Health Information Technology Utilization by Adolescent and Young Adult Aged Inpatients Undergoing Hematopoietic Cell Transplantation

Abstract

This longitudinal study examined feasibility of “Roadmap 1.0,” a modular health information application integrated with the electronic medical record, provided to 30 adolescent and young adult (AYA) inpatients 11–24 years of age undergoing hematopoietic stem cell transplantation (HSCT). Feasibility was demonstrated: 70% accessed the application. Utilization was highest the first 2 weeks of hospitalization, with the laboratory results module used most. Users' tension and fatigue were higher than nonusers' at baseline, but not hospital discharge or day 100. Results suggest AYAs utilize health information technology in ways consistent with the HSCT trajectory and Roadmap 1.0 addressed informational and psychological needs.

Introduction

Hematopoietic stem cell transplantation (HSCT) is an intensive treatment for malignant and nonmalignant conditions.¹ Although implemented for decades to treat various cancers and immune diseases, HSCT can lead to life-threatening complications.¹ Adolescent and young adult (AYA) age patients, who are already in the midst of developmentally expected physical, psychological, and social changes, endure added stress during HSCT owing to both its intensity and uncertain outcomes. The challenges of navigating health care systems further burden AYAs.^2,3 AYAs express a strong need for developmentally appropriate information about their disease, treatment, side effects, and late effects.³

Health technology applications have been proposed as a method to provide health information and encourage AYA engagement in health care management.³ However, few such applications have been developed and investigated empirically⁴ and none for HSCT inpatients of any age. This study presents findings on Roadmap 1.0 (formerly referred to as BMT Roadmap), ^5–10 a health information technology application that was designed on Apple^® (Cupertino, California, USA) iPad and provided to AYA HSCT recipients during their transplant hospital stay. Strengths of Roadmap 1.0 include its flexible modular format that provides patients with real-time patient-specific health information from the MiChart (Epic^®; Verona, WI) electronic medical record (EMR). Modules include laboratory results (labs), medications summarized in plain language, health care provider directory, phases of HSCT care, and an interactive discharge checklist.

Design and feasibility of Roadmap 1.0 among adult HSCT patients and caregivers of adult and child HSCT patients has been previously reported.^5–10 The purpose of this study was to examine feasibility of Roadmap 1.0 among AYA HSCT patients, utilization patterns for each module over the HSCT trajectory, and differences in utilization patterns among older and younger AYAs. An ancillary exploratory aim was to examine whether Roadmap 1.0 utilization and patient-reported outcomes (PROs) were associated with hospital readmission rates.

Methods

Study design

This study assessed the utilization of Roadmap 1.0 in AYAs 11–24 years of age who underwent their first HSCT from September 2015 to December 2018 for relapsed disease, persistent minimal residual disease, and high-risk disease features that are standard of care for transplant criteria. The protocol was approved by Michigan Medicine's Institutional Review Board. Written informed consent/assent was obtained accordingly. Apple iPads loaded with the Roadmap 1.0 application were given to patients at the beginning of their HSCT hospitalization and collected at discharge. A research coordinator provided tutorials on using Roadmap 1.0 and instructed participants to use the application freely while hospitalized.

Utilization of roadmap

Minutes of utilization was the primary metric of patient engagement with Roadmap 1.0. Because length of hospital stay differed among patients, ratio variables were calculated to account for differential length of access to Roadmap 1.0 across participants. See Table 1 for definitions of all utilization variables. Participants who logged on but did not enter any of the modules at any time during their hospitalization were designated “nonusers,” and were treated as a quasi-control group in analyses.

Table 1.

