Estimation of Influenza Vaccine Effectiveness from Observational Studies Open Access

Ainslie, Kylie Ellen Cain (Spring 2018)

Permanent URL: https://etd.library.emory.edu/concern/etds/ns064602j?locale=en
Published

Abstract

Each year seasonal influenza epidemics cause millions of influenza infections worldwide. Influenza vaccination is recommended as the best way to protect against influenza infection; however, a different vaccine must be produced each season due to changes in influenza virus types, subtypes, and phenotypes from one season to the next. To evaluate the reduction in influenza infections caused by vaccination, vaccine effectiveness (VE) is estimated each year. Observational studies are exclusively used to estimate VE in the United States after the universal recommendation of influenza vaccination because randomized clinical trials are now unethical. Here we consider five different types of observational studies used to assess VE. We use probabilistic models to estimate influenza VE from each study design.

 

First, we develop a probability model and accompanying maximum likelihood procedure to estimate vaccine-related protection against transmission of influenza from the household and the community. We apply our method to data from a monitored household study conducted in Michigan during the 2012-2013 influenza season and estimate source-specific transmission parameters and VE. We find that the 2012-2013 influenza vaccine provides a significant protective effect against community-acquired transmission.

 

Second, we develop a dynamic probability model for the evaluation of bias of VE estimates from four commonly used observational study designs: active surveillance cohort (ASC), passive surveillance cohort, test-negative (TN), and traditional case-control. We use the model to evaluate and compare estimates of VE against symptomatic and medically-attended influenza when different sources of bias are present. We show that the preferred study designs for estimating VE against symptomatic influenza and medically-attended influenza are ASC and TN studies, respectively. TN studies are cheaper and involve fewer logistical issues compared to ASC studies; however, if vaccination is suspected to affect the probability of non-influenza acute respiratory illness then one should consider a cohort study.

 

Finally, we extend our dynamic probability model to further evaluate the bias of VE estimates from test-negative studies. First, we allow vaccination to occur over time (as in a pandemic). Several influenza pandemics have occurred throughout the past century, none greater than in 1918, which killed between 50 and 100 million people. Future influenza pandemics have the potential to inflict a tremendous disease burden. Thus, it is important to determine VE in this setting. Second, we assume an all-or-none vaccine model, where a proportion of vaccinated individuals acquire complete immunity from infection, while the remaining vaccinated individuals acquire no protection. This model differs from the vaccine model assumed in the second topic of this dissertation. However, since vaccine model cannot directly be observed it is important to assess the bias of VE estimates under this alternative model. For each extension we assess the bias of TN-based VE estimates when different sources of bias are present.

Table of Contents

Abstract iii

Acknowledgements v

1 Introduction 1

1.1 Influenza . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.1.1 Pandemic Influenza . . . . . . . . . . . . . . . . . . . . . . 5

1.2 Influenza Vaccine . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

1.2.1 History of Influenza Vaccination . . . . . . . . . . . . . . . 7

1.2.2 Vaccine Efficacy and Effectiveness . . . . . . . . . . . . . . 9

Cohort Studies . . . . . . . . . . . . . . . . . . . . . . . . . 10

Case-Control Studies . . . . . . . . . . . . . . . . . . . . . . 13

1.3 Disease Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

1.3.1 Epidemic Models . . . . . . . . . . . . . . . . . . . . . . . . 15

1.4 Aims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

1.4.1 Estimation of influenza vaccine effectiveness in household

studies . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

1.4.2 Evaluation of bias of influenza vaccine effectiveness

estimates from observational studies . . . . . . . . . . . . . 19

1.4.3 Bias of test-negative-based estimates of influenza vaccine

effectiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2 Estimation of influenza VE in household studies 23

2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

2.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.2.1 Probability Model . . . . . . . . . . . . . . . . . . . . . . . 27

2.2.2 Maximum Likelihood Procedure . . . . . . . . . . . . . . . 31

2.2.3 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

2.2.4 Sensitivity Analyses . . . . . . . . . . . . . . . . . . . . . . 34

2.2.5 A Real-Life Example . . . . . . . . . . . . . . . . . . . . . . 35

2.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

2.3.1 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

2.3.2 Sensitivity Analyses . . . . . . . . . . . . . . . . . . . . . . 40

2.3.3 A Real-Life Example . . . . . . . . . . . . . . . . . . . . . . 43

2.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

3 Bias of influenza VE estimates in observational studies 51

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

3.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

3.1.2 Study designs . . . . . . . . . . . . . . . . . . . . . . . . . . 52

3.1.3 Sources of Bias . . . . . . . . . . . . . . . . . . . . . . . . . 52

3.1.4 Outcomes of Interest . . . . . . . . . . . . . . . . . . . . . . 54

3.1.5 Objectives of this Work . . . . . . . . . . . . . . . . . . . . . 55

3.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.2.1 Model Description . . . . . . . . . . . . . . . . . . . . . . . 56

Assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . 59

3.2.2 True VE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

True VE Against SI . . . . . . . . . . . . . . . . . . . . . . . 61

True VE Against MAI . . . . . . . . . . . . . . . . . . . . . 62

3.2.3 VE Estimates . . . . . . . . . . . . . . . . . . . . . . . . . . 62

ASC Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

PSC Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

TN Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

TCC Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

3.2.4 Sources of Bias . . . . . . . . . . . . . . . . . . . . . . . . . 66

3.2.5 Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

3.2.6 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

3.2.7 Sensitivity Analyses . . . . . . . . . . . . . . . . . . . . . . 69

Alternative Values of True VE . . . . . . . . . . . . . . . . . 69

Alternative Probabilities of (X, U) . . . . . . . . . . . . . . 69

Alternative Probabilities of Vaccination . . . . . . . . . . . 70

Sensitivity and Specificity . . . . . . . . . . . . . . . . . . . 71

3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

3.3.1 Sensitivity Analyses . . . . . . . . . . . . . . . . . . . . . . 77

Alternative Values of True VE . . . . . . . . . . . . . . . . . 77

Alternative Probabilities of (X,U) . . . . . . . . . . . . . . . 79

Alternative Probabilities of Vaccination . . . . . . . . . . . 79

Sensitivity and Specificity . . . . . . . . . . . . . . . . . . . 82

3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82

4 Estimation of influenza VE from test-negative studies 87

4.1 Vaccination Over Time . . . . . . . . . . . . . . . . . . . . . . . . . 89

4.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

4.1.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

Model Description . . . . . . . . . . . . . . . . . . . . . . . 90

True VE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

VE Estimates from TN Studies . . . . . . . . . . . . . . . . 94

Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94

Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

Calculations and Simulations . . . . . . . . . . . . . . . . . 96

Sensitivity Analyses . . . . . . . . . . . . . . . . . . . . . . 98

4.1.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

4.1.4 Sensitivity Analyses . . . . . . . . . . . . . . . . . . . . . . 103

Vaccination Coverage . . . . . . . . . . . . . . . . . . . . . 103

True VE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104

Study Length . . . . . . . . . . . . . . . . . . . . . . . . . . 105

4.2 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

4.3 All-or-None Vaccine Model . . . . . . . . . . . . . . . . . . . . . . 110

4.3.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

4.3.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

Model Description . . . . . . . . . . . . . . . . . . . . . . . 110

True VE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

Estimated VE . . . . . . . . . . . . . . . . . . . . . . . . . . 114

4.3.3 Calculations and Simulations . . . . . . . . . . . . . . . . . 115

4.3.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116

4.3.5 Sensitivity Analyses . . . . . . . . . . . . . . . . . . . . . . 117

Alternative Values of True VE . . . . . . . . . . . . . . . . . 117

Vaccination Coverage . . . . . . . . . . . . . . . . . . . . . 119

Alternative Probabilities of (X,U) . . . . . . . . . . . . . . . 120

Alternative Probabilities of Vaccination . . . . . . . . . . . 122

4.3.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

5 Discussion 125

5.1 Work In Progress . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

5.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . 126

Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

5.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

A Appendix to Chapter 2 137

A.1 Source-Specific Probabilities of Infection . . . . . . . . . . . . . . . 137

A.2 Model Adequacy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138

B Appendix to Chapter 3 141

B.1 True VE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

B.1.1 True VE Against SI . . . . . . . . . . . . . . . . . . . . . . . 141

B.1.2 True VE Against MAI . . . . . . . . . . . . . . . . . . . . . 143

B.2 Probability of Being a Case . . . . . . . . . . . . . . . . . . . . . . . 145

B.2.1 ASC Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

B.2.2 PSC Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

B.2.3 TN and TCC Studies . . . . . . . . . . . . . . . . . . . . . . 148

B.3 Probability of Being a TN Control . . . . . . . . . . . . . . . . . . . 152

B.4 Probability of no ARI . . . . . . . . . . . . . . . . . . . . . . . . . . 155

B.4.1 Probability of being a TCC control . . . . . . . . . . . . . . 157

B.5 Input Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

C Appendix to Chapter 4 163

C.1 Vaccination Over Time . . . . . . . . . . . . . . . . . . . . . . . . . 163

C.1.1 True VE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

True VE Against SI . . . . . . . . . . . . . . . . . . . . . . . 163

True VE Against MAI . . . . . . . . . . . . . . . . . . . . . 165

C.1.2 VE Estimates from TN Studies . . . . . . . . . . . . . . . . 166

Probability of Being a Case . . . . . . . . . . . . . . . . . . 166

Probability of Being a TN Control . . . . . . . . . . . . . . 171

C.1.3 Input Parameters . . . . . . . . . . . . . . . . . . . . . . . . 177

C.1.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

C.2 All-Or-None Vaccine Model . . . . . . . . . . . . . . . . . . . . . . 186

C.2.1 True VE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186

True VE against SI . . . . . . . . . . . . . . . . . . . . . . . 186

True VE against MAI . . . . . . . . . . . . . . . . . . . . . . 188

C.2.2 TN-based Estimates of VE . . . . . . . . . . . . . . . . . . . 189

Case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189

Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193

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