This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
Chronic conditions account for 75% of health care costs, and the impact of chronic illness is expected to grow over time. Although subjective well-being predicts better health outcomes, people with chronic conditions tend to report lower well-being. Improving well-being might mitigate costs associated with chronic illness; however, existing interventions can be difficult to access and draw from a single theoretical approach. Happify, a digital well-being intervention program drawing from multiple theoretical traditions to target well-being, has already been established as an efficacious means of improving well-being in both distressed and nondistressed users.
This study aimed to compare change in well-being over time after using Happify for users with and without a chronic condition.
Data were obtained from Happify users, a publicly available digital well-being program accessible via website or mobile phone app. Users work on tracks addressing a specific issue (eg, conquering negative thoughts) composed of games and activities based on positive psychology, cognitive behavioral therapy, and mindfulness principles. The sample included 821 users receiving at least 6 weeks’ exposure to Happify (ranging from 42 to 179 days) who met other inclusion criteria. As part of a baseline questionnaire, respondents reported demographic information (age and gender) and whether they had any of the prespecified chronic conditions: arthritis, diabetes, insomnia, multiple sclerosis, chronic pain, psoriasis, eczema, or some other condition (450 reported a chronic condition, whereas 371 did not). Subjective well-being was assessed with the Happify Scale, a 9-item measure of positive emotionality and life satisfaction. To evaluate changes in well-being over time, a mixed effects linear regression model was fit for subjective well-being, controlling for demographics and platform usage.
At baseline, users with a chronic condition had significantly lower subjective well-being (mean 38.34, SD 17.40) than users without a chronic condition (mean 43.65, SD 19.13). However, change trajectories for users with or without a chronic condition were not significantly different; both groups experienced equivalent improvements in well-being. We also found an effect for time from baseline (
Data from this study support the conclusion that users with a chronic condition experienced significant improvement over time. Despite reporting lower subjective well-being on the whole, their change trajectory while using Happify was equivalent to those without a chronic condition. Consistent with past research, users who completed more activities over a longer period showed the most improvement. In short, the presence of a chronic condition did not prevent users from showing improved well-being when using Happify.
According to the Center for Disease Control, chronic conditions are the leading cause of death and disability in the United States. Chronic conditions affect over 40% of the US population [
People with chronic conditions also account for the greatest use of health care services [
Chronic conditions are associated with lower levels of subjective well-being, which is defined as a combination of greater positive affect and life satisfaction, and lower negative affect [
Conversely, among individuals with chronic conditions, high levels of well-being benefit their physical health. Positive affect, eg, may improve survival and recovery rates among people with physical illnesses by activating the autonomic nervous system and the hypothalamic-pituitary-adrenal axis [
Given the benefits of subjective and psychological well-being among people with chronic conditions, which has been demonstrated in both correlational and experimental studies, a growing body of research has examined the impact of interventions targeting well-being, or mental health, on chronic illness symptoms. A total of 3 key theoretical traditions have been leveraged to improve well-being and mental health in chronic conditions: cognitive behavioral therapy (CBT) [
However, there are numerous barriers to accessing in-person interventions, including cost, logistics, and stigma [
In this study, we offered a digital intervention platform, Happify, which contains activities that draw from each of the 3 key theoretical approaches. The activities on Happify are adapted efficacious interventions, ie, interventions with evidence from at least two separate research studies, in different samples [
Prior research has demonstrated that using Happify can effectively increase subjective and psychological well-being. Moreover, 1 study of existing Happify users demonstrated that usage was associated with more than a 27% increase in positive emotions over the course of 8 weeks, with greater gains among high-usage participants [
In summary, we argue that improving subjective well-being is important for individuals with chronic conditions because it can help improve their physical condition, thereby reducing the associated costs [
In this study, we tested the hypothesis that Happify’s efficacy on users without chronic conditions would generalize to a sample of users who report living with a chronic condition. Specifically, we analyzed observational data using Happify to compare the trajectory of change in well-being over time experienced by users on Happify who do and do not report having chronic conditions.
