Feasibility of Mobile App-based Coping Skills Training for Cardiorespiratory Failure Survivors: The Blueprint Pilot Randomized Controlled Trial

Abstract

Rationale

Psychological distress symptoms are common among patients recently hospitalized with cardiorespiratory failure, yet there are few effective postdischarge therapies that are relevant to their experiences.

Objectives

To determine the feasibility and clinical impact of two different versions of a month-long self-guided mobile app-based coping skills program called Blueprint in comparison to usual care (UC) control.

Methods

Patients hospitalized with a serious cardiopulmonary diagnoses were recruited from adult intensive care units and stepdown units at a large academic medical center. Participants with elevated psychological distress symptoms just after discharge were randomized in a 1:1:1 ratio to Blueprint with a therapist (BP/therapist), Blueprint without a therapist (BP/no therapist), or UC control. All study procedures were conducted remotely. Blueprint is a self-guided, symptom-responsive, mobile app-based adaptive coping skills program with 4 themed weeks with different daily audio, video, and text content. Participants completed surveys via the app platform at baseline and 1 and 3 months later. The primary outcome was feasibility. Additional outcomes included the HADS (Hospital Anxiety and Depression Scale) total score, the PTSS (Post-Traumatic Stress Scale), and a 100-point quality of life visual analog scale.

Results

Of 63 patients who consented, 45 (71%) with elevated distress were randomized to BP/therapist (n = 16 [36%]), BP/no therapist (n = 14 [31%]), and UC (n = 15 [33%]). Observed rates were similar to target feasibility benchmarks, including consented patients who were randomized (71.4%), retention (75.6%), and intervention adherence (97% with weekly use). Estimated mean differences (95% confidence intervals) at 1 month compared with baseline included: HADS total (BP/therapist, −3.8 [−6.7 to −0.6]; BP/no therapist, −4.2 [−7.6 to −0.0]; UC, −3.4 [−6.6 to 0.2]); PTSS (BP/therapist, −6.7 [−11.3 to −2.1]; BP/no therapist, −9.1 [−14.4 to −3.9]; UC, −4.2 [−10.8 to 2.3]); and quality of life (BP/therapist, −4.5 [−14.3 to 4.6]; BP/no therapist, 14.0 [−0.9 to 29.0]; UC, 8.7 [−3.5 to 20.9]).

Conclusions

Among survivors of cardiorespiratory failure, a mobile app-based postdischarge coping skills training intervention demonstrated evidence of feasibility and clinical impact compared with UC control. A larger trial is warranted to test the efficacy of this approach.

Clinical trial registered with ClinicalTrials.gov (NCT04329702).

Millions are hospitalized annually for cardiorespiratory conditions such as acute respiratory distress syndrome, sepsis, congestive heart failure, coronary artery disease, chronic obstructive pulmonary disease, and pneumonia, including coronavirus disease (COVID-19). As medical therapies improve and populations with these conditions grow, a greater number of hospital survivors are experiencing postdischarge physical and emotional disability, financial toxicity, and social disruption (1–3). These challenges, in turn, promote symptoms of psychological distress, including depression, anxiety, and posttraumatic stress disorder (PTSD), which are so common and persistent that they are described in the lay press (4–13). The COVID-19 pandemic also has substantially increased psychological distress symptoms worldwide (14).

Healthy coping is an active response to managing stressful events that involves adaptive cognitive and behavioral skills (e.g., relaxation, activity pacing and planning, cognitive restructuring, and problem-solving), self-efficacy (e.g., confidence in one’s ability to manage challenges), and emotion regulation (15, 16). Although adaptive coping is endorsed by cardiorespiratory disease patients as being important to their quality of life (17), they instead frequently use maladaptive coping (e.g., symptom catastrophizing, activity avoidance, and substance abuse). Maladaptive coping is strongly associated with symptoms of depression, anxiety, PTSD, and poor quality of life (17–20).

Recently, we adapted a coping skills training intervention delivered by telephone and the web to a self-guided, symptom-responsive, fully automated mobile app called Blueprint and refined the target population to include a more clinically diverse population with elevated postdischarge distress (21). The purpose of this pilot randomized trial, therefore, was to determine the intervention’s feasibility, as well as to compare the clinical impact of two different versions (Blueprint plus a therapist vs. Blueprint without a therapist) on psychological distress symptoms to usual care control.

Portions of this work were presented at the 2022 American Thoracic Society International Conference.

Methods

Setting and Oversight

The trial was conducted among patients treated at the Duke University Medical Center. Study methodologies and reporting are guided by the consolidated standards of reporting trials (CONSORT) statement extension to randomized pilot and feasibility trials (22). The Duke Institutional Review Board approved the study protocol (Protocol #00101848). An independent data safety monitoring board approved the protocol and reviewed safety. A detailed safety protocol, including real-time alerts to staff on the basis of concerning symptoms reported via the app platform, was used (see data supplement for details). The trial was registered at ClinicalTrials.gov with identifier NCT04329702. Preliminary descriptive results were previously presented at the 2022 American Thoracic Society International Research Conference.

