Assessment of the Presence of Soil Lead Contamination Near a Former Lead Smelter in Mombasa, Kenya

Background. The informal settlement of Owino Uhuru near an abandoned lead smelter attracted international attention due to an apparent lead poisoning event. Despite this attention, the environmental data collected to date do not indicate high levels of residual contamination. Objectives. To further confirm previous findings and determine any necessary risk mitigation measures, an assessment of surface soil lead concentrations was conducted in the community. Methods. Investigators carried out an assessment of the soil in a ~12,000 m2 section of the Owino Uhuru neighborhood over the course of a single day in June 2017 with the assistance of community leaders. Fifty-nine in situ soil measurements were taken using an Innov-X tube-based (40 kV) alpha X-ray fluorescence instrument (pXRF). Results. The assessment found that mean surface soil lead concentrations in areas conducive to exposure were 110 mg/kg (95% CI: 54–168); below United States Environmental Protection Agency and the Environment Canada screening levels of 400 mg/kg and 140 mg/kg, respectively. Conclusions. There is likely no current need for risk mitigation activities in the community. These results could inform discussions on the allocation of public health spending. Competing Interests. The authors declare no competing financial interests. BE, VOO, CN and JSF are employees of Pure Earth. MPT sits on the Editorial Board of the Journal of Health and Pollution.


Introduction
Lead smelting is a significant source of global soil contamination. 1-4 Surface soil lead concentrations in residential areas can result in exposure to humans through pica (hand-tomouth) behavior and the inhalation and ingestion of contaminated soil and dust. 5, 6 Accordingly, several studies have demonstrated that blood lead levels (BLLs) are strongly associated with soil lead concentrations. 6-11 There is currently no known safe level of exposure to lead, although the United States Centers for Disease Control (CDC) utilizes a reference dose of 5 µg/dL. 12,13 Childhood lead exposure can result in intelligence quotient decrement, decreased lifetime earnings and higher rates of aggravated assault, among other adverse outcomes. [14][15][16][17][18] Early life exposures have also been shown to not remit with age. 17, 19 Although rare, extreme acute lead exposure can result in encephalopathy and death. 20 Lead exposure in adults most significantly results in increased incidence of heart disease, even at low levels of exposure. 20, 21 The Institute for Health Metrics and Evaluation (IHME) estimates that lead exposure accounted for 540,000 deaths globally in 2016. 22 Over the course of the 20th century, environmental lead contamination was most strongly associated with the use of tetraethyl lead additives in gasoline. [23][24][25][26] In the United States, lead-based paint was also a significant source of exposure. 27,28 Following the cessation of the use of lead in these two common products, blood levels have fallen significantly. For example, in the United States, average childhood blood lead declined from around 15 µg/dL in the mid-1970s, falling to < 1 µg/dL today. 27,29 Elsewhere in high-income countries, persistently elevated environmental and biological lead levels continue to be documented around mining and smelting locations. 30,31 In low-and middle-income countries (LMIC), current major sources of lead exposure include traditional ceramic glazes and the manufacture and recycling of lead-acid batteries, particularly when conducted in an informal setting. 32-35 Discrete lead poisoning events have also been identified at mining and smelting locations. 2,36-38 Research and blood lead concentrations have been documented in occupational settings. 39 However, there is a paucity of information about exposures in the home or in residential areas. Data collected as part of the Pure Earth (New York, NY, USA) Toxic Sites Identification Program indicate that domestic lead exposure may be significant. 40 The program has identified and conducted rapid assessments of 60 discrete lead contaminated sites located in residential areas in Kenya since 2009.
A lead poisoning event in 2014 centered around a lead smelter in the coastal city of Mombasa garnered international attention. 41-45 News reports documented elevated BLLs and three deaths in the worker population at the refinery. 46 Limited reports of elevated BLLs were also later found in Owino Uhuru, an informal settlement of approximately 3,000 residents with an area of 28 000 m 2 bordering the northern wall of the facility ( Figure  1). A 2010 study of three children found BLLs of 12, 17 and 23 µg/ dL, while a later report of a separate child found a BLL of 32 µg/dL. 43,46 A community leader in Owino Uhuru and former employee of the smelter was awarded the prestigious Goldman Environmental Prize for her work in raising public awareness that would eventually lead to the closure of the facility in 2014. 45, 47 The community subsequently filed a lawsuit against the national government for USD 1.5 million in compensation for its failure to monitor emissions from the smelter. 48 Following the closure of the facility in 2015, a study of environmental lead concentrations and community BLLs was carried out jointly by the Kenya Ministry of Health and the CDC. The study found a geometric mean surface soil lead concentration in the community of 146.5 mg/kg (geometric Despite the absence of evidence of residual contamination, the site continues to attract national and international attention. 55,56 To further assess the veracity of the claims of contamination and provide the basis for necessary human health intervention strategies, an investigation of surface soil lead concentrations was carried out in Owino Uhuru in June 2017. This type of assessment forms the evidencebased rationale for any subsequent actions required to mitigate potential risk of harm arising from soil and dust contamination. Moreover, the results of the assessment in the context of its high profile potentially provide insight into the setting of public health priorities.
The results of this assessment are presented along with a simple air deposition model developed to estimate likely surface soil lead concentrations resulting from smelter emissions during its operation. The results of these models were used to estimate BLLs in children and adults in the absence of population data on exposures.

