vaccineff

library(vaccineff)

Usage

Vaccines are created to offer protection against diseases that affect human health. Quantifying how well vaccines work in controlled environments and in real-life settings remains a challenge for scientists. Estimating vaccine effectiveness ($VE$) is a key task once a vaccine is available as a control measure within a population, such as during the middle phase of an epidemic like Ebola or COVID-19, and also in the evaluation of regular vaccination programs, such as childhood vaccines.

vaccineff provides tools to estimate $VE$ under different study designs (Halloran, Longini, and Struchiner 2010). The package provides a set of features for preparing the data, estimating crude and adjusted effectiveness, controlling for potential confounders such as age and assessing the performance of the models used to approximate $VE$.

Who are the users / potential users?

vaccineff is useful for local, national, and international health agencies looking for a quick implementation to estimate $VE$ based on their available data. It also provides insights to researchers, data analysts, and epidemiology students on how to approach $VE$ using different methods. We believe that vaccineff would be specially useful for users without advanced training in statistical methods.

What is vaccine effectiveness?

In contrast with vaccine efficacy, which is the percentage reduction of disease incidence in a vaccinated group compared with an unvaccinated group under ideal conditions, $VE$ is the percentage reduction of disease incidence in a vaccinated group compared with an unvaccinated group under routine conditions. The reduction attributable to vaccination is usually assessed from data collected in observational studies (Halloran, Longini, and Struchiner 2010). Evaluating the effectiveness of vaccines in the field is an important aspect of monitoring immunization programs.

For which designs is this package?

vaccineff is a package designed to be used for any infectious disease for which a vaccine strategy has been implemented. This current version only allows measuring $VE$ for cohort study designs. Future version will include other designs such as test-negative/case-control studies, and the screening method (Torvaldsen and McIntyre 2002). For more information, see the vignette Other designs.

Cohort Design

In the cohort design, $VE$ is estimated using the Hazard Ratios ($HR$) between vaccinated and unvaccinated populations,

$$VE = (1-HR(t))\times100.$$

The $HR$ is estimated using the Cox Proportional Hazards model. In particular, we use the vaccine status of the individuals as the only covariate in the regression. Other confounders can be included as matching arguments to adjust for observational bias. The proportional hazards hypothesis is checked using the Schoenfeld test. A visual check is also provided using the log-log representation of the Survival Probability. If the hypothesis is not satisfied, it is recommended to stratify the population into smaller groups using the confounding variables.

What type of data is needed to use the package?

This package is designed to be used with vaccination data sets with the following structure.

Data for Cohort design

Data should be disaggregated at the individual level to track vaccinated and unvaccinated populations over time. The dataset must include the following information:

• Date(s) of vaccination for each individual: The package allows for multiple doses per individual and estimates the immunization date using delay times of outcomes and the timing of vaccine administration.

• Date(s) of outcome(s): The package estimates vaccine effectiveness against various outcomes.

• Date(s) of right censoring: The package allows for the inclusion of information on dates of events that constitute right censoring.

• Individuals’ demographic information (e.g., sex, age group, health status): These can be used as confounding variables to match the population and reduce observational bias.

An example dataset for a cohort design is included, with information on vaccination dates and biological details per dose, as well as relevant demographic information. The level of data aggregation is tailored to the characteristics and needs of the study case. To load this dataset, run the following code

# Load example data
data("cohortdata")
head(cohortdata)
#>         id sex age death_date death_other_causes vaccine_date_1 vaccine_date_2
#> 1 04edf85a   M  50       <NA>               <NA>           <NA>           <NA>
#> 2 c5a83f56   M  66       <NA>               <NA>           <NA>           <NA>
#> 3 82991731   M  81       <NA>               <NA>           <NA>           <NA>
#> 4 afbab268   M  74       <NA>               <NA>     2021-03-30     2021-05-16
#> 5 3faf2474   M  54       <NA>               <NA>     2021-06-01     2021-06-22
#> 6 97df7bdc   M  79       <NA>               <NA>     2021-03-21     2021-05-02
#>   vaccine_1 vaccine_2
#> 1      <NA>      <NA>
#> 2      <NA>      <NA>
#> 3      <NA>      <NA>
#> 4    BRAND2    BRAND2
#> 5    BRAND1    BRAND1
#> 6    BRAND2    BRAND2

Modeling vaccine effectiveness

VE for Cohort design

The current release of the package bases the estimation of $VE$ in the cohort design on the assumption of proportional hazards between vaccinated and unvaccinated populations. The $HR$ is estimated using the Cox proportional hazards model implemented in the R package survival.