Participants

Between December 3, 2020, and December 12, 2021, research coordinators used an electronic health record system-integrated automated screening algorithm daily to identify consecutive eligible patients from adult medical and cardiac intensive care units (ICUs) and stepdown units (see data supplement for details). Inclusion criteria during hospitalization were an age of at least 18 years, treatment in a hospital setting for at least 24 hours, acute cardiorespiratory disease (mechanical ventilation via an endotracheal tube for at least 4 h, noninvasive ventilation for at least 4 h in a 24-h period provided for acute respiratory failure, new use of supplemental oxygen of at least 6 liters per min, use of vasopressors for shock of any etiology, use of inotropes for shock of any etiology, use of pulmonary vasodilators, use of aortic balloon pump or cardiac assist device for cardiogenic shock, and/or use of diuretic intravenous drip), intact cognitive status and the absence of severe or persistent mental illness, and basic functional English fluency. Hospital-based exclusions were complex medical care expected soon after discharge (e.g., surgery, transplantation evaluation, or chemotherapy) and lack of access to either reliable smartphone or WiFi. Postdischarge inclusions were clinically important depression or anxiety symptoms, defined as a HADS (Hospital Anxiety and Depression Scale) total score of at least eight at the time of arrival home (23). Postdischarge exclusions were failure to randomize within 2 months of hospital discharge and failure to access the app within 1 month after randomization in the absence of another explanation (e.g., hospitalization).

This trial was conducted at a time when institutional restrictions prohibited nonclinical personnel from entering hospital settings because of COVID-19 precautions. Therefore, research coordinators approached all potentially eligible participants remotely using telephone or email, often aided by a short animated informational video (see data supplement), and obtained informed consent electronically. At the time a study participant was discharged home, coordinators activated the Blueprint platform to email instructions for downloading the Blueprint app. After doing so, participants completed a login procedure followed by a baseline HADS survey. Those with a qualifying HADS score completed the remainder of the full series of sociodemographic and outcomes measures and were randomized in a 1:1:1 ratio to Blueprint with therapist involvement, Blueprint without therapist involvement, or usual care control. A method of minimization procedure was used to ensure balance across strata, including service (medical vs. surgical), baseline HADS total score (less than 14 vs. at least 14), and age (<50 yr vs. ⩾50 yr) (24, 25).

Blueprint Coping Skills Training Intervention

Blueprint is a standardized program based on a cognitive behavioral theory that is designed to help patients manage psychological distress as well as symptoms such as pain and fatigue. It is delivered via a native mobile app platform for either iOS or Android smartphones on the basis of content developed collaboratively with patients and previously tested in a telephone- and web-based format (21). It provides prompts to complete daily tasks such as viewing original videos (e.g., animated demonstrations of skills and advice from actual patients and clinicians), listening to audio files (e.g., breathing and relaxation exercises), engaging in interactive problem-solving exercises, and learning to use their current stressors as context for real-life application of learned cognitive and behavioral coping skills. The app is complemented by a 36-page printed (or PDF available at the study website) workbook that contains additional exercises, information, and quick response codes linked to specific mobile app content. The app prompts participants to complete the HADS weekly and then displays extra video and text content relevant to the current week’s skills to those with either a HADS total score of at least 15 or a current HADS score that exceeds the previous week’s score. The data supplement details the usability testing for the app, as well as intervention themes by week, app screenshots, workbook examples, and the programmatic plan for app content.

This trial evaluated two versions of Blueprint. The Blueprint plus therapist version included the input of a doctoral-level psychologist (S.A.K.) who called the participant at the time of randomization to explain the rationale for the intervention as well as to perform a brief relaxation exercise. The therapist subsequently called participants with a weekly HADS total score that was either at least 15 or had increased from the previous week’s score. The Blueprint without therapist version included a brief introductory call from a research coordinator to describe the intervention, timeline, and technical support available.

Usual Care Control

The control group received a version of the mobile app that included scheduled study surveys but no additional content. Control participants received the same safety oversight as intervention participants and were provided with phone and email contacts for study staff.

Data Collection

Surveys were completed by all participants at baseline (i.e., time of arrival home from the hospital) as well as both 1 and 3 months postrandomization. The HADS was completed weekly during the month-long active trial phase. All participant data collection was self-reported via the app platform. Participants were compensated $25 for each completed survey. Follow-up was completed in March 2022.

