Effectiveness of interventions for the remediation of lead-contaminated soil to prevent or reduce lead exposure - A systematic review

Objective To summarize the evidence on the effectiveness of soil remediation to prevent or reduce lead exposure. Methods We systematically searched MEDLINE, the Agricultural & Environmental Science Database, Web of Science, and Scopus from 1980 to February 15, 2021. We also performed reference list checking, hand-searched websites, and contacted experts. Eligible studies evaluated the effect of soil remediation to prevent or reduce lead exposure in humans of any age. We screened all records dually; one investigator performed the data extraction; a second checked for completeness and accuracy. Two investigators independently rated the risk of bias of included studies and graded the certainty of evidence. We synthesized findings narratively. Results We identified 6614 potentially relevant publications, all focused on children, of which five studies (six records) fulfilled our prespecified inclusion criteria. The number of evaluated participants ranged from 31 to 1425, with follow-up periods of 11 months to one year. The primary soil remediation method was the replacement of the upper layer with clean soil. Outcomes were limited to blood lead levels (BLL), dust lead levels, and soil lead levels. The largest study, a controlled before– after study (n = 1425) reported favorable effects of soil remediation compared to no intervention. This finding was consistent with results from two cross-sectional studies and one uncontrolled before–after study. One year post-remediation, the mean reduction in BLL was 2.1 μg/dL (p < 0.0001) greater in the intervention group than in the control group. Two randomized controlled trials with a total of 511 participants showed no statistically significant incremental effect of soil remediation when combined with paint and/or dust abatement. The certainty of evidence for all outcomes was low. Conclusion Soil remediation appears to reduce BLL in children when used as a single intervention. The incremental benefit of soil remediation when part of other interventions is limited.


Introduction
The World Health Organization (WHO) has designated lead as one of ten chemicals of major public health concern (World Health Organization, 2018). Lead can cause acute and chronic illnesses of various organ systems, affecting both children and adults. Chronic lead poisoning is more common than acute poisoning. In adults, it typically leads to memory and concentration problems, depression, abdominal and neuromuscular symptoms, fatigue, anemia, sleep disturbance, hypertension, and cardiovascular diseases . In children who are chronically exposed to lead, aggressive behaviour and apathy are the most common symptoms (Pearce, 2007;Patrick, 2006). No level of lead exposure has so far been identified that is without harmful effects (World Health Organization, 2018;Bellinger, 2008). Blood lead levels (BLL) lower than 5 μg/dL were associated with reduced school performances in children (Bellinger, 2008). In adults, BLL of 1 μg/dL were associated with an increased risk of cardiovascular diseases . During pregnancy, both current lead exposure and accrued lead in the mother's bones may harm the developing fetus, which may result in miscarriage, stillbirth, premature birth, and low birth weight (World Health Organization, 2018). The harmful effects of lead exposure, however, are preventable (World Health Organization, 2018). Therefore, the identification and control of lead hazards in residential environments are vital (Chisolm, 2001). Lead exposure can come from various sources (e.g. lead mines, smelters, refineries, recycling and manufacturing sites, leaded petrol and aviation fuel, lead-based paint, water piping, fixtures and solder). Contaminated soil is often an important source of lead exposure (Boreland and Lyle, 2014;von Lindern et al., 2003a;Tirima et al., 2016) for children because it also accumulates as indoor dust (World Health Organization, 2018;Laidlaw et al., 2017). Compared with adults, children have a higher exposure to lead-contaminated soil and indoor dust, as they place hands and objects in their mouths and are closer to the ground, for instance due to crawling and playing. In addition, the absorption and retention of lead is higher in children than in older individuals (World Health Organization, 2018;Bellinger, 2008;Lanphear et al., 1998). Lead can pass the developing blood-brain barrier, making children especially vulnerable because their nervous system is still developing (Mason et al., 2014).
The Center for Disease Control and Prevention (CDC) recommends taking various management actions for children with BLL greater or equal to 5 μg/dL [0.24 μmol/L] (Advisory Committee for Childhood Lead Poisonning Prevention, 2012). As reported in a narrative review from 2017, in some urban neighborhoods in the United States (US), up to 20-40% of children had elevated BLL (≥5 μg/dL), which were, at least partly, attributed to lead in soil (Laidlaw et al., 2017). According to the US Environmental Protection Agency, the acceptable safety standards for lead in bare soil are under 400 ppm (ppm) in children's play areas and under 1200 ppm for bare soil in non-play areas (Agency for Toxic Substances and Disease Registry, 2019).
The remediation of lead-contaminated soil aims to minimize or eliminate the hazard, and a range of soil remediation techniques exists. These approaches, some of which are still experimental, include mechanical, chemical, or biological interventions to clean, stabilize, remove, replace, and/or cover contaminated soil (Laidlaw et al., 2017).
To date, the effectiveness of soil remediation to reduce the negative health effects of lead exposure has not been assessed comprehensively and systematically. A narrative review including soil remediation studies with various techniques (excavation and replacement) to prevent or reduce soil contamination in urban environments reported a BLL decrease that ranged from 35% to 90% after six months to three years post-remediation (Laidlaw et al., 2017). A Cochrane review published in 2020 that assessed the effects of dust abatement in households with high lead exposure, included two studies on the remediation of contaminated soil with insufficient evidence on the effectiveness of soil remediation (Nussbaumer-Streit et al., 2020).
The aim of our review was to support WHO to develop guidelines and systematically assess the effects of soil remediation interventions on human health in rural and urban environments. Specifically, we strove to answer the following research question: -What are the possible effects and/or adverse effects of soil remediation, alone or in combination with other interventions, compared to no interventions or other interventions on BLL and the subsequent health outcomes in humans?

