Inactivation of Influenza Virus by Solar Radiation

Introduction

Influenza A belongs to a virus family of current international concern that includes the H5N1 group of strains that caused the pandemic of 1918–1919 and the strain responsible for bird flu (1,2). The World Health Organization recognizes that influenza pandemics like that of 1918–1919, which caused at least 20 million deaths (with estimates as high as 40–50 million deaths worldwide), are inevitable, and possibly imminent (3). The U.S. National Strategy for Pandemic Influenza published in November 2005 (4) acknowledged that in a typical year, influenza virus causes 36 000 deaths and a cost of $10 billion nationwide and recognized that a pandemic outbreak could dwarf this impact by overwhelming medical capabilities.

Influenza virus is readily transmitted by aerosols (1) and viral shedding in human nasal secretions has been reported to reach up to 10⁷ infectious influenza viral particles per milliliter (5). Viral shedding in human homes and community facilities may occur before symptom onset and continue for several days or weeks after symptoms have ceased (6). However, epidemiological data on the distribution of influenza mortality during the last three centuries in Great Britain could not be explained completely by direct person-to-person transmission, and “the primary agency mediating seasonal control remains unidentified” (7). Thus, abundance of infectious virus shed and epidemiological data not totally accounted by person-to-person transmission suggest that environmental factors play some role in influenza epidemics.

Infectious particles generated by coughing, sneezing and talking can reach distances of 12.5 m (over 40 feet, [8]). Aerosolized influenza virus has been recovered from fomites and environmental surfaces (6,9), and viable influenza virus has survived at least 48–72 h on contaminated surfaces (6,10). Increase in infection rate by vigorously sweeping floors indicates that influenza virus can readily be reaerosolized without much loss in infectivity (11). All this evidence indicates that transmission of influenza includes an environmental phase where viruses remain infectious in aerosols, on fomites, or upon other environmental surfaces.

The survival of influenza virus only varied up to 9% when the relative humidity changed between 50% and 70% (12). Rather extreme changes in relative humidity between 15% and 90% varied survival of influenza 12.5-fold (13), which corresponds to only 1.1 log₁₀ out of up to 7 log₁₀ (10⁷) infectious influenza viral particles per milliliter of nasal secretions (5). In these studies, virus survival was even less influenced by changes in temperature. In agreement with the relative small effect of humidity and temperature on influenza virus inactivation, previous studies (14,15) concluded that the mortality increase in winter was largely independent of temperature, humidity, or incidence of other diseases.

Although humidity, temperature and other variables may affect virus survival, ultraviolet radiation in sunlight is the primary virucidal agent in the environment (16). The relevance of sunlight inactivation of influenza seems supported by the correlation found in Brazil between increased influenza incidence in hospital admission records and solar UV-blocking by smoke during the burning season (17). The preeminent effect of germicidal UV (254 nm) radiation is clearly confirmed by a report whereby inactivation of air-borne virions by UV radiation virtually prevented the spread of influenza among patients in a veterans hospital, during the same time that an epidemic of influenza ravaged similar patients in nearby non-irradiated rooms (18). Although the virucidal effect of germicidal UV light on influenza is documented, the extent to which the UV radiation in sunlight can inactivate influenza virus, and hence the role of solar radiation on virus transmission, remains to be established.

Materials and methods

We estimated influenza A virus inactivation at different U.S. and global locations by combining the published UV (254 nm) sensitivity of the virus with solar spectra modified by an action spectrum previously derived for UV-inactivation of viruses (19). This method was originally developed and utilized to predict the survival of viruses of interest in biodefense.

Results

In this report, we calculated the expected inactivation of influenza A virus by solar UV radiation in various cities of the world during different times of the year. We predicted virus survival following solar exposure by combining the sensitivity of influenza A virus to monochromatic 254 nm radiation obtained in the laboratory (21,22) with solar radiometry data available for a number of sites in the northern latitudes weighted by a 254 nm-normalized UV action spectrum for virus inactivation previously determined in an approach that agreed with experimental data (19). We used 23.5 J m⁻² as the D₃₇ value for influenza (21) to calculate the virucidal (254-nm equivalent) UV fluence for full day solar exposure by the method developed previously for viruses of interest in biodefense (19).

