Rates and Predictors of Nonadherence to the Post Allogeneic Hematopoietic Cell Transplant Medical Regimen in Patients and Caregivers

Abstract

Background:

Allogeneic hematopoietic cell transplant (HCT) requires a complex, multicomponent medical regimen after hospital discharge. Patients must manage multiple medications, care for their catheter, minimize exposure to sources of potential infection, follow diet, exercise, and self-care guidelines, and attend frequent follow up medical appointments. Their caregivers are tasked with helping them manage the regimen. Despite its importance in preventing adverse clinical outcomes, there has been little study of regimen nonadherence and its predictors.

Objectives:

We sought to prospectively determine rates and predictors of nonadherence to components of the post-HCT medical regimen during the first 8 weeks after hospital discharge.

Study Design:

Patients (n=92) and their caregivers (n=91) (total N=183) completed interview assessments pre-HCT, and 4- and 8-weeks after hospital discharge post-HCT. Sociodemographic factors (e.g., age, sex); patient clinical status (e.g., disease type, donor type); patient and caregiver self-reported health-related factors (e.g., medical comorbidities); and patient and caregiver psychosocial factors (e.g., anxiety, depression, HCT task specific and general self-efficacy, relationship quality) were assessed pre-HCT. Nonadherence to each of 17 regimen tasks was assessed at 4- and 8-weeks post hospital discharge via self- and caregiver collateral report.

Results:

Nonadherence rates varied among tasks, with 11.2–15.7% of the sample reporting nonadherence to immunosuppressant medication, 34.8–38.6% to other types of medications, 14.6–67.4% to required infection precautions, and 27.0–68.5% to lifestyle-related behaviors (e.g., diet/exercise). Nonadherence rates were generally stable but worsened over time for lifestyle-related behaviors. The most consistent nonadherence predictors were patient and caregiver pre-HCT perceptions of lower HCT task efficacy. Higher caregiver depression, caregiver perceptions of poorer relationship with the patient, having a nonspousal caregiver, and having non-AML disease also predicted greater nonadherence in one or more areas.

Conclusions:

Rates of nonadherence vary across tasks and both patient and caregiver factors, particularly self-efficacy, predicted nonadherence. The findings highlight the importance of considering not only patient but caregiver factors in post-HCT regimen nonadherence.

Introduction

Successful allogeneic hematopoietic cell transplant (HCT) treatment for hematological disease requires a complex, multicomponent medical regimen be followed by patients with the help of their caregivers after hospital discharge.^{1, 2} Patients are most immunocompromised during the first 100 days post-HCT³, thus the initial period at home after discharge is highly important. During this immunocompromised period, patients are at particular risk for graft rejection, opportunistic infections, graft-versus-host disease, and related complications. The post-HCT medical regimen requires continued timely treatment with multiple medications including immunosuppressants, catheter line care, following daily health practices such as an immunosuppressive diet and getting exercise, and adhering to social restrictions such as avoiding crowds and other potential sources of infection.⁴ Given their physical vulnerability and the potential for adverse events, patients are required to have an informal caregiver (e.g., family member or friend) assist them with the post-transplant regimen after hospital discharge, through at least the first 100 days post-HCT. Patients and their caregivers must navigate this complex care together, managing the timing of multiple clinic appointments as well as infusions both in the clinic and at home, all while the patient recovers from HCT.

Despite the complexity of the post-HCT medical regimen and the potential for negative outcomes if the regimen is not followed, there have been only limited studies of nonadherence in this population. Nonadherence rates for oral medications in the literature range between 21%−66%, with the higher percentages found in longer-term survivors.^5–10 Little is known about nonadherence to other elements of the regimen. Our own pilot work found relatively high nonadherence rates for tasks such as following mouth care guidelines (38%) and wrapping ones port before bathing (20%).¹¹ To promote better health outcomes, it is critical that research investigates the entire set of tasks entailed by the post-HCT medical regimen that patients and their caregivers face, and prospectively determines rates of nonadherence in order to identify areas in which patients and caregivers may need more education and assistance.