Methods
This systematic review was conducted in accordance with Cochrane systematic review methods (Higgins and Thomas, 2019) Throughout the manuscript, we followed the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA) statement (Moher et al., 2009).
We registered the protocol in PROSPERO (the International Prospective Register of Systematic Reviews), ID CRD42019136676.

Eligibility criteria
We were interested in any study designs that evaluated different soil interventions with the aim to prevent or reduce lead exposure. Table 1 presents the a priori-defined eligibility criteria.

Study selection and data extraction
Abstract and full-text review forms were developed and piloted on a sample of 50 abstracts and five full-text articles by all reviewers, working in pairs. Discrepancies were resolved by discussion or by involving a third reviewer. During the study selection process, the abstracts and selected full-text articles were independently reviewed by two investigators using Covidence (Covidence systematic review software, n.d.). After pilot-testing the data extraction forms, one investigator extracted Setting -Any rural or urban settings with lead-contaminated soil -Settings with and without data from the included studies; a second investigator checked for completeness and accuracy. We extracted relevant information related to the characteristics of the study populations, settings, interventions, comparators, study designs, methods, outcomes of interest, and results. We also abstracted data on soil and dust lead levels as well as soil remediation costs

Data synthesis
We summarized the results narratively and grouped them by outcomes of interest. We did not identify enough studies with a similar design to be able to conduct meta-analyses.

Risk of Bias assessment
Two reviewers independently assessed the risk of bias of the included studies. Disagreements were resolved by discussion and consensus or by consulting a third reviewer. We used the Cochrane Risk of Bias Tool 2.0 (RoB2) for assessing the quality of individually randomized controlled trials (RCTs) (Sterne et al., 2019) and a previous version of the tool that is compatible with cluster RCTs (Lyu et al., 2018). For nonrandomized studies that met the Effective Practice and Organization of Care (EPOC) criteria, we used the Cochrane EPOC risk of bias tool (Effective Practice and Organisation of Care Group (EPOC), 2017). For studies that did not meet the EPOC criteria (Cochrane Effective Practice and Organisation of Care (EPOC), 2017), we assessed the risk of bias using the Quality Assessment Tool for Quantitative Studies, developed by the Effective Public Health Practice Project (EPHPP) (Effective Public Health Practice Project, n.d.).
For the RoB2 and EPOC risk of bias criteria, we rated the risk of bias using the categories "low,""some concerns," and "high." For the EPHPP, the categories were "strong," "moderate," and "weak." Weak corresponded to a high risk of bias, moderate to some concerns, and strong to a low risk of bias.

