Differences between major versions of NAADSM

As new versions of NAADSM are released, the NAADSM Development Team makes a concerted effort to ensure that most of these new versions are fully compatible with earlier versions. In most cases, an older model will run in a new version of the NAADSM application. Differences between such versions are deemed "minor": please see NAADSM versioning for more information.

When exceptions to this rule occur and backward compatibility cannot be maintained, a new "major" version of NAADSM is released. This page summarizes the differences between major versions. Additional details are provided in the model specifications and the user's guides for these major versions.

Differences between NAADSM 3.x and NAADSM 4.x

Changes to local-area and airborne spread

In previous versions of NAADSM, users were restricted in their ability to represent non-directional local-area spread of disease (the spread of disease among premises in relatively close proximity to one another by mechanisms that are not readily traced, such as localized aerosol transmission, insects or vermin acting as mechanical fomites, or "across the fence" contact). Users could simulate directional airborne spread of disease, or non-directional local-area spread, but not both. In NAADSM 4, these two mechanisms are now parameterized separately.

In addition, the approach to calculating the probability of local-area and longer-distance directional airborne transmission of disease has been completely revamped. In previous versions of NAADSM, local-area/airborne spread was dependent not only on the source and recipient units involved, but was also influenced by the size of all other units in the study population. This dependence on the entire population not only had little biological or epidemiological justification, but made it very difficult to legitimately use parameters developed for one particular study in any other study that was not based on exactly the same study population. Also, using the previous approach to handling local-area/airborne spread, it was possible to obtain nonsensical probabilities of disease transmission of more than 100%. The new approach used in NAADSM 4 eliminates all of these problems.

Changes to within-herd prevalence of disease

The capability to represent the impact of within-unit prevalence of disease on spread by direct contact and by airborne/local-area transmission was introduced in NAADSM 3.1. This version did not, however, account for the different ways in which within-unit prevalence affected these distance mechanisms. Epidemiologically, it makes sense that local area/airborne transmission might be influenced by the prevalence of animals in a unit that are shedding the infectious agent. By contrast, latent individuals can also contribute to disease spread by direct contact (i.e., the introduction of an infected animal from one unit into another). The prevalences of shedding animals versus all infected animals may be quite different. Furthermore, the probability of disease spread by direct contact is influenced not only by the prevalence of infected animals, but also by the number of animals introduced from an infected unit into another: all other parameters being equal, shipments of more animals are more likely to result in disease transmission than shipments involving fewer animals. NAADSM 4 provides improved capabilities for simulating the effects of within-unit prevalence on disease spread.

Introduction of a "dead from disease" state

Prior versions of NAADSM did not incorporate any notion of disease mortality, which limited their applicability to diseases associated with high levels of mortality, such as avian influenza. NAADSM 4 incorporates a new herd-level disease state called "dead from disease". Users can now specify the probability with which infected units will die from disease or recover and progress to the naturally immune state.

Because NAADSM 4 is still a herd-based model, the most appropriate interpretation of the "dead from disease" state is that all or essentially all animals in the unit have died. This disease state is most applicable to small herds or flocks, which are be more likely to be entirely wiped out by an outbreak of disease. The utility of the "dead from disease" state for production types characterized by large units should be carefully considered by the modeler.

Elimination of exposure delay parameters

NAADSM 3 incorporated several delay parameters for direct, indirect, and airborne transmission of disease, initially designed to account for the time it takes for disease to be transmitted from a source to a recipient unit: for example, long-distance shipments might take longer to arrive at their destinations than short-distance shipments.

In NAADSM 3, however, there was no association between distance and delay: these parameters operated independently and in an unrealistic way. Additionally, because of the way the NAADSM is structured, there is already an implicit one-day delay: the effect of exposures occurs on the simulation day after the exposure takes place. Further complications in NAADSM 3 arose when control measures that would impact the outcome of a contact (e.g., quarantine or destruction) took place while such contacts between units were pending. Behavior of the model under these conditions was not well defined. Readers of earlier editions of the NAADSM user's guide will recall that we consistently advised users not to use these parameters. In NAADSM 4, they have been eliminated altogether.

Clarification of disease spread by direct and indirect contact

In NAADSM 3, users could choose whether latent and subclinical units could spread disease by direct contact, and whether subclinical units could spread disease by indirect contact. We have pointed out elsewhere that, based on definitions of direct and indirect contact, latent and subclinical units should always be able to spread disease by direct contact. Similarly, subclinical units should always be able to spread disease by indirect contact. In NAADSM 4, these two situations are now the case, and the user options described above have been eliminated.

Queuing herds to be vaccinated

Due to a quirk in the model specification for NAADSM 3, it was possible for herds in queue to be vaccinated to be listed in the queue multiple times (see section 9.14 in the model description for NAADSM 3.2). This behavior has been changed in NAADSM 4. This change not only more accurately represents expert opinion with respect to how vaccination is actually carried out, but it also makes it possible to provide useful information to model users about how limited vaccination capacity affects disease control efforts. Please see Understanding vaccination queues in NAADSM 3 and NAADSM 4 for more information.