Outcome Measures

Feasibility and adherence

Feasibility was the primary trial outcome and was assessed by comparing observed to a priori targets for consent (⩾70%), randomization among consented and eligible patients (⩾80%), and 1-month dropout among those randomized (⩽10%) (Table E1 in the data supplement). Analytic data were generated from the Blueprint platform to characterize participants’ use behaviors, including total uses per day, specific sessions viewed and duration of viewing per day, and completion of each day’s suggested content elements (audio and video). The adherence target was at least 75% of participants using the intervention weekly.

Psychological distress symptoms and quality of life

Secondary outcomes were assessed at 1 and 3 months postrandomization. The HADS total score at 1 month postrandomization was the primary clinical outcome. Total HADS scores can range from 0 (no distress) to 42 (high distress), whereas depression and anxiety subscale scores can range from 0 to 21; subscale scores of at least 8 indicate either likely depression or anxiety (23). The minimal clinically important difference in HADS among ICU survivors is approximately 1.7–2.0 units (26, 27). Psychological distress symptoms were measured with the HADS and the PTSS (Post-Traumatic Stress Symptom (PTSS) inventory. PTSS scores range from 10 (no symptoms) to 70 (high burden of symptoms). A score greater than 20 represents clinically important PTSD symptoms, whereas a score of at least 34 suggests likely PTSD (28). Quality of life was measured with a 100-point visual analog scale (range from 0 [worst] to 100 [best]) (29).

Sociodemographics and clinical course

Study staff abstracted clinical data from the electronic health record, including chronic medical comorbidities, service (medical vs. surgical), use of psychiatric medications at the time of admission, type and duration of both respiratory and cardiac supports, major surgery, length of stay, and discharge disposition. Participants reported their age, gender, race, ethnicity, employment status, insurance status, marital status, presence of children at home, limitations in basic and instrumental activities of daily living (30, 31), financial distress (21), perceived social support, use of psychological therapies, and distress associated with eight common physical symptoms (score range from 0 [no symptoms] to 16 [more serious symptoms]) (32).

Statistical Analysis

Although this exploratory trial was not intended to test Blueprint’s efficacy, its planned 45-person sample size was expected to be sufficient both to inform us about the potential challenges of delivering Blueprint in the context of postdischarge care as well as to provide meaningful confidence intervals (CIs) for our estimates such as in our past work (22, 33).

Descriptive analyses compared a priori benchmarks to observed metrics of feasibility and acceptability among participants. General linear models for longitudinal data were used to estimate mean changes and corresponding 95% CIs over time and between groups for the primary and secondary clinical outcomes. These models included coefficients for time, group by time, and the randomization stratification variable and were fit using SAS PROC MIXED (SAS software version 9.4; SAS Institute) with unstructured covariance representing the correlation between patients’ repeated measures. For each group, we report estimated scores and 95% CIs at baseline, 1 month, and 3 months, as well as differences in scores from baseline to 1 month and 3 months. In addition, the difference in differences between intervention groups for each change score is reported. Finally, to help inform the results of this pilot study and future studies, we also calculated the medium effect size for each outcome (i.e., 0.5 of the overall baseline standard deviation [SD]) for a common reference across each outcome (34).

Results

Participants

Of 210 participants approached, 63 (30.0%) provided informed consent, and 45 (71.4%) were randomized to Blueprint plus therapist (n = 16 [35.6%]), Blueprint without therapist (n = 14 [31.1%]), and usual care control (n = 15 [33.3%]) (Figure 1). Participants were middle-aged (mean age, 49.4 [SD, 12.7]), mostly female (n = 27 [60.0%]), either White or Black/African American race (n = 31 [68.9%] and n = 12 [26.7%]), and often reported financial distress (n = 18 [40%]) (Table 1). Groups generally appeared balanced, with the exception that participants in the Blueprint plus therapist group were more often Black/African American (43.8%), whereas those in the Blueprint without therapist group were more often working before hospitalization (78.6%).

Figure 1.

Consolidated standards of reporting trials (CONSORT) diagram for trial. COVID-19 = coronavirus disease; HADS = Hospital Anxiety and Depression Scale; ICU = intensive care unit.

Table 1.

