Sullivan and all authors commented on previous versions of the manuscript.
Data Availability
A de-identified version of the data is publicly available at https://doi.org/10.17605/OSF.IO/YMJQH.
Code Availability
The authors have full control of all primary data and allow the journal to review the data upon request.
Declarations
Ethics Approval
The Institutional Review Boast at Arizona State University (STUDY00011867) approved this study. The study was performed in accordance with the principles of the 1964 Declaration of Helsinki.
Informed Consent
All respondents in this study provided informed consent via an electronic survey.
Conflict of Interest
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- Achtziger A, Gollwitzer PM, Sheeran P. Implementation intentions and shielding goal striving from unwanted thoughts and feelings. Personality and Social Psychology Bulletin. 2008;34(3):381–393. doi: 10.1177/0146167207311201. [DOI] [PubMed] [Google Scholar]
- Argent R. Patient involvement with home-based exercise programs: Can connected health interventions influence adherence? JMIR mHealth and uHealth. 2018;6(3):e47. doi: 10.2196/mhealth.8518. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Armitage CJ, Arden MA. How useful are the stages of change for targeting interventions? Randomized test of a brief intervention to reduce smoking. Health Psychology. 2008;27(6):789–798. doi: 10.1037/0278-6133.27.6.789. [DOI] [PubMed] [Google Scholar]
- Bostock S, Crosswell AD, Prather AA, Steptoe A. Mindfulness on-the-go: Effects of a mindfulness meditation app on work stress and well-being. Journal of Occupational Health Psychology. 2019;24(1):127–138. doi: 10.1037/ocp0000118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Burner ER, Menchine MD, Kubicek K, Robles M, Arora S. Perceptions of successful cues to action and opportunities to augment behavioral triggers in diabetes self-management: qualitative analysis of a mobile intervention for low-income Latinos with diabetes. Journal of Medical Internet Research. 2014;16(1):e25. doi: 10.2196/jmir.2881. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Carden L, Wood W. Habit formation and change. Current Opinion in Behavioral Sciences. 2018;20:117–122. doi: 10.1016/j.cobeha.2017.12.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Carroll N, Sadowski A, Laila A, Hruska V, Nixon M, Ma D, Haines J, on behalf of the Guelph Family Health Study the impact of covid-19 on health behavior, stress, financial and food security among middle to high income Canadian families with young children. Nutrients. 2020;12(8):2352. doi: 10.3390/nu12082352. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Champion VL, Skinner CS. Health behavior and health education: Theory, research, and practice. In: Glanz K, Rimer BK, Viswanath K, editors. The health belief model. Jossey-Bass; 2008. pp. 45–65. [Google Scholar]
- Eberth J, Sedlmeier P. The effects of mindfulness meditation: A meta-analysis. Mindfulness. 2012;3(3):174–189. doi: 10.1007/s12671-012-0101-x. [DOI] [Google Scholar]
- Edenfield TM, Saeed SA. An update on mindfulness meditation as a self-help treatment for anxiety and depression. Psychology Research and Behavior Management. 2012;5:131–141. doi: 10.2147/PRBM.S34937. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Goldberg SB, Imhoff-Smith T, Bolt DM, Wilson-Mendenhall CD, Dahl CJ, Davidson RJ, Rosenkranz MA. Testing the efficacy of a multicomponent, self-guided, smartphone-based meditation app: three-armed randomized controlled trial. JMIR Mental Health. 2020;7(11):e23825. doi: 10.2196/23825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gollwitzer PM. Implementation intentions: Strong effects of simple plans. American Psychologist. 1999;54(7):493–503. doi: 10.1037/0003-066X.54.7.493. [DOI] [Google Scholar]
- Green J, Huberty J, Puzia M, Stecher C. The effect of meditation and physical activity on the mental health impact of covid-19–related stress and attention to news among mobile app users in the united states: Cross-sectional survey. JMIR Mental Health. 2021;8(4):e28479. doi: 10.2196/28479. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Huberty J, Eckert R, Larkey L, Kurka J, Rodríguez De Jesús SA, Yoo W, Mesa R. Smartphone-based meditation for myeloproliferative neoplasm patients: feasibility study to inform future trials. JMIR Formative Research. 2019;3(2):e12662. doi: 10.2196/12662. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hussam R, Rabbani A, Reggiani G, Rigol N. Habit formation and rational addiction: A field experiment in handwashing. Harvard Business School BGIE Unit Working Paper; 2017. pp. 18–030. [Google Scholar]
- IBM SPSS Statistics for Macintosh (Version 27.0). (2020). [Computer software]. IBM Corp.
