Use of outpatient clinics as a health indicator for communities around a coal-fired power plant.

The permit to operate the first coal fired power plant in Israel was issued with the condition that a comprehensive network to monitor its effects on the environment, health, and agriculture must be installed and operated around the plant. The health monitoring system consists of four studies, which started 1 year prior to the operation of the plant and were carried out for 10 years. In the framework of the health monitoring system, a study of requests for health services was carried out. In this survey, 8 clinics of the Sick Fund, served by 16 physicians, were followed up. The clinics were located as near as possible to air pollution monitoring stations and represent expected different levels of pollution. A health recorder summarized each day's visits to each physician and tabulated the total visits for each day and the visits due to respiratory tract complaints. Multivariate stepwise regressions on total as well as on respiratory complaints were carried out. The independent variables in the regressions were sulfur dioxide, meteorological parameters (such as temperature and humidity), and flu epidemics. Temperature was almost always significantly correlated with respiratory complaints, but less correlated with total visits among, adults and children. Sulfur dioxide, most meterological parameters and flu epidemics were not meaningful explanatory factor in the regressions. Ambient air pollution levels did not exceed the Israeli air quality or the more stringent local air quality standards, the monthly and annual average sulfur dioxide and nitrogen oxides values were very low. Imagesp1110-aFigure 1.Figure 2.Figure 3.Figure 4.Figure 5.Figure 6.Figure 7.

a comprehensive network to moinitr its effct on thee:im , health, anda ltre mont be intlledanope around: th pt Ts f Ofr s tQd'K ies, which startedy: r t inpofid 10 .m the fiamework of the healt t g sem, a f requ f lth sices was carned outLIn ti survey, 8  The permit to build and operate the first (1400 megawatt) coal-fired power plant in Israel stipulated that a comprehensive network be installed to monitor its effects on the environment, health, and agriculture. An epidemiological monitoring program was designed in response to the health monitoring requirement (1). Four types of studies were included in the epidemiological monitoring program: mortality analyses, monitoring of requests for health services, studies of pulmonary symptoms and lung function in schoolchildren, and panel studies of adults, both with and without chronic obstructive pulmonary diseases. Baseline data were gathered for at least 1 year prior to the operation of the first unit of the power plant. Subsequent data were collected until the end of 1990.
This epidemiological monitoring program was aimed at detecting any spatial or temporal changes in the health status of the study population, taking into account possible confounding factors (such as meteorological parameters and flu epidemics) that could act similarly to air pollution. Here we evaluate requests for health services in outpatient clinics. This survey was designed to point out short-term health effects with minimal time delay. Ongoing analysis of the data could demonstrate a difference in demand for medical services between areas with different levels of pollution.
Baseline data gathered before the operation of the power plant enabled a comparison with the seasonal distribution of visits before possible changes in air quality.

Methods
About 70,000 people live within a 10km radius of the power plant. The population in this area was relatively stable throughout the time of the follow-up. About 80% of this population was insured under "The General Sick Fund." The use of health services in 8 clinics of the Sick Fund, served by 16 physicians (each clinic has one pediatrician for children younger than 14 years and one general physician for adults) was recorded. The size of the physicians' practices did not change substantially over the years of the follow-up. The number of patients in each of the clinics followed up varied from 600 to 2100. In the area expected to be most polluted, about 7000 patients were followed up. The total number of patients followed up in this study was about 30,000 persons.
The distribution of the eight clinics represent a gradient of expected low, medium, and high pollution, according to the environmental impact statement presented by the Israel Electric Company to the Ministry of Health. The clinics were located as near as possible to continuous air pollution monitoring stations (Fig. 1). The people using these clinics lived in the vicinity; therefore, the location of the clinics served as a good estimate of exposure to air pollutants in the community. The physicians kept records, registering all visits of patients and each patient's diagnosis.
Once a week, a health recorder summarized each day's visits to each physician and separately tabulated the total visits for each day as well as the number of visits for respiratory tract complaints. Air pollution and meteorological data were gathered and analyzed by the Association of Towns for Environmental Protection.
The environmental monitoring network consists of 12 monitoring stations, of which 10 are stationary and 2 are mobile. The monitoring stations are fudly automatic and measure the following data: SO2, NOx, 03, CO, total hydrocarbons, and meteorological parameters. Not all monitoring stations perform all measurements. The instruments are automatically calibrated and continuously measure the levels of ambient air pollution. The data are sent by radio to the Association ofTowns and fed into a computer that stores and analyzes them. Daily averages of air pollution measurements were given to the health recorder for inclusion in the statistical analysis of visits to outpatient clinics.
Data concerning flu epidemics during November to April of each year were gathered nationwide by the epidemiological department of the Ministry of Health. Most flu cases were diagnosed on symptomatology. A combined file was created including daily visits to outpatient clinics, daily air pollution concentrations, daily meteorological data, and daily flu data.
Statistical analysis of use of outpatient clinics in the vicinity of the power plant was carried out using SPSS (McGraw-Articles -Use of outpatient clinics as a health indicator Hill, New York) and BMDP (Berkeley, California) programs. The ratio between the daily number of visits and the size of the physician's practice (size = number of patients assigned by the Sick Fund to the physician) (visit ratio) was calculated both for visits with respiratory tract complaints and for the total number of visits, for every day and physician. The daily averages of these ratios for each day of the week and for each month were calculated. The calculations were carried out separately for visits to pediatricians and for visits to general physicians.
Monthly averages for all the survey years were calculated from the daily averages, both for respiratory diseases and for the total number of visits among children and adults. In the next step, data of environmental measurements from the nearest monitoring stations were added to the database of the outpatient clinics.
From flu monitoring data, daily averages of flu visits were calculated for children and for adults. These flu data were also added to the database of the outpatient clinics. Multivariate stepwise regressions on the daily visit ratios were performed separately for each year of the study for both pediatricians and general physicians. Daily visits due to respiratory tract complaints as well as visits due to all reasons were the dependent variables in the multivariate analysis. Daily ambi-ent SO2 concentrations, meteorological parameters, and flu-data were the independent variables in these regressions. To follow up possible changes over time in the use of outpatient clinics, a one way analysis of variance was carried out, in which the studied effect was the year of study. The meteorological parameters as well as SO2 were entered as covariates in the analysis ofvariance.

