Pollen Trends and Respiratory-Related Emergency Department Visits, Sarasota, Florida
Pollen data are challenging to obtain due to the cost of monitors, the expertise needed to identify species, and the dedication required to regularly, if not daily, count pollen concentrations year after year. There are relatively few long-term pollen monitoring stations in the Eastern United States. A study of the Washington DC metropolitan area from 1998 to 2007 showed an increasing trend in annual tree pollen production, but a study of the New Jersey-New York metropolitan area from 1993 to 2002 showed declining annual pollen production for tree, grasses, and weeds. These apparently contradictory results for a similar time period in relative proximity may illustrate the influence of both increasing temperatures and changing land cover due to urban development. As part of a larger initiative to understand the health effects associated with climate hazards in Florida, we sought to examine the relationship of pollen counts and climate with rates of respiratory-related emergency department (ED) visits in Sarasota, Florida.
Methods: For 2005-2012, daily counts of ED primary complaints for Sarasota County residents treated in Sarasota County facilities were obtained from the Florida Agency for Healthcare Administration. Visits were classified as all respiratory disease (ICD-9 code 460-519), asthma (ICD-9 code 493), and non-asthma (ICD-9 code 460-519 excluding 493). For 2001-2012, daily temperature and precipitation data were obtained from the Sarasota Bradenton Airport through the National Climatic Data Center. Daily total pollen counts from 2001-2012 were collected by Mary Jelks, MD and used with permission from the American Academy of Asthma, Allergies, and Immunology. These pollen data are impressively complete; with the exception of 2012, where only 314 out of 366 daily readings are present (86% complete), all other years are over 94% complete. Over 25 species of pollen were identified, in addition to daily total counts.
Significant interactions were detected between month and all environmental predictors. All analyses were stratified by month. Daily ED visits were modeled as the dependent variable, and Poisson regression was used to test the association with pollen when adjusting for maximum temperature and precipitation. Results: Pollen season is often defined as starting on the day when one percent and ending on the day when 99 percent of the total pollen for the calendar year is observed. Using this definition, between 2001 and 2012, the pollen season in Sarasota averaged 304 days, beginning in mid to late January and ending in November. According to the records, the longest season was in 2012; however, it is possible that this extended season is due more to missing records rather than a truly longer season. There are some concerns raised by this method for describing a “pollen season” in south Florida; due to very infrequent freezes and moderate temperatures, plants grow year round. Between 2001 and 2012, the peak of pollen activity, defined as either the mid-date (50% of total annual pollen) or the date of the highest daily pollen count, usually occurred in early- to mid-March. In 2006 and 2008, these are the same day, and in all years studied, the two dates are within ten days of each other.
From 2001-2012, there was not a clear or statistically significant linear trend in total pollen counts by year. There are, however, distinct monthly trends. February, March, and April, when compared to December, have significantly higher daily total pollen counts (p< 0.0001). When looking at daily or monthly data, 2011 stands out as a year with extremely high pollen counts. The five highest daily pollen counts (all over 10,000 per cubic meter of air) were all recorded in this year, including a record total count of 35,525 on February 25, 2011.
Pollen was significantly and positively associated with all respiratory ED visits in Sarasota County during April through September and December (all p-values ≤ 0.01). Non-asthma respiratory visits mirror this trend, while pollen was only significantly associated with asthma visits in April (p< 0.0001). While the rate ratios are small, an increase in 100 total pollen particles per cubic meter of air would equate to an average daily increase of 1% in any respiratory-related ED visits during April or a 2% increase in May. Later in the year when pollen counts are lower, an increase of 10 total pollen particles per cubic meter of air would equate to an average daily increase of 5% in any respiratory-related ED visits during August. The negative association between respiratory-related ED visits and pollen counts found in January suggests that respiratory-related ED visits decrease as pollen counts increase in this month. This finding may be related to the variable start of the pollen season, usually in January, or due to other climatological factors.
Conclusions: There is a complex relationship between pollen counts, temperature, precipitation, and health outcomes. Obtaining and describing historical pollen data are critical to understanding the contributions of environmental factors on respiratory health outcomes. While Sarasota, Florida may not be representative of the state in terms of length of the pollen season and maximum daily counts, the demonstrable associations with respiratory-related health outcomes is consistent with the literature and evocative of the differences in the growing season/exposure period for residents of south Florida compared to much of the rest of the United States. Work to further analyze the Sarasota pollen record by species and to compare this location to others in the state is ongoing, but dependent on data availability.