Peter A Ubel

Dana-Farber Cancer Institute, Boston, Massachusetts, United States

Are you Peter A Ubel?

Claim your profile

Publications (286)1497.06 Total impact

  • Annals of internal medicine 10/2014; 161(8):605-606. · 13.98 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Purpose: Even when patients are educated about their treatment choices, they are often not activated or able to engage with their physicians regarding their treatment values, goals, or preferences. This study tested whether a DVD that taught and modeled shared decision techniques making to prostate cancer patients would improve their self-efficacy for communicating effectively with their physicians (urologists and radiation oncologists) and their report of whether they engaged in shared decision making. Methods: 564 men undergoing a prostate biopsy were recruited and randomized to receive either a decision aid booklet only or a decision aid booklet + DVD. The DVD follows a patient going through a typical prostate cancer diagnosis visit and modeled shared decision making strategies such as how to better ask questions, question physician recommendations (against active surveillance) and express their values and treatment preferences. Participants who were diagnosed with localized prostate cancer (N=200) completed surveys at recruitment, before learning their diagnosis, and then following their treatment discussion with their medical team. Primary outcome variables included perceived and actual ability to: ask questions, ask for a second opinion, ask for a referral to a radiation oncologist, bring questions, take notes, and participate in shared decision making. Results: Participants who received the DVD reported stronger intentions to: ask questions (p=0.001), ask for second opinions from another urologist (p=0.001), ask for a referral to see a radiation oncologist (p=0.006), take notes (p=0.001), and to participate in shared decision making (p=0.006). However, when surveyed after meeting with their physicians there were no differences in patients’ perceptions of performing these behaviors with either their urologist or their radiation oncologist (measured separately) during their clinic visit(s). Conclusions: Even when patients intend and feel confident about engaging in shared decision making, it is often difficult for them to feel they have succeeded in doing so. It is crucial that we explore methods to better help patients participate in shared decision making at the level they desire.
    The 36th Annual Meeting of the Society for Medical Decision Making; 10/2014
  • Peter A Ubel
    The American Journal of Bioethics 09/2014; 14(9):37-8. · 3.60 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Resources, including space, equipment, funding, personnel, and protected time, are essential in academic medical careers. Negotiation often plays a key role in the distribution of these resources.
    Journal of general internal medicine. 08/2014;
  • [Show abstract] [Hide abstract]
    ABSTRACT: Little is known about the association between patient-oncologist discussion of cancer treatment out-of-pocket (OOP) cost and medication adherence, a critical component of quality cancer care.
    Journal of oncology practice / American Society of Clinical Oncology. 05/2014; 10(3):162-7.
  • Peter A Ubel, Reshma Jagsi
    New England Journal of Medicine 04/2014; 370(14):1280-1. · 54.42 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Whole genome sequencing (WGS) is already being used in certain clinical and research settings, but its impact on patient well-being, health-care utilization, and clinical decision-making remains largely unstudied. It is also unknown how best to communicate sequencing results to physicians and patients to improve health. We describe the design of the MedSeq Project: the first randomized trials of WGS in clinical care.Methods/design: This pair of randomized controlled trials compares WGS to standard of care in two clinical contexts: (a) disease-specific genomic medicine in a cardiomyopathy clinic and (b) general genomic medicine in primary care. We are recruiting 8 to 12 cardiologists, 8 to 12 primary care physicians, and approximately 200 of their patients. Patient participants in both the cardiology and primary care trials are randomly assigned to receive a family history assessment with or without WGS. Our laboratory delivers a genome report to physician participants that balances the needs to enhance understandability of genomic information and to convey its complexity. We provide an educational curriculum for physician participants and offer them a hotline to genetics professionals for guidance in interpreting and managing their patients' genome reports. Using varied data sources, including surveys, semi-structured interviews, and review of clinical data, we measure the attitudes and behaviors of physician and patient participants at multiple time points before and after the disclosure of these results. The impact of emerging sequencing technologies on patient care is unclear. We have designed a process of interpreting WGS results and delivering them to physicians in a way that anticipates how we envision genomic medicine to evolve in the near future. That is, our WGS report provides clinically relevant information while communicating the complexity and uncertainty of WGS results to physicians and, through physicians, to their patients. This project will not only illuminate the impact of integrating genomic medicine into the clinical care of patients but also inform the design of future studies.Trial registration: ClinicalTrials.gov identifier NCT01736566.