Results
Analysis of the daily averages of visit ratios showed an increased use of clinics on Sundays, the first workday of the week in Israel. This trend characterized both visits to pediatricians and visits to general physicians.
In Among adults, a seasonal trend of more frequent use of outpatient clinics due to respiratory tract complaints was observed during the months December through March throughout the followup. A seasonal trend of more frequent use of pediatric outpatient dinics during winter months due to respiratory tract complaints was also observed (Fig. 3). A seasonal trend was not observed for the total visits to practices served by general physicians (Fig. 2). However, the total number of visits to pediatricians peaked during winter months, possibly because respiratory complaints account for a major part of the total visits to pediatricians.
No consistent trend of change in the use of adult clinics due to respiratory tract complaints was observed between 1982 and 1990. A substantial dedine in the total number of visits occurred between 1988 and 1990 (Fig. 2). For children, a trend of moderate decline in the total number of visits, and especially in the number of visits due to respiratory tract complaints, was observed between 1982 and 1984. There is no obvious trend for the following years (Fig. 3). Multivariate statistical analyses were carried out to determine the dominant environmental factors explaining the use of medical services in the power plant area. By means of multiple stepwise regressions, the effects of meteorological parameters (such as temperature, baro-metric pressure, humidity, and precipitation), flu epidemics, and air pollutants on the use of outpatient clinics were analyzed.
For each year of study, a multiple regression analysis which included all the above-mentioned parameters was carried out. Regression equations were calculated separately for respiratory complaints and for total visits, both for children and adults. The coefficients for the variables which appeared in the regression equations are presented in Table 1 for each year of the study. The major explanatory factor for use of outpatient clinics among both children and adults was ambient temperature. Most coefficients for temperature were negative and highly significant; i.e., at lower temperatures, higher numbers of clinic visits were observed. In some regressions, further explanatory factors for use of medical services occurred; a flu epidemic was the most frequent additional explanatory factor. The part of the variance explained by the regressions was between 6% and 32%.
In the analysis of changes in use of outpatient clinics over time, the year of follow-up was entered as the main effect, and all other variables such as tempera-  ture, precipitation, barometric pressure, and SO2 were entered as covariates.
Significant deviations from the average use of pediatric and adult outpatient clinics occurred over time, but the effect of the year of study was not consistent. In other words, a significant increase in use of outpatient clinics was observed for some years, while in other years a decrease was observed.
During the time period of the study, the ambient air pollution levels did not exceed the Israeli air quality standards or the more stringent local standards (half of the Israeli standard), based on measurements of SO2 and NOX at the 12 monitoring stations (Fig. 1). In fact, the annual average S and NOX values were very low. For instance, the 1989 annual SO2 averages did not exceed 20 pg/m3, less than a quarter of the U.S. standard, at any monitoring station ( Table 2).
Because of the difficulty of following multiyear trends of such low levels of annual average concentrations, we decided to focus on half-hourly concentrations, for which there is also an Israeli ambient standard. We counted the number of air pollution "events" in which the half-hourly averages for SO or NO were above an arbitrary threshold. The threshold values chosen were 183rg/m3 (70 ppb) for SO2 and 235 ,ug/m (125 ppb) for NO , which are about onequarter of the local air quality standards and about one-eighth of the Israeli air quality standards. In Figure 4 the summary of events for S and for NOx during 1981-1992 is presented. From this summary, an increase in the total number of events measured by the monitoring stations can be observed. Part of the increase in the number of events originated from the operation of the power plant, but the main increase was connected with other sources, such as industry and especially traffic.
The sources responsible for the events were identified according to the wind direction measured at the time of the event. The highest number of SO2 events from the power plant was measured at the monitoring station located in the area expected to be most polluted (Fig.  5); a markedly lower number of events was registered in the area expected to be moderately polluted. No events were registered during 1987-1992 in the monitoring station located in the expected low-pollution area. The highest number of NOX events from all sources was measured in the monitoring station located in the area expected to be most polluted (Fig. 6); a markedly lower number of  1989  Station  1  2  4  5  6  7  8  10  11  12  January   7  19  8  12  20  7  11  8  0  5  February  5  17  9  15  26  13  8  11  11  6  March  5  12  5  5  15  13  7  7  5  5  April  7  23  8  13  32  12  10  14  23  9  May  9  21  11  10  16  13  5  9  8  8  June  10  21  8  14  13  5  11  7  7  9  July   11  8  4  5  5  18  5  11  0  10  August  13  7  4  5  8  23  8  10  1  10  September   11  11  4  6  4  10  6  7  1  11  October   12  12  12  10  9  7  6  8  7  5  November  5  11  7  10  11  12  4  7  6  5  December  6  19  14  14  18  15  5   events was registered in the expected low pollution area. Only a few events were registered in the area expected to be moderately polluted. A substantial increase in the number of NOx events measured between 1987 and 1992 could be observed. In the monthly distribution of events from the power plant (Fig. 7), a peak in the number of events during summer months (une-September) was registered. The reason that the pollution events occurred at this coastal site primarily during summer days is the strong radiative heating of the ground, combined with an on-shore sea breeze (3-5 m/sec), which together gave rise to a strong convective layer extending to a height of several hundred meters. When the plumes from the high stacks entered the convective layer, fumigation occurred and pollution was brought down to ground level. In contrast, no significant convective layer developed during typical winter weather.

