- Top of page
- Supporting Information
The adverse effects of lead have been known since early times. Throughout the last century, an increasing number of studies were published about childhood lead poisoning. In 1960, the US ‘acceptable’ blood lead value in children was 60 μg/dl. This action level was gradually lowered to 40 μg/dl in 1970 and to 25 μg/dl in 1985. In 1991, the US Centers for Disease Control and Prevention (CDC) established a blood lead level of concern (BLLs) of 10 μg/dl for children (The Centers for Disease Contol & Prevention 1991). In the last decades, toxicity of lead at very low doses in young children has been extensively reported in the medical literature. Evidence from meta-analyses and cohort studies suggests that even at low doses (<10 μg/dl), lead could seriously alter intellectual function in young children. A blood lead increase from <1 to 10 μg/dl is associated with an IQ decrement of 6.2 (95% Confidence interval: 3.8–8.6) (Lanphear et al. 2005). According to the US National Toxicology Program, there is sufficient evidence that BLLs <10 μg/dl are associated with delayed puberty, reduction in post-natal growth and diminution of IQ in young children (U.S.National Toxicology Program 2011). There is also sufficient evidence that children's BLLs <5 μg/dl are strongly associated with decreased academic achievement and increased incidence of attention deficit–hyperactivity disorder. The BLL of concern (i.e. 10 μg/dl) has been revised, and the Advisory Committee on Childhood Lead Poisoning Prevention established the ‘reference value’, currently 5 μg/dl (Centers for Disease Contol & Prevention 2012a). The challenge with lead is linked to the absence of an identified threshold for its toxicity. As a result, the ideal acceptable level should be zero (0 μg/dl). This is unrealistic for children who have previously undergone chronic exposure to lead. Although bone lead resorption is mainly reported and documented in pregnant (or lactating) and post-menopausal women, this phenomenon is likely to occur in young children given the high bone turnover rate observed during bone formation. Much has been written about the toxicokinetics of lead in childhood and adolescence. Details are presented in the ATSDR's report (Agency for Toxic Substances & Disease Registry 2007).
Children living in sub-Saharan Africa (SSA) have every known risk factor for lead poisoning. However, it appears that no previous study synthesised the literature reporting the children's BLLs in SSA. The present systematic review addresses the key question about the levels of blood lead in sub-Saharan African children below the age of 6 years, based on papers published in the last decade.
- Top of page
- Supporting Information
A systematic computerised literature search of PubMed and Web of Science databases was performed for papers published in peer-reviewed journals. Search terms used to identify potential relevant studies were ‘Blood lead’, ‘Lead poisoning’, ‘Lead intoxication’, ‘Child*’, ‘Infant’ and ‘Preschool’. The Boolean operators AND and OR were used, the latter to distinguish between terms relating to the same concept and the former to link concepts to one another. In addition, we used the Medical Subject Headings thesaurus (MeSH) and the CISMeF index of French language health resources to identify synonyms. The MeSH terms identified were then included in the process. The different search strategies we used are provided (See Supplemental Material 1). We also used Google Scholar (http://scholar.google.com) including the same keywords and search terms. All searches were performed by one of the authors (GN) between 30 January 2013 and 30 March 2013.
We considered all studies reporting data on children's BLLs. Studies that met all of the following inclusion criteria were registered: studies published in French or English; studies conducted in SSA; studies published between 1 January 2000 and 30 January 2013 in peer-reviewed journals. Papers published were excluded if one of the following criteria was met: did not report mean/median BLLs in children aged below 6 years; had an ecologic design; did not report original research. However, references of reviews were consulted for identification of potential missed studies.
To ensure the quality of the data extraction, the full texts of all studies identified were independently reviewed by both authors. Both readers used the same data extraction form for their analysis, and data were broken down into five categories: aims of study; descriptive characteristics: place of study, period of study, sampling method, sample size; methods for blood collection or for assessing blood lead data; characteristics of subjects: range of age and socio-economic status, when available; blood lead concentration (mean, median, measures of dispersion, when available). Discrepancies were resolved by consensus.