    Trials 03/2014; 15(1):85. · 2.21 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Female physician-researchers do not achieve career success at the same rate as men. Differences in nonprofessional responsibilities may partially explain this gap. To investigate the division of domestic labor by gender in a motivated group of early-career physician-researchers. Nationwide postal survey between 2010 and 2011. United States. Physician recipients of National Institutes of Health K08 or K23 awards between 2006 and 2009 with active academic affiliation at the time of the survey. Time spent on parenting and domestic tasks was determined through self-report. Among married or partnered respondents with children, a linear regression model of time spent on domestic activities was constructed considering age, gender, race, specialty, MD or MD/PhD status, age of youngest child, number of children, work hours, K award type, and spousal employment. A 74% response rate was achieved, and 1049 respondents were academic physicians. Women were more likely than men to have spouses or domestic partners who were employed full-time (85.6% [95% CI, 82.7% to 89.2%] vs. 44.9% [CI, 40.8% to 49.8%]). Among married or partnered respondents with children, after adjustment for work hours, spousal employment, and other factors, women spent 8.5 more hours per week on domestic activities. In the subgroup with spouses or domestic partners who were employed full-time, women were more likely to take time off during disruptions of usual child care arrangements than men (42.6% [CI, 36.6% to 49.0%] vs. 12.4% [CI, 5.4% to 19.5%]). Analyses relied on self-reported data. The study design did not enable investigation of the relationship between domestic activities and professional success. In this sample of career-oriented professionals, gender differences in domestic activities existed among those with children. Most men's spouses or domestic partners were not employed full-time, which contrasted sharply with the experiences of women. National Institutes of Health.
    Annals of internal medicine 03/2014; 160(5). · 13.98 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Comparative effectiveness research aims to inform health-care decisions by patients, clinicians, and policy makers. However, questions related to what information is relevant, and how to view the relative attributes of alternative interventions have political, social, and medical considerations. In particular, questions about whether cost is a relevant factor, and whether cost-effectiveness is a desirable or necessary component of such research, have become increasingly controversial as the area has gained prominence. Debate has emerged about whether comparative effectiveness research promotes rationing of cancer care. At the heart of this debate are questions related to the role and limits of patient autonomy, physician discretion in health-care decision making, and the nature of scientific knowledge as an objective good. In this article, we examine the role of comparative effectiveness research in the USA, UK, Canada, and other health-care systems, and the relation between research and policy. As we show, all health systems struggle to balance access to cancer care and control of costs; comparative effectiveness data can clarify choices, but does not itself determine policy or promote rationing of care.
    The Lancet Oncology 02/2014; · 25.12 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Objective With increasing exposure, medical students may forget that technical jargon is unfamiliar to laypeople. To investigate this possibility, authors assessed student perceptions of patient understanding across different years in medical school. Methods 533 students at 4 U.S. medical schools rated the proportion of patients likely to understand each of twenty-one different jargon terms. Students were either in the first month of their first year, the middle of their first year, or the middle of their fourth year of medical school. Results Fourth-year students were slightly more pessimistic about patients’ understanding compared to new first-year students (mean percent understanding of 55.1% vs. 58.6%, p = 0.004). Students both over- and under-estimated patient understanding of specific words compared to published estimates. In a multivariate model, other factors did not explain these differences. Conclusion Students do not generally presume that patients understand medical jargon. In many cases they actually underestimate patients’ understanding, and these estimates may become more pessimistic longitudinally. Jargon use in communication with patients does not appear to stem from unrealistic presumptions about patients’ understanding or from desensitization to jargon during medical school. Practice Implications Training about patient knowledge of medical jargon may be a useful addition to communication skills curricula.
    Patient Education and Counseling 01/2014; · 2.60 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: We know little about patient-physician communication during visits to discuss diagnosis and treatment of prostate cancer. To examine the overall visit structure and how patients and physicians transition between communication activities during visits in which patients received new prostate cancer diagnoses. Forty veterans and 18 urologists at one VA medical centre. We coded 40 transcripts to identify major communication activities during visits and used empiric discourse analysis to analyse transitions between activities. We identified five communication activities that occurred in the following typical sequence: 'diagnosis delivery', 'risk classification', 'options talk', 'decision talk' and 'next steps'. The first two activities were typically brief and involved minimal patient participation. Options talk was typically the longest activity; physicians explicitly announced the beginning of options talk and framed it as their professional responsibility. Some patients were unsure of the purpose of visit and/or who should make treatment decisions. Visits to deliver the diagnosis of early stage prostate cancer follow a regular sequence of communication activities. Physicians focus on discussing treatment options and devote comparatively little time and attention to discussing the new cancer diagnosis. Towards the goal of promoting patient-centred communication, physicians should consider eliciting patient reactions after diagnosis delivery and explaining the decision-making process before describing treatment options.
