Chronic Arsenic Exposure and Risk of Infant Mortality in Two Areas of Chile

Children's Health Articles

Chronic Arsenic Exposure and Risk of Infant Mortality in Two Areas of ChileClaudia Hopenhayn-Rich,1 Steven R. Browning,1 Irva Hertz-Picciotto,1 Catterina Ferreccio,3 Cecilia Peralta,1 andHerman Gibb41Department of Preventive Medicine and Environmental Health, University of Kentucky, Lexington, Kentucky, USA; 2Department ofEpidemiology, School of Public Health, University of North Carolina, Chapel Hill, North Carolina, USA; 3GREDIS/Universidad Cat6licaPontificia, Las Condes, Santiago, Chile; 4National Center for Environmental Assessment, U.S. Environmental Protection Agency,Washington, D.C., USA

Chronic arsenic exposure has been aociated with a range of neurologic, vascular, dermatologic,and carcinogenic effects. However, limited research has been directed at the association of arenicexposure and human reprodu health outcomes. The principal aim of this study was to inves-tigate the trends in infant mortality between two geographic locations in Chile: Antofagsta,which has a well-documented history of arsenic exposure from naturally contaminated water, andValparso, a comparable low-exposure city. The arsenic concentration in Antofagsta's publicdrinking water supply rose substantially in 1958 with the introduction ofa new water source, andremained elevated until 1970. We used a retrospective study design to examine time and locationpatterns in infant mortality between 1950 and 1996, using univariate statistics, graphical tech-niques, and Poisson regression analysis. Results of the study document the general declines in latefetal and infant mortality over the study period in both locations. The data also indicate an eleva-tion of the late fetal, neonatal, and postneonatal mortality rates for Antofagasta, relative toValparaiso, for specific time periods, which generally coincide with the period of highest arsenicconcentration in the ddnking water ofAntofagasta Poisson regression analysis yielded an elevat-ed and significant association between arsenic exposure and late fetal mortality [rate ratio (RR) =1.7; 95% confidence intervl (CI), 1.-1.9], neonatal mortality (RR = 1.53; CI, 1.4-1.7), andpostneonatal mortlity (RR = 1.26; CI, 1.2-1.3) after adjustment for location and calendar time.The findings from this investigation may support a role for arsenic exposure in increasing the riskof late fetal and infant mortality. Key words arsenic, Chile, drinking water, infnt mortality,neonatal death, reproductive effect, stillbirth. Environ Health Perspect 108:667-673 (2000).[Online 6 June 20001http:/fehpnetl. niebs.nih.govldos/2000/ 08p667-673hope nbayn-riclabract.hbtml

Arsenic is a naturally occurring element thatis present in the environment in both organ-ic and inorganic forms. Human exposures tothe more toxic inorganic arsenic compoundsresult from exposures in occupational set-tings, such as metal smelting and pesticideproduction, as well as from medicinal treat-ments and environmental sources (1,4. Theuse of drinking water with elevated arsenicconcentrations, primarily from natural cont-amination, has been the main source ofenvironmental exposures in populationsworldwide, including but not limited tocommunities in Taiwan (3), India (4),Bangladesh (5), Thailand (6), Mexico (7),Chile (8), Argentina (9), China (10), andHungary (11). In the United States, it isestimated that 350,000 people obtain theirdrinking water from sources containing> 50 pg/L arsenic, the current maximumcontaminant level (MCL) set by the U.S.Environmental Protection Agency (EPA)(12). Although higher exposures are morecommon in western states, recent concernshave also focused on other geographicregions where private well use is common. Inareas where arsenic has been more extensive-ly measured, levels are near the EPA MCL(12) or the World Health Organization rec-ommended guidance values of 10 pg/L (e.g.,

Minnesota, New Hampshire, and Michigan)(13-15).

Chronic arsenic exposure at high doseshas neurologic, dermatologic, vascular, andcarcinogenic effects (1,2,16,17). Exposure toarsenic from drinking water increases therisks of skin, lung, and bladder cancers(18-20), and also seems to be associated withdiabetes (21,22).

The possible impact of arsenic on repro-ductive effects has been given less attention,but the collective evidence from human andlaboratory studies suggests the potential foradverse effects on several reproductive endpoints. Studies have reported adverse repro-ductive impacts among the offspring ofemployees and nearby residents of a Swedishcopper smelter where arsenic exposures weredocumented (23-26). Female workers gavebirth to lower weight infants than womenwho resided outside the smelter area, and thedifference was greater if the mothers workedin highly exposed jobs (25). An incrementaltrend in the rates of spontaneous abortionswas observed with increasing occupationaland residential exposure (24,25). Congenitalmalformations appeared to be more frequentif the mother was employed in highlyexposed jobs during pregnancy (26).In Bulgaria, the incidence of toxemia of

pregnancy and the mortality from congenitalmalformations were significantly higher thanthe national rates in an area near a smelterwith environmental contamination fromvarious metals (27). A study in Texas foundan increase in the rates of stillbirths in rela-tion to residential exposures from an arsenicpesticide factory (28). Although arsenicexposures were documented in all of thesestudies on reproductive effects, confoundingfrom other metals or from other potentialrisk factors could not be excluded.

Studies of populations exposed to arsenicfrom drinking water have found increasedrates of spontaneous abortions and stillbirthsin Hungary (11) and Argentina (29). In theUnited States, three studies reported adversereproductive effects, including increases inmortality from congenital cardiovascularanomalies (30,31) and spontaneous abor-tions (34. Exposed groups from these stud-ies had arsenic levels that were quite low,making the results difficult to interpret.

