not exist; where available they can provide reliable evidence of long-term effects.

Cross-sectional community studies typically compare communities with different levels of air pollution or populations that live different distances from a hazardous-waste site. All such studies have several problems: Measurement error can occur when the same exposure is assumed for every subject within a group. There are undetected differences between communities for risk factors. There can be “recall bias” if one group knows it is in the high-exposure category. There is little standardization of the equipment used to measure exposure in different locations. Recent studies, such as that by Flessel et al. (1991), suggest that more detailed population exposure measurements over time, in this example to DNA adducts of polycyclic aromatic hydrocarbons, may permit more precise analyses of long-term consequences.

In spite of these difficulties, successful community studies have been done. In contrast to ecologic studies that lack information on potential confounders, community studies generally have collected individual data on nonpollution risk factors. One early study (Holland and Reid, 1965) compared lung function in groups of male postal workers in London and in a number of provincial towns where levels of particulate matter and sulfur dioxide were much lower. In all smoking categories, they found a significant decrement of pulmonary function, measured as Forced Expiratory Volume 1 (FEV1), in the London workers. A more recent study (Groupe Cooperatif PAARC, 1982) found significant regressions in men, women, and children of FEV1 against sulfur dioxide pollution levels in 10 French cities. In the U.S., the best example of such studies is the six-city study that found a strong association between respirable particles and prevalence of acute bronchitis in children, although no differences were found in FEV1 (Dockery et al., 1989).

Other studies of this kind, including one report from Israel (Goren and Hellman, 1988), document that different respiratory symptoms occur in communities with different pollutant exposures. A report from Finland (Jaakkola et al., 1990) indicates that malodorous emissions from kraft pulp mills are associated with eye, nasal, and respiratory symptoms, although reporting bias cannot be ruled out in this case. Schwartz (1989) presents an analysis of pulmonary performance,