Increase in COVID-19 downwind following a wind change

doi:10.21203/rs.3.rs-1741081/v1

Download PDF

Article

Increase in COVID-19 downwind following a wind change

https://doi.org/10.21203/rs.3.rs-1741081/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Aerosolized SARS COV-2 is viable for at least 3–5 hours. Aerosols can rapidly become droplet nuclei and be carried long distances by wind before they settle. We hence investigated the possibility of identifying wind-assisted transmission of COVID-19. COVID-19 cases/100,000 population was calculated for hotspots and surrounding areas. Daily wind direction/speed data for hotspots was collated. Seven-day rolling averages of COVID-19 cases/100,000 population was plotted against wind direction/speed to compare case rate trends in hotspots, upwind and downwind areas. Within 14 days of the wind blowing into an area, case rate trends downwind differed from that of the hotspot. The difference compared to the hotspot could be an increase, plateau, or a slower decrease. This suggests that viral particles carried by the wind can lead to an increase in infections downwind. Research on ways to reduce aerosol transmission in the community is urgently needed.

Within two weeks of the wind blowing from a hot spot

a change in COVID-19 case rates is seen in downwind areas

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) spread to cause a global pandemic in under four months ¹. Lockdown measures helped slow its spread. However, with the ease of lockdown, local hotspots and increased transmission was reported in most countries. It is essential we understand all the mechanisms of transmission to stop the spread of evolving variants and to plan against future pandemics.

Only 46% of 339 adult Americans with coronavirus disease 2019 (COVID-19) reported a known contact ². This suggests that a large proportion of individuals could be infected by less obvious routes. There is increasing evidence of aerosol mediated transmission of viral infections. Aerosol mediated transmission of SARS coronavirus has been demonstrated in a hospital ward ³, and through plumes of aerosols carried by the wind between high-rise buildings ⁴. Possible aerosol mediated transmission of SARS-CoV-2 within a quarantining facility was reported from New Zealand ⁵. A large outbreak in a cruise ship emphasizes the importance of aerosol transmission of SARS-CoV-2 ⁶. An outbreak of COVID-19 has been reported among building workers in Singapore suggesting a role for aerosols in outdoor settings ⁷. In experimental conditions, aerosolized SARS-CoV-2 in 65% relative humidity was reported to be viable for at least 3 hours (which was the duration of the experiment) ⁸. Subsequently viability of aerosolized SARS-CoV-2 for up to 5 hours, (again, the duration of the experiment) has been shown in artificial saliva at 68%-88% relative humidity ⁹. In both experiments, the virus was viable up to the maximum duration tested; the exact duration of viability in aerosols is still unclear. It is hence likely that SARS-CoV-2 particles carried by the wind could remain viable and cause infections downwind. To test this hypothesis, we identified areas in the UK with a high number of COVID-19 cases (hotspots) during the initial phase of the pandemic and assessed the trends in case rates of the hotspot, upwind and downwind within 14 days (i.e., the incubation period of COVID-19) of a period of wind change. If wind assisted transmission is possible, we would expect a change in trends downwind within the incubation period (i.e., 14 days of the start of the wind change). The timeline of measures introduced to stop viral spread and changes to testing strategies were also considered as they could impact case numbers. We then determined if the slopes of the hotspot, upwind and downwind areas were significantly different. Where the trend in case rates upwind of a hotspot was significantly different, we investigated if the changes could be explained by the area being downwind of another hotspot.

Following the lockdown introduced on the 23^rd of March, case rates in the UK started to fall after hitting two high points on the 5^th and the 19^th of April (Fig. S1) ¹⁰. The relationship between case rates and wind for specific regions are described below.

West Midlands and surrounding areas: The wind was southwesterly until the middle of March and variable after that (Fig. 1A). We combined data from Birmingham, Wolverhampton, Walsall, Dudley, Sandwell, Solihull and Coventry into the West Midlands area. West Midlands is straddled by Staffordshire (including Stoke) to the north and Worcestershire, and Warwickshire to the south (Fig. 1B). The wind was northeasterly for two periods of 6-7 days each (26^th– 31^st March, average speed 12.8 km/hour; 17^th-23^rd April, average speed 12.6 km/hour). Since a northeasterly wind would travel from West Midlands into Worcestershire and Warwickshire, these areas were combined for analysis (WosWar). The wind could blow across West Midlands (~30 km along the NE axis) into WosWar in around 3 hours. Worcestershire is ~47 km along the NE axis; therefore, the wind will blow across West Midlands and WosWar in 6-7 hours.

The seven-day rolling average number of cases/100,000 population in West Midlands, WosWar and Staffordshire by date is shown in Fig. 1C. Cases in West Midlands and WosWar peak on the 3^rd April before beginning to fall, a trend similar to that observed for the whole of the United Kingdom (5^th of April) ¹⁰. Unlike West Midlands, case rates in WosWar plateau from the 7^th of April, 12 days after the first wind change (Fig. 1C). The plateauing starts two days before the 9^th of April when the first high throughput (Light House) lab in Milton Keynes, (close to Birmingham) was opened ^11,12. A slight increase in case rates can be seen in both West Midlands and WosWar after the 15^th of April when testing in care homes was started ^11,13. Linear regression lines fitted to the two trends are shown in Fig. 1D. The slopes of the regression lines were significantly different (p=0.012). Nine days after the second wind change, the rate of decline in case rates in WosWar becomes slower compared to West Midlands. Though this was immediately preceded by a testing strategy on the 23^rd of April (testing of all essential workers), this is unlikely to specifically affect WosWar. The slopes of the linear regression lines fitted to the two trends (Fig. 1E) were significantly different (p<0.001).

