Andrew J Epstein

Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania, United States

Are you Andrew J Epstein?

Claim your profile

Publications (75)442.71 Total impact

  • [Show abstract] [Hide abstract]
    ABSTRACT: Based on randomized evidence, expert guidelines in 2011 endorsed shorter, hypofractionated whole breast irradiation (WBI) for selected patients with early-stage breast cancer and permitted hypofractionated WBI for other patients. To examine the uptake and costs of hypofractionated WBI among commercially insured patients in the United States. Retrospective, observational cohort study, using administrative claims data from 14 commercial health care plans covering 7.4% of US adult women in 2013, we classified patients with incident early-stage breast cancer treated with lumpectomy and WBI from 2008 and 2013 into 2 cohorts: (1) the hypofractionation-endorsed cohort (n = 8924) included patients aged 50 years or older without prior chemotherapy or axillary lymph node involvement and (2) the hypofractionation-permitted cohort (n = 6719) included patients younger than 50 years or those with prior chemotherapy or axillary lymph node involvement. Hypofractionated WBI (3-5 weeks of treatment) vs conventional WBI (5-7 weeks of treatment). Use of hypofractionated and conventional WBI, total and radiotherapy-related health care expenditures, and patient out-of-pocket expenses. Patient and clinical characteristics included year of treatment, age, comorbid disease, prior chemotherapy, axillary lymph node involvement, intensity-modulated radiotherapy, practice setting, and other contextual variables. Hypofractionated WBI increased from 10.6% (95% CI, 8.8%-12.5%) in 2008 to 34.5% (95% CI, 32.2%-36.8%) in 2013 in the hypofractionation-endorsed cohort and from 8.1% (95% CI, 6.0%-10.2%) in 2008 to 21.2% (95% CI, 18.9%-23.6%) in 2013 in the hypofractionation-permitted cohort. Adjusted mean total health care expenditures in the 1 year after diagnosis were $28 747 for hypofractionated and $31 641 for conventional WBI in the hypofractionation-endorsed cohort (difference, $2894; 95% CI, $1610-$4234; P < .001) and $64 273 for hypofractionated and $72 860 for conventional WBI in the hypofractionation-permitted cohort (difference, $8587; 95% CI, $5316-$12 017; P < .001). Adjusted mean total 1-year patient out-of-pocket expenses were not significantly different between hypofractionated vs conventional WBI in either cohort. Hypofractionated WBI after breast conserving surgery increased among women with early-stage breast cancer in 14 US commercial health care plans between 2008 and 2013. However, only 34.5% of patients with hypofractionation-endorsed and 21.2% with hypofractionation-permitted early-stage breast cancer received hypofractionated WBI in 2013.
    JAMA. 12/2014;
  • International journal of radiation oncology, biology, physics 08/2014; 89(5):954–957. · 4.59 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The effect of care setting on value of colon cancer care is unknown.
    Cancer 06/2014; · 5.20 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The Centers for Medicare and Medicaid Services require hospitals performing carotid artery stenting (CAS) to recertify the quality of their programs every 2 years, but currently this involves no explicit comparisons of postprocedure mortality across hospitals. Hence, the current recertification process may fail to identify hospitals that are performing poorly in relation to peer institutions. Our objective was to compare risk-standardized procedural outcomes across US hospitals that performed CAS and to identify hospitals with statistically high postprocedure mortality rates.
    Circulation Cardiovascular Quality and Outcomes 06/2014; · 5.66 Impact Factor
  • Andrew J. Epstein, Jonathan D. Ketcham
    [Show abstract] [Hide abstract]
    ABSTRACT: Patients rely on physicians to act as their agents when prescribing medications, yet the efforts of pharmaceutical manufacturers and prescription drug insurers may alter this agency relationship. We evaluate how formularies, and the use of information technology (IT) that provides physicians with formulary information, influence prescribing. We combine data from a randomized experiment of physicians with secondary data to eliminate bias due to patient, physician, drug, and insurance characteristics. We find that when given formulary IT, physicians' prescribing decisions are influenced by formularies far more than by pharmaceutical firms' detailing and sampling. Without IT, however, formularies' effects are much smaller.
