The risk assessment was performed by quantifying carcinogenic and noncarcinogenic health risks that potentially are attributable to detected airborne contaminant exposure. The risk assessment was based on 2 sets of exposure characteristics considered appropriate for either the prison worker or the inmate populations. The main differences between worker and inmate population exposure characteristics are the amount of time spent indoors and the number of years spent at the prison facility. Inmates are considered to be full-time residents for the duration of time that they are incarcerated, and are expected to spend the bulk of their time during this period indoors. In contrast, prison workers occupy the facility only during their 8- or 10-h work shifts. The contaminant concentrations to which both populations potentially are exposed are the same because workers, guards, and inmates almost always occupy the same buildings. Other exposure assumptions summarized in Table 1, such as average time spent indoors, inmate incarceration duration, worker occupational tenure, body weight, inhalation rate, and exposure frequency, are based on USEPA default exposure assumptions (USEPA 1997). In addition to the average exposure scenario, worst-case exposure scenarios for both the worker and inmate populations were evaluated. For each exposure factor, the average value was assumed equivalent to the USEPA 50th percentile, while the worst-case value was assumed equivalent to the 95th percentile. Presenting a range of risk estimates, rather than single point values, underscores the variability and uncertainty typically associated with exposure modeling and risk assessments. Variability and uncertainty in health risk assessments arise from natural variability in exposure characteristics among the receptor populations, as well as lack of full knowledge regarding important factors that affect the risk estimates (NRC 1994; USEPA 1995).
Table Table 1.. Assumed exposure factors used for the risk assessment. Exposure factors were based on U.S. Environmental Protection Agency Exposure Factors Handbook and the risk assessor's judgment. The inmate population is predominantly black and male. The worker population is mixed, both racially and gender-wise. If no data is available for the 95th percentile value, the average value is used
| ||Worker population||Inmate population|
|Exposure factor||Average value||95th percentile value||Average value||95th percentile value|
|Inhalation rate ([IH] m3/h)||0.55||No data||0.633||No data|
|Exposure time ([ET] h/d)||7||9||20||22|
|Body weight ([BW] kg)||75||105.2||75.3||105.4|
|Exposure frequency ([EF] d/y)||250||260||365||365|
|Exposure duration ([ED] y)||6.6||15||10||20|
|Averaging time ([AT] y)||75||79||70||75|
|Bioavailability ([Bs] dimensionless)||1||1||1||1|
The risk assessment was performed in accordance with USEPA human health risk-assessment guidance (USEPA 1989, 2003), and aided by use of the American Petroleum Institute's Decision Support System model (API 1999). American Petroleum Institute's Decision Support System was used to perform the contaminant intake and risk quantification calculations and generate the results. The calculated risk estimates apply to the hypothetical individual member of the specified receptor population, rather than specific individuals within the population.
Although air contaminants were not found in every section of each of the potentially affected buildings (indeed the results were ND for the majority of locations sampled), it was deemed neither practical nor prudent to identify contaminated and clean buildings and to perform the risk assessment separately for each section of an affected building. The absence of measurable air contaminants in some buildings or building sections was assumed not necessarily to indicate absence of poor IAQ. For these reasons, it was assumed conservatively that air contaminants occur site-wide and the risk assessment was performed based on this assumption.
The risk-assessment approach only focused on air contaminants detected in the majority of the prison buildings, including trichloroethylene, tetrachloroethylene, benzene, chloroform, beryllium, cadmium, and mercury. Both 1,3-butadiene and thallium were not included in the risk assessment; 1,3-butadiene was not as widely distributed across the facility as other air contaminants. Furthermore, although 1,3-butadiene has published toxicity information, the half-life in air is extremely short (8 h) as compared to more persistent contaminants such as tetrachloroethylene (3,843 h) and chloroform (6,231 h; Howard 1991). Thallium also was not included in the risk assessment because published inhalation slope factors and reference doses are not available from USEPA.
Two separate receptor populations were evaluated, namely the prison inmate population and the prison staff population. The inmate population was considered to be adult residential, but the workers represent an adult occupational population. It was assumed that both populations were exposed to airborne contaminants primarily through the inhalation pathway.
The air-sampling data were tested for normality using both the Shapiro-Wilk and Kolmogorov-Smirnov tests. The normality tests showed that none of the sampling results for the 7 air contaminants of concern were distributed normally or log-normally. One possible reason the sampling data distributions were skewed (i.e., not normally distributed) was the high percentage of ND values among the sample results. Because the sampling data were neither normally nor log-normally distributed, and the underlying distribution was unknown, a distribution-free approach was used to calculate upper confidence limits for each dataset. The 95% upper confidence limits is considered a reasonable estimate of contaminant exposure point concentrations (USEPA 2002). According to USEPA (2002), the central limit theorem can be used to calculate the 95% upper confidence limits because the approach involves no assumptions regarding the underlying distribution of the dataset (USEPA 2002). The central limit theorem states that, for sufficiently large sample size, the sampling distribution of means (i.e., the sample of sample means) approximately is distributed normally, regardless of the actual underlying distribution of the data (USEPA 2002). The sample sizes for the 7 air contaminants of concern ranged between 60 and 65; ND values in each dataset were replaced by one-half the sample quantitation limit. The 95% upper confidence limits for each air contaminant are summarized in Table 2.
The absorbed dose (DB) associated with the chemical air concentration (Ca) was calculated for the inhalation exposure pathway using Equation 1,
where Bs is the chemical-specific bioavailability for inhalation, IH is the inhalation rate, ET is the exposure time, and BW is the body weight. The chronic daily intake (CDI), which was used to evaluate potential noncarcinogenic effects, was estimated using Equation 2,
where EF is the exposure frequence, ED is the exposure duration, and AT is the average time. The lifetime average daily absorbed dose or intake (LADD or LADI) was used to evaluate carcinogenic risk, and was calculated using Equation 3.