Given the methodologic limitations ofthe current epidemiologic evidence, the liter-ature on arsenic and adverse reproductiveoutcomes is suggestive, but not conclusive.However, the teratogenicity of arsenic in ani-mals is well documented and generally con-sistent, showing that arsenic induces neuraltube defects as well as malformations of othersystems (33-36). The effects are dependenton dose, route, and timing of administration(35). Placental transfer of inorganic arsenicoccurs in both animals and humans (37-44.A recent study in an area of Argentina with

Address correspondence to C. Hopenhayn-Rich,Department of Preventive Medicine andEnvironmental Health, University of Kentucky,1141 Red Mile Road, Suite 201, Lexington, KY40504 USA. Telephone: (859) 257-6569. Fax:(859) 323-1038. E-mail: [email protected] thank A. Dmitrienko, A. G6mez-Caminero,

B. Huang, K. Johnson, S. Novak, D. Remley, andM. Stump for their help in preparing this paper.This work was partially funded by a grant from

the Andrew W. Mellon Foundation to the CarolinaPopulation Center, and by the National Center forEnvironmental Assessment of the U.S. EPA. Workwas performed at Department of PreventiveMedicine and Environmental Health, University ofKentucky.The views communicated in this manuscript are

solely the opinions of the authors and should notbe inferred to represent those of the U.S. EPA.Received 6 October 1999; accepted 10 February

2000.

Environmental Health Perspectives * VOLUME 108 1 NUMBER 7 July 2000 667

Children's Health * Hopenhayn-Rich et al.

high arsenic in drinking water (250 )ig/L)found a close relationship between placentaland cord blood arsenic levels, indicating con-siderable placental transfer of arsenic to thedeveloping fetus during pregnancy (43).

Several areas of northern Chile have hada history of arsenic exposure. The surfacewater that supplies most of the towns and vil-lages in the Atacama Desert region comesfrom rivers originating in the Andes moun-tains. Antofagasta, the largest city in this area,with a current population of approximately250,000 people, has had a unique pattern ofarsenic exposure that has been well describedand which warrants further epidemiologicinvestigation. In 1958, because of insuffi-cient water supply to serve the growing pop-ulation and the decrease in water availability,water from the Toconce River was intro-duced as the new main water source. Theriver was later found to have arsenic concen-trations of approximately 800 pg/L. The firstreported cases of chronic arsenic healtheffects in Antofagasta appeared in 1962 (44),and a number of publications followed(8,20,45,46). After 12 years of high arsenicexposure, an arsenic-removal plant wasinstalled in 1970 at the public water supplycompany, which led to a subsequent decreasein arsenic concentrations.

The purpose of this study was to com-pare the rates of stillbirth and infant mortali-ty over time between Antofagasta, an areawith historically high water arsenic levels,and Valparaiso, a comparison communitywith low arsenic concentrations in drinkingwater. We studied 47 years spanning from1950 to 1996. This study period allows us toexamine changes that may be related to thesharp increase in arsenic levels found in thedrinking water of Antofagasta starting in1958 and to the subsequent decrease startingin 1970. We also examined the general timetrends in infant mortality of these two majorChilean cities over almost half a century.

Materials and Methods

Stdy design andpopulation. This study useda retrospective ecologic design, incorporating

secondary vital statistics data and environ-mental measurement data of arsenic levels inthe public water supplies of Antofagasta andValparaiso. Our primary interest was to com-

pare infant mortality rates over time betweenAntofagasta, an area with historic high water

arsenic levels, and Valparaiso, a comparison

community with no historical evidence ofhigh arsenic water contamination. Both citiesare major Chilean ocean ports, similar in sizeand other sociodemographic characteristics(Table 1). Their locations are shown inFigure 1.

Vital statistics data. We obtained vitalstatistics data from the Instituto Nacional deEstadisticas (INE) in Santiago, Chile, whichcentralizes both local and national vital sta-

tistics and census information. Natality andmortality data were obtained for the period1950 to 1996 from INE annual reports. TheINE infant mortality data are classified sepa-

rately as late fetal deaths (over 28 weeks ofgestation, also commonly referred to as still-births) and infant deaths, dichotomized intocategories of infants younger than 28 days ofage (neonatal) and those 28 days to 1 year ofage (postneonatal).

The birth and mortality data were report-

ed by geographic locations. Since 1976, Chilehas been stratified into regions numbered1-12 from north to south. The regions havebeen further divided into provinces, and theninto comunas (equivalent to counties in theUnited States). Before 1976, Chile was divid-ed into 25 provinces, and throughout theyears, changes in both the number and juris-

diction of geographic boundaries occurred.An in-depth reclassification of geographicareas developed common units of analysis toachieve compatibility across time for allregions of Chile, using the comuna as the

smallest unit of analysis (52). The majority ofthe population in the comunas ofAntofagastaand Valparai'so live in cities bearing the samenames (99 and 97%, respectively), thus mak-ing the comuna and city almost the same.

Births and deaths are reported by bothplace of occurrence and place of maternalresidence. Because our interest is in the rela-tionship between infant mortality andarsenic exposure from drinking water (eitherthrough maternal or infant ingestion), weused the location of maternal residence forcoding the birth and death data.

Environmental exposure data. Weobtained the average levels of arsenic frompublic water supplies for the period1950-1996 from summarized data collectedfrom local water company records and the

Table 1. Demographic characteristics of Valparaiso and Antofagasta.a

Valparaiso AntofagastaYear 1950 1970 1982 1992 1950 1970 1982 1992Total population of county 424,799 254,812 272,520 283,000 165,005 127,967 186,341 228,000or provinceb

Number of persons per 4.9 - 4.0 3.9 4.8 - 4.4 4.3household (average)

Population served by public 80.9 81.0 94.5 97.3 60.9 88.4 93.0 97.9water supply (%)

Population with electricity (%) 81.7 - - 98.0 78.2 - - 98.6Educational status (%)

Basic (grades 1-7) 82.4 67.0 55.0 47.2 73.2 64.7 53.9 45.3Middle (grades 8-12) 16.4 22.2 32.9 40.0 24.2 21.8 30.5 38.3High (postsecondary) 1.2 3.0 5.9 10.5 2.6 4.7 9.8 13.7

Poverty levelIndigent (%) - - - 4.7 - - - 4.2Poor, not indigent(%) - - - 17.5 - - - 12.4

-, data not available in the census reports reviewed by authors.'Data from Census reports (47-50). b1970, 1982, and 1992 data for county; 1950 data for province.

rigure 1. Map o0 unile, incluaing stuay locationsof Antofagasta and Valparaiso. Adapted fromMaps of the Americas (51).