    The RAND Journal of Economics 06/2014; 45(2). · 1.49 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Background: In 2008, the US Congress enacted the Paul Wellstone and Pete Domenici Mental Health Parity and Addiction Equity Act (MHPAEA) requiring insurers to equalize private insurance coverage for mental health and substance use disorder services with coverage for general medical services. Objective: To examine the effects of MHPAEA on substance use disorder treatment. Study Design: We used a difference-in-differences design to compare changes in outcomes among plan enrollees in the years before and after implementation of federal parity (2009-2010) with changes in outcomes among a comparison group of enrollees previously covered by state substance use disorder parity laws. Methods: Insurance claims data from Aetna Inc health plans in 10 states with state parity laws were used to compare outcomes for plan enrollees in fully insured and self-insured health plans (N = 298,339). Results: In the first year of implementation, we found that federal parity did not lead to changes in the proportion of enrollees using substance use disorder treatment. We did find a modest increase in spending on substance use disorder treatment per enrollee ($9.99, 95% confidence interval, 2.54-18.21), but no significant change in identification, treatment initiation, or treatment engagement. Conclusions: Inclusion of substance use disorder services in the federal parity law did not result in substantial increases in health plan spending. It will be critical to study results for year 2 after regulations affecting the management of care (eg, utilization review, network access) take effect.
    The American journal of managed care 01/2014; 20(1):76-82. · 2.12 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Objectives Variation in care within and across geographic areas remains poorly understood. The goal of this article was to examine whether physician social networks—as defined by shared patients—are associated with rates of complications after radical prostatectomy. Methods In five cities, we constructed networks of physicians on the basis of their shared patients in 2004-2005 Surveillance, Epidemiology and End Results-Medicare data. From these networks, we identified subgroups of urologists who most frequently shared patients with one another. Among men with localized prostate cancer who underwent radical prostatectomy, we used multilevel analysis with generalized linear mixed-effect models to examine whether physician network structure—along with specific characteristics of the network subgroups—was associated with rates of 30-day and late urinary complications, and long-term incontinence after accounting for patient-level sociodemographic, clinical factors, and urologist patient volume. Results Networks included 2677 men in five cities who underwent radical prostatectomy. The unadjusted rate of 30-day surgical complications varied across network subgroups from an 18.8 percentage-point difference in the rate of complications across network subgroups in city 1 to a 26.9 percentage-point difference in city 5. Large differences in unadjusted rates of late urinary complications and long-term incontinence across subgroups were similarly found. Network subgroup characteristics—average urologist centrality and patient racial composition—were significantly associated with rates of surgical complications. Conclusions Analysis of physician networks using Surveillance, Epidemiology and End Results-Medicare data provides insight into observed variation in rates of complications for localized prostate cancer. If validated, such approaches may be used to target future quality improvement interventions.
    Value in Health 01/2014; · 2.19 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: We conducted a national survey of general surgeons to address the association between surgeon characteristics and the tendency to recommend surgery. We used a web-based survey with 25 hypothetical clinical scenarios with clinical equipoise regarding the decision to operate. The respondent-level tendency to operate (TTO) score was calculated as the average score over the 25 scenarios. Surgical volume was based on self-report. Linear regression models were used to evaluate the associations between TTO, other covariates of interest, and surgical volume. There were 907 respondents. The mean surgical TTO was 3.05 ± .43. Surgeons had significantly lower TTO scores when responding to questions within their area of practice (P < .0001). There was no association between TTO and malpractice concerns, financial incentives, or compensation structure. Surgeons recommend intervention far less frequently within their area of specialization. Malpractice concerns, volume, and financial compensation do not significantly affect surgical decision making.
    American journal of surgery 12/2013; 206(6):970-8. · 2.36 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The goal of medical education is the production of a workforce capable of improving the health and health care of patients and populations, but it is hard to use a goal that lofty, that broad, and that distant as a standard against which to judge the success of schools or training programs or particular elements within them. For that reason, the evaluation of medical education often focuses on elements of its structure and process, or on the assessment of competencies that could be considered intermediate outcomes. These measures are more practical because they are easier to collect, and they are valuable when they reflect activities in important positions along the pathway to clinical outcomes. But they are all substitutes for measuring whether educational efforts produce doctors who take good care of patients.The authors argue that the evaluation of medical education can become more closely tethered to the clinical outcomes medical education aims to achieve. They focus on a specific clinical outcome-maternal complications of obstetrical delivery-and show how examining various observable elements of physicians' training and experience helps reveal which of those elements lead to better outcomes. Does it matter where obstetricians trained? Does it matter how much experience they have? Does it matter how good they were to start? Each of these questions reflects a component of the production of a good obstetrician and, most important, defines a good obstetrician as one whose patients in the end do well.