Certainty of evidence
We dually rated the certainty of evidence for the outcomes ranked as critical or important by the WHO Guideline Development Group using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach (Guyatt et al., 2011). These ratings incorporate assessments of the risk of bias, inconsistency, indirectness, imprecision, and publication bias for each outcome. Depending on the certainty of evidence, the overall rating for each outcome resulted in one of four categories: high, moderate, low, or very low. Disagreements were resolved by discussion.

Role of the funding source
This review was funded by a contract with WHO. The WHO Guideline Development Committee assisted in the development of the key questions, study inclusion criteria, and outcome measures of interest but was not involved in the data collection, analysis, or manuscript preparation.

Literature search results
Our literature searches detected 6614 unique records after deduplication. We retrieved 72 as full-text publications. Five studies corresponding to six publications met our eligibility criteria (Aschengrau et al., 1994;Weitzman et al., 1993;Farrell et al., 1998;von Lindern et al., 2003b;Lanphear et al., 2003;Gagne, 1994). One study reported controlled short-term and uncontrolled long-term findings of one study arm (Aschengrau et al., 1994;Weitzman et al., 1993). Due to the different study designs, number of participants, and additional interventions during the long-term phase, we present the results of these two publications separately. Fig. 1 depicts the literature review flow. Appendix B lists the studies excluded at the full-text level and the reasons for exclusion.
In the following sections, we first summarize the characteristics of the included studies (Table 2). We then present the effect of soil remediation on the outcomes of interest (Table 3). All the included studies assessed BLL. The costs associated with soil remediation were reported in three studies (Weitzman et al., 1993;Farrell et al., 1998;Gagne, 1994). None of the included studies reported on health outcomes or adverse effects of the intervention.

Effect of soil remediation on blood Lead levels
All studies assessed the effect of soil remediation on BLL in children. Table 3 summarizes the main results. Appendix C presents the certainty of evidence ratings for each outcome.

Soil remediation versus no intervention
Three studies assessed the effects of soil remediation compared with no intervention (von Lindern et al., 2003b;Lanphear et al., 2003;Gagne, 1994). One controlled before-after study with a high risk of bias compared pre-remediation to one-year postremediation BLL in 1425 children under the age of 9 years living near an abandoned industrial lead-zinc mining and smelting complex in the Bunker Hill Superfund Site, Idaho, USA (von Lindern et al., 2003b). The investigators recruited participants over the course of 10 years and grouped them into 10 intervention and control groups. The soil remediation intervention consisted of replacing contaminated surface soil and dust with uncontaminated soil in yards having soil lead levels greater than 1000 ppm. A statistically significantly greater mean reduction in BLL of −2.1 μg/dL (95% CI NR) (p < 0.0001) was observed in the intervention group compared to the control one year post-remediation.
One repeated cross-sectional study with a control group, rated as having some methodological concerns, assessed the effect of the remediation of yards with average soil lead levels greater than 500 ppm in Midvale, Utah, USA (Lanphear et al., 2003). The intervention consisted in excavation and backfill with clean soil. Using an identical protocol, the authors evaluated two distinct cohorts of children between 6 months and 12 years at two time points 10 years apart (Lanphear et al., 2003). No baseline characteristics of the two groups (from 1989 and 1998) were provided. After adjustment for covariates (age, mouthing behaviour, year of study, socioeconomic status), the results indicated no significant reduction in BLL between children, aged 6 and 72 months, who lived in abated and non-abated houses. In contrast, a significant decline in BLL was reported for children from 6 to 36 months old who had not been exposed to lead contaminated yards in early childhood (absolute decline in blood lead concentration 2.5 μg/dL; p = 0.03).
An uncontrolled repeated cross-sectional study by Gagné et al. measured the BLL in 233 children from a residential area close to an active copper smelter in Quebec, Canada (Gagne, 1994). Before the soil remediation, the geometric mean BLLs of the selected group were 10 μg/dL (95th percentile: 20 μg/dL). The soil remediation (1990)(1991) consisted of removing the first 10 cm of soil and replacing it with uncontaminated soil and grass on top (or gravel in parking areas). By the time the last lot was remediated in 1991, the mean BLL decreased to 7.3 μg/dL (p-value not reported). The proportion of children with BLL greater than 10 μg/dL decreased from almost 100% in 1979 to 50% in 1989 and finally to 25% in 1991. No potential confounders were considered. It is not clear which portion can be attributed to soil remediation because of concomitant lead reduction programs (reduction of stark emissions and ban of lead from gasoline) (Gagne, 1994). We rated the certainty of evidence as low that soil remediation compared with no intervention reduces the BLL in children (Appendix C).