Description of Roadmap 1.0 Utilization Variables

Variable name	Type	Definition
Log in	Discrete/frequency	An instance when the participant entered the Roadmap 1.0 (may not imply engagement)
Engagement	Discrete/frequency	An instance when the participant entered one or more of the Roadmap 1.0 modules (requires logging in)
Days of access	Continuous	The number of days the participant had access to Roadmap 1.0a
Days of actual use	Continuous	The number of days the participant entered at least one module after logging in to the Roadmap 1.0
Total minutes of utilization	Continuous	Total minutes spent logged in to Roadmap 1.0 modules
Overall utilization	Ratio	Total minutes spent logged in to Roadmap 1.0 modules per days of actual use
Labs module utilization	Ratio	Total minutes spent logged in to the labs module (results of laboratory studies) per days of actual use
Medications module utilization	Ratio	Total minutes spent logged in to the medication module (medications categorized by indication, dosing, and schedule) per days of actual use
Health care provider module utilization	Ratio	Total minutes spent logged in to the medical provider module (yearbook style) per days of actual use
Phases of care module utilization	Ratio	Total minutes spent logged in to Roadmap 1.0 phases of care module (detailed description of each phase of HSCT) per days of actual use
Discharge checklist module utilization	Ratio	Total minutes spent logged in to Roadmap 1.0 discharge checklist module (interactive 9-item list of discharge criteria) per days of actual use

^aThe number of days the patient had access to Roadmap 1.0 may or may not be the same as length of stay.

HSCT, hematopoietic stem cell transplantation.

Patient-reported outcomes

PROs were assessed through electronic surveys on the Apple iPad at three time points: baseline (at the time of hospital admission for transplant), discharge, and day 100 post-HSCT. Global distress was measured using POMS-2, a 65-item measure that includes anger, tension, vigor, fatigue, confusion, and depression subscales. Higher scores indicate greater distress, with the exception of the vigor subscale, which is reverse scored.¹¹ Anxiety was assessed with the State-Trait Anxiety Inventory, comprising two 20-question surveys assessing State and Trait anxiety. Higher scores reflect greater anxiety.¹²

Health outcome

Hospital readmission data (number of readmissions within the first 100 days) were collected through EMR review as an objective health outcome measure.

Statistical analyses

Analyses were performed using SPSS 24 (IBM Corp.). Descriptive statistics of PROs, overall utilization per week, and specific modular use across time were calculated. Data were cleaved into two age groups consistent with age of consent/assent: those 18 years and older (18+) and those younger than 18 years (<18). Independent means t-tests were calculated to examine differences in PROs between users and nonusers. Regression analysis was used to examine whether gender or age contributed to the variance in Roadmap 1.0 engagement. Pearson's correlations were used to examine associations between Roadmap 1.0 engagement, measures of distress, and hospital readmissions.

Results

Descriptive statistics

Thirty patients 11–24 years of age (mean [M] = 18.10, standard deviation [SD] = 2.99) were enrolled. They primarily identified as male gender (n = 20, 67%) and as non-Hispanic White (n = 22, 73%). More participants underwent allogeneic (n = 25, 84%) than autologous transplantation. Participants' mean levels of distress and anxiety were within half a SD below community norms, indicating low psychological concern. These levels did not significantly change between any time points, although a trend toward worsening psychological functioning appeared over time.

Application utilization

Of the total sample (users and nonusers), 70% engaged with Roadmap 1.0 at least once during the first week of hospitalization. Thus, deploying Roadmap 1.0 is feasible. A hierarchical regression indicated that neither age nor gender contributed to the variance in overall Roadmap 1.0 utilization.

Among users (n = 22), engagement with Roadmap 1.0 was 96% and 91% for the first 2 weeks, respectively, and declined to 58% of those still hospitalized (n = 12) by week 4.

At baseline, users of Roadmap 1.0 reported higher tension (t = −2.58, p < 0.05) and fatigue (t = −2.07, p < 0.05) than nonusers (n = 8) and exhibited a trend toward greater distress (p = 0.08) (Table 2). At discharge and day 100, users and nonusers did not differ significantly in PROs.

Table 2.

