Feb 25, 2010 (CIDRAP News) – Winter influenza outbreaks in the United States typically follow periods of unusually low absolute humidity (AH), and this pattern suggests it may be possible to develop short-term forecasts of flu epidemics, according to a study published this week.
A five-member research team led by Jeffrey Shaman of Oregon State University found the association by comparing humidity records with the timing of flu epidemics nationwide over 30 years. They reported their findings in PLoS Biology.
The team also developed a mathematical model of flu transmission in which transmissibility was governed largely by absolute humidity. Outbreak patterns generated by the model matched up well with actual death rates from pneumonia and flu in five states, they report.
"The results indicate that AH affects both the seasonality of influenza incidence and the timing of individual wintertime influenza outbreaks," the report says. "The association of anomalously low AH conditions with the onset of wintertime influenza outbreaks suggests that skillful, short-term probabilistic forecasts of epidemic influenza could be developed."
Why seasonal flu outbreaks occur in the winter in temperate regions—a pattern that does not hold for pandemic flu—has long been a mystery. The new study builds on recent experimental evidence that aerosolized flu viruses survive much better at low humidity and flu in lab animals spreads much more easily at low humidity.
(AH refers to the actual content of water in the air, as opposed to relative humidity, the amount relative to the maximum possible at a given temperature.)
The authors first sought to determine whether the onset of the flu epidemic each winter is related to a period of abnormally low humidity. They used the increase in the rate of deaths due to pneumonia and flu above a certain threshold to define the onset of the epidemic. Then they identified the onset date in the 48 contiguous states and Washington, DC, for the 30 winters from 1972 to 2002.
To identify periods of unusually low humidity, the researchers measured the daily deviation of AH in each state from the 31-year mean for each day.
Comparing these results, they found that abnormally low AH typically occurs beginning 4 weeks before the onset of an epidemic, with the lowest level occurring 17 days before the outbreak.
Depending on the threshold used to define the start of an epidemic, AH was abnormally low during the 4 weeks preceding outbreaks 55% to 60% of the time, a small but highly significant shift from the expected 50%, the authors report. The findings indicate that unusually low AH readings "are not necessary for wintertime influenza onset but instead presage an increased likelihood of these onset events," they write.
The association between low AH and outbreaks varied by region and was strongest in the Northeast and the Gulf Coast states. The link was not statistically significant in much of the West, though AH was typically below average before outbreaks in that region as well.
To further test their hypothesis, the researchers assessed whether a mathematical model in which flu transmission is driven by AH could reproduce the observed seasonal patterns of pneumonia and influenza death rates. They tested the model in five states with different climates: Arizona, Florida, Illinois, New York, and Washington.
Using daily humidity records to adjust the basic reproductive number, or flu transmission rate in a fully susceptible population, the authors ran multiple simulations for each of the five states. Then they compared the average annual patterns of infection generated by the models with the average annual patterns of observed pneumonia and flu mortality and found that they corresponded well.
"Best-fit model simulations at each site capture the observed seasonal cycle of influenza," they write. The simulations produced consistent results among the five states when similar values were used for variables such as the basic reproductive number and the duration of infectiousness.
The authors say it's difficult to distinguish among the various environmental influences on flu seasonality, including AH, relative humidity, sunlight, and the school calendar. "Nevertheless, our findings indicate that AH is a major (and likely the predominant) determinant of influenza seasonality," given the link between low AH and outbreaks and the consistent response to AH in the simulations for the five states.
The authors say mathematical simulations indicate that school calendars can explain seasonal flu outbreaks, but "the correspondence with observations is not as good as for simulations driven by AH." However, school closures have a well-documented effect on flu transmission, and it's possible that humidity and the school calendar "act in concert" to drive epidemics, they add.
Wladimir J. Alonso, PhD, a research fellow in the Division of International Epidemiology and Population Studies at the Fogarty International Center of the National Institutes of Health, said the study presents good evidence for an influence of humidity on flu transmission in temperate regions. But he suggested that transmission may also be influenced by the effects of stressful weather conditions on the body.
The finding that drops in absolute humidity consistently precede wintertime flu epidemics is "very important for temperate regions, where probably these results can be generalized (given the broad range of US states studied)," Alonso told CIDRAP News by e-mail. "The authors point out that virus survival and transmission—that are indeed affected by absolute humidity, as shown in another study of the authors—are probably the proximate mechanism for this.
"Nevertheless, humidity (absolute or relative) is very high in many tropical regions when influenza is circulating," he added. "That is why another front of explanation might be needed to reconcile the observations in tropical and temperate climates."
Alonso said cold, dry weather conditions can stress the body, causing small lesions, vasoconstriction, and lower mucociliary activity in the respiratory tract. In addition, exposure to cold, which is associated with low humidity, "increases the energy required for thermoregulation, diverting resources from the high energy-consuming immune system."
"In all these cases, our defenses can be compromised, and perhaps this is the explanation—instead of the virus survival in the environment (as suggested by the authors)," Alonso said. "This would be perhaps easier to reconcile with what is observed in tropical regions, where influenza frequently circulates in the rainy season. . . . Higher precipitations—together with high temperatures—probably stress the immune system in similar ways, easing the way to respiratory infections."
Shaman J, Pitzer VE, Viboud C, et al. Absolute humidity and the seasonal onset of influenza in the continental United States. PLoS Biology 2010 Feb 23;8(2) [Full text]