VOLUME 108 1 NUMBER 7 July 2000 * Environmental Health Perspectives668

Children's Health * Chronic arsenic exposure and infant mortality

regional health service (53) (Table 2). Before1958, arsenic levels in the water of Anto-fagasta averaged 90 pg/L. After the introduc-tion of the Toconce River as the main watersource, the mean arsenic levels during theperiod 1958-1970 rose to approximately 860pg/L. In 1970, the city installed an arsenic-removal plant for this water supply. Resultsof subsequent measurements indicated adecline in the arsenic levels over the next 26years, with 1996 levels close to 50 ,ug/L.

In contrast, Valparaiso has no recordedevidence of elevated arsenic concentrationsin the drinking water. Recent water surveysfor Valparaiso, conducted by the ChileanNational Environmental Commission(CONAMA) (54) and by our group, showlow arsenic levels (< 5 pg/L). Data collectedfrom water companies from 1990 to 1994and summarized in the CONAMA study(54) also show that arsenic water levels inValparaiso were below the analytical detectionlimit (20 pg/L). Although we do not havedata before 1990, there is no reason to believethat Valparaiso had past exposures to arsenicfrom drinking water. The natural higharsenic contamination of water from geologicsources is specific to areas in northern Chile.Consequently, no major populations have thehistory of high exposures to arsenic fromdrinking water experienced by Region II,which includes Antofagasta (20). For the pur-pose of this study, we assumed that Valparaisohad low drinking water arsenic concentrationsover the time period of study.

Statistical analysis. We used univariatestatistics and graphical techniques to calcu-late and examine the time trends in infantmortality for Antofagasta and Valparaisoover the study period. We calculated annualinfant mortality rates by dividing the num-ber of deaths by the number of live birthsper location and multiplying by 1,000. Forlate fetal mortality, we divided the numberof fatalities in each year by the number oflive births plus late fetal deaths to obtain thedeath rate per total births. After calculatingyearly mortality rates for each group, wenoted considerable variation from year toyear, given the small number of annualdeaths at each study location. Therefore, wegrouped the rates into 4-year periods (exceptfor the last period, for which we only had 3

Table 2. Average drinking water arsenic levels inAntofagasta.a

Years Concentration (pg/L)1950-1957 901958-1970 8601971-1979 1101980-1987 701988-1 996 40

aData represent an average of existing arsenic water

years of data) to improve the stability of themortality rates while maintaining a distinct

time period of high arsenic exposure inAntofagasta.

For 3 years (1968-1970), the INE infantmortality data were reported as late fetaldeaths and live-born infants who died beforethe end of their first year, without subdivid-ing the latter into neonatal and postneonatalgroups. For these 3 years, we imputed thenumber of neonatal and postneonatal deathsby linear interpolation. We calculated theratio of neonatal deaths to the total numberof infant deaths for each of the three previ-ous years (1965-1967) and the three subse-quent years (1971-1973). We used the slopeof the fitted line through these six datapoints to estimate the proportion of neonataldeaths for the missing years, and multipliedit by the total number of deaths of infantsyounger than 1 year of age to obtain theneonatal deaths. We used the same methodto estimate the number of postneonataldeaths. Rate differences were calculatedbetween Antofagasta and Valparalso for thethree mortality outcomes for each 4-yearperiod. We used rate differences to estimate

the number of excess late fetal, neonatal, andpostneonatal deaths that could be associated,at least in part, with the increase in arsenicexposure from drinking water in Antofagastaover the peak exposure years.

Poisson regression analysis. We usedPoisson regression analysis to fit the mortalityrates (late fetal, neonatal, and postneonatal) as

a function of the estimated exposure toarsenic by log-linear regression models whileadjusting for location and calendar time.Predictor variables in the model includedlocation (Antofagasta or Valparalso), calendartime (by 4-year intervals beginning in 1950),and arsenic exposure, entered into the modeleither as a continuous or dichotomous expo-

sure variable. The inclusion of city in themodels enabled us to control for differencesother than arsenic that can explain variationsin infant mortality, such as socioeconomic or

health-care-related factors. For arsenic as a

continuous variable, we used available histori-cal arsenic levels in drinking water supplies to

estimate averages for each 4-year period ineach location. For the low exposures inValparaiso, we assumed arsenic water levels of5 pg/L. The dichotomous variable (present/absent) was created to serve as an exposure

indicator only for Antofagasta during theyears of high arsenic exposure (1958-1970).Because arsenic levels decreased after theinstallation of the arsenic-removal plant in

March 1970, we only considered three high-arsenic year groups: 1958-1961, 1962-1965,and 1966-1969. For all other time periodsand locations, this indicator variable was

coded as a zero (absent). City, calendar time,

and one of the arsenic exposure variables were

included in the Poisson regression models andgoodness-of-fit was evaluated using graphicalapproaches plotting observed and expectedrates over calendar time.

We calculated adjusted rate ratios (RRs)and the associated 95% confidence intervals(CIs) from the parameters estimated by themodel. The calendar time period 1974-1977served as the referent period for the model,chosen because it was after the end of thehigh arsenic period and quite close to themid point of the entire study time. Theanalysis was performed using the PROCGENMOD procedure provided in the SASsoftware (55). The model was fit with thePoisson distribution with the link functionas the log and the offset as the log of theappropriate denominator (i.e., total births).The results obtained using the arsenic expo-sure variable as either a continuous or as a

categorical variable were not materially dif-ferent; therefore, the results are shown forarsenic as a categorical variable.

ResultsDemographic data for Valparalso andAntofagasta representing several decadeswere obtained from INE census reports(47-50) and are presented in Table 1.These data indicate that Antofagasta andValparaiso are comparable on a number ofcharacteristics, although these indicatorshave changed over time.

Infant and late fetal mortality rates

declined markedly in Chile during the studyperiod of 1950-1996, as illustrated in Figure2. This decrease is characteristic of most

Latin American countries; it follows the gen-eral trend observed in industrialized nations

and is largely due to improvements in livingconditions and health care (56). The rate ofdecline up to the 1980s was the most pro-nounced for postneonatal mortality, withmore gradual declines evident for neonataland late fetal mortality.

