Adherence of Mobile App-Based Surveys and Comparison With Traditional Surveys: eCohort Study

Feasibility, adherence and usability of an observational digital health study built using Apple's ResearchKit among adults aged 18-84 years

Objective

This study evaluated the Labs Without Walls app and paired Apple Watch devices for remote research among Australian adults aged 18-84.

Methods

The study app, built using Apple's open-source ResearchKit frameworks, uses a multi-timescale measurement burst design over 8-weeks. Participants downloaded the app, completed tasks over 8 weeks, and wore Apple Watch devices. Feasibility was assessed by recruitment, remote consent, and data collection without training. Adherence was measured by task completion rates. Usability was assessed by response times, a post-study survey, and qualitative feedback.

Results

228 participants (mean age 53, age range 18-84; 62.7% female) were recruited nationwide, consented remotely, and provided data. 201 (88.16%) completed the 8-week protocol. Task adherence ranged from 100% to 70.61%. Health, environmental, and sleep data were collected passively. Usability feedback was excellent, with 84% rating the app as "extremely" or "a lot" user-friendly, 88% finding alert frequency "just right," and 95.7% finding the schedule manageable. Few age or sex differences were found.

Conclusions

The Labs Without Walls app and paired Apple Watch devices are user-friendly and enable adults aged 18-84 to complete surveys, cognitive and sensory tasks, and provide passive health and environmental data. The app can be used without formal training by males and females living in Australia, including older adults. Future iterations should consider gamification and strategies to improve daily-diary survey user experience.

Feasibility pilot of a novel coaching intervention to optimize cannabis use for chronic pain management among Veterans

INTRODUCTION

Chronic pain is common among Veterans, some of whom use cannabis for pain. We conducted a feasibility pilot study of a novel coaching intervention to help Veterans optimize use of medical cannabis products for pain management (NCT06320470).

METHODS

The intervention drew from scientific literature, consultation with cannabis experts, Veteran input via a Community Advisory Board, and tenets of motivational interviewing. Participants were Veterans with chronic pain who endorsed current use or interest in using cannabis for pain management. Participants received up to 4 individual coaching sessions via videoconference, spaced approximately 2 weeks apart. We assessed feasibility (adherence, satisfaction, acceptability) and preliminary effects on pain symptoms 14 weeks after baseline. The primary outcome was the Patient Global Impression of Change (PGIC), and exploratory outcomes included domains from the Patient-Reported Outcomes Measurement Information System (PROMIS)-29.

RESULTS

Of 22 enrolled participants, 17 attended 4 coaching sessions, 2 attended 3 sessions, and 2 attended 2 sessions. Among those who completed end of intervention surveys (16/21), 87.5% were very or completely satisfied with the intervention and 81.3% rated coaching as very or extremely helpful. All participants reported improvement on the PGIC, with 63% reporting much or very much improvement. Participants reported statistically significant decreased pain intensity (7.1/10 vs. 5.7/10) and pain interference (T-score 66.3 vs. 61.8), and increased social satisfaction (T-score 41.4 vs. 44.3). Participants noted helpful intervention factors, including co-developing a personalized plan, discussing questions/concerns, and trying different approaches to cannabis-based treatment.

CONCLUSIONS

In this feasibility pilot study of coaching on cannabis use for chronic pain among Veterans, participants were satisfied with the intervention and reported clinically significant improvements in pain symptoms. Our results support evaluating this intervention in a larger, efficacy trial.

Identification of key factors related to digital health observational study adherence and retention by data-driven approaches: an exploratory secondary analysis of two prospective longitudinal studies

BACKGROUND

Longitudinal digital health studies combine passively collected information from digital devices, such as commercial wearable devices, and actively contributed data, such as surveys, from participants. Although the use of smartphones and access to the internet supports the development of these studies, challenges exist in collecting representative data due to low adherence and retention. We aimed to identify key factors related to adherence and retention in digital health studies and develop a methodology to identify factors that are associated with and might affect study participant engagement.