The graph for Staffordshire on the other hand looks very different to West Midlands and WosWar (Fig. 1C). Case rates in Staffordshire peak a day later (4^th April) before falling. After the first wind change, a steady increase in case rates well above West Midlands is seen from the 8^th of April. Staffordshire is upwind of West Midlands. We hence investigated if Sheffield, a known hotspot, northeast (and hence upwind) of Staffordshire could account for these changes (Fig. 2). A northeasterly wind blew across Sheffield into Staffordshire at three time points in March and April (Figs. 2A, 2B). Case rates in Sheffield were much higher than in Staffordshire, West Midlands or WosWar (Fig. 2C). Case rates in Staffordshire increase from the 8^th of April, 12 days after the first wind change, peaks on the 15^th of April approaching the rates in Sheffield, before falling. The slopes of the linear regression lines fitted to the two trends (Fig. 2D) were significantly different (p=0.001). Case rates also gradually increase from the 18^th of April, six days after the second wind change. During this increase, the third wind change sets in. Case rates in Sheffield start declining from the 30^th of April, 11 days after the start of the third wind change. However, case rates in Staffordshire plateau until the 5^th of May. Between the 6^th- 14^th of May, case rates in Staffordshire are higher than that in Sheffield. The slopes of the linear regression lines fitted to the two trends for the second (Fig. 2E) and third wind changes (Fig. 2F) were significantly different (p=0.007; p<0.001 respectively). Case rates in Staffordshire did not vary significantly following the opening of the Light House lab in Alderley Park, close to Staffordshire on the 20^th of April ^11,12. A sharp increase in case rates in Sheffield and Staffordshire is however seen after the introduction of essential worker testing on the 23^rd of April ^11,13.

London and surrounding areas: As with Birmingham, the wind was mostly southwesterly until the epidemic began. Two periods of northeasterly wind were observed after the lockdown (26^th-31^st March, average speed 13.0 km/hour; 18^th-22^nd April, average speed 11.0 km/hour) (Fig. 3A), traveling through the London area into Surrey (Fig. 3B). Following the NE compass axis, London is ~49 km across, Surrey is ~44 km across. Therefore, the wind could blow across the whole of London and Surrey in 7-9 hours.

Cases in the London peak between the 5^thand the 8^th of April and fall after that. Cases in Surrey peak on the 6^th of April and starts to drop on the 7^th. It then increases on the 8^th (13 days after start of first wind change), and then plateaus until the 17^th of April. There was a slight increase in numbers after the 15^th followed by a fall from the 18^th of April (Fig. 3C) which coincides with changes in testing strategies on the 15^th and 17^th of April respectively ^11,13. The slopes of the linear regression lines fitted to the two trends (Fig. 3D) were significantly different (p<0.001). Following the second wind change (18^th-22^nd April), cases in London and Surrey continue to fall. However, there appears to be a slight increase in the number of cases in Surrey from the 24^th of April for 4-5 days. This slight change happens right after the introduction of essential worker testing ^11,14. The slopes of the linear regression lines fitted to the two trends (Fig. 3E) were not significantly different (p=0.17)

There are periods of southerly wind blowing from London into Buckinghamshire, Essex and Hertfordshire (BEH) north of London (Fig. 3A). The southerly wind (23^rd-25th March) blew at a time when numbers in BEH are increasing steadily (Fig. 3B). Case rates in BEH peak on the 9^th of April and drop after that (Fig. 3C). No clear change in case rates can be seen (though a slight increase can be seen around the 3^rd of April). There were also southerly winds between the 4^th and 7^th of April. The rate of fall slows down from the 16^th of April (12 days after the 4th). But this change is also immediately after care home testing was introduced ^11,13. Case rates in BEH plateau between the 24^th-29^th of April (following the introduction of essential worker testing) after which it gradually falls ^11,14. Since a change to testing protocol preceded change in case rates these two instances were not investigated further.

Tyne and Wear, County Durham and Northumberland:

The wind was predominately westerly/southwesterly until the outbreak began (Fig. 4A). Data from North and South Tyneside, Newcastle upon Tyne, Gateshead and Sunderland were considered as one area, “Tyne and Wear”. The adjoining counties were County Durham and Northumberland (Fig. 4B). The seven-day rolling average number of cases/100,000 population in Tyne and Wear, County Durham and Northumberland by date is in Fig. 4C.

In April, a southerly (4, 5 April) and southwesterly wind (7 April) blew over three days (Fig. 4A) at an average speed of 8.9 km/hour. This wind would blow into Northumberland from both Tyne and Wear and County Durham. As Tyne and Wear and County Durham is ~54 km across (north to south), the wind would cross into Northumberland in ~6 hours. A significant increase in case rates in Northumberland happens 8 days after the wind change (12-15 April). At this time case rates in Tyne and Wear and County Durham had peaked and were falling (Fig. 4C).

