Elevated blood-lead levels among children living in the rural Philippines

Abstract

Objective

Generally, lead poisoning is not considered a significant environmental hazard for children in rural areas of developing countries. With a prospectively designed policy experiment, the research community and the government are conducting a broad-based investigation to introduce and evaluate the impact of health policy reforms on children in a rural area of the Philippines – the Quality Improvement Demonstration Study (QIDS). As part of this study, we researched lead exposure in children under the age of five.

Methods

We sampled a population of children from the Visayas region in the central Philippines, covering approximately one third of the country’s geographical area. From December 2003 to September 2004, the survey collected blood lead levels (BLL) together with demographic, socioeconomic and child health data points. Supplemental field-testing among a sub-sample of the most exposed children assessed the sources of environmental lead exposure.

Findings

Among children in this study, 21% (601 of 2861 children) had BLL greater than10 µg/dl. BLL were associated independently with age, haemoglobin concentration, water source, roofing material, expenditures and history of breastfeeding. A follow-up assessment of possible environmental exposures among the sub-sample of children with elevated BLL revealed no single or predominant exposure source. Instead, there appear to be multiple potential sources, such as fossil-fuel combustion, lead paint (in or around 38% of homes) and household items.

Conclusion

Elevated BLL are common among children in the Visayas, and may signify an under-recognized threat to children living in rural areas of other developing nations. This setting has varied environmental sources of lead. Observed correlates of BLL may be of clinical, environmental and public health utility to identify and mitigate the consequences of lead toxicity.

Introduction

Lead poisoning is one of the most significant environmental health threats children face. Even low levels of lead exposure are associated with impairment of childhood cognitive function¹ and abnormal infant behaviour.²

Over the past 30 years regulatory and environmental reforms in the developed world have significantly ameliorated lead pollution among children. In developing countries, the extent of environmental lead, and ways to reduce and prevent exposures are only beginning to be understood.³ These nations have many competing health challenges and lead poisoning may be considered less important when there is uncertainty about population prevalence rates and the various correlates and sources of lead poisoning. Regulations designed to limit lead pollution are less common and less likely to be enforced compared with developed countries.⁴

The few investigations carried out in developing countries have studied urban populations. These indicate that the problem may be greater than realized. One cohort study of children in Mumbai and Delhi, India, reported a correlation between blood lead levels (BLL) and age, standard of living, height/weight percentile and parity.⁵ Around the Philippines capital Metro Manila, several small investigations found children’s elevated BLL associated with proximity to lead acid battery recycling or repair activities,⁶ consumption of certain foods⁷ and playground soil.⁸ In Karachi, children’s BLL were linked to proximity to the city centre, application of surma (a traditional eye make-up), parental illiteracy, paternal lead exposure at the workplace and hand-to-mouth activity.⁹

The Quality Improvement Demonstration Study (QIDS) is a large population-based study focused on health policy reforms’ impact on the health status and health-care utilization of children under five living in the Visayas. The baseline QIDS survey collected biomarkers, including blood lead, along with demographic, socioeconomic and child health data points. Data were collected from over 2800 children allowing for assessment of BLL in this population and analysis of the correlates.

This paper describes the prevalence of lead poisoning among these children living in a rural area that covers about one third of the Philippines. We explore the correlations of lead toxicity in this population and describe an environmental investigation to characterize an unexpectedly common toxic health hazard.

Methods

This study was approved by the ethics committees at the University of California, San Francisco, and the University of the Philippines, Diliman. The children’s parents or guardians provided written informed consent before enrolment. The sections below detail the methodological approach taken in our investigation’s three phases: description of the prevalence of lead poisoning in this population, analysis of the correlates of BLL and environmental exposure assessment to identify the potential sources of lead.

Study subjects

Children aged between six months and five years were eligible for inclusion in QIDS. A stratified random sample was used: approximately 100 children from each of the 30 identified health districts were interviewed. Half were selected randomly from households using the current National Statistics Office census frame; the other half had been discharged from hospital within the past four to eight weeks. All answered the same questionnaire. Both separate and combined models were run. The discharged patients had an average of 8% lower levels of lead. For the analysis, we use the combined sample controlling for this covariate in the models reported in the results section. Hospital-based children were tested for lead on the day of discharge; the randomly selected children were tested during a home visit. The analysis population is illustrated in Fig. 1.

Fig. 1.