Participant baseline characteristics

	Total	Blueprint with Therapist	Blueprint without Therapist	Usual Care Control
	(N = 45)	(n = 16)	(n = 14)	(n = 15)
Age (yr), mean (SD)	49.3 (13.2)	49.0 (13.7)	52.7 (11.5)	46.5 (14.1)
<50	18 (40.0)	6 (37.5)	5 (35.7)	7 (46.7)
⩾50	27 (60.0)	10 (62.5)	9 (64.3)	8 (53.3)
Female sex, n (%)	27 (60.0)	11 (68.8)	11 (78.6)	5 (33.3)
Race, n (%)
White	31 (68.9)	9 (56.3)	11 (78.6)	11 (73.4)
Black/African American	12 (26.7)	7 (43.8)	1 (7.1)	4 (26.7)
Asian	1 (2.2)	0	1 (7.1)	0
More than one race	1 (2.2)	0	1 (7.1)	0
Hispanic ethnicity, n (%)	2 (4.4)	0	2 (14.3)	0
Employment status in the month before hospitalization, n (%)
Working, homemaker, or full-time student	28 (62.2)	7 (43.8)	11 (78.6)	10 (66.7)
Working part-time	1 (2.2)	0 (0.0)	1 (7.1)	0 (0.0)
Unemployed	2 (4.4)	1 (6.3)	0 (0.0)	1 (6.7)
Retired	4 (8.9)	2 (12.5)	0 (0.0)	2 (13.3)
Disabled	10 (22.2)	6 (37.5)	2 (14.3)	2 (13.3)
Caring for children at home, n (%)	17 (37.8)	9 (56.3)	4 (28.6)	4 (26.7)
Insurance status, n (%)
Commercial or other	16 (69.6)	5 (71.4)	7 (77.8)	4 (57.1)
Medicaid	3 (13.0)	1 (14.3)	0 (0.0)	2 (28.6)
VA	1 (4.3)	0 (0.0)	1 (11.1)	0 (0.0)
None	3 (13.0)	1 (14.3)	1 (11.1)	1 (14.3)
Missing	22	9	5	8
Social support, n (%)*	28 (62.2)	11 (68.8)	9 (64.3)	8 (53.3)
Financial distress, n (%)†	18 (40.0)	6 (37.6)	6 (42.8)	6 (40.0)
Activities of daily living limitations, mean (SD)	0.8 (1.2)	0.9 (1.4)	0.7 (0.8)	0.9 (1.4)
Instrumental activities of daily living limitations, mean (SD)	3.4 (1.6)	3.2 (1.9)	3.2 (1.1)	3.7 (1.7)
Physical symptoms, mean (SD)‡	9.9 (3.0)	11.3 (2.7)	10.1 (2.7)	8.3 (2.9)
Chronic medical comorbidities, mean (SD)§	2.0 (1.2)	2.3 (1.7)	1.9 (0.9)	1.9 (0.7)
Prescribed at the time of hospital admission, n (%)
Antidepressants	10 (22.2)	5 (31.3)	5 (35.7)	0
Anxiolytics	4 (8.9)	2 (12.5)	1 (7.1)	1 (6.7)
Other psychiatric medication	0	0	0	0
Opioids	4 (8.9)	2 (12.5)	1 (7.1)	1 (6.7)

Definition of abbreviations: SD = standard deviation; VA = Veterans Affairs.

^*Defined as reporting either usually or always.

^† Defined as reporting either barely having enough to pay bills and for basic needs or being short on money and needing more to pay bills.

^‡ Patient Health Questionnaire-15 adapted to the trial as an eight-item version (pain, difficulty concentrating, weakness, feeling your heart pound or race, shortness of breath, stiffness in joints or muscles, constipation or diarrhea, and nausea/gas/indigestion).

^§ Charlson-Deyo index score.

Patients’ key qualifying diagnoses included cardiac failure (n = 25 [57.8%]), respiratory failure (n = 41 [91.1%]), or both (n = 16 [35.6%]) (Table 2). Twenty-two (48.9%) were admitted with a complication of COVID-19. The hospital area in which they received the highest degree of care was evenly split between an ICU (n = 24 [53.3%]) and a stepdown unit (n = 20 [46.7%]). At the time of hospital discharge, participants had a mean of 3.4 (SD, 1.6) limitations in instrumental activities of daily living (IADLs) and a mean physical symptom score of 9.9 (SD, 3.0). Blueprint with therapist participants had the highest physical symptom scores (mean, 11.3 [SD, 2.7]).

Table 2.

