Methodology for Using Real-World Data From Electronic Health Records to Assess Chemotherapy Administration in Women With Breast Cancer.

PURPOSE: Identification of patients' intended chemotherapy regimens is critical to most research questions conducted in the real-world setting of cancer care. Yet, these data are not routinely available in electronic health records (EHRs) at the specificity required to address these questions. We developed a methodology to identify patients' intended regimens from EHR data in the Optimal Breast Cancer Chemotherapy Dosing (OBCD) study. METHODS: In women older than 18 years, diagnosed with primary stage I-IIIA breast cancer at Kaiser Permanente Northern California (2006-2019), we categorized participants into 24 drug combinations described in National Comprehensive Cancer Network guidelines for breast cancer treatment. Participants were categorized into 50 guideline chemotherapy administration schedules within these combinations using an iterative algorithm process, followed by chart abstraction where necessary. We also identified patients intended to receive nonguideline administration schedules within guideline drug combinations and nonguideline drug combinations. This process was adapted at Kaiser Permanente Washington using abstracted data (2004-2015). RESULTS: In the OBCD cohort, 13,231 women received adjuvant or neoadjuvant chemotherapy, of whom 10,213 (77%) had their intended regimen identified via the algorithm, 2,416 (18%) had their intended regimen identified via abstraction, and 602 (4.5%) could not be identified. Across guideline drug combinations, 111 nonguideline dosing schedules were used, alongside 61 nonguideline drug combinations. A number of factors were associated with requiring abstraction for regimen determination, including: decreasing neighborhood household income, earlier diagnosis year, later stage, nodal status, and human epidermal growth factor receptor 2 (HER2)+ status. CONCLUSION: We describe the challenges and approaches to operationalize complex, real-world data to identify intended chemotherapy regimens in large, observational studies. This methodology can improve efficiency of use of large-scale clinical data in real-world populations, helping answer critical questions to improve care delivery and patient outcomes.

Assessment of breast cancer chemotherapy dose reduction in an integrated healthcare delivery system.

PURPOSE: Most cytotoxic drugs are dosed using body surface area (BSA), yet not all cancer patients receive the full BSA-determined dose. Prior work suggests that breast cancer patients who are obese are more likely to experience dose reduction than normal weight patients. However, the factors driving dose reduction remain unclear. METHODS: In 452 women diagnosed with stage I-IIIA primary breast cancer at Kaiser Permanente Northern California, we evaluated the association between obesity and dose reduction, and further explored other factors in relation to dose reduction, including various sociodemographic characteristics, tumor characteristics, and comorbidities. Study participants were a part of the Pathways Study, diagnosed between 2006 and 2013 and treated with cyclophosphamide + doxorubicin, followed by paclitaxel (ACT). Dose reduction was assessed using first cycle dose proportion (FCDP) and average relative dose intensity (ARDI), a metric of dose intensity over the course of chemotherapy. RESULTS: Overall, 8% of participants received a FCDP < 90% and 21.2% had an ARDI < 90%, with dose reduction increasing with body mass index. In adjusted logistic regression models, obese women had 4.1-fold higher odds of receiving an ARDI < 90% than normal weight women (95% CI: 1.9-8.9; p-trend = 0.0006). Increasing age was positively associated with an ADRI < 90%, as was the presence of comorbidity. Dose reduction was less common in later calendar years. CONCLUSION: Results offer insight on factors associated with chemotherapy dosing for a common breast cancer regimen. Larger studies are required to evaluate relevance to other regimens, and further work will be needed to determine whether dose reductions impact outcomes in obese women.

Evaluation of Algorithms Using Automated Health Plan Data to Identify Breast Cancer Recurrences.

