Analyzing spatial correlation between hazardous waste sites and mortality rates in Cincinnati, Ohio.
Industrial plants routinely produce hazardous waste as part of their production processes. Much of this waste is released into the environment at plant sites or at waste treatment facilities. The improper management of these releases at some industrial sites has resulted in contamination posing a confirmed or potential threat to human health or the environment (Ohio EPA 1997). Sites where substantial amounts of hazardous waste are released, or where environmental degradation has taken place, have become a major concern of environmental planners, especially when they are located near densely populated urban neighbourhoods. These urban sites may not be evenly distributed spatially, but rather, create an uneven pattern of resident exposure. Therefore, assessment of the environmental effects of such sites is inherently geographic and spatial analysis can be used to reveal the pattern. In research terms, the assessment results provide an important step in describing problems and in formulating and testing hypotheses about possible links between environment hazards and quality of life. In policy terms, the results have the potential for directing action to areas of greatest need.
The United States is the world's largest producer of hazardous wastes (OTA 1983). In the 1995 U. S. Toxic Release Inventory, 21,951 facilities reported releasing 2.2 billion pounds of listed toxic chemicals into air, water or land (USEPA 1997). The U. S. Environmental Protection Agency (USEPA) lists more than 38,000 uncontrolled hazardous waste sites in the United States. Over 1,400 of the 38,000 sites are listed or proposed to be listed on the USEPA's National Priority List (NPL). The NPL includes sites posing the greatest threat to public health and the environment, and details priorities for long-term investigation and clean up (USEPA 1990). An estimated 41 million persons live within four miles of the NPL sites (NRC 1991). While there has been progress made in recycling, reuse and appropriate disposal of much of this waste, the poor disposal practices of the past, and ongoing problems with some current producers and users of hazardous substances represent continuing threats to human health (OTA 1983).
The U. S. Agency for Toxic Substances and Disease Registry (ATSDR) conducted public health assessments (PHAs) at 1,607 hazardous waste sites in the United States to evaluate the threat posed by those sites to human health (Lichtveld and Johnson 1993). Nearly one percent of the sites were considered to pose urgent public health hazards as the result of short-term exposure. About 19 percent posed public health hazards as the result of long-term exposure. Hazards from 62 percent of the sites surveyed were indeterminate because information was incomplete. Only 10 percent were considered to pose no apparent hazard or no hazard.
The health effects of exposure to environmental contamination have been the subject of considerable controversy. Yet, there is widespread agreement that exposure to hazardous waste may be adding to our disease burden in significant, although as yet not always precisely defined, ways (USDHHS 1980). Investigations of hazardous waste sites have demonstrated health effects in exposed persons, including low birth weight, cardiac anomalies, headache, fatigue, miscarriages, respiratory problems, and neurobehavioural problems (Berry and Bove 1997, Geschwind et al. 1992, Washington 1994, NRC 1991). Some studies have shown higher cancer rates in residents exposed to hazardous waste components (NRC 1991). Clearly, there are reasons to be concerned about the effect of hazardous waste sites on environment and human health. Indeed, the social implications of environmental degradation have drawn increasing attention among scholars and policy makers in environmental, social, and human health studies. Unfortunately, we still lack full knowledge of many of the relationships involved. Further research is especially needed on spatial relationships. The effective handling of spatial information is essential to facilitate an appropriate public policy response. Those charged with safeguarding the public good and making plans that balance the needs of nature and people can turn to geographic variation in health indicators to help identify causes of ill health. An area's mortality rate is certainly one of the better measures of overall health.
Environmental justice has been defined as the equitable sharing of the adverse effects of pollution across racial and income groups (Xia et al. 1997). This implies that public policies and regulations, including the siting of polluting industries or the permitting of toxic releases into the environment, should not disproportionately expose minorities or the poor to environmental hazards (Bullard 1997). Studies investigating environmental justice issues have generally concluded that minorities and the poor are likely to have greater exposure to toxic landfills, waste incinerators, hazardous industrial facilities and other environmentally detrimental activities (Bullard 1993, Bryant & Mohai 1992, Buntin 1994, Ortolano 1997, Ringquist 1997, Bullard 1997, Burby 1997, Vos 1997). Other work has found that commercial hazardous waste treatment, storage and disposal facilities (TSDFs) were not more likely to be located in poor and minority communities (Oakes et al. 1996; Anderton et al. 1994), though a more recent study came to an opposite conclusion (Boer et al. 1997). Several researchers have raised questions about the methods used in some of the early environmental justice studies. Results were found to be different depending on whether a large (county or zip code) or small (census tract) geographic unit was analyzed (Bowen et al. 1995, Anderton et al. 1994).