Hospital characteristics and long-term clinical outcomes

Characteristics or Outcome	Total	Blueprint with Therapist	Blueprint without Therapist	Usual Care Control
	(N = 45)	(n = 16)	(n = 14)	(n = 15)
Primary eligibility diagnosis type, n (%)*
Cardiac failure
Vasopressors or inotropes	25 (55.6)	7 (43.8)	7 (50.0)	11 (73.3)
Diuretic drip	3 (2.2)	1 (6.3)	1 (7.1)	1 (6.7)
Respiratory failure
Mechanical ventilation	29 (64.4)	9 (56.3)	9 (64.3)	11 (73.3)
Noninvasive ventilation	2 (4.4)	0	1 (7.1)	1 (6.7)
High-flow oxygen	10 (22.2)	7 (43.8)	3 (21.4)	0
Cardiac and respiratory failure	16 (35.6)	7 (43.8)	1 (7.1)	8 (53.3)
Admission diagnosis
Acute respiratory failure	18 (40.0)	7 (43.8)	6 (42.9)	5 (33.3)
Cardiac surgery	19 (42.2)	6 (37.5)	7 (50.0)	6 (40.0)
Cardiac diagnosis (not surgery)	6 (13.3)	1 (6.3)	2 (14.3)	3 (20.0)
Other	2 (4.4)	0	1 (7.1)	1 (6.7)
COVID-19–associated admission, n (%)	22 (48.9)	5 (31.3)	6 (42.9)	2 (13.3)
Major surgical procedure during hospitalization, n (%)	22 (48.9)	7 (43.8)	5 (35.7)	10 (66.7)
Care location at the time of eligibility, n (%)
Medical intensive care unit	8 (17.8)	0	7 (50.0)	1 (6.7)
Cardiac intensive care unit	4 (8.9)	2 (12.5)	2 (14.3)	0
Cardiothoracic intensive care unit	15 (33.3)	7 (43.8)	0 (0.0)	8 (53.3)
Surgical intensive care unit	3 (6.7)	1 (6.3)	0 (0.0)	2 (13.3)
Medical stepdown unit	17 (37.8)	6 (37.5)	7 (50.0)	4 (26.7)
Surgical stepdown unit	1 (2.2)	0	1 (7.1)	0
Stepdown unit length of stay, mean (SD)	6.8 (3.8)	7.9 (5.6)	5.7 (1.5)	6.5 (2.1)
ICU length of stay, mean (SD)	7.1 (7.7)	9.3 (10.9)	5.2 (3.1)	7.1 (8.2)
Hospital length of stay, mean (SD)	10.8 (8.6)	12.2 (12.6)	9.2 (3.6)	10.7 (6.9)
Prescribed at the time of hospital discharge, n (%)
Antidepressants	11 (24.4)	5 (31.3)	6 (42.9)	0
Anxiolytics	3 (6.7)	2 (12.5)	1 (7.1)	0
Other psychiatric medication	0	0	0	0
Opioids	19 (42.2)	6 (37.5)	5 (35.7)	8 (53.3)
Readmitted between discharge and 1-mo follow-up, n (%)	4 (8.9)	1 (6.3)†	2 (7.1)‡	1 (6.7)§
Readmitted between 1- and 3-mo follow-up, n (%)	5 (11.1)	1 (6.3)	4 (7.1)	0
Death within 3 mo, n (%)	2 (4.4)	0	2 (14.3)‖	0

Definition of abbreviations: COVID-19 = coronavirus disease; SD = standard deviation.

^*Numbers total greater than n = 45 because some met both cardiac and respiratory criteria.

^† Fall (n = 1), wound infection (n = 1).

^‡ Adrenal insufficiency (n = 1), wound infection (n = 1), redo lung transplantation (n = 1), stroke (n = 1), unexplained cardiac arrest (n = 1), and COVID-19 complications (n = 1).

^§ Surgical complication (n = 1).

^‖ Both deaths postoperative during a readmission.

Feasibility and Adherence

Observed trial feasibility outcomes were generally similar to targeted benchmarks, including rates of postconsent randomization (71.4% observed vs. ⩾80% target) and dropout at 1 month (8.9% vs. ⩽10%) (Table E1). The consent rate (30.8%) was far below our ⩾70% target, likely because of our complete reliance on remote phone and email approaches necessitated by pandemic restrictions instead of in-person encounters. A common reason for refusal was a reluctance to add another activity during a time they felt generally overwhelmed by the ongoing pandemic. Eleven (24.4%) dropped out during the trial, seven because of worsened illness, two changed smartphones, and two (4.4%) were lost to follow-up (Figure 1). During the trial, six (37.5%) participants in the Blueprint plus therapist group received a total of 12 calls (range 1–4 calls per participant) from the study therapist for a weekly HADS score that was either high or had increased in comparison to the previous week (Table E2). Adherence with weekly intervention tasks within the mobile app was high among the 27 (90%) who initiated the intervention. Task completion rates ranged from 77% to 198% (Figure 2). A total of 22 (81.5% vs. 75% target) were active in all 4 weeks of the intervention, whereas 26 (96%) were active in 3 or more weeks. Participants used the app 28.2 (SD, 6.9) mean days and completed 35.3 (SD, 11.2) mean tasks of a possible 33, demonstrating that they generally used it more than required in both intervention groups (Table E2).

Figure 2.

Intervention adherence. This shows weekly and total adherence with mobile app tasks (audio, video, text, and exercises) across the month-long program by Blueprint group. Adherence was defined as the completion of core tasks for the week, a number shown on the horizontal axis. Percentages could exceed 100% if users repeated core tasks.

Psychological Distress Symptoms and Quality of Life

At baseline, participants reported moderate degrees of depression and anxiety symptoms (median HADS total 15.0 [interquartile range [IQR], 11.0–19.0]), relatively high PTSD symptoms (median PTSS, 36.0 [IQR, 27.0–41.0]), and moderately diminished quality of life (median, 60.0 [IQR, 25.0–70.0]) (Table E3 and Figures E1 and E2).

