The Institutional Review Board at Arizona State University approved this study (STUDY00011083). All participants provided electronic consent before participating in the survey. The data sets generated or analyzed during the study are available from the corresponding author upon request.

The study was cross-sectional. Participants were paying subscribers to the Calm meditation app. Participants were recruited via email in December 2019. Calm subscribers were sent a recruitment email if they had (1) completed at least one session of Calm using a sleep-related component in the last 90 days, so as to enrich the sample with those who used sleep-related app content, and (2) opened at least one email from Calm within the last 90 days, to target more active users who would be more likely to see and open the email. Prior to consenting, potential participants were asked to verify their eligibility by confirming that they (1) were at least 18 years old and (2) were able to read and understand questions written in English. Eligible subscribers were then directed to an electronic consent form stating that by choosing to continue to the survey questions on the next page, they were agreeing to participate. They were informed that at the conclusion of the survey, they would have the option to provide their email address to be included in a drawing for one of two US $99 Amazon gift cards.

Participants completed a demographic questionnaire followed by an investigator-developed quantitative survey that included seven questions about using Calm for sleep, which are presented in Textbox 1 (see Multimedia Appendix 1 for the complete sleep survey). Specifically, there were three questions about whether or not sleep difficulties were experienced and, if so, the types of sleep disturbance participants were experiencing when they initially downloaded Calm (ie, type of difficulties, reasons for sleep difficulties, and use of sleep aids), and four questions about their app engagement (ie, reason for downloading Calm, frequency and time of Calm use, and how Calm was used at night). The survey included branching logic so that questions were only presented if they were relevant to the participant based on their previous responses describing their sleep disturbance and app engagement (eg, those who reported that they did not have sleep disturbance were not asked about the reasons they had disturbed sleep).

Respondents were categorized based on their self-reports regarding (1) the presence or absence of sleep disturbance at the time they downloaded Calm and (2) the presence or absence of an insomnia diagnosis at the time they downloaded Calm. Good sleepers reported their sleep disturbance to be “none (ie, no sleep difficulties).” Poor sleepers reported experiencing at least one type of sleep disturbance (ie, difficulty falling asleep, staying asleep, waking too early, or getting restful sleep) but did not have a diagnosis of insomnia. Those who endorsed at least one type of sleep disturbance and self-reported a diagnosis of insomnia were categorized as such (ie, insomnia diagnosis).

Data were analyzed using SPSS Statistics for Mac, version 26.0 (IBM Corp). Chi-square tests were used to compare sleep characteristics and health diagnoses across participants with and without sleep difficulties. Chi-square tests were also used to compare sleep characteristics, reasons for downloading Calm, and app-usage patterns across participants with and without sleep difficulties. Significant chi-square tests were followed up with z tests of column proportions using Bonferroni-corrected P values to adjust for multiple comparisons.

Of the 366,173 subscribers who received an invitation to participate, 130,849 opened the invitations (35.73%). There were 19,341 subscribers who clicked the link to the survey (14.78% of those who opened the email), 14,642 (11.19%) consented to participate, and 11,095 completed the full survey (completion rate 75.78%). This completion rate was likely negatively impacted by the high volume of responses immediately following the release of survey invitations, which initially overtaxed host servers such that survey loading times were substantially slowed (see Limitations section below).

Although all participants had completed at least one session with a Calm sleep component, participants were included in the analyses only if they reported that they used at least one Calm component for the specific purpose of helping their sleep (9868/11,095, 88.94% of survey completers). Approximately 10% of participants were classified as good sleepers (ie, no sleep disturbance; 1008/9868, 10.21%), 78% were poor sleepers (ie, at least one type of sleep disturbance but without an insomnia diagnosis; 7565/9868, 77.66%), and 11% reported a diagnosis of insomnia (ie, at least one type of sleep disturbance and an insomnia diagnosis; 1039/9868, 10.53%). The remaining participants (185/9868, 1.87% of survey completers) were excluded from the analyses because they either (1) reported an insomnia diagnosis but did not endorse any type of sleep disturbance (27/9868, 0.27% of survey completers) or (2) they did not respond to questions about sleep disturbance or health diagnoses (158/9868, 1.60% of survey completers).

On average, respondents were 47.2 years old (SD 14.1). The majority were White, non-Hispanic, and female (see Table 1). The median household income was US $80,000 (IQR US $50,000-$130,000), and most participants had received a higher education degree and were employed.