Differences Between Users and Nonusers

Demographics	Users (n = 22)		Nonusers (n = 8)
	n	%	n	%
Male	13	59	7	87.5
Female	9	41	1	12.5
	M	SD	M	SD
Age, years	18.59	2.58	16.75	3.77

Baseline PROs	Users (n = 20)		Nonusers (n = 6)
	M	SD	M	SD
Total distressa	45.90	8.21	42.83	8.66
Tensionb	48.60	9.92	37.67	4.80
Anger	46.65	7.15	42.67	8.66
Confusion	45.00	7.92	38.67	7.31
Depression	46.45	6.53	50.67	16.98
Fatigueb	47.15	9.75	38.67	3.08
Vigor	48.60	7.43	45.33	12.61

Discharge PROs	Users (n = 21)		Nonusers (n = 8)
	M	SD	M	SD
Total distress	49.90	10.65	48.25	10.42
Tension	45.67	9.98	45.00	11.26
Anger	44.90	9.68	46.13	7.61
Confusion	43.57	8.63	44.63	9.64
Depression	44.90	7.36	50.75	13.90
Fatigue	47.76	9.52	45.50	9.35
Vigor	46.19	6.73	47.00	9.65

Day 100 PROs	Users (n = 18)		Nonusers (n = 8)
	M	SD	M	SD
Total distress	52.72	14.30	47.63	18.20
Tension	48.00	11.28	44.38	14.23
Anger	48.56	10.95	47.50	14.12
Confusion	47.06	11.87	43.88	14.67
Depression	50.94	12.68	55.00	21.40
Fatigue	51.11	11.66	44.63	14.19
Vigor	49.50	9.22	54.50	12.35

^ap = 0.08.

^bp < 0.05.

M = mean; n = sample size.

SD, standard deviation; PROs, patient-reported outcomes.

Users only

Users spent a mean total of 55.39 minutes (SD = 39.80) on Roadmap 1.0 throughout hospitalization, utilizing the modules a mean of 5.41 (SD = 2.81) minutes each day they accessed the application. Participants most often engaged with the labs module. Visual analysis of the data showed that Roadmap 1.0 utilization was highest at pretransplant and during the first 2 weeks post-HSCT and decreased over time. Figure 1 shows changes in usage over time.

FIG. 1.

Average minutes of module use per week for users only, excluding data of participants who did not actually have access to Roadmap 1.0 for the given week.

Gender differences

Female patients (n = 9) spent more minutes utilizing the medication module (M = 1.12, SD = 0.80) than male patients (M = 0.51, SD = 0.35; t = 2.47, p < 0.05). Gender differences were not observed in utilization of any other modules.

Age differences

For the 18+ and <18 groups, mean ages were 20.07 years (SD = 1.94) and 16 years (SD = 0.93), respectively, a marginally significant difference (p = 0.08). Of note, the two age groups did not differ in PROs or application utilization. Within the 18+ group, those who reported more tension at baseline logged in more often (r = 0.61, p < 0.05). Within the <18 group, those who reported more vigor at baseline used the phases module more (r = 0.80, p < 0.05).

Association between utilization and PROs

At baseline, the following associations were found: higher state anxiety was associated with more days of actual use (r = 0.50, p = 0.04), higher confusion was associated with less utilization of the phase module (r = −0.47, p = 0.04) and more utilization of the checklist module (r = 0.50, p = 0.03). Higher anger was associated with more utilization of the labs module (r = 0.48, p = 0.03). No other correlations reached statistical significance at baseline.

At discharge, higher confusion and higher fatigue (r = 0.48, p = 0.03; r = 0.48, p = 0.03, respectively) were associated with more utilization of the checklist module.

PROs at day 100 were not significantly associated with Roadmap 1.0 utilization during hospitalization.

Readmissions for total sample

No indices of utilization were associated with readmissions. However, more readmissions after discharge were associated with higher distress at baseline (r = 0.65, p < 0.001), higher distress at discharge (r = 0.51, p < 0.01), and lower vigor at discharge (r = −0.53, p < 0.01).

Discussion

The first aim of this study was to examine the feasibility of Roadmap 1.0 among AYA HSCT patients. Results revealed engagement with the application was high, with 70% of participants utilizing at least one module. Among users, 96% utilized the application the first week. This demonstrated feasibility of Roadmap 1.0, suggesting that it could be successful in facilitating AYAs involvement in their health care self-management. In addition, the application may have been perceived as particularly worthwhile to more vulnerable patients, as those who engaged with the application most were more fatigued/distressed when coming into the hospital.