Each study included in the review was assessed for quality by both authors using the following criteria: blood collection methods; sampling methods and quality control during lead analyses. We first evaluated the blood collection method. Studies using capillary blood samples were considered to be of a much lower quality for the purposes of this analysis than those using other blood samples, given that capillary sampling is more susceptible to environmental contamination than venous sampling (Schlenker et al. 1994). The selection process was also a key element. Studies with random selection were considered better, as randomisation is an effective safeguard against selection bias (Hernan et al. 2004). Studies less likely to present selection bias, using venous/cord blood samples and reporting adequate procedures for quality control of blood lead analyses were classified as ‘category 1’. Studies missing one of the three last characteristics were classified as ‘category 2’ and those missing more than one of these characteristics were classified as ‘category 3’.
We presented geometric mean of BLLs as reported in the included studies. For follow-up studies, we reported mean BLLs observed at the entry in the study. In addition, we reported the mean BLLs at the end of study (if children were <6 year old at the end of the study). Finally, we restricted our analyses only on studies classified as ‘category 1’. The mean weighted by sample size was then estimated by the following formula:
with GMi being the geometric mean reported in study i and Ni referring to the sample size for study i.
- Top of page
- Supporting Information
A total of 11 148 papers were identified in both electronic databases. After a first screening, we identified 742 relevant papers as shown in the flow chart (Figure 1). Most (N = 726) were not conducted in SSA. From the 16 papers still being considered, a further four papers were set aside because they did not include children below 6 years (Mathee et al. 2002, 2006; Diouf et al. 2003, 2006). Our analysis of the methods of the 12 remaining papers led to another six papers being discarded because they did not specify the age of included children (von Schirnding et al. 2001; Röllin et al. 2007, 2009) or they did not specifically report blood lead concentration in the group of included children aged below 6 years (Nriagu et al. 2008; Were et al. 2008; Haefliger et al. 2009; Lo et al. 2012). After screening the reference lists of the six remaining papers, two additional papers were identified (Pfitzner et al. 2000; Tuakuila et al. 2010) and one more was identified through Google Scholar (Röllin et al. 2009). The end result provided the basis for this systematic review of nine relevant papers (Pfitzner et al. 2000; Wright et al. 2005; Olewe et al. 2009; Röllin et al. 2009; Naicker et al. 2010; Tuakuila et al. 2010, 2013; Keating et al. 2011; Dooyema et al. 2012). Most studies had a cross-sectional design (N = 7) (Pfitzner et al. 2000; Wright et al. 2005; Olewe et al. 2009; Röllin et al. 2009; Tuakuila et al. 2010, 2013; Dooyema et al. 2012), and four were conducted in Nigeria (Pfitzner et al. 2000; Wright et al. 2005; Keating et al. 2011; Dooyema et al. 2012). The age of children ranged from birth to 60 months. Table 1 presents characteristics of included studies.