Assembly of analysis datasets

Study variables

Detailed information about socioeconomic and health status was obtained using household survey instruments. These queried household characteristics, expenditure, sources of water, sanitation, labour force participation and education. Health measures included a general (child) health questionnaire, birth history, anthropometrics and blood tests for haemoglobin, folate and lead. Medical technicians collected the household surveys and physical health measures; project leaders monitored data collection methods to ensure quality and uniformity.

Medical technicians obtained whole venous blood samples using standard techniques. Lead samples were analysed at a central laboratory using the LeadCare analyser, approved for use in clinical measurement of blood lead concentration by the United States of America’s Food and Drug Administration in 1997.¹⁰ Previous field work demonstrates good correlation (r = 0.829 among 51 subjects) between values obtained using the LeadCare device and the gold standard measure, atomic absorption spectroscopy (AAS).¹¹ Accordingly, our study carried out AAS analysis on a sample subset to provide overall calibration and adjustments of the reported LeadCare results. Elevated BLL were defined as greater than 10 μg/dl, conforming to current US Centers for Disease Control and Prevention (CDC) and WHO guidelines.^12,13

Analysis of BLL correlates using QIDS database

All children in the project who had complete household survey data and measured BLL were included in an exploratory analysis to determine the correlates. We logarithmically transformed BLL to normalize the distribution (Fig. 2). Stata software (version 7.0) was used to perform linear regression analyses accounting for multiple independent variables. Regression assumptions were used to assess models’ compliance and standard regression diagnostic techniques included an assessment of the impact of outliers; all data points were retained.

Fig. 2.

Measured blood lead concentrations (µg/dl) in the QIDS baseline cohort

Note: Cluster of values at 65 μg/ml represents upper detection limit of LeadCare device.

Environmental exposure assessment

To increase the likelihood of identifying environmental lead sources, we generated a convenience sample of 100 children with the highest BLL in the QIDS cohort – 50 of these lived in two provinces. Of these 50 children, 26 with the highest lead levels were selected for an environmental exposure assessment. We were unable to locate 5 subjects, and replaced them with alternatives from the original list of 50. All the subjects’ guardians agreed to participate in this assessment. This sample frame yielded subjects with initial lead levels that ranged between 35.7 μg/dl and >65 μg/dl.

We used a structured survey to investigate specific exposures that might contain lead. Environmental sources were assessed by performing assays using LeadCheck swabs; these colorimetric markers qualitatively indicate lead presence in suspect materials. Testing protocols for each material were adapted from protocols developed by Hybrivet Systems in conjunction with the CDC.

Household materials tested for lead were: drinking water, soil, household dust (using dust wipes), paint (interior and exterior), candy wrappers, solder from canned foods, any items made of lead, petrol and motor oil. Any environmental samples that tested positive as well as samples of dust, soil and water were sampled quantitatively by AAS analysis. The sample collection protocols were coordinated with a contracted AAS laboratory that followed methods published by the US Department of Housing and Urban Development.¹⁴

Results

Lead toxicity prevalence

A total of 2861 children were tested for BLL. Of these, 601 (21%) had BLL > 10 μg/dl. The cohort’s mean BLL was 6.9 μg/dl; the median concentration was 4.1 μg/dl (Fig. 2). Leyte had the highest mean (8.5 μg/dl) among the 11 provinces included in the study. BLL were significantly lower in Camiguin, Capiz, Negros Occidental and Siquijor.

BLL correlates

Environmental and clinical correlates with elevated BLL – such as roof construction material, water source, blood haemoglobin concentration, blood folate concentration, age, sex, birth weight, history of breastfeeding and of prematurity – were assessed to determine possible associations. We also investigated a variety of socioeconomic factors including household income, yearly and selected household expenditures, source group (population versus hospital-based) and home province (a community-level variable). After adjustment for covariates, our model showed that roof material, water source, haemoglobin, a history of breastfeeding, age, expenditure on cell phones and home province were all associated significantly with increasing BLL (P < 0.05). Age showed a significant quadratic relationship, indicating that lead levels initially increased with age until 30-36 months, then began to decrease (P < 0.001). No other significant nonlinear relationships were found (Table 1, available at: http://www.who.int/bulletin/volumes/85/9/06.036137/en/index.html).

Table 1. Household and individual factors associated with blood-lead levels (BLL).

