Rui Zhang, Ph.D.2,5, Genevieve B. Melton-Meaux, M.D., Ph.D.2,5

Slides:



Advertisements
Similar presentations
Clinical Trials What Are They and When Are They Right For You? Maura N. Dickler Assistant Attending Physician Breast Cancer Medicine Service Memorial Sloan-Kettering.
Advertisements

Understanding Statistics in Research Articles Elizabeth Crabtree, MPH, PhD (c) Director of Evidence-Based Practice, Quality Management Assistant Professor,
Case-Control Studies (Retrospective Studies). What is a cohort?
Systems Based Practice and Practice Based Learning. How to teach? How to evaluate? Deborah J. DeWaay M.D. Hospitalist/Clerkship Director, General Internal.
Development and Testing of a Risk Assessment Model for Venous Thrombosis in Medical Inpatients: The Medical Inpatients and Thrombosis (MITH) Study Score.
® Introduction Low Back Pain and Physical Function Among Different Ethnicities Adelle A Safo, Sarah Holder DO, Sandra Burge PhD The University of Texas.
NCHS Data – Strengths and Weaknesses from the NHLBI Perspective Paul Sorlie, Ph.D. Chief, Epidemiology Branch National Heart, Lung, and Blood Institute.
Pharmacoepidemiology: Goals and Methods Sean Hennessy, PharmD, PhD Assistant Professor of Epidemiology & Pharmacology Center for Clinical Epidemiology.
Cohort Studies Hanna E. Bloomfield, MD, MPH Professor of Medicine Associate Chief of Staff, Research Minneapolis VA Medical Center.
Presented by Martin H. Cohen, M.D. at the 27 July 2004 meeting of the Oncologic Drugs Advisory Committee.
Electronic Medical Record Use and the Quality of Care in Physician Offices National Conference on Health Statistics August 17, 2010 Chun-Ju (Janey) Hsiao,
Marshall University School of Medicine Department of Biochemistry and Microbiology BMS 617 Lecture 12: Multiple and Logistic Regression Marshall University.
The influence of Breast Cancer Pay for Performance Initiatives on breast cancer survival and performance measures: a pilot study in Taiwan Raymond NC Kuo,
 Definitions  Goals of automation in pharmacy  Advantages/disadvantages of automation  Application of automation to the medication use process  Clinical.
Agency for Healthcare Research and Quality Advancing Excellence in Health Care Improving Administrative Data for Public Reporting Anne Elixhauser.
Author Disclosures Differences in Implantation-Related Adverse Events Between Men and Women Receiving ICD Therapy for Primary Prevention Differences in.
Computers in Healthcare Jinbo Bi Department of Computer Science and Engineering Connecticut Institute for Clinical and Translational Research University.
File #1 Table of Contents Introduction Presenter’s Notes Oncology Rehabilitation: Web-based Learning for Physical Therapists Who Provide Rehabilitation.
Survey of Medical Informatics CS 493 – Fall 2004 October 11, 2004 V. “Juggy” Jagannathan.
Instructor: Mary “Stela” Gallegos, ABD, (RT), (R), (M) Seminar 4.
Understanding Medical Articles and Reports Linda Vincent, MPH UCSF Breast SPORE Advocate September 24,
Tim Wiemken PhD MPH CIC Assistant Professor Division of Infectious Diseases University of Louisville, Kentucky Confounding.
Flowsheet Analysis Design Concepts Introduction Acknowledgements Proposed Design By studying PICU flowsheets in the context of their actual use, we have.
The Role of Residential Segregation in Disparity Research: A Case Example of ADHD Diagnosis and Treatment Dinci Pennap, MPH, 1 Mehmet Burcu, MS, 1 Daniel.
A Claims Database Approach to Evaluating Cardiovascular Safety of ADHD Medications A. J. Allen, M.D., Ph.D. Child Psychiatrist, Pharmacologist Global Medical.
A collaborative scientific program E gyptian H ypertension S ociety.
C-1 Pegfilgrastim (Neulasta  ) Oncologic Drugs Advisory Committee Pediatric Subcommittee October 20, 2005 Amgen Inc.
Course: Research in Biomedicine and Health III Seminar 5: Critical assessment of evidence.
Biomedical Research. What is biomedical research?  It is the area of science devoted to: The study of life processes The prevention & treatment of disease.
Biostatistics Board Review Parul Chaudhri, DO Family Medicine Faculty Development Fellow, UPMC St Margaret March 5, 2016.
BREAST CANCER BY STAGE OF DISEASE AT DIAGNOSIS, CENTRAL OKLAHOMA Arthur Owora, MPH; Aaron Wendelboe, PhD; David Thompson, PhD; Janis Campbell, PhD The.
Bussara Sukpanichnant, Human Subject Protection Office, USAMD-AFRIMS Unanticipated Problems 15 th FERCAP International Conference 24 Nov 15 Nagasaki, Japan.
“Reading and commenting papers” (Scientific English) Alexis Descatha INSERM, UMS UVSQ- Unité de pathologie professionnelle, Garches.
Summary of “A randomized trial of standard versus intensive blood-pressure control” The SPRINT Research Group, NEJM, DOI: /NEJMoa Downloaded.
Differential Effect of Glaucoma Medications on Dry Eye Incidence Scott D. Walter, MD, MS, Sarah R. Wellik, MD, William J. Feuer, MS, Anat Galor, MD, MSPH.
8/12/2010 Nursing 1. 8/12/2010 Nursing2 Associate Professor Family and Community Medicine Department King Saud University By Epidemiological Methods.
Public Health February 2017
Constructing a Predictor to Identify Drug and Adverse Event Pairs
Contact information Does daily tobacco smoking affect outcomes after microdecompression for degenerative central lumbar spinal stenosis? - A multicenter.
x-squared= p= /10 patients had no pathology results
. Troponin limit of detection plus cardiac risk stratification scores for the exclusion of myocardial infarction and 30-day adverse cardiac events in ED.
Evidence-based Medicine
Patient Registries and Health Outcomes in Diabetes: A Retrospective Study Nipa Shah, MD1; Fern Webb, PhD1; Liane Hannah, BSH1; Carmen Smotherman, MS2;
Qualitative research: an overview
Mortality Prognostic Model for Peripheral Arterial Disease
Presentation for the SCTR Scientific Retreat on Aging Related Research
Martha Carvour, MD, PhD March 2, 2017
Amanda L. Do, MPH1,2, Ruby Y. Wan, MS1,2, Robert W
Introduction to Clinical Pharmacy
Component 11: Configuring EHRs
Risk of post-operative stroke in patients with known extra-cranial carotid artery disease undergoing Non-Cardiac Surgery Heart and Vascular.
Basic Statistics Overview
Paula Braun Entrepreneur-in-Residence
Medication Use Pattern Mining for Childhood Pneumonia Using Six Year Inpatient Electronic Medical Records in a Shanghai Hospital, China Chunlei Tang, PhD1,2,3,
A SOAP Approach to Clinical Research Design and Analysis
Medical Laboratory Science
Improving Identification and Diagnosis of Hypertensive Patients Hiding in Plain Sight (HIPS) in Health Centers  Margaret Meador, MPH, C-PHI, Jerome A.
Latent TB Infection among Diabetic patients
Measures of Disease Occurrence
Megan Eguchi, MPh Sana karam, md, phd
Research Techniques Made Simple: Interpreting Measures of Association in Clinical Research Michelle Roberts PhD,1,2 Sepideh Ashrafzadeh,1,2 Maryam Asgari.
Effect of Sample size on Research Outcomes
Introduction to Basic Research Methods
Cancer is not a risk factor for bullous pemphigoid
NAACCR/IACR Combined Annual Conference 6/11/2019, Vancouver, Canada
When to Submit a Urine Specimen for Testing?
Medical Informatics and Explainable AI
Receipt of Adjuvant Endometrial Cancer Treatment According to Race NRG Oncology/Gynecologic Oncology Group (GOG) 210 Study Ashley Felix, PhD, MPH Assistant.
Jill S. Barnholtz-Sloan, PhD
Gregory Levin, FDA/CDER/OTS/OB/DBIII
Presentation transcript:

