Zaata Yamni did you try zero-inflated negative binomial regression? Size of data set? Do you have a minimum of 400? You may use Markov chain Monte Carlos (MCMC) techniques for sampling before probability distributions of interested parameters or projected state variables can be obtained for quantification of uncertainty. .For the model check if any variable is causing singularity problem.

The hierarchical Bayesian modeling described by Cressie et al. (2009) is a promising approach.

Good luck

Did you try run the example code at the bottom of the function? This might provide clues to modifying the output and what the input ought to look like.

OK. "A simple inflation model where all zero counts have the same probability of belonging to the zero component can by specified by the formula y ~ x1 + x2 | 1."

This formula will fit a zero inflated model where the count part has two predictors x1 and x2. The logistic part of the model has no predictors just "1" which is the intercept. So the probability of a zero is fixed for all values of x1 and x2. For example if the intercept in the logistic part is -0.5 then for any observation there is around a 32% probability of a zero from this part of the model and if not a zero from this route an observation count dependent on the intercept + x1 and x2 effects for the count part (which can in theory produce zeroes in non-hurdle models). One would normally fir predictors to both parts (though not necessarily the same predictors in each part). For example the model could be: y ~ x1 + x2 | x3 + x4 if you thought that x3 and x4 were likely to predict zeroes and x1 and x2 how many counts for non-zero events.