Table 1. Characteristics of included studies
|Study||Period of study||Place of study||Children's age (Sample size)||Method for blood collection||Method for recruitment|
|Pfitzner et al. (2000)||March–April 1997||Jos (Nigeria)||6–35 months (N = 218)||Venous blood||Randomised cluster sampling|
|Wright et al. (2005)||1991||Jos (Nigeria)||6 months–5 years (N = 64)||Venous blood||Randomised cluster sampling|
|Olewe et al. (2009)||April–August 2007||Nairobi (Kenya)||6–59 months||Capillary blood||Hospital-based sampling|
|Röllin et al. (2009)||2005–2006||7 sites (South Africa)||Newborn (N = 67)||Cord blood||Hospital-based recruitment|
|Naicker et al. (2010)||1990–2010||Johannesburg–Soweto (South Africa)||Newborns (N = 618)||Cord blood||Hospital-based sampling|
|Tuakuila et al. (2010)||May 2003–June 2004||Kinshasa (Democratic Republic of Congo)||0–5 years (N = 100)||Venous blood||Mobilisation campaign|
|Dooyema et al. (2012)||May–June 2010||North-western area (Nigeria)||2–59 months (N = 68)||Venous blood||Door-to-door survey|
|Keating et al. (2011)||1998–2000||Jos (Nigeria)||12–18 months (N = 218)||Venous blood||Care-centre-based sampling|
|Tuakuila et al. (2013)||Unclear, but surely after 2004||Kinshasa (Democratic Republic of Congo)||1–5 years (N = 55)||Venous blood||Word-of-mouth campaign|
Based on the quality criteria previously defined, five studies were classified in ‘category 1’ (Wright et al. 2005; Naicker et al. 2010; Tuakuila et al. 2010, 2013; Dooyema et al. 2012). As shown in Table 2, other studies did not meet all criteria for being classified in this category. Hospital- and school-based recruitments were considered to be more susceptible to selection bias.
Table 2. Quality rates of included studies
|Study||Selection bias||Venous or cord blood||Details on quality control assessment||Category|
|Pfitzner et al. (2000)||Less likely||Yes||No||2|
|Wright et al. (2005)||Less likely||Yes||Yes||1|
|Olewe et al. (2009)||Likely||No||Yes||3|
|Röllin et al. (2009)||Likely||Yes||Yes||2|
|Naicker et al. (2010)||Less likely||Yes||Yes||1|
|Tuakuila et al. (2010)||Less likely||Yes||Yes||1|
|Dooyema et al. (2012)||Less likely||Yes||Yes||1|
|Keating et al. (2011)||Likely||Yes||Yes||2|
|Tuakuila et al. (2013)||Less likely||Yes||Yes||1|
Recent studies reporting BLLs in sub-Saharan African children below six are scant. Data on BLLs are shown in Table 3. Eight studies reported BLL data in children from urban areas and two studies from rural areas. Authors did not systematically report the presence of lead industries/mining sites in the areas of study. The geometric means vary between 5.9 and 153.3 μg/dl, and the available prevalence of BLLs ≥10 μg/dl varies from 7.0% to 70.9% depending on the place of study. It was not possible to estimate the pooled estimate weighted by inverse variance given the small number of studies reporting the variance statistic. However, the mean weighted by sample size was 13.1 μg/dl. According to the analyses involving only ‘category 1’ studies, the mean weighted by sample size was 16.2 μg/dl.
Table 3. Average Blood lead levels (BLLs) in sub-Saharan African children from studies published between 2000 and 2013
|Study||Settings||Blood lead levels (μg/dl)|
|Geometric mean||Median||Range||Prevalence of BLLs ≥10 μg/dl (%)|
|Pfitzner et al. (2000)||Urban||15.2||NA||1.0–>60.0||70.0|
|Wright et al. (2005)||Urban||11.2||NA||NA||55.0|
|Olewe et al. (2009)||Urban||6.0||5.4||3.3–24.7||7.0|
|Röllin et al. (2009)||Rural||NA||NA||NA||NA|
|Naicker et al. (2010)||Urban||5.9||NA||2.0–17.0||>50.0|
|Tuakuila et al. (2010)||Urban||12.4||NA||NA||63.5|
|Dooyema et al. (2012)||Rural||Village A: 153.3 Village B: 107.5||Village A: 143.8 Village B: 87.5||Village A: 55.9–331.0 Village B: 36.5–445.0||NA|
|Keating et al. (2011)||Urban||11.1||9.0||1.0–43.0||44.7|
|Tuakuila et al. (2013)||Urban||11.2||11.5||3.0–37.8||70.9|
- Top of page
- Supporting Information
A total of nine studies conducted in SSA and published between January 2010 and January 2013 were identified as reporting mean/median BLLs in children aged below 6 years. The geometric mean weighted by sample size was 13.1 μg/dl and reached 16.2 μg/dl by restricting analyses to high-quality studies. In six of the studies, at least half of children had BLLs ≥10 μg/dl. Overall, it is clear that the geometric mean of BLLs in young children living in SSA remains high relative to the US children. Results from the NHANES 1999–2004 indicated that the geometric mean BLL in 1–5-year-old US children was 1.9 μg/dl (95% CI: 1.8–2.0 μg/dl) (Jones et al. 2009). This survey also indicated that the distribution of BLLs for the US children has shifted towards lower BLL categories between the 1988 and 1991 period and the 1999–2004 period. Conversely, the mean BLL estimated in this systematic review remains approximately similar to mean values reported in SSA between 1982 and 1995 (Nriagu et al. 1996). The high mean BLLs observed in sub-Saharan African children probably reflect high levels of lead in children's environments. Sources of lead are identified and largely described in the literature.