Much of the more recent research has moved beyond simply identifying where environmental injustice may exist, to assessing how social, political and economic institutions have contributed to the current state (Been 1994, Heiman 1996, Penderhughes 1996). Indeed, simply sorting out which came first, the environmentally hazardous facility or the residential area, can be a complex and trying exercise. Such efforts must consider how public institutions have allowed or encouraged the siting of new developments or the expansion of existing ones. The absence of minority representation in the siting decision process (Vos 1997) and the need to limit industrial expansion near residential areas (Burby 1997) have been cited as two of the areas where public institutions have contributed to environmental injustice.
Planners, public officials and residents concerned with quality of life and environmental justice require analytical tools that enable them to identify and initiate responses to potential threats. These tools must allow for determining the possibility of a significant health threat, and the need for more in depth epidemiological, environmental or land use analyses. The combination of geographic information systems (GIS) and statistical analysis comprises just such a tool to analyze health, environmental and demographic data. Two features offered by GIS are especially useful to help accomplish this. First, GIS allows the construction of maps, identification of nearest neighbours, and display of spatial relationships. A series of patterns, each for a different variable of interest, may be created and combined to reveal correspondences and disparities. Second, GIS functions, such as the storage, retrieval and manipulation of spatially related data, allow the aggregation of data collected from varying sources. This data can be used to develop composite descriptions, and to explore associations, such as the use of distance buffers to identify how the effects of a potential hazard may change as one moves father away (Wartenberg 1993).
It is within this context that this study uses GIS and statistical analysis to describe and analyze the spatial relationships between the proximity to hazardous waste sites and mortality rates in Cincinnati. Hazardous waste sites used in this study are sites of reported industrial releases for toxic chemicals or where improper hazardous waste management has resulted in environmental contamination, and not simply TSDFs. This study seeks to determine whether residential areas close to hazardous waste sites have higher mortality rates than those farther away. The relationship between the mortality rates and socio-economic conditions of the population living at different distances from hazardous waste sites is also addressed.
The study area was the City of Cincinnati, Ohio (1990 city population 364,040; 1990 consolidated metropolitan statistical area population 1,744,124). Census block groups from the 1990 U.S. Census were used as the unit of analysis. To integrate the data for this study, five types of digital data files were compiled: 1) death records, including location of residence, date and causes of death for all persons who died in Cincinnati from January 1, 1986 to December 31, 1994; 2) locations of hazardous waste sites; 3) 1990 U.S. census block group population, housing and other socioeconomic characteristics for Cincinnati; 4) census block group boundaries for Cincinnati; and 5) streets in Cincinnati.
Records of all deaths reported to the Cincinnati Health Department for the years 1979 to 1994 were obtained on tape. They were in compressed text format and subsequently converted to a dBase (.dbf) format. The data file contained 35 data items for each of the death records, including last residence of the diseased, cause of death, birth and death dates, and limited socioeconomic information such as race, number of years of education, and state of birth (CHD 1994). Records for Cincinnati residents who may have died outside the city were not included. Also, we excluded the records for non-Cincinnati residents who died in Cincinnati since the study area was limited to the City of Cincinnati.
Two types of environmental contamination and hazardous waste release data were collected: the 1994 Master Sites List (MSL) from the Ohio Environmental Protection Agency's (OEPA) Division of Emergency and Remedial Response (DERR); and the 1992 Toxic Release Inventory (TRI) Annual Report, from OEPA's Division of Air Pollution Control.
The MSL is a database of sites in Ohio where: (1) there is evidence of, or it is suspected that, improper hazardous waste management has resulted in the contamination of air, water or soil; and (2) there is a confirmed or potential threat to human health or the environment. The MSL includes a diversity of sites of varied environmental concerns. In addition to those sites in the USEPA's Comprehensive Environmental Response, Compensation, and Liability Information System (CERCLIS) prior to 1989, sites were added to the MSL listing by DERR staff based on inter-program referrals, citizen complaints, or DERR's discovery efforts. The DERR updates the MSL annually and sites may be delisted if formal remediation has been completed (DERR 1994). Examples of the MSL sites are chemical companies and landfills. The MSL database records contain a field with the street address for each site.
The TRI report contained annually compiled data on the quantity and location of industrial releases for approximately 300 toxic chemicals and 20 chemical categories (Bowen et al. 1995). Manufacturing firms subjected to Title III, Section 313 of the federal Emergency Planning and Community Right-to-Know Act of 1986 are included in the TRI list. These firms are required to report the location and amount of toxic chemicals released to the air, water, or land. In Ohio, the TRI Program within the Division of Air Pollution Control of the OEPA coordinates the collection, digitizing and distribution of TRI data. The TRI sites include a broad range of industrial facilities, from manufacturing and food processing, to chemical plants. Like the MSL database, the locations of the TRI sites are stored in a street address field.
