Clinical Trial Participation Among Older Adult Medicare Fee-for-Service Beneficiaries With Cancer.

Importance: Clinical trials play a critical role in the development of novel cancer therapies, and precise estimates of the frequency with which older adult patients with cancer participate in clinical trials are lacking. Objective: To estimate the proportion of older adult Medicare Fee-for-Service (FFS) beneficiaries with cancer who participate in interventional cancer clinical trials, using a novel population-based methodology. Design, Setting, and Participants: In this retrospective cohort study evaluating clinical trial participation among older adult patients with cancer from January 1, 2014, through June 30, 2020, claims data from Medicare FFS were linked with the ClinicalTrials.gov to determine trial participation through the unique National Clinical Trial (NCT) identifier. The proportion of patients with newly diagnosed or newly recurrent cancer in 2015 participating in an interventional clinical trial and receiving active cancer treatment from January 2014 to June 2020 was estimated. Data analysis was performed from November 18, 2020, to November 1, 2021. Exposures: Patients with cancer aged 65 years or older with Medicare FFS insurance, with and without active cancer treatment. Main Outcomes and Measures: Enrollment in clinical trials among all patients with cancer 65 years and older and among patients receiving active cancer treatments as defined by the presence of at least 1 NCT identifier corresponding to an interventional cancer clinical trial in Medicare claims. Results: Among 1 150 978 patients (mean [SD] age, 75.7 [8.4] years; 49.9% men and 50.1% women) with newly diagnosed or newly recurrent cancer in 2015, 12 028 (1.0%) patients had a billing claim with an NCT identifier indicating enrollment in an interventional cancer clinical trial between January 2014 and June 2020. In a subset of 429 343 patients with active cancer treatment, 8360 (1.9%) were enrolled in 1 or more interventional trials. Patients enrolled in a trial tended to be younger, male, a race other than Black, and residing in zip codes with high median incomes. Conclusions and Relevance: Findings of this cohort study show that clinical trial enrollment among older adult patients with cancer remains low, with only 1.0% to 1.9% of patients with newly diagnosed or recurrent cancer in 2015 participating in an interventional cancer clinical trial as measured by the presence of NCT identifiers in Medicare claims. These data provide a contemporary estimate of trial enrollment, persistent disparities in trial participation, and only limited progress in trial access over the past 2 decades.

Validation of a Population-Based Data Source to Examine National Cancer Clinical Trial Participation.

Importance: The Centers for Medicare & Medicaid Services requires health care organizations to report the National Clinical Trial (NCT) identifier on claims for items and services related to clinical trials that qualify for coverage. This same NCT identifier is used to identify clinical trials in the ClinicalTrials.gov registry. If linked, this information could facilitate population-based analyses of clinical trial participation and outcomes. Objective: To evaluate the validity of a linkage between fee-for-service (FFS) Medicare claims and ClinicalTrials.gov through the NCT identifier for patients with cancer enrolled in clinical trials. Design, Setting, and Participants: This cohort study included 2 complementary retrospective analyses for a validation assessment. First, billing data from 3 health care institutions were used to estimate the missingness of the NCT identifier in claims by calculating the proportion of known participants in cancer clinical trials with no NCT identifier on any submitted Medicare claims. Second, the Surveillance Epidemiology and End Results-Medicare data set, which includes a subset of all FFS Medicare beneficiaries for whom health insurance claims are linked with cancer registry data, was used to identify adult patients diagnosed with cancer between 2006 and 2015 with an NCT identifier in claims corresponding to an interventional cancer clinical trial. To estimate the accuracy of the NCT identifier when present, the proportion of NCT identifiers that corresponded to trials that were aligned with the patients' known primary or secondary diagnoses was calculated. Data were analyzed from March 2020 to March 2021. Exposures: An NCT identifier present in Medicare claims. Main Outcomes and Measures: The main outcome was participating in a clinical trial relevant to patient's cancer diagnosis. Results: A total of 1 171 816 patients were included in analyses. Across the 3 participating institutions, there were 5061 Medicare patients enrolled in a clinical trial, including 3797 patients (75.0%) with an NCT identifier on at least 1 billing claim that matched the clinical trial on which the patient was participating. Among 1 171 816 SEER-Medicare patients, 29 138 patients (2.5%) had at least 1 claim with a value entered in the NCT identifier field corresponding to 32 950 unique patient-NCT identifier pairs. There were 26 694 pairs (81.0%) with an NCT identifier corresponding to a clinical trial registered in ClinicalTrials.gov, of which 10 170 pairs (38.1%) were interventional cancer clinical trials. Among these, 9805 pairs (96.4%) were considered appropriate. Conclusions and Relevance: In this cohort study, this data linkage provided a novel data source to study clinical trial enrollment patterns among Medicare patients with cancer on a population level. The presence of the NCT identifiers in claims for Medicare patients participating in clinical trials is likely to improve over time with increasing adherence with the Centers for Medicare & Medicaid Services mandate.