Figures 3-5 show secular trends in latefetal, neonatal, and postneonatal mortalityrates, respectively, by 4-year intervals, forAntofagasta and Valparaiso. Between 1950and 1996, Antofagasta experienced an 86%decline in the late fetal mortality rate, an81% decline in the neonatal mortality rate,and a 92% decline in the postneonatal mor-

tality rate. Similarly, the declines in infantmortality rates for Valparaiso were 64, 77,and 92%, respectively. It is evident thatAntofagasta experienced a more substantialdecline in the late fetal mortality rate overthis calendar period, whereas the relativedecreases in neonatal and postneonatal mor-tality rates were similar in both locations.Despite the overall decline, rates for all out-comes increased in Antofagasta during the

Environmental Health Perspectives * VOLUME 108 1 NUMBER 7 July 2000

measurements, as presented by Pedreros (53).

669

Children's Health * Hopenhayn-Rich et al.

calendar period 1958-1961 and declinedthereafter. The increases and declines vary byoutcome, but overall they coincide with theperiod of higher arsenic levels in the drink-ing water supply of Antofagasta. Table 3shows the comparison of rates between thetwo areas. During the period of maximumdifference in rates between the two commu-nities (1958-1961), Antofagasta had anexcess rate of approximately 20 late fetaldeaths, 24 neonatal deaths, and 24 post-neonatal deaths per 1,000 births (Table 3).The total estimated excess of 68 deaths perthousand births is equivalent to approximate-ly 770 of the 1,900 combined late fetal andinfant deaths which occurred during thattime period. Subsequent to 1974, mortalityrates exhibit a more gradual decline and arevery similar between the two communities.

Results from the Poisson regressionanalysis of the rates of late fetal, neonatal,and postneonatal mortality are given inTable 4. We observed significantly elevatedRRs for high arsenic exposure in association

100 _ 2

869:5

Ra 7ub. 0.a.4'

1t

U.I

with each of the three mortality outcomes,after adjustment for location and calendartime. The association between arsenic expo-sure and late fetal mortality exhibited thestrongest association (RR = 1.72; CI,1.54-1.93), with neonatal mortality (RR =1.53; CI, 1.40-1.66), and postneonatalmortality (RR = 1.26; CI, 1.18-1.34) alsoyielding elevated RRs, respectively. The vari-able for location demonstrated a significantassociation with each outcome, although themagnitude was less than that betweenarsenic exposure and each outcome. Therate ratios for the 4-year calendar groupsexhibited the predictable secular declinesover time.

DiscussionThe results of this study indicate that expo-sure to inorganic arsenic from public watersupplies may be associated with an increasedrisk of infant mortality. Specifically, the datasuggest that arsenic exposure may represent agreater risk for late fetal mortality with a

lower, but still elevated, risk for neonatal andpostneonatal mortality. This association wasevident after controlling for geographiclocation and calendar time in Poissonregression models.

The mortality rates in both Antofagastaand Valparaiso reflect the well-establisheddecline in late fetal, neonatal, and post-neonatal mortality that occurred in Chileover the study period (57. Improvement inthe standard of living, development of pro-grams for maternal and infant care (prenatal,nutritional supplementation, and healtheducation, among others), and the decreasein birth rate are considered the primary fac-tors accounting for the substantial improve-ment in infant mortality rates during thisperiod (56,58,59). The steeper decline inpostneonatal mortality supports these chang-ing patterns of infant mortality because diar-rheal infections and other childhood diseasesthat are much less fatal under improvedhealth and living conditions are commonthroughout the first year of life.

45

to m

t:

10.a. '.

* isi 11

553 54-7 58-61 62-65 6669 70-73 74-77 76-61 8245 86-9 9093 94-6

Figure 2. Late fetal, neonatal, and postnatal mortality rates for Chile, by 4-yearperiods from 1950 to 1996.

50~.

e 40

20 10 ~*W- I ~ .~ . ~ ~ 4

U. - '-

10 _

5553 54-57 5861 62-65 6 70-73 74-77 76881 82-45 869 98-93 946

YearsFigure 3. Late fetal mortality rates for Antofagasta and Valparaiso, by 4-yearperiods from 1950 to 1996.

50-53 54-57 5861 62-65 6669 70-73 74-7 7881 82-65 869 93 996

Figure 4. Neonatal mortality rates for Antofagasta and Valparaiso, by 4-yearperiods from 1950 to 1996.

90 A*?~ -~0 Antfda.a1 .E

C@D

..........~

* 9 ........ ... .

U *_

51-53 5457 5661 82-45 66-9 70-73 74-77 7881 8285 8-89 90-93

Figure 5. Postneonatal mortality rates for Antofagasta and Valparaiso, by 4-year periods from 1950 to 1996.

VOLUME 108 1 NUMBER 7 1 July 2000 * Environmental Health Perspectives670

Children's Health * Chronic arsenic exposure and infant mortality

Table 3. Mortality rates and rate differences for Antofagasta and Valparaiso, by 4-year periods.a

Late fetal Neonatal PostneonatalYears ANTOF VALP RD Cl ANTOF VALP RD Cl ANTOF VALP RD Cl1950-1953 32.2 15.9 16.3 15.4-17.3 38.4 34.7 3.7 3.0-4.4 69.8 78.4 -8.6 (7.8-9.4)1954-1957 25.9 9.3 16.7 15.8-17.4 39.0 28.3 10.7 9.9-11.5 74.8 65.5 9.2 8.5-10.11958-1961 29.3 9.6 19.7 18.9-20.5 45.7 22.1 23.6 22.7-24.5 93.0 69.3 23.7 22.8-24.61962-1965 28.6 12.3 16.4 15.6-17.0 43.2 25.2 18.0 17.2-18.8 73.1 55.4 17.6 16.8-18.61966-1969 25.2 11.0 14.3 13.5-14.9 33.5 23.2 10.3 9.6-11.0 62.1 45.8 16.3 15.4-17.21970-1973 19.4 8.3 11.1 10.5-11.7 23.8 23.1 0.7 0.2-1.3 31.0 26.9 4.0 3.5-4.71974-1977 9.1 10.9 -1.7 (1.3-2.3) 20.9 21.3 -0.5 (0.1-0.9) 30.5 22.2 8.4 7.6-9.01978-1981 5.3 5.9 -0.5 (0.3-0.9) 16.5 14.1 2.3 1 .9-2.9 13.0 13.5 -0.5 (0.03-1.0)1982-1985 5.4 3.8 1.6 1.3-1.9 13.2 9.9 3.3 2.8-3.8 11.2 10.1 1.1 0.7-1.51986-1989 5.4 6.5 -1.0 (0.7-1.5) 11.4 9.3 2.2 1.7-2.5 7.1 9.3 -2.2 (1.8-2.6)1990-1993 4.3 6.3 -2.1 (1 .6-2.4) 8.5 8.0 0.5 0.1-0.9 7.0 6.7 0.3 0.04-0.61994-1996 4.5 5.8 -1.2 (0.9-1.7) 7.3 8.0 -0.7 (0.3-1.2) 5.4 6.1 -0.8 (0.3-1.11