The most common type of sleep disturbance was difficulty falling asleep, which was endorsed by 73% of those reporting a sleep disturbance (5523/7565, 73.00%) (see Table 2).

Among those who reported sleep disturbance, the most commonly reported reasons for having sleep disturbance were racing thoughts (6220/7565, 82.22%), followed by stress or anxiety (5508/7565, 72.81%) (see Table 3). Other reasons, such as pain, were relatively infrequent, occurring in less than 20% of poor sleepers or those with insomnia. Attribution of sleep problems to racing thoughts did not differ between poor sleepers and those with insomnia; however, compared to poor sleepers, respondents with insomnia were significantly more likely to report that sleep difficulties were caused by stress or anxiety, physical pain, fluctuating hormones, nightmares, or use of medications that interfere with sleep.

Across all participants, rates of reported use of sleep aids were highest among those with insomnia and lowest among good sleepers, with the exception of practicing yoga before bed, which did not significantly differ across sleeper types (see Table 4). Those with insomnia were the most likely to use almost all listed sleep aids, followed by poor sleepers, and then good sleepers; all pairwise differences were significant. However, good sleepers were more likely than poor sleepers and those with insomnia to report using other relaxation apps for sleep, although poor sleepers and those with insomnia did not differ with regard to the use of other apps for sleep.

We conducted a cross-sectional survey of subscribers to the Calm mobile app that had used a sleep-related component within the past 90 days. First, we categorized participants as good sleepers, poor sleepers, or having a diagnosis of insomnia, and described the specific types of sleep disturbance reported by individuals within each category (eg, falling asleep, staying asleep, not waking too early, and getting restful sleep). Next, we described and compared how good sleepers, poor sleepers, and those with insomnia engage with Calm, including reasons for downloading the app, frequency of app usage, and how the app was used specifically for sleep. This is the first study to report the use of mindfulness meditation apps for sleep among mobile app consumers.

Only 10% of the respondents indicated they had no difficulty with sleep. Given the high rates of sleep disturbance among US adults, it is not surprising that sleep disturbance is highly represented in our sample. We do, however, acknowledge that the rates of sleep disturbance and insomnia in our sample may be inflated, due to our specific recruitment of subscribers who had recently used sleep components of the Calm app. Despite this, it is of interest to describe this sample and examine their perceptions regarding the causes of their sleep disturbance, other sleep remedies they may be exploring, and how they use Calm for sleep.

Across the entire sample, the most common type of sleep disturbance reported was difficulty falling asleep, closely followed by difficulty staying asleep. Poor sleepers and those with insomnia indicated that the most common reasons for disturbed sleep were racing thoughts and stress or anxiety; however, reports of sleep disturbance due to stress or anxiety were more prevalent among those with insomnia than in poor sleepers without insomnia. These results are consistent with both the larger sleep-related literature [9,10,16] and insomnia-specific literature [17-19]. Further, findings are consistent with research on the comorbidity between sleep and anxiety disorders [9,10,16].

Those with insomnia were more likely than good or poor sleepers to use Calm to help them sleep, and they used Calm for sleep more often. Those with insomnia diagnoses were also more likely to use Calm in conjunction with other sleep aids. This is consistent with research showing that current treatment options for insomnia and sleep disturbance (eg, pharmacological and behavioral therapies) are not meeting the needs of individuals in this population [20]. Given this, mindfulness-based apps that can deliver sleep-specific tools, on demand, may be an innovative strategy to improve sleep difficulties and self-manage sleep, beyond other treatment options [21].

Although most of Calm’s sleep content is designed for general relaxation and is intended to allow for flexible usage at any time of day, some components (eg, Sleep Stories) are specifically designed to be used right before bed, such that it is not surprising that most participants used Calm at night. However, rates of nighttime usage were significantly higher among poor sleepers and those with insomnia compared to rates among good sleepers (eg, 81% vs 51% used at bedtime). Although both poor sleepers and those with insomnia primarily used Calm when laying down for bed to fall asleep, those with insomnia were significantly more likely than poor sleepers to use Calm to fall back asleep after waking up at night. While falling asleep may be primarily a function of high presleep arousal, insomnia is often characterized by middle-of-the-night awakenings and challenges falling back asleep [22,23]. Problems with middle-of-the-night awakening may, to some extent, differentiate less severe sleep disturbance from more severe sleep disturbance that may warrant diagnosis and clinical attention [22,23], highlighting a specific need within this population. Our findings show that, in addition to using Calm to fall asleep at night, some Calm subscribers with insomnia may also be using the app as a means of returning to sleep after nighttime awakenings.