Since a southerly and southwesterly wind will blow from both County Durham and Tyne and Wear into Northumberland, the combined case rates for these areas and for Northumberland is shown in Fig. 4D. The trend in case rates in the combined areas of Tyne and Wear + County Durham and in Northumberland are similar until the 11^th of April. Eight days after the southerly wind, while case rates in the combined areas of Tyne and Wear + County Durham start to decline, the rates in Northumberland increase significantly until the 15^th of April. The slopes of the linear regression lines fitted to the two trends (Fig. 4E) were significantly different (p=0.001). Case rates in both areas then start declining and the patterns looks similar from the 20^th of April.

Between the 16^th and 24^th of April the wind blew in an easterly/northeasterly direction (Fig. 4A) with an average speed of 12.2 km/hour. The wind hence blew from Tyne and Wear into County Durham. Following the NE compass axis, the distance between North Tyneside to County Durham is ~21 km across. County Durham is ~61 km across. Therefore, it would take 6-7 hours for the wind to go across Tyne and Wear and County Durham. Cases in County Durham continue to fall until the 28^th of April (12 days after) when the case rates plateaued for a week until the 5^th of May (Fig. 4C). The slopes of the linear regression lines fitted to the two trends (Fig. 4F) were significantly different (p=0.001).

The slight increase in the number of cases in County Durham after the 23^rd of April is likely to be due to the change in testing strategy ¹³. There was also a change in testing strategy on the 28^th of April (testing to protect the most vulnerable) ¹⁵ and the announcement of 100,000 home testing kits being sent across the UK on the 29^th of April ¹⁶. It is possible that the plateau from 28^th April - 5^th May is due to these testing strategies. This trend is not seen in Tyne and Wear or Northumberland or in any of the other areas analyzed for that matter. Home testing by its very nature takes time for the results to be available (i.e., time taken for collection device to be posted, sample collected, returned and tested), the plateau is likely to be following the wind change and not due to a change in testing strategy.

Details on population, prevalence of infection, trends in case rates of hotspots and downwind areas shown in the order of outcomes downwind are given in table 1. No clear relationship between these parameters in the hotspot and downwind areas are obvious. In situations where case rate trends increased downwind, case rate trends in the hotspots were increasing in two and was stable in the third; all three hotspots also had a mean case/100,000 above 8.8. There were two wind changes from areas with case/100,000 over 8.8 which led to a plateau or a slow decline in case rates downwind. In both instances, the trend in case rates in the hotspot was not increasing steadily.

We have only analyzed data from the initial wave of the pandemic. Various factors, including lockdown and travel restrictions, could affect the spread of infection. Nationwide lockdown introduced on the 23^rd of March dramatically changed movement patterns in the UK. Travel was permitted for key workers and for essential purposes (shopping for food within a 5-mile radius, exercise close to the house). Key workers could therefore contract and transmit the virus back to the areas they are travelling from. However, any such transmission is unlikely to affect downwind areas alone. Infections within care homes were a major problem during the first wave. It was not possible to obtain reliable data on the number and distribution of care homes in different areas, so we are unable to account for this. Though we cannot rule out the possibility of a local outbreak just after a wind change, the probability of local outbreaks causing the observed changes in all the downwind regions investigated is small (see figures 1C, 2C, 3C, 4C and 4D).

Since changes to the testing policy and reporting definitions could affect the number of confirmed cases, data for all analysis were collated on one day (28/06/2020). Our dataset was hence collated prior to the removal of duplicates on the 2^nd of July. Though the issue was UK wide, this is may or may not have an impact on case rates downwind. Since the areas analyzed are from England, changes in testing strategies between countries should not have an impact. We have not analyzed the data specifically for rainfall or temperature. However, having looked at the pattern of rain during the periods of sustained wind and the following days, we do not think rain had an obvious impact. Given that we are looking at downwind areas, in a 14 day period following wind change, temperature is unlikely to be an important factor.

The increase in cases downwind is likely to be due to the wind carrying virus particles from a hot spot to the adjacent area. Whether the change is easily identifiable probably depends on factors such as the population density in the two areas, the difference in prevalence between the two areas and the duration of the wind change. A change in wind direction of two days or more was observed all four areas analyzed. Increasing case rate trends in the hotspot were associated with increasing (Staffordshire, Northumberland) or plateauing (Surrey and WosWar) of case rate trends in downwind areas. In the only instance where a wind change did not lead a significant difference in the slope (London vs Surrey - second wind change), case rates in Surrey (downwind area) were higher than that of London (the hotspot) when the wind change happened.

Since the flow of wind across the UK does not begin or end at the edge of a county, it is likely that places outside of the areas analyzed could also have an effect. On initial analysis, we expected Staffordshire case rates not to be affected by viral particles carried from West Midlands (being upwind) and therefore to have a similar pattern as West Midlands. However, case rates in Staffordshire increased from the 7^th of April and behave very differently to West Midlands and WosWar (Fig. 1C). Staffordshire is ~50 km southwest of Sheffield, a known hotspot we choose not to include for analysis initially in the assumption that the Pennine Hills could affect wind currents. A check on Ventusky suggested that a northeasterly wind from Sheffield would blow into Staffordshire and hence we collated data from Sheffield and included it in the analysis (Fig. 2A-C). Unlike the West Midlands, case rate trends in Sheffield were constantly high from the beginning of lockdown (23^rd of March) until the 1^st of May after which case rates gradually reduced (Fig. 2C). The increasing case rate trends in Staffordshire from the 7^th of April and the long duration of sustained transmission is temporally associated with the wind blowing from Sheffield into Staffordshire (Fig. 2C). There were also 4 days of southerly wind when the wind would have blown from West Midlands into Staffordshire (4^th-7^th of April). This, along with the second northeasterly wind from Sheffield into Staffordshire could have contributed to the increase in case rates seen in Staffordshire on the 18^th of April.

