Results

Most of the correlations made between total pollen levels and the meteorological and pollution factors were weak and insignificant. This is likely because total pollen includes tree, grass, and weed pollen, which respond differently to different factors, produce different pollen levels, and have different pollen seasons. Some of the notable correlations that it did create were between total pollen and wind speed (r=.3, p=.008) and humidity (r=-.321, p=.06) by monthly average. These correlations indicate that total pollen counts are greater in warm, dry months, which aligns with our knowledge of when pollen is released. Wind picks up the pollen grains and carries them further away from the plant, therefore, an increased wind speed causes a greater distribution of pollen in the air. When there is less moisture in the air, this can also increase the levels of airborne pollen because pollen grains are not being picked up by water droplets. Yearly averages indicate that precipitation may play a role in helping plants grow to full potential before releasing pollen, though as previously stated, more collection of data must be done before yearly averages can be statistically relied upon.

PM₁₀ (particulate matter) was one factor that we expected to be correlated at the daily level with total pollen. We have been told by Dr. Leru and other researchers in the field that some pollen has the potential to be smaller than 10 microns in diameter, making it considered a PM₁₀ particle. Dr. Leru has informed us that Ambrosia pollen is too large for this confusion to occur, except in the case of extreme weather. We did not observe this effect with our total pollen and PM₁₀ daily (r=-.005, p=.89) and monthly (r=.107, p=.61) correlations. Individually correlating each species of pollen to PM₁₀ revealed that Artemisia and Ambrosia are the only types of pollen that could potentially affect the PM₁₀ count.

As previously stated, when one observes the correlations between certain categories, one must consider the context of how they are related over time. For example, plants need time to fully grow before environmental factors have the ability to impact pollen production. Daily and monthly temperature show little to no correlation with the level of Ambrosia pollen, while yearly correlations are better, but have a high p-value. The Ambrosia plants are well suited to produce more pollen as temperature increases, as we have studied in our background. Daily and monthly correlations do not reflect this statistic because rising temperatures is not a direct trigger for the ragweed plant to release its pollen. In fact, Ambrosia pollen production peaks from mid-August to September as the temperatures are cooling down. Yearly correlations are the way we would find this correlation to be true, but it currently has too high of a p-value to be considered statistically significant (p= -0.821, r= 0.088). As more data is collected over years, we expect to see p-values drop and yearly correlations with temperature becoming more reliable. Additionally, we expect to see a positive correlation based on extensive research of this relationship from other studies, as opposed to the negative correlation we observed. We cannot confirm that the correlations we found with temperature are accurate due to the patching of weather data from different sources for the beginning of 2014 and the end of 2019. Based on our data, these years appear to have drastically lower yearly average temperatures compared to the rest of the data, although the pollen counts were relatively high these years. Since the source of our temperature data is not consistent, we cannot verify that the temperature these years were truly lower than the others, and therefore cannot confidently report the relationship between temperature and ragweed pollen.

The strongest correlations we found between ragweed pollen and meteorological factors are with relative humidity (no average: r= -0.243, p=0; monthly: r= -0.389, p= 0.06) and precipitation (no averages: r= -0.229, p= 0; yearly: r= -0.8, p= 0.2) during the full pollen season. From these results, it is clear that there is a negative correlation with both of these factors; when there is less moisture in the air, the pollen count is higher. These correlations were not found to be very strong, although they exceeded the strength of many of the correlations with other factors. Unlike with temperature data, these results make sense in the context of daily and monthly correlations, since precipitation and humidity do not follow a defined pattern within the pollen season. This validates that we are seeing some statistically significant results that we expect, since it is known that the pollen is attracted to the water molecules, thus reducing the ambient pollen levels and the detected pollen count. On the other hand, we found no correlation between Ambrosia pollen counts and wind speed. Perhaps there were other factors present on windy days that had a stronger influence on reducing the pollen counts, such as precipitation, thus mitigating the increased pollen levels we expected to observe on windy days.

Of the chemical pollution factors carbon monoxide (CO) was the most strongly correlated to Ambrosia pollen counts, especially during the peak season of mid-August to mid-September (no averages: r=0.2 p=0.019, monthly: r=0.367 p=0.331, yearly: r=0.872 p=0.053). Since Bucharest is an urban area, it is exposed to higher levels of CO than rural areas due to the concentrated abundance of fossil fuel burning. While there is limited research available on how carbon monoxide may directly affect pollen production, it is possible that there is another factor that directly impacts the levels of both CO and ragweed pollen, making them linked to each other.

