Environmental and potential human health legacies of non-industrial sources of lead in a Canadian urban landscape – the case study of St John's, Newfoundland

Abstract

Residential soil and house dust were collected in St John's to assess the levels of lead exposure and potential human health risk. Although St John's does not have an identified, major point source for lead, nor is it a heavily industrialized or populated city, 51% of all analysed soil samples (n = 1231) exceeded the Canadian Council of Ministers of the Environment (CCME) residential soil lead guideline of 140 ppm, 26% exceeded the 400 ppm United States Environmental Protection Agency (US EPA) guideline for soil in children's play areas, and 9% exceeded the 1200 ppm US EPA guideline for soil outside of play areas. High soil lead concentrations, particularly those above 1200 ppm, are clustered in the older downtown core. Samples located along exterior house walls (dripline) have the highest mean soil lead concentrations, followed by open spaces in yards (ambient) and then roadside sites. Lead in dripline soil samples from older housing stock is sourced to lead-bearing paint. Lead from both dripline and ambient samples on properties developed between the 1940s and 1980s have a mixture of sources including coal ash, paint, and leaded gasoline. Approximately 12% of analysed house dust samples (n = 96) exceeded the US EPA guidelines for lead in indoor dust, all from pre-1950s housing and all associated with dripline soil lead concentrations greater than 900 ppm. Human health risk predictions suggest that, although the wider St John's community may not be at risk of adverse health effects, children living in pre-1970s housing may be exposed to increased risk.

Keywords:

• lead contamination
• urban environment
• medical geology
• St John's
• Newfoundland

Introduction

‘With 100 years of coal combustion, 40 years of leaded gasoline use, and continued use of bunker crude by large facilities, the city's soils may have become a large reservoir of potentially toxic metals’. Christopher et al. (1993) reached this conclusion on the basis of the sediment chemistry record of four urban lakes in St John's, Newfoundland and Labrador, Canada (Figure 1). Chronological sequencing of lake sediment cores showed steadily increasing loadings of anthropogenic lead throughout the late nineteenth century and early to mid-twentieth century, followed by a particularly intense period of lead contamination between about 1940 and 1970. An exploratory survey of surface soil lead levels in St John's revealed that 60% of all samples and 89% of samples from residential properties exceeded Canadian recommended guidelines (Bell 2003). Public concern over the high levels of lead in soil and the potential health risk for children (e.g. Dyer 2003) prompted a broader and more systematic survey of environmental lead exposure in St John's.

Figure 1. Location map for St John's in eastern Newfoundland, Atlantic Canada. The oldest parts of the city are located in the downtown area, on the north side of St John's Harbour.

This article has two goals: first, to provide a review of recent studies and public health impacts of soil and indoor dust lead exposure in Canadian cities, and second, to report the results of the expanded St John's soil and dust lead study in the broader context of the potential human health risk. Soil and dust lead exposure has been a recognized environmental heath concern for several decades in major cities in the USA and Europe and in communities where lead smelters and other point-source emissions have existed. In contrast, St John's does not have an identified, major point source for lead, nor is it a heavily industrialized or populated city; hence understanding the sources and pathways for elevated lead levels has broader implications for similar communities across Canada and raises potential concern for heath effects on local residents.

Lead and human health

Health effects

The health effects of lead exposure on humans depend on the amount of lead taken into the body. In general, lead affects the developing nervous system, causing mental and behavioural problems, the haematological and cardiovascular systems, and the kidneys (ATSDR 2005). The neuro-developmental effects of lead, including decreased IQ scores and other intelligence and developmental deficits, have been recently studied because of their occurrence at low blood lead concentrations.

Children are the most sensitive to the neurological impacts of lead. Children have a higher gastrointestinal absorption rate (Ziegler et al. 1978) and, because their nervous system is not fully developed, it is easier for lead to cross the blood–brain barrier (Johanson 1980; Adinolfi 1985). Children also display a high rate of hand-to-mouth behaviour, which increases the amount of potentially contaminated soil and dust consumed (Moya et al. 2004).

Meta-analysis of comparable studies has revealed significant relationships between IQ score and blood lead concentration (Needleman and Gastonis 1990). In general, an increase of 10 μg/dL in blood lead concentration has been shown to cause a decrease in IQ of between 1 and 5 points. Some longitudinal studies that followed children through time found a significant correlation between high prenatal and/or postnatal blood lead concentration and poor performance on mental development tests, both during the toddler years and into older childhood (Boston, Bellinger et al. 1985; Port Pirie, Australia, Vimpani et al. 1985; Cincinnati, Dietrich et al. 1987). Other studies, however, found no association between blood lead and performance on development tests within a blood lead range of 4–32 μg/dL (Lansdown et al. 1986; Harvey et al. 1989). Because a variety of other factors may play a role in mental development (e.g. birth weight, socio-economic status), it can be difficult to isolate and control confounding variables in studies of lead effects on mental health (MOEE 1994; US EPA 1998).

Guidelines

The United States Centers for Disease Control and Prevention (US CDC) set the screening guideline for blood lead at 10 μg/dL. This level is a tool for risk management and most studies conclude that there is no threshold in blood level concentration at which health effects first occur (US CDC 1991; Lanphear et al. 2000; Bellinger 2004). In 1994 the Federal Provincial Committee on Environmental and Occupational Health Lead Working Group (FPLWG) in Canada recommended that individual intervention should occur when a child's blood lead concentration exceeds 10 μg/dL and that community-level intervention be undertaken when blood lead concentrations from a sample of children exceed the arithmetic mean from the general population by three standard deviations or when the percentage of children with values above 10 μg/dL is double that seen in the general population (FPLWG 1994).

Canadian blood levels

Blood lead concentrations in children have been studied in Canada, particularly for those areas impacted by an industrial source of lead pollution. In general, blood lead concentrations have steadily declined since the late 1970s as sources of lead in the environment started being phased out. Recent studies indicate that Canadian children have relatively low blood lead concentrations, between 1.9 and 6.7 μg/dL (Table 1). For example, in residential areas near the coke oven site in Sydney, Nova Scotia (NS), the geometric mean (GM) for blood lead in children aged between 1 and 5 years was 1.86 μg/dL, with the maximum observation just below 9 μg/dL (NSDH and CBDHA 2001). In Port Colborne, Ontario (ON), where there is a nickel refinery, the GM for blood lead in children under 7 years of age was 2.3 μg/dL and no children exceeded the 10 μg/dL guideline (Decou et al. 2001).