Characteristics	No. of children	Mean blood level μg/dl		Multiple regression model
				Coefficienta (95% CI)		P-value
Roof construction material
Strong	760	6.3		0.13	0.323	0.323
Light	800	8.1		0.36	0.009	0.009b
Mixed but predominantly strong	633	6.6		0.17	0.206	0.206
Mixed but predominantly light	563	6.6		0.16	0.219	0.219
Mixed but predominantly salvaged	32	6.6		0.54	0.010	0.010c
Salvaged/makeshift	60	4.7		(referent category)
Water source
Own use, tubed/piped well	178	5.9		–0.16	0.052	0.052
Shared, tubed/piped well	649	6.8		–0.10	0.048	0.048c
Dug well	577	6.5		–0.03	0.510	0.510
Spring, river, stream etc.	134	6.9		0.05	0.554	0.554
Rain	20	10.1		0.54	0.039	0.039c
Peddler	9	3.1		–0.17	0.529	0.529
Others	45	9.2		0.17	0.296	0.296
Communal water system	1238	6.9		(referent category)
Blood haemoglobin concentration				–0.04	(–0.07, –0.02)	0.001b
<11 g/dl	672	7.8
11–13 g/dl	1622	6.8
>13 g/dl	523	6.3
Blood folate concentration				–0.0001	(–0.0003, 0.0002)	0.583
<1.20 ng/ml	548	6.6
1.20–1.40 ng/ml	1294	7.2
>2.40 ng/ml	1018	6.3
HOME percentage score				–0.01	(–0,02, –0.01)	<0.00b
Low (<50%)	2503	6.7
Middle (50-75%)	12	7.0
High (>75%)	2	2.2
Age				0.03	0.000	<0.001b
6 months to 1 year	443	5.9
1 year	1061	6.7
2 years	644	7.5
3 years	455	7.4
4 years	250	7.6
Age²				–0.0004	0.001	<0.001b
Sex				–0.09	0.015	0.015c
Female	1330	6.5
Male	1530	7.3		(referent category)
Birth weight				0.00003	(–0.00001, 0.00007)	0.144
<2500 g	419	6.7
2500–3500 g	1293	6.6
>3500 g	475	7.2
Breastfeeding				0.22	(0.09, 0.34)	0.001c
History of breastfeeding	2576	7.1
No breastfeeding	274	5.7		(referent category)
Urbanity				–0.10	(–0.18, –0.02)	0.019c
Rural	1953	7.0
Urban	907	6.8		(referent category)
Income proxy				0.15	(0.03, 0.27)	0.013c
Owns mobile phone	535	7.1
Does not own mobile phone	2303	6.9		(referent category)
Household type				–0.08	(–0.16, –0.002)	0.044c
Follow-home	1425	6.9
Random	1435	6.9		(referent category)

CI, confidence interval.
^a These coefficients indicate percentage changes of the blood-lead level for every unit change of the regressors. A coefficient of 0.10 indicates that 1 unit increase in a continuous explanatory variable results in an estimated 10% increase in BLL. Similarly, a coefficient of 0.10 for a categorical variable indicates that BLL for the associated category is estimated to be 10% higher than that for the reference category. Individual p-values for categorical variables are derived from the t-statistics of individual regression terms.
^b Statistical significance at P = <0.001 level.
^c Statistical significance at P = <0.05 level.

Clinically, we observed that the blood haemoglobin concentration was inversely associated with BLL – 1 g/dl increase in haemoglobin = 3% decrease in BLL (95% confidence interval, CI: 1, 5%; P = 0.043). A history of breastfeeding was associated with a 19% increase in BLL (95% CI: 5, 35%; P = 0.006); females showed an 8% (95% CI: 2, 14%; P = 0.016), marginally significant, decrease in BLL.

Environmental exposure assessment

A structured survey among individuals with very elevated BLL identified multiple possible sources of lead exposure (Table 2, available at: http://www.who.int/bulletin/volumes/85/9/06.036137/en/index.html). For example, all families had potential sources in food and traditional remedies. Tests in households and nearby environments detected lead in soil, paint, household dust, fishing weights, canned fish and fossil fuels. The presence of lead in all the positive field samples was confirmed using the quantitative methods described (Table 3).

Table 2. Reported prevalence of potential lead exposure sources (n=26).

Potential exposure source	Households reporting presence (%)
Water system with lead piping/solder	3 (12)
Soil	26 (100)
Traditional remedies	13 (50)
Canned food	23 (88)
Fish	26 (100)
Wrapped candy	26 (100)
Painted home/other painted buildings	16 (62)
Family lives near road designated as a highway	7 (27)
Family owns vehicle	10 (38)
Family sells petrol	0 (0)
Family has auto batteries in house	1 (4)
Family melts/recycles lead	1 (4)
Someone in family paints	4 (15)
Someone in family makes fishing weights	1 (4)
Occupational exposure: manufacturing	1 (4)
Occupational exposure: mining	0 (0)
Occupational exposure: fertilizers and pesticides	5 (19)

Table 3. Summary of lead content of environmental samples.