Foetuses may be exposed to lead throughout pregnancy given that lead can easily cross the placental barrier. Previous studies suggest that bone lead release increases during pregnancy and constitutes the dominant contributor to blood lead (Flegal & Smith 1992; Leggett 1993; Agency for Toxic Substances & Disease Registry 2007). As a result, there may be an inequality in the risk to foetuses of contact with lead, depending on the past exposure of their mothers. It has been reported that 78.9% of Nigerian pregnant women had BLLs ≥10 μg/dl (range: 0.5–448 μg/dl), although this study suffered from a small sample size (Njoku & Orisakwe 2012). A recent study indicated that the geometric mean of BLLs in pregnant Nigerian women (15–40 years old) varies between 2.7 and 73.8 μg/dl (Adekunle et al. 2009; Ugwuja et al. 2012). In South Africa, the lead levels in cord blood ranged from 2 to 17 μg/dl (Naicker et al. 2010). All these values are higher than those recently reported in North Carolina women (USA) (0.07–0.13 μg/dl) (Sanders et al. 2012) or from NHANES 2003–2004 (1.78 μg/dl) (Lee et al. 2005) or from the Duke cohort (more than 75% of pregnant women with BLLs <1.00 μg/dl) (Miranda et al. 2010) or in Quebec (1.50 μg/dl) (Rhainds et al. 1999). On that basis, young children living in SSA are more likely to be exposed to lead in utero than US children.
Breastfeeding constitutes a potential source of exposure if a child's mother has a high body burden of lead. Urban women living in South Africa are reported as having high BLLs (Median: 32.9 μg/l; range: 16.3–81.5 μg/l) (Röllin et al. 2009). In Kinshasa (Democratic Republic of Congo), women showed a median BLL of 9.4 μg/dl (range: 2.9–37.8 μg/l) (Tuakuila et al. 2013). Studies estimating milk lead levels in African women are not common. However, one study reported the geometric mean of 10.2 μg/dl in Fulani women in Nigeria (VanderJagt et al. 2001).
Drinking water is another potential source of lead for young children, particularly for those living in homes with lead service lines (Brown et al. 2011). Like people from Western countries, Africans used lead-based materials for plumbing systems. No study appears to have been conducted for estimating the proportion of homes with lead service lines in urban areas of SSA. Unlike North American countries, programmes for monitoring lead in water have not been clearly established in SSA. Studies reporting tap water lead concentration in SSA are scant. In Kampala (Uganda), children using bore hole/dug well as a water source are reported to be less likely to develop BLLs ≥10 μg/dl than children using water from community systems (Graber et al. 2010). However, authors did not report lead concentrations in drinking water. More recently, in Kinshasa (Democratic Republic of Congo), the geometric mean of lead concentration in drinking water was reported to be 0.24 μg/l (95% CI: 0.16–0.37) (Tuakuila et al. 2013). This value does not exceed the maximum permissible WHO levels (10 μg/l). Yet, authors did not state whether these values referred to flushed or stagnated water samples. Water lead concentration may drastically increase during stagnation, and this increase is supposed to be proportional to water temperature (Cartier et al. 2011). This is a great concern in sub-Saharan countries given the high ambient temperature throughout the year.
