Spatial indoor radon distribution in Mexico City.

We present a spatial analysis of residential radon concentrations in the Mexico City Metropolitan Area, which we intend to use to assign radon exposure in an ongoing case-control study. As part of a probabilistic household survey, carried out between May and June 1999, 501 dwellings were selected for indoor placement of solid state nuclear track detectors (LR 115) in a cup array over a period of approximately 90 days. As part of the sampling design, the city was grid partitioned into nine zones and a sample of dwellings was selected in each zone. All zones were simultaneously surveyed. The stratified sampling design allowed us to obtain radon geometric means, adjusted for household characteristics, week of detector placement and number of days of measurement for these zones. Additionally, adjusted geometric means were estimated for the 100 census tracts surveyed and this information was used to obtain a more detailed spatial distribution of residential radon levels through kriging interpolation and surface contouring. Radon levels depended on the room of placement, the floor level and the ventilation habits but not on building materials. Regarding the city zone, the highest adjusted geometric mean was found in the southwest (136 Bqm(-3)), where 46% of the households had an estimated radon level in excess of 200 Bqm(-3). In the rest of the city, the geometric mean concentration ranged between 41 and 98 Bqm(-3). A more detailed spatial distribution showed that, in general, most of the eastern and middle zones of the city had estimated radon geometric means below 74 Bqm(-3), while the western ones had geometric means above this concentration. Very high geometric means, exceeding 111 Bqm(-3) and even reaching 288 Bqm(-3), are estimated for some areas located in the southern and western zones of Mexico City. The obtained spatial distribution shows that the areas with very high estimated residential radon concentrations are close to inactive volcanic mountains. We believe that the geo-statistical techniques, we have used, offer reasonably good estimates of the average spatial residential radon distribution in Mexico City under average ventilation in homes. The use of this indirect approach for radon exposure measurement in epidemiological studies is an inexpensive alternative to direct radon exposure measurement but may be subject to non-differential misclassification error. The effect of such error on the detection of a real increase in lung cancer risk from indoor radon remains to be determined.