Risk factors that may predict nonadherence must also be identified so that intervention strategies can be developed that successfully target modifiable factors to improve outcomes. The relatively larger literature on solid organ transplant recipients^{12, 13} indicates a significant role for psychosocial factors in adhering to medical regimens including lower levels of anxiety and depression symptoms as well as higher levels of self-efficacy. These psychosocial factors may also be important in HCT, although they have received limited examination to date, either in patients or their caregivers. Mishkin and colleagues’ recent retrospective medical chart review in HCT patients found that higher overall pre-HCT psychosocial risk was positively associated with the subsequent occurrence of a nonadherence event (e.g., intentional interaction with sick others).⁶ Ice and colleagues found that younger age and higher levels of distress among patients were positively associated with medication nonadherence.⁵ Given the importance of the caregivers in the post-HCT medical regimen, the contribution to nonadherence risk of psychosocial and other factors in caregivers has received surprisingly little research attention. In other clinical literatures, caregiver factors have been shown to be important risk factors for nonadherence to chronic disease and organ transplant regimens.^{12, 13} Although caregiver factors have been shown to be important contributors to HCT patients’ psychosocial and behavioral functioning,^{14, 15} their influence on adherence to the post-HCT medical regimen has received scant attention.

The purpose of the present study was to prospectively investigate the rates and predictors of nonadherence to the medical regimen during the first 8 weeks after hospital discharge. It is already established that patients must adhere to their regimen because nonadherence is known to be related to poorer clinical outcomes (e.g., GVHD).^{5, 7} Thus, we focused in the present paper on correlates and predictors of adherence. We examined patient-related and caregiver-related risk factors assessed pre-HCT that, based on previous research, we hypothesized would be associated with degree of nonadherence in a sample of patients with hematological malignancies who underwent allogeneic HCT and their caregivers. These factors included: sociodemographic factors (e.g., age, sex); patient clinical status factors (e.g., disease type, donor type); self-reported health-related factors (e.g., medical comorbidities); and psychosocial factors (e.g., anxiety and depression symptoms, perceptions of self-efficacy, patient-caregiver relationship quality).

Materials and Methods

Participants

Participants were adult patient and caregiver dyads recruited prior to allogeneic HCT from the University of Pittsburgh Medical Center - Hillman Cancer Center. Eligible patients were scheduled for their first allogeneic HCT, as part of a curative treatment protocol for hematological disease. They were also required to be English-speaking and have no known history of serious mental illness (e.g., psychosis). At our institution, patients are required to identify one main caregiver (typically a family member or close friend) who will help them after HCT. They are encouraged to identify other caregivers who may also assist. For this study, we sought participation from the main caregiver. All caregivers were also English speaking. Both the patient and caregiver had to agree to participate in the study for the dyad to be enrolled. All patients who underwent allogeneic HCT during our recruitment period were screened for study eligibility. Two patients were excluded during the screening process due to not meeting the eligibility criterion of having their first transplant.

Procedure

The study was approved by the University of Pittsburgh Institutional Review Board. All eligible patients were initially asked by the transplant treatment team if study staff could contact them to discuss the study. Patients provided their caregiver’s name and contact information Both patient and caregiver participants provided written informed consent and responded to all study measures and questionnaires during interviews conducted a 3 timepoints: 4–14 days prior to HCT, and 4- and 8-weeks post-hospital discharge after HCT. Patients and caregivers were each interviewed separately. Interviews in which questionnaires were administered were approximately 30–45 minutes in length, coincided with medical visits when possible, and were conducted by a trained bachelor’s-level study interviewer.

Measures

Outcomes:

Adherence to the post-HCT medical regimen.

To assess adherence at 4- and 8-weeks post-HCT, we used a modified version of the Health Habits Assessment (HHA) administered by a trained interviewer.¹² The HHA asks about adherence to the behaviors in the post-transplant medical regimen (e.g., taking immunosuppressants) and has been used in organ recipients/caregivers.¹² We created our modified version by examining our institution’s post-transplant discharge instructions and education resources given to patients, which are based on recommendations from patient education materials from the National Marrow Donor Program guidelines⁴. We have used this measure in our prior work with HCT patients/caregivers.¹¹ Our HCT-specific version of the HHA retains 7 original items (take immunosuppressant medication, take other medications, attend clinic appointments, follow diet guidelines, follow exercise guidelines, avoid tobacco, and avoid alcohol) and contains 10 items specific to the post HCT regimen (take IV meds/supplements, flush central venous catheter (CVC) line, wrap port prior to bathing, bathe, follow hand hygiene guidelines, follow mouthcare guidelines, use sunblock, avoid crowds and sick people, avoid activity with high germ exposure, and take temperature).

