Exposure assessment
We measured exposure to greenness in the residential environment as our
exposure of interest. Greenness was measured using the mean Normalized
Difference Vegetation Index (NDVI) within 100 m, 300 m, 500 m, and 1000
m of the participant’s residence(s) during pregnancy and the first two
years of life in both spring and summer seasons., We
selected the 300 m radius for NDVI for the main analyses, as this has
been recommended by the WHO and UNICEF. The calculation of the NDVI was
based on land surface reflectance of visible red (VISR) and
near-infrared (NIR) wavelengths, applying the following equation (Eq.
(1)):
\(NDVI=\ \frac{NIR-VISR}{NIR+VISR}\) (1)
The NDVI values range from -1 to 1, where negative values represent
water, zero values rock, sand or snow (without vegetation), and positive
values indicate density of green vegetation (i.e., the highest NDVI
values are observed from photosynthetically active and healthy
vegetation). We used only cloud-free images during the spring
(April–May) and summer (June–August) seasons from Landsat 5 (spatial
resolution: 30 m) between conception (1983-90) and the first two years
of life (1984-92) to capture maximum spatial contrasts in greenness
(Figure 1). The used images represented the highest available quality
and processing level (https://earthexplorer.usgs.gov/). We included NDVI
values higher than zero to investigate the role of green vegetation on
the risk of asthma. In southern Finland, the spring usually begins in
April and the summer begins in late May
(https://en.ilmatieteenlaitos.fi/seasons-in-finland). ArcMap 10.5 was
used to process satellite images, and QGIS 3.8 was used to extract the
average NDVI within each buffer size.
We calculated individual-level cumulative exposure to NDVI (cumNDVI)
during pregnancy and early life (first two years of life) by taking the
average of monthly exposure over the residential history, while
weighting by the days spent in each residential address and time of
pregnancy in the spring and summer seasons. Estimates were based on the
number of pregnancy or early-life days during spring and summer [Eq.
(2)].
\(\text{\ Cumulative\ exposure\ to\ NDVI}_{\ residence,\ \ i}=\ \sum{\text{mean\ exposure\ to\ NDVI}_{j}*duration\ of\ exposure}_{j}\text{\ \ \ \ \ \ \ \ \ \ \ \ }\)(2)
where residencei corresponds to
ith participant, mean exposure to
NDVIj corresponds to the mean exposure to NDVI in
jth residential addresses during months 4-5
(April-May) for spring and during months 6-8 (June-August) for summer,
and the duration of exposurej corresponds to the
number of estimated days within specific month during spring/summer
periods in jth residential addresses during pregnancy
or early life.
We defined individual cumNDVI during the entire pregnancy, the first
trimester (defined as gestational months 1 to 3), the second trimester
(gestational months 4 to 6) and the third trimester (gestational months
7 to the end of pregnancy) by taking the average of exposure during each
pregnancy month. The child was exposed to the average spring NDVI level
during pregnancy months from April to May, to the average summer NDVI
level during pregnancy months from June to August, and to NDVI level 0
from September to March. The latter corresponded to a lack of exposure
outside of the growing season when NDVI values decrease significantly
for all types of land use in vegetation zones like southern Finland in
the Baltic Sea region.
We conducted a sensitivity analysis where mean exposure to
NDVIj corresponded to the mean exposure to NDVI in
jth residential address during the months 3-5
(March-May) for spring and the months 6-9 (June-September) for summer.