Solar radiometric data were available at 33 reported sites in North America and one in New Zealand, and provided on a continual year-round basis by the USDA UV-B Monitoring and Research Program (http://uvb.nrel.colostate.edu/UVB/). We used solar radiometry data from 20 sites for 4 weeks centered around the times for winter and summer solstices and the spring and fall equinoxes and determined the typical maximum levels, representing clear, sunny days for each time during 2005. Because of radiometric equipment malfunction during 2005 at the Hawaii site, data for the first three time periods in 2006 were used for this location. All the radiometric data were weighted by the 254 nm-normalized action spectrum for UV inactivation of viruses, corrected to account for all UVB wavelengths to give a 254-nm equivalent UV flux at solar noon for the selected site and time of year (19).

The total inactivating flux (I) was obtained by combining the solar flux at UVB solar wavelengths with the 254 nm-normalized virus action spectrum:

This approach is the essence of integrating the area under a bar graph of virus inactivation at different wavelengths of interest. The flux value at each wavelength for which there is a datum must be multiplied by some factor to account for the flux from wavelengths for which there are no data. For example, some reports (23–25) include data at 295, 300, 305, 310 nm, etc., spaced evenly at 5 nm intervals, then each flux value was multiplied by five to account for the missing data. When the intervals between data points were not uniform, then the multiplication factors were adjusted to account for the missing data, being larger when the interval were larger, and vice versa.

We found empirically that the virucidal effectivenesses of 300 nm radiation (solar flux at 300 nm × 0.05 [the action spectrum value]) added to that of 305.5 nm (solar flux at 305.5 nm × 0.0064 [the action spectrum value]) contributed 75% of the total effective solar flux (data not shown). Therefore, solar flux measurements at those two wavelengths were used to estimate the total effective solar flux at each location by the equation:

in units of W m⁻²_254effect.

The daily fluence (flux × time) was obtained by multiplying this total effective solar flux by 3600 to change seconds to hours and divided by 0.2 because 1 h at noon contributes 20% of the daily total effective solar UVB:

The maximum daily virucidal solar UV radiation values calculated for sites for which radiometric data were available are presented in Fig. 1 as a function of radiometry site latitude for each selected time period. We assumed the shapes of the individual curves to be primarily determined by the angle the earth’s surface subtends in the solar radiation beam and the extent of attenuation of the UVB by the ozone layer. Although the SZA at a given location is the same at the spring and fall equinoxes, the solar UV radiation received in the northern hemisphere was generally greater in the fall than in the spring, except for the location furthest south, Hawaii (latitude 19.5°N). Data for Alexandra, New Zealand, in the southern hemisphere where the seasons are reversed, demonstrated the same trend with spring UV radiation being lower than fall UV radiation (data not shown). Note that the level of solar UV radiation needed to inactivate 90% of influenza A virus in 1 day (23.5 J m⁻²) is only available during winter solstice at latitudes below 30°. On the other hand, at summer solstice all sites reported have more than 23.5 J m⁻² day⁻¹, including Fairbanks, Alaska (latitude 65.1°).

