UCSF and MDEpiNet Coordinating Center are collaborating to study the adoption pattern of cardiac surgical procedures (e.g. TAVR), percutaneous coronary interventions, pacemaker, and defibrillator implantations. The New York State discharge data is used for the projects and California all-payer data will be acquired.

Piloting Women’s Health Registries in EHRs – the ULTRA Study at UCSF:
ULTRA study is a pilot study for Uterine Fibroid led by Dr. Vanessa Jacoby (https://fibroids.ucsf.edu/). This collaborative effort aims to help patients and doctors understand how the treatment changes fibroid symptoms, affects fertility and pregnancy, and impacts the need for additional fibroid treatment in the future.

In a post-market observational cohort study, ULTRA assesses safety and effectiveness of women undergoing AcessaTM treatment. The project recruits women through voluntary referrals from clinicians and engaged clinical sites in a multicenter study. UCSF follows up for 3 years after surgery. The outcomes of the study are operative morbidity, change in fibroid, treatment failure, and pregnancy outcomes. The patient population is queried every 6 months with questionnaires to collect PRO, and medical records are obtained for follow-up imaging. ULTRA collaborates with COMPARE-UF for harmonization of outcome assessments (CRFs and questionnaires) as well as WHT-CRN for opportunities to pilot new data elements in an ongoing post-market device study.

The collaborators created a core minimum data set for surgeries and procedures to treat uterine fibroids with a focus on the use of medical devices during these procedures. The minimum data set draws from data collected at the time of surgery (i.e. intraoperative outcomes and events). These data points are structured to serve as a template for a case report form in the research environment (e.g. post-market surveillance) and/or to embed within the electronic health record for use in a general clinical care workflow.


EVAR in the treatment of ruptured abdominal aortic aneurysms (rAAA) is an MDEpiNet Coordinating Center -supported clinical trial project nested under the ICVR. UAB has been established as an MDEpiNet collaborator for this and future projects.

The aim of this project is to evaluate the safety and effectiveness of EVAR devices used to treat rAAA (compared to open rAAA repair) by international evaluation in the existing ICVR registries. The project is intended to provide manufacturers of currently approved EVAR devices (Cook, Endologix, Gore, Medtronic) with real-world data allowing them to understand better how their devices perform in the setting of ruptured aneurysm.

As EVAR is now the primary treatment for AAA, there is increasing importance to evaluate these parameters from a surgical quality and outcome perspective. Additionally, these data could be used as support for a regulatory application to modify the labeling of the endovascular grafts evaluated in this study with respect to rAAA.

The central purpose of this project is to evaluate in-hospital mortality after EVAR for rAAA in a multinational registry collaboration using mortality associated with standard open repair to establish performance goals. Given that untreated rAAA carries a mortality approaching 100%, the project will focus specifically on survival to discharge. Further, the long-term safety and effectiveness of these EVAR devices has been extensively studied and established for elective AAA repair. The major endpoint for this project is improving initial survival based on the improvement of outcomes after rAAA repair.

This figure (Beck AW et al. 2019) demonstrates the modality of repair for intact (A) and ruptured (B) aneurysms internationally27. EVAR has become the dominant method of repair for elective aneurysms internationally. In the United States, the proportion of ruptured aneurysms treated with EVAR exceeds 50%, despite a lack of formal evaluation in this clinical setting.

Comparative effectiveness of peripheral vascular intervention versus surgical bypass for critical limb ischemia in the Vascular Study Group of Greater New York

Journal Journal of Vascular Surgery
Authors Meltzer AJ, Sedrakyan A, Isaacs A, Connolly PH, Schneider DB
Year Published 2016
Link to publication



In this study, the effectiveness of peripheral vascular intervention (PVI) was compared with surgical bypass grafting (BPG) for critical limb ischemia (CLI) in the Vascular Study Group of Greater New York (VSGGNY).


Patients undergoing BPG or PVI for CLI at VSGGNY centers (2011-2013) were included. The Society for Vascular Surgery objective performance goals for CLI were used to directly compare the safety and effectiveness of PVI and BPG. Propensity score matching was used for risk-adjusted comparisons of PVI with BPG.