Other chemical pollutants appeared correlated to the ragweed pollen counts, but in context were not statistically significant and often fluctuated between positive and negative values in different situations. These inconsistencies are likely due to the small holes in the availability of the pollution data and patching of data from different stations to try and make up for this. Hopefully, the continued collection and improved consistency of data would help confirm these trends, but for now no further conclusions can be made.

Similarly to the ragweed pollen correlation results, most of the meteorological and pollution factors did not show a strong correlation with grass pollen levels when looking at the broader context. The most significant factors we found correlated to grass pollen were ozone (O₃) and solar radiation, specifically during the peak grass season from April 25 until the end of June. Both showed a positive correlation; O₃, solar radiation and grass pollen levels increase and decrease at similar times. While the strongest correlations with O₃ were found looking at yearly averages, the monthly averages were the only scenario with a strong correlation and a statistically significant p-value (yearly: r=0.6, p=0.208; monthly: r=0.521, p=0.056). While we cannot show causation between these factors and grass pollen levels, this correlation suggests that there may be some link between the O₃ levels or solar radiation and grass pollen over time. We can only speculate; we suspect that the link is potentially carbon dioxide (CO₂), although we did not have any data on this pollutant. CO₂ is a significant contributor to global warming, especially in urban environments. It converts to O₃ in the atmosphere, and has been shown to increase pollen levels, since it serves as an energy source for plants (Schmidt, 2016).

The only meteorological factor we found correlated to grass pollen was precipitation level (monthly average: r= -0.337, p=0.3403; yearly: r=-0.8, p=0.2). However, there was no statistical significance and the negative correlation could be more related to how particles are distributed in the air than the pollen production of the plant itself. Based on our research, other studies have found that grass pollen is not strongly correlated to meteorological factors (Ariano et al., 2010) and pollution is more related to allergy symptoms than pollen counts themselves. Therefore, our results for grass pollen were not so different from what we expected.

Regarding public interest, we found from Google Trends that the terms “Ragweed” and “Allergy”in Romania became increasingly popular over the past 5 years. This is the data we used to represent public interest in pollen and seasonal allergies.

We found many correlations between public interest and climate and pollution data. We treated the interest in allergies as a measure of how many people are affected by allergies, as well as how severe their symptoms are. Both interest in the terms ‘Allergies’ and ‘Ragweed’ aligned with the total pollen levels on the daily, yearly, and monthly scale.

1. Climate vs Allergy Interest

There are several meaningful correlations that can be drawn between climate factors and the level of pollen allergy symptoms observed by the Romanian population. The factors that positively correlate strongest with interest are solar radiation and temperature. A strong negative correlation exists between relative humidity and allergy interest as well. These three once again point to the idea that a hot, dry, sunny day will be the worst meteorological conditions for a pollen allergy sufferer. The most dry, hot, and low-wind months cause the most allergy interest.

Wind speed has a weak negative correlation with allergy interest. It points to the idea that the worst allergy symptoms come on days where the wind is not blowing. This effect could be due to wind in cities helping to diffuse pollutants, making the pollutants less concentrated and therefore less inflammatory during windy days. This can help explain why total pollen levels exhibited a positive correlation to wind while allergies show a negative correlation. These results further cement the idea that pollen allergies will be worse on a hot, dry summer or fall day. Knowing this is true in Bucharest allows for better predictions to be made about what days will be the worst for allergy sufferers.

2. Pollutants vs Allergy Interest

One of the most noteworthy findings is ozone’s strong correlation to the interest in allergies. While some of this correlation is due to seasonality, yearly (r=0.5, p=0.39), monthly (r=0.626, p=<0.005), and daily correlations (r=0.461, p=<0.005), O3 shows positive correlations. Ozone is known to have an irritating respiratory effect similar to allergies, and is also known to amplify the effect of pollen on the lungs, similar to combustion particles. This relationship in Bucharest could be investigated as a leading cause of irritation for allergy sufferers within the city. Bucharest’s ozone levels peaked at 136μg/m3 in 2016, exceeding the World Health Organization’s recommendation of 100μg/m3 for daily ozone (Zhang, Wei, & Fang, 2019).