Census population data were directly extracted from the U.S. Census Bureau's 1990 Summary Tape File (STF) 3A available on CD-ROM. The census data are considered the most reliable source for population information by geographic area. Data used in this study include population by age group for each census block group within the city of Cincinnati and population by age group for the state of Ohio. The total numbers of deaths in Ohio by age group were obtained from the Ohio Department of Health (ODH 1992).
The Census block group boundary data were extracted from the First Street geographic data files produced by Wessex Inc. The First Street files were compiled and enchanced from the U.S. Census Bureau's 1992 Topographically Integrated Geographic Encoding and Referencing (TIGER) files. The files contained graphic data--i.e., maps--defining census block group boundaries and associated attribute data. Unique block group identification numbers (IDs) were used to link attribute data to graphic data. There are 417 census block groups in Hamilton County, of which 22 had no residents in 1990. These were eliminated, thus providing a total of 395 census block groups for this study. The street address data used in the study were the 1994 TIGER files. The street files contained street name and address ranges for street sections in Hamilton County.
The above data files were integrated based on their spatial location. GIS functions, including geocoding, buffering, and overlay analysis, were used to complete the tasks. The statistical tools were then used to calculate census block group based mortality rates and to test the relationship between the mortality rate and the distance to the hazardous waste sites. Arc/Info, a GIS software program (Environmental System Research Institute, Inc.) was used for geographic analyses. SPSS, a statistical analysis software program (SPSS Inc.), was used for statistical analysis.
The geocoding function was used to identify the location of hazardous waste sites and residences of the deceased. During the geocoding process, the address for each data record was compared with the address data in the street file. Each address received a score, in a range from 0 to 100, corresponding to the confidence of the address matching result. A higher score indicated a higher degree of confidence that an address had been correctly located on the map. A threshold score, 96, was used in this study. That is, any record with a matching score greater than 96 was considered to have been address matched successfully. The geocoding products were three Arc/Info point coverages for mortality data and MSL and TRI sites.
The spatial overlay function was used to complete two tasks. The first task was to calculate the proximity to hazardous waste sites from census block groups. Grids of 100 feet (approximately 30 metres) were overlaid on the city area and the proximity from each grid to nearest MSL or TRI site was calculated. After the scattergrams of mortality rate versus distance were created and evaluated, the census block groups were divided into different buffer zones of equal distance to the nearest MSL and TRI site, respectively. The use of the buffers allows the aggregation of block groups based on their distance to the hazardous waste sites so the effect of proximity on mortality rates can be assessed.
The second task was to assign each of the residences of deceased persons into a specific census block group. Points for the residences of deceased persons were overlaid on the census block groups to obtain the block group number for each residence. Then the number of residences of diseased within each block group was calculated.
Calculating Age-Adjusted Mortality Rates
Nine-year average crude mortality rates were used to calculate the age-adjusted mortality rates for each block group. The average annual number of deaths in each block group from 1986 to 1994 was divided by the 1990 total population of each block group. This was done to minimize the effect of random error since the number of deaths in any one-block group might be small.
Age-adjusted rates for each census block group were used to control for the age effects when comparing mortality rates by block group. To calculate the ageadjusted rates, the 1990 total population and the annual average number of deaths for each census block group were sorted into five age cohorts: less than one year; one to 14 years; 15 to 34 years; 35 to 64 years; and 65 years and over. From these two numbers the mortality rate was calculated for each age cohort in each census block group. The total population of Ohio was then sorted into the same age cohorts. Finally, the age-adjusted rate for any census block group was calculated as follows:
where, Dk: adjusted mortality rate for census block group k
Psj: Ohio population for age cohort j
Ps: total Ohio population
djk: mortality rate for age cohort j
djk: = Cjk/Pjk
where, Cjk: the bannual number of deaths for age cohort j in census block group k
Pjk: population for age cohort j in block group k
The independent t-test for equality of means was used to determine the significance of the difference in mean age-adjusted mortality rates between adjacent buffer zones. The independent t-test rather than the matched-pair t-test was used since the measurement of mortality rate in each zone was assumed to be entirely independent. The t-test results provide a basis for accepting or failing to accept the null hypothesis that the means of the age-adjusted mortality rates for the census block groups in buffer zones closer to the hazardous waste sites were not significantly different than the means for those in buffer zones farther away.
A bivariate linear correlation was employed to illustrate the association between the mean age-adjusted mortality rate of each block group and selected socioeconomic indicators for each block group. A value of the linear correlation coefficient (r) close to 1.0 or -1.0 indicates highly positively or negatively correlated variables, while a value close to 0 indicates no relationship between the variables.