    Health expectations: an international journal of public participation in health care and health policy 12/2013; · 1.80 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To explore aspects of mentoring that might influence medical faculty career satisfaction and to discover whether there are gender differences. In 2010-2011, the authors surveyed 1,708 clinician-researchers who received (in 2006-2009) National Institutes of Health K08 and K23 awards, which provided mentoring for career development. The authors compared, by gender, the development and nature of mentoring relationships, mentor characteristics, extent of mentoring in various mentor roles, and satisfaction with mentoring. They evaluated associations between mentoring and career satisfaction using multivariable linear regression analysis. The authors received 1,275 responses (75% response rate). Of these respondents, 1,227 (96%) were receiving K award support at the time and constituted the analytic sample. Many respondents had > 1 designated mentor (440/558 women, 79%; 410/668 men, 61%; P < .001). Few were dissatisfied with mentoring (122/1,220, 10.0%; no significant gender difference). Career dissatisfaction was generally low, but 289/553 women (52%) and 268/663 men (40%) were dissatisfied with work-life balance (P < .001). Time spent meeting or communicating with the mentor, mentor behaviors, mentor prestige, extent of mentoring in various roles, and collegiality of the mentoring relationship were significantly associated with career satisfaction. Mentor gender, gender concordance of the mentoring pair, and number of mentors were not significantly associated with satisfaction. This study of junior faculty holding mentored career development awards showed strong associations between several aspects of mentoring and career satisfaction, indicating that those concerned about faculty attrition from academic medicine should consider mentor training and development.
    Academic medicine: journal of the Association of American Medical Colleges 12/2013; · 2.34 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Purpose: The current study tested whether having access to a decision aid prior to physician consultation resulted in stable treatment preferences. Method: 1,023 patients presenting to one of four VAs for a prostate cancer biopsy completed baseline measures and were randomized to receive 1 of 2 decision aids. 244 patients diagnosed with prostate cancer completed 2 additional surveys (73% response rate). Time 2 occurred after they had read their decision aid but just prior to receiving their diagnosis. Patients were re-interviewed 1 week later (Time 3), which was following their diagnosis. Electronic medical records were used to determine the treatment patients received. Surveys included measures of treatment preference, literacy, patients’ perceptions of their physicians recommendations and their communication with their physician. Result: Patients’ preferences for treatment from before they saw their doctor (Time 2) and 1 week later (Time 3) changed significantly: 39% of patients who originally indicated they wanted surgery did not want surgery at Time 3, 43% of those originally expressing interest in active surveillance decided not to pursue that treatment and 77% of those who originally wanted external beam radiation (XRT) did not decide to have XRT. Instability of preferences from Time 2 to the treatment they received was similar: 44% for surgery, 53% for active surveillance, and74% for XRT. In contrast, patients’ preferences at Time 3 were highly related to their physicians’ recommendation the week before: 81% of patients who said their urologists recommended surgery ultimatiely decided on surgery. 82% of those receiving an XRT recommendation chose radiation. Finally, 67% of those whose physician recommended active surveillance reported that as their treatment decision at Time 3. Similarly physicians’ recommendations were strongly related to the treatment patients received: 73% for surgery, 63% for external bema radiation, and 93% for active surveillance. Predictors of stability of patients preferences were prostate cancer related knowledge (p<0.05), literacy (p<0.05) and confidence in decision (from COMRADE, p<0.05). Conclusion: Even after reading a decision aid prior to talking with their urologist, many prostate cancer patients’ treatment preferences changed after talking with their physician. Patients’ treatment decisions were highly concordant with their urologist’s recommendation. Patients who had greater knowledge of prostate cancer, higher literacy scores and were confident in their decision were more likely to have stable treatment preferences.