The second aim of the study was to examine modular utilization patterns over the HSCT trajectory. The modules did not require lengthy engagement for users to extract information, a finding that is consistent with other health application studies.¹³ Utilization was initially high, followed by a steady decline that matched the content of modules and the trajectory of treatment. Of note, participants utilized different parts of the application at different stages of the transplant trajectory. For example, use of the labs module to monitor blood counts peaked ∼2 weeks post-transplant, when signs of engraftment were eagerly awaited. A few PRO variables were associated with the use of specific modules. The significant correlations between the psychological variables of confusion and state anxiety with measures of utilization at various time points suggest that the utility of Roadmap 1.0 generalizes beyond being purely informational to being potentially psychologically transformational, as well.

The third aim of the study was to investigate differences in utilization patterns among older and younger AYAs, dichotomized at age of consent/assent. Two distinctive functional groups did not emerge, as teens did not differ from young adults in application utilization or PRO variables. This outcome is consistent with theorized continuity of development between the second and third decades of life.¹⁴

An exploratory aim of the study was to examine whether PRO variables and Roadmap 1.0 utilization were associated with readmission rates. Distress at both baseline and discharge, but not day 100, were each associated with higher readmission rates, suggesting future studies should explore potential relationships between acute physical well-being, distress, and hospital readmissions. However, Roadmap 1.0 utilization was not associated with readmissions, suggesting the value of Roadmap 1.0 was most felt during the time it was available in the hospital. Future investigations could examine the impact of extending access to health information technology after hospital discharge.

In addition to testing the main hypotheses, this study also illuminated several methodological lessons. In investigations of application use, it is critical to decipher and report levels of engagement with fidelity (e.g., simply logging in to the application does not indicate engagement). Data must be considered contextually (e.g., minutes of actual use should be calculated in relation to minutes of potential access to the application). Furthermore, application utilization alone may not be useful as an outcome without considering the timing of usage. Although the application was used somewhat sparingly depending on the recovery trajectory, an in-depth contextual analysis of utilization demonstrated that participants may have been using modules when most salient (e.g., checking labs often when engraftment was anticipated and reconsulting the medications module as discharge approached).

This study is not without its limitations. First, this primarily male, White, non-Hispanic sample makes our findings difficult to generalize. In addition, although patients were provided with individual log-in credentials, we were unable to discriminate whether caregivers or other family members used the application on behalf of the patient. Finally, not all patients completed all PROs at all time points, but this would not be unexpected among patients undergoing HSCT and coping with significant post-HSCT challenges.

Nonetheless, our study had several notable strengths, including being the first to examine AYA inpatient health information technology application usage during HSCT hospitalization. Furthermore, detailed analysis of the timing of utilization provided important insight into timing of patients' perceived need for additional informational and educational support. Finally, with an age range bridging AYAs, this study demonstrated that future research into health technology applications should embrace AYAs as members of a continuous developmental group, specifically with regard to technological affinity and facility.

This study is at the forefront of application usage within HSCT populations and, therefore, much can be gleaned that would be helpful in future research. Of importance, innovative applications that leverage technology platforms are urgently needed, particularly as a means to engage HSCT AYAs. Health information technology offers the potential to overcome constraints in health care delivery limited by provider time, complicated health information, and financial pressures.¹⁵ Given the tremendous expenditure of resources required for application development, future researchers would be wise to optimize interactive components and to carefully target their content to critical times in the course of care. Overall, this study suggests that AYAs are willing to utilize applications to become true collaborators in their medical care.

Disclaimer

The content is solely the responsibility of the authors and does not necessarily represent the official views of the above-mentioned parties. No portion of this article has previously been presented or published.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

The study herein was supported by a grant from the National Institutes of Health Agency for Health care Research and Quality (“Exploratory and Developmental Grant to Improve Health Care Quality through Health Information Technology”: 1R21HS023613).