Figure 3 shows survey scores by study group throughout follow-up. Within-group improvement compared with baseline at both 1 and 3 months for Blueprint groups was observed at a level exceeding 0.5 SD of the baseline survey score, representing a medium effect size on the HADS total, HADS depression, PTSS, and quality of life scales (Table 3). Notable within-group differences compared with baseline were seen at 1 month for the Blueprint with therapist and Blueprint without therapist groups in the HADS total score (−3.8 [95% CI, −6.7 to 0.9] and −4.2 [−7.6 to 0.9], respectively), HADS depression subscale (−2.1 [−3.7 to −0.4]), and −1.8 [−3.7 to 0.1]), HADS anxiety subscale (−1.7 [−3.2 to 0.2], and −2.4 [−4.1 to 0.6]), PTSS (−6.7 [−11.3 to 2.1], and −9.1 [−14.4 to 3.9]), and quality of life visual analog scale (−4.5 [−13.5 to 4.6], and 8.3 [−1.2 to 18.7]).

Figure 3.

Estimated means and 95% confidence intervals (CIs) of outcomes over time by study group. This demonstrates estimated mean scores and 95% CIs for survey scores at baseline as well as 1 and 3 months after randomization for each of the three study groups. Asterisks reflect within-group differences in survey scores at each time point that are at least moderate (i.e., greater than 0.5 standard deviations of the baseline score). HADS = Hospital Anxiety and Depression Scale; PTSS = Post-Traumatic Stress Scale; QOL = quality of life; SD = standard deviation.

Table 3.

Estimated means and 95% confidence interval of outcomes over time by intervention arm

Outcome	Timepoint	Blueprint with Therapist, Mean (95% CI)	Blueprint without Therapist, Mean (95% CI)	Control, Mean (95% CI)	Difference between Blueprint with Therapist vs. Control, Mean (95% CI)	Difference between Blueprint without Therapist vs. Control, Mean (95% CI)	0.5 SD at Baseline
HADS Total	Baseline	15.5 (13.6 to 17.3)	15.5 (13.6 to 17.3)	15.5 (13.6 to 17.3)	n/a	n/a	1.85
	1 mo	11.7 (8.5 to 14.8)	11.3 (7.7 to 14.8)	12.1 (8.7 to 15.6)	−0.4 (−4.7 to 3.8)	−0.8 (−5.4 to 3.7)
	3 mo	11.0 (7.2 to 14.8)	11.4 (6.8 to 16.1)	13.5 (9.5 to 17.5)	−2.5 (−7.8 to 2.8)	−2.1 (−8.0 to 3.9)
	Difference 1 mo  − baseline	−3.8 (−6.7 to −0.9)	−4.2 (−7.6 to −0.9)	−3.4 (−6.6 to −0.2)	−0.4 (−4.7 to 3.8)	−0.8 (−5.4 to 3.7)
	Difference 3 mo  − baseline	−4.5 (−8.3 to −0.7)	−4.1 (−8.7 to 0.5)	−2.0 (−5.9 to 2.0)	−2.5 (−7.8 to 2.8)	−2.1 (−8.0 to 3.9)
HADS Depression	Baseline	7.7 (6.6 to 8.8)	7.7 (6.6 to 8.8)	7.7 (6.6 to 8.8)	n/a	n/a	1.7
	1 mo	5.6 (3.9 to 7.4)	5.9 (3.9 to 7.9)	6.0 (4.1 to 7.9)	−0.3 (−2.7 to 2.0)	−0.1 (−2.7 to 2.5)
	3 mo	5.5 (3.4 to 7.7)	6.6 (4.0 to 9.2)	6.5 (4.2 to 8.7)	−0.9 (−3.9 to 2.1)	0.1 (−3.2 to 3.5)
	Difference 1 mo  − baseline	−2.1 (−3.7 to −0.4)	−1.8 (−3.7 to 0.1)	−1.7 (−3.5 to 0.1)	−0.3 (−2.7 to 2.0)	−0.1 (−2.7 to 2.5)
	Difference 3 mo  − baseline	−2.1 (−4.3 to 0.0)	−1.1 (−3.7 to 1.5)	−1.2 (−3.5 to 1.0)	−0.9 (−3.9 to 2.1)	0.1 (−3.2 to 3.5)
HADS Anxiety	Baseline	7.8 (6.8 to 8.8)	7.8 (6.8 to 8.8)	7.8 (6.8 to 8.8)	n/a	n/a	3.05
	1 mo	6.1 (4.4 to 7.8)	5.4 (3.5 to 7.3)	6.1 (4.3 to 7.9)	−0.0 (−2.3 to 2.2)	−0.7 (−3.1 to 1.7)
	3 mo	5.6 (3.7 to 7.5)	4.9 (2.6 to 7.2)	6.9 (4.9 to 8.9)	−1.3 (−3.9 to 1.3)	−2.0 (−4.9 to 0.9)
	Difference 1 mo  − baseline	−1.7 (−3.2 to −0.2)	−2.4 (−4.1 to −0.6)	−1.7 (−3.4 to 0.0)	−0.0 (−2.3 to 2.2)	−0.7 (−3.1 to 1.7)
	Difference 3 mo  − baseline	−2.2 (−4.1 to −0.4)	−2.9 (−5.1 to −0.6)	−0.9 (−2.8 to 1.1)	−1.3 (−3.9 to 1.3)	−2.0 (−4.9 to 0.9)
PTSS	Baseline	35.5 (32.0 to 39.1)	35.5 (32.0 to 39.1)	35.5 (32.0 to 39.1)	n/a	n/a	5.95
	1 mo	28.8 (23.7 to 33.9)	26.4 (20.6 to 32.1)	30.5 (25.0 to 36.1)	−1.7 (−8.4 to 4.9)	−4.1 (−11.3 to 3.0)
	3 mo	28.9 (22.0 to 35.9)	29.4 (21.0 to 37.8)	31.3 (24.2 to 38.4)	−2.3 (−11.5 to 6.8)	−1.9 (−12.1 to 8.4)
	Difference 1 mo  − baseline	−6.7 (−11.3 to −2.1)	−9.1 (−14.4 to −3.9)	−5.0 (−10.1 to 0.1)	−1.7 (−8.4 to 4.9)	−4.1 (−11.3 to 3.0)
	Difference 3 mo  − baseline	−6.6 (−13.0 to −0.2)	−6.1 (−14.0 to 1.8)	−4.2 (−10.8 to 2.3)	−2.3 (−11.5 to 6.8)	−1.9 (−12.1 to 8.4)
Quality of life	Baseline	58.4 (52.7 to 64.2)	58.4 (52.7 to 64.2)	58.4 (52.7 to 64.2)	n/a	n/a	9.55
	1 mo	54.0 (44.6 to 63.4)	66.7 (56.0 to 77.4)	61.8 (51.5 to 72.1)	−7.8 (−20.9 to 5.3)	4.9 (−9.1 to 18.9)
	3 mo	61.4 (50.6 to 72.3)	72.5 (58.3 to 86.7)	67.1 (55.9 to 78.3)	−5.7 (−21.3 to 9.8)	5.4 (−12.7 to 23.4)
	Difference 1 mo  − baseline	−4.5 (−13.5 to 4.6)	8.3 (−2.1 to 18.7)	3.4 (−6.6 to 13.4)	−7.8 (−20.9 to 5.3)	4.9 (−9.1 to 18.9)
	Difference 3 mo  − baseline	3.0 (−8.9 to 14.9)	14.0 (−0.9 to 29.0)	8.7 (−3.5 to 20.9)	−5.7 (−21.3 to 9.8)	5.4 (−12.7 to 23.4)

