Antonia V Bennett

University of North Carolina at Chapel Hill, North Carolina, United States

Are you Antonia V Bennett?

Claim your profile

Publications (28)297.34 Total impact

  • Source
    The Oncologist 04/2015; 20(5). DOI:10.1634/theoncologist.2014-0431 · 4.54 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The standard approach for documenting symptomatic adverse events (AEs) in cancer clinical trials involves investigator reporting using the National Cancer Institute’s (NCI’s) Common Terminology Criteria for Adverse Events (CTCAE). Because this approach underdetects symptomatic AEs, the NCI issued two contracts to create a patient-reported outcome (PRO) measurement system as a companion to the CTCAE, called the PRO-CTCAE. This Commentary describes development of the PRO-CTCAE by a group of multidisciplinary investigators and patient representatives and provides an overview of qualitative and quantitative studies of its measurement properties. A systematic evaluation of all 790 AEs listed in the CTCAE identified 78 appropriate for patient self-reporting. For each of these, a PRO-CTCAE plain language term in English and one to three items characterizing the frequency, severity, and/or activity interference of the AE were created, rendering a library of 124 PRO-CTCAE items. These items were refined in a cognitive interviewing study among patients on active cancer treatment with diverse educational, racial, and geographic backgrounds. Favorable measurement properties of the items, including construct validity, reliability, responsiveness, and between-mode equivalence, were determined prospectively in a demographically diverse population of patients receiving treatments for many different tumor types. A software platform was built to administer PRO-CTCAE items to clinical trial participants via the internet or telephone interactive voice response and was refined through usability testing. Work is ongoing to translate the PRO-CTCAE into multiple languages and to determine the optimal approach for integrating the PRO-CTCAE into clinical trial workflow and AE analyses. It is envisioned that the PRO-CTCAE will enhance the precision and patient-centeredness of adverse event reporting in cancer clinical research.
    JNCI Journal of the National Cancer Institute 09/2014; 106(9):dju244-dju244. DOI:10.1093/jnci/dju244 · 15.16 Impact Factor
  • Source
    William A. Wood, Antonia V. Bennett, Ethan Basch
    [Show abstract] [Hide abstract]
    ABSTRACT: Recent advancements in consumer directed personal computing technology have led to the generation of biomedically-relevant data streams with potential health applications. This has catalyzed international interest in Patient Generated Health Data (PGHD), defined as “health-related data – including health history, symptoms, biometric data, treatment history, lifestyle choices, and other information-created, recorded, gathered, or inferred by or from patients or their designees (i.e. care partners or those who assist them) to help address a health concern.”(Shapiro et al., 2012) PGHD offers several opportunities to improve the efficiency and output of clinical trials, particularly within oncology. These range from using PGHD to understand mechanisms of action of therapeutic strategies, to understanding and predicting treatment-related toxicity, to designing interventions to improve adherence and clinical outcomes. To facilitate the optimal use of PGHD, methodological research around considerations related to feasibility, validation, measure selection, and modeling of PGHD streams is needed. With successful integration, PGHD can catalyze the application of “big data” to cancer clinical research, creating both “n of 1” and population-level observations, and generating new insights into the nature of health and disease.
    Molecular Oncology 08/2014; 9(5). DOI:10.1016/j.molonc.2014.08.006 · 5.94 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Context Patient-reported outcomes (PROs) provide a way to understand the effects of hematopoietic cell transplantation (HCT)-related stress upon patients' lives. We previously reported that weekly collection of PROs is feasible. Objectives Here, we report on the feasibility of daily patient-reported symptom collection and we examine the relationship between daily vs. weekly symptom reporting over time. Methods We analyzed data from 32 autologous and allogeneic HCT patients obtained until Day (D)+100. We used questions from the PRO version of the Common Terminology Criteria for Adverse Events to capture symptoms. Results We found that overall rates of daily survey completion were moderate to high (range 67%-86%). The effect size of the difference between the maximum daily severity score and the weekly severity score ranged from 0.15 to 0.35, and the concordance correlation coefficient (CCC) ranged from 0.513 to 0.834. Concordance of daily and weekly surveys was higher for maximum daily severity rating and mean daily severity rating than for minimum daily severity rating or most recent daily severity rating. Conclusion We conclude that a seven-day recall period for symptom severity provides acceptable accuracy and precision in the first 100 days following HCT. Further studies to explore the utility of daily symptom reporting within specific clinical contexts may be warranted.