METHODS

In this exploratory secondary analysis, we used data from two separate prospective longitudinal digital health studies, conducted among adult participants (age ≥18 years) during the COVID-19 pandemic by the BIG IDEAs Laboratory (BIL) at Duke University (Durham, NC, USA; April 2, 2020 to May 25, 2021) and Evidation Health (San Mateo, CA, USA; April 4 to Aug 31, 2020). Prospective daily or weekly surveys were administered for up to 15 months in the BIL study and daily surveys were administered for 5 months in the Evidation Health study. We defined metrics related to adherence to assess how participants engage with longitudinal digital health studies and developed models to infer how demographic factors and the day of survey delivery might be associated with these metrics. We defined retention as the time until a participant drops out of the study. For the purpose of clustering analysis, we defined three metrics of survey adherence: (1) total number of surveys completed, (2) participation regularity (ie, frequency of filling out surveys consecutively), and (3) time of activity (ie, engagement pattern relative to enrolment time). We assessed these metrics and explored differences by age, sex, race, and day of survey delivery. We analysed the data by unsupervised clustering, survival analysis, and recurrent event analysis with multistate modelling, with analyses restricted to individuals who provided data on age, sex, and race.

FINDINGS

In the BIL study, 5784 unique participants with the required demographic data completed 388 600 unique daily surveys (mean 67 [SD 90] surveys per participant). In the Evidation Health study, 89 479 unique participants with the required demographic data completed 2 080 992 unique daily surveys (23 [32] surveys per participant). Participants were grouped into adherence clusters based on the three metrics of adherence, and we identified statistically discernible differences in age, race, and sex between clusters. Most of the individuals aged 18-29 years were observed in the clusters with low or medium adherence, whereas the oldest age group (≥60 years) was generally more represented in clusters with high adherence than younger age groups. For retention, survival analysis indicated that 18-29 years was the age group with the highest risk of exiting the study at any given point in time (BIL study, hazard ratio [HR] for 18-29 years vs ≥60 years, 1·69 [95% CI 1·53-1·86; p<0·0001]; Evidation Health study, HR 1·50 [1·47-1·53; p<0·0001]). Sex and race were not discernible predictors of retention in the BIL study. In the Evidation Health study, male participants (vs female participants; HR 0·96 [0·94-0·98]; p<0·0001) and White participants (vs Asian participants; HR 0·96 [0·93-0·98; p=0·0004) had a lower risk of study exit, and Other race participants (vs Asian participants) had a higher risk of study exit (HR 1·10 [1·06-1·14; p<0·0001]). Recurrent event analysis confirmed age as the factor most associated with adherence; for the 18-29 years age group (vs ≥60 years group), the transition intensity from an active to inactive state per day in the BIL study was 1·661 (95% CI 1·606-1·718) and in the Evidation Health study was 1·108 (1·094-1·121). Participation patterns were variable by race and sex between the studies.

INTERPRETATION

Our analyses revealed that age was consistently associated with adherence and retention, with younger participants having lower adherence and higher dropout rates than older participants. Unsupervised clustering and survival analyses are established methods in this field, whereas the use of recurrent event analysis, was, to our knowledge, the first instance of the application of this method to remote digital health data. These methods can help to understand participant engagement in digital health studies, supporting targeted measures to improve adherence and retention.

FUNDING

US National Science Foundation, US National Institutes of Health, Microsoft AI for Health, Duke Clinical and Translational Science Institute, North Carolina Biotechnology Center, Duke MEDx, Duke Bass Connections, Duke Margolis Center for Health Policy, and Duke Office of Information Technology.

Association of Depressive Symptom Trajectory With Physical Activity Collected by mHealth Devices in the Electronic Framingham Heart Study: Cohort Study

BACKGROUND

Few studies have examined the association between depressive symptom trajectories and physical activity collected by mobile health (mHealth) devices.

OBJECTIVE

We aimed to investigate if antecedent depressive symptom trajectories predict subsequent physical activity among participants in the electronic Framingham Heart Study (eFHS).

METHODS

We performed group-based multi-trajectory modeling to construct depressive symptom trajectory groups using both depressive symptoms (Center for Epidemiological Studies-Depression [CES-D] scores) and antidepressant medication use in eFHS participants who attended 3 Framingham Heart Study research exams over 14 years. At the third exam, eFHS participants were instructed to use a smartphone app for submitting physical activity index (PAI) surveys. In addition, they were provided with a study smartwatch to track their daily step counts. We performed linear mixed models to examine the association between depressive symptom trajectories and physical activity including app-based PAI and smartwatch-collected step counts over a 1-year follow-up adjusting for age, sex, wear hour, BMI, smoking status, and other health variables.