    The 35th Annual Meeting of the Society for Medical Decision Making; 10/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: Purpose: Research has shown that racial minorities often have different treatment preferences than Whites. One question is whether this is the result of knowledge deficits, differences in literacy or numeracy levels, or any of a number of factors. In the present research, men receiving prostate cancer biopsies were given either a low or high literacy decision aid (DA), and then treatment preferences were assessed. The purpose of the present research was to explore racial differences in treatment preferences following exposure to low versus high literacy DAs. Method: 1023 men from four VA hospitals were recruited at their prostate cancer biopsy appointment; 77% of the participants were White, and 23% were non-White (of which 92% were African American). At Time 1 participants were randomly assigned to receive either a low or high literacy prostate cancer decision aid. Literacy (using the REALM), and subjective numeracy were also assessed. At Time 2, which occurred after participants read the DA but just prior to receiving a prostate cancer diagnosis, men were asked to report their treatment preferences by indicating whether they were considering each treatment option. Knowledge about the treatment options was also assessed. Result: Across both DAs, more White men were interested in active surveillance (41%) than non-White men (23%). However this effect was moderated by the type of DA, B=1.40, SE=.37, p=.02: Relative to the high literacy DA, the low literacy DA increased interest in active surveillance for Whites but not non-Whites. The low literacy DA also decreased interest in external beam radiation for Whites but not non-Whites, B=-1.29, SE=.55, p=.02. These interactions were still significant when controlling for numeracy, literacy, and knowledge. There were no effects involving race for interest in any other treatment option (i.e., surgery, brachytherapy). Conclusion: Results showed that White men’s preferences were influenced by the low literacy DA, but the preferences of non-White men stayed the same regardless of the DA type. Moreover, the effect of the different DAs did not change when controlling for numeracy, literacy, or knowledge. These results suggest that Whites may be more likely to change their treatment preferences in response to a lower-literacy DA, but the treatment preferences of racial minorities are rooted in other factors that were not identified from the current analysis.
    The 35th Annual Meeting of the Society for Medical Decision Making; 10/2013
  • [Show abstract] [Hide abstract]
    ABSTRACT: Purpose: Shared decision making is considered the gold standard for making effective, informed treatment decisions. Yet, a persistent physician concern about engaging in the process is that it will significantly increase clinic time. To address this question empirically, we evaluated transcripts of diagnosis visits in early stage prostate cancer to determine the correlation between informing quality and time in the encounter. Methods: 258 men with localized prostate cancer (PSA<20, Gleason score of 6 or 7) from 4 VA hospitals had their diagnosis visits recorded. As reported previously, men had received one of two decision aids, varying by literacy level, but not content. Time in minutes was obtained from audio recordings. Quality of informing was measured using the Informed Decision Making (IDM) method developed by Braddock, et al. (IDM scale possible range=0-18). Transcripts were scored independently by two trained raters and disagreements resolved by consensus. Treating urologists were mostly senior residents who did not receive special training in shared decision-making. Results: Preliminary data analysis in 211 encounters with complete IDM scores showed a correlation of .25 between time in the encounter and quality of informing. Observed range of IDM scores was 0-15, suggesting physician performance from poor to excellent. IDM scores showed modest quality (IDM M+SD=7.63±2.47). Observed range of encounter times was 6 to 59 minutes (min M+SD =23±10). Analysis of times by quartile quality (IDM) scores shows that the top quartile (IDM M+SD =10.6±1.2) and the bottom quartile (IDM M±SD =4.4±1.5) were similar in the length of the appointment (min M±SD =26±11; min M±SD =22±12 respectively). Times were highly variable, as were IDM scores. Conclusions: The low correlation between time spent in the encounter and the quality of physician informing suggests that higher quality informing does not require more time even in complex, multiple option decision encounters. This suggests it is very important to identify critical shared decision making skills for physician training. A parsimonious skill model for high quality informed and shared decision making may be attainable, though even in the top quartile of IDM scores, improvement is needed.