The HHA was designed to maximize accurate reporting (attention to item wording and responses scales to encourage truthful answers; use of trained interviewers who establish rapport with respondents and are independent from the clinical team). HHA items have demonstrated predictive validity for health outcomes in organ recipients.¹² Meta-analyses show that self-report (alone or combined with other strategies, e.g., self-report+informant report) is a more sensitive measure of nonadherence than other approaches (e.g., indirect methods like medication monitoring or biologic measures).^{13, 16, 17}

The HHA uses ordinal scales (1=not at all to 7=daily). Participants were asked about 17 tasks over the past four weeks, a recall time period consistent with the optimal recall period for various chronic disease populations discussed by the NIH Adherence Network.¹⁸ In particular, they note that a 4-week timeframe is suitable for recall of adherence to multiple-dose/multiple medication regimens, and that this recall period reduces ceiling effects relative to shorter periods. Whether or not a set task was performed adherently was determined based on the requirements of the transplant team, which are grounded in the National Marrow Donor Program guidelines.⁴ Patients and caregivers had received written and verbal education from the medical team on adherence requirements before transplant as part of standard clinical care; they also receive additional information from staff during outpatient care post-HCT. Scores were dichotomized (adherent or not adherent) based on medical team standards for each element of the regimen. For example, a patient who missed any immunosuppressant dose was classified as nonadherent to immunosuppressants.

Patients and caregivers were interviewed separately, and each answered the same series of questions about patients’ adherence to the regimen; data from the caregiver served as a collateral source of information. For each dyad at each post-HCT time point, a single measure of adherence for each task was determined by comparing responses reported separately by the patient and caregiver. For reports that conflicted (e.g., patient reported no missed medication, caregiver reported missed doses), we choose the least adherent response (e.g, in this example, the patient was classified as nonadherent). For immunosuppresant nonadherence, we did not correlate the self-reported findings with actual therapeutic levels because blood levels of immunosuppressants and other medications may be too high or low for reasons unrelated to adherence, e.g., related to patient metabolism.¹⁹

We assessed nonadherence to 17 key tasks. In order to limit our number of outcomes, we grouped the tasks into 7 conceptual components, each with 1–4 items. Within each component, we counted how many patients had no nonadherence to any task, were nonadherent to one task, or were noneadherent to two or more tasks. These separate groupings served as our primary outcomes and correspond to major regimen components: (a) how many of three categories of medications (immunosuppressants, other medications, magnesium) were missed (3 items, i.e., score range of 0–3); (b) the number of elements of line/port care not performed (2 items); (c) the number of lifestyle requirements were not met (e.g., diet; 4 items); and (d) use of alcohol or tobacco (2 items); (e) the number of personal infection precautions not taken (e.g., daily bathing; 3 items), (f) the number of social infection precautions not taken (e.g., isolation from others; 2 items). Attending clinic visits was assessed on its own (1 item) as the seventh component of adherence.

Predictors:

All predictors were assessed at baseline (prior to transplant). Sociodemographic information, including age, sex, race, financial status, education level, and type of patient-caregiver relationship (e.g., spousal), was obtained from all participants. Financial status for both patients and caregivers was assessed with one item, “Thinking about your money situation, would you say you” , (a) “do not have enough for basic requirements”, (b) “have just enough to get along on”, or (c) “are comfortable.” ²⁰

Clinical characteristics, including disease type that indicated transplant and type of donor (matched unrelated, sibling, or haploidentical nonsibling), were obtained from patients’ medical records. Patients and caregivers also completed the Brief Charlson Comorbidity Index²¹ about their own health status prior to HCT. This 11-item Index assesses the presence of comorbid medical conditions, with a higher total score indicating more serious and more comorbid conditions. For the present report, we calculated patients’ total score excluding the condition which necessitated the transplant.

Psychosocial characteristics.

Both patients and caregivers also completed the following self-report measures at baseline.

Anxiety and Depression Symptoms were assessed with the Hospital Anxiety and Depression Scale²² which contains two 7-item subscales to measure the severity of anxiety and depression symptoms and has been widely used in medical populations. Subscale scores range from 0 to 21 with higher scores indicating greater distress (among patients in the present sample, α=.86 and .87 for anxiety and depression, respectively; among caregivers, α= .87 and .71 for anxiety and depression, respectively).

Perceived Self-efficacy was measured with the General Self-Efficacy Scale ²³ which is a 10-item measure assessing beliefs about personal ability to handle new and difficult tasks, with a mean score calculated (α=.85 and .88 for patients and caregivers respectively in the present sample).