Soil remediation with co-interventions versus co-interventions alone
Two studies (three publications) compared soil remediation plus other interventions versus other interventions alone and provided findings about the incremental benefits of soil remediation when used in combination with other lead-reducing interventions (Aschengrau et al., 1994;Weitzman et al., 1993;Farrell et al., 1998).
A cluster RCT by Farrell et al., which evaluated 408 children over 6 months, randomized contaminated neighborhoods to exterior paint stabilization with or without the replacement of 15 cm of soil (Farrell et al., 1998). We rated the study as having a high risk of bias because of the high attrition (55%). The geometric mean BLL of participants who completed the study decreased from 12.1 μg/dL to 9.7 μg/dL due to soil remediation. After adjustment for covariates (time, seasonality, socioeconomic status, age, and mouthing behaviour), the authors reported no statistically significant difference between the intervention and the control group (covariate-adjusted effect estimate: −0.05 μg/ dL [standard error: 0.04]).
The second study, the Boston project, included children between 6 months and 5 years exposed to lead, and assessed the effect of soil remediation in contaminated areas. Phase one of the project was an RCT reported by Weitzman et al. (Weitzman et al., 1993) The interventions consisted of loose interior paint removal and interior dust abatement with and without the replacement of 15 cm of contaminated soil (Weitzman et al., 1993). We rated the study as having some methodological concerns. After 11 months, the mean BLL of the intervention group including soil remediation decreased by 1.53 μg/dL more than that of the comparison group (95% CI: −2.87, −0.19 μg/dL). When confounding variables were adjusted for (race and ethnicity, socioeconomic status, playing or sitting on the floor, and baseline BLL), the adjusted mean decrease in BLL was −0.80 μg/ dL (95% CI: −2.05 to 0.45).
Using an uncontrolled before-after study design, Aschengrau et al. reported the results of the second phase of the Boston study (Aschengrau et al., 1994). Participants who received paint stabilization and dust abatement or dust abatement alone during phase one received soil remediation in phase two. The study was rated as having a high risk of bias because of the study design and a high dropout rate. When the analysis was restricted for the season or to only one child from each family and adjusted for baseline BLL, the decrease in BLL after the soil remediation was −5.24 μg/dL and − 2.57 μg/dL, respectively (Aschengrau et al., 1994).
We rated the certainty of evidence as low that soil remediation when added to paint stabilization and dust abatement leads to a small incremental reduction in BLL in children (Appendix C).

Effect of soil remediation on health outcomes
No evidence on patient-relevant health outcomes was identified.

Soil lead levels
Two RCTs (Weitzman et al., 1993;Farrell et al., 1998), two beforeafter studies (one with a control group) (Aschengrau et al., 1994;von Lindern et al., 2003b), and one repeated cross-sectional study (Lanphear et al., 2003) assessed the effect of soil remediation on soil lead levels. All reported a statistically significant reduction after the intervention.
Two RCTs and one uncontrolled before-after study also assessed recontamination (Aschengrau et al., 1994;Weitzman et al., 1993;Farrell et al., 1998). One RCT reported no statistically significant increase in soil lead levels 9 months post-remediation (Weitzman et al., 1993); the other RCT mentioned "significant lead reaccumulation" (no data available) two years post-remediation (Farrell et al., 1998). Aschengrau et al. reported that 44% of the remediated yards from the first intervention group and 62% from the second intervention group showed evidence of recontamination at 6-10 months post remediation (Aschengrau et al., 1994).