An assessment of estimation methods for generalized linear mixed models with binary outcomes.

Two main classes of methodology have been developed for addressing the analytical intractability of generalized linear mixed models: likelihood-based methods and Bayesian methods. Likelihood-based methods such as the penalized quasi-likelihood approach have been shown to produce biased estimates especially for binary clustered data with small clusters sizes. More recent methods using adaptive Gaussian quadrature perform well but can be overwhelmed by problems with large numbers of random effects, and efficient algorithms to better handle these situations have not yet been integrated in standard statistical packages. Bayesian methods, although they have good frequentist properties when the model is correct, are known to be computationally intensive and also require specialized code, limiting their use in practice. In this article, we introduce a modification of the hybrid approach of Capanu and Begg, 2011, Biometrics 67, 371-380, as a bridge between the likelihood-based and Bayesian approaches by employing Bayesian estimation for the variance components followed by Laplacian estimation for the regression coefficients. We investigate its performance as well as that of several likelihood-based methods in the setting of generalized linear mixed models with binary outcomes. We apply the methods to three datasets and conduct simulations to illustrate their properties. Simulation results indicate that for moderate to large numbers of observations per random effect, adaptive Gaussian quadrature and the Laplacian approximation are very accurate, with adaptive Gaussian quadrature preferable as the number of observations per random effect increases. The hybrid approach is overall similar to the Laplace method, and it can be superior for data with very sparse random effects.

Assessment of rare BRCA1 and BRCA2 variants of unknown significance using hierarchical modeling.

Current evidence suggests that the genetic risk of breast cancer may be caused primarily by rare variants. However, while classification of protein-truncating mutations as deleterious is relatively straightforward, distinguishing as deleterious or neutral the large number of rare missense variants is a difficult on-going task. In this article, we present one approach to this problem, hierarchical statistical modeling of data observed in a case-control study of contralateral breast cancer (CBC) in which all the participants were genotyped for variants in BRCA1 and BRCA2. Hierarchical modeling permits leverage of information from observed correlations of characteristics of groups of variants with case-control status to infer with greater precision the risks of individual rare variants. A total of 181 distinct rare missense variants were identified among the 705 cases with CBC and the 1,398 controls with unilateral breast cancer. The model identified three bioinformatic hierarchical covariates, align-GV, align-GD, and SIFT scores, each of which was modestly associated with risk. Collectively, the 11 variants that were classified as adverse on the basis of all the three bioinformatic predictors demonstrated a stronger risk signal. This group included five of six missense variants that were classified as deleterious at the outset by conventional criteria. The remaining six variants can be considered as plausibly deleterious, and deserving of further investigation (BRCA1 R866C; BRCA2 G1529R, D2665G, W2626C, E2663V, and R3052W). Hierarchical modeling is a strategy that has promise for interpreting the evidence from future association studies that involve sequencing of known or suspected cancer genes.

Physician visits prior to treatment for clinically localized prostate cancer.

