Indoor birch pollen concentrations differ with ventilation scheme, room location, and meteorological factors

Indoor pollen concentrations are an underestimated human health issue. In this study, we measured hourly indoor birch pollen concentrations on 8 days in April 2015 with portable pollen traps in five rooms of a university building at Freising, Germany. These data were compared to the respective outdoor values right in front of the rooms and to background pollen data. The rooms were characterized by different aspects and window ventilation schemes. Meteorological data were equally measured directly in front of the windows. Outdoor concentration could be partly explained with pheno logical data of 56 birches in the surrounding showing concurrent high numbers of trees attaining flowering stages. Indoor pollen concentrations were lower than out door concentrations: mean indoor/outdoor (I/O) ratio was highest in a room with fully opened window and additional mechanical ventilation (.75), followed by rooms with fully opened windows (.35, .12) and lowest in neighboring rooms with tilted window (.19) or windows only opened for short ventilation (.07). Hourly I/O ratios depended on meteorology and increased with outside temperature and wind speed oriented per pendicular to the window opening. Indoor concentrations additionally depended on the previously measured concentrations, indicating accumulation of pollen inside the rooms even after the full flowering period.

detailed information of the indoor pollen concentrations that have the more meaningful influence on humans compared to outdoor or even background pollen concentrations.
Although there are a few studies on indoor pollen or mold spores, [8][9][10][11][12][13][14][15][16] there is still a need for a comprehensive study that combines outdoor concentrations on rooftop level, ground level as well as indoor concentrations. 17 Most of the studies reported decreased values of pollen concentration inside buildings, but some studies also showed that these concentrations were not correlated with outdoor levels (eg, Ref. 12).
O'Rourke and Lebowitz 11 stated that atmospheric transport plays a negligible role for indoor pollen concentrations and identified feet and bodies of people and animals as main vectors. The great influence of pollen transport into houses via clothing was also supported by Jantunen and Saarinen. 18 Furthermore, indoor pollen concentrations were found to increase when rooms were more frequently accessed and outdoor activities of people were higher. 8,9 Equally for birch pollen antigens in dust, it has been suggested that they are carried indoors via footwear and clothes. 19 However, there is a lack of knowledge about how meteorological parameters are able to influence the indoor concentration of pollen in relation to its outdoor concentration. We additionally realize that little is known about differences caused by different ventilation schemes. A deeper understanding of the influence of window ventilation will allow a better adaptation of the individual behavior.
Therefore, this study aimed to answer the following questions: • Is there a significant correlation between outdoor and indoor pollen concentrations?
• Will indoor pollen concentrations and indoor/outdoor ratios change under different ventilation schemes and room locations?
• How do meteorological conditions, especially wind direction, influence the number of floating pollen in indoor air?

| Study site and rooms
The study was conducted inside and outside the forest faculty building of the Technical University of Munich at Freising, Germany (48°24′N, 11°45′E). The three-story building is situated at the western edge of the green campus area on which agricultural fields and forests border.
The edifice itself is surrounded by extensively managed meadows, hedges, and groups of trees comprising different species, including some birch (Betula pendula Roth) specimen ( Figure 1). Indoor (I) and outdoor (O) concentrations of birch pollen were assessed for five rooms in the building: three office rooms, one large combined laboratory/seminar room, and one small laboratory room (Table 1, Figure 1).
Ventilation schemes and other properties of the rooms are listed in Table 1. All rooms have a central heating system with heating elements under the windows.
In all rooms, the windows were opened for starting and stopping the respective outdoor personal pollen samplers (see section Pollen monitoring) which were placed on the window sills. Windows were closed at the end of the day. All five rooms, especially the office rooms, were frequently entered by co-workers, students, and the regular users. All sampling days except the first one (April 19, 2015, DOY 109) were working days. Thus, the experimental conditions were in accordance with real-life situations. The rooms East-Tilt and East-Vent lie directly next to each other; one tall birch tree that was flowering during the measurements is situated right in front of their windows. Air is aspirated at 10 L min −1 through a vertically oriented intake, and pollen is deposited on microscope slides that are coated with white and pharmaceutic Vaseline (Molyduval). Microscope slides were inserted every second hour for 60 minutes during 8 and 7 p, resulting in six measurements per day (8-9 , 10-11 , 12 -1 p, 2-3 p, 4-5 p, and 6-7 p). In total, the sampling campaign resulted in 480 pollen samples of which five had to be discarded due to failure in the sampling. To prepare permanent samples, we applied a mixture of distilled water, gelatine, gelvatol, and safranin (staining) to cover slips and fixed them to the microscope slides. The edges were sealed with common transparent nail varnish. Samples were assayed under a light microscope at 400× magnification (Axio Lab. A1 connected to a Motic Moticam 3, 3.0 MP; Zeiss Microscopy GmbH, Jena, Germany).