Table 1 shows the expected levels of daily solar virucidal UV radiation as interpolated from Fig. 1, together with expected influenza virus inactivation expressed in log₁₀ form for populous North American metropolitan areas that could be particularly impacted by an influenza pandemic. Although the actual values obtained from individual radiometry sites near the selected metropolitan areas (e.g. the Beltsville, MD site is 12 miles from Washington, DC) differ somewhat from the values interpolated from Fig. 1, they show the same temporal trend and lead to the same conclusions. Table 2 presents similar values interpolated or extrapolated from Fig. 1 for populous metropolitan areas on other continents. The values in the tables clearly illustrate that influenza virus will be distinctly inactivated in different cities and at different times of the year. For example, at winter solstice (December, in the northern hemisphere), just after the beginning of the usual flu season (which peaks in January [26]) influenza virus exposed to full sunlight will be reduced by at least 99% (2 log₁₀) during a single day in Mexico City and other cities between the Tropics of Cancer and Capricorn, and by at least 90% (1 log₁₀) in Miami and Houston in the U.S. and Cairo, Egypt and Shanghai, China in the lower latitudes of the temperate zone, but will remain largely unaffected by the sun in Boston or Seattle in the U.S. or any of the European cities. Of course, the same trend applies to the Southern Hemisphere where winter begins in June.

Discussion

On a clear sunny day, the two components of solar radiation, direct and atmospheric-scattered indirect radiation, may be roughly equal in intensity. Any location that does not receive total solar exposure, both direct and indirect, for the entire day will have less virus inactivation than has been calculated in this report. Lower exposure may be caused by shadows from natural or artificial structures, e.g. trees, buildings, etc., as well as cloud cover and chemical and dust pollution. Varying ozone levels will also affect the level of UVB and thus the virus inactivation levels.

Among the environmental factors that affect survival of influenza virus, our calculation of virus inactivation by solar UV radiation appeared to produce an effect stronger than expected. The potential solar UV inactivation ranges from near negligible to 9 log₁₀ or more per day, depending on location and season (Tables 1 and 2). Therefore, it appears that the germicidal impact of solar UV radiation may be several orders of magnitude more relevant to environmental virus inactivation that the other primary physical factors (temperature and relative humidity).

However, other factors may be also involved in virus persistence as influenza epidemics still occur in the tropics (3) in spite of the strong virucidal effect produced there by solar UV radiation. Person-to-person transmission, aerosolization and environmental contamination are likely affected by seasonal human behavior. For example, people in the developed countries located in the temperate zone spend their work time indoors, minimizing exposure to outdoor environmental sources of influenza, while maximizing direct indoor person-to-person transmission. Thus, although our calculations suggest that solar UV radiation might have a significant role in influenza outbreaks, it is apparent that influenza epidemics will ultimately be understood only after considering the interactions among a number of variables.

The inactivation rates in the tables indicate that influenza virions should remain infectious after winter release from the host for several days in many higher latitude cities, with continued risk for reaerosolization and human infection. These findings are supported by increased mortality during winter months recorded between 1959 and 1999 when influenza was identified as the primary determinant of excess winter mortality (15). By spring equinox, solar inactivation improves in parallel with a general decrease in flu cases. Also paralleling the sharp seasonal differences in radiation at higher latitude, influenza epidemics have a marked seasonal occurrence in the northern and southern temperate zones (latitudes 30–70°), while the occurrence at lower latitudes is spread more evenly over the year (7).

The analysis described above should apply to all viral strains within the influenza A group as (1) the different strains that were studied have similar UV sensitivities (23.5, 25.3 and 22.1 J m⁻² for the Melbourne, NIB-4 and NIB-6 strains, respectively [21,22]) and (2) viruses within a family have a similar UV target and sensitivity (19). Therefore, periods of reduced solar radiation could foster persistence of infectious virus also in lake water and bird droppings where high incidence of “bird flu”influenza in local waterfowl has been observed (27,28).

The higher virucidal solar fluxes that we calculated during the fall compared with the spring values (Fig. 1) closely correlates with the reported lower incidence of household and day-care fomites contaminated with influenza A virus (29), where 23% of fomites collected in the fall were contaminated compared with 53% contaminated of fomites collected in the spring. Interestingly, no differences between dry and moist surfaces were encountered, while the seasonal effect was pronounced.

The correlation between low and high solar virucidal radiation and high and low disease prevalence, respectively, suggest that inactivation of viruses in the environment by solar UV radiation plays a role in the seasonal occurrence of influenza pandemics.