Table 1. Details of blood lead studies conducted in Canada

Year	Location	Point source of pollution	Age range	Blood lead level (μg/dL)	Above 10 μg/dL	Reference
1978	Canada Health Survey	N			25%	Statistics Canada 1982
1979	Rouyn-Noranda, QC	Y		Mean 21.4		Goulet et al. 1996
1984	Ontario Blood Lead Study	N		Mean 12.0		O'Heany et al. 1988
1984	South Riverdale, ON	Y	<6 years	GM 12		Langlois et al. 1996
1987	Alberta	N	0–16 years	Mean 5.8–6	5	FPLWG 1994
1989	Quebec City, QC	N	1–6 years	Mean 5.6	10.8%	Levallois et al. 1991
1989	Vancouver, BC	N	2–3 years	GM 5.9	5%	Jin et al. 1995
1989	Rouyn-Noranda, QC	Y		Mean 11.1		Letourneau and Gagnon 1989
1989	St-Jean-sur-Richelieu, QC	Y	6 months–10 years	GM 9.2		Goulet et al. 1996
1990	Murdochville, QC	Y	6 months–5 years	Mean 5.9		FPLWG 1994
1990	Murdochville, QC	Y	5–12 years	Mean 6.7		FPLWG 1994
1991	Trail, BC	Y	<6 years	13.5	83%	Hilts et al. 2001
1991	St-Jean-sur-Richelieu, QC	Y	6 months–10 years	GM 5.0		Goulet et al. 1996
1992	South Riverdale, ON	Y	<6 years	GM 3		Langlois et al. 1996
2000	Trail, BC	Y	<6 years	6.7	27%	Hilts et al. 2001
2001	Sydney, NS	Y	1–5 years	GM 1.9	0%	NSDH and CBDHA 2001
2001	Port Colborne, ON	Y	<7 years	GM 2.3	0%	Decou et al. 2001

In studies that included lead abatement measures, blood lead concentrations have declined dramatically in recent years. A 10-year-long soil and dust abatement study in South Riverdale (ON) showed a decrease from 12 to 3 μg/dL in the GM blood lead of children less than 6 years of age between 1984 and 1992 (Langlois et al. 1996). In Trail, British Columbia (BC), the site of a lead-smelter, the GM blood lead concentrations for the same age range of children declined from 13.5 to 6.7 μg/dL between 1991 and 2000 following implementation of community education and remediation programmes (Hilts et al. 2001).

Sources of lead in the environment

Inhaled emissions from leaded gasoline combustion and industrial activities were the primary sources of lead exposure for Canadians (Health Canada 2004a). Since the phasing out of leaded gasoline and increasing restrictions on industrial emissions, airborne lead has become less of a concern while soil lead and household dust have drawn increased attention as exposure pathways in urban environments (US EPA 1989; Mielke et al. 1999). Other pathways may occur through consumption of home-grown produce or the direct ingestion of leaded paint, for example. As the scope of this article and the focus of the St John's study are limited to exposure from soil and indoor dust lead, the reader is directed to reviews of other lead sources and pathways. For example, Health Canada provides a brief overview (Health Canada 2004a) and the Ontario government has an older document that goes into more detail (MOEE 1994).

Outdoor soil

Naturally occurring background concentrations of lead have an average concentration of 16 parts per million (ppm) in crustal rock, whereas surface soils range from 20 to 100 ppm, reflecting pervasive low-level contamination at all but the most remote sites (McKeague and Wolynetz 1980; Davies 1990). Surface soil can be contaminated with lead from several sources, including the deposition of airborne combustion materials from point-source emitters and automobiles as well as from the addition of weathered lead-based paint. Lead in soil is a particular health concern because it does not biodegrade and is not rapidly absorbed by plants, so it remains in the soil at elevated levels or is removed to local drainage systems through soil disturbance and erosion.

The Canadian Council of Ministers of the Environment (CCME) recommended a soil lead guideline of 140 ppm for residential properties (CCME 1999). This is a conservative guideline designed to limit lead exposure for children as much as possible. On the other hand, the United States Environmental Protection Agency (US EPA) used a more moderate cost–benefit process to determine its soil lead guidelines (US EPA 2001a). Several models were used to generate soil and dust lead levels at which less than 5% of children would have an estimated blood lead concentration greater than 10 μg/dL. The health benefits of having reduced blood lead concentrations were then compared to the costs of lowering lead levels in all affected US homes. The greatest benefit for the most reasonable cost was chosen. The US EPA soil lead guidelines are 400 ppm for bare soil in children's play areas and an average of 1200 ppm across a sodded area where children have regular access, such as a residential yard or a school playground. The US EPA states that the guidelines are just a beginning and that, if lead levels are lower than the guidelines, it might still be beneficial to reduce lead levels even further if children under 6 spend much time in the area, or if the soil or dust contributes to lead in play areas or dwellings (US EPA 2001a).

Direct comparison of soil lead concentrations in cities across Canada and the USA is compromised by variations in sampling strategy (single or composite samples), location (open areas or near buildings), sample digestion (partial or full), and reporting metric (mean or median). Nevertheless, the median soil lead concentration (203 ppm) reported in the pilot study of St John's was higher than most other community surveys in Canada (13–167 ppm), with the exception of Sydney (340 ppm) and Trail (756 ppm), which have point sources of lead pollution (Table 2). Compared to much larger cities in the USA, the median soil lead value for St John's is similar to inner-city open areas of New Orleans (212 ppm), but lower than Chicago (1773 ppm).

Table 2. Comparisons of soil lead concentrations for Canadian cities. The data for St John's are from this study

City	Population	Metric	Lead (ppm)	Sample type
St John's, NLPilot studyThis study	95,000	MedianGM Median	203162148	Residential soil collected from open spaces, along foundations, and by roadsides
Belledune, NB^a	1711	Median	43–136	Garden soil
Sydney, NS^b	24,115	Median	340	Residential soil collected away from buildings and roads near the coke ovens
Victoria, BC^c	75,000	Median	90	Boulevards, parks, schoolyards
Trail, BC^d	7575	GM	756	Residential soil collected from two to three areas of exposed soil where children play
Port Colborne, ON^e	18,600	Median	167	Residential topsoil collected at least 1 m away from driveways, walkways, buildings, and fences
Sudbury, ON^f	157,857	Mean	30	Various locations downwind from three Ni- and Cu-smelters
Ottawa, ON^g	774,000	GM	42	Residential soil collected from five locations in yard
Iqaluit, NU^h	4220	Median	13	Commercial and residential sites sampled at grid intersections and also targeted samples from playgrounds, roads, and culverts
^aGovernment of New Brunswick 2005; ^bLambert and Lane 2004; ^cBowman and Bobrowsky 2003; ^dHilts 2003; ^eMOE 2002; ^fAdamo et al. 2002; ^gRasmussen et al. 2001; ^hPeramaki and Decker 2000

Studies have shown that soil lead levels of 1000 ppm may contribute 2–7 μg/dL to overall blood lead findings (Lanphear et al. 2000). The relative impact of soil lead exposure on blood lead levels, however, is controversial and elevated soil lead levels do not necessarily correspond with elevated blood lead levels. In some cases results are scale dependent, showing a relationship for community averages, but not for individuals (Mielke et al. 1999; Johnson and Bretsch 2002). Thus, lowering of elevated soil lead levels does not necessarily correspond with reductions in blood lead levels. In the Lead-in-Soil Demonstration Project carried out in Boston, USA, a large decline in soil lead (from 2075 to 105 ppm) was associated with only a 12% decline in children's mean blood lead (Weitzman et al. 1993).

Canadian studies that examined both soil lead and blood lead found a range of associations. In both Sydney, NS, and Port Colborne, ON, GM blood lead levels for children did not exceed 10 μg/dL, although median soil lead concentrations were 340 and 167 ppm, respectively (Decou et al. 2001; NSDH and CBDHA 2001). In contrast, soil samples from around the lead- and zinc-smelter in Trail, BC, had a median lead concentration of 756 ppm and 27% of children tested exceeded 10 μg/dL (Hilts et al. 2001).