Item	Samples tested	Lead concentration	Hazard limita
Soil	2	< 10–93 ppm	420 ppm
Paint chips	10	530–120 000 ppm	5000 ppm
Dust wipes	7	< 10–110 µg/ft²	100 µg/ft²
Canned tuna	2	0.14–0.16 ppm	0.2 ppm
Candy wrapper	1	< 42 ppm	0 ppm
Petrol	2	4.6–24.6 ppm	13 ppm
Motor oil	2	13.8–15.4 ppm	(not regulated)
Fishing weights	3	89–100%	(not regulated)

^a Current hazard limits as determined by applicable United States of America regulatory agencies (Department of Housing and Urban Development, Environmental Protection Agency, Centers for Disease Control and Prevention, Food and Drug Administration and Department of Agriculture) or Philippine regulatory agency (e.g. Department of Environment and Natural Resources for petrol).

Paint

Interior and exterior paint was found in 10 out of 26 households (38%) and tested for lead. All 10 paint chip samples had detectable amounts of lead; 3 (12%) exceeded the 0.5% allowable limit.¹⁴ In nearly all instances, paint was chipped and/or peeling. We also tested two local schools and three hospitals. Lead paint was found on interior and exterior walls and trim in one school, and on walls, trim, cribs and beds in all three hospitals.

Water

The primary source of drinking water tested negative in all 26 households. We learned of widespread upgrades to the water system in the study area with the highest lead levels (Leyte) between initial study visits and follow-up exposure assessment visits.

Soil and dust

Qualitative tests on composite soil samples from each household were all negative. Two control samples were sent for quantitative AAS testing: one was negative, the other had a detectable level of 93 ppm (420 ppm allowable limit).¹⁵ No dust samples tested positive in qualitative sampling. Four of seven dust wipes submitted for quantitative AAS testing had detectable lead levels. The two highest lead-dust concentrations were found in samples obtained from the floor under a child’s bed where fishing weights were stored; both exceeded the recognized hazard threshold.¹⁴

Fishing weights

Four sets of fishing weights were found, and all tested positive (at least 89% by weight) for lead. These were stored under a child’s bed in one household and were melted down for shaping in the kitchen of another.

Canned tuna

Two locally obtained samples of canned tuna were tested. Both contained detectable levels of lead at 0.14 ppm and 0.16 ppm respectively, just below the allowable limit of 0.2 ppm.¹⁶

Fossil fuels

Samples of petrol (one premium, one regular) and motor oil were obtained in Leyte, the province with the highest number of children with elevated BLL. The Philippine Department of Energy has set the maximum allowable limit for lead in petrol at 13 ppm.¹⁷ There is no established maximum for motor oil, specifically 2T motor oil commonly added to the fuel mixture in two-stroke engines, motorcycles and tricycles (motorcycles with sidecars). The premium petrol sample had a lead concentration of 4.6 ppm; regular contained 24.6 ppm. The motor oil samples contained 13.8 ppm and 15.4 ppm of lead respectively. Lead in petrol was first regulated in the Philippines in 1993 and reduced to its current allowable level of 0.13 g/L on 31 December 2000.¹⁷ Government experts believe that residual lead oxide remains in dust, soil and water, especially in areas with high traffic volumes.

Discussion

We report that 21% of 2861 children living in the rural Philippines had elevated levels of lead in whole blood. This is a previously unreported environmental hazard for children living in the Visayas. Smaller studies have found elevated BLL in developing countries, but these were in urban areas where exposure to industrial paint sources and fossil-fuel burning are relatively high.^5,9,18

Elevated lead levels in rural settings have not been reported widely or characterized as a problem for developing countries. Yet the associations between elevated lead levels and socioeconomic factors are different from those in more affluent countries.¹⁹ The combination of elevated lead and high poverty in our population is a particular concern. The average annual income in our sample was US$ 1179; the lowest quintile earned less than US$ 460 (based on direct assessment of 2004 annual expenditures). Constrained public resources in the Philippines, and other developing countries, make it unlikely that surveillance or screening will occur unless awareness is increased and the consequences of lead toxicity are understood. This paper looks at the prevalence, associations and possible causes of lead exposure, but further analyses will be needed to assess the consequences of these exposures.