The socioeconomic indicators were selected to reflect known mortality risk factors, as well as to provide insight into the conditions found in each block group. The following indicators were selected: length of residence in unit (percent of households living in unit more than 10 years); median household income; median housing unit value (owner-occupied units); median household rent (renteroccupied units); percent of persons 25 years or older with less than a ninth grade education; and percent of African American population. Bivariate correlation coefficients were also calculated for the relationship between the distance of the census block group from a hazardous waste site and the socioeconomic indicators.
Results and Discussion
Among the 1979-1994 death records in the Cincinnati Health Department's data file, 96,440 were identified as Hamilton County residents. Of these, 96,235 were geocoded (99.8%) based on the recorded street addresses. After excluding accidental causes of death, this study used records for 31,526 decedents who were Cincinnati residents at the time of their death and died between January 1,1986 and December 31, 1994.
The locations for 75 MSL sites and 131 TRI sites in Hamilton County were geocoded based on the street addresses of the sites. Only 12 sites were found on both the TRI and MSL lists. Figure 1 plots block group mortality rate versus minimum distance to the nearest TRI or MSL sites. It clearly shows that the block groups with high mortality rates are closer to TRI or MSL sites. The mortality rates for census blocks about two kilometres away from TRI sites or one kilometre away from MSL sites are near or below the mean mortality rate of 9.2 deaths per thousand residents. The vast majority (95%) of the census block groups is within four kilometres of a hazardous waste site. To further test the effects of proximity on the mortality rates, we divided the first four kilometres into five buffer zones of equal distance. The first buffer zone covered the block groups within 800 metres (approximately 0.5 miles) of the nearest hazardous waste site. Because only ten block groups (<3%) were 4.8 kilometres or more away from the nearest hazardous waste site, all block groups four kilometres or beyond were included in Zone Six. Block groups in the second through the fifth zones fell in between, at 800-metre increments.
Figure 2 displays the MSL sites with corresponding buffer zones. The MSL sites were scattered within the central part of the city while there were no MSL sites in the northwestern and southeastern portions of the city. Several MSL sites were located in the southern part of the city, just to the west of the central business district. Most MSL sites were near major highways or close to waterways. The census block groups within 800 metres of a MSL site were almost all connected, except for those in a small area in the western part of the city.
The distribution of TRI sites shows a similar pattern to that of the MSL sites (Figure 3). There were only a few TRI sites in the western portion of the city, mostly located along the narrow corridor in the southwestern corner. Similarly, there were only a few TRI sites in the southeastern portion of the city. The TRI sites were even more concentrated along the Mill Creek and two major highways, I-71 and I-75, than were the MSL sites. It should be noticed that a number of TRI sites were found in the two communities of Norwood and Elmwood Place which are surrounded by the city of Cincinnati. Those sites were included in the analysis since their impact would not stop at the political boundaries.
The age adjusted mortality rate by census block groups in Cincinnati is shown in Figure 4. Mortality rates appear higher in the older, less affluent areas in the centre and west portions of the city. The newer areas to the east generally have lower mortality rates. Table 1 shows the distribution of those census block groups in relation to the buffer zones for MSL and TRI sites. About half the census block groups (190) were within 800 metres of MSL sites and 14 were beyond four kilometres from any MSL sites. When census block groups were compared with TRI buffer zones, about 40 percent of the census block groups were within 800 metres of TRI sites and 20 were more than four kilometres from any TRI sites.
Table 1: Distribution of Block Group Mean Age Adjusted Mortality Rates among Buffer Zones MSL Sites TRI Sites Mean Mean Buffer Number of Mortality Number of Mortality Zone Census Block Rates Census Block Rates Groups (per 1000 Groups (per 1000 population) population) One (closest) 190 9.9 163 9.6 Two 108 9.6 112 9.8 Three 64 8.9 71 9.4 Four 14 5.0 20 7.3 Five 5 6.9 9 5.3 Six (farthest) 14 3.4 20 6.5
Citywide 395 9.2 395 9.2
The mean mortality rates in Table 1 confirm that census block groups in buffer zones closer to MSL sites had higher mortality rates than those farther away. Census block groups in Zone One through Zone Three had mean mortality rates ranging from 8.9 to 9.9 per 1000 population. This compares with the mean mortality rate for all block groups of 9.2 deaths per 1,000 population. Census block groups in Zone Four and beyond had mean mortality rates ranging from 3.4 to 6.9 per 1000 population. Like the MSL sites, census block groups in buffer zones closer to TRI sites generally had higher mortality rates than those farther away. Census block groups in Zone One through Zone Three had mean mortality rates ranging from 9.4 to 9.8 per 1000 population. Census block groups in Zones Four and beyond had mean mortality rates ranging from 5.3 to 7.3 per 1000 population. This study shows that, though the threat to human health or the environment from many TRI sites may be less severe than the MSL sites, the general patterns between the two groups are quite similar. The significant differences in mean mortality rates among buffer zones suggest an explanation in the characteristics common to MSL and TRI sites - location. The sites in both groups are concentrated in current or previous industrial areas within the central part of the city. Thus given the similarity in the spatial distribution of the MSL and TRI sites, it is not surprising that relationship between proximity of residence and mortality rates is similar for both.