Discussion
Epidemiological investigations of health effects of ambient air pollution have often focused on community or regional morbidity and mortality (2-4). Shortterm effects on morbidity have been sought by examining correlations of pollution levels based on networks of monitoring stations with rates of hospital admissions (3)(4)(5)(6), emergency room visits (2,(7)(8)(9)(10), or outpatient clinic visits (11)(12)(13). Such community-based epidemiological studies are subject to a number of limitations; primary among these is the use of community exposure values rather than use of exposures of individuals in the community. Nevertheless, when morbidity data are collected according to well-defined and uniform criteria and pollution data are collected by reliable networks of monitoring stations, such studies provide a relatively inexpensive and valuable tool. In our study morbidity data were collected according to well-defined and uniform criteria. However, we have not been able to distinguish between acute and planned visits nor could we identify multiple visits of the same patients. Air pollution data were collected by a reliable network of monitoring stations, located in the vicinity of the Sick Fund clinics. All available data collected over the entire follow-up period (13) also used morbidity data as recorded by pediatricians in the community. They registered a higher incidence of respiratory diseases (p<O.O 1) with higher mean duration for contaminated versus clean areas. In our survey we could not find an effect of SO2 exposure on use of outpatient clinics, apparently due to extremely low SO2 levels even in the regions expected to be more polluted. Moreover, the number of events of SO2 peaked during summer months, whereas visits due to respiratory complaints peaked during winter months. Similar to our findings, Rebmann et al. (14), in their health study carried out among preschool children in southwestern Germany, did not find a correlation between cases of croup and the degree of air pollution, apparently because of relatively low pollution levels. The epidemiological monitoring program carried out in the vicinity of the Hadera power plant, included a health study among schoolchildren in which pulmonary symptoms and lung functions had been assessed. In the framework of this study we also could not show any deleterious health effect among the children residing  -4, 1996 This European ISSX Workshop will take place Saturday, June I-Tuesday, June 4 in the lovely seashore city of Sitges, located 30 km south of Barcelona. Workshop attendance will be limited. The objective of the workshop is to bring together both senior and young scientists to present and discuss their latest contributions in diverse areas of host mechanisms, such as mechanisms of toxicity, role of biotransformation enzymes, and inhibitory and inducing effects which condition the response of xenobiotics. There will be particular emphasis on compounds present in diet. In addition to the opportunity for poster and oral presentations, the following subjects will be covered in scientific sessions: * mechanisms of toxicity * inhibitory and inducing effects * role of biotransformation enzymes * natural and artificial food toxins