    Journal of Pain and Symptom Management 08/2014; 49(3). DOI:10.1016/j.jpainsymman.2014.06.011 · 2.74 Impact Factor
  • Source
    Ethan Basch, Antonia V Bennett
    [Show abstract] [Hide abstract]
    ABSTRACT: The science of measuring patient-reported outcomes (PROs) has advanced substantially in recent decades, allowing evaluation of how patients feel and function in clinical research. Assessment of the patient experience in populations with rare diseases can be successfully achieved using PRO measures when careful planning and rigorous methods are employed. A number of challenges exist when designing and implementing PRO analyses in rare disease contexts, including heterogeneity of outcomes, availability of suitable measures, recruitment, and selection of appropriate data collection methods. Strategies to address these exist and have been employed in past clinical research, particularly in pediatric populations. PRO assessments in rare disease clinical trials have been particularly successful through partnerships between investigators, PRO methodologists, and patient organizations. The overall goal of PRO measurement is to understand the patient experience and it provides an essential part of evaluating the impact of disease and treatment.
    Journal of General Internal Medicine 07/2014; 29(S3). DOI:10.1007/s11606-014-2892-z · 3.42 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The objective of this report was to address the use and mixing of data collection modes within and between trials in which patient-reported outcome (PRO) end points are intended to be used to support medical product labeling. The report first addresses the factors that should be considered when selecting a mode or modes of PRO data collection in a clinical trial, which is often when mixing is first considered. Next, a summary of how to “faithfully” migrate instruments is presented followed by a section on qualitative and quantitative study designs used to evaluate measurement equivalence of the new and original modes of data collection. Finally, the report discusses a number of issues that must be taken into account when mixing modes is deemed necessary or unavoidable within or between trials, including considerations of the risk of mixing at different levels within a clinical trial program and mixing between different types of platforms. In the absence of documented evidence of measurement equivalence, it is strongly recommended that a quantitative equivalence study be conducted before mixing modes in a trial to ensure that sufficient equivalence can be demonstrated to have confidence in pooling PRO data collected by the different modes. However, we also strongly discourage the mixing of paper and electronic field-based instruments and suggest that mixing of electronic modes be considered for clinical trials and only after equivalence has been established. If proceeding with mixing modes, it is important to implement data collection carefully in the trial itself in a planned manner at the country level or higher and minimize ad hoc mixing by sites or individual subjects. Finally, when mixing occurs, it must be addressed in the statistical analysis plan for the trial and the ability to pool the data must be evaluated to then evaluate treatment effects with mixed modes data. A successful mixed modes trial requires a “faithful migration,” measurement equivalence established between modes, and carefully planned implementation to minimize the risk of increased measurement error impacting the power of the trial to detect a treatment effect.
    Value in Health 07/2014; 17(5):501–516. DOI:10.1016/j.jval.2014.06.005 · 2.89 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Pain negatively affects quality of life for cancer patients. Preliminary data in metastatic castration-resistant prostate cancer (mCRPC) suggested a benefit of the oral tyrosine kinase inhibitor cabozantinib to pain palliation. Prospective evaluation of cabozantinib's benefits on pain and narcotic use in mCRPC. This was a nonrandomized expansion (NRE) cohort (n=144) of a phase 2 randomized discontinuation trial in docetaxel-refractory mCRPC patients. Pain and interference of symptoms with sleep and general activity were electronically self-reported daily for 7-d intervals at baseline and regularly scheduled throughout the study. Mean per-patient scores were calculated for each interval. Narcotic use was recorded daily during the same intervals. Open-label cabozantinib (100mg or 40mg). The following stringent response definition was used: clinically meaningful pain reduction (≥30% improvement in mean scores from baseline) confirmed at a later interval without concomitant increases in narcotics. Only patients with moderate or severe baseline pain were analyzed. Sixty-five patients with moderate or severe baseline pain were evaluable. Of these, 27 (42%) experienced pain palliation according to the stringent response definition. Thirty-seven patients (57%) had clinically meaningful pain relief at two consecutive intervals, reported ≥6 wk apart in the majority. Forty-four patients (68%) had palliation at one or more intervals; 36 (55%) decreased narcotics use during one or more intervals. Clinically meaningful pain reduction was associated with significant (p ≤ 0.001) improvements in sleep quality and general activity. A limitation of this study was its open-label design. Cabozantinib demonstrated clinically meaningful pain palliation, reduced or eliminated patients' narcotic use, and improved patient functioning, thus meriting prospective validation in phase 3 studies. We evaluated the potential of cabozantinib to improve symptoms in patients with metastatic prostate cancer that no longer responds to standard therapies. We saw a promising reduction in pain and reduced need for narcotic painkillers. Larger, well-controlled trials are necessary to confirm these findings.