Indoor house dust

Sources of lead in interior house dust include contaminated soil and the weathering of leaded interior paint. Dust lead levels from houses in Sydney, NS, indicated that lead loading was highest in doorways, on some premises an order of magnitude higher than elsewhere in the house, suggesting an exterior soil lead source (Lambert and Lane 2004). Experimental studies estimated that 20–30% of lead measured in indoor dust may originate in outdoor contaminated soil (Rutz et al. 1997). Conversely, residential surveys in Rochester, New York, and Ottawa, ON, suggested that lead sources other than soil – likely lead-based paint – contributed more to indoor dust lead levels (Lanphear and Roghmann 1997; Rasmussen et al. 2001).

Ingestion of house dust can be a major contributor (50% or more) to blood lead levels in children (Thornton et al. 1994; Lanphear et al. 1996; Lanphear and Roghmann 1997; Lanphear et al. 1998; Yiin et al. 2000). House dust intake rivalled only food intake as the primary component of the total lead exposure for children in Ottawa, ON, accounting for 30–70%, depending on the concentration statistic used (Rasmussen et al. 2001). In a pooled epidemiological study, Lanphear et al. (1998) determined that floor dust loading (amount of lead per area wiped for dust) was the most significant environmental predictor of blood lead, and that at floor dust loadings above 13.5–27 μg/m² (5–10 μg/ft²) the occurrence of elevated blood levels increased dramatically. Indoor dust lead may present a more significant health concern because the relatively high organic component of dust effectively accumulates metals (Rasmussen et al. 2001) and, thus, lead in dust may be more bioavailable than other media (Rasmussen 2004). Preschoolers spend the majority of their time indoors, making the impact of lead in house dust potentially greater than that in outdoor soil (Yiin et al. 2000).

US EPA dust lead loading standards are 107.6 μg/m² (40 μg/ft²) for floors, 672.8 μg/m² (250 μg/ft²) for window sills, and 1076 μg/m² (400 μg/ft²) for window troughs (US EPA 2001a). Neither Health Canada nor Environment Canada has any guidelines for lead levels in house dust in Canada (Health Canada 2004a).

Historical sources of environmental lead in St John's

St John's is a historic port city on the east coast of Canada that is over 500 years old (Figure 2a). The city itself remains small, with only 90,000 residents, although the development of satellite communities brings the population in the region to slightly above 180,000 (2006 census). Historically, small-scale commercial and industrial enterprises served the local community, but there is no history of major industrialization or a long-standing point source of lead pollution. The urban soils of St John's may have elevated lead levels for several reasons. The cycle of demolition and construction of buildings built with painted clapboard in the high-density downtown core of the city may have added large amounts of lead to the soil through the deposition of paint chips and weathered paint by-products. Additionally, several devastating fires in the 1800s and early 1900s would have generated ash and other lead-laden combustion products that were deposited in local soils (Figure 2b). The combustion of coal from the early nineteenth century to the late twentieth century on incoming ships as well as in both industrial and residential settings would have contributed large amounts of particulate matter and pollution including lead into the air and directly into the soil as stove ashes were commonly disposed of in back gardens. One of the most substantial contributions of lead into the environment would have been through emissions from motor vehicles using leaded gasoline, which first appeared in St John's in 1903 and lasted until the 1970s (Poole 1994).

Figure 2. (a) View looking west across downtown St John's c. 1920s. The harbour is visible on the far left. The downtown was (and remains) predominantly residential, comprised of clapboarded, wooden-framed row houses with small rear gardens. The paint regularly applied to clapboard contained lead. Coal was a common heating fuel for much of the 1800s through to the 1950s and contained high lead and arsenic levels. (b) View looking west across downtown St John's soon after the fire of 1892. Photographer's position is slightly to the left (south) of the one in Figure 2(a). Only stone and brick structures survived the fire. Note the two towers of The Basilica Cathedral of St John the Baptist in the right background. They are also visible on the far right in Figure 2(a). Source: The Geography Collection, Archives and Manuscripts Division, Memorial University of Newfoundland, St John's.

In addition to these anthropogenic sources of lead, there is also a natural source from underlying bedrock. Much of the city, including the downtown core, overlies the St John's Group, which is composed of grey and black shale and sandstone. Measurements of sediment geochemistry in lakes overlying this group provided a range of lead concentrations from 15 to 139 ppm, whereas similar measurement from adjacent sandstone- and siltstone-dominated bedrock groups gave values of <9 to 20 ppm (Conception and Signal Hill groups; Christopher et al. 1993).

Sampling strategy

Surface soil samples were collected from residential properties in 2004 and 2005. Neighbourhood areas as defined by the Community Accounts information system of the Newfoundland and Labrador Statistics Agency (www.communityaccounts.ca) were chosen as the spatial sampling units. Each neighbourhood roughly represents a population of 1000 and is defined by discrete postal codes. Neighbourhood areas are smaller and population density higher in the downtown core and to the north and southwest, reflecting the growth pattern of the city. Basic demographic data, including population, age structure, gender, dwelling type, and age, are available for each of the 95 neighbourhoods in St John's.

A minimum of three houses were sampled in each neighbourhood. An opportunistic sampling method was used with some effort made to spatially distribute the sample locations throughout the neighbourhood. Ultimately the choice of property largely depended on who was home at the time of sampling, whether or not they owned the property, and if they agreed to participate in the study.

Multiple soil samples were taken from each property, including, where possible, one sample from within 5 m of an adjacent road (roadside sample), within 1 m of the foundation of the house (dripline sample), and in an open area of the property away from buildings and roads (ambient sample). Soil lead concentrations from these discrete sample locations represent the relative contribution from vehicular emissions, leaded exterior paint, and ambient atmospheric deposition, respectively (Campbell 2008). This spatial sampling design follows the approach of Mielke (1994) in New Orleans and US EPA guidelines (US EPA 2000). However, sample composites were not created for individual properties (e.g. US EPA 1995; Hilts et al. 2001; Rasmussen et al. 2001; MOE 2002; Lambert and Lane 2004; Government of New Brunswick 2005); rather, each sample was analysed separately. Surface soil samples were also collected along urban–rural transects and at various depths below the surface in both urban and rural settings to establish surface soil lead levels away from urban sources and naturally occurring background levels of soil lead, respectively.

Indoor dust lead samples were collected from a small sub-sample of properties originally tested for soil lead. Sampling was conducted during a 2-week period at the end of September 2005 when indoor dust lead levels from tracking-in of soil or blowing-in of outdoor dust were likely to be at a maximum. Three samples were collected in each house: one from the most frequently used entrance floor, one from the kitchen floor, and one from the window sill of a frequently opened window. Indoor dust was collected using Ghost Wipes purchased from the Maine State Health and Environmental Testing Laboratory as per US EPA protocol (US EPA 1995). The main limitation of using the wet wipe method is that only the dust lead loading, not the actual concentration of lead in the dust, can be determined. Nevertheless, some studies have indicated that dust lead loadings are a better predictor of childhood blood lead levels than dust lead concentrations, especially on non-carpeted surfaces (Lanphear et al. 1995; Yiin et al. 2000).

Participants were asked to refrain from cleaning the selected surfaces for a few days prior to sampling to ensure there would be sufficient dust to collect. For comparative purposes, the number of days since a sampling surface had been cleaned was recorded, together with a visual inspection of the degree of dustiness. The type of sill surface and its general condition was noted to distinguish between a plastic window sill and one that was painted and peeling. Some homeowners neglected or forgot the cleaning instructions, especially the kitchen floor, and this may have resulted in under-representation of lead loading on these surfaces.