    Academic medicine: journal of the Association of American Medical Colleges 11/2013; · 2.34 Impact Factor
  • JAMA The Journal of the American Medical Association 10/2013; 310(14):1501-1502. · 29.98 Impact Factor
  • Value in Health 05/2013; 16(3):A211. · 2.19 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: To assess the association between obstetricians' years of experience after training and the maternal complications of their patients during their first 40 years of post-residency practice. Retrospective cohort analysis. Obstetrical discharges from acute care hospitals in Florida and New York between academic years 1992 and 2009. 6 704 311 deliveries performed by 5175 obstetricians. Three composite measures of maternal complication rates per physician year from vaginal and cesarean births separately and combined, adjusted for secular trends. Obstetricians' maternal complication rates declined during the first three decades after completion of residency. The improvement was largest in the first decade and diminished thereafter. For all deliveries, the change was -0.21 (95% confidence interval -0.23 to -0.19) percentage points per year in the first decade, -0.11 (-0.13 to -0.09) percentage points per year in the second decade, and -0.05 (-0.08 to -0.01) percentage points in the third decade (P<0.001 for second to first decade comparison; P=0.001 for third to second decade comparison). The patterns were comparable for cesarean deliveries and vaginal deliveries and across several sensitivity checks. Among obstetricians practicing in Florida and New York, those with more years of experience had fewer maternal complications. This association persisted over the first three decades of practice but diminished in magnitude.
    BMJ (online) 03/2013; 346:f1596. · 17.22 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: IMPORTANCE As many surgical procedures have undergone a transition from a standard, open surgical approach to a minimally invasive one in the past 2 decades, the diffusion of minimally invasive surgery may have had sizeable but overlooked effects on medical expenditures and worker productivity. OBJECTIVE To examine the impact of standard vs minimally invasive surgery on health plan spending and workplace absenteeism for 6 types of surgery. DESIGN Cross-sectional regression analysis. SETTING National health insurance claims data and matched workplace absenteeism data from January 1, 2000, to December 31, 2009. PARTICIPANTS A convenience sample of adults with employer-sponsored health insurance who underwent either standard or minimally invasive surgery for coronary revascularization, uterine fibroid resection, prostatectomy, peripheral revascularization, carotid revascularization, or aortic aneurysm repair. MAIN OUTCOMES AND MEASURE Health plan spending and workplace absenteeism from 14 days before through 352 days after the index surgery. RESULTS There were 321 956 patients who underwent surgery; 23 814 were employees with workplace absenteeism data. After multivariable adjustment, mean health plan spending was lower for minimally invasive surgery for coronary revascularization (-$30 850; 95% CI, -$31 629 to -$30 091), uterine fibroid resection (-$1509; 95% CI, -$1754 to -$1280), and peripheral revascularization (-$12 031; 95% CI, -$15 552 to -$8717) and higher for prostatectomy ($1350; 95% CI, $611 to $2212) and carotid revascularization ($4900; 95% CI, $1772 to $8370). Undergoing minimally invasive surgery was associated with missing significantly fewer days of work for coronary revascularization (mean difference, -37.7 days; 95% CI, -41.1 to -34.3), uterine fibroid resection (mean difference, -11.7 days; 95% CI, -14.0 to -9.4), prostatectomy (mean difference, -9.0 days; 95% CI, -14.2 to -3.7), and peripheral revascularization (mean difference, -16.6 days; 95% CI, -28.0 to -5.2). CONCLUSIONS AND RELEVANCE For 3 of 6 types of surgery studied, minimally invasive procedures were associated with significantly lower health plan spending than standard surgery. For 4 types of surgery, minimally invasive procedures were consistently associated with significantly fewer days of absence from work.
