Sarah M Belcher, Paul Scott, Susan M Sereika, Catherine Bender, Jacqueline Dunbar-Jacob, Margaret Q Rosenzweig, Benyam Muluneh, Lindsay M Sabik, Valire Copeland, Mounzer Agha, Abigail A Lustyik, Emily S He, Corrine Bozich, Grace Booze, Petra Duran Basso, J Devin Peipert
Reliability, two-factor dimensionality, and construct validity were supported, with some evidence of concurrent criterion validity with electronic event monitored adherence data. Additional validation testing is needed to support findings. Evidence supports the feasibility of longitudinal oral anticancer medication adherence assessment monitoring using PMAS.
BACKGROUND: Oral anticancer medications are standard care for cancer. Medication adherence influences health outcomes, but valid, reliable measures assessing self-reported medication adherence are limited. Psychometric properties of the PROMIS® Medication Adherence Scale (PMAS) were evaluated in patients prescribed oral anticancer medications for multiple myeloma.
METHODOLOGY: This was a secondary analysis from a longitudinal observational study examining medication adherence, symptoms, quality of life, and financial hardship among individuals prescribed oral anticancer medications for multiple myeloma. PMAS measured self-reported medication adherence. Self-report and medical record data assessed participant characteristics and adherence correlates. Objective medication adherence indices were generated from continuous electronic event monitored data. Internal consistency reliability was estimated using Cronbach's alpha. Dimensionality was assessed with confirmatory factor analysis. Construct validity was assessed considering adherence correlates. Spearman rank-order correlations summarized associations between PMAS and electronic event monitored data.
RESULTS: PMAS items had limited variability, with high adherence over time. Reliability of PMAS scores was adequate (T1 α = 0.82, 95% CI: 0.75, 0.87; T2 α = 0.84, 95% CI: 0.78, 0.89). Confirmatory factor analysis fit was better for sub-scales (Medication Beliefs and Knowledge and Medication Taking Behaviors) than Total scale, particularly for the Medication Beliefs and Knowledge subscale. Adherence correlates were observed as expected between PMAS and age, self-reported cognitive function, symptom severity, and depression. Weak/moderate positive associations were found between PMAS and electronic event monitored data.
CONCLUSIONS: Reliability, two-factor dimensionality, and construct validity were supported, with some evidence of concurrent criterion validity with electronic event monitored adherence data. Additional validation testing is needed to support findings. Evidence supports the feasibility of longitudinal oral anticancer medication adherence assessment monitoring using PMAS.