Stakeholders, including Nigerian state health officials and MSF, recommended the two most affected villages (A and B) for participation. The investigation protocol was reviewed and approved by both the Nigerian government and the CDC. The investigation was conducted in accordance with the Declaration of Helsinki developed by the World Medical Association (2008)
After obtaining consent of village leaders, investigators conducted a cross-sectional, door-to-door survey from 23 May to 4 June 2010 to interview parents, sample blood from children < 5 years of age, and collect soil from households. Informed consent to administer the survey, draw blood, and collect environmental samples was obtained from the head of the household before survey initiation. Local health professionals and translators were trained in a 4-hr session on survey administration, venous blood drawing, and environmental sampling.
Each village consisted of numerous family dwellings separated by low walls. Investigators defined a compound as several multigenerational and multifamily dwellings enclosed by a common wall. All compounds in village A were eligible to participate in the survey. Because of time and logistical constraints, only compounds in the central area of village B were eligible to participate. The central area of village B included most of the compounds and common gathering places such as mosques, markets, and the residence of the head of the village.
The survey collected information about children < 5 years of age in each compound, including the number living in the compound, the number with a history of convulsions, the number who had died in the preceding 12 months, and their approximate date of death.
Before the investigation, the team heard that artisanal gold ore–processing activities were occurring in both villages. These ore-processing activities included a) breaking rocks into small gravel-size pieces (breaking); b) grinding rocks into a fine powder with a flour mill or mortar and pestle (grinding); c) washing ground ore powder with water to separate gold particles (washing); d) drying ground ore after washing (drying); e) using liquid mercury to amalgamate gold flakes (separating); and f) using heat to vaporize mercury from the gold mixture after amalgamation (melting). Both men and women in the villages participated in ore processing. Men typically processed ore in central locations around the village, and women processed ore inside family compounds. The survey included questions about gold ore–processing activities inside and outside the compound, household member and maternal processing activities, history of animal deaths within the compound, and the primary water source of the family. Before data collection, investigators mapped each village and marked the location of each family compound. Global positioning system (GPS) coordinates were taken at the entrance of each compound to facilitate compound identification and follow-up on blood and environmental testing results. At the end of the investigation, MSF staff visited the compounds and provided parents with blood lead test results of their children as well as information on chelation therapy and the medical management of lead poisoning.
Venous blood was collected from children 2 months to 5 years of age. Phlebotomists attempted to draw blood from every available child in village A and from children in every other surveyed compound in village B within time constraints. Samples of manufacturer lots of materials used for blood collection were prescreened by CDC laboratories and determined to be free of lead. To prevent sample contamination, all blood collection supplies were kept in plastic gallon-size storage bags before sample collection. In addition, the venipuncture site was thoroughly cleaned with alcohol wipes before the specimen was obtained. One 1- to 3-mL blood sample was collected in a laboratory tube and analyzed for lead using a portable blood lead analyzer, LeadCare II® (Magellan Biosciences, Chelmsford, MA, USA) in an uncontaminated area away from the villages. This portable instrument can reliably determine BLLs from 3.3 to 65 µg/dL with an accuracy level of ± 3 µg/dL (Freeney and Zink 2007
A second blood sample was collected from every third child and analyzed for lead, total mercury, and manganese by the Inorganic and Radiation Analytical Toxicology Laboratory at the CDC National Center for Environmental Health in Atlanta, Georgia. Inductively coupled plasma mass spectroscopy was used to analyze lead and total mercury. Detailed explanations of CDC blood lead and total blood mercury laboratory methods have been published elsewhere (Caldwell et al. 2009
; Jones et al. 2007
). Inductively coupled dynamic reaction cell plasma mass spectrometry was used for whole-blood manganese analysis as previously described (Jones et al. 2010
). The limit of detection for blood lead is 0.25 µg/dL, total blood mercury 0.33 µg/L, and manganese 0.8 µg/L. Precision was evaluated by monitoring the replicate results of internal quality control (QC) materials. QC tests, included at the beginning and end of each analytical run, help ensure the accuracy and precision of the analysis process. The low-level QC should be in the low to normal range for blood levels in the U.S. population, and the high-level QC should be less than that found in the high-normal population range of the U.S. population. The population ranges are taken from the Fourth National Report on Human Exposure to Environmental Chemicals
). The low-level QC for mercury had an interday coefficient of variation (CV) of 15.1 at 0.516 µg/L and a CV of 2.3 at 5.857 µg/L. For lead, the low-level QC had an interday CV of 1.7 at 2.876 µg/dL and a CV of 1.2 at 12.754 µg/dL. For manganese, the low-level QC, with a mean of 7.983, has a CV of 4.8%, whereas the high-level QC with a mean of 14.929 has a CV of 6.7%. Accuracy for blood lead and mercury was verified by analyzing standard reference material from the National Institute of Standards and Technology (Gaithersburg, MD, USA) . Blood manganese accuracy was verified by participating in proficiency testing with the Wadsworth Center of New York Trace Elements in Whole Blood Program (Albany, NY, USA). The level of concern for blood lead is 10 µg/dL (CDC 2002
). Although there is no blood level of concern for metallic mercury exposure, long-term effects, largely neurological, have been noted at total blood mercury levels < 1.0 µg/L (Agency for Toxic Substances and Disease Registry 1999
). Investigators used the reference range of 7.7–12.1 µg/L from Tietz Textbook of Clinical Chemistry
for blood manganese (Milne 1999
). Although some heavy metals, such as mercury, are better evaluated using urine biomarkers, logistical constraints precluded collection of urine specimens.