Abbreviations: ANTOF, Antofagasta; VALP, Valparaiso; RD, rate difference.&Mortality rates were calculated per 1,000 live births for neonatal and postneonatal period, and per 1,000 total births (live + late fetal) for the late fetal period.

Although both Antofagasta and Val-paraiso experienced substantial improve-ments in their overall infant mortality ratesover the study period, the pattern of infantmortality for these two communities wasquite different before the 1970s. Antofagastaexhibited marked increases in late fetal,neonatal, and postneonatal mortality in1958-1961. As compared to Valparaiso, theserates remained elevated until 1974-1977 forlate fetal mortality, and until 1970-1973 forneonatal and postneonatal mortality. Thepattern of increase and subsequent decline inAntofagasta's mortality rates show a closetemporal relationship with the rather suddenand sharp rise in the levels of arsenic in thecity's public water supply from 1958 untilMarch 1970, when an arsenic-removal facili-ty was installed, and suggest a possible rolefor inorganic arsenic in the observed increasein infant mortality.

As in the rest of the country, the declinein mortality rates in Antofagasta during theoverall study period was likely a result ofother nonarsenic-related factors such asimprovement in health care and standard ofliving, which affected all of the regions inChile. Moreover, starting in the 1960s,Region II, which includes the city of Anto-fagasta, experienced a surge of economicexpansion, with a rate of growth and devel-opment greatly surpassing that of the rest ofthe country: the increase in adjusted annualper-capita income was greater by far inRegion II than any other region (60). Thisexpansion brought improvements in region-al health care as well, which may explain, atleast in part, the fact that in the periodbefore 1958 Antofagasta had higher latefetal and infant mortality rates than Val-paraiso, whereas after 1970 the ratesdeclined to comparable levels between thetwo towns. It appears that the increase inarsenic in the water supply served to delaythe improvement in fetal and infant healthuntil the arsenic-removal plant was installed(Figures 3-5).

The results of this study should be con-sidered within the context of limitations inthe study design and in the type of dataobtained. In general, ecologic studies are sus-ceptible to biases that pertain to the lack ofindividual data on exposure, outcomes, andconfounders. These biases, known as theecologic fallacy, have been discussed in detailelsewhere (61). In this study, although expo-sure was measured at the group level, publicwater was consumed by most residents ineach of the communities, the water supplycame from one main source at each location,and we adjusted for community differencesin the model. However, we lacked data onindividual-level confounders, and thereforewe cannot definitely exclude the possibilityof an ecologic bias. Nevertheless, the distincttemporal pattern of infant mortality rates inAntofagasta as compared to Valparaisoargues against this bias. That is, residual con-founding from factors not available wouldhave to closely relate to the timing of thehigh arsenic exposure documented inAntofagasta. The change in the arsenic level

of the water supply was an indisputableevent, and no other environmental pollutanthas been described to account for the manydocumented health effects experienced bythe population of Antofagasta during thestudy time period (8,20,44,46,62-65).Thus, it would not be surprising if thesereproductive events were also due to the con-sumption of water with high arsenic.

The underascertainment of both birthand death registrations is typically a poten-tial concern of studies of this design. Withregard to birth registration, there is evidenceof underascertainment, which has substan-tially declined over time. In Chile, childrenwhose births are not registered by March ofthe year after they were born are categorizedseparately. A study of birth registrationsfrom 1955-1988 showed variations rangingfrom 11.4 to 4.2% in late birth registrations(66). We also observed variations betweenregions; Antofagasta and Valparaiso wereconsistently among those with the lowestlate registration rates. The increase in birthsthat occur in hospitals (42.3% in 1953 vs.

Table 4. Poisson regression analysis for secular trends in fetal, neonatal, and postneonatal mortality byarsenic exposure, location, and calendar time.

Fetal mortality Neonatal mortality Postneonatal mortalityVariable RR Cl RR Cl RR ClArsenic exposureNo 1.0 (Referent) 1.0 (Referent) 1.0 (Referent)Yes 1.7 1.5-1.9 1.5 1.4-1.7 1.3 1.2-1.3

LocationValparaiso 1.0 (Referent) 1.0 (Referent) 1.0 (Referent)Antofagasta 1.5 1.4-1.6 1.1 1.1-1.2 1.1 1.0-1.1

Calendar time1950-1953 2.1 1.8-2.3 1.7 1.6-1.9 3.0 2.8-3.31954-1957 1.4 1.2-1.6 1.5 1.4-1.6 2.7 2.5-2.91958-1961 1.2 1.1-1.4 1.2 1.1-1.3 2.8 2.6-3.01962-1965 1.4 1.2-1.6 1.3 1.1-1.4 2.2 2.1-2.41966-1969 1.2 1.1-1.4 1.1 1.0-1.2 1.8 1.7-2.01970-1973 1.2 1.1-1.4 1.1 1.0-1.2 1.1 1.0-1.21974-1977 1.0 (Referent) 1.0 (Referent) 1.0 (Referent)1978-1981 0.6 0.5-0.7 0.7 0.6-0.8 0.5 0.5-0.61982-1 985 0.4 0.4-0.5 0.5 0.5-0.6 0.4 0.4-0.51986-1989 0.6 0.5-0.7 0.5 0.4-0.5 0.3 0.3-0.41990-1993 0.5 0.4-0.6 0.4 0.3-0.4 0.3 0.2-0.31994-1 996 0.5 0.4-0.6 0.4 0.3-0.4 0.2 0.2-0.3

Environmental Health Perspectives * VOLUME 108 1 NUMBER 7 July 2000 671

Children's Health * Hopenhayn-Rich et al.