    European Urology 02/2014; DOI:10.1016/j.eururo.2014.02.013 · 12.48 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Rectal cancer patients' expectations of health and function may affect their disease- and treatment-related experience, but how patients form expectations of postsurgery function has received little study. We used a qualitative approach to explore patient expectations of outcomes related to bowel function after sphincter-preserving surgery for rectal cancer. This was a cohort study of patients who were about to undergo sphincter-preserving surgery for rectal cancer. The study was conducted through individual telephone interviews with participants. Twenty-six patients (14 men and 12 women) with clinical TNM stage I to III disease were enrolled. The semistructured interview script contained open-ended questions on patient expectations of postoperative bowel function and its perceived impact on daily function and life. Two researchers analyzed the interview transcripts for emergent themes using a grounded theory approach. Participant expectations of bowel function reflected 3 major themes: 1) information sources, 2) personal attitudes, and 3) expected outcomes. The expected outcomes theme contained references to specific symptoms and participants' descriptions of the certainty, importance, and imminence of expected outcomes. Despite multiple information sources and attempts at maintaining a positive personal attitude, participants expressed much uncertainty about their long-term bowel function. They were more focused on what they considered more important and imminent concerns about being cancer free and getting through surgery. This study was limited by context in terms of the timing of interviews (relative to the treatment course). The transferability to other contexts requires further study. Patient expectations of long-term functional outcomes cannot be considered outside of the overall context of the cancer experience and the relative importance and imminence of cancer- and treatment-related events. Recognizing the complexities of the expectation formation process offers opportunities to develop strategies to enhance patient education and appropriately manage expectations, attend to immediate and long-term concerns, and support patients through the treatment and recovery process.
    Diseases of the Colon & Rectum 02/2014; 57(2):151-157. DOI:10.1097/DCR.0000000000000036 · 3.20 Impact Factor
  • Source
    Value in Health 11/2013; 16(7):A601. DOI:10.1016/j.jval.2013.08.1701 · 2.89 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Purpose The National Cancer Institute’s Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) is a library of question items that enables patient reporting of adverse events (AEs) in clinical trials. This study contributes content validity evidence of the PRO-CTCAE by incorporating cancer patient input of the relevance and comprehensiveness of the item library. Methods Cognitive interviews were conducted among patients undergoing chemotherapy or radiation therapy at multiple sites to evaluate comprehension, memory retrieval, judgment, and response mapping related to AE terms (e.g., nausea), attribute terms (regarding frequency, severity, or interference), response options, and recall period. Three interview rounds were conducted with ≥20 patients completing each item per round. Items were modified and retested if ≥3 patients exhibited cognitive difficulties or if experienced by ≤25 % patients. Results One hundred and twenty-seven patients participated (35 % ≤high school, 28 % non-white, and 59 % female). Most AE terms (63/80) generated no cognitive difficulties. The remaining 17 were modified without further difficulties by Round 3. Terms were comprehended regardless of education level. Attribute terms and response options required no modifications. Patient adherence to recall period (7 days) was improved when the reference period was incorporated. Conclusions This study provides evidence confirming comprehension of the US English language versions of items in the PRO-CTCAE library for measuring symptomatic AEs from the patient perspective within the context of cancer treatment. Several minor changes were made to the items to improve item clarity, comprehension, and ease of response judgment. This study helps to establish the content validity of PRO-CTCAE items for patient reporting of AEs during cancer treatment.