RESULTS

We identified 3 depressive symptom trajectory groups from 722 eFHS participants (mean age 53, SD 8.5 years; n=432, 60% women). The low symptom group (n=570; mean follow-up 287, SD 109 days) consisted of participants with consistently low CES-D scores, and a small proportion reported antidepressant use. The moderate symptom group (n=71; mean follow-up 280, SD 118 days) included participants with intermediate CES-D scores, who showed the highest and increasing likelihood of reporting antidepressant use across 3 exams. The high symptom group (n=81; mean follow-up 252, SD 116 days) comprised participants with the highest CES-D scores, and the proportion of antidepressant use fell between the other 2 groups. Compared to the low symptom group, the high symptom group had decreased PAI (mean difference -1.09, 95% CI -2.16 to -0.01) and the moderate symptom group walked fewer daily steps (823 fewer, 95% CI -1421 to -226) during the 1-year follow-up.

CONCLUSIONS

Antecedent depressive symptoms or antidepressant medication use was associated with lower subsequent physical activity collected by mHealth devices in eFHS. Future investigation of interventions to improve mood including via mHealth technologies to help promote people's daily physical activity is needed.

Associations of stressful life events with subthreshold depressive symptoms and major depressive disorder: The moderating role of gender

BACKGROUND

Stressful life events (SLEs) are high-risk factors for subthreshold depressive symptoms (SDS) and major depressive disorder (MDD). This study sought to assess the association of SLEs with SDS and MDD, with a focus on gender effects.

METHODS

A total of 4132 participants were recruited from 34 primary health care settings. The Stressful Life Events Screening Questionnaire (SLESQ) was used to measure SLEs that participants had experienced in the past time. The Patient Health Questionnaire 9 (PHQ-9) was used to assess SDS, and the Mini-International Neuropsychiatry Interview (MINI) depression module was used to assess the diagnosis of MDD by trained psychiatrists.

RESULTS

In our sample (N = 4132), exposure to any SLEs was more common in individuals with SDS and MDD than in non-depressed population, and the proportion of emotional abuse was relatively high (SDS: 10.6 %; MDD: 33.9 %). After adjusting for control variables, people who experienced SLEs were at a higher risk of SDS and MDD. For males, those experiencing only one event were not at a higher risk of SDS (P = 0.061). For individuals who had experienced multiple SLEs, the association between SLEs and SDS was stronger in females than males. However, the association between SLEs and MDD was stronger in males than females.

LIMITATIONS

The cross-sectional study design and self-reported SLEs.

CONCLUSIONS

SLEs were associated with the increased risks of SDS and MDD. The associations of SLEs with SDS were more robust for females than males. In contrast, the association between SLEs and MDD was stronger in males than females.

Factors associated with long-term use of digital devices in the electronic Framingham Heart Study

Long-term use of digital devices is critical for successful clinical or research use, but digital health studies are challenged by a rapid drop-off in participation. A nested e-cohort (eFHS) is embedded in the Framingham Heart Study and uses three system components: a new smartphone app, a digital blood pressure (BP) cuff, and a smartwatch. This study aims to identify factors associated with the use of individual eFHS system components over 1-year. Among 1948 eFHS enrollees, we examine participants who returned surveys within 90 days (n = 1918), and those who chose to use the smartwatch (n = 1243) and BP cuff (n = 1115). For each component, we investigate the same set of candidate predictors for usage and use generalized linear mixed models to select predictors (P < 0.1, P value from Z test statistic), adjusting for age, sex, and time (app use: 3-month period, device use: weekly). A multivariable model with the predictors selected from initial testing is used to identify factors associated with use of components (P < 0.05, P value from Z test statistic) adjusting for age, sex, and time. In multivariable models, older age is associated with higher use of all system components. Female sex and higher education levels are associated with higher completion of app-based surveys whereas higher scores for depressive symptoms, and lower than excellent self-rated health are associated with lower use of the smartwatch over the 12-month follow-up. Our findings show that sociodemographic and health related factors are significantly associated with long-term use of digital devices. Future research is needed to test interventional strategies focusing on these factors to evaluate improvement in long-term engagement.

Digital Connectedness in the Jackson Heart Study: Cross-sectional Study

BACKGROUND

Although new approaches for data collection, such as mobile technology and teleresearch, have demonstrated new opportunities for the conduct of more timely and less costly surveys in community-based studies, literature on the feasibility of conducing cardiovascular disease research using mobile health (mHealth) platforms among middle-aged and older African Americans has been limited.

OBJECTIVE

The purpose of this study was to contribute to the knowledge regarding the penetrance of internet and mobile technologies, such as cellphones or smartphones in existing large cohort studies of cardiovascular disease.

