Estimating population distributions when some data are below a limit of detection by using a reverse Kaplan-Meier estimator.

BACKGROUND: Data with some values below a limit of detection (LOD) can be analyzed using methods of survival analysis for left-censored data. The reverse Kaplan-Meier (KM) estimator provides an effective method for estimating the distribution function and thus population percentiles for such data. Although developed in the 1970s and strongly advocated since then, it remains rarely used, partly due to limited software availability. METHODS: In this paper, the reverse KM estimator is described and is illustrated using serum dioxin data from the University of Michigan Dioxin Exposure Study (UMDES) and the National Health and Nutrition Examination Survey (NHANES). Percentile estimates for left-censored data using the reverse KM estimator are compared with replacing values below the LOD with the LOD/2 or LOD/ radical2. RESULTS: When some LODs are in the upper range of the complete values, and/or the percent censored is high, the different methods can yield quite different percentile estimates. The reverse KM estimator, which is the nonparametric maximum likelihood estimator, is the preferred method. Software options are discussed: The reverse KM can be calculated using software for the KM estimator. The JMP and SAS (SAS Institute, Cary, NC) and Minitab (Minitab, Inc, State College, PA), software packages calculate the reverse KM directly using their Turnbull estimator routines. CONCLUSION: The reverse KM estimator is recommended for estimation of the distribution function and population percentiles in preference to commonly used methods such as substituting LOD/2 or LOD/ radical2 for values below the LOD, assuming a known parametric distribution, or using imputation to replace the left-censored values.

Estimation of background serum 2,3,7,8-TCDD concentrations by using quantile regression in the UMDES and NHANES populations.

BACKGROUND: The goal of the present study was to quantify the population-based background serum concentrations of 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) by using data from the reference population of the 2005 University of Michigan Dioxin Exposure Study (UMDES) and the 2003-2004 National Health and Nutrition Examination Survey (NHANES). METHODS: Multiple imputation was used to impute the serum TCDD concentrations below the limit of detection by combining the 2 data sources. The background mean, quartiles, and 95th percentile serum TCDD concentrations were estimated by age and sex by using linear and quantile regressions for complex survey data. RESULTS: Any age- and sex-specific mean, quartiles, and 95th percentiles of background serum TCDD concentrations of study participants between ages 18 and 85 years can be estimated from the regressions for the UMDES reference population and the NHANES non-Hispanic white population. For example, for a 50-year-old man in the reference population of UMDES, the mean, quartiles, and 95th percentile serum TCDD concentrations are estimated to be 1.1, 0.6, 1.1, 1.8, and 3.3 parts per trillion, respectively. The study also shows that the UMDES reference population is a valid reference population for serum TCDD concentrations for other predominantly white populations in Michigan. CONCLUSION: The serum TCDD concentrations increased with age and increased more over age in women than in men, and hence estimation of background concentrations must be adjusted for age and sex. The methods and results discussed in this article have wide application in studies of the concentrations of chemicals in human serum and in environmental samples.

Hierarchical cluster analysis of polychlorinated dioxins and furans in Michigan, USA, soils: evaluation of industrial and background congener profiles.

As part of the University of Michigan Dioxin Exposure Study, soil samples were collected from 766 residential properties near the Tittabawassee River between Midland and Saginaw; near the Dow Chemical Facility in Midland; and, for comparison, in the other areas of Midland and Saginaw Counties and in Jackson and Calhoun Counties, all located in the state of Michigan, USA. A total of 2,081 soil samples were analyzed for 17 polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs). In order to better understand the distribution and sources of the PCDD/F congeners in the study area, hierarchical cluster analysis (HCA) was used to statistically group samples with similar congener patterns. The analysis yielded a total of 13 clusters, including: 3 clusters among the soils impacted by contamination present in the Tittabawassee River sediments, a cluster comprised mainly of samples collected within the depositional area of the Dow incinerator complex, a small cluster of samples with elevated 2,3,7,8-tetrachlorinated dibenzo-p-dioxin (TCDD), and several clusters exhibiting background patterns. The clusters related to the Tittabawassee River floodplain contamination all contained elevated PCDF levels and were differentiated from one another primarily by their relative concentrations of higher-chlorinated PCDDs, a difference likely related to both extent and timing of impacts from Tittabawassee sediments. The background clusters appear to be related to combustion processes and are differentiated, in part, by their relative fractions of TCDD. Thus, HCA was useful for identifying congener profile characteristics in both contaminated and background soil samples.