Carbon monoxide, or CO, had an inverse effect: when carbon monoxide levels rise, interest in allergies falls. Again, this applies across the yearly(r=-0.6, p=0.28), monthly(r=-0.581, p=<0.005), and daily(r=-0.343, p=<0.005) scales. Pollen seems to be uncorrelated with the levels of CO, yet it exhibits an inverse relationship with interest in allergies. There is no documented research of CO’s effect on pollen symptoms.

Our research suggested that, like ozone, PM10 and PM2.5 are lung irritants that can lead to the development of Ambrosia allergies. Both PM10 and PM2.5 have been recorded many times above the legal limit in Bucharest. (Wall-Street, 2019). However, our PM10 data returned low or negative daily, monthly, and yearly correlation coefficients for both interest in ‘Ragweed’ and ‘Allergies’. Additionally, PM2.5 had a negative correlation to the interest in ‘Allergies’ regarding daily(r=-0.34, p=<0.005), monthly(-0.617, p<0.005), and yearly(r=-0.8, p=0.2) averages. These results suggest that neither PM10 or PM2.5 plays a role in aggravating allergies. PM10 and PM2.5 also have seasonal factors which come into play, raising the levels over the cold months. This may throw off daily and monthly correlations, however yearly correlations are also showing a strong negative correlation, forcing us to look deeper into why this is happening. Additionally, our research has found that the combination of ozone and PM2.5 exacerbates asthma symptoms, yet the link between pollen and the combination of these two pollutants has not been established (Gleason et al., 2014). Our results suggest that there is little synergy between ozone and PM2.5, because allergy interest is not increased by an increase in PM2.5.

We are not able to obtain data on the exact number of new people who develop an allergy. More data, including health data, is needed to obtain accurate correlations that can help to prove this connection. The correlations between the yearly average of PM10/PM2.5 and new allergy patients could help describe the development of allergies as a result of the pollutants.

CO, PM2.5, NOx, NO, and NO2 can all play into the creation of ozone via solar radiation, meaning that the higher levels of pollution in cities can increase ozone, thereby exacerbating allergy symptoms. This interaction could also explain why PM2.5 and CO are negatively correlated to allergy interest; they are consumed in the chemical process of creating ozone (Zhang, Wei, & Fang,2019). Knowing this, we conclude that ozone is the worst lung irritant, based on our results. The public remains largely unaffected by the raw PM2.5, CO, and NOx levels, according to our results. However, when the sun beams down and converts these volatile chemicals into ozone, the public seems to experience more severe and widespread allergy symptoms. Therefore, negative correlations found between NOx, CO, and PM2.5 should not allow for the dismissal of these factors as lung irritants, because their concentrations drop when they are turned into ozone, the worst lung irritant from our findings.

After compiling all of our results together, we compared our results to studies that have been done in other countries. These comparisons should be able to help us predict which correlations that we found will become stronger over time, or which correlations may not be true due to our limited amount of data.

Northern Italy

A study conducted in the Imperia Province of Italy by Ariano et al. examined correlations between pollen levels and temperatures in this region. They measured pollen from birch, cypress, olive, grass and Parietaria groups. Their data spanned over 27 years, starting in 1981 and ending in 2007. The meteorological data used in this study included yearly average values for temperature, rainfall, relative humidity, wind speed and solar radiation. The most important result from this data was that radiation and temperature both showed to linearly increase significantly over time. However, the total amount of grass pollen did not show a significant change over time, so temperature and grasses were not strongly correlated (r=0.02, p=0.06). There were no significant correlations between rainfall, humidity or wind speed with pollen in this data (Ariano et al., 2010).

In our project we did not find significant correlations between grass pollen with temperature, relative humidity, precipitation, wind speed or solar radiation. Our findings also showed that temperature and relative humidity increased almost linearly over time, while grass pollen remains relatively constant. The only significant factor we found correlated to grass pollen was ozone (O3) during the peak grass season. Based on this study that spans over 27 years, and how similar the findings were to our study, we would expect our data to continue this pattern in the future. This information may contribute to predicting grass pollen levels.