METHODS

A digital connectedness survey was conducted in the Jackson Heart Study (JHS), a Mississippi-based African American cohort study, as part of the annual follow-up calls with participants from July 2017 to February 2019.

RESULTS

Of the 4024 participants contacted, 2564 (63.7%) completed the survey. Among survey respondents, 2262 (88.2%) reported use of internet or cellphone, and 1593 (62.1%) had a smartphone. Compared to nonusers (n=302), internet or cellphone users (n=2262) were younger (mean age 80.1, SD 8.0 vs 68.2, SD 11.3 years), more likely to be affluent (n=778, 40.1% vs n=39, 15.4%), and had greater than high school education (n=1636, 72.5% vs n=85, 28.1%). Internet or cellphone users were less likely to have cardiovascular disease history compared to nonusers (136/2262, 6.6% vs 41/302, 15.8%). The prevalence of current smoking and average BMI were similar between internet or cellphone users and nonusers. Among internet or cellphone users, 1316 (58.3%) reported use of email, 504 (22.3%) reported use of apps to track or manage health, and 1269 (56.1%) expressed interest in using JHS-developed apps.

CONCLUSIONS

Our findings suggest that it is feasible to use mHealth technologies to collect survey data among African Americans already enrolled in a longitudinal study. Our findings also highlight the need for more efforts to reduce the age and education divide in access and use of internet and smartphones for tracking health and research in African American communities.

Predicting asthma attacks using connected mobile devices and machine learning: the AAMOS-00 observational study protocol

INTRODUCTION

Supported self-management empowering people with asthma to detect early deterioration and take timely action reduces the risk of asthma attacks. Smartphones and smart monitoring devices coupled with machine learning could enhance self-management by predicting asthma attacks and providing tailored feedback.We aim to develop and assess the feasibility of an asthma attack predictor system based on data collected from a range of smart devices.

METHODS AND ANALYSIS

A two-phase, 7-month observational study to collect data about asthma status using three smart monitoring devices, and daily symptom questionnaires. We will recruit up to 100 people via social media and from a severe asthma clinic, who are at risk of attacks and who use a pressurised metered dose relief inhaler (that fits the smart inhaler device).Following a preliminary month of daily symptom questionnaires, 30 participants able to comply with regular monitoring will complete 6 months of using smart devices (smart peak flow meter, smart inhaler and smartwatch) and daily questionnaires to monitor asthma status. The feasibility of this monitoring will be measured by the percentage of task completion. The occurrence of asthma attacks (definition: American Thoracic Society/European Respiratory Society Task Force 2009) will be detected by self-reported use (or increased use) of oral corticosteroids. Monitoring data will be analysed to identify predictors of asthma attacks. At the end of the monitoring, we will assess users' perspectives on acceptability and utility of the system with an exit questionnaire.

ETHICS AND DISSEMINATION

Ethics approval was provided by the East of England - Cambridge Central Research Ethics Committee. IRAS project ID: 285 505 with governance approval from ACCORD (Academic and Clinical Central Office for Research and Development), project number: AC20145. The study sponsor is ACCORD, the University of Edinburgh.Results will be reported through peer-reviewed publications, abstracts and conference posters. Public dissemination will be centred around blogs and social media from the Asthma UK network and shared with study participants.

Safety of mRNA vaccines administered during the initial 6 months of the US COVID-19 vaccination programme: an observational study of reports to the Vaccine Adverse Event Reporting System and v-safe

BACKGROUND

In December, 2020, two mRNA-based COVID-19 vaccines were authorised for use in the USA. We aimed to describe US surveillance data collected through the Vaccine Adverse Event Reporting System (VAERS), a passive system, and v-safe, a new active system, during the first 6 months of the US COVID-19 vaccination programme.

METHODS

In this observational study, we analysed data reported to VAERS and v-safe during Dec 14, 2020, to June 14, 2021. VAERS reports were categorised as non-serious, serious, or death. Reporting rates were calculated using numbers of COVID-19 doses administered as the denominator. We analysed v-safe survey reports from days 0-7 after vaccination for reactogenicity, severity (mild, moderate, or severe), and health impacts (ie, unable to perform normal daily activities, unable to work, or received care from a medical professional).

