Hematopoietic cell transplantation is a high-risk but potentially curative therapy for life-threatening blood diseases [1-3]. Hematopoietic cell transplantation patients require a committed informal family caregiver or care partner (relative or friend) to provide unpaid assistance for long durations [4]. Caregivers of hematopoietic cell transplantation patients [5] perform complex medical tasks, transport and accompany patients during appointments, manage medications, monitor vital signs and fluid intake, assist with activities of daily living, and provide emotional support [6,7]. Caregivers experience immense psychological and physical risks resulting from the stresses of managing the care recipients’ as well as their own needs [8].

Caregiving demands often exceed the resources available to caregivers [7]. In particular, patients undergoing hematopoietic cell transplantation require caregiving for an extended time, and demands vary based on stage of disease at diagnosis, treatment intensity, and possible treatment complications [4]. If caregivers have to relocate with their care recipient to be close to the transplant center, financial toxicity and social isolation may further compound care demands [9]. Caregiving has also been described as a rewarding and positive experience; however, ensuring quality of life among caregivers of hematopoietic cell transplantation patients requires broad consideration of their physical, social, psychological, and spiritual demands and needs [4].

Caregivers of hematopoietic cell transplantation patients have significant needs for information about their care recipient’s laboratory results, appointments, health conditions, or treatment regimens [4,10-12]. These data are available through electronic health record portals, a secure online website allowing patients access to their personal health information [10]. Caregivers may use their care recipient’s patient portal to help them with role demands, such as managing medications, keeping up-to-date with medical diagnoses and treatments, and communicating with health care providers [13,14]. Use of the patient portal can support caregivers in managing their own and their care recipient’s health [13,15]. However, little is known about hematopoietic cell transplantation caregivers’ uptake of their own portal use (self) and use of their care recipient’s portal. Information accessed via the patient portal can be critical for reducing caregiver role ambiguity and anxiety, increasing engagement in care, and meeting information needs among hematopoietic cell transplantation caregivers [10].

The purpose of this study was to learn more about family caregivers’ use of electronic health record patient portals. Building on inpatient and outpatient interviews, we developed a survey to be distributed nationally (in the United States) to family caregivers of hematopoietic cell transplantation patients—the National Caregiver Health Survey [3,16,17]. We drew upon a nationally representative sample to (1) characterize hematopoietic cell transplantation caregivers; (2) describe their mental health and coping behaviors; and (3) examine the relationship between caregiver characteristics, mental health and coping, and caregiver self and care recipient portal use.

The survey is part of a larger multiphase project and was developed through cognitive interviews with hematopoietic cell transplantation caregivers, using verbal probing and think-aloud techniques [3,10,17-25].

The sampling frames were email distribution lists from the National Bone Marrow Transplant Link (nBMTLINK) and Blood and Marrow Transplant Information Network (BMT InfoNet); both are nonprofit patient advocacy organizations in the United States devoted to serving transplant patients and family caregivers. With institutional review board approval, the nBMTLINK and BMT InfoNet advertised and provided access (ie, through hyperlinks) to the survey in their electronic newsletters and through email distribution lists. All listed members were presumed to have been sampled. Recruitment into the lists was voluntary and opt-in. Total counts of members in the lists and noncoverage of the target population were unknown. Additional survey responses were obtained by distributing a study brochure that contained the survey URL and QR code at BMT InfoNet’s Celebrating a Second Chance at Life Survivorship Symposium (May 2-5 2019, Orlando, Florida). A waiver of informed consent documentation was obtained, and information about the survey was provided on the first screen.

Although there is no sampling error in a census (ie, all members of the email lists were sampled), there were other sources of potential error in surveys, such as nonresponse and measurement errors. The survey was implemented by the Center for Survey Research at Indiana University (LY); cognitive interview techniques [3] were used to minimize error in the development of the questionnaire.

The survey was programmed for web administration in Qualtrics (Qualtrics XM) software. The field period was May 2 to June 30, 2019. Eligibility criteria included being an unpaid informal caregiver of an hematopoietic cell transplantation recipient, an adult, and able to complete the survey online in English. A US $20 gift card was offered to respondents for survey completion. The survey duration was approximately 16 minutes.

The survey included 5 components: (1) caregiver characteristics (age, gender, race, ethnicity, marital status, educational status, employment, annual household income, relationship with care recipient, and caregiver medical conditions, for example, high blood pressure, heart disease, diabetes, arthritis, asthma, mental health disorder, cancer); (2) caregiving characteristics, responsibilities, and life experiences posttransplant (eg, care recipient’s age, gender, timing of transplant, transplant type, and transplant source, stem cell donor relationship, care duration, care burden, whether residing with the care recipient, whether caring for others in addition to the hematopoietic cell transplantation patient); (3) use of information technology, including the patient portal; (4) depression and anxiety; and (5) coping strategies [3]. For items 4 and 5, the Patient Health Questionnaire (PHQ-4), and Brief COPE were incorporated [26,27].

The PHQ-4 screens has depression and anxiety subscales consisting of 2 items each [26,28]. Respondents rate symptoms of depressed mood (eg, having little interest or pleasure in doing things) and anxiety (eg, not being able to stop or control worrying), over the past 2 weeks on a scale from 0 (not at all), to 3 (nearly every day). Subscale scores range from 0 to 6 with a cut-off score of 3, suggestive of clinically significant depressive or anxiety disorders, respectively. Higher scores indicate worse depression and anxiety, with Cronbach α=.85 when measured in a large general population sample [29].