Dust lead levels
Two studies (three publications) reported floor dust levels and showed a statistically significant decrease after soil remediation (Aschengrau et al., 1994;Weitzman et al., 1993;Lanphear et al., 2003). Weitzman et al. found that 4 to 5 weeks after phase one of the Boston project, the median floor dust lead concentrations in the intervention (soil remediation, loose interior paint removal, interior dust abatement) and control groups (loose interior paint removal, interior dust abatement) were reduced by 53% and 49%, respectively (Weitzman et al., 1993). In the second phase of the same project, the mean floor lead levels were unchanged at 6 to 12 months after the soil remediation intervention in both groups, compared with baseline levels (Aschengrau et al., 1994).
Lanphear et al. reported a statistically significant reduction in the floor dust lead level in the soil remediation group (geometric mean decrease: 409 μg/dL) compared with the control group (geometric mean decrease: 157 μg/dL), p < 0.001 (Lanphear et al., 2003).

Adverse events of soil remediation
None of the included studies reported on adverse effects of the soil remediation intervention.

Discussion
To our knowledge, this is the first systematic review on the effectiveness of soil remediation to prevent or reduce lead exposure and, subsequently, to avoid or mitigate negative health effects in humans of all ages.
Overall, the available evidence is limited in quality and quantity. Findings from four nonrandomized studies suggested favorable effects of soil remediation compared to no intervention regarding BLL in children (Aschengrau et al., 1994;von Lindern et al., 2003b;Lanphear et al., 2003;Gagne, 1994). In addition, they reported statistically significant decreases of soil-and dust-lead levels (Weitzman et al., 1993;Farrell et al., 1998) (Aschengrau et al., 1994;von Lindern et al., 2003b) (Lanphear et al., 2003). The significance of the estimated effect sizes from a clinical and public health perspective, however, is unclear. The incremental benefit of soil remediation when added to other interventions to reduce lead exposure in children was modest. Two RCTs reported no statistically significant incremental effects of soil remediation when combined with paint stabilization or dust abatement (Weitzman et al., 1993;Farrell et al., 1998). Due to limitations in the conduct and design of the studies, we rated the certainty of evidence as low which means that future studies might have a substantial impact on these effect estimates.
We also identified 10 studies, which did not meet our inclusion criteria because they assessed combinations of different interventions including soil remediation (Braun et al., 2018;Boreland et al., 2009;Calder et al., 1994;Chowdhury et al., 2021;Ericson et al., 2018;Goulet et al., 1996;Lalor et al., 2001;Langlois et al., 1996;Maynard et al., 2003;Schoof et al., 2016). Because the effect of the soil remediation could not be isolated, we excluded these studies from our analyses. Results, however, pointed to a potentially greater effect of the combined interventions compared with soil remediation alone.
The internal validity of the included studies is limited, as they all carry at least some concerns for bias, most having a high risk of bias, either inherent to their study design (Lanphear et al., 2003;Gagne, 1994) or due to factors such as low response or completion rates (Farrell et al., 1998) and unaccounted differences between groups at baseline (von Lindern et al., 2003b;Lanphear et al., 2003).
The included studies of our review have several strengths and limitations. Two studies were RCTs (Weitzman et al., 1993;Farrell et al., 1998) and two out of four non-randomized studies included a control group (von Lindern et al., 2003b;Lanphear et al., 2003). One study included more than 1400 participants (von Lindern et al., 2003b). Almost all of the studies adjusted the analyses for potential confounders (Aschengrau et al., 1994;Weitzman et al., 1993;Farrell et al., 1998;Lanphear et al., 2003). Two RCTs and one uncontrolled before-after study assessed soil recontamination (Aschengrau et al., 1994;Weitzman et al., 1993;Farrell et al., 1998).
A notable limitation are the many gaps in reporting, which may be due to the fact that the studies are from the 1990s and early 2000s, when fewer reporting standards existed.
The analysis methods were usually not prespecified and, hence, multiple testing cannot be ruled out. Selective outcome reporting cannot be ruled out, either, when no study protocol exists.