Perceived HCT task efficacy was assessed via items adapted from a measure originally developed to assess patient and caregiver efficacy in completing behavioral tasks required in heart disease care regimens.²⁴ We modified items to reflect the HCT specific tasks (e.g., catheter line care). Patients and caregivers were asked about their confidence in doing each task or ensuring it gets done (e.g., patient’s port is wrapped before bathing) rated on a 1 (not at all confident) to 10 (completely confident) scale; α=.86 for both patients and caregivers in the present sample. Two tasks, refraining from tobacco and refraining from alcohol, were not included in this efficacy measure. Because many participants reported complete confidence for many of the tasks, we scored each item on this measure in a bimodal manner for statistical analysis, thus creating a score of 1 for those who rated themselves completely confident to the task, and a score of 0 for everyone else. We further created separate task efficacy summary scores in a similar manner for each medical regimen component by matching the specific task efficacy items to the specific medical regimen tasks within each component. For example, we assigned a score of 1 for the Medication component if the participant felt completely confident on all 3 medication tasks and a score of 0 for anything less than complete confidence on all 3 tasks.

Patient-Caregiver Relationship Quality was reported by both the patient and caregiver via a 20-item measure adapted from the work of Spanier²⁵ and Pearlin and Schooler²⁶, and used in our own work with HCT dyads²⁷ as well as extensively in organ transplant populations as well as their caregivers.^28–32 This instrument yields an overall relationship quality score of 1=poor relationship to 5=excellent relationship (α=.95 and .94) for patients and caregivers, respectively.

Statistical Analyses

Descriptive statistics were used to characterize levels of nonadherence to individual medical regimen tasks, the 7 components of tasks at both 4- and 8 weeks post-hospital discharge, and potential patient-related and caregiver-related predictors (sociodemographic, clinical, and psychosocial characteristics) of these areas of nonadherence. Paired t-tests (for continuous variables and McNemar x² tests (for categorical variables) were conducted to examine change in nonadherence in each component between the 4- and 8-weeks post-hospital discharge timepoints, and to compare patients and caregivers on the sociodemographic, clinical, and psychosocial factors. Correlation coefficients were calculated to examine bivariate relationships of patient and caregiver predictors to nonadherence for each of medical regimen components at each timepoint. To examine independent predictors of nonadherence levels, multiple linear regression analyses were then conducted, entering as predictors the sociodemographic, clinical health status, and psychosocial factors which showed correlations of at least modest size (r≥.20) with a given nonadherence outcome. Separate regression analyses were performed for each nonadherence component. Data analyses were conducted using IBM SPSS version 25.

Results

One hundred eleven of 121 consecutive eligible patients approached agreed to participate. Reasons for refusal included: not wanting the caregiver to be burdened or caregiver refusal (n= 4), patient feeling too sick (n=3) and patient not interested (3). One patient-caregiver dyad withdrew prior to questionnaire completion due to not wanting an additional burden and one caregiver in another dyad was unable to complete the questionnaires due to time constraints. Thus 111 patients and 110 caregivers provided study data. Among 111 patient-caregiver dyads enrolled before transplant, 92 patients and 91 caregivers are included in this analysis. (One caregiver was enrolled but never completed any assessments; the patient’s data are included in our analyses.) Of 19 patients enrolled but not included in analyses, 3 did not undergo transplantation, 16 were not assessed after transplant (10 died, 6 were too ill or were in a skilled nursing facility and thus neither they nor their caregiver were responsible for their medical regimen). Among the 92 patients included in the analyses, 89 patients provided 4-week post-discharge assessment data (3 patients were too ill to participate). At the 8-week post-discharge assessment, 84 patients were included, with 12 patients not included for the following reasons: 7 patients died, 4 patients were too ill, and 1 patient withdrew (and thus caregiver not assessed). As shown in Table 1, most participants were white, with 34% of patients and 43% of caregivers having at least a college education. Most caregivers were spouses (60%) and slightly more patients were male (56%) consistent with the transplant population, while more caregivers were female (70%). A majority of patients had acute myelogenous leukemia (AML; 60%) and about half (52%) had an unrelated donor. Patients and caregivers had similar ages and financial situations. Table 1 also shows descriptive statistics for patient and caregiver health-related and psychosocial factors. Caregivers reported higher levels of anxiety and depression symptoms than patients, but no differences in general or task specific self-efficacy were observed. Not including the condition that necessitated the transplant, patients reported more medical comorbidities than caregivers, as well as higher perceived patient-caregiver relationship quality than caregivers. Patients also reported greater perceived relationship quality than caregivers.

Table 1.