The main sources of exposure of African children to lead include leaded gasoline (Nriagu et al. 1996). In Nigeria, automobiles account for about 90.0% of all lead released annually to the atmosphere (Obioh. 1993 cited by Nriagu et al. 1997). Phase-out of lead gasoline began in the United States in 1970s (Agency for Toxic Substances & Disease Registry 2007), and an important part of notable reductions in BLLs of the US children during 1970–1990 is attributed to this action. Unfortunately, leaded gasoline has been used in SSA up to the 21st century. While decreasing in the United States, the contribution of lead to global atmospheric pollution increased in SSA. Authors reported that this contribution increased from <5% in 1980 to about 20% in 1992 (Pacyna. 1993 cited by Nriagu et al. 1996). In the mid-1990s, all vehicles in Nigeria used leaded gasoline and the gasoline lead level was about 0.74 g/l (Ogunsola et al. 1995). Moreover, gasoline with lead concentration ≥0.50 g/l was reported as being used in most sub-Saharan countries during the 1990s (Alliance to End Childhood lead poisoning, 1994 cited by Nriagu et al. (1996). By comparison, gasoline lead concentrations were 0.15 g/l in Belgium, 0.13 g/l in Spain, 0.11 g/l in Italy, 0.05 g/l in the United Kingdom and Switzerland, and 0.00 g/l in Canada, in the United States and in Sweden as reported elsewhere (Thomas et al. 1999). Lead was removed from South African petrol only in the beginning of 2006 (Röllin et al. 2007, 2009) and since 2005 in Democratic Republic of Congo (Tuakuila et al. 2013). As reported elsewhere, most sub-Saharan countries began using unleaded gasoline only in 2004–2005 (United Nations Environmental Programme 2011). In Johannesburg, children's BLLs showed a huge decrease of 2.8 μg/dl (on average) since the introduction of unleaded gasoline, used by approximately 30% of vehicles (Röllin et al. 2007, 2009).
Lead-contaminated residential dusts and paints also are an important route of exposure in young American children (Lanphear et al. 1998; Ryan et al. 2004; Agency for Toxic Substances & Disease Registry 2007; Dixon et al. 2009). High paint lead concentration has been observed in a large number of South African homes, and authors reported that 20% of homes investigated had paint lead concentration >5000 μg/g (The US reference level) (Mathee et al. 2007). The median concentration reported in a survey in Kinshasa (Democratic Republic of Congo) was 13 μg/g (range: 9–1890 μg/g) (Tuakuila et al. 2013). One study conducted in Johannesburg indicated that 55% of suburbs contained at least one home with lead-based paint (Montgomery & Mathee 2005). More recently, a survey conducted in Cameroon reported that 66% of the new paint had lead concentrations >90 ppm and 98% of samples that exceeded 90 ppm were in excess of 600 ppm (Gottesfeld et al. 2013). However, the US regulation stated that ‘paint and similar surface-coating materials for consumer use that contain lead or lead compounds and in which the lead content (calculated as lead metal) in excess of 0.009% are banned hazardous products (effective from 14 August 2009)’ (The Consumer Product Safety Commission 2008).
Urban areas in sub-Saharan countries are very dusty given the low proportion of paved road and the high traffic density. Reported dust lead concentrations are ≥7000 μg/g in Lagos (Nigeria) and 4000 μg/g in Nairobi (Kenya) (Ogunsola et al. 1995). Reported indoor dust loading was very high in Kinshasa with a median value reaching 76.4 μg/ft2 (range: 8–1038 μg/ft2) (Tuakuila et al. 2013).
