Follow-up of schoolchildren in the vicinity of a coal-fired power plant in Israel.

This study was carried out in the framework of a health monitoring system set up in the vicinity of a 1400 megawatt coal-fired power plant in Israel. Second- and fifth-grade school children were followed up every 3 years; they performed pulmonary function tests (PFT), and their parents filled out American Thoracic Society-National Heart and Lung Institute health questionnaires. Among the cohort of second graders (in 1983) living in the area expected to be most polluted, a significant increase in the prevalence of part of the respiratory symptoms (such as cough and sputum, wheezing with and without cold and wheezing accompanied by shortness of breath) was evident in 1986. The prevalence of asthma among fifth graders in this area doubled (p = 0.0273) compared with prevalence when they were second graders. Among the children from the older cohort (fifth graders in 1983) living in this community, a similar although milder trend could be observed, especially in regard to an increased prevalence of asthma in 1986 compared with 1983 (13.9% versus 8.1%). Annual increases in PFT in the four groups of children (boys and girls from both cohorts) were found to be higher in the community expected to be polluted (especially in the younger cohort) compared with the two other communities. The discrepancy between the increased prevalence of respiratory symptoms and diseases and the higher annual increase in PFT among children from the expected more polluted community may be partly attributable to differential annual increase in height and to different distribution of background variables (such as socioeconomic status, passive smoking, heating, and respiratory diseases among parents) in the three communities.


Introduction
The first coal-fired power plant in Israel is located on the seacoast in a semi-rural region midway between Tel-Aviv and Haifa. The plant consists of four 350 megawatt units. In 1981 the units were gradually put into operation until the summer of 1984, at which time all were operating. The permit to build and operate this power plant was given provided that three monitoring systems, environmental, agricultural, and long-term health monitoring, be set up near the plant. An advisory epidemiological committee decided on a health monitoring system, which included four health surveys: mortality survey, requests for health services, schoolchildren health survey, and an adult panel study (1). This paper deals with the results ofthe third round of tests carried out among schoolchildren. The comparison ofthe data from the two earlier rounds are discussed elsewhere (2).

Subjects and Methods
A follow-up is regularly conducted in three cohorts of schoolchildren, i.e., second-, fifth-, and eighth-grade pupils *Research Institute for Environmental Health, Ministry of Health and Sackler School of Medicine, University of Tel-Aviv, Israel. tEpidemiology Unit Faculty of Health Sciences, Ben Gurion University ofthe Negev, Beer-Sheva, Israel.
Address reprint requests to A. I. Goren, Research Institute for Environmental Health, Ministry of Health and Sackler School of Medicine, University of Tel-Aviv, Israel. living in three communities with different expected levels of air pollution located within 19 km of the power plant (Fig. 1). Baseline data for this study were gathered in 1980, before the first unit was put into operation. The second data set was gathered in 1983 when two units ofthe plant were operating, and the third set was gathered in spring 1986-when all units had operated for at least 2 years. The health questionnaire used in this survey is a translated version of the American Thoracic Society-National Heart and Lung Institute (ATS-NHLI) self-administered questionnarie (3), to be completed by the children's parents. The questionnaires were distributed to the children by the school nurses who also collected them after they were completed. After receiving these questionnaires, the survey technician conducted pulmonary function tests (PFT) in the participating schools using a portable, digital spirometer (Minato AS-1000). The PFT parameters obtained were forced vital capacity (FVC), forced expiratory volume in 1 sec (FEVy.o), FEVI. /FVC, and peak expiratory flow (PEF).
The pupils performed the test standing and repeated the maneuver at least three times until two similar tests (within 10%) were obtained. The best test (highest FVC + FEV1.0) was chosen. In each round the same technician carried out the PFT in all the participating schools. The school nurses weighed the children and measured their height. All tests were done between April and June during the morning hours. Statistical analysis was carried out using the SPSS program (4). Prevalences of background variables, respiratory symptoms, and diseases among children from the three communities with expected different levels ofair pollution were analyzed by means ofchi-square test for examination ofindependence between two variables. The annual changes in pulmonary function test (PFT) among boys and girls in the studied cohorts living in the three communities were analyzed using one-way analysis of variance. Regression of annual changes in PFT on annual changes in height was carried out.