The production of droplets from coughs and sneezes are well known. Small droplets and aerosols are also generated while breathing ¹⁷ and speaking ^18-20. People infected with influenza viruses (A/B/dual infection) also shed fine aerosols (≤5µ) during normal breathing and speaking apart from coughing ²⁰. Infectious influenza virus could be cultured from 39% of these individuals pointing to the important role aerosol mediated transmission of influenza ²⁰. Droplets ≤100µm are the largest fraction generated during breathing ^21,22, coughing ²² or sneezing ²². The distance a droplet spreads depends on its size, temperature, humidity and air currents ^22-24. As per the Wells evaporation-falling curve. larger droplets >100µm in diameter take very little time to drop 2 meters and fall close to the individual ^25,26. Droplets ~100µm in diameter settle at 30 cm/second and those around 10µm in diameter settle at 0.3 cm/second, taking ~10 minutes to fall 2 meters ^25,26. The estimated time for falling two meters is 6 seconds for a droplet of 100µm and 600 seconds for a 10µm droplet. At 18˚C, in unsaturated air, particles that are 100µm and 50µm in diameter evaporate to become “droplet nuclei” in 1.7 seconds and 0.4 seconds respectively ²⁵. Hence, in normal conditions, a large proportion of particles under 100µm are likely to become droplet nuclei. The estimated time to settle for a 1µm particle being 16.6 hours ^25,26, these droplet nuclei are likely to be carried long distances in the presence of air currents. We now know that aerosolized SARS-CoV-2 is twice as stable as influenza ⁹ and aerosolized SARS-CoV-2 in artificial saliva more stable at higher relative humidity (68-88% - decay rage: 0.40% min^-1) compared to moderate relative humidity (40-60% - decay rate: 2.27% min^-1) ⁹. The average relative humidity is stable across the UK (Edinburgh: 81% over 6-years ²⁷; England and Wales: 80.3% over 10 years ²⁸). This suggests that environmental conditions are conducive for the virus to survive in the UK. Rapid inactivation of the virus exposed to sunlight calibrated for 40˚ North latitude (Equivalent to Madrid / Philadelphia) in under 20 minutes has been reported ²⁹. However, areas situated above 50˚ North latitude (London - 51˚ North) are unlikely to get the high-intensity of sunlight used for the experiment and will probably get the mid-intensity levels at noon, at the height of summer, in the south of England ^30,31.

COVID-19 transmission indoors via aerosols is known. In a choir practice, 87% (53 individuals) of those exposed to a symptomatic individual were infected ³². Data from a cruise ship outbreak of COVID-19 highlight the importance of aerosol mediated transmission ⁶. After quarantining measures were introduced in the cruise ship, fomite transmission was reduced and short-range transmission (i.e., within cabins) increased. The contribution of aerosol transmission was similar to droplet transmission pre-quarantine (median: 60% vs 40%. p=0.32). However, the contribution of aerosol transmission increased significantly during quarantine (median: 85% vs 15%. P<0.0001), suggesting transmission via air currents in the cruise ship ⁶. The reduction in other common respiratory viral infections following the introduction of lockdown and social distancing measures at a time of increased COVID-19 transmission suggests that of SARS-CoV-2 can spread from a distance ³³.

Apart from aerosols generated from the respiratory tract, the importance of aerosol generated from stool and urine also needs to be considered. Aerosol mediated transmission of the SARS coronavirus via vertical soil stacks is known ⁴. These aerosols rose through air shafts and were then blown through open windows to cause transmission in three blocks of flats downwind ⁴. SARS-CoV-2 is shed in high concentrations in stool and urine ³⁴, with many having diarrhea associated with COVID-19. The virus in urine and stool is also viable and infective as shown by infection experiments in ferrets (urine: days 11 and 13; stool day 15) ³⁴. Aerosols are generated from toilets on flushing, with a significant increase in particles <3µm in size ³⁵. Hence a large amount of virus must be flushed from infected households, residential homes and hospitals generating plumes within the toilet which could then be vented into the atmosphere either through open windows, exhaust fans or home ventilation systems. Soil pipes are also vented to the outside atmosphere, and in the UK, soil pipe venting to the outside air can be seen on the roof. This offers a direct route for aerosols generated by flushing to be vented into the surrounding air.

The data hence suggests that small droplets are generated during normal day-to-day living, irrespective of whether someone is coughing or sneezing. Since most small droplets are likely to evaporate and become droplet nuclei before they settle, they are likely to be transported by wind. We also know that individuals also shed the virus when asymptomatic (either pre-symptomatic or truly asymptomatic). These individuals are likely to be outdoors, increasing the possibility of droplet nuclei being produced and transported by wind. The recommendation from the NHS is to keep windows open to reduce transmission of COVID-19 to household contacts ³⁶. Hence, a proportion of droplet nuclei generated indoors could also be transported outside by the wind.

