no association was found (Richards et al., 1981; Goldstein and Weinstein, 1986). Because the mix of pollutants varies, and a number of other factors affect hospital emergencies, such results are not unexpected.
Hospital studies that include all hospitals in the area can provide data comparable with those generated in population-based studies. Hospital-based studies are very relevant to future research on hazardous wastes and point sources of pollution in several ways. First, they provide broad support for the use of systematic and longitudinal symptom reporting, or hospital visits, as valid outcome measurements and investigative tools. Such reports are easier to obtain than are physiologic measurements, and in the case of respiratory illness they appear to be as useful in some cases as are measurements of function test change. Second, the critical feature of such studies is their longitudinal nature. Many of the difficulties that arise from reporting bias in using data from questionnaires administered once for hazardous-waste exposures (detailed in the next section) could be avoided by using monitored data of temporal fluctuations in exposure in a population, together with diary information. This might be more effective than depending on cross-sectional comparisons between groups of subjects who probably are aware of their relative exposure ranking and who therefore are prone to recall bias.
Recent analyses of the Health Interview Survey of the U.S. National Center for Health Statistics have permitted studies of associations between the number of days people report being restricted due to respiratory conditions and levels of air pollutants (Portney and Mullahy, 1986; Ostro and Rothschild, 1989). Strong associations were found with fine particles and weaker ones with ozone. The large numbers of subjects in these studies have permitted significant associations to be found. An important strength of their analyses is the random selection of the subjects.
Initial reports of the occurrence of mild symptoms in people working in sealed, usually recently constructed, office buildings were generally discounted. However, the syndrome has been firmly established for several reasons. First, a remarkable concordance has been found in the kinds of complaints made by workers in different loca-