Accumulation of DNAPL waste in subsurface clayey lenses and layers.

Chlorinated solvents like trichloroethylene (TCE) and tetrachloroethylene (PCE) are prevalent groundwater contaminants being detected more than half of the Superfund Sites. They are introduced into the subsurface due to improper disposal of hazardous wastes containing these chlorinated dense non-aqueous phase liquids (DNAPL). DNAPL wastes sink in the subsurface and form pools on clay layers. Further transport into the underlying clay layer is dominated by diffusion due to the low permeability of these soils and results in accumulation of contaminants therein overtime through diffusion. However, field evidence reported that mass storage of solvents in such zones was higher than what can be attributed to simple diffusion. To evaluate this finding, mass storage of TCE in a hypothetical aquitard after 30 years of diffusion was calculated. The diffusion coefficient reported in the field study resulted in mass storage of 137.7 g in this hypothetical aquitard whereas the mass calculated using the measured diffusion coefficient of TCE from pure solvent into water-saturated clayey soil was 25-65% of it (35.0-89.4 g). The calculated mass storage was even lower (8.2 g) if the measured diffusion coefficient of TCE from DNAPL waste into waste-contacted soil was used. So, mass of TCE accumulated through sole diffusion was not able to explain the one observed in the field. This excess mass in the field might be the mass of DNAPL entered into the cracks which were reported to form in clay layers as a result of the direct contact between water saturated clay and DNAPL waste. Mass of TCE in the cracks was calculated using reported average crack size and an assumed crack depth, and it appeared that cracks filled with DNAPL could increase the stored mass up to a minimum of 334 g which could easily account for the enhanced mass storage observed in the field.

Measurement of asbestos emissions associated with demolition of abandoned residential dwellings.

Many cities are revitalizing their urban cores through the demolition of abandoned residential dwellings (ARDs). However, data regarding the emissions of asbestos during such an operation are sparse. We measured airborne asbestos emissions from emergency demolitions (demolitions on structures deemed too dangerous to enter and remove asbestos) of ARDs in Detroit. High-flow air sampling was conducted during ARD demolitions. Air samples were analyzed using Phased Contrast Microscopy (PCM), and a subset using Transmission Electron Microscopy (TEM). One hundred and one air samples were collected on 25 emergency demolitions. Fifty-four of the 101 PCM samples (53%) exceeded the limit of detection (LOD). However, only 2 of 46 TEM samples (4%) exceeded the LOD for asbestos; these latter samples were from two different demolitions and each contained a single chrysotile asbestos fiber. Using conservative exposure assumptions and commonly-accepted risk estimation formulae, we estimated the lifetime risk of mesothelioma and lung cancer combined to be less than one case per one million people. Emissions of airborne asbestos during emergency (unabated) ARD demolition operations appear to be negligible. As a result, the associated health risk for asbestos-related disease is also negligible. Reconsideration of current regulatory mandates for asbestos abatement in ARDs may be warranted.

An investigation of homes with high concentrations of PCDDs, PCDFs, and/or dioxin-like PCBs in house dust.