Table Table 2.. The 95% upper confidence limits (UCLs) for contaminants
|Contaminant||Sample size||Concn. range (μg/m3)||95% UCL (μg/m3)|
The slope factors (SF) and reference doses (RfD) used to quantify carcinogenic and noncancer health risks are summarized in Table 3 and were obtained from the USEPA Integrated Risk Information System database (USEPA 2003) and USEPA National Center for Environmental Assessments. Using Equation 4, the product of multiplying the LADD by the SF is the theoretical incremental cancer risk (IELCR) associated with chemical exposure
Using Equation 5, noncarcinogenic risk is calculated by dividing the CDI for each chemical by the chemical-specific RfD. The ratio, which is referred to as the hazard quotient (HQ), determines whether exposure is likely to result in noncarcinogenic adverse health impacts. The hazard index (HI) is an indicator of the likelihood of adverse (noncarcinogenic) health effects occurring due to cumulative chemical exposure, and is obtained by summing the HQs associated with each contaminant of concern. If the HI is less than or equal to a value of 1, then cumulative exposure to different contaminants is not expected to cause adverse health effects.
Risk-assessment results for the worker population
Risk estimates are summarized in Table 4. Figure 1 shows the theoretical cancer risk by chemical assuming average worker exposure conditions. The average, or best estimate, incremental lifetime cancer risk was 1.6 × 10–5. This indicates that, based on assumed average exposure conditions, the additional risk of cancer attributable to cumulative exposure to the contaminants of concern is approximately 2 in 100,000, which is equivalent to 20 in 1 million. For upperbound, or worst-case, exposure conditions, the estimated incremental lifetime cancer risk was 7.1 × 10–5, which is equivalent to a population cancer risk of approximately 70 in 1 million. The USEPA (1989) regards cancer risks ranging between 1 in 10 million (i.e., 10–7) and 100 in 1 million (equivalent to 10–4) as within the range of acceptable risk. Both the average and worst-case results were within the USEPA (1989) range of acceptable risk.
Table Table 3.. Contaminant toxicity characteristics and target organsa
|Contaminant||SFbc (mg/kg/d)–1||RfDcd (mg/kg/d)–1||Critical effects for RfD||Target organs|
|Benzene||2.7 × 10–2 (I)||8.6 × 10–3 (I)||Decreased lymphocyte count||Eyes, skin, respiratory system blood, CNSe|
|Chloroform||8.1 × 10–2 (I)||1.4 × 10–2 (E)||Liver cysts||Eyes, skin, liver, kidneys, heart, CNS|
|Tetrachloroethylene||2.0 × 10–2 (R)||1.4 × 10–1 (E)||Hepatoxicity (in mice)||Eyes, skin, respiratory system, liver, kidneys, CNS|
|Trichloroethylene||4.0 × 10–1 (E)||1.00 × 10–2 (E)||Liver, kidneys, fetus development||Eyes, skin, respiratory system, heart, liver, kidneys, CNS|
|Beryllium||8.4 (I)||5.7 × 10–6 (I)||CBDf (lung lesions)||Eyes, skin, respiratory system|
|Cadmium||6.3 (I)||5.7 × 10–5 (E)||Proteinuria (kidney impairment)||Respiratory system, kidneys, prostate, blood|
|Mercury||No Data||8.6 × 10–5 (I)||Hand tremors, autonomic dysfunction, memory loss||Eyes, skin, CNS, PNS,g kidneys|
Table Table 4.. Summary of risk-assessment results
|Risk||Worker HIa||Worker CRb||Inmate HI||Inmate CR|
|Average estimates||4.5 × 10–1||1.6 × 10–5||2.1||1.2 × 10–4|
|Worst-case estimates||8.4 × 10–1||7.1 × 10–5||3.3||4.0 × 10–4|
Figure 2 shows calculated HI values for each air contaminant assuming average worker exposure conditions. Assuming average exposure conditions, the best estimate HI was 0.45 (Table 4). The upperbound, or worst-case, HI was 0.84. Both results were less than a value of 1, indicating that, even under assumed worst-case exposure conditions, the detected contaminant concentrations in air are lower than the concentrations that would be expected to cause adverse health effects.
Risk-assessment results for the inmate population
Figure 1 shows the theoretical cancer risk by chemical assuming average inmate exposure conditions, and indicates that the main chemical of concern was trichloroethylene. The average, or best estimate, incremental lifetime cancer risk was 1.2 × 10–4 (i.e., 100 in 1 million), which falls within the USEPA (1989) acceptable risk range. Assuming upperbound, or worst-case, exposure conditions, the incremental lifetime cancer risk was 4.0 × 10–4 (400 in 1 million), which significantly exceeded the USEPA (1989) acceptable risk range. It is highly unlikely, however, that worst-case exposure conditions would occur at the prison facility.
Figure 2 summarizes the HI by chemical assuming average inmate exposure conditions, and indicates that the main chemicals of concern are mercury and beryllium. The average and worst-case HIs for the inmate population were 2.1 and 3.3, respectively. Inmate population exposure levels for the contaminants of concern exceeded levels that potentially could result in adverse (noncarcinogenic) health impacts. Compared to worker exposure conditions, the single most significant exposure characteristic responsible for the high inmate HI results is the exposure time, that is, the number of hours per day spent indoors at the prison. Table 1 shows that exposure time is the characteristic that varies the most between both populations. The exposure frequency (i.e., number of days per y that a receptor is exposed) also affects the risk assessment results significantly.