The t-test results demonstrate somewhat different patterns for MSL and TRI sites (Table 2). When the buffer zones for the MSL sites were compared, there was a significant difference in mean age-adjusted mortality rates between the third and fourth buffer zones. This was not true for buffer zones based on the TRI sites. When we consolidated buffer zones into two groups (Table 3), depending on their distance to MSL or TRI sites, a significant difference (p=0.000) was measured for all MSL groups, but only beyond the second buffer for the TRI groups. The MSL sites are so designated because there is evidence, or it is strongly suspected, that hazardous contamination has taken place, and there is a confirmed or potential threat to human health or to environmental resources, such as ground water or soil. The TRI sites, in aggregate, likely represent a reduced potential for hazard to human health or the environment since the site list includes firms that use or transport approximately 300 toxic chemicals regardless of whether there was a release.
The bivariate correlation analysis shows moderately statistically significant associations between the mean age-adjusted mortality rate and the selected socioeconomic indicators (Table 4). The negative coefficients between mortality rate and length of residence, median household income, median home value, and median rent were expected and indicated that when these indicators increased, the mortality rate decreased. The positive correlation coefficients between mortality rate and the percent of African Americans and the percent of people completing less than ninth grade implied a higher mortality rate in areas with higher percentage of minority residents and lower education level. The strongest socioeconomic indicator correlated with mortality rate was income, followed by race, rent, length of residence, home value, and education level.
The bivariate correlation coefficients for the relationship between the distance of census block groups to MSL sites and all six socioeconomic indicators showed significant associations (Table 5). The percent of African Americans and the median household income showed the strongest relationship and length of residence the weakest. When the six socioeconomic indicators were correlated with the distance to the TRI sites, four of them had statistically significant correlation coefficients: race, median household income, median home value, and education level. The correlation with indicators for median rent, and length of residence were not significant.
Table 2: t-test for Comparision of Mean Age Adjusted Mortality Rates in Adjacent Buffer Zones Buffer Zones Mean Mortality Two-tailed t-test Sample Size Compared Rate Difference probability MSL Sites One/Two 0.3 0.618 190/108 Two/Three 0.8 0.239 108/64 Three/Four 3.8 0.000 (1) 64/14 Four/Five -1.9 0.094 14/5 Five/Six 3.4 0.007 (1) 5/14 TRI Sites One/Two -0.2 0.796 163/112 Two/Three 0.4 0.654 112/71 Three/Four 2.1 0.075 71/20 Four/Five 1.9 0.199 20/9 Five/Six -1.2 0.370 9/20 (1) significant at 0.05 level. Table 3: t-test for Comparision of Mean Age Adjusted Mortality Rates in Combined Buffer Zones Buffer Zones Compared Mean Mortality Two-tailed t-test Sample Size Rate Difference probability MSL Sites One/Two-Six 1.4 0.008 (1) 190/205 One, Two/Three-Six 2.4 0.000 (1) 298/97 One-Three/Four-Six 5.0 0.000 (1) 362/33 One-Four/Five,Six 5.1 0.000 (1) 376/19 One-Five/Six 6.0 0.000 (1) 381/14 TRI Sites One/Two-Six 0.6 0.220 163/232 One,Two/Three-Six 1.4 0.013 (1) 275/120 One-Three/Four-Six 3.0 0.000 (1) 346/49 One-Four/Five-Six 3.3 0.000 (1) 366/29 One-Five/Six 2.9 0.000 (1) 375/20 (1) significant at 0.05 level. Table 4:Bivariate Correlation Coefficients for Proximity of MSL and TRI Sites (Buffer Number) and Selected Socioeconomic Indicators (Sample Size =
Indicator Mortality Rate r p (1) Length of Residence -0.2096 0.000 (2) Median Household Income -0.3487 0.000 (2) Median Home Value -0.1850 0.000 (2) Median Rent -0.2974 0.000 (2) Percent African American 0.3251 0.000 (2) Educational Level 0.1406 0.005 (2) (1) Significance level for the null hypothesis that the r-value in not different from O. (2) Significant at the 0.05 level. Table 5: Bivariate Correlation Coeefficients for Proximity of MSL and TRI Sites (Buffer Number ) and Selected Socioeconomic Indicators (Sample Size
Indicator MSL Buffer TRI Buffer r p(1) r p(1) Length of Residence 0.1231 0.014(2) 0.0304 0.548 Median Household Income 0.3286 0.000(2) 0.2101 0.000(2) Median Home Value 0.1685 0.001(2) 0.1240 0.019(2) Median Rent 0.2815 0.000(2) 0.0586 0.249 Percent African American -0.3316 0.000(2) -0.3278 0.000(2) Educational Level -0.1781 0.000(2) -0.1699 0.001(2) (1) Significance level for the null hypothesis that the r-value in not different from O. (2) Significance at the 0.05 level.