99.7% in 1994) and births assisted by med-ical professionals (57.5% in 1953 vs. 100%in 1994) also impacts the completeness ofbirth registration. Regional statistics showthat Antofagasta and Valparaiso were two ofthe four provinces with the highest percent-age of physician-assisted births in 1953,compared to the rest of Chile (6/).

Death registration is also characterizedby omissions, especially for newborns thatdied in the first hours or days after birth. Astudy conducted in Santiago in 1968-1969showed that over one-half of the hospital-born babies who died were not registered(68). Another study in Santiago found that13% of the deaths of children younger than5 years of age (69) were not registered, andmost of this underregistration occurred fordeaths in the neonatal period. In Chile, thecertification of infant deaths by a physicianincreased from 55% in 1952 (68) to 96% in1994 (58). The omissions in death reportingand the late birth registrations affect mortali-ty rates, which will be biased up or downdepending on whether birth or death omis-sions were higher. However, given the muchgreater number of births than deaths, under-reporting of deaths will have a much greatereffect on the rates. In a study of 1976-1978mortality by region, medical certificationwas used as an indirect measure of relativeaccuracy in the reporting of death (69,70).Antofagasta and Valparaiso were among thefew provinces with < 5% of deaths lackingmedical certificates (69,70). There is no evi-dence that underreporting was significantlydifferent between the two study areas, butwe did not find any published report withspecific comparisons of fetal and infant mor-tality omissions by geographical locationsacross time.

In addition to the ecologic nature of thestudy and the potential effects of underascer-tainment of births or deaths, our analysis didnot include other contaminants. For exam-ple, differences in exposures to pesticides orother metals could have an effect on infantmortality. Although we do not have specificinformation on pesticides, their use has beenhistorically very limited in the Antofagastaregion because it is a desert area with onlylocal small-scale agriculture. On the contrary,the Valparaiso region is one of the most pro-ductive agricultural areas of the country. Ifpesticide exposure increased infant mortality,it would tend to increase the rates inValparaiso so that the true RRs due to arseniccould be higher than those observed. On theother hand, the Antofagasta region has a his-tory of mining and smelting that could havethe opposite effect. However, most exposedactivities are not close to the city ofAntofagasta. Most importantly, it is unlikelythat the clear sharp increase in arsenic water

levels, followed by the decrease 12 years laterwith the installation of the arsenic-removalplant, would be accompanied by anothercontaminant with a similar temporal pattern.

Summary andRecommendationsAlthough there is suggestive evidence forarsenic-related human developmental toxici-ty, the existing literature falls short of estab-lishing a clear causal association betweenenvironmental arsenic exposure and repro-ductive health effects. The findings presentedin this paper, albeit not definitive, support arole for arsenic in the increased late fetal andinfant mortality observed in Antofagasta.Further studies are needed to investigate thebroad range of reproductive and develop-mental effects, ranging from low birth weightto infant death, that may be causally relatedto arsenic in environmentally exposed popu-lations around the world. The study designsof these investigations should consider usingdefined populations, with individual-leveldata and well-delineated methods of exposureassessment for arsenic and collection of datafor potential confounding variables.

Results from the present study, theprospective cohort study under way in Chileby our group, and additional studies of thisnature are important in the context of theongoing review of the arsenic drinkingwater standard.

REFERENCES AND NOTES

1. IARC. Arsenic and arsenic compounds. IARC MonogrEval Carcinog Risk Hum 23:39-141 (1980).

2. ATSDR. Toxicological Profile for Arsenic, Update. Rpt NoTP-92V102. Atlanta, GA:Agency for Toxic Substances andDisease Registry, 1993.

3. Tseng W, Chu HM, How SW, Fong JM, Lin CS Yeh S.Prevalence of skin cancer in an endemic area of chronicarsenicism in Taiwan. J NatI Cancer Inst 40:453-463(1968).

4. Mazumder DNG, Das Gupta J, Santra A, Pal A, Ghose A,Sarkar S. Chronic arsenic toxicity in West Bengal-theworst calamity in the world. J Ind Med Assoc 96:4-7,18(1998).

5. Chowdhury AM, Jakariya M. Testing of water for arsenicin Bangladesh [Letter]. Science 284:1622 (1999).

6. Williams M, Fordyce F, Paijitprapapon A, CharoenchaisriP. Arsenic contamination in surface drainage andgroundwater in part of the southeast Asian tin belt,Nakhon Si Thammarat Province, southern Thailand.Environ Geol 47:16-33 (1996).

7. Cebrian ME, Albores A, Aguilar M, Blakely E. Chronicarsenic poisoning in the North of Mexico. Hum Toxicol2:121-133 (1983).

8. Borgono JM, Vincent P, Venturino H, Infante A. Arsenicin the drinking water of the city of Antofagasta: epidemi-ological and clinical study before and after the installa-tion of a treatment plant. Environ Health Perspect19:103-105 (1977).

9. Biagini RE. Hidroarsenicismo cronico en la RepublicaArgentina. Med Cut ILA 6:423-432 (1975).

10. Luo ZD, Zhang YM, Ma 1, Zhang GY, He X, Wilson R, ByrdDM, Griffiths JG, Lai S, He L, et al. Chronic arsenicismand cancer in Inner Mongolia-consequences of well-water arsenic levels greater than 50 micrograms perliter. In: Arsenic Exposure and Health Effects (AbernathyCD, Calderson RL, Chappell WR, eds). London:Chapmanand Hall, 1997;55-68.

11. Borzsonyi M, Bereczky A, Rudnai P, Csanady M, HorvathA. Epidemiological studies on human subjects exposedto arsenic in drinking water in Southwest Hungary. ArchToxicol 66:77-78 (1992).

12. Science Applications International Corporation.Estimated National Occurrence and Exposure to Arsenicin Public Drinking Water Supplies. Revised Draft pre-pared for U.S. Environmental Protection Agency undercontract 68-01-766. Washington, DC:U.S. EnvironmentalProtection Agency, 1987.