BACKGROUND: We updated algorithms to identify breast cancer recurrences from administrative data, extending previously developed methods. METHODS: In this validation study, we evaluated pairs of breast cancer recurrence algorithms (vs. individual algorithms) to identify recurrences. We generated algorithm combinations that categorized discordant algorithm results as no recurrence [High Specificity and PPV (positive predictive value) Combination] or recurrence (High Sensitivity Combination). We compared individual and combined algorithm results to manually abstracted recurrence outcomes from a sample of 600 people with incident stage I-IIIA breast cancer diagnosed between 2004 and 2015. We used Cox regression to evaluate risk factors associated with age- and stage-adjusted recurrence rates using different recurrence definitions, weighted by inverse sampling probabilities. RESULTS: Among 600 people, we identified 117 recurrences using the High Specificity and PPV Combination, 505 using the High Sensitivity Combination, and 118 using manual abstraction. The High Specificity and PPV Combination had good specificity [98%, 95% confidence interval (CI): 97-99] and PPV (72%, 95% CI: 63-80) but modest sensitivity (64%, 95% CI: 44-80). The High Sensitivity Combination had good sensitivity (80%, 95% CI: 49-94) and specificity (83%, 95% CI: 80-86) but low PPV (29%, 95% CI: 25-34). Recurrence rates using combined algorithms were similar in magnitude for most risk factors. CONCLUSIONS: By combining algorithms, we identified breast cancer recurrences with greater PPV than individual algorithms, without additional review of discordant records. IMPACT: Researchers should consider tradeoffs between accuracy and manual chart abstraction resources when using previously developed algorithms. We provided guidance for future studies that use breast cancer recurrence algorithms with or without supplemental manual chart abstraction.

Missing data strategies for the Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) in Alliance A091105 and COMET-2.

PURPOSE: Missing scores complicate analysis of the Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) because patients with and without missing scores may systematically differ. We focus on optimal analysis methods for incomplete PRO-CTCAE items, with application to two randomized, double-blind, placebo-controlled, phase III trials. METHODS: In Alliance A091105 and COMET-2, patients completed PRO-CTCAE items before randomization and several times post-randomization (N = 64 and 107, respectively). For each trial, we conducted between-arm comparisons on the PRO-CTCAE via complete-case two-sample t-tests, mixed modeling with contrast, and multiple imputation followed by two-sample t-tests. Because interest lies in whether CTCAE grades can inform missing PRO-CTCAE scores, we performed multiple imputation with and without CTCAE grades as auxiliary variables to assess the added benefit of including them in the imputation model relative to only including PRO-CTCAE scores across all cycles. RESULTS: PRO-CTCAE completion rates ranged from 100.0 to 71.4% and 100.0 to 77.1% across time in A091105 and COMET-2, respectively. In both trials, mixed modeling and multiple imputation provided the most similar estimates of the average treatment effects. Including CTCAE grades in the imputation model did not consistently narrow confidence intervals of the average treatment effects because correlations for the same PRO-CTCAE item between different cycles were generally stronger than correlations between each PRO-CTCAE item and its corresponding CTCAE grade at the same cycle. CONCLUSION: For between-arm comparisons, mixed modeling and multiple imputation are informative techniques for handling missing PRO-CTCAE scores. CTCAE grades do not provide added benefit for informing missing PRO-CTCAE scores. CLINICALTRIALS: gov Identifiers: NCT02066181 (Alliance A091105); NCT01522443 (COMET-2).

Evaluating Treatment Tolerability Using the Toxicity Index With Patient-Reported Outcomes Data.

CONTEXT: Summarizing longitudinal symptomatic adverse events during clinical trials is necessary for understanding treatment tolerability. The Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) provides insight for capturing treatment tolerability within trials. Tolerability summary measures, such as the maximum score, are often used to communicate the potential negative symptoms both in the medical literature and directly to patients. Commonly, the proportions of present and severe symptomatic adverse events are used and reported between treatment arms among adverse event types. The toxicity index is also a summary measure previously applied to clinician-reported CTCAE data. OBJECTIVES: Apply the toxicity index to PRO-CTCAE data from the COMET-2 trial alongside the maximum score, then present and discuss considerations for using the toxicity index as a summary measure for communicating tolerability to patients and clinicians. METHODS: Proportions of maximum PRO-CTCAE severity levels and median toxicity index were computed by arm using all trial data and adjusting for baseline symptoms. RESULTS: Group-wise statistical differences were similar whether using severity level proportions or the toxicity index. The impact of adjusting for baseline symptoms was equivalently seen when comparing arms using severity rates or the toxicity index. CONCLUSION: The toxicity index is a useful method when ranking patients from those with the least to most symptomatic adverse event burden. This study showed the toxicity index can be applied to PRO-CTCAE data. Though as a tolerability summary measure, further study is needed to provide a clear clinical or patient-facing interpretation of the toxicity index.