    Quality of Life Research 07/2013; 106(9). DOI:10.1007/s11136-013-0470-1 · 2.86 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Contemporary tumor-directed therapies for metastatic castration-resistant prostate cancer (mCRPC) are approved to prolong life, but their effects on symptoms such as pain are less well understood as a result of the lack of analytically valid assessments of pain prevalence and severity, clinically meaningful definitions of therapeutic benefit, and methodologic standards of trial conduct. This study establishes pain characteristics in the mCRPC population using a PRO measure. Patients with prostate cancer participated in an anonymous survey at five US comprehensive cancer centers in the Prostate Cancer Clinical Trials Consortium that incorporated the Brief Pain Inventory (BPI), analgesic use, and interference with daily activities. Prevalence and severity of cancer-related pain and analgesic use were tabulated according to castration-resistant status and exposure to docetaxel chemotherapy. Four hundred sixty-one patients with prostate cancer participated, of whom 147 had mCRPC involving bone (61% [89 of 147] docetaxel exposed, 39% [58 of 147] docetaxel naive). Pain of any level was more common among docetaxel-exposed versus docetaxel-naive patients with mCRPC (70% [62 of 89] v 38% [22 of 58], respectively; P < .001). BPI score ≥ 4 was reported by 38% (34 of 89) of docetaxel-pretreated and 24% (14 of 58) of docetaxel-naive patients with mCRPC; 40% of these patients with pain intensity ≥ 4 reported no current narcotic analgesic. Pain prevalence and severity were higher in patients with prior docetaxel exposure. Analgesics were underutilized. These results provide a method for estimating accruals along the disease continuum, and for enabling design of trials appropriately powered to assess pain.
    Journal of Oncology Practice 06/2013; DOI:10.1200/JOP.2013.000876
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSEPatient-reported outcomes are increasingly used in routine outpatient cancer care to guide clinical decisions and enhance communication. Prior evidence suggests good patient compliance with reporting at scheduled clinic visits, but there is limited evidence about compliance with long-term longitudinal reporting between visits. PATIENTS AND METHODS Patients receiving chemotherapy for lung, gynecologic, genitourinary, or breast cancer at a tertiary cancer center, with access to a home computer and prior e-mail experience, were asked to self-report seven symptomatic toxicities via the Web between visits. E-mail reminders were sent to participants weekly; patient-reported high-grade toxicities triggered e-mail alerts to nurses; printed reports were provided to oncologists at visits. A priori threshold criteria were set to determine if this data collection approach merited further development based on monthly (≥ 75% participants reporting at least once per month on average) and weekly compliance rates (60% at least once per week).ResultsBetween September 2006 and November 2010, 286 patients were enrolled (64% were women; 88% were white; median age, 58 years). Mean follow-up was 34 weeks (range, 2 to 214). On average, monthly compliance was 83%, and weekly compliance was 62%, without attrition until the month before death. Greater compliance was associated with older age and higher education but not with performance status. Compliance was greatest during the initial 12 weeks. Symptomatic illness and technical problems were rarely barriers to compliance. CONCLUSION Monthly compliance with home Web reporting was high, but weekly compliance was lower, warranting strategies to enhance compliance in routine care settings.
