• air pollution;
  • asthma;
  • generalized additive models;
  • pollen;
  • relative risk

Airborne pollen levels have been associated with an increase in asthma morbidity (1). This relationship has usually been formulated in terms of a linear form. However, effects of environmental exposures on health could also be nonlinear (2). Approximations based on linear modals present limitations making arbitrary assumptions about shape of the relationship; also, categorical analyses usually have low efficiency and cut-off points are in most cases opportunistic (3). To solve these limitations, a nonparametric method to compute point-wise estimation of non-linear exposures has recently been proposed (4).

Under a Generalized Additive Model (GAM) a nonparametric estimate of the relative risk (RR) can be defined as inline image, where xref is the reference value of the continuous exposure and f(x) any of usual smoothers. Asymptotic variance of ln(RR(x,xref)) can be expressed in terms of the covariance matrix of the smoother f(x) as

  • image

Figueiras and Cadarso-Suárez (4) provide exact formulae estimation.

We applied this method to derive GAM-based risk estimates for the effects of airborne pollen levels on asthma emergency room (ER) admissions in Madrid, Spain. Daily number of asthma ER admissions from Gregorio Marañón Hospital was studied for 1995–1998. Pollen types collected were those with the highest allergenic capacity in Madrid (Olea, Plantago, Poaceae, and Urticaceae). Data on air pollutants (PM10, SO2, NO2 and O3), temperature and relative humidity were also collected. Additional information has been described elsewhere (5). Trend and seasonality were fitted using a cubic smoothing spline with 72 degrees of freedom (df). Cubic smoothing splines with 2 and 4 df were included to adjust for humidity and temperature, respectively. Dummy variables for day of the week and holidays were also fitted. Air pollutants were adjusted using cubic smoothing splines with 2 df for PM10 (lag 3) and SO2 (lag 3), 3 df for NO2 (lag 3), and 1 df for O3 (lag 1). Finally, all airborne pollen levels were adjusted using cubic smoothing splines with 3 df: Olea (lag 1), Plantago (lag 2), Poaceae (lag 2) and Urticaceae (lag 1).

Figure 1shows evident non-linear relationships between airborne pollen levels and risk of asthma ER admissions. Reference values were taken as the 90th percentile, as it is a minimum risk value (5) (Olea: 6.3 grains/m3, Plantago: 15.3 grains/m3, Poaceae: 27.8 grains/m3, and Urticaceae: 6 grains/m3). Airborne pollen levels with statistically significant (P < 0.05) risk were those placed between 30 and 60 grains/m3 of Plantago, more than 200 grains/m3 of Poaceae, and between 10 and 15 grains/m3 of Urticaceae.


Figure 1.  Relationship between airborne pollen levels and risk of asthma emergency admissions [log relative risk, ln(RR)] in Madrid, Spain (1995–1998). Vertical lines represent the reference value placed at the 90th percentile of type of pollen.

Download figure to PowerPoint

The use of nonparametric methods provided more informative estimations than traditional linear approaches for continuous environmental factors. The advantage of this innovative approach is that we do not need to make any a priori assumption about the functional form of the relationship. In comparison with categorical analyses, this method can be more efficient (3), also removing the need to select cut-off points for categories in an opportunistic manner. We observed evident nonlinearities for the estimated relationships between airborne pollen levels with allergenic capacity and risk of asthma emergency room admissions, updating our previous results (5) and being similar to those described for grass pollen in England (6). Note that GAMs are very conservative at the extremes of the curve, having wide confidence intervals, as the Fig. 1 shows. This is because usually there are few observations at the extremes of the exposure factor, and because of a slight boundary effect inherent in all flexible methods. Furthermore, results can be presented only in relation to a reference category. Then a graphical distortion can lead to misinterpretations as well as the arbitrary selection of the reference point (3). However, the highest risk of asthma morbidity in Madrid was that related to the highest airborne pollen levels when abruptly released into Madrid’s environment.


  1. Top of page
  2. Acknowledgments
  3. References

This study was funded by the Advisory Committee to the Madrid Regional Asthma Prevention & Control Programme (Comisión Asesora del Programa Regional de Prevención y Control del Asma de la Comunidad de Madrid). Aurelio Tobías had a postgraduate fellowship of Universidad Autónoma de Madrid. Marc Saez was partially funded by Fondo de Investigación Sanitaria, FIS 00/0010–02.


  1. Top of page
  2. Acknowledgments
  3. References
  • 1
    Anderson HR, Ponce de Leon A, Bland JM, Bower JS, Emberlin J, Strachan DP. Air pollution, pollens, and daily admissions for asthma in London 1987–92. Thorax 1998;13:842848.
  • 2
    Schwartz J, Ballester F, Saez M, Perez-Hoyos S, Bellido J, Cambra K et al. The concentration-response relation between air pollution and daily deaths. Environ Health Perspect 2001;109:10011006.
  • 3
    Greenland S, Michels KB, Robins JM, Poole C, Willett WC. Presenting statistical uncertainty in trends and dose-response relations. Am J Epidemiol 1999; 149: 10771086.
  • 4
    Figueiras A, Cadarso-Suarez C. Application of nonparametric models for calculating odds ratios and their confidence intervals for continuous exposures. Am J Epidemiol 2001;154:264275.
  • 5
    Tobías A, Galán I, Banegas JR, Aranguez E. Short-term effects of airbonne pollen concentrations on asthma epidemic. Thorax 2003;58:708710.
  • 6
    Newson R, Strachan D, Archibald E, Emberlin J, Hardaker P, Collier C. Acute asthma epidemics, weather and pollen in England, 1987-1994. Eur Respir J 1998;11:694701.