13. Small-Johnson J, Soule R, Durking D, MinnesotaDepartment of Health. Human exposure and healtheffects associated with low-level arsenic in drinkingwater [Abstract]. Epidemiology 9(4):S152 (1998).

14. Karagas MR, Tosteson TD, Blum J, Morris JS, Baron JA,Klaue B. Design of an epidemiologic study of drinkingwater arsenic exposure and skin and bladder cancer riskin a U.S. population. Environ Health Perspect 106(suppl4):1047-1050 (1998).

15. Michigan Department of Public Health. Arsenic inDrinking Water: A Study of Exposure and a ClinicalSurvey. Lansing, MI:Michigan Department of Health,1982.

16. Wu MM, Kuo TL, Hwang YH, Chen CJ. Dose-responserelation between arsenic concentration in well waterand mortality from cancers and vascular diseases. Am JEpidemiol 130:1123-1132 (1989).

17. Chen CJ, Chen CW, Wu MM, Kuo TL. Cancer potential inliver, lung, bladder and kidney due to ingested inorganicarsenic in drinking water. Br J Cancer 66:888-892 (1992).

18. Chen CJ, Kuo TL, Wu MM. Arsenic and cancers [Letter].Lancet 1(8582):414-415 (1988).

19. Hopenhayn-Rich C, Biggs ML, Smith AH. Lung and kidneycancer mortality associated with arsenic in drinking waterin C6rdoba, Argentina. Int J Epidemiol 27:561-569 (1998).

20. Smith AH, Goycolea M, Haque R, Biggs ML. Markedincrease in bladder and lung cancer mortality in a regionof northern Chile due to arsenic in drinking water. Am JEpidemiol 147:660-669 (1998).

21. Shibata A, Ohneseit PF, Tsai YC, Spruck CH, Nichols PW,Chiang H, Lai M, Jones PA. Mutational spectrum in the p53gene in bladder tumors from the endemic area of blackfoot disease in Taiwan. Carcinogenesis 15:1085-1087(1994).

22. Rahman M, Tondel M, Ahmad SA, Axelson 0. Diabetesmellitus associated with arsenic exposure in Bangladesh.Am J Epidemiol 148:198-203 (1998).

23. Nordstrom S, Beckman L, Nordenson I. Occupationaland environmental risk in and around a smelter in north-ern Sweden. I: Variations in birth weight. Hereditas88:43-46 (1978).

24. Nordstrom S, Beckman L, Nordenson I. Occupationaland environmental risk in and around a smelter in north-ern Sweden. Ill: Frequencies of spontaneous abortion.Hereditas 88:51-54 (1978).

25. Nordstrom S, Beckman L, Nordenson I. Occupationaland environmental risks in and around a smelter innorthern Sweden. VI: Congenital malformations.Hereditas 90:297-302 (1979).

26. Nordstrom S, Beckman L, Nordenson I. Occupationaland environmental risk in and around a smelter in north-ern Sweden. V: Spontaneous abortion among femaleemployees and decreased birth weight in their offspring.Hereditas 90:291-296 (1979).

27. Zelikoff JT, Bertin JE, Burbacher TM, Hunter ES, MillerRK, Silbergeld EK, Tabacova S, Rogers JM. Health risksassociated with prenatal metal exposure. Fundam AppIToxicol 25:161-170 (1995).

28. lhrig MM, Shalat SL, Baynes C. A hospital-based case-control study of stillbirths and environmental exposure toarsenic using an atmospheric dispersion model linked toa geographical information system. Epidemiology9:290-294 (1998).

29. Castro JA. Efectos carcinogenicos, mutagenicos y ter-atogenicos del arsenico. Acta Bioquim Clin Latinoam16:3-17 (1982).

30. Engel RE, Smith AH. Arsenic in drinking water and mor-tality from vascular disease: an ecological analysis in 30counties in the United States. Arch Environ Health49:418-427 (1994).

31. Zierler S, Theodore M, Cohen A, Rothman KJ. Chemicalquality of maternal drinking water and congenital heartdisease. IntJ Epidemiol 17:589-594(1988).

672 VOLUME 1081 NUMBER 71 July 2000 * Environmental Health Perspectives

Children's Health * Chronic arsenic exposure and infant mortality

32. Aschengrau A, Zierler S, Cohen A. Quality of communitydrinking water and the occurrence of spontaneous abor-tion. Arch Environ Health 44:283-290 (1989).

33. Golub MS. Maternal toxicity and the identification ofinorganic arsenic as a developmental toxicant. ReprodToxicol 8:283-295 (1994).

34. Shalat SL, Walker DB, Finnell RH. Role of arsenic as areproductive toxin with particular attention to neuraltube defects. J Toxicol Environ Health 48:253-272 (1996).

35. Golub MS, Macintosh MS, Baumrind N. Developmentaland reproductive toxicity of inorganic arsenic: animalstudies and human concerns. J Toxicol Environ Health BCrit Rev 48:199-241 (1998).

36. DeSesso JM, Jacobson CF, Scialli AR, Farr CH, HolsonJF. An assessment of the developmental toxicity of inor-ganic arsenic. Reprod Toxicol 12:385-433 (1998).

37. Squibb KS, Fowler BA. The toxicity of arsenic and its com-pounds. In: Biological and Environmental Effects of Arsenic(Fowler BA, ed). Amsterdam:Elsevier, 1983;233-269.

38. Hood RD, Vedel-Macrander GC, Zaworotko MJ, TatumFM, Meeks RG. Distribution, metabolism, and fetal uptakeof pentavalent arsenic in pregnant mice following oral orintraperitoneal administration. Teratology 35:19-25 (1987).

39. Nicholson WJ, Perkel G, Selikoff IJ. Occupational expo-sure to asbestos: population at risk and projected mor-tality-1980-2030. Am J lnd Med 3:259-311 (1982).

40. Tabacova S, Baird DD, Balabaeva L, Lolova D, Petrov I.Placental arsenic and cadmium in relation to lipid perox-ides and glutathione levels in maternal-infant pairs fromcopper smelter area. Placenta 15:873-881 (1994).

41. Ferm VH. Arsenic as a teratogenic agent. Environ HealthPerspect 19:215-217 (1977).

42. Willhite CC, Ferm VH. Prenatal and development of toxi-cology of arsenicals. In: Nutritional and ToxicologicalAspects of Food Safety (Friedman M, ed). NewYork:Plenum Publishing Corp., 1984;205-228.