We know that dust particles, bacteria, fungi and insects are carried across the globe by wind ³⁷. Aerosols are also carried by the wind over hundreds of miles. Aerosolized sea salt left behind when sea water spray evaporates is carried inland and affects chloride levels in inland water. When measured, the lines of equal chloride content was almost parallel to that off the shore off Massachusetts ²⁵. Sea salt aerosols have been detected hundreds of miles inland, having travelled 250-300km in 10-24 hours ³⁸ demonstrating the long distances small droplet nuclei can be transported.

Our manuscript is the first to report the impact of wind on case numbers downwind. Others have reported the impact of wind on local case numbers (i.e., in the hotspot and not downwind). In Suffolk County, USA, warmer temperatures (16-28˚C) and low wind speeds (≤8.5km/hour) were significantly associated with increased COVID-19 incidence following multivariate analysis ³⁹. The authors also report that COVID-19 incidence appeared to increase slightly on cooler, windier days as a function of wind speed, but this was not significant on multivariate analysis ³⁹. The study however does not investigate the directionality of wind. Suffolk county is surrounded by the water on three sides. Wind from the South and East will blow from the Atlantic Ocean. Wind from the North will blow from Connecticut across kilometers of sea and that from the West will blow from New York. Hence, it is possible that a significant difference might be observed if wind direction is included for analysis. A negative association between wind speed and COVID-19 cases has already been reported ^40,41. In 14 day lagged analysis, an increase in wind speed by 5Km/hour is associated with a 6% drop in local COVID-19 cases; a positive association was also seen with maximum temperature (5˚C increase = 7% increase in COVID-19 cases) and absolute humidity between 5-10% g/m³(23% increase in COVID-19 cases) ⁴¹. Absolute humidity was consistent even on 7 day lagged analysis ⁴¹. An analysis of global weather suggests positive association between wind, other meteorological parameters and increase in local case numbers ⁴². An inverse association between wind speed and secondary cases of SARS CoV was reported from China and two local regions within China ⁴³. Unlike SARS CoV, lower wind speeds are associated with lower incidence of MERS CoV ⁴⁴. These reports suggest that wind might be an important factor in viral transmission, and its impact on different viruses may need to be investigated in detail.

The possibility that SARS-CoV-2 can be transmitted by the wind has significant implications in stopping viral spread. Apart from vaccines, the available options are reducing the production of aerosols, contact with aerosols and prophylactic measures which could protect post-exposure to aerosols. Facemasks have a role in reducing viral shedding and in the inhalation of viral particles. Masks helped reduce secondary transmission to family members in China ⁴⁵. Countries in Asia that enforced mask wearing over lockdown measures had lower total cases of COVID-19. Cloth masks are sufficient to reduce the spread ⁴⁶, and possibly reduce the burden of virus that infects a person in the community ⁴⁶ allowing critical PPE to remain available to healthcare workers. UK guidelines initially required a 2-meter distance from others in public spaces. After lockdown was eased, this was reduced to 1 meter+ and people advised to wear a mask when in proximity to others. Since the data suggests that virus is carried long distances, compulsory mask-wearing is likely to be an important control measure, both to stops aerosols from an infected person and possibly to reduce the chance of infection in a susceptible individual. Since most individuals are unlikely to know if they were exposed during the day, it is important that we make use of the innate antiviral immune response in cells as the first line of defense against respiratory viral infections ⁴⁷. In this mechanism, infected cells produce hypochlorous acid with which they fight viral infections. Supplying chloride ion locally to nasal epithelial cells via saltwater helps the cells clear viral infections quicker and reduce viral shedding ⁴⁸. Others have shown the efficacy of local saltwater application (nasal irrigation/ sprays) in preventing viral upper respiratory tract infections in adults and children ^49,50. A prophylactic study to stop URTI including COVID-19 is urgently needed.

Data on COVID-19 cases was taken from the Gov.UK ‘Coronavirus (COVID-19) cases in the UK’ dashboard ¹⁰ which is updated daily and displays numbers of laboratory-confirmed cases of COVID-19 in the UK. Data extract was obtained on 28/06/20 and we have shown data up to 31/05/20 to allow for any reporting delays. Time period was chosen to be during lockdown – to keep a consistent environment. The dashboard displays numbers for each constituent country in the UK. However, due to population sizes we have focused on England as the other regions reported information aggregated over larger geographical areas. Case numbers in England was available aggregated to region, upper tier local authority (UTLA) and lower tier local authority (LTLA) levels. Number of cases for each area were adjusted for population size using 2020 numbers from the Office for national statistics ⁵¹ and presented as 7-day rolling averages per 100,000 population to account for any reporting bias over the weekend.

Birmingham, London, and Newcastle were chosen as hotspots, being metropolitan areas with a high number of cases within the UTLA level, and with surrounding areas without major adjoining cities. Areas were also selected for being relatively flat as hills are more likely to impede wind flow. Sheffield was initially ruled out as a hotspot for this reason as we anticipated that the Pennine Hills could affect wind direction. However, Sheffield had to be included to explain the trend in Staffordshire. Data from Ventusky did not suggest major variations in wind direction and we were able to complete the analysis. Some hotspots were not selected for analysis as there were other cities with large number of cases close by (e.g., Liverpool and Manchester) with commuting areas in between making it difficult to tease out changes that could be caused by a wind change. Some areas were combined to account for commuter towns. These include West Midlands (Wolverhampton, Walsall, Dudley, Sandwell, Birmingham, Solihull and Coventry), and Tyne and Wear (made up of Newcastle upon Tyne, Gateshead, North Tyneside, South Tyneside and Sunderland). For our analysis data from Stoke-on-Trent was included as part of Staffordshire. Other areas were combined to consider a particular wind direction, for example ‘WosWar’ (Worcestershire and Warwickshire) denotes the area downwind of West Midlands if the wind was from the northeast.