BACKGROUND: The 2 primary therapeutic interventions for localized prostate cancer are delivered by different types of physicians, urologists, and radiation oncologists. We evaluated how visits to specialists and primary care physicians (PCPs) by men with localized prostate cancer are related to treatment choice. METHODS: Using the Surveillance, Epidemiology, and End Results (SEER)-Medicare linked database, we identified 85 088 men with clinically localized prostate cancer diagnosed at age 65 years or older, between 1994 and 2002. Men were categorized by primary treatment received within 9 months of diagnosis: radical prostatectomy (n = 18 201 [21%]), radiotherapy (n = 35 925 [42%]), androgen deprivation (n = 14 021 [17%]), or expectant management (n = 16 941 [20%]). Visits to specialists and PCPs were analyzed by patient characteristics and primary therapies received and were identified using Medicare claims and the American Medical Association Physician Masterfile. RESULTS: Overall, 42 309 men (50%) were seen exclusively by urologists, 37 540 (44%) by urologists and radiation oncologists, 2329 (3%) by urologists and medical oncologists, and 2910 (3%) by all 3 specialists. There was a strong association between the type of specialist seen and primary therapy received. Visits to PCPs were infrequent between diagnosis and receipt of therapy (22% of patients visited any PCP and 17% visited an established PCP) and were not associated with a greater likelihood of specialist visits. Irrespective of age, comorbidity status, or specialist visits, men seen by PCPs were more likely to be treated expectantly. CONCLUSIONS: Specialist visits relate strongly to prostate cancer treatment choices. In light of these findings, prior evidence that specialists prefer the modality they themselves deliver and the lack of conclusive comparative studies demonstrating superiority of one modality over another, it is essential to ensure that men have access to balanced information before choosing a particular therapy for prostate cancer.

Disenrollment from Medicare managed care among beneficiaries with and without a cancer diagnosis.

BACKGROUND: Medicare managed care may offer enrollees lower out-of-pocket costs and provide benefits that are not available in the traditional fee-for-service Medicare program. However, managed care plans may also restrict provider choice in an effort to control costs. We compared rates of voluntary disenrollment from Medicare managed care to traditional fee-for-service Medicare among Medicare managed care enrollees with and without a cancer diagnosis. METHODS: We identified Medicare managed care enrollees aged 65 years or older who were diagnosed with a first primary breast (n = 28 331), colorectal (n = 26 494), prostate (n = 29 046), or lung (n = 31 243) cancer from January 1, 1995, through December 31, 2002, in Surveillance, Epidemiology, and End Results (SEER) cancer registry records linked with Medicare enrollment files. Cancer patients were pair-matched to cancer-free enrollees by age, sex, race, and geographic location. We estimated rates of voluntary disenrollment to fee-for-service Medicare in the 2 years after each cancer patient's diagnosis, adjusted for plan characteristics and Medicare managed care penetration, by use of Cox proportional hazards regression. RESULTS: In the 2 years after diagnosis, cancer patients were less likely to disenroll from Medicare managed care than their matched cancer-free peers (for breast cancer, adjusted hazard ratio [HR] for disenrollment = 0.78, 95% confidence interval [CI] = 0.74 to 0.82; for colorectal cancer, HR = 0.84, 95% CI = 0.80 to 0.88; for prostate cancer, HR = 0.86, 95% CI = 0.82 to 0.90; and for lung cancer, HR = 0.81, 95% CI = 0.76 to 0.86). Results were consistent across strata of age, sex, race, SEER registry, and cancer stage. CONCLUSION: A new cancer diagnosis between 1995 and 2002 did not precipitate voluntary disenrollment from Medicare managed care to traditional fee-for-service Medicare.

Estimating the empirical Lorenz curve and Gini coefficient in the presence of error with nested data.