43. Concha G, Vogler G, Lexcano D, Nermell B, Vahter M.Exposure to inorganic arsenic metabolites during earlyhuman development. Toxicol Sci 44:185-190 (1998).

44. Zaldivar R. A morbid condition involving cardiovascular,bronchopulmonary, digestive and neural lesions in chil-dren and young adults after dietary arsenic exposure.Zentralbl Bakteriol [Orig B] 179:44-56 (1980).

45. Zaldivar R, Prumes L, Ghai GL. Arsenic dose in patientswith cutaneous carcinomata and hepatic haemangioen-dothelioma after environmental and occupational expo-sure. Arch Toxicol 47:145-154 (1981).

46. Arroyo-Meneses A. Hidroarsenicismo cr6nico en la regi6nde Antofagasta. In: Primera Jornada Sobre ArsenicismoLaboral y Ambiental II Regi6n. Santiago, Chile:AsociacionChilena de Seguridad, 1991;35-47.

47. Instituto Nacional de Estadisticas, Ministerio de Salud.Censo de Poblacion y Vivienda, Chile. Santiago,Chile:lnstituto Nacional de Estadisticas, 1992.

48. Instituto Nacional de Estadisticas, Ministerio de Salud.Censo de Poblacidn y Vivienda, Chile. Santiago,Chile:lnstituto Nacional de Estadisticas, 1982.

49. Instituto Nacional de Estadisticas, Ministerio de Salud.Censo de Poblaci6n y Vivienda, Chile. Santiago,Chile:lnstituto Nacional de Estadisticas, 1970.

50. Instituto Nacional de Estadisticas, Ministerio de Salud.Censo de Poblacion y Vivienda, Chile. Santiago,Chile:lnstituto Nacional de Estadisticas, 1950.

51. Maps of the Americas. The Perry-Castaneda Library MapCollection, The University of Texas at Austin. Available:http://www.lib.utexas.edu/Libs/PCL/Map-collection/americas/Chile_sm97.gif [cited 12 April 20001.

52. Facultad de Ciencias Fisicas y Matemdticas de laUniversidad de Chile. Impacto en salud atribuible aexposici6n a arsenico: un estudio ecologico. In: ProyectoFONDEF 2-24. Santiago, Chile:Universidad de Chile, 1997.

53. Pedreros R. Presencia de Arsenico en 11 Regi6n y EstudioRetrospectivo de Cohortes Expuestas. Segundas Jornadasde Arsenicismo Laboral y Ambiental. Antofagasta,Chile:Servicio de Salud de Antofagasta, 1994.

54. Facultad de Ciencias Fisicas y Matematicas de laUniversidad de Chile. Determinaci6n de linea base dearsenico ambiental. In: Proyecto FONDEF 2-24. Santiago,Chile:Universidad de Chile, 1997.

55. SAS Institute, Inc. SAS/STAT User's Guide. 6th ed. Cary,NC:SAS Institute, Inc., 1990.

56. Bahr J, Wehrhahn R. Life expectancy and infant mortali-ty in Latin America. Soc Sci Med 36:1373-1382 (1993).

57. Zapatero JT, Chavez RT, Torres GO. Mortalidad perinatalen Chile, 1956-1986. Bol Oficina Sanit Panam 106:287-295(1989).

58. Instituto Nacional de Estadisticas, Ministerio de Salud.Anuario de Demografia. Santiago, Chile:Servicio RegistroCivil e Identificaci6n, 1994.

59. Castafeda T. Determinantes del descenso de la mortali-dad infantil en Chile: 1975-1982. Cuad Economia22:195-214 (1985).

60. Corporaci6n de Investigaciones Econ6micas paraLatinoamerica, Ministerio del Interior, Subsecretaria deDesarrollo Regional y Administrativo 1994. Evoluciondel Producto por Regiones 1960-1992, Santiago,Chile:CIEPLAN, 1994.

61. Morgenstern H. Ecologic studies in epidemiology: con-cepts, principles, and methods. Annu Rev Public Health16:61-81 (1995).

62. Zaldivar R. Arsenic contamination of drinking water andfood-stuffs causing endemic chronic poisoning. BeitrPathol 151:384-400 (1974).

63. Zaldivar R, Guillier A. Environmental and clinical investi-gations on endemic chronic arsenic poisoning in infantsand children. Zentralbl Bakteriol lOrig B] 165:226-234(1977).

64. Sandoval H, Venturino H. Contaminaci6n ambiental porarsenico en Chile. Cuad Med 28:30-37 (1987).

65. Rosenberg HG. Systemic arterial disease and chronicarsenicism in infants. Arch Pathol 97:360-365 (1974).

66. Instituto Nacional de Estadisticas y CentroLatinoamericano de Demografia Chile, Estimaci6n de laOportunidad de Inscripci6n de los Nacimientos-TotalPais y Regiones 1955-1988. Santiago, Chile:Ministerio deEconomia, Fomento y Reconstrucci6n, 1990.

67. Instituto Nacional de Estadisticas, Ministerio de Salud.Movimiento Demogrdfico. Santiago, Chile:ServicioRegistro Civil e Identificaci6n, 1953.

68. Legarreta A, Aldea A, Lopez L. Omisi6n del registro dedefunciones de nifnos ocurridas en aternidades, Santiago,Chile. Bol Oficina Sanit Panam 75:308-314 (1973).

69. Rice Puffer R, Serrano CV. Caracteristicas de Ia mortali-dad en la nifiez. In: Scientific Publication No. 262.Washington, DC:Pan American Health Organization,1973;36.

70. Haynes R. The geographical distribution of mortality bycause in Chile. Soc Sci Med 17:355-364 (1983).

-usnu.usEifSfflrUaEnvironmental

m Technical and Toxicity Repoo* Chemical Health and Safety Database * Rodent Historical Control Database

Visit us online!^..............,

http:IIe is*nleFts.ni ov/

,........ Zs .4

l rm xtfr:n~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~er~

Environmental Health Perspectives * VOLUME 1081 NUMBER 71 July 2000 673