The creation of healthcare administrative databases provides the unprecedented opportunity to evaluate the clinical effectiveness of treatments strategies, and the quality and efficiency of health care delivery systems in large and heterogeneous populations. The imperatives to reduce health disparities and to evaluate health care delivery systems have never been greater. These assessments must be done with the highest methodological rigor, with the best possible data, must engage patients, and must become a routine part of our health system. I am a Professor of Biostatistics and Associate Dean of Information Technology at the Harvard School of Public Health (HSPH). During my career at Johns Hopkins University, 1999-2009, I have developed statistical methods for the analysis of large databases on air pollution and health (see item 2). I have gained experience with the analysis of Medicare data and their linkage by geography and time to other data sources, such us air pollution, weather, and socioeconomic status. During this phase of my career (from post-doctoral fellow to Professor), I have developed statistical methods for the analysis of these large data (e.g. methods for the adjustment of measured and unmeasured confounders, Bayesian hierarchical models, causal inference methods, and missing data methods.) In 2009 I was recruited as a Full Professor in the Department of Biostatistics at Harvard School of Public Health. Within the first two years of my appointment, I have been awarded leadership positions in administration (Associated Dean of Information Technology), mentorship (co- PI of a NIEHS funded training grant in Environmental Biostatistics) and in cancer research (PI of a resubmission of NCI P01 on Cancer Informatics). I have started to develop new collaborations with colleagues at the Dana Farber Cancer Institute (DFCI) (Drs. Schrag, Alexander, Block) and Health Care Policy (HCP) at Harvard Medical School (HMS) (Dr. Normand). As a result of being in a new environment and with new colleagues I started to realize that my expertise in the analysis of Medicare data, in development of causal inference methods for the assessment of health benefits of environmental interventions could be easily transportable and enhanced to address questions in CER of critical importance. In particular, both for personal reasons and because of the growing discussions surrounding the health care reform, I became increasingly interested in better understanding how the use of claims data can address questions regarding ways to best deliver care for patients that have a terminal cancer. In fact, for these populations how to deliver the best care, achieve the best outcomes, and at the same time, containing medical costs is very challenging. The gaps of knowledge in this area are enormous and questions regarding best ways of deliver care in presence of several concomitant factors (e.g. multiple treatments, coordination among medical teams, palliative care) cannot be addressed with randomized clinical trials, only. The linkage and analysis of healthcare administrative databases (e.g. Medicare), of which I have developed expertise as environmental scientist, provides the unprecedented opportunity to evaluate the clinical effectiveness of treatments strategies, and the quality and efficiency of health care delivery systems in cancer research. I am seeking this K18 to: 1) improve my ability to pursue future research in patient centered outcomes in end of life (EOL) care for cancer patients;2) mentor the junior biostatisticians that increasingly want to get involved in CER;3) more effectively collaborate with clinical investigators and stakeholders and 4) further develop statistical methods so these complex CER questions can be addressed with the highest methodological rigor.
The specific aims are: 1) Participate in an intense, mentored career development experience in comparative effectiveness (CER) with a special focus on cancer (see Item 4 and mentoring plan). The proposed training will be directly targeted to address the specific aims detailed in my research plan (Item 11); 2) Conduct a research project focused on addressing key challenges on how to best deliver care for a well defined population: elderly diagnosed with Glioblastoma. These will be the largest populations studied to date (N=24,142, Part A Medicare data, N=9,343 SEER-Medicare, and N=9,320, Medicaid-Tumor Registry, N=GBM cases). The research plan will be a vehicle to experience, first hand, how to provide a solid evidence base to impact policy and to improve health care experience of the elderly with an advanced cancer (Item 11);3) Leverage the local wealth of expertise in CER in cancer and EOL using the oversight of a prestigious science and stakeholder advisory board. The board includes experts in health care policy, CER, oncology, health disparities, palliative care, neuro-oncology, and decision making for health care reimbursement. The board also includes two patient advocates and a Pulitzer Prize-winning journalist expert in EOL (Diana Sugg) (Item 4). The successful completion of this project will provide capacity building (data and methods), identify important data and methodological gaps, and address questions of paramount importance in health care delivery and health disparities for the largest population of elderly GBM patients studied to date. Data, methods, and results will also advance CER research in other cancer populations. This career development award will also provide me with multidisciplinary expertise and fruitful collaborations necessary for training the next generation of scientists in CER for PCOR.

Public Health Relevance

Little is known about the salient characteristics of the health care delivery systems that provide excellent and cost-effective health care for patients diagnosed with a terminal cancer. The creation of healthcare administrative databases (e.g. Medicare, SEER-Medicare, Medicaid) provides the unprecedented opportunity to evaluate the clinical effectiveness of treatments strategies, and the quality and efficiency of health care delivery systems in large and heterogeneous populations of cancer patients. With this K18 I am seeking an intense, mentored career development experience in CER that will substantially improve my ability to pursue future research in PCOR for patients with cancer at an advanced stage. My longer-term goal is to become an independent investigator that will champion this field.

National Institute of Health (NIH)
Agency for Healthcare Research and Quality (AHRQ)
The Career Enhancement Award (K18)
Project #
Application #
Study Section
Special Emphasis Panel (ZHS1-HSR-A (01))
Program Officer
Willis, Tamara
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Harvard University
Biostatistics & Other Math Sci
Schools of Public Health
United States
Zip Code
Arvold, Nils D; Cefalu, Matthew; Wang, Yun et al. (2017) Comparative effectiveness of radiotherapy with vs. without temozolomide in older patients with glioblastoma. J Neurooncol 131:301-311
Braun, Danielle; Gorfine, Malka; Parmigiani, Giovanni et al. (2017) Propensity scores with misclassified treatment assignment: a likelihood-based adjustment. Biostatistics 18:695-710
Cefalu, Matthew; Dominici, Francesca; Arvold, Nils et al. (2017) Model averaged double robust estimation. Biometrics 73:410-421
Wang, Yun; Eldridge, Noel; Metersky, Mark L et al. (2016) Association Between Hospital Performance on Patient Safety and 30-Day Mortality and Unplanned Readmission for Medicare Fee-for-Service Patients With Acute Myocardial Infarction. J Am Heart Assoc 5:
Lee, Kyu Ha; Dominici, Francesca; Schrag, Deborah et al. (2016) Hierarchical models for semi-competing risks data with application to quality of end-of-life care for pancreatic cancer. J Am Stat Assoc 111:1075-1095
Dominici, Francesca; Cefalu, Matthew (2015) Confounding adjustment and exposure prediction in environmental epidemiology: additional insights. Epidemiology 26:e28
Bell, Michelle L; Son, Ji-Young; Peng, Roger D et al. (2015) Ambient PM2.5 and Risk of Hospital Admissions: Do Risks Differ for Men and Women? Epidemiology 26:575-9
Wang, Chi; Dominici, Francesca; Parmigiani, Giovanni et al. (2015) Accounting for uncertainty in confounder and effect modifier selection when estimating average causal effects in generalized linear models. Biometrics 71:654-65
Bobb, Jennifer F; Obermeyer, Ziad; Wang, Yun et al. (2014) Cause-specific risk of hospital admission related to extreme heat in older adults. JAMA 312:2659-67
Obermeyer, Ziad; Makar, Maggie; Abujaber, Samer et al. (2014) Association between the Medicare hospice benefit and health care utilization and costs for patients with poor-prognosis cancer. JAMA 312:1888-96

Showing the most recent 10 out of 11 publications