Descriptive characteristics of participants (N=183)

	Patients (n=92)	Caregiversa (n=91)	Test statistic, b p value
Sociodemographic characteristics
Age (mean; range; SD)	53.9; 22–68; 9.9	52.2; 18–86; 13.7	1.26, .212
Sex, Female, n (%)	42 (46)	64 (70)	c ---, .009
Race, White, n (%) d	87 (95)	85 (93)	c ---, 1.00
Financial situation, n (%) e			c ---, .497
Do not have enough for basic requirements	10 (11)	6 (7)
Have just enough to get along on	38 (41)	37 (43)
Are comfortable	44 (48)	44 (50)
Highest education level, ≥ college graduate, n (%)	32(35)	39 (43)	c ---, .220
Type of Pt-Cg relationship, n (%)
Spousal		55 (60)
Pt parent-Cg Adult child		12 (13)
Pt Adult child – Cg parent		11 (12)
Sibling/sibling in law		12 (13)
Friend		2 (2)
Clinical characteristics
Disease type, n (%)
AML	55 (60)
ALL	8 (9)
MDS	14 (15)
Other	15 (16)
Type of Donor, n (%)
MUD	48 (52)
Sibling	34 (37)
Haploidentical nonsibling	10 (11)
Health-related characteristics, M (SD)
Brief Charlson Comorbidityf (0=low, 12=high)	0.86 (1.0)	0.44 (0.8)	3.25, ** .002
Psychosocial characteristics, M (SD)
Relationship quality (1=low, 5=high)	4.45 (0.5)	4.26 (0.6)	3.31, *** <.001
Anxiety (0=low, 21=high)	6.58 (4.3)	8.44 (4.0)	−3.25, ** .002
Depression (0=low, 21=high)	4.08 (3.1)	5.13 (2.8)	−2.74, ** .007
General Self-efficacy (1=low, 4=high)	3.27 (0.3)	3.22 (0.4)	0.69, .493
Task efficacy, n (%)			c ---, .523
Indicated highest possible score	29(32)	23(26)
Indicated lower than highest possible score	62(68)	65(74)

^aone caregiver signed consent but did not provide data;

^bχ² for nominal variables and paired t tests for continuous variables based on 91 dyads;

^cno test statistic is reported because McNemar test p value is reported;

^dall 5 Black patients had Black caregivers; 1 Native American caregiver;

^en may not total group number as some demographic data missing;

^fPatients’ mean score excluded the condition which necessitated the transplant.

^**p<.01;

^***p<.001

AML = Acute Myelogenous Leukemia; ALL = Acute Lymphocytic Leukemia; MDS = Myelodysplastic Syndrome; MUD = Matched Unrelated Donor; Pt = Patient; Cg = Caregiver

Nonadherence rates at 4- and 8-weeks post-hospital discharge

Figure 1 shows nonadherence rates for the 17 tasks at each time point. In general, patients reported greater levels of nonadherence to the infection precautions and lifestyle behaviors than other types of behaviors. Roughly half of the sample did not fully adhere to handwashing, avoiding crowds and sick people, and following mouth care guidelines. The highest rate of nonadherence was for daily bathing, followed by using sunblock. Table 2 shows levels of nonadherence to each component, broken down by no nonadherence to any task within the component, nonadherence to one task, and nonadherence to one or more tasks. Nonadherence appeared descriptively to worsen over time for these infection precautions and lifestyle behaviors, but significantly so only for the Lifestyle component (t=−4.09, p=.001). The factors within the Lifestyle component that appeared to be responsible for this significant change were diet and exercise: diet nonadherence increased from 37% to 56% by the 8-week assessment and exercise nonadherence increased from 27% to 42% (Figure 1). Nonadherence rates were descriptively lower for the more medical tasks, with levels of nonadherence to immunosuppressant at 11.2% at 4 weeks and 15.7% at 8 weeks. However, over one third of patients at each timepoint were nonadherent to other medications (Figure 1). Nearly twice as many patients reported nonadherence to their IV magnesium at 8 weeks, compared to 4 weeks. Substance use nonadherence remained fairly steady at 3.4%−8.3%. as did clinic appointment attendance at 7% (Figure 1).

Figure 1.

Percentage of sample nonadherent to regimen tasks at 4 and 8 weeks post-hospital discharge

Table 2.

Levels of nonadherence to medical regimen components assessed at 4- and 8-weeks post hospital discharge after HCT