The integrated dataset cohortdata serves as a minimal example of the package’s input. The data is accessed using data("cohortdata").

vaccineff has two main functions: 1. make_vaccineff_data: This function returns an S3 object of the class vaccineff_data with the relevant information for the study. This function also allows to create a matched cohort to control for confounding variables by setting match = TRUE and passing the corresponding exact and nearest arguments. make_vaccineff_data supports the methods summary() to check the characteristics of the cohort, the matching balance and the sizes of matched, excluded, and removed populations, and plot() to plot the cohort coverage.

2. estimate_vaccineff: This function provides methods for estimating VE using the $HR$. A summary of the estimation can be obtained using summary() and a graphical representation of the methodology is generated by plot().

# Create `vaccineff_data`
vaccineff_data <- make_vaccineff_data(
  data_set = cohortdata,
  outcome_date_col = "death_date",
  censoring_date_col = "death_other_causes",
  vacc_date_col = "vaccine_date_2",
  vaccinated_status = "v",
  unvaccinated_status = "u",
  immunization_delay = 15,
  end_cohort = as.Date("2021-12-31"),
  match = TRUE,
  exact = c("age", "sex"),
  nearest = NULL
)

# Print summary of vaccineff data object
summary(vaccineff_data)
#> Cohort start:  2021-03-26
#> Cohort end:  2021-12-31
#> The start date of the cohort was defined as the mininimum immunization date. 
#> 65 registers were removed with outcomes before the start date.
#> 
#> Nearest neighbors matching iteratively performed.
#> Number of iterations:  4
#> Balance all:
#>               u         v         smd
#> age   63.917069 62.997438 -0.08593156
#> sex_F  0.520277  0.573474  0.10701746
#> sex_M  0.479723  0.426526 -0.10701746
#> 
#> Balance matched:
#>               u         v smd
#> age   63.751784 63.751784   0
#> sex_F  0.521457  0.521457   0
#> sex_M  0.478543  0.478543   0
#> 
#> Summary vaccination:
#>               u     v
#> All       10973 19905
#> Matched   10789 10789
#> Unmatched   184  9116
#> 
#> // tags: outcome_date_col:death_date, censoring_date_col:death_other_causes, vacc_date_col:vaccine_date_2, immunization_date_col:immunization_date, vacc_status_col:vaccine_status

# Plot the vaccine coverage of the total population
plot(vaccineff_data)

# Estimate the Vaccine Effectiveness at 90 days
ve90 <- estimate_vaccineff(vaccineff_data, at = 90)

# Print summary of VE
summary(ve90)
#> Vaccine Effectiveness at 90 days computed as VE = 1 - HR:
#>      VE lower.95 upper.95
#>  0.6943   0.4889   0.8172
#> 
#> Schoenfeld test for Proportional Hazards assumption:
#> p-value = 0.027
#> Warning:
#> 
#> p-value < 0.05. Please check loglog plot for Proportional Hazards assumption

# Loglog plot to check proportional hazards
plot(ve90, type = "loglog")

For details on the estimation of VE in cohort studies see the vignette Introduction to cohort design with vaccineff

Key references

Halloran, Elizabeth, Ira Longini, and Claudio Struchiner. 2010. Design and Analysis of Vaccine Studies. Springer.

Torvaldsen, Siranda, and Peter B McIntyre. 2002. “Observational Methods in Epidemiologic Assessment of Vaccine Effectiveness.” Communicable Diseases Intelligence Quarterly Report 26 (3).