Sample collection and analysis

A 250 ml soil sample was collected from either the surface of bare ground or the uppermost 5–10 cm, where a vegetation mat was present for a total of 1231 samples. Duplicate soil samples were collected at about every 10th site. Samples were air-dried at 40°C, dry-sieved to <63 μm, and then digested using a multi-acid method following standard laboratory protocols for sediment analysis (Finch 1998), producing a complete sample digestion, and yielding total lead concentration.

Samples were analysed for a wide array of elements, including lead, using inductively coupled plasma emission spectrometry (ICP-ES) at the Geological Survey of Newfoundland and Labrador's Geochemical Laboratory. Canadian-certified reference materials were also analysed for comparison. Every 20th sample was split in the laboratory and run as a duplicate to assess analytical precision.

A 0.25 m² square plastic template was used to guide floor dust collection, whereas for window sills the total area of the sill was measured and wiped. Sample duplicates and blank wipes were collected after every 10th sample. Indoor dust wipes were digested according to the US EPA ‘Modified SW-846 Method 3050A Acid Digestion Procedure for Single-Wipe Samples’ (US EPA 1996). Digested samples were analysed for the full suite of trace elements by inductively coupled plasma mass spectrometry (ICP-MS) in the Earth Science geochemistry laboratory at Memorial University. The lead concentrations reported by the laboratory reflected the mass of lead in the digested sample that was then transformed into dust lead loadings (μg/ft²) for comparison with US EPA guidelines for indoor dust levels (US EPA 2001a, 2001b).

Results

Analysis of laboratory and field duplicate soil samples and laboratory standards indicated that laboratory methods were accurate and precise. Both the laboratory (n = 75, r = 0.999, p = 0.000) and field (n = 104, r = 0.934, p = 0.000) duplicates were highly correlated (Campbell 2008). Because of the skewed nature of the data, logarithmic transformations were used to improve data distribution; however, the Anderson–Darling test indicated that normality was still not achieved (p < .05). Nevertheless, GMs and their associated confidence intervals (CIs) are used instead of medians because they better represent the data.

Soil lead

Background soil lead concentrations

Twenty-nine sub-surface soil samples were collected from 19 sites in new suburbs of St John's and adjacent rural areas. It is assumed that these samples, being isolated from anthropogenic sources of lead, would represent background values of lead in soil. Material typically consisted of undisturbed glacial diamicton (till) and locations ranged from city construction sites to hand-dug pits between 5 and 10 km away from the city centre along arterial roadways. Sampling depths ranged between 16 and 130 cm with an average of 45 cm. The GM lead concentration of the 29 sub-surface soil samples was 17 ppm (Table 3). In contrast, the GM concentration for 18 surface soil samples, also from rural and suburban settings, was 37 ppm, more than double the sub-surface concentrations (Table 3). At rural locations, where continuous soil profiles were sampled, there were either no differences in surface and sub-surface lead values or samples above 20 cm gave values between 5 and 10 times higher. The latter locations were typically closer to the city and situated near major roads.

Table 3. Summary statistics on lead concentrations (ppm) in soil samples from surface and sub-surface levels in rural and suburban (non-urban) settings around St John's. Sub-surface samples are further divided by the geological group underlying the sample location

	Non-urban samples		Sub-surface samples overlying geological groups
	Surface	Sub-surface	St John's (urban)	St John's (non-urban)	Conception	Signal Hill
n	18	29	5	5	17	7
Median	36	17	37	22	16	21
GM	37	17	37	23	15	22
Range	9–171	8–45	18–73	16–30	15–32	12–45

Sub surface soil lead values were also classified by underlying rock type to explore the influence of bedrock geology on soil lead concentrations. The GM soil lead concentration for urban sub-surface samples overlying the St John's Group was 37 ppm, almost twice that of all other groups, including those samples overlying the St John's Group in rural/suburban settings (Table 3). Samples overlying the Conception Group, on the eastern side of the city, have the lowest associated soil lead values of the three bedrock groups (Table 3).

City-wide soil lead concentrations

In total, 1231 surface soil samples were collected and analysed, 514 from ambient locations, 328 from dripline locations, and 389 from along roadways (Table 4). The lead concentrations are highly skewed with 50% of the samples below 147 ppm and 10% above 1000 ppm (Figure 3). Dripline samples have the highest GM soil lead concentration (219 ppm), followed by ambient and road samples (Table 4). Both dripline and ambient samples have a wide range of soil lead concentrations (>12,000 ppm), whereas road samples have concentrations more or less below 1200 ppm. The majority of highly elevated soil samples came from dripline and to a lesser extent ambient locations. Fifty-one per cent of all soil samples exceeded the CCME residential soil lead guideline of 140 ppm, 26% exceeded the 400 ppm US EPA guideline for soil in children's play areas, and 9% exceeded the 1200 ppm US EPA guideline for soil outside play areas (Figure 3).

Table 4. Descriptive statistics for soil lead concentrations by sample category for St John's

Sample category	n	Mean	SE	GM	Min.	Percentile			Max.
						25th	50th	75th
Ambient	514	411	40	154	9	50	138	424	12,738
Dripline	328	766	112	219	15	57	194	831	24,477
Road	389	222	12	136	16	57	136	306	1765
All	1231	446	35	162	9	55	148	415	24,477
All values in ppm. SE is standard error.

Figure 3. The percentage distribution of soil samples by sample type in each of four soil lead concentration categories defined by Canadian and US recommended guidelines: CCME residential guideline (140 ppm), US EPA guideline for bare soil in play areas (400 ppm), and US EPA bare soil outside play areas (1200 ppm).

Spatial distribution

The spatial distribution of soil lead concentrations across the city of St John's shows a clustering of high lead values in the downtown core (Figure 4a). This is especially true for lead concentrations above 1200 ppm. In contrast, samples with soil lead concentrations below 140 ppm are found throughout the city. Almost every dripline sample collected downtown is above the CCME soil lead guideline of 140 ppm, and high lead concentrations outside the downtown core also are typically from dripline locations (Figure 4b).

Figure 4. Spatial distribution of soil lead concentrations by sample location for (a) all samples and (b) dripline samples only. Mapped categories are defined by Canadian and US recommended soil lead concentrations. The high density of soil samples in the downtown area causes overlapping of category symbols and obscuring of samples with higher lead concentrations in (a). These high-soil-lead locations are predominantly associated with dripline samples and are more clearly visible in (b).

Relationship to property age

In total, 939 soil samples were collected on 311 properties whose age could be independently verified. The data show a decrease in GM soil lead concentration for all three sample types as property age declines (Figure 5). Ambient and dripline samples, however, contain markedly higher GM lead concentrations for houses built before 1926 compared to those built between 20 and 35 years later. Road samples show a more modest decrease in soil lead concentrations with property age and are consistently lower than the other sample types until the late 1970s, since when all three display relatively low mean concentrations.

Figure 5. The distribution of GM soil lead concentrations by sample type for each of six property age categories for St John's.

Almost all samples collected from properties built before 1926 exceed the CCME residential soil guideline of 140 ppm and the proportion remains high for properties older than the 1960s but is less than 10% for those dated to the late 1970s (Figure 6). The proportion of samples exceeding the US EPA 400 ppm guideline is above 80% for ambient and dripline samples from houses built before 1926, and less than 10% in houses built after 1961. The only significant proportion of samples exceeding the US EPA 1200 ppm guideline occurs on properties built before 1926 and consists of both dripline (∼50%) and ambient (45%) samples.