Rui Zhang, Ph.D.2,5, Genevieve B. Melton-Meaux, M.D., Ph.D.2,5 S27: A Study on Phenotyping and Visualizing Black Box Warnings of Antineoplastic Breast Cancer Medications Deyu Sun, Ph.D.1, Gopal Sarda, M.S.3, Steven Skube, M.D.2, Anne Blaes, M.D.4, Rui Zhang, Ph.D.2,5, Genevieve B. Melton-Meaux, M.D., Ph.D.2,5 1School of Statistics, 2Department of Surgery, 3Department of Computer Science, 4Department of Medicine, 5Institute for Health Informatics, University of Minnesota March 29, 2017 automated detection

The authors have nothing to disclose. Disclosure The authors have nothing to disclose.

Learning Objectives After participating in this activity, the learner should be better able to Gain a better understanding of phenotyping algorithms Explore the benefits of visualizing integrated knowledge 1

INTRODUCTION Phenotyping In medical contexts, phenotyping refers to classify patients with specific diseases and traits. E.g., adverse events caused by medication Phenotyping shows who the patient is, based on which physicians can decide what can be done to cure the patient Phenotyping provides more insight than ICD codes In this study, we focus on detecting most serious adverse events of two first-line breast cancer drugs: tamoxifen and docetaxel Q: lower case for all Phenotyping: not equal to Deviation: more about the characteristics, current state … … 2

INTRODUCTION Visualization Visualization uses graphic displays of health information that make sense and effectively communicate insight behind data Visualization in phenotyping helps render causality between exposure and outcome Visualization assists physicians integrate information Most current visualization systems focus on showing structured data alone, without integrating additional knowledge 3

INTRODUCTION Objectives We focus on detecting “potential” adverse events caused by docetaxel & tamoxifen We proposed a visualization prototype rendering potential adverse events associated with medication usage In this study, we focus on detecting most serious adverse events of two first-line breast cancer drugs: tamoxifen and docetaxel Q: lower case for all Phenotyping: not equal to Deviation: more about the characteristics, current state … … 4