As part of the University of Michigan Dioxin Exposure Study, the 29 congeners of polychlorinated dibenzo-p-dioxins, polychlorinated dibenzofurans, and dioxin-like polychlorinated biphenyls that have World Health Organization consensus toxic equivalency factors were measured in house dust from 764 homes using a population-based sampling design over selected regions in five Michigan counties. Twenty homes had a total toxic equivalency in house dust that was more than 2.5 standard deviations above the mean (i.e., defined to be outliers). This follow-up investigation describes the outlier house dust measurements and corresponding soil measurements and explores possible sources of these toxins in house dust. The congener distributions in the house dust outliers varied and were dominated (i.e., >50% of the total toxic equivalency) by either polychlorinated dibenzo-p-dioxins (n = 9), polychlorinated dibenzofurans (n = 1), or dioxin-like polychlorinated biphenyls (n = 9). Likely sources of contamination of house dust were identified in only three cases. In two cases, dust contamination appeared to be related to contaminated soil adjacent to the home; in one case, contamination was related to a source within the home (a carpet pad). In most cases, the source(s) of contamination of house dust could not be identified but appeared likely to be related to uncharacterized sources within the homes.

Case report: human exposure to dioxins from clay.

CONTEXT: For the general population, the dominant source of exposure to dioxin-like compounds is food. As part of the University of Michigan Dioxin Exposure Study (UMDES), we measured selected polychlorinated dibenzo-p-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs), and dioxin-like polychlorinated biphenyls (PCBs) in serum of 946 subjects who were a representative sample of the general population in five Michigan counties. CASE PRESENTATION: The total toxic equivalency (TEQ; based on 2005 World Health Organization toxic equivalency factors) of serum from the index case was 211 ppt on a lipid-adjusted basis, which was the highest value observed in the UMDES study population. This subject had no apparent opportunity for exposure to dioxins, except that she had lived on property with soil contaminated with dioxins for almost 30 years, and had been a ceramics hobbyist for > 30 years. Soil from her property and clay that she used for ceramics were both contaminated with dioxins, but the congener patterns differed. DISCUSSION: The congener patterns in this subject's serum, soil, and ceramic clay suggest strongly that the dioxin contamination in clay and not soil was the dominant source of dioxin contamination in her serum. RELEVANCE TO PUBLIC HEALTH PRACTICE: It appears that ceramic clay, in particular the process of firing clay with unvented kilns, can be a significant nonfood and nonindustrial source of human exposure to dioxins among ceramics hobbyists. The extent of human exposure from ceramic clay is unclear, but it may be widespread. Further work is needed to more precisely characterize the routes of exposure.

Wettability hysteresis and its implications for DNAPL source zone distribution.

Subsurface heterogeneity at sites contaminated with nonaqueous phase liquids (NAPLs) reduces the effectiveness of traditional remediation measures. One cause may be the increased proportion of NAPL that is hydraulically isolated due to capillary trapping in heterogeneously-wetted materials. This study examines the wettability of ten materials, ranging from minerals, such as calcite and dolomite, to carbonaceous materials, such shale and coal, in air and water, NAPL and air, and NAPL and water systems. The wettability differed depending on which phase the solid material was initially immersed in: the less crystalline solids, if initially contacted by water were water-wet, but if initially contacted by NAPL were NAPL-wet. This difference, termed here wettability hysteresis, was observed for a suite of halogenated NAPLs and was independent of equilibration time. The degree of wettability hysteresis was greatest in the NAPL and water systems, with the magnitude of the difference increasing with the carbonaceous materials. Since the degree of capillary trapping in subsurface materials is related to wettability, the phenomenon of wettability hysteresis suggests that system history is a factor that may increase the heterogeneity of NAPL source zones.

Effective diffusion coefficients of DNAPL waste components in saturated low permeability soil materials.

Diffusion is regarded as the dominant transport mechanism into and out of low permeable subsurface lenses and layers in the subsurface. But, some reports of mass storage in such zones are higher than what might be attributable to diffusion, based on estimated diffusion coefficients. Despite the importance of diffusion to efforts to estimate the quantity of residual contamination in the subsurface, relatively few studies present measured diffusion coefficients of organic solutes in saturated low permeability soils. This study reports the diffusion coefficients of a trichloroethylene (TCE), and an anionic surfactant, Aerosol OT (AOT), in water-saturated silt and a silt-montmorillonite (25:75) mixture, obtained using steady-state experiments. The relative diffusivity ranged from 0.11 to 0.17 for all three compounds for the silt and the silt-clay mixture that was allowed to expand. In the case in which the swelling was constrained, the relative diffusivity was about 0.07. In addition, the relative diffusivity of 13C-labeled TCE through a water saturated silt-clay mixture that had contacted a field dense non-aqueous phase liquid (DNAPL) for 18months was measured and equaled 0.001. These experimental results were compared with the estimates generated using common correlations, and it was found that, in all cases, the measured diffusion coefficients were significantly lower than the estimated. Thus, the discrepancy between mass accumulations observed in the field and the mass storage that can attributable to diffusion may be greater than previously believed.

