Oncology · Medical Literature
Medication adherence in breast cancer: a comparative analysis of self-report tools and prescription refill data.
Abstract
Adherence to oral anticancer medications (OAMs) is critical for clinical outcomes in breast cancer (BC), yet assessment methods lack standardization. The goal is to evaluate OAMs adherence in BC patients utilizing the most common adherence assessment methods on the same population, as well as to test the feasibility and concordance of these methods when used separately or in combination. 500 BC patients on OAMs at the National Cancer Institute, Cairo, were randomized into a two-period crossover study evaluating self-report tools (Arm A: Morisky Medication Adherence Scale (MMAS-8) and the Medication Adherence Report Scale (MARS-5); Arm B: the Adherence to Refills and Medications Scale (ARMS) and the Simplified Medication Adherence Questionnaire (SMAQ)) with a three-month interval to minimize respondent burden and order effects. Prescription refill data were calculated via Medication Possession Ratio (MPR) and Proportion of Days Covered (PDC). Agreement, diagnostic performance, and receiver operating characteristic analyses were performed. Refill-based measures estimated higher adherence than self-reported tools. A multi-method approach significantly increased the detection of nonadherent patients (49.4%-69.0%). PDC-based combinations detected more nonadherent cases than MPR. Among self-reported measures, ARMS demonstrated the strongest concordance with refill-based metrics (κ = 0.51-0.55; AUC≈0.80), followed by MARS-5 and MMAS-8, whereas SMAQ showed weaker performance. Methodological variability significantly impacts reported adherence rates. While no single "gold standard" exists, integrating brief self-reported scales with objective metrics is feasible and provides a more comprehensive and clinically informative assessment of medication use. This multi-method approach is essential for identifying at-risk patients and tailoring adherence-support interventions in oncology.
Clinical relevance
Among breast cancer patients receiving oral anticancer medications (OAMs), how do four validated self-report adherence tools (MMAS-8, MARS-5, ARMS, SMAQ) compare to each other and to prescription refill-based measures (MPR and PDC) in estimating adherence rates, concordance, and detection of non-adherence? __SNAPSHOT__ {"studyDesign":"Randomized two-period crossover study (methodological comparative study of adherence measurement instruments)","setting":"National Cancer Institute (Breast Cancer Comprehensive Center), Cairo University, Cairo, Egypt — single tertiary oncology center in a low- and middle-income country (LMIC)","sampleSize":"n=500 breast cancer patients receiving OAMs (Arm A: MMAS-8 and MARS-5; Arm B: ARMS and SMAQ; crossover at 3 months); refill-based metrics (MPR and PDC) applied to the full cohort","population":"Adult female breast cancer patients (metastatic and non-metastatic) receiving oral anticancer medications, including endocrine therapies and CDK4/6 inhibitor-based regimens, at a tertiary cancer center in Egypt. Specific inclusion/exclusion criteria not reported in the abstract.","intervention":"Four self-report adherence scales administered in a crossover design (MMAS-8, MARS-5, ARMS, SMAQ), with a 3-month washout interval between periods to minimize respondent burden and order effects","comparator":"Prescription refill-based adherence metrics: Medication Possession Ratio (MPR) and Proportion of Days Covered (PDC), applied to the same patient population","primaryOutcome":"Concordance, diagnostic performance (including area under the receiver operating characteristic curve [AUC]), and agreement (Cohen's κ) between self-report tools and refill-based measures, individually and in combination; detection rate of non-adherent patients","followUp":"Minimum 3 months (duration of crossover interval); total observation window not explicitly reported in the abstract","fundingSource":"Not reported"}
Key findings
Refill-based measures (MPR and PDC) consistently estimated higher adherence rates than all four self-report tools, a finding consistent with known social desirability bias inherent to patient-reported instruments. A multi-method approach combining self-report with refill data significantly increased the detection of non-adherent patients, with the non-adherence detection rate rising from 49.4% (single method) to 69.0% (multi-method combination) — an absolute increase of 19.6 percentage points (calculated). PDC-based combinations outperformed MPR-based combinations in identifying non-adherent cases. Among the four self-report instruments, ARMS demonstrated the strongest concordance with refill-based metrics (κ = 0.51–0.55; AUC ≈ 0.80), followed by MARS-5 and MMAS-8, while SMAQ showed the weakest performance. The κ values for ARMS represent moderate agreement by standard thresholds (κ 0.41–0.60 = moderate). Specific absolute adherence rates for individual tools other than refill-based metrics are not fully reported in the available abstract. No clinical outcome data (e.g., disease-free survival, recurrence) are reported. NNT and NNH are not applicable to this measurement concordance study.
