Evolving Healthcare Database Methods to Advance Pharmacoepidemiology Open Access

Barberio, Julie (Spring 2023)

Permanent URL: https://etd.library.emory.edu/concern/etds/1j92g867s?locale=en
Published

Abstract

The field of pharmacoepidemiology often uses real-world data (e.g., electronic health records, health insurance claims) to evaluate safety and effectiveness of products throughout drug development. Scientific and regulatory communities have been hesitant to rely on real-world evidence for regulatory and clinical decision-making due to the potential for epidemiologic biases, which threaten the validity of all observational research. The overarching goal of this dissertation was to evaluate multiple aspects of healthcare database research, assessing how pharmacoepidemiologic methods can be applied to appropriately chosen real-world data sources to inform medication safety and effectiveness.

In Aim 1, we assessed fitness for regulatory purpose of a mother–infant linked cohort in the Japan Medical Data Center claims database for postapproval pregnancy safety studies. Although accurate identification of the complete mother–infant population was possible, limitations of gestational age estimation may impede valid assignment of pregnancy onset and delivery dates as needed to define critical in utero exposure windows.

In Aim 2, we evaluated the risks of severe cytopenias in relapsed multiple myeloma patients who received sequential treatment with immunomodulatory agents (IMiDs) versus IMiD-free regimens in the Flatiron Health electronic health records database. Results suggest sequential exposure to IMiDs may increase the risks of severe cytopenias, specifically those related to white blood cells and especially among patients with recent cytopenia histories.

In Aim 3, we investigated the impact of incomplete death information in United States claims data by comparing cardiovascular cumulative risk estimates from models in which death was treated as a censoring event (cause-specific) versus competing event (sub-distribution). Differences in cause-specific versus sub-distribution cumulative risks in the claims-based cohort increased over follow-up time and were largest in the oldest age group, where cardiovascular outcome and mortality risks were the highest. Simulation results demonstrated the differences in cumulative risks to increase in response to doubled and tripled mortality rates.

The results of this dissertation demonstrate the importance of using appropriately chosen real-world databases, high-quality study designs, and rigorous analytic methods to produce valid real-world evidence. With such methods, we can inform trustworthy uses of fit-for-purpose real-world data for regulatory and clinical decision-making, which has important implications for real-world populations (at the practice, provider, and patient levels).

Table of Contents

Chapter 1: Introduction and Background. 1

Overarching Goal and Specific Aims. 4

Chapter 2: Characterizing Fit-for-Purpose Real-World Data: An Assessment of a Mother–Infant Linkage in the Japan Medical Data Center Claims Database. 6

Abstract. 6

Introduction. 8

Methods. 14

Study Population. 14

Data Considerations. 16

Statistical Analyses. 17

Results. 28

1. Data Relevancy. 29

1.1. Availability of Key Data Elements. 29

1.1.1. Exposure (Maternal Medication Exposure). 29

1.1.2. Outcome (Infant Major Congenital Malformations). 30

1.1.3. Covariates. 31

1.1.4. Patient-Level Linking (Linkage of Maternal and Infant Records). 31

1.2. Representativeness. 32

1.3. Sufficient Subjects. 37

1.4. Longitudinality. 41

2. Data Quality. 43

2.A. Mother–Infant Matches. 43

2.A.1. Completeness. 43

2.A.2. Accuracy. 45

2.A.2.1. Validity. 45

2.A.2.2. Conformance. 45

2.A.2.3. Logical Plausibility. 46

2.A.2.4. Consistency. 48

2.A.3. Transparency of Data Processing. 49

2.A.4. Provenance. 49

2.B. Gestational Period. 50

2.B.1. Completeness. 50

2.B.2. Accuracy. 51

2.B.2.1. Validity. 52

2.B.2.2. Conformance. 55

2.B.2.3. Logical Plausibility. 57

2.B.2.4. Consistency. 57

2.B.3. Transparency of Data Processing. 57

2.B.4 Provenance. 57

Summary of Assessment. 58

Discussion. 60

Chapter 3: Real-World Risk of Severe Cytopenias in Multiple Myeloma Patients Sequentially Treated with Immunomodulatory Drugs. 69

Abstract. 69

Background. 71

Methods. 76

Data Source. 76

Study Population. 77

Outcome. 79

Exposure. 80

Covariates. 81

Statistical Analyses. 86

Results. 89

Study Population. 89

Overall Treatment Effect 94

Treatment Effect Stratified by Prior Exposure. 97

Treatment Effect according to Cytopenia History. 100

Treatment Effect according to Age. 107

Treatment Effect according to Cytogenetic Risk. 114

Exploratory Analysis. 121

Discussion. 123

Chapter 4: Influence of Incomplete Death Information on Cumulative Risk Estimates in United States Claims Data. 135

Abstract. 135

Background. 137

Methods. 140

Data Source. 140

Study Population. 140

Outcome. 145

Exposure. 145

Covariates. 146

Statistical Analyses. 149

Simple Simulation. 151

Plasmode Simulation. 152

Exploratory Analysis. 153

Results. 155

Optum Cohort. 155

Simple Simulation. 162

Plasmode Simulation. 170

Exploratory Analysis. 177

Discussion. 184

Chapter 5: Summary of Results and Future Research. 191

Review of Major Findings. 192

Future Directions. 194

Conclusions. 197

References. 199

About this Dissertation

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
School
Department
Degree
Submission
Language
  • English
Research Field
Keyword
Committee Chair / Thesis Advisor
Committee Members
Last modified

Primary PDF

Supplemental Files