A mechanism of basal spacing reduction in sodium smectitic clay materials in contact with DNAPL wastes.

There has been concern regarding the possible attack of clays in aquitards, slurry walls and landfill liners by dense nonaqueous phase liquid (DNAPL) wastes, resulting in cracking. Despite the fact that a reduction in basal spacing in sodium smectitic clay materials has been linked to cracking, no plausible mechanism by which this reduction occurs in contact with waste DNAPLs has been formulated. To elucidate a mechanism, screening studies were conducted that showed that the combination of an anionic surfactant (AOT), a nonionic surfactant (TritonX-100) and a chlorinated solvent, tetrachloroethylene (PCE), could replicate the basal spacing reduction and cracking behavior of water-saturated bentonite caused by two waste DNAPLs obtained from the field. FTIR measurements of this system showed a displacement of the HOH bending band of water symptomatic of desiccation. Sorption measurements showed that the uptake of AOT by bentonite increased eight fold in the presence of TritonX-100 and PCE. The evidence presented here supports a mechanism of syneresis, involving the extraction of water from the interlayer space of the clay through the synergistic sorption of a nonionic and anionic surfactant mixture. It is speculated that the solvation of water in reverse micellar aggregates is the process driving the syneresis.

Prediction of two-phase capillary pressure-saturation relationships in fractional wettability systems.

Capillary pressure/saturation data are often difficult and time consuming to measure, particularly for non-water-wetting porous media. Few capillary pressure/saturation predictive models, however, have been developed or verified for the range of wettability conditions that may be encountered in the natural subsurface. This work presents a new two-phase capillary pressure/saturation model for application to the prediction of primary drainage and imbibition relations in fractional wettability media. This new model is based upon an extension of Leverett scaling theory. Analysis of a series of DNAPL/water experiments, conducted for a number of water/intermediate and water/organic fractional wettability systems, reveals that previous models fail to predict observed behavior. The new Leverett-Cassie model, however, is demonstrated to provide good representations of these data, as well as those from two earlier fractional wettability studies. The Leverett-Cassie model holds promise for field application, based upon its foundation in fundamental scaling principles, its requirement for relatively few and physically based input parameters, and its applicability to a broad range of wetting conditions.

Serum concentrations of polychlorinated dibenzo-p-dioxins among ceramicists.

Polychlorinated dibenzo-p-dioxins (PCDDs) occur naturally in ball clay at elevated concentrations. Thus, persons who habitually work with clay may be at risk for exposure to PCDDs. An earlier case report provided some evidence of elevated PCDD levels in serum for long-term hobby ceramicists; however, no previous study has measured serum dioxin concentrations among ceramicists. This study measured PCDD serum levels for 27 individuals involved in ceramics making. The average residual, defined as the average of the [log measured serum lipid concentration ­ log background serum lipid concentration], was calculated and then tested to determine whether it was significantly different from zero. The p-values for the average residuals indicated that the serum lipid concentrations for several PCDD congeners were elevated relative to background. The number of significant residuals increased dramatically if the background concentrations were adjusted to account for the fact that they were not contemporaneous with the measurements for the ceramicists. The ratio of the 1,2,3,6,7,8-HxCDD concentration to the 1,2,3,7,8,9-HxCDD concentration was greater than 1.0, unlike in ball clay, suggesting that although long-term working with ball clay elevates the PCDD levels in serum somewhat, it is not the predominant source of the PCDD body burden for ceramicists.

Dependence of particle concentration effect on pH and redox for arsenic removal by FeS-coated sand under anoxic conditions.