Practical Implications
• The assessment of indoor pollen is crucial for human wellbeing as people stay most of the day inside buildings.
Although restricted to one season and five rooms, our study demonstrated that outdoor pollen concentrations varied with room location and that weather and ventilation schemes strongly influenced indoor/outdoor ratios.
In addition to the wise choice of room location and

| Meteorological data
Meteorological Although in meteorology, wind direction is reported by the direction from which it originates (eg, southerly wind), for clarity in the analyses, we refer to the direction it is going to (eg, wind toward north).

| Statistical analyses
Linear regressions were performed to determine the combined influ-   where i = 1…235 is the observation id, and the rest as above. Residual plots showed that errors were heteroscedastic with respect to outdoor pollen concentrations; thus, we included a variance function that is an exponential of the outdoor pollen concentration with different parameters for each room. This corresponds to following formulation:

| Birch pollen and flowering season
The aerobiologically defined birch pollen season started on April 13th and ended on April 25th when 5% or 95% of the annual sum had been collected ( Figure 3). The birch pollen concentration sharply increased to its annual peak on April 16th with 1722 pollen grains m −3 .

| Outdoor meteorological conditions
Under the influence of a high-pressure system, the weather from April 19th to 22nd was predominantly sunny with only a few clouds on April 22nd. 21,23 Dry air masses from subpolar origin subsequently warmed up, and daily maximum temperatures increased up to ~20°C.
During nights, some frost was observed. Winds in this period were generally weak and toward west and south. On April 23rd, a small low-pressure system in the higher troposphere reached Bavaria from northwest and led to lower air temperatures. On the following day, it was sunny and warm again with weak winds toward east. There was light rain on the non-sampling days 25th and 26th. On April 27th, a low-pressure system moved over Bavaria and this was the only sampling day when 6 mm precipitation was registered. On April 29th, the weather was cooler but sunny again. The outdoor measurements ( Figure 4) reflected the diurnal patterns with maximum temperatures around midday or in the early afternoon, depending on the aspect (east-midday, south and west-early afternoon). On the south and sometimes also on the west side, the highest temperatures were recorded, corresponding to the lowest air humidity values. Air pressure records mirror the frontal system passing on April 27th.

| Wind speed and directions
Wind speeds were generally low during the sampling days, not exceeding the category of moderate breeze (Beaufort scale 4, 5.5-7.9 m s −1 ).
Opposite to the general pattern toward east, 20% of winds were also toward west ( Figure 5). The outdoor wind field in front of the win-

| Outdoor and indoor birch pollen concentrations
With respect to outdoor pollen concentrations, the eight sampling days can be divided into two periods: medium-to-high pollen concentration on the first four sampling days (April 19th till 22nd) of up to 600 pollen grains m −3 and low concentrations rarely exceeding 50 pollen grains m −3 on the last four sampling days (April 23rd, 24th, 27th, and 29th) (see Figure 6). This dichotomy is equally seen in the Background (building rooftop) and outdoor (window) pollen concentrations were highly correlated for all rooms (all P<.001, Table 2).
Outdoor and indoor pollen concentrations were highly correlated for South-Open and less for the other rooms, with no significant correlation for North-Open. The correlations between indoor and background concentrations were very similar to the indoor-outdoor correlations.

| Modeling
The significant explanatory variables for modeling indoor pollen con-

| DISCUSSION
Background pollen concentration and phenological observations largely matched; however, some small discrepancies are obvious which have to be carefully interpreted in light of a three-day temporal resolution of the phenological observations. The peak concentration West-Open, the value was smaller (.77), probably due to its largest distance to the roof trap and the building structure (see Figure 1).
For complex city structures, the representativeness of pollen traps is known to be limited. 27 The presence of high buildings and complex surfaces may, for example, increase turbulence, thereby causing pollen concentrations to differ considerably over short distances both vertically and horizontally. [28][29][30][31] For most allergic people, due to their living and working conditions, indoor pollen concentrations are more relevant than outdoor background concentrations and the suitable siting of office or living rooms and their ventilation matters. 13,32 How relevant these differences between outdoor and indoor conditions may be is underlined by our study. First symptoms in people allergic to Betula pollen occur when airborne pollen concentrations exceed ~20 pollen grains per m³, 33,34 a threshold which was exceeded on nearly all days of the birch pollen season (see Figure 3). This is the range which can be underrun by the "best performing" room in our study (16 pollen grains per m³ in East-Vent) that even has a birch tree at 5 m distance in front of its window.
The average I/O ratio of birch pollen grains found in this study (.33) largely matches results reported in the literature; especially the F I G U R E 8 Same as Figure 7, but for effects on the I/O ratio of pollen average rate for the four rooms which were not influenced by an installed ventilation system (.22) is in line with previous studies. 8,9,13,32 However, our study revealed distinct and significant differences between the five rooms.
The highest mean I/O ratio of .75 was found for South-Open, strongly supported by the highest correlation between indoor and outdoor pollen concentration (.95, see Table 2). This relatively small laboratory room is characterized by a high window/room volume ratio (see Table 1); thus, the constantly opened window allowed considerable exchange of air, boosted by the exhaust hood working at a very small extraction rate which obviously constituted an effective venti- however, their window to room volume ratio was similar ( Table 1) Table 1  date coefficients had P<.05) and were higher in the second-half of the sampling period (t tests comparing the first to the last four measurement days, all P<.05) During this time, both outdoor and indoor concentration of pollen grains decreased substantially. The linear model for indoor pollen concentrations revealed still an additional dependence on the previous concentration after accounting for outdoor pollen concentration; thus, most likely pollen grains also accumulated inside the rooms, which at the end may also influence I/O ratios.
As the sampling sites were not cleaned daily, pollen grains probably have settled down and accumulated over time. Even if the ground is cleaned, pollen grains from inaccessible areas of a room can be moved to open areas and appear in the samples beyond the pollen season. 38 In addition, pollen grains are able to accumulate in house dust. Thus, they can reach a peak even a long time after pollination season and maintain their antigenic activities until the next pollination season. 39,40 Yli-Panula 41 and Enomoto et al. 42