Data were drawn from registered users of Happify, a publicly available digital platform that offers games and activities based on research in positive psychology, CBT, and mindfulness. Although Happify is located in the United States, the platform is available worldwide and has been localized in 8 different languages to date. Of the 821 users included in our analyses, the majority used the English language version of the platform (605/821, 73.7% of sample); the remaining users used Happify as translated into German (25/821, 3.1% of sample), Spanish (15/821, 1.8% of sample), Japanese (8/821, 1.0% of sample), French (4/821, 0.5% of sample), Portuguese (3/821, 0.4% of sample), and Chinese (1/821, 0.1% of sample).
When registering with Happify, users provided semipassive consent that their data could be used for research purposes. Specifically, to access Happify content, users were asked to agree to the following statement: “Information that we collect about you also may be combined by us with other information available to us through third parties for research and measurement purposes, including measuring the effectiveness of content, advertising, or programs. This information from other sources may include age, gender, demographic, geographic, personal interests, product purchase activity or other information.” Data from all users aged 18 years and older who created accounts on the site between October 29, 2018, and April 4, 2019 (when data were queried), were initially considered; before October 29, 2018, Happify did not ask users about their chronic condition status. Our secondary analysis of Happify consumer data was performed under the supervision of IntegReview, an independent institutional review board.
Screenshots of Happify can be found in a previous publication [
The respondent’s subjective well-being was assessed with the Happify Scale, a 9-item measure that includes a positive emotionality component and a life satisfaction component, with higher scores indicating greater well-being [
Participants were prompted to complete the Happify Scale on the day after completing the platform registration onboarding process and every 2 weeks thereafter. In each case, the assessment was optional, and users were able to exit out of the assessment without completing it if they wished. As a result, there was considerable variability in terms of how many assessments users completed and when those assessments were completed. For each individual, we calculated an average time between any 2 assessments (in days). The average of this average across the sample is 30.69 days (SD=21.10), ranging from 11.83 to 149.
Descriptive statistics were stratified by self-reported chronic condition status (yes vs no). Group differences in baseline variables were examined using chi-square tests for categorical characteristics and
To evaluate changes in well-being over time, a mixed effects linear regression model was fit for subjective well-being. The predictor variable of key interest was self-report of any of the 8 chronic conditions gathered at baseline. A binary variable was created to indicate having 1 or more chronic conditions vs none. The following covariates were included as control variables: gender, age category, number of activities completed on Happify, baseline anxiety [
Normally distributed person-specific random effects were included to account for varying numbers of follow-up assessments. To test whether changes in outcome measures differed between those reporting a chronic condition and those not reporting a chronic condition, all interaction terms between time from baseline, number of activities completed, and chronic condition status were included. Adherence to modeling assumptions was tested using residual plots (eg, Q-Q plots to examine if residuals followed a Gaussian distribution) and was met.
All computations were done in R, version 3.6.1 [
For
Change in well-being over time for users with and without a chronic condition. Facets are broken into the 25th, 50th, and 75th percentile of activities completed. Gray bands around the chronic condition lines reflect 95% CIs.
During the study period, 6801 new users created accounts and completed the baseline assessment. Of these, data were excluded from those users who never completed any activities (n=1931) or who did not complete their self-report measures in a way that makes logical sense in relation to their usage of the platform (n=1058). Specifically, to be included, they were required to use Happify within 30 days of taking their initial baseline assessment—otherwise, their baseline assessment may or may not have accurately represented their state when they started using Happify. They were also required to have taken their final assessment within 30 days of their final activity to maximize chances that self-report scores were representative of the user’s psychological state when usage was terminated. In addition, users were excluded if they did not receive a minimum of 6 weeks’ exposure to Happify. They were not required to use Happify at any particular level during that time, but they were required to at least have had access to Happify for 6 weeks or more (2939 users were excluded by this criterion). Finally, 52 participants had missing onboarding questions because of a server error and were excluded. The final sample consisted of 821 users who had access to the platform between 42 and 179 days.
