GIVE

Sentinel Chicken Surveillance and West Nile Virus in Florida

In 1999 and 2000, Florida watched other regions of the U.S. react to West Nile (WN) virus. We heard about the failure of sentinel chickens as transmission indicators when compared with dead wild birds. There was published speculation that white chickens were not attractive to the WN mosquito vectors. The use of sentinel chickens to effectively monitor WN transmission was questioned.

Elsewhere we have discussed the importance of obtaining accurate arboviral "surveillance" information and distinguishing this from "detection" (BuzzWords Oct./Nov. 2000 - West Nile in North America: A Florida perspective). The power of using wild birds to detect WN virus was confirmed in July 2001 when the first Florida WN virus infected dead bird was collected in Jefferson County. Subsequently, a dead bird was the first indication of the presence of the virus in most Florida counties. However, in a few counties an infected horse was detected first, in a few others it was a sentinel chicken, and then in Monroe County the first indication of virus transmission was an infected human.

What have we learned in 2001? Should our emphasis on viral "detection" continue? Once WN virus is "detected", what further "surveillance" information is required to predict human risk?

Here are some facts:

  • An infected dead bird is likely to be the first indication of WN virus in a new location due to the large number of exposed birds, the relative ease of finding and collecting dead birds, and the widely dispersed bird populations found throughout Florida.
  • Although Florida was forced to take action using dead bird detection in 2001, there was no predictive value in the number of infected birds and human or horse WN cases.
  • Horses were excellent indicators of WN transmission in Florida during 2001. This is due to their numbers, mosquito exposure, susceptibility to infection, and tendency to show clinical signs.
  • The number of infected horses in a region did not predict human cases.

In summary, once "detected", WN infected wild birds and horses are excellent "indicators" of the presence of WN virus in an area. However, this information provides little prediction of human risk to WN infection. When a reporter recently asked if dead birds or horses were a true "gauge" of human risk, we responded that we felt more comfortable with the term "indicator". Unfortunately, the bird or horse "gauge" has yet to be calibrated. It has been proposed that the gauge can be calibrated by tracking the number of dead birds per square mile, as was done for Staten Island, NY in 2000. Perhaps, but without an accurate measure of local virus transmission by mosquitoes, a dead bird count provides little concrete information about human risk, as was evidenced in the Florida Keys during the summer of 2001. In addition, it is possible that we will see an attenuated effect of WN on wild birds as the virus settles into Florida ecosystems. The introduction of this virus to the New World has been hard on certain avian species. However, many birds survive infection and pass this resistance trait on to their offspring. It is likely that in future years, we may see little or no WN-associated avian mortality, in much the same way that SLE and EEE viruses now interact with native wild bird populations. Finally, without an understanding of the relationships between mosquito transmission levels, numbers of infected birds in the overall avian population, and the proportions of infected mosquitoes and birds in an affected region, one is unable to accurately assess human risk. Even if we had a strong statistical correlation between a dead bird index and human WN cases, the assessment of human risk would remain difficult without a thorough understanding of all the factors that lead to mosquito infection and transmission levels in the field.

Some Florida facts about sentinel chickens and WN virus:

  • Thirty-eight Florida Mosquito Control or County Health Programs maintained sentinel chicken flocks during 2001 as part of their historical sentinel chicken surveillance effort for SLE and EEE viruses.
  • Sentinel flocks proved less effective for the initial "detection" of WN virus, mainly because they were not maintained in counties, especially Jefferson and Madison, where the initial WN transmission focus was located.
  • Soon after WN virus was detected in north Florida, a number of sentinels seroconverted to WN virus. This was because the virus expanded out of areas that had no sentinel chicken surveillance program into areas, including Leon and Duval Counties that did. Sentinel chickens and horses provide a direct measure of the number of WN virus transmitting mosquitoes in an area. If the mosquito transmission rate is 1:10,000, few sentinels will be bitten by an infected mosquito and seroconvert. Horses, on the other hand, will frequently seroconvert, because of their high level of mosquito contact. Humans, under this situation, will have a relatively small risk of infection. If the mosquito transmission rate is 1:1,000, sentinel chickens and horses will frequently be bitten by infected mosquitoes and will seroconvert. Humans will be at a much higher risk of WN infection when mosquito transmission rates approach 1:1,000.
  • As of Nov. 1, 2001, 23% (147 sentinels of 639 infected birds) of the WN-positive birds in Florida have been sentinel chickens, effectively ending the belief that white chickens are not attractive to WN virus mosquito vectors.

Wild bird surveillance, in its current form, is not a gauge of human risk to WN infection. Numerous well-placed sentinel chicken flocks can be a gauge of human risk. Sentinel chicken seroconverion rates can be used to assess human risk. Florida must expand its sentinel chicken surveillance program, and, in the process, reconfigure flock placement and the number of individual chickens exposed each week to meet the challenge of WN virus risk assessment. The three essential components of sentinel chicken surveillance are: location, location, and location. Seroconversion rates in a sentinel chicken flock provide a quantitative assessment of human risk in the area surrounding that flock. The total size of this area is still poorly understood. Horse seroconversion rates provide similar information, but will be costly and difficult to interpret, especially in light of the recent equine WN vaccination program and the extended lag period between the infection of the horse and the report of a confirmed WN case in that horse. Florida must support effective and efficient arboviral surveillance using the time-proven sentinel chicken surveillance methods that have been pioneered here in Florida.

Was the WN transmission observed in Florida during 2001 an EEE or an SLE type of event? It started out like EEE, focal in the Florida Panhandle, but later became more like a widely dispersed sporadic SLE transmission event. Was it a "big event "comparable to what Florida has seen in the past"? We submit that Florida, with fewer than 15 human WN cases in 2001, escaped the big event. The 237 horse cases as of Nov. 1, 2001 are indeed troubling, but we have little by way of comparison because SLE does not cause clinical disease in horses and many Florida horses are protected by EEE vaccinations. The dead wild bird index does not appear to have been helpful in predicting where or when human cases would appear. Under the best of circumstances sporadic human cases caused by WN, EEE, or SLE virus are difficult to predict. We appear to have escaped a big WN transmission event during 2001. Will we be prepared to predict and rationally react to a big event, be it WN, SLE, or EEE, in 2002?

Walter J. Tabachnick, Director
Florida Medical Entomology Laboratory

Jonathan F. Day, Professor
Florida Medical Entomology Laboratory