    JAMA surgery. 03/2013;
  • Emily Carter Paulson, Xiaoying Fu, Andrew J Epstein
    [Show abstract] [Hide abstract]
    ABSTRACT: BACKGROUND: We sought to identify colon cancer patients within the Veterans Affairs (VA) system who experienced lengthy wait times for surgery or chemotherapy. We looked specifically at the relationship between location of treatment and timing of care. METHODS: We performed a retrospective cohort study of 4635 patients diagnosed with colon cancer in the VA Health System during 2002-2010 and treated with surgery followed by chemotherapy. We used VA administrative databases, including the VA Outpatient Clinic, Patient Treatment, and Fee Basis inpatient and outpatient files. Time from diagnosis to surgery and time from surgery to initiation of chemotherapy were the primary outcome measures. RESULTS: Patients who required referral to a hospital different from their home VA facility for surgery experienced delays in surgical intervention compared with patients treated at their home VA medical center. For patients referred outside of the VA system, this delay was almost 2 wk (13.5 d, P < 0.001). When these patients then went to another hospital for chemotherapy, they experienced further delays in care. Patients treated surgically outside the VA system who returned to the VA system for chemotherapy were more likely to initiate chemo >8 wk following surgery (OR 1.69, P = 0.01). The average adjusted time from surgery to chemotherapy for these patients compared with those treated wholly within the VA system was 11.4 d (P = 0.003). CONCLUSIONS: VA patients who require treatment at multiple hospitals for colon cancer, especially those who require surgery outside of the VA system, are more likely to experience delays in care compared with patients treated at a single hospital.
    Journal of Surgical Research 03/2013; · 2.02 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: PURPOSE: National attention has focused on whether urology-radiation oncology practice integration - known as integrated prostate cancer centers (IPCCs) - contributes to use of intensity-modulated radiation therapy (IMRT), a common and expensive treatment for prostate cancer. METHODS: We examined prostate cancer treatment patterns pre- and post-conversion of a urology practice to an IPCC in July, 2006. Using the SEER-Medicare database, we identified patients age ≥ 65 years diagnosed in one state-wide registry with non-metastatic prostate cancer between 2004 and 2007 and classified patients into 3 groups: (1) those seen by IPCC physicians (exposure group); (2) those living in the same hospital referral region (HRR) and not seen by IPCC physicians (HRR-control group); and (3) those living elsewhere in the state (state-control group). We compared changes in treatment among the 3 groups, adjusting for patient, clinical, and socio-economic factors. RESULTS: Compared with the 8.1 percentage point (ppt) increase in adjusted IMRT use in the state-control group, IMRT increased 20.3 ppts (95% confidence interval [CI] 13.4, 27.1) in the IPCC group and 19.2 ppts (95% CI 9.6, 28.9) in the HRR-control group. Androgen-deprivation therapy (ADT), for which Medicare reimbursement declined sharply, decreased similarly in the IPCC and HRR-control groups. Prostatectomy declined significantly in the IPCC group. CONCLUSIONS: Coincident with the conversion of a urology group practice to an IPCC, we observed increases in IMRT and decreases in ADT among patients seen by IPCC physicians and those seen in the surrounding healthcare market that were not observed in the remainder of the state.
    The Journal of urology 02/2013; · 3.75 Impact Factor
  • Andrew J Epstein, David A Asch, Colleen L Barry
    [Show abstract] [Hide abstract]
    ABSTRACT: Concerns about the pharmaceutical industry's influence in academic medical centers and on medical education have led many medical schools and teaching hospitals to adopt conflict-of-interest (COI) policies. Although the restrictiveness of these policies differs, the goal is the same: to shield physicians-in-training from the persuasive aspects of pharmaceutical promotion. But do these policies work? This Issue Brief examines how COI policies affect the prescribing patterns of antidepressants, one of the most heavily promoted drug classes in the past decade. As such, it provides the first empirical evidence of the effects of COI policies in residency on the subsequent prescribing patterns of practicing physicians.
    LDI issue brief 01/2013; 18(3):1-4.