Task	N(%) 4 wks post discharge 89 dyads	N(%) 8 wks post discharge 84 dyads	Comparison of components at 4 and 8 weeksa (McNemar testb) p value
Medication component			0.370
No nonadherence	57 (64.0)	49 (58.3)
Nonadherent to one task	15 (16.9)	16 (19.0)
Nonadherent to more than one task	17 (19.1)	19 (22.6)
Infection Precautions Personal component			0.480
No nonadherence	11 (12.4)	14(16.7)
Nonadherent to one task	25 (28.1)	16 (19.0)
Nonadherent to more than one task	53 (59.6)	54 (64.3)
Infection Precautions Social component			0.930
No nonadherence	32 (36.0)	31 (36.9)
Nonadherent to one task	44(49.4)	38 (45.2)
Nonadherent to more than one task	13 (14.6)	15 (17.9)
Lifestyle component			0.038*
No nonadherence	15 (16.9)	10 (11.9)
Nonadherent to one task	37(41.6)	27 (32.1)
Nonadherent to more than one task	37 (41.6)	47 (56.0)
Catheter Care component			0.372
No nonadherence	58 (65.2)	44 (53.0)
Nonadherent to one task	28 (31.5)	34 (41.0)
Nonadherent to more than one task	3 (3.4)	5 (6.0)
Substance Use component			0.453c
No nonadherence	81 (91.0)	74 (88.1)
Nonadherent to one task	8 (9.0)	9 (10.7)
Nonadherent to more than one task	0 (0)	1 (1.2)
Attend clinic appointments			1.00
No nonadherence	83 (93.3)	78 (92.9)
Nonadherence to clinic appointments	6 (6.7)	6 (7.1)

^acomparison based on 81 dyads with data at both timepoints;

^bno test statistic in McNemar test;

^cnonadherence categories collapsed due to small n in cells;

^*p<.05

Predictors of adherence components at 4- and 8-weeks post hospital discharge

We examined independent associations of pre-HCT patient factors and caregiver factors with post-HCT levels of nonadherence to each of 4 components of the regimen. Nonadherence was rare for the domains of substance use and clinical appointment attendance, so we did not examine predictors for them. Nonadherence to catheter care was bimodally distributed, thus precluding linear regression. Results are shown in Table 3 for the nonadherence outcomes at both 4- and 8-weeks post-discharge.

Table 3.

Predictors of nonadherence to four medical regimen components at 4- and 8-weeks post-hospital discharge: zero order correlations and regression coefficients.^a

	Medications		Infection Precautions, Personal		Infection Precautions, Social		Lifestyle
	r	b	r	b	r	b	r	b
4 Weeks Post-Discharge
Patient factors
Sociodemographic characteristics
Sex, female	.03	---	−.23*	−.11	−.05	---	.16	---
Type of relationship, spousal	−.09	---	−.04	---	−.26*	−.26*	−.16	---
Clinical characteristics
Donor type (MUD verus all others)	−.07	---	−.11	---	−.01	---	−.20	−.19
Psychosocial characteristics
Relationship quality	−.14	---	−.12	---	−.20	−.22	−.03	---
General self-efficacy	−.14	---	−.23*	−.14	−.05	---	.05	---
Task specific self-efficacy	−.10	---	−.28**	−.24*	−.29**	−.13	−.17	---
Caregiver factors
Psychosocial characteristics
Relationship quality	−.18	---	−.13	---	−.25*	.09	−.32**	−.31**
Depression	−.01	---	.07	---	.20	.14	.10	---
Task specific efficacy	−.30**	−.30**	−.43***	−.40***	−.41***	−.32**	−.19	−.18
Regression model	F(8,527); p=.004		F(7,618); p<.001		F(5,464); p<.001		F(6,482); p=.002
8 Weeks Post-Discharge
Patient factors
Sociodemographic characteristics
Age	.06	---	−.09	---	−.14	---	−.21*	−.18
Education, ≥college degree	.35**	.32**	.03	---	.16	---	.02	---
Financial situationb	.21	.10	.07	---	.12	---	−.14	---
Clinical characteristics
Disease type, AML (vs. all others)	−.06	---	.03	---	−.32***	−.26*	−.06	---
Psychosocial characteristics
Relationship quality	−.13	---	−.12	---	−.23*	−.05	−.10	---
General self-efficacy	−.05	---	−.20	−.17	−.03	---	−.12	---
Task specific self-efficacy	−.04	---	−.26*	−.27*	−.10	---	−.17	---
Caregiver factors
Sociodemographic characteristics
Age	.20	.15	.10	---	−.23	−.09	−.10	---
Education, ≥college degree	−.01	---	−.01	---	.24*	.18	−.05	---
Psychosocial characteristics
Relationship quality	−.17	---	−.02	---	−.34**	−.11	−.20	−.10
Anxiety	−.04	---	.01	---	.32**	−.04	.04	---
Depression	.03	---	.08	---	.42***	.30*	.01	---
General self-efficacy	−.03	---	−.04	---	−.25*	−.19	.04	---
Task specific efficacy	−.03	---	−.22**	−.21*	−.26**	−.10	−.29**	−.26*
Regression	model F(5,440); p=.002		F(4,628); p=.005		F(4,923); p<.001		F(3,691); p=.015

^aVariables were included in the multiple regression for a given outcome only if they were associated at r ≥.20 with the outcome at 4- or 8-weeks post discharge. The following variables did not meet this criterion and were omitted from the table: Ablation type; Patient and Caregiver Brief Charlson Comorbidity Index; Patient Age, Anxiety, and Depression (4 wks); Caregiver Age, Sex, and Financial Situation (4 wks).