The Lorenz curve is a graphical tool that is widely used to characterize the concentration of a measure in a population, such as wealth. It is frequently the case that the measure of interest used to rank experimental units when estimating the empirical Lorenz curve, and the corresponding Gini coefficient, is subject to random error. This error can result in an incorrect ranking of experimental units which inevitably leads to a curve that exaggerates the degree of concentration (variation) in the population. We consider a specific data configuration with a hierarchical structure where multiple observations are aggregated within experimental units to form the outcome whose distribution is of interest. Within this context, we explore this bias and discuss several widely available statistical methods that have the potential to reduce or remove the bias in the empirical Lorenz curve. The properties of these methods are examined and compared in a simulation study. This work is motivated by a health outcomes application that seeks to assess the concentration of black patient visits among primary care physicians. The methods are illustrated on data from this study.

Cystectomy for muscle-invasive bladder cancer: patterns and outcomes of care in the Medicare population.

OBJECTIVES: To describe the population-based patterns of care among patients with muscle-invasive bladder cancer. METHODS: A retrospective cohort study using the Surveillance, Epidemiology and End Results (SEER)-Medicare database identified 4664 patients aged 65 years or older with muscle-invasive bladder cancer diagnosed between 1991 and 1999. The use of particular treatment modalities was evaluated according to the clinical and demographic characteristics available in the SEER-Medicare database. RESULTS: Considerable variation was found in the treatments delivered to the cohort members. Overall, 39% had undergone cystectomy; 30% of Stage II, 57% of Stage III, and 38% of Stage IV patients underwent this operation within 6 months of diagnosis. The frequency of resection declined with age, such that 55% of patients aged 65 to 69 years and 27% of those aged 80 to 84 years underwent cystectomy. For 36% of Stage II, 18% of Stage III, and 27% of Stage IV patients, no evidence was found of surgery, chemotherapy, or radiotherapy within 6 months of diagnosis. Other management strategies included chemotherapy alone (14% Stage II, 6% Stage III, and 12% Stage IV), radiotherapy alone (11% for each stage), or combined modality chemoradiotherapy (10% Stage II, 8% Stage III, and 12% Stage IV). Multivariate analyses suggested that cystectomy conferred a survival advantage. CONCLUSIONS: A marked heterogeneity exists in the strategies used to treat muscle-invasive bladder cancer. The extent to which this variation can be attributed to the lack of informative clinical trials, the presence of comorbid illness, patient or physician preferences, or access to care warrants further evaluation.

Adjuvant chemotherapy use for Medicare beneficiaries with stage II colon cancer.

PURPOSE: Clinical trials have not demonstrated that adjuvant chemotherapy improves survival for patients with resected stage II colon cancer. Nevertheless, patients may receive this treatment despite its uncertain benefit. The objective of this study was to determine the extent to which adjuvant chemotherapy is used for patients with stage II colon cancer. PATIENTS AND METHODS: Using the Surveillance, Epidemiology, and End Results-Medicare linked database, we identified 3,151 patients aged 65 to 75 with resected stage II colon cancer and no adverse prognostic features. The primary outcome was chemotherapy use within 3 months of surgery ascertained from claims submitted to Medicare. Relationships between patient characteristics and adjuvant chemotherapy use were measured and their significance was assessed using multivariable logistic regression. Survival for treated and untreated patients was compared using a Cox model. RESULTS: Twenty-seven percent of patients received chemotherapy during the 3 postoperative months. Younger age at diagnosis, white race, unfavorable tumor grade, and low comorbidity were each associated with a greater likelihood of receiving treatment. Sex, the number of examined lymph nodes in the tumor specimen, the urgency of the surgical admission, and median income was each unrelated to treatment. Five-year survival was 75% for untreated patients and 78% for treated patients. After adjusting for known between-group differences, the hazard ratio for survival associated with adjuvant treatment was 0.91 (95% confidence interval, 0.77 to 1.09). CONCLUSION: A substantial percentage of Medicare beneficiaries with resected stage II colon cancer receive adjuvant chemotherapy despite its uncertain benefit.