- Judah G, Gardner B, Aunger R. Forming a flossing habit: An exploratory study of the psychological determinants of habit formation. British Journal of Health Psychology. 2013;18(2):338–353. doi: 10.1111/j.2044-8287.2012.02086.x. [DOI] [PubMed] [Google Scholar]
- Kaushal N, Rhodes RE. Exercise habit formation in new gym members: A longitudinal study. Journal of Behavioral Medicine. 2015;38(4):652–663. doi: 10.1007/s10865-015-9640-7. [DOI] [PubMed] [Google Scholar]
- Kaushal N, Rhodes RE, Meldrum JT, Spence JC. The role of habit in different phases of exercise. British Journal of Health Psychology. 2017;22(3):429–448. doi: 10.1111/bjhp.12237. [DOI] [PubMed] [Google Scholar]
- Kerst A, Zielasek J, Gaebel W. Smartphone applications for depression: A systematic literature review and a survey of health care professionals’ attitudes towards their use in clinical practice. European Archives of Psychiatry and Clinical Neuroscience. 2020;270(2):139–152. doi: 10.1007/s00406-018-0974-3. [DOI] [PubMed] [Google Scholar]
- Kouchaki M, Smith IH. The morning morality effect: the influence of time of day on unethical behavior. Psychological Science. 2014;25(1):95–102. doi: 10.1177/0956797613498099. [DOI] [PubMed] [Google Scholar]
- Lacaille J, Sadikaj G, Nishioka M, Carrière K, Flanders J, Knäuper B. Daily mindful responding mediates the effect of meditation practice on stress and mood: The role of practice duration and adherence. Journal of Clinical Psychology. 2018;74(1):109–122. doi: 10.1002/jclp.22489. [DOI] [PubMed] [Google Scholar]
- Lally P, Gardner B. Promoting habit formation. Health Psychology Review. 2013;7(sup1):S137–S158. doi: 10.1080/17437199.2011.603640. [DOI] [Google Scholar]
- McArthur LH, Riggs A, Uribe F, Spaulding TJ. Health belief model offers opportunities for designing weight management interventions for college students. Journal of Nutrition Education and Behavior. 2018;50(5):485–493. doi: 10.1016/j.jneb.2017.09.010. [DOI] [PubMed] [Google Scholar]
- O’Carroll RE, Chambers JA, Dennis M, Sudlow C, Johnston M. Improving adherence to medication in stroke survivors: A pilot randomised controlled trial. Annals of Behavioral Medicine. 2013;46(3):358–368. doi: 10.1007/s12160-013-9515-5. [DOI] [PubMed] [Google Scholar]
- Phillips LA, Cohen J, Burns E, Abrams J, Renninger S. Self-management of chronic illness: The role of ‘habit’ versus reflective factors in exercise and medication adherence. Journal of Behavioral Medicine. 2016;39(6):1076–1091. doi: 10.1007/s10865-016-9732-z. [DOI] [PubMed] [Google Scholar]
- Pignatiello GA, Martin RJ, Hickman RL. Decision fatigue: A conceptual analysis. Journal of Health Psychology. 2020;25(1):123–135. doi: 10.1177/1359105318763510. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Prestwich A, Lawton R, Conner M. The use of implementation intentions and the decision balance sheet in promoting exercise behaviour. Psychology and Health. 2003;18(6):707–721. doi: 10.1080/08870440310001594493. [DOI] [Google Scholar]
- Puzia ME, Huberty J, Eckert R, Larkey L, Mesa R. Associations between global mental health and response to an app-based meditation intervention in myeloproliferative neoplasm patients. Integrative Cancer Therapies. 2020;19:153473542092778. doi: 10.1177/1534735420927780. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Qualtrics. (2020). Qualtrics. https://www.qualtrics.com
