Skip to content

RFC 20121122: Partial Vaccine Immunity

Jump to bottom
ndh2 edited this page Dec 3, 2014 · 21 revisions

Applies to: Model description v1.0.0

Type of change: New feature, for next minor revision of the software.

Summary: This RFC adds variable levels of vaccine immunity.

Justification: The existing vaccination model does not capture vaccination of only some animals in a unit, or vaccine efficacy.

Background:

ADSM is herd-level with the possible states being: Susceptible, Latent, Subclinical Infectious, Clinical Infectious, Naturally Immune, Vaccine Immune, and Destroyed.

The states are not completely all-or-nothing: there is a “prevalence curve” overlaid during the Latent, Subclinical Infectious, and Clinical Infectious states to express the prevalence of infected animals within the unit.

ADSM takes an all-or-nothing approach to vaccination:

  • When a unit is vaccinated, there is a delay for immunity to develop.
  • If the unit receives an effective exposure during this delay period, the unit becomes infected and the vaccine has no effect.
  • Once immunity has developed, the unit is 100% protected against subsequent exposures.
  • If an already-infected unit is vaccinated, the vaccination has no effect.

Problem statement:

The modeling group wants to explore the effects of varying:

  • The proportion of animals within a unit that are vaccinated
  • The vaccine efficacy.

For ease-of-use, the modeler should enter these values — proportion of animals vaccinated, vaccine efficacy — directly, since these represent concrete real-world numbers. The challenge is to translate those numbers, in some biologically justified way, into effects in the herd-level model.

Experiment 1:

In the February 2013 meeting in Fort Collins, it was suggested to use the within-herd disease progression model developed for NAADSM v5 to explore the effects of vaccinating at different levels of coverage and at different times relative to the effective exposure. The research question was to see if those numbers translate predictably into flattening or shortening of the prevalence curve.

For these experiments, within-herd parameters were taken from a document written by G. Garner in which he was preparing prevalence parameters for the QUADS comparison exercise.

  • Number of animals: 293
  • Latent period: 4 days
  • Infectious clinical period: 6 days
  • Daily adequate contact rate: triangular distribution 1.1 - 1.4 - 2.2
  • Uniform mixing
  • Natural immunity period (after transition from infectious): forever (365)
  • Delay to vaccine immunity: uniform distribution 3 - 5
  • Vaccine immunity period: triangular distribution 21 - 28 - 35

(Latent, immune, etc. period are all animal-level.)

With no vaccination, the disease progression looks like this. Note that disease spreads through the entire population and at the end, all animals are naturally immune.

80% vaccination was used in the following experiments because 80% has been mentioned as the coverage that achieves herd-level immunity. (Here, 80% is the percentage of animals who become vaccine immune assuming there is no infection in the scenario. It is the product of percent of animals vaccinated × percent of vaccinated animals who become immune.) A series of runs were done where vaccination happens before infection, and another series of runs were done where infection happens before vaccination.

These examples from the vaccination-then-infection runs:

show vaccination not protecting against disease, but just delaying it. The disease persists at a low level among the susceptible animals. (Given the 6-day clinical period, even a small daily adequate contact rate is enough to keep at least 1 animal in the population infected until vaccine immunity runs out.) Then the disease proceeds as usual.

How much does the adequate contact rate need to be reduced before 80% vaccine immunity is protective? Some additional runs were done using uniform distributions 0.1 wide for the adequate contact rate. With a uniform distribution between 0.475 and 0.575, the vaccination was mostly protective, with only a few rare runs in which disease persisted until vaccine immunity ran out.

In the infection-then-vaccination runs, three distinct behaviors appear.

  1. If vaccination happens early on, the prevalence curve is flattened somewhat, but eventually the disease proceeds as normal once vaccine immunity runs out.
1. If vaccination happens a bit later, there is a “just right” timing where the disease is suppressed and does _not_ come back after vaccine immunity runs out. 1. If vaccination happens even later, the prevalence curve is still suppressed somewhat, and some animals never get infected. (Note that at the end, about 15% of the animals are still in the susceptible state: they never got infected.)

Experiment 2:

On the July 3 conference call some changes to the parameters were suggested.

For conventional vaccine:

  • longer delay to immunity, 5-8 days
  • keep 21-28-35 vaccine immune period
  • booster at 21 days
  • immunity will last 6 months after booster

It was noted that this could be modeled as a single vaccination event that gives 6 month immunity, but the modeler would need to later double the number of doses used, as recorded by the model.

For high potency vaccine:

  • keep 3-5 day delay to immunity
  • 6 month vaccine immune period
Clone this wiki locally