Environmental sampling. Places where children ate or slept, which were identified by the eldest mother in each compound, were targeted for environmental sampling in each surveyed compound. Samples of soil were swept, placed in a plastic bag, and analyzed for lead content using a portable, hand-held X-ray fluorescence spectrometer (XRF) (Innov-XSystems, Woburn, MA, USA and Thermo-Scientific Niton, Billerica, MA, USA) in an uncontaminated area away from the villages. A certified industrial hygienist or environmental engineer also assessed each village and took XRF readings throughout the villages using U.S. Environmental Protection Agency (EPA) method 6200 to determine areas of high contamination (U.S. EPA 2007). Inside the villages, priority for XRF assessment and analysis was given to areas potentially affected by ore processing. The limit of detection for lead by XRF was approximately 40 ppm, whereas the upper reporting limit was 100,000 ppm. Samples of processed ore were also collected and analyzed for lead using XRF. After XRF analysis, a subset of soil and ore samples was sent to the U.S. Geological Survey (USGS) for further analysis using inductively coupled plasma mass spectroscopy to verify XRF field results and to perform mineralogy analysis to determine the chemical composition of the samples.
The U.S. EPA’s Lead: Renovation, Repair, and Painting Program (U.S. EPA 2008) defines a soil lead hazard as bare soil containing total lead ≥ 400 ppm in a child’s play area or 1,200 ppm in bare soil in other parts of a yard where a child lives (U.S. EPA 2008). U.S. EPA standards were used to categorize XRF soil lead results from surveyed compounds into three groups: ≤ 400 ppm, 401–1,200 ppm, and > 1,200 ppm. A limited number of water samples were collected by a certified industrial hygienist. Water was drawn up in a bucket, and the sampling bottle was immediately submerged to obtain the sample. Water samples were stabilized with 1 mL nitric acid. Unfortunately, no field blanks were taken. Water samples were sent to a commercial environmental testing laboratory in the United States and analyzed using U.S. EPA method 200.8 for lead, arsenic, and manganese. The U.S. EPA action level of 15 ppb for drinking water was the reference standard for water results (U.S. EPA 1991).
Statistical analysis. Data from the household survey, blood lead results, and environmental sample results were entered into Epi-Info version 3.5.1 (CDC, Atlanta, GA, USA). Statistical analyses were performed using SAS software (SAS Institute Inc., Cary, NC, USA). Univariate analyses, including calculation of daily and overall under age 5 mortality rates (U5MR) were performed. The daily U5MR was calculated by taking the number of deaths per 10,000 children per day for the 6-month period of December 2009—May 2010. This rate was compared with the United Nations High Commissioner for Refugees (UNHCR) threshold of > 2.0/10,000/day, which indicates deteriorating conditions in a refugee relief situation (UNHCR 2007). To calculate the overall U5MR, the number of deaths was divided by the number of children alive in the compounds in the 12-month period of May 2009–May 2010 and multiplied by 1,000. This was compared with the U5MR per 1,000 live births for this region of northwestern Nigeria (217 in 1,000 live births) as a way to crudely estimate excess mortality in the villages (NPC and ICF Macro 2009).
Bivariate and multivariate logistic regression was used to identify risk factors for child mortality from suspected lead poisoning. Using this outcome, compounds without any children < 5 years of age were excluded from bivariate and multivariate analyses. Risk factors were grouped into demographic, ore processing, and other environmental risk factors. Demographic risk factors included age, sex, and village of residence. Ore-processing risk factors included whether members of the compound or a child’s mother participated in any of the six ore-processing activities. Investigators focused on maternal risk factors because children < 5 years of age in this population typically spend the majority of time with their mothers in the compound. Environmental risk factors included proximity to ore-grinding activities, presence of ground material inside the compound, soil lead concentrations within the compound, primary water source, and history of animal death (as a proxy for environmental contamination) within the compound. The outcome of interest was death of a child. Multicollinearity, intervariable correlation, the Hosmer–Lemeshow goodness-of-fit assessment, and the effect of each risk factor on the outcome were considered. Variables from bivariate analysis with a p-value < 0.1 were tested in the model using backward selection. Variables with a p-value of < 0.05 remained in the final model.