- Rhodes RE, De Bruijn G-J. Automatic and motivational correlates of physical activity: Does intensity moderate the relationship? Behavioral Medicine. 2010;36(2):44–52. doi: 10.1080/08964281003774901. [DOI] [PubMed] [Google Scholar]
- Saghafi-Asl M, Aliasgharzadeh S, Asghari-Jafarabadi M. Factors influencing weight management behavior among college students: An application of the Health Belief Model. PLoS ONE. 2020;15(2):e0228058. doi: 10.1371/journal.pone.0228058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Shen H, Chen M, Cui D. Biological mechanism study of meditation and its application in mental disorders. General Psychiatry. 2020;33(4):e100214. doi: 10.1136/gpsych-2020-100214. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Souza TC, Oliveira LA, Daniel MM, Ferreira LG, Della Lucia CM, Liboredo JC, Anastácio LR. Lifestyle and eating habits before and during COVID-19 quarantine in Brazil. Public Health Nutrition. 2021;25(1):65–75. doi: 10.1017/S136898002100255X. [DOI] [PMC free article] [PubMed] [Google Scholar]
- StataCorp (Version 17). (2021). Computer software. StataCorp LLC..
- Stecher C, Berardi V, Fowers R, Christ J, Chung Y, Huberty J. Identifying app-based meditation habits and the associated mental health benefits: longitudinal observational study. Journal of Medical Internet Research. 2021;23(11):e27282. doi: 10.2196/27282. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Stecher C, Mukasa B, Linnemayr S. Uncovering a behavioral strategy for establishing new habits: Evidence from incentives for medication adherence in Uganda. Journal of Health Economics. 2021;77:102443. doi: 10.1016/j.jhealeco.2021.102443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Stecher C, Sullivan M, Huberty J. Using personalized anchors to establish routine meditation practice with a mobile app: randomized controlled trial. JMIR MHealth and UHealth. 2021;9(12):e32794. doi: 10.2196/32794. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Tang, Y.-Y., Lu, Q., Fan, M., Yang, Y., & Posner, M. I. (2012). Mechanisms of white matter changes induced by meditation. Proceedings of the National Academy of Sciences, 109(26), 10570–10574. 10.1073/pnas.1207817109. [DOI] [PMC free article] [PubMed]
- Thorneloe RJ, Griffiths CEM, Emsley R, Ashcroft DM, Cordingley L, Barker J, Benham M, Burden D, Evans I, Griffiths C, Hussain S, Kirby B, Lawson L, Mason K, McElhone K, Murphy R, Ormerod A, Owen C, Reynolds N, Warren R. Intentional and unintentional medication non-adherence in psoriasis: the role of patients’ medication beliefs and habit strength. Journal of Investigative Dermatology. 2018;138(4):785–794. doi: 10.1016/j.jid.2017.11.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wood W, Neal DT. A new look at habits and the habit-goal interface. Psychological Review. 2007;114(4):843–863. doi: 10.1037/0033-295X.114.4.843. [DOI] [PubMed] [Google Scholar]
- Wood W, Neal DT. Healthy through habit: Interventions for initiating & maintaining health behavior change. Behavioral Science & Policy. 2016;2(1):71–83. doi: 10.1353/bsp.2016.0008. [DOI] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
A de-identified version of the data is publicly available at https://doi.org/10.17605/OSF.IO/YMJQH.
The authors have full control of all primary data and allow the journal to review the data upon request.