The correlation results demonstrate the important relationship of socioeconomic factors with health outcomes. The positive correlation coefficient for the association between distance to MLS sites and length of residence suggests that people move more often in areas closer to the hazardous waste sites. Block groups with higher percentage of minority residents are found closer to the hazardous waste sites. It has long been established that poorer and less educated people suffer more health problems and are more likely to die from such problems than those having higher income. The results from this study seem to corroborate those findings. The fact that income, home values, rent and education levels are higher in census block groups farther away from the hazardous waste sites suggests that as one moves away from the industrial areas, the property becomes more desirable and therefore more valuable for residential uses. Poorer persons who are less able to afford the higher home prices and rents of the neighbourhoods farther from the hazardous waste sites thus buy and rent homes closer to these sites. Their poorer health status is reflected in the higher mortality rates for these areas. However, whether living closer to the hazardous waste sites causes additional health problems is unclear. Research is needed to compare the health conditions in census block groups with similar socioeconomic characteristics at different distances from hazardous waste sites.
Similar strengths of association would be expected with the TRI sites. The similarity was found with race, median household income, education level, and median home value. However, the results show that the correlation of mortality rate with the length of residence and median rent were not statistically significant. Again, this disparity in the results may reflect differences in the MSL and TRI sites. While the MSL sites have been identified as environmental hazards, the TRI sites are simply the sources of toxic releases to the air, water, land or sewers. It is not clear how this difference may affect the rents and length of residence for nearby residents. A closer examination and comparison of the characteristics of the census block groups near and far from each type of hazardous waste site is needed to clarify this issue.
After examining the spatial distribution of hazardous waste sites, mortality rate and six socioeconomic indicators at the census block group level in Cincinnati, we found associations between the distance to hazardous waste sites and the mortality rate and selected socioeconomic indicators. The study has demonstrated the need for more in depth investigations. Four issues raised by the results of this study are especially pertinent for environmental planners and policymakers. The issues are: 1) the utility of GIS for assessing environmental hazards; 2) the need for targeted follow up study of human health effects; 3) the impact of land use controls and public policy in systematically locating people who are poor, less educated and minority in environmentally threatened neighbourhoods; and 4) the availability of land use and other planning options for remediating the environmental hazards posed by living near hazardous waste sites.
Utility of GIS
We employed a spatial analysis approach to examine the distribution of mortality rates in relation to contamination and hazardous waste release sites. GIS appears to be an effective means for analyzing the multiple kinds of environmental and social factors encountered in this study. The task of finding proper locations for over 96,000 mortality records would not have been feasible without using the geocoding functions provided by GIS. Just as important is the higherorder systematization of GIS spatial analyses. The use of GIS allowed data to be effectively and efficiently aggregated based on spatial placement. It also prepared the data for statistical analysis. The calculation of proximity, overlay analysis and the integration of maps and attribute data allowed for the determination of relationships between environmental and human health in a way that could not be done with only tabular data. Researchers, planners and policymakers can and should use this tool to better understand complex environmental relationships, and to communicate them to the public.
Need for Targeted Human Health Studies
The analysis clearly shows that census block groups closest to the MSL and TRI sites in Cincinnati have significantly higher age-adjusted mortality rates than those further away. However, the reasons for this association are not clear, and require further study. Although exposures of greater intensity and longer duration to hazardous wastes might be causally related to higher mortality rates, this study has not sought to identify such a causal relationship. Rather, this is an exploratory effort to determine if proximity and health status are statistically associated in some way. Given that an association has been confirmed, a natural expansion of the research will be to explain the nature of the relationships involved. There may be direct causal relationships, or there may be synergistic effects involving several factors. Also, it is likely that there are other confounding factors which are related to both proximity and health status. For example, those census block groups closer to the MSL and TRI sites had significantly lower incomes, lower levels of education and larger proportions of minority residents, and these are important determinants of health status. Closer analysis of the variation in mortality rates among subgroups may provide additional insight, as may scrutinizing other factors such as property values, health care access and utilization, and health-related personal habits. An interdisciplinary team from urban and environmental planning, public health, and environmental engineering is needed to address the complex issues involved. Clearly, developing a better understanding of the fundamental nature of the relationships would be of great benefit to planners involved in a wide-range of land use-related decisions.