The generalizability of findings may be also compromised by several factors. Four out of five studies evaluated children under 9 years of age, so the results may not be applicable to other groups at particular risk, such as pregnant women as well as adolescents and the general adult population (Aschengrau et al., 1994;Weitzman et al., 1993;Farrell et al., 1998;von Lindern et al., 2003b;Gagne, 1994).
The included studies' BLL were only mildly elevated at baseline, between 5.6 μg/dL (Lanphear et al., 2003) and 15.3 μg/dL (von Lindern et al., 2003b). The effect of soil remediation in high-level exposure could not be assessed.
In three publications, the intervention and control groups received other interventions in addition to soil remediation (paint stabilization and/or dust abatement) (Aschengrau et al., 1994;Weitzman et al., 1993;Farrell et al., 1998); another publication mentioned concomitant lead reduction programs (reduction of stark emissions and ban of lead from gasoline) (Gagne, 1994). Hence, the results showed the added effect of soil remediation, which might not be equal to that of soil remediation alone. Albeit not ideal for a strictly separate effect estimation, this mirrors real-life scenarios. Usually, remediation programs address different sources of exposure (e.g. air, dust, chipping paint, drinking water, consumer products). They are designed to eliminate all possible sources of lead contamination and not primarily to assess and compare the effectiveness of individual interventions.
Finally, the follow-up periods of identical study populations in the included studies did not exceed one year (Weitzman et al., 1993;Farrell et al., 1998;von Lindern et al., 2003b;Gagne, 1994) and all studies were conducted in high income countries. Hence, the generalizability of findings to other settings, such as low-and middle-income countries and the intervention's long-term effects and sustainability cannot be established.
We found no studies that assessed the adverse effects or the effectiveness of soil remediation on patient-relevant health outcomes. The BLL is a surrogate outcome and therefore can only indirectly indicate the health status of an individual. It has limitations because lead stores in bones and can continue to contribute to BLL in the absence of further exposure; post-remediation BLL may still represent past exposures (Laidlaw et al., 2017;Gulson et al., 1996) BLL is hence better suited to measure recent or current exposures at low or moderate levels (Barbosa et al., 2005). Many experts consider BLL the primary biomarker for measuring lead exposure because it has a causal relation with health outcomes (Lanphear et al., 1998;Barbosa et al., 2005;Wang et al., 2015;Schober et al., 2006). New studies focusing on health outcomes are probably not realistic. Considering the complexity of health outcomes (e.g. cognitive and neurobehavioral outcomes, physical development in children, adverse pregnancy outcomes, cardiovascular, renal or fertility outcomes), those studies would need larger sample sizes and a longer follow-up for exposed persons during different stages of life (early childhood, childhood, adolescence, adulthood) to provide valuable results.
Our study has some methodological limitations. Although we applied a vigorous methodology, adhering to the standards provided in the Cochrane Handbook (Higgins and Thomas, 2019), we cannot rule out that we may have missed relevant studies or findings. We limited our eligibility criteria to studies published in English, French and German. A recent systematic review reported a negligible impact on the effect estimates and conclusions of language restrictions for most medical topics (Dobrescu et al., 2021). Nevertheless, we might missed studies published in languages other than English, French and German. While we searched for grey literature, reporting bias is a potential limitation of any systematic review. We were not able to conduct a metaanalysis due to differences in the study designs, BLL at baseline, contamination sources, follow-up times, or sampling methodology.

Conclusion
Soil remediation appears to reduce BLL in children when used as a single intervention. The incremental benefit of soil remediation, however, is limited, when it is part of other interventions that aim to reduce lead exposure. The quality and quantity of the evidence assessing soil remediation to reduce or prevent lead exposure is limited. No evidence is available to assess the effects of soil remediation on pregnant women or other vulnerable populations, in lowand middle-income countries, and on the health outcomes of exposed individuals. No conclusions about long-term effects are possible.

Funding
This systematic review was commissioned and funded by the World Health Organization.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.