FINDINGS

During the study period, 298 792 852 doses of mRNA vaccines were administered in the USA. VAERS processed 340 522 reports: 313 499 (92·1%) were non-serious, 22 527 (6·6%) were serious (non-death), and 4496 (1·3%) were deaths. Over half of 7 914 583 v-safe participants self-reported local and systemic reactogenicity, more frequently after dose two (4 068 447 [71·7%] of 5 674 420 participants for local reactogenicity and 4 018 920 [70·8%] for systemic) than after dose one (4 644 989 [68·6%] of 6 775 515 participants for local reactogenicity and 3 573 429 [52·7%] for systemic). Injection-site pain (4 488 402 [66·2%] of 6 775 515 participants after dose one and 3 890 848 [68·6%] of 5 674 420 participants after dose two), fatigue (2 295 205 [33·9%] participants after dose one and 3 158 299 participants [55·7%] after dose two), and headache (1 831 471 [27·0%] participants after dose one and 2 623 721 [46·2%] participants after dose two) were commonly reported during days 0-7 following vaccination. Reactogenicity was reported most frequently the day after vaccination; most reactions were mild. More reports of being unable to work, do normal activities, or of seeking medical care occurred after dose two (1 821 421 [32·1%]) than after dose one (808 963 [11·9%]); less than 1% of participants reported seeking medical care after vaccination (56 647 [0·8%] after dose one and 53 077 [0·9%] after dose two).

INTERPRETATION

Safety data from more than 298 million doses of mRNA COVID-19 vaccine administered in the first 6 months of the US vaccination programme show that most reported adverse events were mild and short in duration.

FUNDING

US Centers for Disease Control and Prevention.

Screening Depressive Symptoms and Incident Major Depressive Disorder Among Chinese Community Residents Using a Mobile App-Based Integrated Mental Health Care Model: Cohort Study

Background

Depression is associated with significant morbidity and human capital costs globally. Early screening for depressive symptoms and timely depressive disorder case identification and intervention may improve health outcomes and cost-effectiveness among affected individuals. China’s public and academic communities have reached a consensus on the need to improve access to early screening, diagnosis, and treatment of depression.

Objective

This study aims to estimate the screening prevalence and associated factors of subthreshold depressive symptoms among Chinese residents enrolled in the cohort study using a mobile app–based integrated mental health care model and investigate the 12-month incidence rate and related factors of major depressive disorder (MDD) among those with subthreshold depressive symptoms.

Methods

Data were drawn from the Depression Cohort in China (DCC) study. A total of 4243 community residents aged 18 to 64 years living in Nanshan district, Shenzhen city, in Guangdong province, China, were encouraged to participate in the DCC study when visiting the participating primary health care centers, and 4066 (95.83%) residents who met the DCC study criteria were screened for subthreshold depressive symptoms using the Patient Health Questionnaire-9 at baseline. Of the 4066 screened residents, 3168 (77.91%) with subthreshold depressive symptoms were referred to hospitals to receive a psychiatric diagnosis of MDD within 12 months. Sleep duration, anxiety symptoms, well-being, insomnia symptoms, and resilience were also investigated. The diagnosis of MDD was provided by trained psychiatrists using the Mini-International Neuropsychiatric Interview. Univariate and multivariate logistic regression models were performed to explore the potential factors related to subthreshold depressive symptoms at baseline, and Cox proportional hazards models were performed to explore the potential factors related to incident MDD.

Results

Anxiety symptoms (adjusted odds ratio [AOR] 1.63, 95% CI 1.42-1.87) and insomnia symptoms (AOR 1.13, 95% CI 1.05-1.22) were associated with an increased risk of subthreshold depressive symptoms, whereas well-being (AOR 0.93, 95% CI 0.87-0.99) was negatively associated with depressive symptoms. During the follow-up period, the 12-month incidence rate of MDD among participants with subthreshold depressive symptoms was 5.97% (189/3168). After incorporating all significant variables from the univariate analyses, the multivariate Cox proportional hazards model reported that a history of comorbidities (adjusted hazard ratio [AHR] 1.49, 95% CI 1.04-2.14) and anxiety symptoms (AHR 1.13, 95% CI 1.09-1.17) were independently associated with an increased risk of incident MDD. The 5-item World Health Organization Well-Being Index was associated with a decreased risk of incident MDD (AHR 0.90, 95% CI 0.86-0.94).

Conclusions

Elevated anxiety symptoms and unfavorable general well-being were significantly associated with subthreshold depressive symptoms and incident MDD among Chinese residents in Shenzhen. Early screening for subthreshold depressive symptoms and related factors may be helpful for identifying populations at high risk of incident MDD.