A statistical power analysis was performed for sample size estimation based on data from a randomized study examining Happify’s efficacy [
Baseline sample characteristics and platform usage of the users with and without chronic conditions.
Characteristic | Chronic conditions (n=450) | No chronic conditions (n=371) | ||
Female, n (%) | 372 (82.7) | 305 (82.2) | .94 | |
|
<.001 | |||
|
18-24 | 73 (16.2) | 103 (27.8) |
|
|
25-34 | 139 (30.9) | 116 (31.3) |
|
|
35-44 | 109 (24.2) | 88 (23.7) |
|
|
45-54 | 86 (19.1) | 41 (11.1) |
|
|
55-64 | 35 (7.8) | 20 (5.4) |
|
|
65+ | 8 (1.8) | 3 (0.8) |
|
Activities completed, mean (SD) | 56.20 (86.63) | 50.50 (64.51) | .29 | |
Total time elapsed between first and last activity (in days), mean (SD) | 72.19 (27.19) | 76.23 (30.79) | .046 | |
Total number of assessments, mean (SD) | 4.18 (1.76) | 4.20 (1.69) | .86 | |
|
<.001 | |||
|
Not at all | 0 (0.0) | 346 (93.3) |
|
|
Yes, not major | 259 (57.6) | 23 (6.2) |
|
|
Yes, very much | 191 (42.4) | 2 (0.5) |
|
Arthritis, n (%) | 31 (8.5) | 0 (0.0) | <.001 | |
Chronic pain, n (%) | 75 (20.5) | 0 (0.0) | <.001 | |
Insomnia, n (%) | 92 (25.2) | 0 (0.0) | <.001 | |
Multiple sclerosis, n (%) | 3 (0.8) | 0 (0.0) | .33 | |
Psoriasis, n (%) | 7 (1.9) | 0 (0.0) | .05 | |
Diabetes, n (%) | 13 (3.6) | 0 (0.0) | .003 | |
Eczema, n (%) | 0 (0.0) | 0 (0.0) | Not applicable | |
Other chronic condition, n (%) | 269 (73.7) | 0 (0.0) | <.001 | |
Number of chronic conditions, median (IQR) | 1.00 (1.00-2.00) | 0.00 (0.00-0.00) | <.001a | |
Positive emotion score, mean (SD) | 32.34 (17.65) | 36.43 (20.52) | .002 | |
Life satisfaction score, mean (SD) | 44.78 (21.55) | 51.36 (22.55) | <.001 | |
Subjective well-being, mean (SD) | 38.34 (17.40) | 43.65 (19.13) | <.001 | |
Generalized anxiety disorder 2-item scores, median (IQR) | 4.00 (2.00-6.00) | 3.00 (2.00-5.00) | .002 |
aMann Whitney
The 2 groups were significantly different at baseline in terms of positive emotionality, as users with a chronic condition (mean 32.34, SD 17.65) had lower positive emotion scores than users without a chronic condition (mean 36.43, SD 0.52); life satisfaction baseline scores were similarly statistically different for users with a chronic condition (mean 44.78, SD 21.55) and those without a chronic condition (mean 51.36, SD 22.55), with users with a chronic condition scoring lower. Users with a chronic condition also had lower scores on the composite subjective well-being scale (mean 38.34, SD 17.40) than users without a chronic condition (mean 43.65, SD 19.13).
For those users with a chronic condition, the most common reported category was “other,” followed by insomnia, chronic pain, and arthritis. The most common number of reported conditions was 1; however, 136 users (136/450, 30.2% of the users with chronic conditions) reported having 2 or more.