Figure 6. The percentage distribution of soil samples by sample type in each of four soil lead c-oncentration categories defined by Canadian and US recommended guidelines grouped by property age.

Indoor dust lead

Ninety-six dust samples were collected from 32 homes in St John's. Seventy-five per cent of the samples have dust lead loading values below 61.9 μg/m² (23 μg/ft²), with individual loadings as high as 8528 μg/m² (3169 μg/ft²; Table 5). The GM dust lead loading is 21.5 μg/m² (8 μg/ft²; 95% CI; 3–10 μg/ft²), but there is a broad range of values depending on the sample location within the home (Table 5). Window sills have a GM dust lead loading three times higher than entrance floors and six times higher than kitchen floors. Window sills, and to a lesser degree entrance floors, have several very high lead loadings, whereas kitchen floors are consistently low.

Table 5. Descriptive statistics for indoor lead loadings by dust sample location. All values in μg/ft² for comparison with US EPA guidelines for indoor dust levels

Sample Type	n	GM	Min.	Percentile			Max.
				25th	50th	75th
Entrance	32	6.7	0.1	2.9	4.7	13.5	3169.4
Kitchen	32	3.3	0.2	0.9	2.7	9.4	79.5
Window	32	24.3	0.3	6.7	15.5	82.5	2938.2
All	96	8.1	0.1	2.3	7.0	23.1	3169.4

Eleven samples from seven different houses exceeded the US EPA guidelines for lead in indoor dust. They represent 16% of window sill samples, 13% of entrance floor samples, and 6% of kitchen floor samples. As expected, those surfaces that were reportedly cleaned most recently prior to sampling produced lower dust lead loadings. For instance, kitchen floors were cleaned on average 3 days before sampling, whereas entrance floors were cleaned on average 12 days before and window sills 34 days before.

Relationship with housing age

GM dust lead loadings for all household samples decrease with declining age for houses pre-dating the late 1940s; there is little temporal change in loadings for houses built after the 1940s (Figure 7). The highest dust lead loadings are recorded mostly in houses built before 1926, especially for window sill samples. In general, window sill loadings are consistently higher than floors in sampled houses built between the 1920s and 1990s. Houses built since the 1990s contained low loading values irrespective of the sampling location (Figure 7). Of the 11 dust samples that exceeded the US EPA dust lead loading guidelines, nine are from houses built before 1926 and two are from houses built between 1926 and 1948.

Figure 7. The distribution of GM dust lead loadings by sample location for each of six property age categories for St John's.

Relationship between dust lead and soil lead

A strong positive linear relationship (r ² = .677, p = .000) exists between indoor dust lead loadings and dripline soil lead concentration for sampled houses in St John's (Figure 8). More specifically, the relationship with dripline soil lead values is strongest for entrance floor samples (r ² = .955, p = .000), less so for window sill samples (r ² = .747, p = .000), and least for kitchen floor samples (r ² = .597, p = .000).

Figure 8. Log-log plot of GM soil lead concentration and dust lead loading by sample location on sampled properties in St John's.

All 11 samples that exceeded the US EPA guideline for dust lead loading were associated with dripline soil lead concentrations greater than 900 ppm; however, not all properties with high dripline soil lead concentrations were associated with high dust lead loadings. Eight of the 11 dust samples are associated with GM soil lead concentrations greater than 900 ppm for the entire property; the other three are from properties with high dripline but very low ambient and/or road soil concentrations.

Discussion

Soil lead

Background levels

Naturally occurring levels of lead in the greater St John's region appear to vary according to local bedrock type. Geochemical analysis of undisturbed till overlying similar geological groups to St John's on the Bay de Verde Peninsula 50 km to the west revealed that lead concentrations were noticeably higher in the St John's Group, with many samples ranging from 1 to 89 ppm, a few over 100 ppm, and three samples over 200 ppm (Batterson and Taylor 2003). Lead concentrations for till overlying the Signal Hill and Conception groups were lower; most samples were less than 21 ppm, with only an occasional sample between 21 and 89 ppm (Batterson and Taylor 2003). Data for sub-surface samples overlying the same three geological groups in St John's fall within these ranges and therefore a bedrock source likely explains the spatial pattern in background soil samples in the city and specifically the higher lead concentrations found in undisturbed sub-surface samples overlying the St John's Group.

In the absence of local anthropogenic sources of lead, sub-surface and surface soil samples outside the city of St John's have similar concentrations, although surface soil values may be slightly elevated – on average by 20 ppm – probably due to long-range transport and deposition of airborne lead from mainland Canada, as noted in rural lake sediments in eastern and south-central Newfoundland (Davenport et al. 1992; Christopher et al. 1993). Proximity to historic roadways may also result in slightly higher surface lead concentrations in rural areas.

On the basis of strictly undisturbed sub-surface samples in urban settings, the mean background soil lead concentration in St John's is 37 ± 18 ppm. Assuming a long-range pollution effect extrapolated from nearby rural settings (+20 ppm), the surface soils in St John's should have a mean lead concentration of 57 ppm and a potential maximum of 75 ppm. For surfaces overlying the St John's Group, the occasional soil sample may exceed 100–200 ppm, which means that in rare circumstances natural lead concentrations in surface soil in St John's may be above the CCME guideline of 140 ppm.

City-wide patterns and potential sources

The GM soil lead concentration for the complete St John's data-set is 162 ± 12 ppm (95% CI) and the median value is 148 ppm, 55 ppm lower than that reported in the pilot study. The pilot project included a higher proportion of samples from downtown St John's, where soil lead values are much higher than in the city in general. The present study with its expanded sampling programme detected much higher individual soil lead concentrations compared to the pilot study (six samples higher than the previous maximum of 7047 ppm, with values up to 24,477 ppm), indicating that the local exposure risk is larger than previously thought on some individual properties.

Notwithstanding differences in sampling strategy, the GM and median soil lead concentrations for the city of St John's are higher than other comparable Canadian cities without an industrial lead pollution source, such as Victoria, Sudbury, Ottawa, and Iqaluit (Table 2). St John's soil lead concentrations are either similar to or lower than Canadian communities impacted by industrial sources of lead (Table 2).

Soil lead concentrations are highest around the harbour and in the downtown core of St John's and roughly decrease with distance from the city centre. A similar distance–decay pattern was observed in New Orleans and other cities in Minnesota (Mielke et al. 1984/1985; Mielke 1994), which has been attributed to two main factors: inner cities have a larger proportion of older houses with leaded paint, and there is a history of higher traffic volume and more buildings with which to trap the associated (past) leaded gasoline emissions (Mielke 1994). Another relevant factor for St John's is coal burning for domestic heating, light industrial uses, and coal-fired steamship traffic. Soil surveys in England, for example, indicated that heavy metal associations in lead-contaminated soil matched those in coal and they identified two distinct pathways along which coal contamination had entered the soil: through the atmospheric deposition of aerosols from domestic and industrial coal combustion, and by the spreading of coal ash on back gardens as a soil amendment or for waste disposal purposes (Kelly and Thornton 1996; Rawlins et al. 2002). Similar sources and pathways have been reported for soil lead in St John's with severe smog conditions in the old city centre attributed to smoke from steamships in the harbour and domestic coal fires (Figure 2a).