Definition of abbreviations: CI = confidence interval; HADS = Hospital Anxiety and Depression Scale; PTSS = Post-Traumatic Stress Scale; SD = standard deviation.

Bold font denotes values greater than half of the baseline standard deviation. For all scales, a negative difference indicates improvement, with the exception of quality of life, for which a positive difference indicates improvement.

Compared with control, we observed among intervention groups no differences between baseline at 1 month of at least a medium effect size in HADS, PTSS, or quality of life. At 3 months compared with control, both Blueprint groups demonstrated improvements from baseline in HADS total scores that exceeded a medium effect size (Blueprint plus therapist group −2.5 [95% CI −7.8 to 2.8], Blueprint without therapist group −2.1 [−8.0 to 3.9]). Though less than a medium effect size, numerically significant differences compared with control were largest for Blueprint groups in scores on the HADS anxiety at 3 months and the PTSS at 1 and 3 months, as well as for the Blueprint no therapist group with 3-month quality of life. Other 3-month improvements compared with control among Blueprint participants were of small effect size.

At the time of the final (3-mo) follow-up, five (11.1%) of those randomized had been readmitted at least once, and two (4.4%) died (Table 2 and Figure 1). A total of 11 (32.4%) had persistent financial distress, and participants reported a limitation in an average of 1.3 (SD, 1.8) instrumental activities of daily living (Table E3).

Discussion

In this pilot randomized controlled trial (RCT), we found support for the Blueprint mobile app-based coping skills intervention’s feasibility and clinical impact. The intervention performed similarly whether a therapist was included or not in its effect on depression, anxiety, and PTSD symptoms as well as quality of life. Adherence rates were also similar between the two groups. Therefore, this intervention shows promise as a scalable person-centered approach to addressing postdischarge psychological distress among those with a variety of cardiorespiratory conditions.