For subjective well-being scores at final assessment, there were main effects for chronic condition status (
A key objective of this study was to explore whether a digital intervention could reliably improve subjective well-being among users living with a chronic condition. We were particularly interested in testing the impact of an intervention that targets subjective well-being because of the demonstrated benefits of subjective well-being, and especially positive affect, among individuals with chronic conditions such as greater self-management of their condition [
Consistent with other studies demonstrating the effectiveness of PPIs [
Although previous research typically focused on specific conditions such as chronic pain [
This study provides preliminary evidence that Happify can significantly improve subjective well-being among people with chronic conditions, despite the fact that people with chronic conditions also are more likely to suffer from more serious psychological distress. For example, although depression has a prevalence of 10% to 20% in the general population, among individuals living with a chronic condition, depression rates range from 35% to 50% [
Importantly, the burden of chronic illness can be amplified when poor mental health, especially depression, is also present. Chronic pain patients with comorbid depression and anxiety report greater pain severity and pain-related disability as well as poorer health-related quality of life than people with pain alone [
Another important direction for future research is to explore how, specifically, Happify usage helps to improve mental well-being. Previous research suggests that mindfulness programs have been effective in reducing depressive symptoms among individuals with chronic pain by reducing pain catastrophizing and psychological distress [
This study was a naturalistic, observational study of existing Happify users. Although observational studies are an important tool in the assessment of health-related outcomes [
Usage patterns observed in this study were also naturally occurring, as compared with those that may be observed in a more controlled study with participation incentives and more frequent (potentially annoying or invasive) reminders. Nevertheless, given that the Happify platform tested in this study is a commercial product, freely available to the public, and just as easy to quit as it is to sign up, dropout levels were higher than would be observed in a more controlled setting. The resulting sample consisted of only the most dedicated users. Therefore, self-selection is a concern for this type of study design. However, even randomized controlled trials (RCTs) can suffer from this, as unmeasured moderating variables may influence a participant’s willingness to participate in a randomized study [
Finally, because this study is observational, we cannot rule out the possibility that the users with a chronic condition were different from those without a chronic condition in ways we did not measure; as chronic condition status cannot be randomized, we did not have the benefit of randomization to address systematic biases. Moreover, although we collected respondent data on a number of chronic conditions, our list of chronic conditions was not exhaustive. Approximately 73.7% (605/821) of respondents in the chronic condition group self-identified as having “some other condition” for which we have no additional information. It may also be that users with and without a chronic condition differ on meaningful but unmeasured covariates. For example, as Happify is a commercial product and the analyses included in this study are secondary analyses from Happify’s consumer base and not a randomized clinical trial, we did not have access to user information that might be relevant here, such as access to other mental health-related treatments. Such differences between groups or omitted variables can contribute to biased estimates of treatment effects [
It is all too easy in the world of digital well-being interventions, the use of which is largely unregulated, to assume that an intervention that works in one population can safely be generalized to other populations. We would argue that it is important to understand who may be in the sample of consumers interacting with a digital intervention and to evaluate whether there are subgroups of users for whom the intervention fails to produce results. Although in the case of this paper, we were able to ascertain that Happify’s effects on well-being do not differ significantly between users with chronic conditions and those without chronic conditions, we could also have found that users with chronic conditions need something else; only by evaluating subgroup data can we gain confidence in our ability to generalize, as a freely available digital intervention inevitably will. In summary, this study provides valuable observational evidence of the efficacy of Happify’s use among real users living with chronic conditions under naturalistic conditions. Given these data, future research should seek to replicate these effects under more controlled conditions, such as RCTs, and explore the impact of Happify’s use on other important outcomes associated with chronic conditions such as depressive and anxiety symptoms as well as physical and health-related outcomes.
cognitive behavioral therapy
effect size
generalized anxiety disorder 2-item
mindfulness-based stress reduction
positive psychological interventions
randomized controlled trial
AP contributed substantially to the study aims and scope, wrote the initial draft of the Introduction and Discussion, and oversaw the writing team. AW contributed substantially to the revision of all sections and to the drafting of the Methods section, and also provided support in responding to review comments on methodology and drafted the revision response. GK provided substantive support in the drafting of the Introduction and Discussion sections. JS provided substantive support in the drafting of the Introduction and Discussion sections as well as general manuscript support. EB led the revision of the Introduction and Discussion sections in response to reviewer feedback. RH crafted the statistical approach and drafted and revised the Results section.
ACP, ALW, GMK, EMB, and JLS are employees of Happify. RDH is a paid consultant with Happify.