^bParticipants endorsed Financial Situation as 1=not enough for basic requirements; 2=just enough to get along on; 3=comfortable

^*p < .05;

^**p < .01;

^***p<.001;

Abbreviations: AML=Acute Myelogenous Leukemia; MUD=Matched Unrelated;Donor; Patient n=92; Caregiver n=91

At 4 weeks post hospital discharge, the most consistent predictor across all areas was caregiver task efficacy which was an independent predictor for three out of four components: Medications, Personal infection precautions, Social infection precautions. For example, for the Medication component, lower caregiver task efficacy predicted greater nonadherence (i.e., b=−.30, p<.01, as shown in Table 3). In addition, lower patient task efficacy predicated greater nonadherence to Personal infection precautions and having a spousal caregiver predicted lower nonadherence to Social infection precautions. Poorer caregiver perceived relationship quality was related to worse nonadherence to the Lifestyle component.

At 8 weeks post hospital discharge, low caregiver task efficacy was again the most common predictor for two out of the four components: Personal infection precautions and Lifestyle. However, low patient task efficacy was also important for Personal infection precautions. We found caregiver depression symptom severity was a significant predictor of nonadherence to Social infection precautions and that higher patient education level predicted greater Medication nonadherence. Finally, having a disease other than AML was the only clinical factor that was an independent predictor of nonadherence to any component (here to Social infection precautions).

Discussion

Like many populations with complex medical conditions, the post HCT regimen is multifaceted. However, immune reconstitution following HCT necessitates additional task demands on the patient in order to minimize potential infection and other negative outcomes.^{33, 34} To our knowledge, this study is the first to examine nonadherence rates to the multiple tasks required by the post-HCT medical regimen during the critical early time period following hospital discharge, and the first to concurrently examine potential predictive factors in both hematological cancer patients and their caregivers prior to HCT. A better understanding of the predictors related to nonadherence is essential for developing effective targeted interventions to improve the clinical management of this patient population.

We found that levels of nonadherence varied substantially according to type of task. For the critical task of taking the immunosuppressant medication, 11.2 – 15.7% of the patients were nonadherent (missed at least one dose in the month prior to the assessment) over the two study assessments, which is particularly worrisome given the well-established window of heightened vulnerability of patients during the initial post-discharge period. Nonadherence to immunosuppressant therapy has been related to mild GVHD⁵ and chronic GVHD,⁷ and adherence rates below 95% have been shown to lead to acute rejection and graft loss in kidney transplant recipients.³⁵ Further, slightly more than a third of patients were nonadherent to other types of medications, and a similar percentage had trouble adhering to catheter care requirements by the time of the second assessment at 8 weeks post hospital discharge. Tasks related to minimizing exposure to potential sources of infections had even higher levels of nonadherence, with about half of patients showing nonadherence to the majority of these preventive activities. Given that nonadherence tends to increase over time in patient populations generally³⁶, this is worrisome for this population who must endure a long period of being immunocompromised. Nonadherence to tasks related to lifestyle was also high, for example, ranging from 37–56% of the patients for diet to 54–48% of the patients for mouthcare across the two assessment time points. We took a conservative view of exercise, with any type of physical activity counting as exercise which could be split up over the day in any increments adding up to 30 minutes on “most days.” Time spent walking from one’s car to the clinic, for example, counted towards exercise minutes as did any walking inside one’s house. Regardless, 27–42% of patients did not meet this criterion. Further, we found that nonadherence generally worsened over time for these lifestyle behaviors.