FeS has been recognized as a good scavenger for arsenic under anoxic conditions. To create a suitable adsorbent for flow-through reactors such as permeable reactive barriers, it has been suggested that this material may be coated onto sand. However, previous work on FeS-coated sand has focused on batch reactors, while flow-through reactors usually have higher solid-solution ratios. To ascertain whether differences in the solid-solution ratio (SSR) are important in this system, batch sorption experiments were conducted as a function of pH using As(III) and FeS-coated sands at various solid-solution ratios. The results showed little variation in the distribution coefficient with SSR at pH 7 and 9. However, at pH 5, the results showed lower values of the distribution coefficient at lower SSRs, the reverse of typically reported SSR effects. Measured pe values showed a dependence on SSR, which, when coupled with chemical modeling of the Fe-As-S-H2O system, suggested a change in the removal mechanism with SSR, from adsorption to a reduced Fe(II) oxyhydroxide phase (represented by Fe2(OH)5) to precipitation as As2S3 or AsS. On the other hand, at pH 7 and 9, arsenite adsorption is the most probable removal mechanism regardless of the pe. Thus, this study identified variations in pH and redox conditions, and the removal mechanisms that these parameters govern, as the reason for the apparent SSR effect.

Estimation of primary drainage three-phase relative permeability for organic liquid transport in the vadose zone.

The modeling of transport of organic liquid contaminants through the vadose zone often requires three-phase relative permeabilities. Since these are difficult to measure, predictive models are usually used. The objective of this study is to assess the ability of eight common models to predict the drainage relative permeability to oil in a three-phase system (water-oil-air). A comparison of the models' estimates using data set from Oak [Oak, M.J., 1990. Three-phase relative permeability of water-wet Berea. In: Seventh Symposium on Enhanced Oil Recovery, Paper SPE/Doe 20183. Tulsa, OK, April 22-25] showed that they provide very different predictions for the same system. The goodness of the models does not increase with the amount of data or computation that the models require. Also, the calculations showed how different interpretations of the models and of the terminology associated with them can significantly impact the predictions. Thus, considerable error may be introduced into the simulations of organic liquid transport in the vadose zone depending on the selection and interpretation of the three-phase relative permeability model.

Impact of DNAPL contact on the structure of smectitic clay materials.

Smectitic clays have a flexible structure that may be impacted by contact with dense nonaqueous phase liquids (DNAPLs) present at hazardous waste sites. Measurements of the basal spacing of air-dry clays contacted with pure chlorinated solvents and chlorinated DNAPL wastes showed that the intraparticle spacing is similar to that in air. Basal spacings of water-saturated clays contacted with pure chlorinated solvents are similar to those in contact with water, even after extended equilibration times (300 d). In contrast, contact with chlorinated DNAPL wastes reduced the basal spacing of water-saturated sodium smectites in a relatively short time frame, resulting in cracks that were as large as 1mm in aperture. The penetration of these wastes into the intraparticle spacing of clay and the resultant cracking may contribute to the accumulation of chlorinated compounds in clay layers observed in the field and the extended remediation times associated with this mass storage.

Evaluation of the release of dioxins and PCBs during kiln-firing of ball clay.

Ball clay is known to be naturally contaminated with high levels of polychlorinated di-benzo-p-dioxins (PCDDs). This study evaluated the potential for PCDD, polychlorinated dibenzofuran (PCDF) and polychlorinated biphenyl (PCB) release during the kiln firing of ball clay in an art studio. Toxic equivalence (TEQ) were calculated using World Health Organization (WHO) 2005 toxic equivalence factors (TEF) and congener concentrations. Ten bags of commercial ball clay were found to have an average TEQ of 1,370 nanograms/kilogram (ng kg(-1)) dry weight (dw), almost exclusively due to PCDDs (99.98% of TEQ). After firing, none of the 29 dioxin-like analytes was measured above the limits of detection (LOD) in the clay samples. Air samples were taken during firings using both low-flow and high-flow air samplers. Few low-flow air samples contained measurable levels of dioxin congeners above the LOD. The mean TEQ in the high volume air samples ranged from 0.07 pg m(-3) to 0.21 pg m(-3) when firing ball clay, and was 0.11 pg m(-3) when no clay was fired. These concentrations are within the range measured in typical residences and well-controlled industrial settings. The congener profiles in the high-flow air samples differed from the unfired clay; the air samples had a considerable contribution to the TEQ from PCDFs and PCBs. Given that the TEQs of all air samples were very low and the profiles differed from the unfired clay, it is likely that the PCDDs in dry ball clay were destroyed during kiln firing. These results suggest that inhalation of volatilized dioxins during kiln firing of dry ball clay is an unlikely source of exposure for vocational and art ceramicists.