    Journal of Clinical Oncology 06/2013; 31(20). DOI:10.1200/JCO.2012.47.6804 · 17.88 Impact Factor
  • Source
    A.V. Bennett, K Keenoy, E Basch, L K Temple
    Value in Health 05/2013; 16(3):A33. DOI:10.1016/j.jval.2013.03.192 · 2.89 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: OBJECTIVE: The Brief Pain Inventory (BPI) was designed to yield separate scores for pain intensity and interference. It has been proposed that the pain interference factor can be further broken down into unique factors of affective (e.g., mood) and activity (e.g., work) interference. The purpose of this analysis was to confirm this affective/activity interference dichotomy. PATIENTS AND METHODS: A retrospective confirmatory factor analysis was completed for a sample of 184 individuals diagnosed with castrate-resistant prostate cancer (age 40-86, mean = 65.46, 77% White non-Hispanic) who had been administered the BPI as part of Cancer and Leukemia Group B trial 9480. A one-factor model was compared against two-factor and three-factor models that were developed based on the design of the instrument. RESULTS: Root mean squared error of approximation (0.075), comparative fit index (0.971), and change in chi-square, given the corresponding change in degrees of freedom (13.33, P < 0.05) values for the three-factor model (i.e., pain intensity, activity interference, and affective interference), were statistically superior in comparison with the one- and two-factor models. This three-factor structure was found to be invariant across age, mean prostate-specific antigen, and hemoglobin levels. CONCLUSIONS: These results confirm that the BPI can be used to quantify the degree to which pain separately interferes with affective and activity aspects of a patient's everyday life. These findings will provide clinical trialists, pharmaceutical sponsors, and regulators with confidence in the flexibility of the BPI as they consider the use of this instrument to assist with understanding the patient experience as it relates to treatment.
    Pain Medicine 10/2012; 13(11). DOI:10.1111/j.1526-4637.2012.01498.x · 2.24 Impact Factor
  • Source
    Antonia V Bennett, Roxanne E Jensen, Ethan Basch
    [Show abstract] [Hide abstract]
    ABSTRACT: Answer questions and earn CME/CNE Patient-reported outcome (PRO) questionnaires assess topics a patient can report about his or her own health. This includes symptoms (eg, nausea, fatigue, diarrhea, pain, or frequent urination), physical functioning (eg, difficulty climbing stairs or difficulty fastening buttons), and mental health (eg, anxiety, fear, or worry). Electronic PRO (ePRO) systems are used in oncology clinical care because of 1) their ability to enhance clinical care by flagging important symptoms and saving clinicians time; 2) the availability of standardized methods for creating and implementing PROs in clinics; and 3) the existence of user-friendly platforms for patient self-reporting like tablet computers and automated telephone surveys. Many ePRO systems can provide actionable links to clinical care such as summary reports in a patient's electronic medical record and real-time e-mail alerts to providers when patients report acute needs. This review presents 5 examples of ePRO systems currently in use in oncology practice. These systems support multiple clinical activities, including assessment of symptoms and toxicities related to chemotherapy and radiation, postoperative surveillance, and symptom management during palliative care and hospice. Patient self-reporting is possible both at clinical visits and between visits over the Internet or by telephone. The implementation of an ePRO system requires significant resources and expertise, as well as user training. ePRO systems enable regular monitoring of patient symptoms, function, and needs, and can enhance the efficiency and quality of care as well as communication with patients. CA Cancer J Clin 2012. © 2012 American Cancer Society.
    CA A Cancer Journal for Clinicians 07/2012; 62(5):336-47. DOI:10.3322/caac.21150 · 162.50 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We used the Female Sexual Function Index (FSFI) to investigate the prevalence of sexual dysfunction (SD) and factors associated with diminished sexual functioning in early stage endometrial cancer (EC) patients treated with simple hysterectomy and adjuvant brachytherapy. A cohort of 104 patients followed in a radiation oncology clinic completed questionnaires to quantify current levels of sexual functioning. The time interval between hysterectomy and questionnaire completion ranged from <6 months to >5 years. Multivariate regression was performed using the FSFI as a continuous variable (score range, 1.2-35.4). SD was defined as an FSFI score of <26, based on the published validation study. SD was reported by 81% of respondents. The mean (± standard deviation) domain scores in order of highest-to-lowest functioning were: satisfaction, 2.9 (± 2.0); orgasm, 2.5 (± 2.4); desire, 2.4 (± 1.3); arousal, 2.2 (± 2.0); dryness, 2.1 (± 2.1); and pain, 1.9 (± 2.3). Compared to the index population in which the FSFI cut-score was validated (healthy women ages 18-74), all scores were low. Compared to published scores of a postmenopausal population, scores were not statistically different. Multivariate analysis isolated factors associated with lower FSFI scores, including having laparotomy as opposed to minimally invasive surgery (effect size, -7.1 points; 95% CI, -11.2 to -3.1; P<.001), lack of vaginal lubricant use (effect size, -4.4 points; 95% CI, -8.7 to -0.2, P=.040), and short time interval (<6 months) from hysterectomy to questionnaire completion (effect size, -4.6 points; 95% CI, -9.3-0.2; P=.059). The rate of SD, as defined by an FSFI score <26, was prevalent. The postmenopausal status of EC patients alone is a known risk factor for SD. Additional factors associated with poor sexual functioning following treatment for EC included receipt of laparotomy and lack of vaginal lubricant use.