The inverse gradient between haemoglobin concentration and BLL (Table 1) is another important finding. Some reports suggest that this is the clinical consequence of parallel iron and lead uptake in iron-deficient children.²⁰ Related studies have demonstrated that children with iron deficiency anaemia are at increased risk of developing lead poisoning.²¹ This effect may be mediated through a common absorptive receptor; intestinal iron ligands may also bind lead.²⁰ Another possible mechanism for the observed association is confounding, since both lead poisoning and iron deficiency are associated with a variety of circumstances related to lower socioeconomic status. However, this study found that the relationship persisted despite adjustment for multiple confounders, suggesting that the association is more likely to be biological.

The association between breastfeeding and a significant increase in BLL (7.1 μg/dl mean BLL in breastfed children; 5.7 μg/dl mean BLL with no history of breastfeeding) also may have a biological basis. Previous research has indicated that lead may be transferred via breast milk.²² In sub-group analyses, the association between breastfeeding and BLL was strongest among the youngest children, those most likely to be breastfeeding. This indicates that breastfeeding remains a significant predictor of increased BLL in a population ranging from 6 months to 5 years of age.

Concurrently collected data from the broad-based (QIDS) population survey link roofing material and water source with elevated lead levels. Roofing materials serve as a means of collecting rainwater and thus potential ingestion. Alternatively, both roofing materials and water are proxies for expenditure, which is also significant in our study. Our confirmation testing did not find significant lead levels in rainwater or in any roofing material. However, lead levels in paint, petrol and motor oil were far higher than regulatory standards. The fact that lead levels in the petroleum products were above mandated limits indicates that aerosolized lead is likely to be an important source. This indicates the difficulty of enforcing environmental regulations within the country.

The positive association between lead levels and income proxies, such as roofing and water sources, further suggests that ingested or inhaled lead from paints and fossil fuel consumption may be an important source of the widespread exposure we observed. For example, we know that relatively better-off families are more likely to live along major thoroughfares, use motorcycles and tricycles for transportation, paint their houses and not breastfeed their newborns. For lead exposure this suggests that it is safer to be very poor rather than relatively affluent if you live in the countryside.

This study has important limitations. The cross-sectional nature of the sample limits any causal inferences, and the non-random environmental assay also is a limitation. We are confident there is no single overwhelming source of exposure, but follow-up assessments are needed to identify all potential sources.

Several lead sources identified in this study appear open to intervention. Over a third of the households (as well as public hospitals and schools) contain lead-based paint, often in poor condition. As the sale of lead-based home paint is not regulated in the Philippines, this offers the opportunity for targeted interventions at national and local levels.

The prevalence of lead toxicity could be reported to the public and preventive messages about BLL directed towards at-risk populations. The relationships between BLL and blood haemoglobin concentration or the lack of breastfeeding could be exploited. Our findings and others suggest that iron supplementation may have the double benefit of reducing anaemia and decreasing gastrointestinal lead uptake.²³ At the very least, anaemia should be considered a clinical risk factor for elevated BLL and its presence may warrant screening.

Breast milk is widely considered to be the optimal mode of nutrient delivery to term infants.²⁴ The health benefits for this population clearly outweigh any benefit from lowering BLL by reducing breastfeeding. This association indicates the importance of limiting maternal lead exposure and eventual child exposure through some of the simple measures described.

Lead is an anthropogenic neurotoxicant now dispersed widely around the globe. Its presence in toxic concentrations among child populations seems to closely follow economic and industrial development.²⁵ This study population had minimal direct exposure to industry but extensive exposure to lead-containing products brought into the region. Lead pollution also may be an indicator of wider environmental degradation in rural areas of developing nations. Accordingly, careful environmental, societal and economic assessments of environmental lead exposure are important. Development and implementation of comprehensive strategies are matters for global and local policy.

Lead poisoning is a threat to the potential cognitive development of children living in poor, rural areas. Other threats are malnutrition and chronic poverty, poor hygiene, incomplete access to healthcare and limited educational opportunity. Yet in comparison with such large and complex problems, lead exposures may be relatively straightforward to mitigate, and greater awareness and further investigation are warranted. Successful childhood lead-poisoning prevention programmes in other countries can serve as models for regulating anthropogenic lead pollution and enforcing existing laws. ■

Ethics committee approval

The QIDS study protocol and the lead evaluation were approved by the Committee on Human Research, University of California, San Francisco (approval number H10609-19947-03) and by the University of the Philippines School of Economics (UPEcon) Institutional Review Board (FWA number 000005371). All study participation was strictly voluntary.