Permeability of iron sulfide (FeS)-based materials for groundwater remediation.

Iron sulfide (FeS) has been extensively assessed as a reactive medium to remove both metals and halogenated organics from groundwater. However, to address its suitability as a material for permeable reactive barriers (PRBs), its propensity for solids and gas production, which result in reduced permeability, must be evaluated. The reduction in permeability for sands coated with FeS (as mackinawite), under the anoxic conditions often encountered at contaminated groundwater sites, was examined through column experiments and geochemical modeling under conditions of high calcium and nitrate, which have been previously shown to cause significant permeability reduction in zero-valent iron (ZVI) systems. The column experiments showed negligible production of both solids and gases. The geochemical modeling predicted a maximum reduction in permeability of 1% due to solids and about 30% due to gas formation under conditions for which a complete loss of permeability was predicted for ZVI systems. This difference in permeability reduction is driven by the differences in thermodynamic stability of ZVI and FeS in aqueous solutions. The results suggest that geochemical conditions that result in high permeability losses for ZVI systems will likely not be problematic for FeS-based reactive materials.

Impact of dissolved silica on arsenite removal by nano-particulate FeS and FeS-coated sand.

This work evaluated the inhibitory effect of dissolved silica on arsenite adsorption to nanoparticulate FeS (NP-FeS) or mackinawite and FeS-coated sand (CS-FeS) sorbents. Arsenite retention by the NP-FeS solid was not affected by dissolved silicate over a wide range in pH, in contrast to the known inhibitory effect of dissolved silica on As(III) uptake by Fe-(hydr)oxide systems. However, some inhibition was observed in CS-FeS system at pH 9. This latter result is attributed to the co-existence of both FeS and small amounts of Fe-(hydr)oxide phases on the sand surface. Given the ubiquitous presence of dissolved Si in groundwater, FeS-based sorbents may have an advantage for As retention compared to those based on Fe-(hydr)oxides in reducing subsurface environments.

Estimation of age- and sex-specific background human serum concentrations of PCDDs, PCDFs, and PCBs in the UMDES and NHANES populations.

Age- and sex-specific estimates of serum dioxin concentrations are important for comparisons among populations. However, such comparisons are problematic because populations have different age and sex structures and values are typically reported only in broad age ranges that are not comparable across studies. There are few studies that report congener-specific serum concentrations, and none that provide these by sex for age as a continuous function. We combined the NHANES 2003-2004 data with the University of Michigan Dioxin Exposure Study (UMDES) referent population 2005 data to achieve stable and accurate estimates of mean and quantiles of serum dioxins by sex over ages 18-85. Survey-weighted linear and quantile regression models were fitted on the combined data with the log-transformed congener concentration as outcome and age, sex, and data source as covariates. Formulas are provided to allow calculations of age- and sex-specific mean and quantile estimates over ages 18-85. For instance, the geometric mean, median, 75th percentile, and 95th percentile of serum TEQ for men aged 50 can be estimated, respectively, from the formulas as 18.33, 19.02, 22.60, and 30.37 pg g(-1) lipids among the Michigan general population, and as 15.71, 15.89, 22.60, and 29.90 pg g(-1) lipids among US non-Hispanic whites. These methods and results are useful for comparing the congener-specific human serum dioxin concentrations in any individual to the general population mean, median, 75th percentile, and 95th percentile, and for comparing the serum dioxin concentration in any group of interest to the US and the Michigan general populations.