Despite the general city-wide trends, there are, however, some high soil lead concentrations in outlying areas and low concentrations in the urban core. High concentrations in peripheral areas of the city tend to be associated with older properties or near older roads around which the city has grown. It is likely that sources associated with high soil lead in the inner city, such as leaded paint and coal and gasoline combustion emissions, also apply to these sites. Additionally, high concentrations in ambient samples outside the downtown may be due to sample proximity to old structures (sheds or garages) that have since been removed from the properties. Low soil lead concentrations in the downtown core may result from the use of imported or amended soil and sod for landscaping and renovation purposes. Imported fill is also used in the construction of road medians and sidewalks.

Property-scale patterns

In this study, GM soil lead values were highest for dripline locations (219 ppm) compared to ambient and roadside sampling sites (154 and 136 ppm, respectively), reflecting the influence of past and/or present leaded paint deterioration on exterior wooden clapboard. Road samples have lower concentrations than might be expected, which may reflect the relatively small size of St John's, the corresponding low traffic density, and the relatively new road network in the city, post-dating the removal of lead from gasoline. Roadside locations in suburban residential areas have concentrations similar to nearby ambient samples, as might be expected.

Mielke (1994) used a similar sampling strategy for soil lead characterization of inner-city, mid-city, and suburban locations in New Orleans. Despite the significant difference in population size, ambient and dripline samples in St John's, for the most part, exceeded those of New Orleans (Table 6) – by as much as 500% in the case of ambient values for the inner city. However, the differences in concentrations between the two cities generally decreased from the inner city to the suburbs. The marked difference in ambient soil lead values in the inner city may reflect the close spacing of buildings in downtown St John's (Figure 2a) and the compounding factor of multiple lead sources in small gardens and open areas. Dripline and ambient samples have higher median soil lead concentrations in mid-city areas of St John's, possibly due to a higher lead paint contribution from painted clapboard houses in these areas in St John's, or perhaps a younger housing stock in the mid-city of New Orleans compared to St John's.

Table 6. Comparisons of soil lead concentrations (ppm) in three different locations on properties in New Orleans and St John's

City location	Sample location	New Orleans median	St John's median
Inner city	All samples		981
(pre-1926)	Ambient	212	1077
	Dripline	840	1253
	Road	342	376
Mid city	All samples		224
(1926–1960)	Ambient	40	201
	Dripline	110	412
	Road	110	168
Suburban	All samples		58
(post-1960)	Ambient	28	49
	Dripline	50	62
	Road	86	64

Housing age and soil lead levels

Soil lead concentrations are high for all samples taken on properties developed before 1926 and dripline sample concentrations are also high on properties dating from the late 1940s. The age of these properties correspond to the period when lead concentrations were the highest in paint, up to 50% by weight (CMHC 2005). It also corresponds with a period of widespread airborne pollution and ash disposal from coal combustion. GM dripline and ambient soil lead concentrations are below the CCME guideline of 140 ppm on properties developed after 1961, mirroring the decline in the amount of lead used in paints. This year (1961) is mentioned specifically by the Canada Mortgage and Housing Corporation (CMHC, 2005) as the building date after which homeowners should be less concerned about leaded paint in homes. The use of coal for residential heating also declined in St John's during the 1950s, which may have also contributed to the lower soil lead concentrations. GM roadside lead levels are only slightly above CCME guidelines on properties built before 1948, which may simply reflect the smaller vehicular traffic volume in St John's in the first half of the century (Poole 1994), or the dilution or replacement of -lead-contaminated roadside soils in the last 50–60 years during infrastructure improvements.

Lead isotopic composition of soils and contaminants

Magalios (2005) used lead isotopes to differentiate the sources of lead contamination in St John's soils. Relative proportions of ²⁰⁶Pb, ²⁰⁷Pb, and ²⁰⁸Pb were measured in 27 surface soil samples from road, dripline, and ambient locations, seven undisturbed sub-surface soil samples, two exterior clapboard paint chips (one each from 50- and 100-year-old paint surfaces), and a piece of archived coal. Most of these samples define a linear array trending from bedrock-derived lead in sub-surface samples (natural background) with high ²⁰⁶Pb/²⁰⁷Pb (∼1.18–1.20) and low ²⁰⁸Pb/²⁰⁶Pb (∼2.06–2.08) ratios, to paint-derived lead in dripline samples from pre-1929 houses with low ²⁰⁶Pb/²⁰⁷Pb (∼1.12–1.13) and high ²⁰⁸Pb/²⁰⁶Pb (∼2.13–2.14) ratios (Figure 9; Sylvester et al. 2006). Although the ranges in lead isotopic ratios appear quite narrow, especially for ²⁰⁸Pb/²⁰⁶Pb, these ratios do have the ability to fingerprint specific lead sources that have characteristic isotopic values. For example, the known lead isotopic composition of gasoline has varied from a low ²⁰⁶Pb/²⁰⁷Pb ratio in ore from Trail, BC (1.064) or from Broken Hill, Australia (1.037) to a higher ratio characteristic of ore from southeastern Missouri, USA (1.28–1.33) or Bathurst, New Brunswick (1.16; Rosman et al. 1994).

Figure 9. Lead isotope composition of contaminated urban soils, undisturbed background samples (C-horizon tills), and potential contaminants (coal, paint chips, ore, and atmospheric aerosol). Soil samples from open spaces and yards are equivalent to ambient samples.

On the basis of their relative ratios of ²⁰⁶Pb/²⁰⁷Pb and clustering along the linear array, three groupings of soil samples with discrete sources, ages, and pathways can be deduced:

•	Low 206Pb/207Pb (∼1.12) samples, found mostly in dripline locations of 100-year-old or so houses, likely had a source in leaded paint from the 1800s to early 1900s, which was applied to exterior wooden clapboard. Weathering of these painted surfaces would have added lead dust and chips to dripline soils adjacent to old housing stock.
•	Medium 206Pb/207Pb (∼1.15–1.165) samples are found mainly in downtown residential yards and along city roads and probably had sources in coal ash, lead-based exterior paint from the mid-1900s, and residues of leaded gasoline from automobile exhaust. The close association between these sample ratios and those for single source contaminants plotted on Figure 9 (50-year-old paint chip, archival coal sample, and Bathurst ore used in leaded gasoline) strongly suggests one or more lead sources in coal ash, lead-based exterior paint from the mid-1900s and residues of leaded gasoline from automobile exhaust. Much of the lead loadings from these sources would likely have occurred after World War II, a period of population growth and urban expansion in St John's, and prior to the early 1980s, since when the use of lead-based paint and leaded gasoline was discontinued (Poole 1994; Sylvester et al. 2006).
•	High 206Pb/207Pb (∼1.17–1.20) samples are found in open spaces and near roads. Although this isotopic lead signature forms part of the contemporary atmospheric pollution in St John's, largely associated with long-range transport from the USA, the close association with road samples suggests that the source may be gasoline with lead additives derived from SE Missouri ore (206Pb/207Pb 1.28–1.33), which was used commonly in the USA during the 1960s and 1970s (Rosman et al. 1994).