This RCT was designed to directly remedy lessons learned about eligibility criteria and intervention delivery in our previous multicenter RCT that tested a telephone- and web-based version of the intervention among ICU patients who required mechanical ventilation for 48 or more hours (21). The previous trial had no exclusion for low baseline distress, illness severity, or discharge disposition and targeted those with very severe illness. We subsequently learned that low distress symptoms soon after discharge generally did not increase substantially and that illness severity and subsequent distress were poorly correlated (5, 35). As a result, although the intervention had a significant impact on depression symptoms and quality of life among the 40% with elevated baseline distress, its overall impact was diluted by a healthy patient floor effect from the 60% with low symptoms. Furthermore, many participants’ persistently serious illnesses compromised postdischarge adherence to the intervention and retention in the study.

In the current pilot trial, we expanded eligibility to include a variety of cardiorespiratory conditions managed in both ICU and stepdown locations, added a requirement for elevated baseline distress symptoms, and delivered the intervention completely through a self-directed mobile app. In comparison to the previous RCT, whose participants had nearly identical baseline HADS scores (21), we observed higher intervention initiation rates (90% vs. 63%), active participation in all weeks of the intervention (82% vs. 34%), and overall task completion (mean, 35.3 [SD, 11.2] of a possible 33 tasks completed vs. mean of 2.7 [SD, 2.8] of a possible six phone calls), and occasionally greater impacts on outcomes. Dropout (24.4%) at 3 months was also lower than the 38.9% rate observed in the previous trial.

Unique elements of the Blueprint intervention are worth noting. First, it is a completely self-directed 1-month program that includes different daily video, audio, text, and problem-solving content augmented with a workbook. Particularly popular video content included actual ICU survivors discussing how they applied adaptive coping techniques in their own recovery. Second, Blueprint automatically provides additional personalized content if psychological distress symptoms either remain high or worsen in comparison to the previous week. Third, the entire pilot RCT was conducted completely remotely with peak enrollment during the delta COVID-19 wave, including informed consent, intervention activation and delivery, and data collection. Also, Blueprint’s unique functionalities, content that includes actual patients and clinicians, safety monitoring features, and advanced security and privacy features further differentiate it from other app-based interventions (36). Although an app in itself is not novel, Blueprint is unique because it was built from the ground up with cardiorespiratory patient input and emphasizes function and usability (the quality of the patient’s experience) across a wide variety of characteristics, including age (37).

Blueprint’s placement within past work conducted in this population should be considered. Psychological well-being is central to quality of life, yet systematic review within this population demonstrates that most interventions have little or no effect on depression, anxiety, and PTSD symptoms (38–40). Many psychological distress interventions have evaluated medications or in-person cognitive behavioral therapy among populations of stable outpatients (41–43). However, many patients prefer nonpharmacological therapy and content relevant to their unique experiences (17, 44). Furthermore, they often need easily accessible interventions because of physical disability, distance from specialized centers, financial stress, lack of mental health professionals, and worries about exposure to COVID-19 in healthcare settings (17, 45–47).

Limitations

Several challenges remain to deliver easily accessible and relevant therapies to this growing population. Although the intervention screens for the presence of elevated distress symptoms after discharge, it is still difficult to identify populations at high risk for symptoms while they are still hospitalized. Despite our enhanced eligibility criteria, nearly 30% did not possess elevated symptoms on arrival home. In addition, although dropout was lower than our goal of less than 25% and related in all but two cases because of unforeseeable illness complications, it highlights the challenge of navigating the uncertainties of often serious, persistent illnesses. The informed consent rate of 30% is the lowest such rate we have ever experienced. However, we believe that this reflects the challenge of approaching seriously ill patients at the height of a global pandemic by telephone and email within a culture of call screening and is significantly lower than rates in similar past studies that employed in-person consent procedures. Last, the importance of a human touch beyond technology remains unclear. Although dropout was lower in the intervention group with therapist involvement and participant feedback about this feature was good, the overall clinical impact appeared similar to that observed in the intervention group without therapist involvement.

Additional limitations are noteworthy. Importantly, this pilot RCT was designed to evaluate Blueprint’s feasibility and general clinical impact. Although the results are compelling, the sample size doesn’t permit conclusions to be made about efficacy. Although the study population was diverse in race, sex, and diagnoses, this pilot RCT was conducted within a single health system and, therefore, may not be generalizable to other locations. We observed the highest dropout in the group assigned to Blueprint without a therapist, potentially raising questions about this approach. Last, clinicians should note that although we excluded those with unstable medical and psychiatric conditions because of their high likelihood of dropout (21), these populations may have high degrees of distress that likely require a more comprehensive intervention.

Conclusions

A self-directed, symptoms-responsive, mobile app-based adaptive coping skills training intervention for survivors of cardiorespiratory failure called Blueprint demonstrated evidence of feasibility, usability, and impact on both psychological distress symptoms and quality of life. A larger clinical trial is warranted to determine the long-term efficacy of this promising and highly scalable intervention.