Abstract
Introduction: First- (imatinib) and second-generation (dasatinib, nilotinib) tyrosine kinase inhibitors are the standard of care in the management of chronic myeloid leukemia. Despite their high efficacy and the convenience of oral administration, studies have reported variation in patient medication behavior with non-adherence rates varying from low to moderate based on definition and measurement method. We conducted study-level meta-analyses stratified by measurement method to quantify adherence prevalence rates in chronic myeloid leukemia patients as reported in non-controlled "real-life" studies.
Methods: We searched PubMed, Embase, and Cochrane Library for non-controlled studies reporting adherence or non-adherence rates to tyrosine kinase inhibitor treatment in chronic myeloid leukemia patients across various methods of measurement. For retained studies, adherence rates and 95% confidence interval (95% CI) were extracted or calculated; and grouped by method of measurement. Random-effects meta-analyses were performed to account for estimated (Q, I2, tau2) within and between study heterogeneity, and associated forest plots were generated. Analyses were done using Comprehensive Meta-Analysis V.3.
Results: From 649 publications yielded by the search, 40 articles and abstracts were retained. Measurement methods included structured interview, medical/pharmacy chart review, medication possession ratio, proportion of days covered, electronic monitoring, and self-report. Electronic monitoring and self-report were used in one study each and thus excluded from meta-analysis. Table 1 summarizes, by the remaining four methods, the number of studies and patients included in each meta-analysis, the estimated adherence event rates with 95%CI, and heterogeneity indices. In random-effects analyses, adherence rate estimates as measured by each method ranged (in descending order) from 0.75 (95%CI=0.66-0.82) for structured interview, 0.68 (95%CI=0.54-0.79) for medical/pharmacy chart review, 0.57 (95%CI=0.47-0.67) for medication possession ratio, to 0.56 (95%CI=0.36-0.74) for proportion of days covered. All four analyses showed significant heterogeneity.
Conclusion: Our meta-analyses using clinical data (structured interview; medical/pharmacy chart review) indicate that, while the majority of chronic myeloid leukemia patients are adherent to their tyrosine kinase inhibitor regimens, between 1/3rd and 1/4th of them are not. Indirect methods using prescription claims data (medication possession ratio; proportion of days covered) yielded lower adherence rates, though caution about such indirect results is warranted. Considering evidence linking adherence to impaired cytogenetic (Noens et al, Blood 2009) and molecular response (Marin et al, J Clin Oncol 2010), clinicians should integrate adherence assessment and enhancement into routine clinical practice.
MacDonald:Matrix45: Employment, Equity Ownership; Ex Ante International: Equity Ownership. Abraham:Matrix45: Equity Ownership; Belgamis: Equity Ownership; Ex Ante International: Equity Ownership.