Weather data was collected from two sources – 1. the Ventusky web application, developed by the Czech meteorological company InMeteo, and 2. the Met Office. In most instances, met office weather stations were not close to the place of interest. Hence data from Ventusky was gathered manually and cross-checked by a second person prior to being used. We hence used Ventusky data for Birmingham, Sheffield and London regions and Met Office data (Albemarle, Northumberland) was used as representing Newcastle and surrounding regions. Wind speed data at 10m above ground in km/hour (Ventusky) or knots (Met Office – multiplied by 1.852 to convert to km/hour) was available in both sources. Ventusky report wind speed three hourly while the Met Office records data hourly. The mean of these values gave the daily average wind speed. In Ventusky, wind direction is represented by moving arrows. Based on the direction of the arrow, the wind direction of wind determined to be north, south, east, west and northeast, northwest, southeast and southwest. The wind direction at the eight available time points was transcribed to an Excel database and cross-checked by a second person. The wind direction for the day was then determined. Being a manual process, a few days were marked “variable” when the wind direction changed substantially and often within the day. Since wind direction from the Met Office was reported in degrees, each day had an average value, and the direction could hence be clearly assigned for all days. Compass degrees were divided into 8 and assigned a compass direction: >337.5 and ≤22.5 = ‘N’; >22.5 and ≤67.5 = ‘NE’; >67.5 and ≤112.5 = ‘E’; >112.5 and ≤157.5 = ‘SE’; >157.5 and ≤202.5 = ‘S’; >202.5 and ≤247.5 = ‘SW’; >247.5 and ≤292.5 = W; >292.5 and ≤337.5 = ‘NW’. Since Met Office data was available hourly, we used the mode of direction over the 24-hour period as the wind direction for the day. The wind maps from Ventusky with that from the closest available Met office weather station appeared similar. Though a few individual days could be different, the periods of sustained wind were the same.

We investigated periods when the wind blew from a hotspot into the area of interest for two or more days. In most situations, the wind blew in one direction over the period. Sometimes the wind direction could be different, but due to the positioning of the hotspot and the downwind areas, the net effect would be the wind blowing from the hotspot into the downwind areas. We observed case rates in the hotspot and downwind areas for two weeks from the first day the wind blew from the hotspot into the area of interest for any changes associated with the wind change. We compared the rate of change in case rates in downwind area to hotspot by fitting a linear regression line to the case rates following a wind change. The difference in slopes were compared by t-statistic ⁵².

Since aerosols could potentially be washed away by rain, we also looked for periods of significant rainfall in Birmingham, London, Newcastle, Ashford, and Sheffield on Ventusky. Intensity of rain was classified as light (<1mm/day), moderate (1-4mm/day) and heavy (>4mm/day) and duration of rain was classified as short (≤25% of the day), medium (>25% and ≤50% of the day) and long (>50% of the day). The highest rating of intensity and/or duration helped identify if there was rain was light, moderate or severe.

Data Availability:

Publically available data used for analysis can be obtained at the websites mentioned below.

Weather data:

UK Met Office https://www.metoffice.gov.uk/ .
Ventusky web application: https://www.ventusky.com/?p=52.41;-1.97;8&l=wind-10m

UK Covid-19 cases:

Gov.UK: Coronavirus (COVID-19) cases in the UK https://coronavirus.data.gov.uk/?_ga=2.71257488.1607018182.1590504796-115249015.1588603027

ONS figures:

Office for National Statistics: Population projections for local authorities: https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/populationprojections/datasets/localauthoritiesinenglandtable2. Table 2: 2018 based projections

Acknowledgments:

We would like to thank the Met Office and Ventusky for the weather data and for answering our doubts.

Funding: SR is supported by NHS Research Scotland (NRS) and the Scottish Chief Scientist Office (CSO)

Author contributions: SR, CG and AH designed the study. AH, SR collated the data. AH, SR and CG analyzed the data. AH and SR wrote the manuscript; CG reviewed the manuscript.

Competing interests: The authors declare no competing interests.

Materials and Correspondence:

Correspondence: Dr. Sandeep Ramalingam, Consultant Virologist, Department of Laboratory Medicine, Royal Infirmary of Edinburgh; Edinburgh, UK

