Medication Adherence in Patients with Hematologic Malignancies Undergoing Hematopoietic Stem Cell Transplantation

Background: Patients with hematologic malignancies recovering from allogeneic hematopoietic stem cell transplantation (HSCT), along with their caregivers, must manage complex medication regimens. Medication adherence, defined as the extent to which patients follow their prescribed medications, is crucial for a successful recovery post-HSCT. Suboptimal medication adherence can lead to several short-and long-term negative outcomes, such as increased infection rates, acute and chronic graft-versus-host disease, and increased healthcare utilization (e.g., hospitalizations, intensive care unit admissions). However, there is limited data on the best methods to measure medication adherence and the patient factors associated with it.

Methods: We conducted a prospective longitudinal study involving 150 patients with hematologic malignancies undergoing HSCT to describe medication adherence using a multi-measure approach. We employed well-established subjective measures (e.g., medication diaries and logs) and objective measures (e.g., pill counts) to assess adherence. Additionally, we used the Medication Adherence Response Scale-5 (MARS-5) and obtained drug levels of immunosuppressants (e.g., tacrolimus) from the electronic health record. Assessments were conducted on days 30, 60, and 180 post-HSCT. For pill counts, we calculated an adherence ratio: (number of doses used / total number of doses prescribed). From the medication log data, we calculated dosage adherence ratio (the number of doses reported as taken / the number of doses expected) and timing adherence (medications taken within three hours of the prescribed dosing time). All adherence ratios were computed as binary outcomes (adherent versus nonadherent) based on cutoff points [0.8-1.2 (adherent) or outside that range (nonadherent)]. We characterized drug levels as adherent versus nonadherent using an adherence drug level >80% in the therapeutic range as a cutoff point for adherence across all time points. We classified MARS-5 scores as adherent if the score was ≥23 and nonadherent if the score was < 23, consistent with other studies of patients with hematologic malignancies. We used Kappa analysis to explore the agreement between measures of medication adherence and linear regression models to assess the relationship between medication adherence and patient-, clinical-, and patient-reported outcomes such as cognitive functioning and self-efficacy.

Results: Participants were 57.5 (SD=13.5) years old; 40.7% (n=61) female, 85.3% (n=128) non-Hispanic White, 73.3% (n=110) married or living with someone, 62.6% (n=94) Christian, and 54.0% (n=81) some college or college educated. The majority of participants underwent reduced intensity conditioning (62.7%, n=94) and did not receive total body radiation (83.3%, n=125) or experience GVHD by Day 30 (93.3%, n=140). Across all three time points, the medication adherence rates were as follows: 52-64% (pill counts), 96-98% (dose adherence), 83% (time adherence), 18-24% (medication levels), and 97-98% (MARS-5). There was no agreement between the adherence ratios calculated using different adherence measures; pill count | drug level: kappa=0.04, pill count | medication dose: kappa=0.03, pill count | medication timing: kappa=0.10, pill count | MARS-5: kappa=0.04, medication dose | medication timing: kappa=0.12, medication dose | MARS-5: kappa=0.03, medication dose | drug level: kappa=0.006, medication timing | MARS-5: kappa=0.04, medication timing | drug level : kappa=0.008, MARS-5 | drug level: kappa=0.004. Patient and clinical factors, as well as patient-reported outcomes such as cognitive functioning and self-efficacy, were not significantly associated with medication adherence across all adherence measures.

Conclusions: Medication adherence in patients with hematologic malignancies undergoing HSCT can be evaluated using a multi-measure approach that includes both objective and subjective methods. However, these measures likely capture different aspects of medication adherence and suggest differences in how patients take their medications, resulting in little to no agreement across measures. Larger studies involving diverse sociodemographic and ethnic groups could offer deeper insights into the patient factors associated with medication adherence in this population.