Efforts to better understand why HCT patients do not fully adhere to their regimens would be helpful towards developing targeted interventions to help improve adherence. One study found that HCT patients who perceived more serious consequences of their disease and believed treatment could control their cancer ate a healthier diet after transplant.³⁷ Some qualitative work in solid organ populations suggest patients do not adhere due to factors such as perceived burdensome treatment and responsibilities ³⁸ and this may be true for HCT patients. It may be that the number and complexity of required behaviors is too overwhelming and tiring to manage simultaneously. We note that nonadherence is problematic for many types of patients; nonadherence to oral medications in the general population is estimated to be at least 33%³⁹, with rates between 7%−83.5% in populations with multiple morbidities.⁴⁰ We further note that, compared to organ transplant recipients, diet and exercise nonadherence of HCT patients appears to be worse.¹³

We assessed a number of sociodemographic, clinical, and psychosocial characteristics in both patients and their caregivers prior to HCT as possible predictors of nonadherence. It should be noted first that psychosocial factors were more often determined to be independent predictors of nonadherence than sociodemographic or clinical characteristics. Second, although both patient and caregiver factors were important, overall, there were more significant caregiver-related psychosocial predictors than patient-related psychosocial predictors. Third, the most common predictor was perceived task efficacy (either patient or caregiver), which was an independent predictor for each of the four medical regimen components for at least one timepoint. Thus, the perception that they will be able to accomplish the required tasks appeared to be particularly important: lower ratings of task efficacy predicted greater nonadherence. Of note, caregiver task efficacy was more often a predictor than patient task efficacy; this finding highlights the central role of the caregiver in helping HCT patients adhere. We note that task self-efficacy was more important than general self-efficacy, perhaps because the post-HCT regimen requires the performance of novel tasks. Thus, a specific focus on the perceived ability to do these HCT specific tasks may be more relevant than overall perceived ability of oneself to accomplish things in general. Others have noted patient-reported self-efficacy for symptom management is important in HCT patients⁴¹, and the results here add to this self-efficacy literature. Importantly, self-efficacy in the area of health is modifiable⁴² and could serve as a potential target to decrease nonadherence. Fourth, we found the greatest influence of caregiver and dyad-related factors in nonadherence to social infection precautions. Overall, dyads that were nonspousal and in which the caregiver reported lower task-efficacy and greater levels of depressive symptoms had greater nonadherence. Further, we found that nonadherence to lifestyle behaviors was predicted by caregiver perceived relationship quality and caregiver task efficacy. We are not surprised by the influence of dyad-related and caregiver-related factors for both social infection precautions and lifestyle tasks as they are more likely to be part of the caregiver’s experience as well (e.g., both patient and caregiver likely overlap in their social activity, eat similar diets).

The only clinical characteristic we found to be predictive of nonadherence was whether the patient had AML or not; those with AML had less nonadherence to social infection precautions. We speculated that this may be due to their (and their caregiver) having more experience with such precautions because their type of disease necessitated prior long hospital stays for induction chemotherapy, during which they needed to follow similar social infection precaution rules. Finally, for the demographic factors, we were puzzled to find greater patient education level to predict medication nonadherence at one timepoint. We do note that the literature is mixed on the role of education level in adherence in other diseases.⁴³

This study had several limitations. Although it is unique in considering both patients and caregivers as predictors of nonadherence, our sample consisted predominantly of patient-caregiver spousal dyads. We did not have enough other types of dyads (e.g., patient-adult child caregiver) to be able to examine types of patient-caregiver relationships as predictors of nonadherence. Further, most of the sample was white and non-Hispanic, and we were thus unable to examine race/ethnicity as predictors of nonadherence. We also focused our follow ups on the first two months after hospital discharge and did not examine longer-term nonadherence, thus we are unable to provide information on adherence trajectories. We also note that we did not assess any behavior, such as diet or exercise, prior to HCT, so we cannot know whether patients changed from their pre-HCT behavior. Further, we note that this is a single-site study and many not represent or be applicable to the general HCT population. Strengths of this study included the prospective design, repeated measurements, and assessment of rates of nonadherence to the entire post-HCT medical regimen range of tasks

In conclusion, our data show that rates of nonadherence to the post-HCT medical regimen appear to be relatively high, even to the critical immunosuppressant medication. Nonadherence to important health behaviors such as diet, exercise, and mouthcare is surprisingly high and generally worsens over the short-term. Both patient- and caregiver-related factors are related to nonadherence, highlighting the importance of considering the caregiver as well as the patient in patient care. Targeting task efficacy in efforts to help patients and caregivers manage the medical regimen may be one feasible way to improve adherence. Psychological interventions in cancer patients have been shown to improve self-efficacy⁴⁴ and a recent psychosocial intervention with HCT caregivers improved their self-efficacy in managing the impact of HCT.⁴⁵ Future randomized clinical intervention trials need to investigate the impact of psychosocial interventions on both potential mediating factors and nonadherence following HCT in this patient population.

Highlights.

• Task related self-efficacy is a key predictor of task adherence

• Caregivers contribute to patient self-management

• Patients and caregivers are not fully adherent to the post-HCT medical regimen