Systematic Bias Against the Poor, Uneducated and Minority
The results here suggest the need for further investigation to clarify why residential areas closest to the hazardous waste sites are home to poorer, less educated, and minority groups. Recent research in this area suggests institutional factors may contribute to the locational bias (Burby 1997, Vos 1997). Such areas are generally considered to be less desirable locations to live and property values reflect this. The market economy in the U.S. dictates where people live based on their ability to pay. Consequently, it might be argued that the disparity identified is simply the result of "the invisible hand" of the market.
There are at least two reasons to suggest that non-market institutional factors may be major contributors to the apparent bias against the poor, less educated and minority. First, several of the largest residential concentrations closest to the hazardous waste sites are public housing complexes built in the past thirty years. Decision-makers may have determined that the nearby hazardous waste sites posed no hazards, or simply did not consider the issue. That the real estate was relatively inexpensive, and separated from existing neighbourhoods made these sites politically attractive for situating public housing. In either case, hindsight suggests this decision making was flawed and biased against the current public housing residents.
Second, many of the facilities releasing hazardous wastes were built or substantially expanded in the past thirty years, often on existing industrial sites that had been in use for a century of more. While the industry may have been there first, the residences may have been built when the amount and toxicity of the contamination and releases of nearby industry posed a greatly reduced threat. In such a situation, the residences have remained the same, but the hazards posed by the physical environment in which they are located have increased substantially. In these cases it would seem that environmental permitting and land-use regulations have failed to adequately protect residents from the evolving hazards of modern industrial processes. Responses have been proposed, such as a limit on the number of industrial facilities that locate near residential areas (Burby 1997). Others have recommended historical analysis of siting to search for evidence of bias in permit decisions (Boer et al. 1997). To the residents it makes little difference whether this failure was the result of a deliberate conspiracy or simply a reflection of institutional neglect fostered by existing power structures. Both problems recommend changes in planning processes that would give greater consideration to the hazards resulting from the way in which industrial land is used. For example, performance zoning might be used, based on an enlightened, more comprehensive view of the hazards of industrial use, rather than the current system that assumes all permitted hazardous waste releases pose no hazard to nearby residents.
Remediating the Environmental Hazards
The locations of hazardous waste sites should be considered in making decisions regarding the appropriate land use for nearby properties, as well as environmental remediation and pollution prevention. Planners and communities must take steps to ensure that hazardous waste sites are managed in effective ways in order to prevent the exposure of nearby residents to hazardous levels of toxins. Industrial facilities cannot be allowed to release hazardous wastes, even in small quantities, if the cumulative effect of multiple small releases creates health hazards. Steps must be taken to avoid multiple small hazardous releases that are effectively equivalent to the hazards posed by more seriously contaminated individual sites. Further, while the synergistic effects of likely chemical combinations are not well understood, explicit consideration of the potential for such effects would seem prudent, and should become an integral part of standard land use planning. Also, when resources are allocated for environmental remediation, higher priority generally should be given to sites that individually pose more serious problems, however the impact of other nearby sites must also be considered.
This paper reports on research partially sponsored by a grant from the University Research Council, University of Cincinnati.
Anderton, D., A. Anderson, P. Rossi, M. Oakes and M. Fraser. 1994. "Hazardous waste facilities: Environmental equity issues in metropolitan areas." Evaluation Review. 18(2):123-140.
Been, V. 1994. "Unpopular neighbors: Are dumps and landfills sited equitably?" Resources. Spring: 16-19.
Berry, M. and F. Bove. 1997. "Birth Weight Reduction Associated With Residence Near a Hazardous Waste Landfill." Environmental Health Perspectives. 105(8): 856-861.
Boer, J. T., M. Pastor, J. L. Sadd, and L. D. Snyder. 1997. "Is there environmental racism? The demographics of hazardous waste in Los Angeles County." Social Science Quarterly. 78(4): 793-810.
Bowen, W. M., M. J. Salling, K. E. Haynes and E. J. Cyran. 1995. "Toward Environmental Justice: Spatial Equity in Ohio and Cleveland." Annals of the Association of American Geographers. 85: 641-663.