Additional research is needed to assess the app content that is most attractive to, and utilized by, Calm users with sleep disturbance, including those with insomnia diagnoses, and to test the effects of different types of sleep content on sleep improvements. Findings from this line of research will likely be relevant in determining the extent to which specific results may generalize to other mindfulness meditation apps. Additionally, of course, these same questions should be pursued in examining sleep-related and general meditation features of other mindfulness-based apps.

When asked to describe how they used Calm at night, most participants reported that they “try to use Calm at night on a regular basis” (ie, rather than using it on an occasional or as-needed basis). This finding is encouraging and suggests perceived benefit in using the app for regular sleep. Interestingly, good sleepers were almost twice as likely to report that they used Calm at night sometimes or occasionally, as opposed to most nights. Research on adherence to sleep interventions indicates that participants who perceive the intervention to be relevant or necessary are more consistently engaged in their prescribed treatment [24,25]. Future studies using app-based interventions for sleep should consider insomnia, or severity of baseline sleep disturbance, as a potential moderator of adherence and, by extension, treatment effects. Additionally, given that some poor sleepers and participants with insomnia (over 20%) reported that they use Calm for sleep only when experiencing acute or immediate sleep disturbance, more research on the effects of as-needed use of meditation apps is warranted.

Although this was the first study to report usage characteristics among a large sample of subscribers to a mobile app with and without self-reported sleep disturbance, there are some limitations to note. First, participants were mostly White, female, and highly educated, limiting generalizability across populations. Also, because participants had to have opened at least one email from Calm in the last 90 days, the sample was likely made up of highly engaged users of Calm. This is reflected in that, of those who opened the invitation, less than 15% chose to participate (3% of those who were sent emails to participate). Additionally, the large influx of traffic shortly following the release of email invitations temporarily overloaded the host server, and survey loading times were substantially slowed, such that survey completers were likely more committed to sharing their feedback compared to those who did not complete the entire survey. Second, as the sample had to have used a sleep component on the app in the last 90 days, we cannot assume the findings are representative of Calm users who do not use the app for sleep. However, even with this eligibility requirement, there was a portion of the sample who did not endorse sleep difficulties and those who did not use the app for sleep, such that we may be capturing at least some of the patterns associated with other uses of sleep-related content among good sleepers. Third, this was a cross-sectional survey, relying on participants’ retrospective self-reports. Future research that includes objective usage data in addition to self-reported usage would allow for a more nuanced understanding of these findings and broaden the range of research questions that could be explored. This may be particularly useful for addressing questions regarding if and how various types of sleep-related content within the app may be used differently, and the extent to which those with sleep difficulties prefer different types of sleep content than those without sleep difficulties. Additional research is needed to understand the extent to which these factors may drive or explain some findings presented here. Finally, the study was limited in scope because it did not address the efficacy of Calm’s sleep content. Given the potential for using mobile apps to deliver highly scalable, accessible interventions for sleep disturbance, future RCTs are warranted to determine the effectiveness of mindfulness meditation apps for sleep and to assess the usage and effects of different types of sleep content among those with sleep disturbance.

We are the first to report how subscribers to the Calm meditation app use the app for sleep. Our data suggest that a high proportion of Calm subscribers experiencing sleep disturbance, including insomnia, use the app to improve their sleep. Poor sleepers and those with insomnia are specifically using Calm at night and on a regular basis, suggesting they are intentionally using the app for the self-management of sleep. We describe subscribers’ use of other pharmacological and nonpharmacological sleep aids, highlighting the extent to which sleep support is needed among Calm users experiencing sleep disturbance.

Our findings have important implications for future research and the development of content and interventions for meditation apps that aim to reduce sleep disturbance. There is a need for RCTs that can determine whether consumer-based meditation mobile apps are efficacious for improving sleep in those experiencing sleep disturbance, including those diagnosed with insomnia. Additional studies are necessary to determine the dose (ie, extent of app engagement) that is necessary for reducing sleep disturbance. Our findings also offer directions for designers of consumer-based apps that aim to develop content to help users fall asleep and return to sleep after waking in the night.