In summary, isotopic ratios support the view that leaded paint was a major source for lead-contaminated dripline and ambient soils on pre-1926 housing in St John's. A slightly higher isotopic ratio fingerprints ‘younger’ leaded paint, coal, and leaded gasoline, which were widely in use during the urban expansion in St John's around the middle of the twentieth century. Understanding the sources of anthropogenic lead in St John's soils may also be important for consideration of lead bioaccessibility for health risk assessment and lead mobility for exposure mitigation and site rehabilitation.

Indoor dust lead

GM window sill dust lead loadings are much higher than those for entrance and kitchen floors in homes in St John's. This is most likely due to the presence of leaded paint on exterior and interior window frames. Window sill lead loadings were also elevated compared to floor samples in a study in Rochester, New York (GM of 393–476 μg/ft² compared to 8 μg/ft²; Lanphear et al. 1999). Window sill loadings in Rochester are much higher than those reported here, which probably reflects a sampling concentration on urban homes in the Rochester study compared to both urban and suburban homes in this study.

Entrance floors have only slightly higher loading values than kitchen floors have in St John's, in contrast to results from Sydney, NS, where doorway loadings were found to be an order of magnitude higher than kitchen floor ones (Lambert and Lane 2004). Entrance floor loadings were expected to be higher than those on kitchen floors in St John's because of tracking-in of elevated lead soil on shoes or pets in entranceways and the more frequent cleaning of kitchen surfaces as reported by study participants.

Relationship between soil lead and indoor dust lead

In general, dust lead loadings in sampled houses in St John's were more strongly associated with soil lead concentrations from dripline sample locations. Those houses with dust lead loadings in excess of the US EPA guidelines had dripline soil lead concentrations higher than 900 ppm. The relationship between dripline soil lead and indoor dust loadings is the strongest for samples from entrance floors and window sills, which is not unexpected considering their proximity and exposure opportunities (e.g. tracking-in on footwear; Lambert and Lane 2004). In contrast, Rasmussen et al. (2001) reported no statistically significant relationship between outdoor soil concentrations and indoor dust concentrations for sampled properties in Ottawa, which they attributed to the much greater influence of indoor leaded paint sources. In support of this view, it is apparent that many properties in St John's with dripline soil lead concentrations higher than 900 ppm did not have elevated dust lead loadings. Other factors, including the exposure and deterioration of indoor leaded paint and the cleaning regime within the home, must also be important for indoor dust lead loadings.

Housing age and indoor dust lead

For St John's samples, dust lead loadings are much higher in sampled houses built before 1926 and overall GM dust lead loadings decrease with declining age in pre-1948 housing stock. This time period was also used to distinguish between homes with high (pre-1950s) and low (post-1950s) dust lead levels in Ottawa (Rasmussen et al. 2001). Around this time, restrictions on lead concentration in paint began to be implemented while interior paint became lead-free in the late 1970s. Surprisingly, there does not appear to be a dramatic decline in dust lead loadings for houses built after the 1970s in St John's; perhaps renovations and remodelling blur age-specific trends for indoor dust lead in homes built in the last 30 years or so.

Health risk

Soil lead concentrations and soil lead guidelines

There is reason to be concerned about soil lead levels in St John's as 51% of the samples collected and 43% of the neighbourhoods sampled have GM lead concentrations above the CCME residential soil lead guideline of 140 ppm. Based on the findings reported here, all residents may be at risk on properties developed before 1960. Considering that 30% of all houses in St John's were built prior to 1960, there is a potential health risk; however, based on a simple analysis of neighbourhood demographics, most children live in the suburbs and older housing is concentrated in the city centre. Therefore, while the majority of the city's children may not be at risk, for those that do live in pre-1960s housing, and especially those that live in pre-1926 housing, exposure risk to soil lead is potentially high.

Estimated blood lead levels

Lanphear et al. (1998) constructed a multivariate regression model to generate a simple predictive relationship for blood lead concentrations from a suite of variables, including interior floor dust lead loading, exterior soil or dust lead concentrations, and a range of other variables related to household lead sources, exposure pathways, and socio-economic status. By using data on soil lead concentrations and indoor dust loadings from this study together with standard values for the remaining variables, estimates of mean blood lead concentrations in children between 6 and 36 months were generated for St John's (Table 7). For young children living in pre-1926 housing stock, the predicted GM blood lead concentrations were as high as 6.5 μg/dL and the probability that concentrations would exceed the clinical threshold for lead poisoning (10 μg/dL) was 21%. In contrast, the same estimates for post-1950s housing were roughly 3 μg/dL or less and 2% or lower, respectively.

Table 7. Predicted blood lead concentrations for children between 6 and 36 months in St John's based on data from Lanphear et al. (1998)

Property age	Measured GM soil lead concentration (ppm)	Measured GM floor dust lead loading (μg/ft^2)	Predicted GM blood lead (μg/dL)	Predictedprobability >10 μg/dL (%)
>1926	901	15.5	6.5	21.3
1926–1948	334	6.1	4.6	7.7
1949–1960	191	1.0	3.0	1.6
1961–1976	95	2.1	3.2	2.2
1977–1992	49	0.2	<2.6	<0.8
1993–Present	38	1.9	2.8	1.3

Campbell (2008) investigated the potential human health risk associated with environmental lead exposure in St John's using two approaches. The first was a screening-level risk assessment using the Canadian Government's Preliminary Quantitative Risk Assessment (PQRA) procedures (Health Canada 2004b). The PQRA standardizes screening-level risk assessment methodology for federal contaminated sites to facilitate comparisons between sites and to help establish priority areas for remediation (Health Canada 2004b). An average risk scenario was created using 50th percentile soil and dust lead concentrations along with more moderate exposure parameters. According to Campbell's PQRA results, the cumulative lifetime risk (i.e. risk averaged over a lifetime of continuous exposure) associated with environmental lead exposure in St John's is negligible for all residents, except for those who live and consume home-grown garden produce on pre-1926 properties (soil lead concentration >900 ppm). Because of the specific health effects of lead on toddlers, it is possible that this age group is at risk at a soil lead concentration of 900 ppm, even without eating produce from the garden, and at much lower soil lead concentrations if they do.

Second, Campbell used the US EPA Integrated Exposure Uptake Biokinetic (IEUBK) model to estimate lead exposure in St John's children and generate a distribution of blood lead values (US EPA 2002). The model is based on three different components: exposure to different media containing lead, uptake of the lead based on differing bioavailability in the lungs and gastrointestinal tract, and a biokinetic component that models the storage, transportation, and excretion of lead within the body. As also identified in the PQRA, the IEUBK model output for St John's is sensitive to the consumption of home-grown garden produce and, when considered in the model runs, makes a significant contribution to total daily lead intake and blood lead concentrations. For example, in the absence of garden produce consumption, the IEUBK model suggested 41% of children living in pre-1926 housing may have a blood lead concentration higher than 10 μg/dL, whereas with the consumption of garden produce all children living in pre-1977 housing would be expected to exceed this blood lead level.

There is a large amount of uncertainty associated with the estimation of lead concentrations in garden produce, which makes interpreting the risks associated with garden produce exposure difficult. Both the PQRA and IEUBK models generate high lead exposures, which over the short harvest season appears to greatly increase the health risk. Because of the tendency of the IEUBK model to overestimate blood lead levels under extremely high soil lead concentrations (White et al. 1998), its results should be interpreted with caution. Comparisons with other data from St John's and across Canada (Table 1) indicate that predicted blood lead concentrations for the scenario without garden produce may be the most realistic.