METHODS Data Collection 4,084 patients diagnosed with breast cancer in University of Minnesota Medical Center (2010-16) Diagnosis codes (ICD-9, 10) Vital signs (e.g., blood pressure, oxygen saturation) Lab tests (e.g., white blood count, blood lactate concentration) Black box warnings are strictest drug associated adverse events put in the labeling of prescription drugs by the FDA WBC: ful name More e.g. 5

Extraction of Black Box Warnings METHODS Extraction of Black Box Warnings Medication Black Box Warnings docetaxel neutropenia bronchospasm febrile neutropenia anaphylactic shock infection fluid retention thrombocytopenia oedema stomatitis pleural effusion hypersensitivity dyspnoea at rest rash generalized cardiac tamponade erythema abdominal distension hypotension tamoxifen cerebrovascular accident endometrial adenocarcinoma pulmonary embolism sarcoma uterus Black box warnings are strictest drug associated adverse events put in the labeling of prescription drugs by the FDA What’s the relationship among adverse events? 6

Phenotyping Algorithms METHODS Phenotyping Algorithms Title: Hypo /ou/ Example : Flow chart: Hypo- vs. No hypo- Details: how to search, when & how to develop algorithms Physician: 4th year resident 7

Validation of Algorithms METHODS Validation of Algorithms Hypothesis Prevalence of adverse events detected in the group who take the medication is higher than the group without taking the medication 8

RESULTS Medication Black box warning Incidence in exposed patients Incidence in non-exposed patients p-value 95% CI of RR docetaxel neutropenia 0.5466*† 0.3312 1x10-15 (1.3251,2.0554) febrile neutropenia 0.4405*† 0.3386 1x10-4 (1.0439,1.6213) infection 0.5627*† 0.4314 2x10-6 (1.0461,1.6264) thrombocytopenia 0.3312* 0.2834 0.03 (0.9265,1.4742) stomatitis 0.1286* 0.0988 0.04 (0.9398,1.8027) hypersensitivity 0.0386 0.0266 0.10 (0.8252,2.5519) rash generalized 0.2476 0.2100 0.055 (0.9154,1.5187) erythema 0.0032 0.0014 0.21 (0.3413,15.3062) hypotension 0.8328 0.8582 0.90 (0.7242,0.8965) bronchospasm 0.0064 0.0179 0.94 (0.0894,1.4297) anaphylactic shock 0.3344*† 0.1707 4x10-14 (1.5564,2.4657) fluid retention 0.0289 0.0658 1.00 (0.2273,0.8487) oedema 0.2122 0.2616 0.97 (0.6203,1.0607) pleural effusion 0.0932 0.0929 0.49 (0.6885,1.4619) dyspnoea at rest 0.2830 0.2980 0.70 (0.7473,1.2146) cardiac tamponade 0.0037 0.86 NA abdominal distension 0.1640*† 0.1157 4x10-3 (1.0563,1.9021) The proposed phenotyping algorithms detected significantly higher incidence of 5 (by RR) or 7 (by z-test) out of 17 black box warnings in the patients taking Docetaxel, while the incidence in the patients taking Tamoxifen is not significantly different from the population. * denotes a significant difference for one-tailed two-proportion z-test at the level of 0.05. † denotes if RR is significantly larger than 1. Why not significant? Why p-value differ from RR? Conservative. Color to highlight the difference. 9

RESULTS Medication Black box warning Incidence in exposed patients Incidence in non-exposed patients p-value 95% CI of RR tamoxifen cerebrovascular accident 0.0143 0.0217 0.83 (0.2754,1.5766) pulmonary embolism 0.3657 0.3557 0.35 (0.8302,1.3505) endometrial adenocarcinoma 0.0314 0.0678 1.00 (0.2553,0.9964) sarcoma uterus 0.0200 0.0344 0.93 (0.2772,1.2194) Why not significant? Why p-value differ from RR? Bcs comparing with docetaxel, the AE is longer time scale, e.g. heart diseases, cancer, etc. 10

RESULTS Visualization of Patient 1 Visualization of Patient 2 11 ? Explain why sarcoma uterus, endo- happens right after ? 11 Visualization of Patient 2

Discussion We proposed use cases to detect black box warnings of two first-line chemotherapy drugs: docetaxel and tamoxifen A visualization prototype is devised to integrate medication and outcome information Study limitations Need to validate these techniques by gold standard Need to incorporate clinical notes Summary of what we achieved: Use case drug Viz -> Limitations: rephrase: improve standard, deeper analysis, confounding factors, advanced modeling 12

Acknowledgement  This research was supported and publication costs were covered by the Agency for Healthcare Research & Quality grant (#1R01HS022085) (GBM), and the University of Minnesota Clinical and Translational Science Award (#8UL1TR000114) (Blazer) Summary of what we achieved: Use case drug Viz -> Limitations: rephrase: improve standard, deeper analysis, confounding factors, advanced modeling 13

Questions? Thank you! sunxx697@umn.edu