WHO. Timeline: WHO's COVID-19 response, <https://www.who.int/emergencies/diseases/novel-coronavirus-2019/interactive-timeline#event-0> (2020).
Tenforde, M. W. et al. Characteristics of Adult Outpatients and Inpatients with COVID-19 - 11 Academic Medical Centers, United States, March-May 2020. MMWR Morb Mortal Wkly Rep69, 841-846, doi:10.15585/mmwr.mm6926e3 (2020).
Wong, T. W. et al. Cluster of SARS among medical students exposed to single patient, Hong Kong. Emerg Infect Dis10, 269-276, doi:10.3201/eid1002.030452 (2004).
Yu, I. T. et al. Evidence of airborne transmission of the severe acute respiratory syndrome virus. N Engl J Med350, 1731-1739, doi:10.1056/NEJMoa032867 (2004).
Eichler, N. et al. Transmission of Severe Acute Respiratory Syndrome Coronavirus 2 during Border Quarantine and Air Travel, New Zealand (Aotearoa). Emerg Infect Dis27, doi:10.3201/eid2705.210514 (2021).
Azimi, P., Keshavarz, Z., Cedeno Laurent, J. G., Stephens, B. & Allen, J. G. Mechanistic transmission modeling of COVID-19 on the Diamond Princess cruise ship demonstrates the importance of aerosol transmission. Proc Natl Acad Sci U S A118, doi:10.1073/pnas.2015482118 (2021).
Leclerc, Q. J. et al. What settings have been linked to SARS-CoV-2 transmission clusters? [version 2; peer review: 2 approved] Wellcome Open Research5 (2020).
van Doremalen, N. et al. Aerosol and Surface Stability of SARS-CoV-2 as Compared with SARS-CoV-1. N Engl J Med382, 1564-1567, doi:10.1056/NEJMc2004973 (2020).
Smither, S. J., Eastaugh, L. S., Findlay, J. S. & Lever, M. S. Experimental aerosol survival of SARS-CoV-2 in artificial saliva and tissue culture media at medium and high humidity. Emerg Microbes Infect9, 1415-1417, doi:10.1080/22221751.2020.1777906 (2020).
GOV.UK. Coronavirus (COVID-19) in the UK <https://coronavirus.data.gov.uk/?_ga=2.71257488.1607018182.1590504796-115249015.1588603027> (2020).
Dunn, P., Allen, L., Cameron, G. & Alderwick, H. COVID-19 policy tracker. A timeline of national policy and health system responses to COVID-19 in England, <https://www.health.org.uk/news-and-comment/charts-and-infographics/covid-19-policy-tracker> (2020).
GOV.UK. Health Secretary launches biggest diagnostic lab network in British history to test for coronavirus <https://www.gov.uk/government/news/health-secretary-launches-biggest-diagnostic-lab-network-in-british-history-to-test-for-coronavirus> (2020).
GOV.UK. Government to extend testing for coronavirus to more frontline workers, <https://www.gov.uk/government/news/government-to-extend-testing-for-coronavirus-to-more-frontline-workers> (2020).
GOV.UK. Coronavirus testing extended to all essential workers in England who have symptoms <https://www.gov.uk/government/news/coronavirus-testing-extended-to-all-essential-workers-in-england-who-have-symptoms> (2020).
GOV.UK. Further expansion of access to coronavirus testing helps protect the most vulnerable, <https://www.gov.uk/government/news/further-expansion-of-access-to-coronavirus-testing-helps-protect-the-most-vulnerable> (2020).
GOV.UK. Major home testing programme for coronavirus will track levels of infection in the community, <https://www.gov.uk/government/news/major-home-testing-programme-for-coronavirus-will-track-levels-of-infection-in-the-community> (2020).
Wurie, F. B., Lawn, S. D., Booth, H., Sonnenberg, P. & Hayward, A. C. Bioaerosol production by patients with tuberculosis during normal tidal breathing: implications for transmission risk. Thorax71, 549-554, doi:10.1136/thoraxjnl-2015-207295 (2016).
Anfinrud, P., Bax, C. E., Stadnytskyi, V. & Bax, A. Could SARS-CoV-2 be transmitted via speech droplets? medRxiv, doi:10.1101/2020.04.02.20051177 (2020).
Asadi, S. et al. Aerosol emission and superemission during human speech increase with voice loudness. Sci Rep9, 2348, doi:10.1038/s41598-019-38808-z (2019).
Yan, J. et al. Infectious virus in exhaled breath of symptomatic seasonal influenza cases from a college community. Proc Natl Acad Sci U S A115, 1081-1086, doi:10.1073/pnas.1716561115 (2018).
Stadnytskyi, V., Bax, C. E., Bax, A. & Anfinrud, P. The airborne lifetime of small speech droplets and their potential importance in SARS-CoV-2 transmission. Proc Natl Acad Sci U S A117, 11875-11877, doi:10.1073/pnas.2006874117 (2020).
Duguid, J. P. The size and the duration of air-carriage of respiratory droplets and droplet-nuclei. J Hyg (Lond)44, 471-479, doi:10.1017/s0022172400019288 (1946).
Tellier, R., Li, Y., Cowling, B. J. & Tang, J. W. Recognition of aerosol transmission of infectious agents: a commentary. BMC Infect Dis19, 101, doi:10.1186/s12879-019-3707-y (2019).
Li, X., Shang, Y., Yan, Y., Yang, L. & Tu, J. Modelling of evaporation of cough droplets in inhomogeneous humidity fields using the multi-component Eulerian-Lagrangian approach. Build Environ128, 68-76, doi:10.1016/j.buildenv.2017.11.025 (2018).
Wells, W. F. On air-borne infection: Study II. Droplets and droplet nuclei. American Journal of Hygiene20, 611–618 (1934).
WHO. Natural ventilation for infection control in health-care settings, <https://www.who.int/water_sanitation_health/publications/natural_ventilation/en/> (2009).