A total of 414 patients (268 PVI, 146 BPG) were treated for tissue loss (69%) or rest pain (31%). Patients undergoing PVI were more likely to have tissue loss (74.6% vs 57.5%; P < .001) and comorbidities such as diabetes (69.3% vs 57.5%; P = .02), heart failure (22% vs 13.7%; P = .04), and severe renal disease (13.1% vs 4.1%; P = .004). No significant differences were found between the groups across a panel of safety objective performance goals. In unadjusted analyses at 1 year, BPG was associated with higher rates of freedom from reintervention, amputation, or restenosis (90.4% vs 81.7%; P = .02) and freedom from reintervention or amputation (92.5% vs 85.8%, P = .045). After propensity score matching, PVI was associated with improved freedom from major adverse limb events and postoperative death at 1 year (95.6% vs 88.5%; P < .05).


By unadjusted comparison, early reintervention and restenosis are more prevalent with PVI. However, risk-adjusted comparison underscores the safety and effectiveness of PVI in the treatment of CLI.

Regulator loss functions and hierarchical modeling for safety decision making

Journal Medical Decision Making
Authors Hatfield LA, Baugh CM, Azzone V, Normand S-LT
Year Published 2017
Link to publication



Regulators must act to protect the public when evidence indicates safety problems with medical devices. This requires complex trade-offs among risks and benefits, which conventional safety surveillance methods do not incorporate.


To combine explicit regulator loss functions with statistical evidence on medical device safety signals to improve decision-making.


In the Hospital Cost and Utilization Project National Inpatient Sample, we select pediatric inpatient admissions and identify adverse medical device events (AMDEs). We fit hierarchical Bayesian models to the annual hospital-level AMDE rates, accounting for patient and hospital characteristics. These models produce expected AMDE rates (a safety target), against which we compare the observed rates in a test year to compute a safety signal. We specify a set of loss functions that quantify the costs and benefits of each action as a function of the safety signal. We integrate the loss functions over the posterior distribution of the safety signal to obtain the posterior (Bayes) risk; the preferred action has the smallest Bayes risk. Using simulation and an analysis of AMDE data, we compare our minimum-risk decisions to a conventional Z score approach for classifying safety signals.


The two rules produced different actions for nearly half of hospitals (45%). In the simulation, decisions that minimize Bayes risk outperform Z score-based decisions, even when the loss functions or hierarchical models are misspecified.


Our method is sensitive to the choice of loss functions; eliciting quantitative inputs to the loss functions from regulators is challenging.


A decision theoretic approach to acting on safety signals is potentially promising, but requires careful specification of loss functions in consultation with subject matter experts.

Nursing Home Use After Implantable Cardioverter-Defibrillator Implantation in Older Adults: Results from the National Cardiovascular Data Registry

Journal Journal of the American Geriatrics Society
Authors Kramer DB, Reynolds MR, Normand SL, Parzynski CS, Spertus JA, Mor V, Mitchell SL.
Year Published 2017
Link to publication


To evaluate the incidence and characteristics of nursing home (NH) use after implantable cardioverter-defibrillator (ICD) implantation.


Cohort study.


Medicare beneficiaries in the National Cardiovascular Data Registry-ICD Registry.


Individuals aged 65 and older receiving ICDs between January 1, 2006, and March 31, 2010 (N = 192,483).


Proportion of ICD recipients discharged to NHs directly after device placement, cumulative incidence of long-term NH admission, and factors associated with immediate discharge to a NH and time to long-term NH admission.


Over 4 years, 40.6% of the cohort died, and 35,939 (18.7%) experienced at least one NH admission, including 4.0% directly discharged to a NH after ICD implantation and 2.8% admitted to long-term NH care during follow-up. The cumulative incidence of long-term NH admission, accounting for the competing risk of death, was 1.7% at 1 year, 3.8% at 3 years, and 4.6% at 4 years; 20.1% of individuals admitted to a NH died there. Factors most strongly associated with direct NH discharge and time to long-term NH care were older age (adjusted odds ratio (AOR) = 2.09, 95% confidence interval (CI) = 2.01-2.17 per 10-year increment; adjusted hazard ratio (AHR) = 1.88, 95% CI = 1.80-1.97, respectively), dementia (AOR = 2.60, 95% CI = 2.25-3.01; AHR = 2.50, 95% CI = 2.14-2.93, respectively), and Medicare Part A claim for NH stay in prior 6 months (AOR = 3.96, 95% CI = 3.70-4.25; AHR = 2.88, 95% CI = 2.65-3.14, respectively).


Nearly one in five individuals are admitted to NHs over a median of 1.6 years of follow-up after ICD implantation. Understanding these outcomes may help inform the clinical care of these individuals.