Additional sources of lead may include medicinal products. Traditional medicine practices are spread throughout SSA. For example, in Nigeria, use of herbal medicine at home constitutes the primary line for fighting high fever from malaria in 60.0% of children (World Health Organization 2003). Most recent studies assessing lead concentration in medicinal/herbal products commonly used indicate that none of them exceeded the WHO maximum limit value of 10 μg/g (Steenkamp et al. 2000, 2002; Ajasa et al. 2004; Maiga et al. 2005; Adepoju-Bello et al. 2012; Ebrahim et al. 2012; Samali et al. 2012). However, care should be taken because traceability of most medicinal products is not easy in sub-Saharan Africa. The lead content may depend on the place where plants have been collected. For example, plants growing near industries or highways are more likely to contain lead. A recent case report from CDC indicated infant lead intoxication associated with the use of tiro, a Nigerian folk remedy containing 82.6% lead (Centers for Disease Contol & Prevention 2012b).
As a whole, results from this systematic review suggested that the mean BLL in young children living in SSA is elevated relative to the US children. The new reference value of 5 μg/dl for identifying children with ‘elevated BLLs’ will surely impact public health policies elsewhere because many nations use CDC benchmarks as their own. This standard is mainly based on the 97.5th percentile of the BLLs distribution among children 1–5 years old in the United States. However, the reality is very different in SSA. The fact is that the lower limit will allow CDC authorities to maintain a high budget for lead poisoning prevention. Furthermore, it also implies that important efforts should be directed towards the initiation (or reinforcement) of prevention programmes and research studies in SSA. Health surveillance for lead should be improved (or initiated) in children living in SSA. Blood lead testing, physical examination and biological monitoring are required.
This review has a number of limitations. First, we restricted our search to studies published in peer-reviewed journals; results described in grey literature were not taken into account. Second, the study suffers from an absence of statistical measures of dispersion, and as a result, it is not possible to assess the variability of weighted mean of BLLs. Finally, of all sub-Saharan African countries, studies we identified were conducted in only four countries (Nigeria, South Africa, Democratic republic of Congo and Kenya) and none of these studies were nationally representative. As a result, data we report here may not completely generalisable.
This review leaves many avenues for future investigations. A large study is needed for assessing the trend of BLLs in young children over the past 10 years given that leaded gasoline has just been removed in a few African countries at the beginning of the 21st century. Studies assessing lead exposure are required as lead exposure from drinking water, soil, dust and paint remains largely unknown in sub-Saharan African countries. A specific protocol for sampling household water is needed and exposure characterisation should include rainwater and well water because some studies suggested that atmospheres with high lead burden may be affecting rainwater lead concentration (Sanderson & Marchand 1984; Cizmecioglu & Muezzinoglu 2008; Farahmandkia et al. 2010). Due to non-availability of potable water supply, rainwater is a main source of drinking water for some people in both urban and rural areas. Most published studies are conducted in South Africa, and studies reporting recent data elsewhere are scant. A zonal-stratified analysis including a large and representative household sample is required. As a whole, thorough and geographically widespread studies need to be carried out to find the true incidence of lead poisoning in SSA children. However, even without data on the full extent of childhood lead poisoning in SSA, known effective public health measures including the elimination of residential lead paint, control of emissions from formal and informal ore processing and metal recycling and increased availability of lead safe cosmetics and traditional medicines should be undertaken.
It should be noted that in the early 2000s, the mental disability burden in young children accounted for 74.6% of the total disability-adjusted life years attributable to lead exposure (Norman et al. 2007). As children in SSA have every known risk factor for lead poisoning, one might expect to find high BLLs and an elevated prevalence of lead poisoning in children living in SSA. The available data, although scanty, support this expectation. In conclusion, while primary prevention measures to reduce lead exposure are being reinforced in Western countries, a close attention should be paid to sub-Saharan African children below 6 years.