Price, R. H. M., Graham, C. & Ramalingam, S. Association between viral seasonality and meteorological factors. Sci Rep9, 929, doi:10.1038/s41598-018-37481-y (2019).
Warren-Gash, C. et al. Circulating influenza virus, climatic factors, and acute myocardial infarction: a time series study in England and Wales and Hong Kong. J Infect Dis203, 1710-1718, doi:10.1093/infdis/jir171 (2011).
Schuit, M. et al. Airborne SARS-CoV-2 Is Rapidly Inactivated by Simulated Sunlight. J Infect Dis222, 564-571, doi:10.1093/infdis/jiaa334 (2020).
GOV.UK. Solar ultraviolet radiation in Great Britain (1989 to 2008), <https://www.gov.uk/government/publications/solar-ultraviolet-radiation-in-great-britain-1989-to-2008> (01/06/2011).
Bilbao, J. & de Migue, A. Erythemal Solar Irradiance, UVER, and UV Index from Ground-Based Data in Central Spain. Applied Sciences10, 6589 (2020).
Hamner, L. et al. High SARS-CoV-2 Attack Rate Following Exposure at a Choir Practice - Skagit County, Washington, March 2020. MMWR Morb Mortal Wkly Rep69, 606-610, doi:10.15585/mmwr.mm6919e6 (2020).
PHE. Weekly national Influenza and COVID-19 surveillance report: Week 19 report (up to week 18 data), <https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/986168/Weekly_Flu_and_COVID-19_report_w19.pdf> (2021).
Jeong, H. W. et al. Viable SARS-CoV-2 in various specimens from COVID-19 patients. Clin Microbiol Infect, doi:10.1016/j.cmi.2020.07.020 (2020).
Knowlton, S. D. et al. Bioaerosol concentrations generated from toilet flushing in a hospital-based patient care setting. Antimicrob Resist Infect Control7, 16, doi:10.1186/s13756-018-0301-9 (2018).
NHS. How to avoid spreading coronavirus to people you live with., <https://www.nhs.uk/conditions/coronavirus-covid-19/self-isolation-and-treatment/how-to-avoid-spreading-coronavirus-to-people-you-live-with/> (2020).
UNEP, W., UNCCD. Global Assessment of Sand and Dust Storms. United Nations Environment Programme, Nairobi., <https://wedocs.unep.org/handle/20.500.11822/7681> (2016).
Bondy, A. L. et al. Inland Sea Spray Aerosol Transport and Incomplete Chloride Depletion: Varying Degrees of Reactive Processing Observed during SOAS. Environ Sci Technol51, 9533-9542, doi:10.1021/acs.est.7b02085 (2017).
Clouston, S. A. P., Morozova, O. & Meliker, J. R. A wind speed threshold for increased outdoor transmission of coronavirus: an ecological study. BMC Infect Dis21, 1194, doi:10.1186/s12879-021-06796-z (2021).
Rendana, M. Impact of the wind conditions on COVID-19 pandemic: A new insight for direction of the spread of the virus. Urban Clim34, 100680, doi:10.1016/j.uclim.2020.100680 (2020).
Islam, N. et al. COVID-19 and climatic factors: A global analysis. Environ Res193, 110355, doi:10.1016/j.envres.2020.110355 (2021).
Xu, R. et al. Weather, air pollution, and SARS-CoV-2 transmission: a global analysis. Lancet Planet Health5, e671-e680, doi:10.1016/S2542-5196(21)00202-3 (2021).
Cai, Q. C. et al. Influence of meteorological factors and air pollution on the outbreak of severe acute respiratory syndrome. Public Health121, 258-265, doi:10.1016/j.puhe.2006.09.023 (2007).
Altamimi, A. & Ahmed, A. E. Climate factors and incidence of Middle East respiratory syndrome coronavirus. J Infect Public Health13, 704-708, doi:10.1016/j.jiph.2019.11.011 (2020).
Wang, Y. et al. Reduction of secondary transmission of SARS-CoV-2 in households by face mask use, disinfection and social distancing: a cohort study in Beijing, China. BMJ Glob Health5, doi:10.1136/bmjgh-2020-002794 (2020).
Peeples, L. Face masks: what the data say. Nature586, 186-189, doi:10.1038/d41586-020-02801-8 (2020).
Ramalingam, S. et al. Antiviral innate immune response in non-myeloid cells is augmented by chloride ions via an increase in intracellular hypochlorous acid levels. Sci Rep8, 13630, doi:10.1038/s41598-018-31936-y (2018).
Ramalingam, S., Graham, C., Dove, J., Morrice, L. & Sheikh, A. A pilot, open labelled, randomised controlled trial of hypertonic saline nasal irrigation and gargling for the common cold. Sci Rep9, 1015, doi:10.1038/s41598-018-37703-3 (2019).
Rabone, S. J. & Saraswati, S. B. Acceptance and effects of nasal lavage in volunteer woodworkers. Occup Med (Lond)49, 365-369 (1999).
Slapak, I., Skoupa, J., Strnad, P. & Hornik, P. Efficacy of isotonic nasal wash (seawater) in the treatment and prevention of rhinitis in children. Arch Otolaryngol Head Neck Surg134, 67-74, doi:10.1001/archoto.2007.19 (2008).
ONS. Population projections for local authorities: Table 2: 2018 based projections, <https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/populationprojections/datasets/localauthoritiesinenglandtable2> (2020).
Currell, G. Scientific data analysis. (Oxford University Press, 2015).

Table 1 is available in the Supplemental Files section.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Increase in COVID-19 downwind following a wind change

Status:

Version 1

Abstract

Figures

Introduction

Discussion

Methods

Declarations

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1