Dose response models for Giardia sp.

From QMRAwiki
Revision as of 21:49, 11 November 2011 by Sidgunawardena (talk | contribs) (Giardia infection in large rodents by human-derived strains)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Giardia infection in humans

Rendtorff (1954) administered Giardia sp. (unknown human-derived strain) to male prisoners by the oral route. The dose units were cysts, and the response was infection.

  • k = 0.01991, 95% CI 0.01050 to 0.03290

The deviance of this model fit is 8.371 with 7 degrees of freedom.
The p-value for the null hypothesis of goodness of fit is 0.30.
The p-value for the null hypothesis that the beta-Poisson model fits no better than the exponential model is 1.00.
At a dose of 34.8, the expected percentage of hosts that would suffer the response would be 50%.

Histogram of k
Dose response model with confidence intervals

Istre et al. (1984) investigated an waterborne outbreak of giardiasis by interviewing ill people and well controls. The dose units were glasses of water (as defined by the interviewee), and the response was infection. The amount of contamination in the water was not measured.

  • k = 0.06976, 95% CI 0.04030 to 0.1050

The deviance of this model fit is 8.371 with 6 degrees of freedom.
The p-value for the null hypothesis of goodness of fit is 0.31.
The p-value for the null hypothesis that the beta-Poisson model fits no better than the exponential model is 0.62.
At a dose of 9.9, the expected percentage of hosts that would suffer the response would be 50%.

Histogram of k
Dose response model with confidence intervals

Giardia infection in large rodents by human-derived strains

Erlandsen et al. (1969) administered Giardia sp. (unknown human-derived strain) to wild beavers and muskrats by the oral (beavers) and intragastric (muskrats) route. The dose units were cysts, and the response was infection. Much higher doses were required to infect these rodents with Giardia sp. from humans.

For beavers:

  • N50 = 14597.96, 95% CI 820.35 to 8.25E11
  • α = 0.13700, 95% CI 0.02236 to 220.65

The deviance of this model fit is 1.219 with 2 degrees of freedom.
The p-value for the null hypothesis of goodness of fit is 0.54.
The p-value for the null hypothesis that the beta-Poisson model fits no better than the exponential model is 4.0E-6.
At a dose of 14597.96, the expected percentage of hosts that would suffer the response would be 50%.

Scatterplot of N50 and alpha
Dose response model with confidence intervals

For muskrats:

  • k = 1.73E-6, 95% CI 9.49E-7 to 2.91E-6

The deviance of this model fit is 8.399 with 2 degrees of freedom.
The p-value for the null hypothesis of goodness of fit is 0.078.
The p-value for the null hypothesis that the beta-Poisson model fits no better than the exponential model is 0.36.
At a dose of 400663, the expected percentage of hosts that would suffer the response would be 50%.

Histogram of k
Dose response model with confidence intervals