    International journal of radiation oncology, biology, physics 05/2012; 84(2):e187-93. DOI:10.1016/j.ijrobp.2012.03.030 · 4.18 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Patient reporting of symptoms in a questionnaire with a 7-day recall period was expected to differ from symptom reporting in a 7-day symptom diary on the basis of cognitive theory of memory processes and several studies of symptoms and health behaviors. A total of 101 adults with chronic obstructive pulmonary disease (COPD) completed a daily diary of items measuring symptoms and impacts of COPD for 7 days, and on the seventh day they completed a questionnaire of the same items with a 7-day recall period. The analysis examined concordance of 7-day recall with summary descriptors of the daily responses, examined the magnitude and covariates (patient characteristics and response patterns) of the difference between 7-day recall and mean of daily responses, and compared the discriminant ability and ability to detect change of 7-day recall and mean of daily responses. A 7-day recall was moderately concordant with the mean and maximum of daily responses and was 0.34 to 0.50 SDs higher than the mean of daily responses. Only the weekly report itself was a covariate of the difference. The discriminant ability and ability to detect change were equivalent. In measuring the weeklong experience of COPD symptoms and impacts on groups of patients, the 7-day recall scores were higher than the daily diary scores, but equivalent in detecting change over time.
    Value in Health 05/2012; 15(3):466-74. DOI:10.1016/j.jval.2011.12.005 · 2.89 Impact Factor
  • Source
    Ethan Basch, Antonia Bennett, M Catherine Pietanza
    CancerSpectrum Knowledge Environment 12/2011; 103(24):1808-10. DOI:10.1093/jnci/djr493 · 15.16 Impact Factor
  • E. Basch, A. V. Bennett, H. I. Scher
    Molecular Cancer Therapeutics 11/2011; 10(Supplement 1):B57-B57. DOI:10.1158/1535-7163.TARG-11-B57 · 6.11 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Adverse symptom event reporting is vital as part of clinical trials and drug labeling to ensure patient safety and inform risk-benefit decision making. The purpose of this study was to assess the reliability of adverse event reporting of different clinicians for the same patient for the same visit. A retrospective reliability analysis was completed for a sample of 393 cancer patients (42.8% men; age 26-91, M = 62.39) from lung (n = 134), prostate (n = 113), and Ob/Gyn (n = 146) clinics. These patients were each seen by two clinicians who independently rated seven Common Terminology Criteria for Adverse Events (CTCAE) symptoms. Twenty-three percent of patients were enrolled in therapeutic clinical trials. The average time between rater evaluations was 68 min. Intraclass correlation coefficients were moderate for constipation (0.50), diarrhea (0.58), dyspnea (0.69), fatigue (0.50), nausea (0.52), neuropathy (0.71), and vomiting (0.46). These values demonstrated stability over follow-up visits. Two-point differences, which would likely affect treatment decisions, were most frequently seen among symptomatic patients for constipation (18%), vomiting (15%), and nausea (8%). Agreement between different clinicians when reporting adverse symptom events is moderate at best. Modification of approaches to adverse symptom reporting, such as patient self-reporting, should be considered.
    Quality of Life Research 10/2011; 21(7):1159-64. DOI:10.1007/s11136-011-0031-4 · 2.86 Impact Factor

Publication Stats

145 Citations
297.34 Total Impact Points

Institutions

  • 2013–2015
    • University of North Carolina at Chapel Hill
      • Department of Health Policy & Management
      North Carolina, United States
  • 2010–2013
    • Memorial Sloan-Kettering Cancer Center
      • Epidemiology & Biostatistics Group
      New York City, New York, United States
  • 2008
    • University of Washington Seattle
      • Department of Health Services
      Seattle, Washington, United States