Poland

Another study which was conducted in the city of Szczecin, Poland investigated Spearman’s correlation coefficients between the Poaceae grass family and multiple meteorological factors. Similar to our project, they analyzed the correlation between Poaceae with wind speed, rainfall, relative humidity, SO2, ozone and PM10 and with data spanning from 2004 to 2008. Most notably they found statistically significant correlations between Poaceae and relative humidity (p=0.00, r=-0.39), between Poaceae and ozone (p=0.00, r=0.46), and between Poaceae and PM10 (p=0.00, r=0.32) (Puc, 2010). In our study we consider correlation above r=0.4 to be strong. Using this information, this study found the same results as we did, that only ozone correlates strongly with grasses. For the other two correlations, grasses correlated with relative humidity and grasses correlated with PM10, our project findings showed correlations of the opposite sign (+/-) as this study. We found the correlation between grasses and relative humidity to be positive (r=0.257), and the correlation between grasses and PM10 to be negative (r=-0.392) and both were weak correlations. This demonstrates that more data needs to be collected both in Poland and Romania before considering these correlations as significant.

Central Europe

Matyasovszky et al. collected data from 1995-2010 from 66 pollen stations across Europe including France, Austria, Hungary and seven additional countries. This data included Ambrosia pollen counts, Ambrosia season start and end dates, as well as climate factors such as temperature and precipitations. Then, they used linear regression analysis and determined that total pollen counts depended on maximum daily ragweed pollen counts, as well as the start and the duration of the ragweed pollen season. Pollen variables did not have any significant correlations with the daily mean temperature or precipitation (Matyasovszky et al., 2018). These findings are slightly different from our project considering that we did find a strong correlation between precipitation and Ambrosia pollen. This difference suggests that as more data is added, this correlation may become weaker, or that geography may be a contributing factor to this correlation.

United States

A study in the United States used linear regression for data collected in 1995-2009 to calculate correlations between Ambrosia season duration, number of frost free days, change in days to first frost, and latitude from Georgetown, Texas, US to Saskatoon, Canada. They determined that the duration of the ragweed pollen season correlates with the number of frost free days (r2=0.74), and also correlates with latitude (r2=0.95). This finding suggested that areas at higher latitude are more affected by climate change, shown by the increase in their frost free periods over years. With longer frost free periods, these areas also have longer Ambrosia seasons. For latitudes above ~44oN, the season increases as much as 13-27 days in 2009 compared to 1995. Even though analysis with temperature within our project was inconclusive, Romania has latitude of 45.9432° N, therefore, it is likely that over the years, with more data collected, we can also observe the increase in Ambrosia season duration due to climate change (Ziska et al., 2010).

Another study done on 5277 children in Southern California looking at the relationship between air pollutants (NO2, O3, PM2.5, PM10) and pollen allergies, showed that children living near major roads containing high levels of these pollutants have increased risk of respiratory disease, as well as deficits in lung function (p < 0.001). Children living within 500m of a major freeway are 10 times more likely to develop pollen related rhinitis if they are already sensitized to at least one pollen group. The pollen types examined belong to highly allergenic species from grass, weed, and tree pollen: Ambrosia, Olea europaea, Quercus agrifolia, Asteraceae, Phleum pratense, and Cynodon dactylon (Zhou et al., 2018). Based on the results of our project, we agree that pollutants can intensify pollen allergy symptoms, especially ozone.

Dr. Leru was interested in hearing what we might expect the future to hold for pollen allergies in Bucharest. We developed these prospective trends by observing what known events are taking place in Bucharest, understanding the implications of these events on the factors involving pollen allergies, and using our own correlations, and other researchers’ correlations to tell us what the effects will be in terms of pollen levels and allergy symptoms. We have looked at two major events taking place in Bucharest, climate change and pollution awareness.

Bucharest Climate Outlook

Based upon the correlations we have drawn above, there is research available that we may combine with our findings to produce predictions. The globe is currently warming in a widely accepted process known as climate change. Average yearly temperatures have been generally climbing as a result of global production of CO₂ and other greenhouse gasses. A simple linear regression of the past 30 years of climate research shows a warming of 0.08℃ per year since 1990, meaning that by the year 2030 the average temperature in Bucharest could be around 11℃, if this trend were to continue. Humidity is also expected to drop according to our regressions.