Practice implications
This study establishes that choice of adherence measurement tool substantially affects the observed adherence rate in breast cancer patients on OAMs — methodological variability is not a trivial concern but a clinically consequential one. Refill data (particularly PDC) inflate apparent adherence relative to patient self-report, likely reflecting that dispensed medication is not synonymous with ingested medication. The ARMS scale demonstrated the best psychometric performance among self-report instruments in this LMIC oncology context, making it a potentially preferable tool when resource constraints preclude multi-method assessment. The principal actionable finding is that a combined approach — pairing at least one brief self-report scale (preferably ARMS) with PDC-based pharmacy data — identifies substantially more non-adherent patients than either method alone, enabling earlier intervention. Importantly, this study does not establish whether improved detection of non-adherence translates into improved clinical outcomes, nor does it define an optimal intervention strategy once non-adherence is identified. Findings are drawn exclusively from a single Egyptian tertiary cancer center, limiting direct generalizability to other healthcare systems, particularly those with different pharmacy dispensing infrastructure, digital health integration, or patient populations (e.g., HER2-positive or triple-negative breast cancer in high-income settings). The crossover design controls for inter-individual variation in questionnaire responses, which is a methodological strength, but the 3-month washout may be insufficient to fully eliminate carry-over effects in self-report behavior. Because only the abstract is available, granular subgroup data, regression analyses, and operational definitions of adherence thresholds for each instrument cannot be fully assessed. __LIMITATIONS__ Single-center study conducted at a tertiary cancer center in Egypt; generalizability to other healthcare systems, pharmacy dispensing models, and non-LMIC settings is uncertain.,Abstract-only access: complete methodological details (eligibility criteria, dropout rates, statistical handling of missing data, definitions of adherence thresholds per tool) are not available for full critical appraisal.,No gold-standard biological measure of adherence (e.g., drug-level assays, electronic monitoring via MEMS caps) was included; refill data serve as a proxy and cannot verify actual ingestion.,Prescription refill data are inherently subject to institutionally specific dispensing practices (e.g., partial fills, stockpiling), which may not apply uniformly across settings; PDC and MPR may overestimate true adherence.,Social desirability bias is an inherent and unquantified limitation of all four self-report instruments, particularly in a face-to-face clinical setting.,The 3-month crossover interval may be insufficient to eliminate carry-over or learning effects in self-reported behavior.,No clinical outcome data (survival, recurrence, toxicity) were reported; the clinical significance of improved non-adherence detection remains untested.,Funding source not reported; potential conflicts of interest cannot be assessed.,OAM regimen heterogeneity (tamoxifen, aromatase inhibitors, CDK4/6 inhibitor combinations) may introduce variation in adherence behavior that was not fully stratified in the abstract-reported results.,The study population appears predominantly female; applicability to male breast cancer patients is not addressed. __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ In breast cancer patients receiving OAMs, the choice of adherence measurement tool materially alters estimated adherence rates — refill-based measures (particularly PDC) detect greater apparent adherence than self-report tools, while a combined multi-method approach increases identification of non-adherent patients by approximately 20 percentage points (49.4% to 69.0%). Among self-report instruments, ARMS demonstrated the strongest concordance with refill data (κ = 0.51–0.55; AUC ≈ 0.80) and should be considered preferentially when a brief validated tool is needed. Clinicians and pharmacy teams should recognize that relying on any single adherence measurement method will underestimate the true burden of non-adherence; integration of brief patient-reported tools with dispensing records is both feasible and diagnostically superior, though whether this translates to improved patient outcomes requires prospective investigation.