Methods
Investigators carried out an assessment over the course of a single day in June 2017 with the assistance of community leaders. Fifty-nine in situ soil measurements were taken using an Innov-X tube-based (40 kV) alpha X-ray fluorescence instrument (pXRF) over a ~12 000 m 2 section of the Owino Uhuru neighborhood that is adjacent to the facility. The pXRF has a lower detection limit of 5 mg/ kg. 57 Fifty-seven measurements were taken directly from surface soil, while two were taken at a depth of 10 cm. Two of the surface soil measurements were taken from an area within the perimeter wall of the smelter, which are unlikely to be accessed by humans and thus are not indicative of community exposure. The pXRF was calibrated before the assessment using an alloy-grade 316 steel clip and measurement accuracy was evaluated by assessment of a National Institute of Standards and Technology (NIST) standard (2702: Inorganics in Marine Sediment) during the assessment. 58 The NIST reference material contains a known value for lead of 132.8 mg/ kg. The pXRF measurement of this material found a value of 137 mg/kg (+/-10) for lead and was thus within acceptable range. The inside of the facility was not accessible and was not assessed.

Spatial and statistical analysis
Latitude and longitude for each sample point were collected using World Geodetic System 1984 format using a Garmin eTrex 10 with an accuracy of < 3 meters. 59 Spatial and statistical analyses were performed using ArcMap 10.5 and Stata 15. 60, 61 Basic descriptive statistics of the data were generated to assess exposure. In addition, simple linear regression was conducted to assess any relationship between lead concentration and proximity to the smelter.

Aerial deposition model
To determine soil lead concentrations resulting from aerial emissions, a simple algorithm was developed based on known deposition rates of lead smelters in different settings. To determine the relevant mass of soil, a volume was calculated based on a likely penetration of deposited lead to a maximum depth of 2 cm. The 2 cm value is based on studies of the isotopic composition of soil lead at a smelter in Mount Isa, Australia. 73,74 Mackay et al. found that lead found below this depth tended to be associated with naturally occurring deposits, rather than aerial deposition from the smelter. 73 The soil type at the site is Haplic Lixisol with an approximate clay, silt and sand content of 18%, 27% and 55%, respectively and a mass of roughly 1.4 grams/cm 3 . 75 These soils are conducive to metals mobility more generally, as discussed below. For the purpose of the sensitivity analysis the model was also run with a depth of 5 cm. Studies at a smelter in Boolaroo, Australia found deposited lead at a maximum depth of 5 cm with 80% less lead in the lower 2.5 cm than the top 2.5 cm. 76 Background lead concentrations for the study area were not available. As an alternative, the mean background lead concentration for the earth's crust of 17 mg/kg was used. 77  The IEUBK model was also used to estimate likely environmental Pb levels in air and soil required for a hypothetical 2-year-old child to have a BLL of 20 µg/dL. The IEUBK model assigns this age a higher BLL than younger or older age groups. It was selected to provide the most sensitivity to environmental levels.

Results
The results of both the in situ surface soil measurements and aerial deposition modeling indicate that environmental lead levels in Owino Uhuru are within or slightly above US regulatory screening levels and generally consistent with urban areas globally.

Surface soil assessment
The mean surface soil lead concentration in the areas assessed with the pXRF was 224 mg/kg (95% CI: 15-434). The median value was 47 mg/kg. Within the targeted 12,000 m 2 sample area, soil lead measurements were spaced an average of 9.4 m apart (95% CI: 7.8-11.2) (Figure 1). There was no statistically significant association between proximity to the smelter and soil lead concentrations (p<0.05). The mean soil lead concentration of the eight samples taken within 3 m of the facility perimeter wall was 1,026 mg/ kg (95% CI: 611-2663). The mean for the eight samples taken from 3 m to 10 m was 231 mg/kg (95% CI: 183-646) and the mean for the 43 samples taken beyond 10 m was 66 mg/kg (95% CI: 47-86). The sample taken closest to the smelter site was at the base of the perimeter wall, while the furthest was taken at a distance of 130 meters.