Studies from the USA have not only shown that the number of frost free days correlates to the amount of pollen produced in a season, but also that Ambrosia is able to generate more pollen under high CO₂ conditions (Rogers, 2006). This link is well established and will be observable as time continues. We can therefore expect that as the climate continues to warm, more pollen will be generated. The CO₂ levels in Romania, as well as more data pollen levels of Ambrosia in Romania would help bolster this claim, as we were not able to observe it directly with the limited Ambrosia data available. Global CO₂ levels continue to rise, meaning that the CO₂ in Romania will also rise. Under these conditions, Ambrosia will almost certainly continue to spread and grow in the Bucharest region, and we expect an increase in the amounts of pollen these plants produce.

Hot, dry days have been shown by our research to be the worst days for pollen allergy symptoms. The lower average humidity and higher average temperatures we have crudely observed will produce more of these days and therefore worsen allergy symptoms in the future.

Bucharest Pollution Outlook

Currently, Bucharest’s levels of pollution are extremely high. As stated previously, the level of PM_2.5 and PM₁₀ both exceed EU legal limits. If the level of PM₁₀ and PM_2.5 were allowed to continue to grow, they will continue to cause lung irritation which can lead to the development of pollen allergies. Volatile PM_2.5 particles can also play a part in the generation of ozone, an additional lung irritant which we have correlated strongly with the public interest about allergies. While we do not have enough data to firmly claim whether PM₁₀ and PM_2.5 continue to rise, we are able to look to the recent interests in both. Recent action has been taken by Romanias to attempt to reduce the amount of PM₁₀ and PM_2.5 pollution. Two independent pollutant monitoring stations have been established to better present information to the public eye, Airly and Aerlive. Both companies present a user-friendly way to view live, forecasted, and historic pollution data on a map of Romania. The Airly company provides over 30 pollutant monitoring stations in Bucharest alone, and has a network across the country.

Recent pollution activism in Bucharest has begun due to a spike in PM_2.5 in March, reaching a level 10 times the legal limit. Protestors argue the lack of public transportation, green spaces, and regulations on waste has led to this problem. The government has pointed to recent fires as the cause of this spike in pollution (Gherasim, 2020). Additionally, a tax on operating old cars was removed by the mayor of Bucharest. We expect that as older cars stay on the roads, larger amounts of pollution that they produce will lead to higher levels of PM_2.5 and other harmful pollutants such as NOx.

As pollution has been brought to the attention of the Romanian public via these independent companies, we observe legal action being taken to reduce the amount of pollution in Bucharest. The European Union has begun the process to hold Romania responsible for its pollution in Iasi, Bucharest, and Brasov (Gherasim, 2020). This may force Romania to begin working toward reducing its pollution levels. We hope the levels of air pollution will fall as Romania’s public and leadership become more aware of their pollution problems and begin working together to help reduce the amount of pollution. Reducing the amount of pollution will help alleviate and reduce the yearly growth in the number of allergy patients that allergists in Romania have observed over the past years.

COVID-19 and Pollution

The correlation between pollution and pollen allergies is a trend that must be mapped over decades. The COVID-19 pandemic is a rapid development in comparison. Airly has conducted studies comparing air quality to the areas of outbreak of the COVID-19 virus and found that average annual PM_2.5 was highest in the most infected areas in Europe, especially Northern Italy. It is believed that PM_2.5 may increase the infectivity and death rate of the COVID-19 virus, as the related SARS virus exhibited similar characteristics of being more severe for people in polluted regions (Airly, 2020). Figure 19 below shows the annual averages of PM_2.5 data from the B-1 air quality measuring station in Bucharest. It aligns with the map data presented in the study which demonstrates that while relatively speaking, the annual pollution in Bucharest is not nearly as bad as Northern Italy, it is still categorized as moderate to severe. It has not been clearly stated whether chronic or acute PM_2.5 exposure is responsible for making the lungs more vulnerable to a COVID-19 infection. This distinction is important because PM_2.5 counts will decrease over the summer and have decreased annually, yet PM_2.5 spikes can come at any time due to fires or excessive traffic. Regardless, ensuring that there are measures in place to protect the population from pollution spikes and continuing to decrease the average amounts of PM_2.5 would certainly be an action to protect against COVID-19.