Feasibility of Implementing the Patient-Reported Outcomes Version of the Common Terminology Criteria for Adverse Events in a Multicenter Trial: NCCTG N1048.

Purpose The US National Cancer Institute (NCI) Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) was developed to enable patient reporting of symptomatic adverse events in oncology clinical research. This study was designed to assess the feasibility and resource requirements associated with implementing PRO-CTCAE in a multicenter trial. Methods Patients with locally advanced rectal cancer enrolled in the National Cancer Institute-sponsored North Central Cancer Treatment Group (Alliance) Preoperative Radiation or Selective Preoperative Radiation and Evaluation before Chemotherapy and Total Mesorectal Excision trial were asked to self-report 30 PRO-CTCAE items weekly from home during preoperative therapy, and every 6 months after surgery, via either the Web or an automated telephone system. If participants did not self-report within 3 days, a central coordinator called them to complete the items. Compliance was defined as the proportion of participants who completed PRO-CTCAE assessments at expected time points. Results The prespecified PRO-CTCAE analysis was conducted after the 500th patient completed the 6-month follow-up (median age, 56 years; 33% female; 12% nonwhite; 43% high school education or less; 5% Spanish speaking), across 165 sites. PRO-CTCAE was reported by participants at 4,491 of 4,882 expected preoperative time points (92.0% compliance), of which 3,771 (77.2%) were self-reported by participants and 720 (14.7%) were collected via central coordinator backup. Compliance at 6-month post-treatment follow-up was 333 of 468 (71.2%), with 122 (26.1%) via backup. Site research associates spent a median of 15 minutes on PRO-CTCAE work for each patient visit. Work by a central coordinator required a 50% time commitment. Conclusion Home-based reporting of PRO-CTCAE in a multicenter trial is feasible, with high patient compliance and low site administrative requirements. PRO-CTCAE data capture is improved through centralized backup calls.

Exploring differences in adverse symptom event grading thresholds between clinicians and patients in the clinical trial setting.

PURPOSE: Symptomatic adverse event (AE) monitoring is essential in cancer clinical trials to assess patient safety, as well as inform decisions related to treatment and continued trial participation. As prior research has demonstrated that conventional concordance metrics (e.g., intraclass correlation) may not capture nuanced aspects of the association between clinician and patient-graded AEs, we aimed to characterize differences in AE grading thresholds between doctors (MDs), registered nurses (RNs), and patients using the Bayesian Graded Item Response Model (GRM). METHODS: From the medical charts of 393 patients aged 26-91 (M = 62.39; 43% male) receiving chemotherapy, we retrospectively extracted MD, RN and patient AE ratings. Patients reported using previously developed Common Terminology Criteria for Adverse Events (CTCAE) patient-language adaptations called STAR (Symptom Tracking and Reporting). A GRM was fitted to calculate the latent grading thresholds between MDs, RNs and patients. RESULTS: Clinicians have overall higher average grading thresholds than patients when assessing diarrhea, dyspnea, nausea and vomiting. However, RNs have lower grading thresholds than patients and MDs when assessing constipation. The GRM shows higher variability in patients' AE grading thresholds than those obtained from clinicians. CONCLUSIONS: The present study provides evidence to support the notion that patients report some AEs that clinicians might not consider noteworthy until they are more severe. The availability of GRM methodology could serve to enhance clinical understanding of the patient symptomatic experience and facilitate discussion where AE grading discrepancies exist. Future work should focus on capturing explicit AE grading decision criteria from MDs, RNs, and patients.