Brief COPE is a 28-item instrument used to assess 14 different coping strategies: self-distraction, active coping, denial, alcohol and drug use, use of emotional support, use of instrumental support, behavioral disengagement, venting, positive reframing, planning, use of humor, acceptance, and religion [30]. The author provides permission to choose or adapt selected scales for use. Thus, based on cognitive interviews of hematopoietic cell transplantation caregivers, 16 items were included in the final survey [3]. Factor analysis yielded a set of 4 unique coping factors. The mean response to the component items in each factor served as each caregiver’s score for that factor.

We summarized continuous variables with means and standard deviations, and we summarized categorical variables with percentages. Logistic regression models were fit in 3 stages. First, we assessed the univariate and multivariate association of caregiver characteristics with use of the health care portal for the care recipient’s health. The multivariate model was determined by entering all variables at once and then removing one variable at a time (backward selection) until all remaining variables were statistically significant (ie, had odds ratios with 95% confidence intervals that excluded the value of 1.0). Second, we assessed univariate and multivariate associations of caregiver mental health measures with use of the health care portal for the care recipient’s health using the same approach. Third, we combined all the variables from the two multivariate models into a single combined multivariate model and further reduced variables with backward selection. These three modeling approaches were also repeated with the outcome changed to the caregiver’s use of a health care portal for their own health. The fit of all multivariate models was summarized by area under the curve (AUC), which ranges from 0.5 for a random model to 1.0 for a perfect model and quantifies how well the fitted logistic regression probabilities discriminate among caregivers who use the portal and caregivers who do not. Data were analyzed using R (version 3.6.02) in R Studio (version 1.2.5033).

A flow diagram of the survey respondents (N=948) is shown in Figure 1, and demographics are summarized in Table 1. Note that percentages are based on denominators that vary from the overall sample size of 948 due to missing data. The median age of the study population was 40 years (range 18-89 years). Most caregivers identified as female (620/944, 65.7%), were married (823/943, 87.3%), were employed (743/940, 79.0%), were White (746/940, 79.4%), were of non-Hispanic ethnicity (783/941, 83.2%), were college educated (665/945, 70.4%), and had annual household income greater than $50,000 (623/872, 71.4%). Caregiver relationships to care recipients were parent (311/946, 32.9%), adult child (274/946, 28.9%), spouse (257/946, 27.1%), and other (104/946, 11.1%; eg, grandparent, cousin, friend).

The majority of caregivers supported another individual in addition to the care recipient (644/945, 68.1%) and resided in the same household as the care recipient (786/942, 83.4%). Care demands varied from ≤20 hours per week (343/943, 36.4%), through 20 to 40 hours per week (376/943, 39.9%), to >40 hours per week (224/943, 23.7%). Duration of caregiving was almost evenly split between ≤6 months (443/944, 46.9%) and >6 months (501/944, 53.1%). Two-thirds of caregivers (629/948, 66.4%) indicated they had at least one chronic medical condition.

Caregiver mental health variables are summarized in Figure 2; 28.6% of caregivers (259/904) exceeded the cut-off score of 3 for clinically significant depression, and 21.5% (194/903) exceeded the cut-off score of 3 for clinically significant anxiety. The means of the 4 coping scales ranged from 2.5 to 3.0, suggesting the 4 coping processes were used sometimes by the average caregiver.

Care recipient demographics are summarized in Table 2. Most (658/944, 69.7%) were adults and 63.3% (598/945) were male; 50.9% (476/934) received a transplant from a related donor, 35.1% (328/934) from an unrelated donor, and 13.9% (130/934) received an autologous transplant. Cell sources for the transplants varied among bone marrow (470/935, 50.3%), peripheral blood (113/935, 37.6%), and cord blood (113/935, 12.1%).

Caregivers (597/937, 64%) accessed a health care portal for information regarding their care recipient’s health, 49% (463/937) accessed a health care portal for checking their own health information; 36.2% (340/937) accessed a health care portal for checking both (ie, self as well as care recipient’s), while 23.1% (217/937) did not access a portal for either purpose.

We report univariate correlations between demographics, mental health variables, and caregiver access of health care portals for their care recipients in Multimedia Appendix 1.

In the multivariate model of caregiver demographics, care recipient portals were more likely to be accessed by White caregivers, 40 years old or younger, married, earning an income less than $50 000, caring for their parent, and living with their care recipient (Figure 3A; AUC 0.885). In the multivariate model of caregiver mental health variables, care recipients’ portals were more likely to be accessed by caregivers with higher depression, anxiety, and emotional coping (Figure 3B, AUC 0.668). However, in the final multivariate model that included both caregiver demographics and mental health variables, caregiver depression was the only mental health variable that remained associated with caregiver use of the care recipient portal while controlling for caregiver demographics (Figure 4; AUC 0.856).

In the multivariate model of caregiver demographics, caregivers’ use of their own health care portal was more likely among White caregivers, age 40 years or younger, without a college degree, with high income (>$50,000), with care duration <6 months, and an increased number of medical comorbidities (Figure 5A; AUC 0.823). In the multivariate model of caregiver mental health variables, self-portal use was more likely with greater strategic and social support coping (Figure 5B; AUC 0.624). However, in the final multivariate model, lack of college degree, care duration, and strategic and social support coping were no longer associated with portal use (Figure 6; AUC 0.790) partially due to multicollinearity. Specifically, higher anxiety was correlated with shorter duration of caregiving, and increased use of social support coping was correlated with higher levels of education.