The future of West Nile virus: a prediction

June 2005

My interest in West Nile virus (WNV) infection in the United States derives mainly from my role as an associate editor of ProMED-mail (www.promedmail.org), where there has been daily tracking of outbreaks of emerging diseases, including the WNV epidemic, and where viral disease moderators, such as Craig Pringle, PhD, and Charles Calisher, PhD, have commented on the evolution of the outbreak.

There are two purposes in writing this editorial. One is to draw attention to some facts about the epidemiology of WNV infection in the United States, which to my knowledge has not been widely noted. Another is to make a prediction that could either prove prescient or disastrously incorrect.

The rise of WNV

WNV infection was first noted in humans in the United States in 1999 in New York City and surrounding counties. There were 62 cases of central nervous system (CNS) disease reported, including seven deaths (case-fatality rate (CFR) 11%). The more common, milder forms of the infection, asymptomatic infection and West Nile fever, were undetected as there was no epidemiological serologic testing.

In 2000, the virus had spread in birds and mammals up and down the East Coast from Vermont to Washington. There were 21 human cases reported in New York, New Jersey and Connecticut, including two deaths (CFR 10%); all had CNS disease.

In 2001, the virus reached Florida and Canada and spread west from the East Coast about 800 miles. There were 66 cases of CNS disease reported with nine deaths (CFR 14%).

In 2002, the virus spread farther west to encompass over two-thirds of the United States and reached Mexico. There was an explosion of human infection with 4,156 cases reported. Serology became available to a limited extent, and diagnoses were made in people without CNS disease. In fact only 69% of those confirmed to have WNV infection had CNS disease. There were 284 deaths reported (6.8% of all confirmed cases and 10% of those with CNS disease). Almost two-thirds of human cases were from five states — Illinois, Michigan, Ohio, Louisiana and Indiana.

In 2003, more cases of non-CNS disease were detected through epidemiological testing and screening of blood donors and others. There were 9,862 cases confirmed, but only 29% had CNS disease. The absolute number with CNS disease was about the same as in 2002. There were 264 deaths (2.7% of all confirmed and 9% of those with CNS disease). About three-fourths of the cases were from five states – Colorado, Nebraska, South Dakota, Texas and North Dakota.

In 2004, 2470 cases were confirmed, of which 36% had CNS disease. The absolute number with CNS disease was less than one-third of those with CNS disease in 2002 or 2003. There were 88 deaths in 2004 (3.6% of all confirmed cases and 9.8% of those with CNS disease). This represented less than one-third of the deaths in 2002 or 2003. Most of the cases were in three states – California, Arizona and Colorado. By the end of 2004, WNV had been found in all states except Alaska and Hawaii.

Several conclusions can be drawn from these data. First, despite activity over an enlarging area of the United States, the number of cases of WNV infection as determined by CNS disease and mortality peaked in 2002, remained level in 2003 and significantly decreased in 2004. Second, the mortality rate of CNS disease has been constant in the 10% range over the six years of the outbreak. Third, as case recognition increased, the percent with CNS disease and the overall mortality rate decreased.

Although the data are not given here, as the wave of cases moved west, there were many more cases in an area during the second year in which the virus was recognized than there were during the first year.

Furthermore, in those states with many cases, the year after the peak of cases, there was a marked decrease in the number of cases reported.

The fall of WNV?

The reasons for the apparent decrease in WNV infection in humans in the United States are not known. There are many possible explanations and combinations of explanations for the decrease, none of which have been demonstrated.

1) The virus may possibly lose infectivity for birds or humans as it persists from year to year in an area.

2) The mosquito population may change in some way (eg, decrease in mosquitoes from weather changes or control measures) or become less competent vectors because of a change in the species. It is of interest that in the United States more than 40 species of mosquitoes belonging to at least 10 genera have been found to be West Nile positive. The most important belong to the genus Culex. Some species such as Culex pipiens bite mainly birds: they are ornithophagic and transmit virus among susceptible birds, such as the crow. Other mosquitoes bite humans (anthropophagic) and not birds. For humans to become infected a bridge vector is needed, a mosquito that bites birds but will sometimes also bite humans. In the United States such species include hybrids of the bird biter C. pipiens and a very closely related mosquito, C. molestus, that bites people. Other bridge vectors are C. restuans, C. salinarius and Aedes vexans. The actual mosquito vectors will depend on their geographical distribution.

3) Humans may be less likely to become infected because of mosquito avoidance among those humans most likely to be exposed. Herd immunity from prior exposure may play a role.

4) In my opinion, most likely, but again with no evidence, the bird population may have changed in some way. Reports indicate that large numbers of die-offs of birds have decreased. Perhaps herd immunity in birds has played a role. If this is the case, another wave of disease may be observed in the future if herd immunity is lost over a period of years.

5) There may be an artificial element in the decrease as well. Following the major WNV activity in a location, authorities may have stopped doing rigorous testing, so the counts of animal (bird) and human cases were artificially lowered by the “seek and ye shall find” theory of disease reporting.

Many states stopped collecting dead birds other than for sentinel testing, and many clinicians may have been diagnosing WNV infection clinically but not sending appropriate sera for testing.

Humans and equines, the two most susceptible mammalian hosts, do not develop sufficient viremias to significantly re-infect mosquito vectors and represent dead-end hosts for the virus. Any evolution in the virulence/pathogenicity of WNV must take place in the vector or the amplifying host (avian species).

To conclude, I will speculate on the future direction of WNV infection in the United States.

My speculation is that, with the exception of more detection of mild or asymptomatic disease by more widespread testing, the number of cases of WNV infection reported in the United States will continue to decrease and may never again reach the peak levels seen in 2002 and 2003. However, we can be certain that WNV will never disappear from the United States.

For more information:

• Donald Kaye, MD, Drexel University, College of Medicine, Philadelphia.

Editorial comments serve as a timely reminder as we enter the 2005 WNV season. One can only hope that Dr. Kaye’s prediction will prove true! — Theodore C. Eickhoff, MD