We distributed the survey to the Romanian Ambrosia allergy sufferers Facebook group. We received 92 responses and all participants are from different parts of Romania, mostly the Muntenia region where Bucharest is located. The age of participants ranges from 18 to 74, with 74% being female. The majority (93%) of the participants are allergy sufferers, among which 87% are affected by pollen (including ragweed, trees, and grasses), 28% are affected by dust mites, and the rest are affected by other allergens such as animals, chemicals, food, metals and histamine intolerance. Many people have more than one type of allergies.

Sneezing, ocular itching, and nasal conditions such as runny nose, nasal congestion, and itchy nose are the most common symptoms, with a few sufferers experiencing rashes or hives. They reported to experience these symptoms mainly from August to October (Figure 20), which is the Ambrosia season, suggesting that allergy to Ambrosia is the most common. This finding is biased because our respondents are from the Ambrosia sufferer group. However, seeing that people have their symptoms even outside of the ragweed season suggests that many are affected by more than just ragweed alone. Additionally, coping with these symptoms in Romania is largely done by visiting a doctor, followed by over-the-counter medicine (Figure 21).

The majority reported no history of allergy from personal or parental records (Figure 22). Instead, our respondents developed their allergies within their lifetimes. 68.5% have been suffering from their symptoms for over 3 years (up to survey time). Even though genetics and family history plays a role, the exact numbers and biomechanics for this is still uncertain (Agnew et al., 2018). Our collaborator, Dr. Leru, also mentioned that personal atopy is more typical in grass pollen, not so much for Ambrosia, and sensitization to Ambrosia usually happens after childhood. As the majority of the respondents are Ambrosia sufferers, this might be a biased result.

Instead of genetic causes, there has been more evidence linking allergies to environment factors, including household conditions and outdoor exposure to chemical pollutants. Our survey showed that 30% of allergy sufferers reported that they smoke tobacco products, 25% have lived in a space with mold (Figure 23), 52% have or used to have pets, and 63% lived in an urban area during the age of 0-6.

We also found that 52% choose cars as their primary mode of transportation. However, 54% of them only spend on average less than 1 hour driving per day (Figure 24), so it is unclear whether Romanians’ car usage has an impact on pollution level.

Bucharest is a highly congested city with abundant air pollutants that are harmful to allergy sufferers (Wall-Street, 2019). The suburbs directly surrounding cities are most affected due to their combination of unmaintained growing space, urban pollution, and higher amount of plant life than the densely populated city. This is further evidenced by Google Trends data, where Ragweed was searched in Ilfov county, which surrounds Bucharest, more than anywhere else in Romania in the past 5 years (Figure 25).

The interest in allergy symptoms align with the height of ragweed season, with a secondary peak in March, indicating allergies to other pollens such as trees and grasses. Transportation statistics, however, did not provide conclusive data to link vehicles with allergies. To be able to make a concrete correlation, a more thorough study about vehicle usage over time compared to chemical emissions and allergy symptoms will need to be conducted.

Another aspect in our survey is diet. Studies in our Background linked a poor diet to the likelihood of developing an allergy. However, our findings regarding diet did not align with those studies. Of the allergy sufferers surveyed, 90% reported to not have a poor diet (Figure 26), with food choices from 6 categories (Vegetable, Fruit, Grains, Protein, Dairy, and Oils/Butter). A few reasons that perhaps explain this mismatch could be the inaccuracy in self-evaluation, the broad definitions for each food category (which do not specify elements of a poor diet), and perhaps focus on diet during the period of the development of allergies rather than current diet.

Another important finding is that 75% of Romanians experience their symptoms at work (Figure 27). This finding is significant because it means allergies can affect their productivity, the quality of their work, as the symptoms can be disruptive or may cause them to leave work for doctor visits, and sometimes even their own safety. It especially stresses the importance of more research to help people cope with the effects of seasonal allergies.

These findings give us an insight into what Romanian allergy sufferers are facing, however, the amount of responses we received was minimal and targeted to Ambrosia sufferers, making our results somewhat biased towards those types of respondents. Therefore, more studies are necessary to have results that are representative of the Romanian population, especially for different sub-groups of Ambrosia and Grass allergy sufferers.