Measuring complications of cancer treatment using the SEER-Medicare data.

BACKGROUND: The linkage of SEER registry data with Medicare claims allows the longitudinal tracking of health care and outcomes for patients after a cancer diagnosis. One category of outcomes amenable to research using Medicare claims is complications of cancer treatments: the unintentional, adverse side effects or sequelae of interventions used to treat or palliate cancer patients. RESEARCH DESIGN: The authors review some of the methods and limitations of using Medicare claims to identify both acute and chronic complications of cancer treatments, and present an original analysis comparing survey-based and claims-based complications following radical prostatectomy for prostate cancer to illustrate some of the potential limitations inherent in using claims for this purpose. RESULTS: Utility of the Medicare claims for identifying postdischarge complications varies by the patient type, the initial treatment used, and any subsequent treatment of complications. For patients undergoing surgical interventions, Medicare claims can be used to identify most acute inpatient complications. However, claims data cannot be used as effectively in the long-term to capture chronic complications, particularly when the complication does not consistently prompt an intervention. CONCLUSION: Researchers who use the SEER-Medicare-linked database to assess long-term complications of cancer treatments should exercise caution when designing and interpreting studies. Ideally, for studies of most chronic complications of cancer care, validation studies similar to the one performed here would provide valuable additional evidence to assess the credibility of conclusions based on claims data.

Variations in morbidity after radical prostatectomy.

BACKGROUND: Recent studies of surgery for cancer have demonstrated variations in outcomes among hospitals and among surgeons. We sought to examine variations in morbidity after radical prostatectomy for prostate cancer. METHODS: We used the Surveillance, Epidemiology, and End Results-Medicare linked data base to evaluate health-related outcomes after radical prostatectomy. The rates of postoperative complications, late urinary complications (strictures or fistulas 31 to 365 days after the procedure), and long-term incontinence (more than 1 year after the procedure) were inferred from the Medicare claims records of 11,522 patients who underwent prostatectomy between 1992 and 1996. These rates were analyzed in relation to hospital volume and surgeon volume (the number of procedures performed at individual hospitals and by individual surgeons, respectively). RESULTS: Neither hospital volume nor surgeon volume was significantly associated with surgery-related death. Significant trends in the relation between volume and outcome were observed with respect to postoperative complications and late urinary complications. Postoperative morbidity was lower in very-high-volume hospitals than in low-volume hospitals (27 percent vs. 32 percent, P=0.03) and was also lower when the prostatectomy was performed by very-high-volume surgeons than when it was performed by low-volume surgeons (26 percent vs. 32 percent, P<0.001). The rates of late urinary complications followed a similar pattern. Results for long-term preservation of continence were less clear-cut. In a detailed analysis of the 159 surgeons who had a high or very high volume of procedures, wide surgeon-to-surgeon variations in these clinical outcomes were observed, and they were much greater than would be predicted on the basis of chance or observed variations in the case mix. CONCLUSIONS: In men undergoing prostatectomy, the rates of postoperative and late urinary complications are significantly reduced if the procedure is performed in a high-volume hospital and by a surgeon who performs a high number of such procedures.

Two-stage designs for gene-disease association studies.

The goal of this article is to describe a two-stage design that maximizes the power to detect gene-disease associations when the principal design constraint is the total cost, represented by the total number of gene evaluations rather than the total number of individuals. In the first stage, all genes of interest are evaluated on a subset of individuals. The most promising genes are then evaluated on additional subjects in the second stage. This will eliminate wastage of resources on genes unlikely to be associated with disease based on the results of the first stage. We consider the case where the genes are correlated and the case where the genes are independent. Using simulation results, it is shown that, as a general guideline when the genes are independent or when the correlation is small, utilizing 75% of the resources in stage 1 to screen all the markers and evaluating the most promising 10% of the markers with the remaining resources provides near-optimal power for a broad range of parametric configurations. This translates to screening all the markers on approximately one quarter of the required sample size in stage 1.