Results
In Table 1 a summary of the schoolchildren population by grade and year of examination is presented. Of the second and fifth graders in 1983, 915 and 885, respectively, were reexamined in 1986. No difference in the prevalence of background variables (e.g., socioeconomic status, parental smoking) could be observed between followed-up children and those who were not reexamined. Changes over time were analyzed only among children who were reexamined.
In the younger cohort, e.g., second graders in 1983, the prevalence of most respiratory symptoms became lower when children grew up in the community expected to be least polluted. In the expected medium-polluted community, part of the symptoms became more prevalent and part of them less common. In the community expected to be polluted most, symptoms became more prevalent; the change over time was statistically significant (p = 0.0012) for cough and sputum ( Table 2). As regards the temporal changes in the prevalence of respiratory diseases for children in the younger cohort, almost all of them became less common in the community expected to be low polluted, whereas the changes over time in the other communities were mixed, part of them became less and part of them more common.
It is interesting to stress the significant rise in the prevalence of asthma in the community expected to be polluted (Table 3). For the older cohort-fifth graders in 1983, the temporal changes in the prevalence of respiratory symptoms in the three communities were somewhat different. Most respiratory symptoms became less common in the community expected to be medium polluted, while in the other two communities the trend was mixed ( Table 4). As regards the temporal changes in the prevalence of respiratory diseases among the children ofthe older cohort, most ofthem tended to be less common in the communities expected to be low and medium polluted, while inthe community expectd to be polluted, part of them, as asthma, became more common ( Table 5). As can be seen from Table 6, the annual increase in FVC, FEVI .o, and PEF was higher in the older cohort compared with the younger one in the three studied communities. The annual increase in FVC and FEVI.o in the four subgroups was highest in the community expected to be most polluted. For boys and girls from the younger cohort, the lowest annual increase in FVC and in FEVy.o was observed in the community expected to be least polluted, while for the older cohort the lowest annual increase was observed in the community expected to be medium polluted. The annual increase in PEF in the younger cohort did not differ significantly between the three communities. The differences in the older cohort were statistically significant only among boys; the highest annual increase in PEF characterized boys from the community expected to be least polluted.
The annuai increase in height was highest in the community expected to be most polluted and lowest in the community expect to be medium polluted in the four studied subgroups (boys and girls from the two cohorts). By means ofregression analysis association of towns and fed into a computer that stores and ofannual changes in PFT on annual changes in height, we tried analyzes them. As can be seen from Table 7, the range of monto explain the temporal changes in PFT in the three communities. thly average concentrations for SO2 are between 0 and 22 tglm3. In three subgroups ofchildren the annual changes in height ex-The highest monthly concentrations for SO2 characterize the plained only 4 to 8% of the variance in PFT in the three com-community expected to be medium polluted. CO is not regularly munities. Only among boys from the older cohort could the an-measured in Pardes-Hana, and for the other two communities nual changes in height explain about 40% ofthe variance ofanmeasurements were available only for January to August 1985. nual changes in PFT.
Here again the values in the city of Hadera are considerable The association oftowns for environmental protection operates higher compared with the other community. As regards NO,, a network of 12 air monitoring stations in the radius of 25 km values they do not seem to vary considerably among the three from the power plant. The monitoring stations are fully communities. automatic and measure the following parameters: SO2, NO, NO2, and total NO,,; part of them also measure total hydrocar-Discussion bons, 03, CO, and TSP (total suspended particulate matter). Atmospheric data such as wind speed, wind direction, temperature, The community expected to be medium polluted, as a result relative humidity, barometric pressure, solar radiation, and ofthe operation ofthe power plant, is a city, while the other two precipitation are also collected. The instruments are automatical-communities are more rural. Air pollution levels measured in the ly calibrated, and the data sent by radio to the center of the vicinity of the power plant are extremely low. Hadera, the community expected to be medium-polluted, are apparently a result ofurban pollution rather than a result ofemissions from the power plant. It seems that the expectations of "low," "medium," and "high" pollution were not met.
The distribution of respiratory symptoms and diseases among children from three communities with different expected levels ofpollution seems to be connected with several other factors (5) rather than with air pollution levels, which were very low in all three communities. The community expected to be least polluted is characterized by lowest socioeconomic status (by crowding index, lack ofheating ofhouses, fathers' and mothers' education) in both cohorts. The prevalence of respiratory symptoms and most respiratory diseases among children ofthe younger cohort growing up in this community was, in 1983, much higher compared with the prevalence among children from the other two communities. In the older cohort this trend was less obvious. It seems that adverse health effects of low socioeconomic status diminish with age ofexposed children. This fact partly explains the sharp decline in the prevalence of respiratory symptoms and diseases in the expected low-pollution community. Respiratory problems among parents ofchildren growing up in the community expected to be most polluted became more prevalent in both cohorts. This could be a partial explanation of the rise in the prevalence of asthma and wheezing among their children.
The annual increase in FVC and in FEVy.1 in the four subgroups (boys and girls from both cohorts) was highest in the community expected to be most polluted. Since the annual increase in height was also highest in this community, we made a regression analysis of annual changes in PFT on annual changes in height. Only among boys from the older cohort could the annual changes in height explain a relatively high proportion ofthe variance of annual changes in PFT in the three communities.
Still, a discrepancy between the rise in prevalence of most respiratory symptoms and diseases among children from the younger cohort (and to a certain extent from the older cohort) growing up in the community expected to be most polluted and the highest annual rise in FVC and FEV1.o among the same children exists. This annual rise in PFT can only partly be explained by annual changes in height. The lowest annual increase in FVC and in FEVy *0 among children from the younger cohort growing up in the community expected to be least polluted may be partly attributed to the lower socioeconomic status in this community, the effect ofwhich diminishes with age. The lowest increase in FVC and in FEVI.0 among children from the older cohort living in the community expected to be medium polluted may be partly attributed to the lowest rise in height among them and to a city effect in this community. In the analysis carried out in 1983 (2), the lowest increase in respiratory symptoms characterized the children growing up in the community expected to be most polluted. On the other hand, the rise in their PFT (FVC and FEVy. ) as well as the increase in their height was lowest compared with the other two communities.
These are preliminary results ofthe third round oftests, carried out in 1986; further analysis of existing data as well as analysis of further data gathered in the fourth round in 1989 will perhaps lead to a better understanding of factors influencing the longitudinal trends in the studied cohorts in the three communities.