Bryant, B. and P. Mohai, eds. 1992. Race and the incidence of environmental hazards: A time for discourse. Boulder, CO: Westview Press.
Bullard, R., ed. 1993. Confronting environmental racism: Voices from the grass-roots. Boston: South End Press.
Bullard, R. 1997. Dismantling Environmental Racism in the Policy Area: The Role of Collaborative Social Research. College Station, TX: Texas A&M University Press.
Buntin, B. 1994. Environmental Justice: Issues, Policies, and Solutions. Washington, DC: Island Brothers.
Burby, R. 1997. "People and pollution: What can government and industry do to foster environmental justice?" Paper prepared for presentation at the Annual Meeting of the Association of Collegiate Schools of Planning, Ft. Lauderdale, FL, November 6-9.
CHD (Cincinnati Health Department). 1994. Unpublished internal documents provided by Joseph Geisland, Data Center Manager, 1525 Elm Street, December 29th.
DERR (Division of Emergency and Remedial Response). 1994. Master Sites List 1994. Columbus, OH: Ohio Environmental Protection Agency.
Geschwind, S.A., J.A. Stolwijk, M. Bracken, E. Fitzgerald, A. Stark, C. Olsen and J. Melius. 1992. "Risk of Congenital Malformations Associated with Proximity to Hazardous Waste Sites." American Journal of Epidemiology. 135(11): 1197-1207.
Heiman, M.K. 1996. "Race, waste and class: New perspectives on environmental justice." Antipode. 28(2): 111-121.
Lichtveld, M. Y. and B. L. Johnson. 1993. Public health implications of hazardous waste sites in the United States. Paper presented at Hazardous Wastes and Public Health Approaches: Hazardous Waste Conference. Washington DC: Agency for Toxic Substances and Disease Registry.
NRC (National Research Council). 1991. Environmental epidemiology: public health and hazardous wastes. Vol. 1. Washington, D.C.: National Academy Press.
Oakes, J. M., D. L. Anderton, and A. B. Anderson. 1996. "A longitudinal analysis of environmental equity in communities with hazardous waste facilities." Social Science Research. 25: 125-148.
ODH (Division of Vital Statistics, Ohio Department of Health). 1992. Annual Report: Vital Statistics.Columbus: Ohio Department of Health.
Ohio EPA (Ohio Environmental Protection Agency). 1997. Master Sites List. Columbus, OH: Division of Emergency and Remedial Response.
Ortolano, L. 1997. Environmental Regulation and Impact Assessment. New York: John Wiley and Sons.
OTA (Office of Technology Assessment). 1983. Technologies and Management, Strategies for Hazardous Waste Control. Washington, D.C.: Government Printing Office.
Penderhughes, R. 1996. "The impact of race on environmental quality: An empirical and theoretical discussion." Sociological Perspectives. 39(2): 231-248.
Ringquist, E. J. 1997. "Equity and the distribution of environmental risk: The case of TRI facilities." Social Science Quarterly. 78(4): 811-829.
USDHHS (U.S. Department of Health and Human Services). 1980. Health Effects of Toxic Pollutants. Report prepared for the U.S. Senate by the Surgeon General, Serial No. 96-15. Washington, DC: U.S. Government Printing Office.
USEPA (U.S. Environmental Protection Agency). 1990. "Hazard ranking system." Federal Register. 55:51532.
USEPA (U.S. Environmental Protection Agency). 1997. 1995 Toxic Release Inventory Public Data Release, Washington D.C.: http://WWW.epa.gov/opptintr/tri/pdr95/drover01.htm, last updated: April 14th.
Vos, J. 1997. "The role of local planners and decision-makers in the occurrence of environmental injustice." Paper prepared for presentation at the Annual Meeting of the Association of Collegiate Schools of Planning, Ft. Lauderdale, FL, November 6-9.
Wartenberg, D. 1993. "Use of geographic information systems for risk screening and epidemiology." Paper presented at Hazardous Wastes and Public Health Approaches: Hazardous Waste Conference, Washington DC: Agency for Toxic Substances and Disease Registry.
Washington, R. 1994. "Environmental Equity: Dilemmas and Challenges for Public Health and Social Work for the 1990s." Journal of Health and Social Policy. 62(2): 1-17.
Xia, H., B.P. Carlin, and L. A. Waller. 1997. "Hierarchical models for mapping Ohio lung cancer rates." Environmetrics. 8(2): 107-120.
|Printer friendly Cite/link Email Feedback|
|Author:||Wang, Xinhao; Auffrey, Christopher|
|Date:||Jan 1, 1998|
|Previous Article:||Land resource management planning: the Kamloops experience.|
|Next Article:||Environmental equality: pollution, race and socio-economic status in Michigan.|