All 3 systems work from the premise that practicing veterinarians hold key animal health information, which could improve means for early detection of emerging disease if aggregated efficiently through advanced information technology. While all systems capture basic epidemiologic data, they each represent a different approach to emerging disease surveillance.

VetPAD has a syndromic surveillance approach. It can include all farm animals. It collects data describing every case. Cases are categorized by using dropdown lists, check boxes, and a clinical diagnosis. Based on the categorizations, cases can be flexibly aggregated for syndromic surveillance. The strategy to minimize the surveillance reporting impact is to provide a tool capturing the ordinary business data veterinarians must manage anyway (medical records, inventory, and accounts). Surveillance data are a subset of these other data.

The RSVP-A system employs an aggregation-based syndromic surveillance but focuses on a restricted set of syndromes (nonneonatal diarrhea, neurologic dysfunction or recumbency, abortion or birth defect, unexpected death, erosive or ulcerative lesions, and unexplained feed refusal or weight loss). These syndromes are defined to cover clinical signs of emerging disease other than the common production problems on which most livestock enterprises are focused. Practitioners determine the specific syndrome each case best fits and record demographic data about the diseased animals. The RSVP-A system also requests additional clinical observations potentially useful to further characterize incident patterns. The strategy to minimize the reporting impact is to focus on less common clinical syndromes and to make data capture for each case require <1 minute.

"émergences" has a different approach as it targets atypical cases and specific diseases, which correspond to known diseases hypothesized to be emerging. Forms are available (see an example of atypical case form, Figure) for reporting epidemiologic and clinical data. The system requests a follow-up description of each case's evolution and monthly confirmations of vigilance from veterinarians reporting no cases. Moreover, atypical cases can be categorized by the system administrator according to clinical description similarities to facilitate exploration of their potential links. The system has generic features, making it available for any country, any disease, and any domestic species. Description of atypical cases for "émergences" is a less frequent and more open process than the syndromic surveillance methods.

In all these systems, routine data recording is simplified by the use of pick-up lists. However, free text fields are also available, as the unexpected often does not fit in predefined fields. VetPAD and RSVP-A use mobile telephones or personal data assistants for data capture. "émergences" primarily uses the Internet.

A successful surveillance system must be able to keep veterinarians engaged and continuing to submit data after the novelty of the new system wears off. Systems can provide value to a veterinarian with useful management tools, which are available in VetPAD, and by enhancing their clinical expertise and intellectual curiosity. To trigger interactions and learning from participants' experiences, practitioners participating in "émergences" have access to all case descriptions. In addition, illness and death rates are available in real time either at the clientele level ("émergences") or at a custom-made level ("émergences," RSVP-A). In VetPAD, customized reports are available to involved parties.

One output of these surveillance systems is an indication of unusual events that require additional investigation. This investigation might include communication with other veterinarians to find additional cases, targeted epidemiologic studies, research projects, or control programs.

Other outputs are data upon which analyses can be conducted. A challenge is the categorization of reports to identify possible etiologic links. Procedures based on contextual analysis must be developed to analyze pick-up list data as well as free text (26). Each system must also address the challenge of detecting increased incidence of a rare event. Two types of situations can be considered. The first is the emergence from a "zero case" situation (e.g., BSE occurred probably as erratic cases before its amplification [27]). Incidence threshold analysis needed for this situation requires methods such as the evaluation of record process (28). Moreover, the constructed statistics should be robust with a small number of cases and allow differentiation of sporadic cases from emergence (29). The second situation is the emergence of clusters of highly pathogenic variants of an endemic disease. Spatial-temporal analysis can provide helpful insights concerning baseline patterns of clinical syndromes and aberrations from them, which can trigger further investigation.

Atypical case detection is limited by practitioners' experience, knowledge, vigilance, and willingness to report findings (30). Multiple, similar reports of atypical cases improve confidence that a new disease is emerging. Making case data available through surveillance systems, such as the 3 we have indicated, will also foster basic common knowledge and shared practical experience among veterinarians. Because surveillance for the unknown requires a mindset different from surveillance of the known, notification quality and vigilance should be enhanced by specific training courses (31).

A substantial limitation of syndromic surveillance is the need to establish baseline levels for defined syndromes. This step requires time and resources; however, without them, we cannot know when the incidence of a syndrome has significantly increased. VetPAD and RSVP-A are developing such baselines.

Economic consideration leaves few alternatives to clinical detection of farm animal diseases. Laboratory analyses are infrequently performed and generally more basic compared to human medicine (32). However, slaughterhouses and other assembly points do provide surveillance opportunities.

Finally, a clinical reporting tool alone is only the first step to determine if the cases share an etiologic pathway. Review by expert clinicians, necropsy findings, immunologic screenings, and focused epidemiologic studies play key roles in such determination (33). Similarities between distinct submitted atypical cases provide additional evidence. For example, BSE was identified as a novel syndrome through epidemiologic, clinical, and pathologic findings (8).

To determine whether to extend an information system, several points must be reviewed. First, the activity and number of participating veterinarians can be evaluated by quantifying indicators such as number of entries submitted, number of atypical cases entered, and participants' levels of accessing posted results. Moreover, all systems include reference diseases or symptoms for which descriptive statistics are available, which can serve to check quality recording (e.g., babesiosis in the "émergences" pilot study). In addition, the likelihood of detecting an emerging event is high. Many rare diseases are not defined in cattle, so a dedicated information system should detect >1 unexpected event over the test period. For example, the initial "émergences" pilot found 3 sets of clinical signs not linked to a known disease (persistent, ultimately fatal paraplegia, without general clinical signs [Figure]; weight loss, depilation at the extremities leading to death; and congenital cataract neither linked to bovine virus diarrhea nor familial history) and 1 rare known syndrome (facial eczema). Finally, the decision to extend a detection system will depend largely on the interest veterinarians hold and on the inclusion of new diseases as a national surveillance objective (6,34).