Overall, the results of Campbell's (2008) risk assessments for both the PQRA and the IEUBK models suggest that cumulative lifetime risk of negative health effects occurring in St John's may be a problem for those individuals living on the oldest properties with the highest soil lead concentration (>900 ppm). Toddlers living on these properties may also have a temporary increase in health risk even if no garden produce is eaten.

Preliminary screening for children's blood lead

Results from preliminary blood lead studies in St John's and a retrospective chart review for the province of Newfoundland and Labrador indicate that children's GM blood lead concentrations are as high as 3.13 μg/dL, with only 4% or less predicted to have concentrations above 10 μg/dL (Table 8). These measured lead concentrations are much lower than the Lanphear et al. model predictions for children living in pre-1926 housing, but roughly mirror the estimates for post-1950s housing (Table 7). These measured concentrations are also much lower than those predicted by the IEUBK model, even for model runs excluding garden produce consumption. It should be emphasized that this study was based on opportunistic sampling rather than systematic sampling, and that the sample size was very small. As such it is unwise to draw firm conclusions from the results.

Table 8. Summary data for blood lead measurements in children across Newfoundland (unpublished chart review by O'Brien, personal communication 2006) and in the St John's region (unpublished pilot surveillance projects, Office of the Medical Officer of Health, Eastern Health, St John's)

	Newfoundland	St John's region
Dates	1993–2004	August 2004–April 2005	September 2005–December 2006
Sample size	Under 6 years – 77	0.5–6 years– 69	0.5–6 years – 24
GM blood lead conc. (μg/dL)	0–12 years 3.13	0.5–6 years 2.04	0.5–6 years 2.2
% above 10 μg/dL threshold	0–18 years 4.2	0.5–6 years 4.4	0.5–6 years 0

The difference in measured and modelled blood lead concentrations in St John's children may be due in part to some or all of the following factors: (1) small sample size, (2) poor representation of children from older housing stock, (3) low proportion of St John's children living in older housing, (4) higher nutrition levels, (5) good housekeeping practices, or (6) invalid model assumptions or default settings for St John's. In the case of the first two factors, only a targeted study of children living in older properties would address these data gaps. With respect to where children live in St John's, it is roughly estimated from using demographic data that 95% of children 4 years of age or younger live in post-1950s housing and, interestingly, the measured concentrations for children's blood lead are most similar to the modelled results for children living in post-1950s housing stock. Higher nutrition levels tend to reduce the absorption of lead in children, whereas regular house cleaning minimizes the amount of indoor dust and lead exposure. Finally, Campbell (2008) provided an extensive evaluation of the limitations and uncertainties associated with the PQRA and IEUBK model results and recommended that local measurements be taken for a variety of exposure pathways, including lead levels in home-grown garden produce, to increase the confidence of modelled risk scenarios for St John's.

Conclusions and implications

This study has confirmed the speculation that the legacy of past coal combustion and leaded gasoline use in St John's has indeed contributed to a large reservoir of lead in the city's soils (Christopher et al. 1993). Our analysis has shown that more than half the residential soil sampled in the city had lead concentrations exceeding national guidelines of 140 ppm. Of greater concern are the exceptionally high concentrations (>1200 ppm) that tend to occur along the dripline of houses built before the 1960s and anywhere on residential properties that date to the 1920s or earlier. Although most of these properties are focused in the older downtown core, others are located along well-established transportation routes and in older suburbs. In addition to the sources identified by Christopher et al. (1993), leaded paint decayed from clapboarded houses appears to have been an important contributor of lead to dripline soils and, to a lesser extent, open spaces and ambient samples. The tracking and blowing in of lead-contaminated soil and dust in part contribute to elevated indoor dust lead levels, which together with deteriorating interior leaded paint are responsible for a high lead exposure risk, especially for young children in older housing in St John's.

The levels of lead in St John's soils, particularly on pre-1950s properties, are well above those from Canadian cities of equal or greater population size (e.g. Victoria and Ottawa) and are closer to those with point-source emissions of lead or densely populated North American cities (e.g. New Orleans). St John's high soil lead levels are the product of centuries of urbanization, together with traditional practices of painted clapboard housing and coal burning, both of which released lead to the local environment. High soil lead levels in urban environments therefore do not require an industrial source, and consequently other communities of similar longevity and tradition (e.g. in north-eastern North America) may have also inherited a reservoir of lead in soils associated with their older housing stock.

Based on the results of the pilot study, estimated blood lead levels suggest that, while the wider St John's community may have a low risk for negative health effects from lead-contaminated soil and indoor dust, there is a small exposure risk for toddlers living in pre-1950s housing, which for the most part is concentrated downtown and in surrounding neighbourhoods. A census of children living in pre-1920s housing and a targeted blood lead survey of these children at risk is desirable but a targeted education programme to mitigate and reduce environmental lead exposure could prevent potential problems. Reducing exposure risk to soil lead is relatively straightforward and involves covering over bare soil with grass sod, wood chips, or ornamental patio products, especially in children's play areas and along the dripline of older houses (CMHC 2005). As the consumption of garden produce grown on elevated lead soils may increase exposure risk, care should be taken to avoid, remediate, or replace contaminated soils in vegetable and fruit gardens (e.g. Finster et al. 2004; Hough et al. 2004)

Pilot surveillance programmes for children's blood lead in St John's have to date reported relatively low levels in contrast to predictions based on empirical models, although sampling was not systematic and sample size small. This discrepancy is discussed above in terms of model parameters and data quality; it is also instructive, however, to identify factors about St John's that might cause overestimation of model predictions. For example: (1) the bioaccessibility of sources for soil lead in St John's – thought to be primarily paint – may be less than model default values; (2) soil lead levels are particularly high at dripline locations on pre-1950s properties in St John's and hence this spatially restricted source may limit the overall exposure risk; (3) the seasonal exposure to bare soil for children in St John's is limited to the snow-free period, which is normally 5 months and considerably shorter than in many US cities; (4) based on a limited questionnaire survey, only about 10–15% of houses in St John's have gardens where edible produce is grown and most of the produce is fruit, not vegetables, which have a lower lead exposure risk (Finster et al. 2004; Campbell 2008); (5) the older housing stock has experienced considerable renovations in St John's, causing a reduction in indoor dust lead exposure; and (6) in contrast to major US cities, St John's inner core has relatively few children living in the housing with higher soil exposure risk (Campbell 2008). Together these factors may explain why predicted blood lead levels are much higher than measured blood lead levels in St John's to date, even though it has relatively high soil and moderate indoor dust lead levels. It is important, however, for St John's residents not to become complacent about lead exposure for they must live with the legacy of lead in their environment for some time to come.

Acknowledgements

We acknowledge the participation of St John's homeowners in this study. Funding was provided by the Research Advisory Committee of The Janeway Children's Health and Rehabilitation Centre, the Natural Sciences and Engineering Research Council of Canada, and the Faculty of Arts research grant programme, Memorial University. Melissa Putt and Dominique St Hilaire participated in soil lead sampling and Lakmali Hewa and Pam King assisted with the dust sample analysis. Marc Poujol performed the lead isotope analyses. We thank Chris Finch at the Geochemical Laboratory of the Geological Survey of Newfoundland and Labrador for soil lead analysis. The manuscript benefited from the comments of two journal reviewers.