    The 35th Annual Meeting of the Society for Medical Decision Making; 10/2013
  • Source
    New England Journal of Medicine 10/2013; 369(16):1484-6. · 54.42 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Introduction: Tamoxifen and raloxifene are chemopreventive drugs that can reduce women’s relative risk of primary breast cancer by 50%; however, most women eligible for these drugs have chosen not to take them. The reasons for low uptake may be related to women’s knowledge or attitudes towards the drugs. We aimed to examine the impact of an online breast cancer chemoprevention decision aid (DA) on informed intentions and decisions of women at high risk of breast cancer. Methods: We conducted a Randomized Clinical Trial, assessing the effect of a decision aid about breast cancer chemoprevention on informed choices about chemoprevention. Women (n=585) aged 46-74 completed online baseline, post-test, and 3-month follow-up questionnaires. Participants were randomly assigned to either an intervention group, a standard control group that answered questions about chemoprevention at baseline, or a 3-month control group that did not answer questions about chemoprevention at baseline. The main outcome measures were whether women’s intentions and decisions regarding chemoprevention drugs were informed, and whether women who viewed the DA were more likely to make informed decisions than women who did not view the DA, using a dichotomous composite variable “informed choice” [yes/no] to classify informed decisions as those reflecting sufficient knowledge and concordance between a woman’s decision and relevant attitudes. Results: Analyses showed that more intervention than standard control participants (52.7% vs. 5.9%) made informed decisions at post-test, p<0.001. At 3-month follow-up, differences in rates of informed choice between intervention (16.9%) and both control groups (11.8% and 8.0%) were statistically non-significant, p=0.067. Conclusions: The DA increased informed decision making about breast cancer chemoprevention, although impact on knowledge diminished over time. This study was not designed to determine how much knowledge decision makers must retain over time. Examining informed decisions increases understanding of the impact of DAs. A standard for defining and measuring sufficient knowledge for informed decisions is needed. Trial registration: ClinicalTrials.gov; number NCT00967824
    Breast Cancer Research 10/2013; · 5.33 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Studies have suggested that male physicians earn more than their female counterparts. The authors examined whether this disparity exists in a recently hired cohort. In 2010-2011, the authors surveyed recent recipients of National Institutes of Health (NIH) mentored career development (i.e., K08 or K23) awards, receiving responses from 1,275 (75% response rate). For the 1,012 physicians with academic positions in clinical specialties who reported salary, they constructed linear regression models of salary considering gender, age, race, marital status, parental status, additional doctoral degree, academic rank, years on faculty, specialty, institution type, region, institution NIH funding rank, K award type, K award funding institute, K award year, work hours, and research time. They evaluated the explanatory value of spousal employment status using Peters-Belson regression. Mean salary was $141,325 (95% confidence interval [CI] 135,607-147,043) for women and $172,164 (95% CI 167,357-176,971) for men. Male gender remained an independent, significant predictor of salary (+$10,921, P < .001) even after adjusting for specialty, academic rank, work hours, research time, and other factors. Peters-Belson analysis indicated that 17% of the overall disparity in the full sample was unexplained by the measured covariates. In the married subset, after accounting for spousal employment status, 10% remained unexplained. The authors observed, in this recent cohort of elite, early-career physician-researchers, a gender difference in salary that was not fully explained by specialty, academic rank, work hours, or even spousal employment. Creating more equitable procedures for establishing salary is important.
    Academic medicine: journal of the Association of American Medical Colleges 09/2013; · 2.34 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Tamoxifen and raloxifene are chemopreventive drugs that can reduce women's relative risk of primary breast cancer by 50%; however, most women eligible for these drugs have chosen not to take them. The reasons for low uptake may be related to women's knowledge or attitudes towards the drugs. We aimed to examine the impact of an online breast cancer chemoprevention decision aid (DA) on informed intentions and decisions of women at high risk of breast cancer. We conducted a Randomized Clinical Trial, assessing the effect of a decision aid about breast cancer chemoprevention on informed choices about chemoprevention. Women (n=585) aged 46 to 74 years old completed online baseline, post-test, and 3-month follow-up questionnaires. Participants were randomly assigned to either an intervention group, a standard control group that answered questions about chemoprevention at baseline, or a 3-month control group that did not answer questions about chemoprevention at baseline. The main outcome measures were whether women's intentions and decisions regarding chemoprevention drugs were informed, and whether women who viewed the DA were more likely to make informed decisions than women who did not view the DA, using a dichotomous composite variable "informed choice" [yes/no] to classify informed decisions as those reflecting sufficient knowledge and concordance between a woman's decision and relevant attitudes. Analyses showed that more intervention than standard control participants (52.7% vs. 5.9%) made informed decisions at post-test, P<0.001. At the 3-month follow-up, differences in rates of informed choice between intervention (16.9%) and both control groups (11.8% and 8.0%) were statistically non-significant, P=0.067. The DA increased informed decision making about breast cancer chemoprevention, although the impact on knowledge diminished over time. This study was not designed to determine how much knowledge decision makers must retain over time. Examining informed decisions increases understanding of the impact of DAs. A standard for defining and measuring sufficient knowledge for informed decisions is needed. Trial registration: ClinicalTrials.gov; number NCT00967824.