Aerial deposition model results
Using van Alphen's mean deposition rate of 18 mg/m 2 /day and a 2 cm estimate for the likely maximum penetration of lead into surface soil resulted in an additional accumulation of 0.64 mg/kg/day within 750 meters of the facility while it was operating. Using the same deposition rate and the less conservative surface soil penetration estimate of 5 cm results in an additional accumulation of 0.26 mg/kg/day. These rates would have resulted in a surface soil concentration of 686-1,688 mg/ kg after ten years of operation. To arrive at the mean value identified in pXRF sampling of 110 mg/kg, a daily deposition rate of 1-2.51 mg/m 2 /day (0.036 mg/kg/day) would be required.

Blood lead level assessment
Current BLLs for 0-to 7-year-olds were estimated to be from 1. Ericson et al

Table 1 -Descriptive Statistics of Surface Soil Concentrations in Owino Uhuru
However, even in locations where all or some of these conditions are present, very limited mobility of lead through soil profiles has been reported. 87- 89 Teutsch et al., for example, found similar lead contamination profiles at the same location 15 years apart, with measurable but very small amounts of lead migrating at a rate of up to 1 cm/ year. 89 The Haplic Lixisol soil in Owino Uhuru has an approximate cation exchange capacity of 1.8 cmol(+) kg and a pH of 5.9. 75 Its clay, silt and sand content are roughly 18%, 27% and 55%, respectively. 75 Thus, while these soils are more amenable to migration than others, migration is unlikely to account for any significant difference in surface soil concentration over the 10-year period between the opening of the smelter and the execution of this study.
A second possible explanation for the discrepancy between past BLLs and the current environmental levels present is that the primary exposure pathway to the community while the smelter was operating was the inhalation of airborne lead. Air lead concentrations would have declined immediately following the closing of the smelter. Reducing airborne sources of lead exposure near smelters has been strongly associated with declines in BLLs. 76 Given that the key sources of exposure were most likely associated with the operation of the facility, the need for mitigation work in Owino Uhuru may not be as pressing as presented elsewhere. The former facility very probably contains high levels of lead on site, which should be appropriately considered in any future land use plans.

Informal housing
There

Setting public health priorities
A number of factors potentially influence the allocation of public health resources. These could include political forces, societal values and economic justifications, among others. 100-102 In response to the various challenges implicit in allocating resources in public health, there has been a general coalescence in recent decades around rational and transparent approaches. Foremost among these is the notion of evidencebased decision making, commonly articulated in burden of disease, cost effectiveness, or equity analysis approaches. 102 These approaches seek to make the best use of finite resources through "the conscientious, explicit, judicious and reasonable use of modern, best evidence in making decisions. " 103,104 Within global public health, it is arguable that the issue of environmental lead poisoning receives proportionately less attention relative to its impact than other public health risks. IHME, for instance, calculated that lead exposure attributed to 13.9 million disability-adjusted life years and 540,000 deaths globally 2016. 22 As context, this amounts to about 0.6 % of all disability-adjusted life years and about 1% of all deaths globally in the same year. 22 In addition, there are indications that this value may be an underestimation. 33 Major sources of environmental lead poisoning globally include industrial mining and smelting operations, lead-based ceramic glazes, and the recycling of used lead acid batteries. 105 Lead exposure also results in a range of adverse societal impacts, including increased rates of violence and decreased economic output, that are not captured in disease burden approaches. 15,106,107 Despite this considerable impact, there is currently no international convention or multi-lateral funding mechanism Research to support work related to lead contamination, resulting in a resourcepoor environment for potential implementers.
In the context of disproportionately limited funding for lead interventions, the importance of evidence-based decision making is augmented. Put differently, the burden of proof to justify interventions should be high. The residents of Owino Uhuru are subject to myriad environmental health and livelihood risks common to life in informal settlements. 108 However, this study has found that those related to lead exposure were likely mitigated with the closing of the facility. The lead poisoning event in Owino Uhuru continues to receive national and international media attention despite the absence of current information regarding the health impact. 56,109,110 In evaluating the relative importance of an intervention in Owino Uhuru versus similar projects elsewhere, policy makers should be encouraged to utilize evidence like that presented here to form the basis of their decisions.

Limitations
A key limitation in the present study is the heavy reliance on pXRF measurements. The instrument was calibrated, and accuracy was confirmed in accordance with the manufacturer's instructions. When conducted in sufficient quantity, pXRF measurements have been shown to closely approximate wet chemistry techniques. 111 However, no samples from this study were sent for laboratory analysis. Measurements of the NIST reference material during the assessment found that the pXRF was functioning within acceptable standards.
A similar limitation relates to the spatial distribution of the surface soil measurements taken in the present study. The assessment, which focused on the residential area of Owino Uhuru, did not determine concentrations in the industrial areas to the west, south, and east of the facility. Moreover, within Owino Uhuru, significant assessment gaps are evident in Figure 1, particularly to the northwest of the facility. While this area is unlikely to present elevated concentrations resulting from aerial deposition, it is possible that the area could be contaminated by other means such as manual deposition of waste. However, interviews with community members did not indicate that this was the case. Nevertheless, that this area was not assessed here represents a limitation.