  • Andrew Epstein, Sean Nicholson, David A. Asch
    [Show abstract] [Hide abstract]
    ABSTRACT: Our understanding of the determinants of physician skill and the extent to which skill is valued in the marketplace is superficial. Using a large, detailed panel of new obstetricians, we find that, even though physicians’ maternal complication rates improve steadily with years of practice, initial skill (as measured by performance in a physician’s first year of practice) explains most of the variation in physician performance over time. At the same time, we find that the trajectories of new physicians’ delivery volume develop in a way partially consistent with Bayesian learning about physician quality. In particular, as physicians gain experience, their volume becomes increasingly sensitive to the information in their accumulated prior.Institutional subscribers to the NBER working paper series, and residents of developing countries may download this paper without additional charge at
  • [Show abstract] [Hide abstract]
    ABSTRACT: BACKGROUND:: Academic medical institutions have instituted conflict of interest (COI) policies in response to concerns about pharmaceutical industry influence. OBJECTIVE:: To determine whether exposure to COI policies during psychiatry residency training affects psychiatrists' antidepressant prescribing patterns after graduation. RESEARCH DESIGN:: We used 2009 physician-level national administrative prescribing data from IMS Health for 1652 psychiatrists from 162 residency programs. We used difference-in-differences estimation to compare antidepressant prescribing based on graduation before (2001) or after (2008) COI policy adoption across residency program groups with maximally, moderately, and minimally restrictive COI policies. The primary outcomes were shares of psychiatrists' prescribing of heavily promoted, brand reformulated, and brand antidepressants. RESULTS:: Rates of prescribing heavily promoted, brand reformulated, and brand antidepressants in 2009 were lower among post-COI graduates than pre-COI graduates at all levels of COI restrictiveness. However, differences between pre-COI and post-COI graduates' prescribing of heavily promoted medications were larger for maximally restrictive programs than both minimally restrictive programs [-4.3 percentage points; 95% confidence interval (CI), -7.0, -1.6] and moderately restrictive programs (-3.6 percentage points; 95% CI, -6.2, -1.1). The difference in prescribing reformulations was larger for maximally restrictive programs than minimally restrictive programs (-3.0 percentage points; 95% CI, -5.3, -0.7). Results were consistent for prescribing of brand drugs. CONCLUSIONS:: This study provides the first empirical evidence of the effects of COI policies. Our results suggest that COI policies can help inoculate physicians against persuasive aspects of pharmaceutical promotion. Further research should assess whether these policies affect other drug classes and physician specialties similarly.
    Medical care 11/2012; · 3.24 Impact Factor
  • Source
    Value in Health 06/2012; 15(4):A11. · 2.19 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Implantable cardioverter-defibrillators and cardiac resynchronization therapy-defibrillators (ICD/CRT-Ds) are evidence-based preventative treatments for many patients with heart failure (HF), yet large numbers of eligible patients remain untreated. It is uncertain if localities with more frequent ICD/CRT-D use have had better rates of HF survival. To determine if US Hospital Referral Regions (HRRs) with larger increases in the rate of ICD/CRT-D utilization during 2002 to 2007 also had commensurate increases in HF survival. Retrospective cohort. Medicare beneficiaries age 66 to 80 nonelectively hospitalized for HF from 2002 to 2007. Each HRR's annual ICD/CRT-D rate was estimated from the cohort's Medicare procedure claims. Survival duration was determined from Medicare mortality records. HRR-year-level panel regression models were estimated to assess whether an HRR's ICD/CRT-D rate predicted HF survival, adjusting for baseline differences in survival across HRRs and secular trends. A total of 883,002 HF patients were propensity-score matched within HRR across 2002 to 2007. Across HRRs, growth in ICD/CRT-D use among such patients varied from 1 to 12 percentage points. Regression models indicated that a 1 percentage point increase in an HRR's ICD/CRT-D utilization among hospitalized HF patients was associated with an increase in 1-year survival of 0.12% [95% confidence interval (CI), 0.03%-0.21%, P=0.009] and with a 0.26% increase in HF survival at 2 years (95% CI, 0.14%-0.37%, P<0.001). Localities with greater increases in ICD/CRT-D utilization from 2002 to 2007 also had greater improvements in HF survival. Areas with persistently low ICD/CRT-D use may be good targets for programs designed to increase the evidence-based use of defibrillators.
    Medical care 01/2012; 50(1):10-7. · 3.24 Impact Factor

Publication Stats

897 Citations
442.71 Total Impact Points


  • 2014
    • Hospital of the University of Pennsylvania
      Philadelphia, Pennsylvania, United States
  • 2004–2014
    • University of Pennsylvania
      • Center for Health Equity Research
      Philadelphia, Pennsylvania, United States
  • 2004–2010
    • Yale University
      • • Section of Cardiovascular Medicine
      • • School of Public Health
      • • Department of Internal Medicine
      New Haven, Connecticut, United States
  • 2005–2009
    • Yale-New Haven Hospital
      • Department of Laboratory Medicine
      New Haven, Connecticut, United States
  • 2006–2008
    • Arizona State University
      Phoenix, Arizona, United States
  • 2007
    • Icahn School of Medicine at Mount Sinai
      Manhattan, New York, United States