    Breast cancer research: BCR 09/2013; 15(5):R74. · 5.87 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: and OBJECTIVE: Existing research concludes that measures of general numeracy can be used to predict individuals' ability to assess health risks. We posit that the domain in which questions are posed affects the ability to perform mathematical tasks, raising the possibility of a separate construct of "health numeracy" that is distinct from general numeracy. The objective was to determine whether older adults' ability to perform simple math depends on domain. Community-based participants completed 4 math questions posed in 3 different domains: a health domain, a financial domain, and a pure math domain. Participants were 962 individuals aged 55 and older, representative of the community-dwelling US population over age 54. We found that respondents performed significantly worse when questions were posed in the health domain (54% correct) than in either the pure math domain (66% correct) or the financial domain (63% correct). Our experimental measure of numeracy consisted of only 4 questions, and it is possible that the apparent effect of domain is specific to the mathematical tasks that these questions require. These results suggest that health numeracy is strongly related to general numeracy but that the 2 constructs may not be the same. Further research is needed into how different aspects of general numeracy and health numeracy translate into actual medical decisions.
    Medical Decision Making 07/2013; · 2.89 Impact Factor

Publication Stats

5k Citations
1,497.06 Total Impact Points

Institutions

  • 2014
    • Dana-Farber Cancer Institute
      Boston, Massachusetts, United States
  • 2010–2014
    • Duke University
      • Fuqua School of Business
      Durham, North Carolina, United States
    • Tufts Medical Center
      • Institute for Clinical Research and Health Policy Studies
      Boston, MA, United States
  • 2013
    • Fred Hutchinson Cancer Research Center
      Seattle, Washington, United States
  • 2002–2013
    • Concordia University–Ann Arbor
      Ann Arbor, Michigan, United States
  • 2001–2013
    • University of Michigan
      • • Center for Bioethics and Social Sciences in Medicine
      • • Department of Radiation Oncology
      • • Department of Internal Medicine
      • • Department of Psychology
      • • Division of General Medicine
      Ann Arbor, MI, United States
    • Penn State Hershey Medical Center and Penn State College of Medicine
      • Department of Humanities
      Hershey, Pennsylvania, United States
  • 2012
    • Stony Brook University
      • Health Sciences Center
      Stony Brook, NY, United States
  • 2011–2012
    • Grand Valley State University
      • Department of Psychology
      Grand Rapids, Michigan, United States
  • 2008–2011
    • Bryant University
      Smithfield, Rhode Island, United States
    • U.S. Department of Veterans Affairs
      Washington, Washington, D.C., United States
  • 2003–2009
    • Université de Montréal
      • Department of Health Administration
      Montréal, Quebec, Canada
    • Johns Hopkins University
      • Division of General Internal Medicine
      Baltimore, MD, United States
  • 1995–2009
    • University of Pennsylvania
      • • Department of Medicine
      • • Center for Clinical Epidemiology and Biostatistics
      • • Department of Psychology
      Philadelphia, Pennsylvania, United States
  • 2000–2008
    • Carnegie Mellon University
      • • Department of Social and Decision Sciences
      • • Department of Engineering and Public Policy
      Pittsburgh, PA, United States
    • Queen Elizabeth Hospital Birmingham
      Birmingham, England, United Kingdom
  • 2007
    • National Institutes of Health
      • Division of Cancer Control and Population Sciences
      Bethesda, MD, United States
  • 2005
    • McGill University
      Montréal, Quebec, Canada
  • 1997–2005
    • Hospital of the University of Pennsylvania
      • • Department of Psychiatry
      • • Department of Biostatistics and Epidemiology
      Philadelphia, Pennsylvania, United States
    • University of the Sciences in Philadelphia
      Philadelphia, Pennsylvania, United States
  • 1993–2005
    • University of Pittsburgh
      • School of Medicine
      Pittsburgh, PA, United States
  • 2004
    • University of Chicago
      Chicago, Illinois, United States
    • University of Toledo
      • Division of General Internal Medicine
      Toledo, OH, United States
  • 2000–2001
    • Christiana Care Health System
      Wilmington, Delaware, United States
  • 1997–1998
    • University of Maryland, Baltimore
      • Department of Pediatrics
      Baltimore, Maryland, United States
  • 1995–1996
    • Minneapolis Veterans Affairs Hospital
      Minneapolis, Minnesota, United States