Strengths and limitations
Besides the modest sample size and somewhat limited external validity
due to loss to follow-up, primarily of families with low school
education, the largest limitation of our study is that it is based on a
single study area. Therefore, our findings cannot be generalized to
other areas. Future studies should further investigate our observed
associations in urban, suburban and rural contexts across different
climates and countries. Moreover, we cannot exclude residual confounding
of our findings by allergenic shrubs and herbs and even by trees on
private grounds, as these were not part of the tree registry. We
co-adjusted our models with tree exposures for general vegetation
degree, but this is not sufficient to rule out potential impacts of
other allergenic vegetation. Limited sample size together with
relatively low prevalence of asthma and AR precluded us from testing
more effect modifiers, in particular, socioeconomic status.
Nevertheless, LISA is an established cohort study meant to explore
various risk factors of allergic diseases. All allergic outcomes were
collected by using state-of-the-art methods at many timepoints. Repeated
measurements analysis provides enough power to detect even small
associations. The impact of confounding was reduced by controlling for
many perinatal and early postnatal allergy risk factors. Finally, we
were able to classify all tree genera into allergenic vsnon-allergenic by several different criteria, something that was never
done before in epidemiological analyses.