Urinary nickel concentrations and selected pregnancy outcomes in delivering women and their newborns among arctic populations of Norway and Russia Jon Øyvind Odland,*a Evert Nieboer,ac Natalya Romanova,b Yngvar Thomassen,b Tor Norsethb and Eiliv Lunda aInstitute of Community Medicine, University of Tromsø, N-9037 Tromsø, Norway bNational Institute of Occupational Health, P.O. Box 8149 Dep, N-0033 Oslo, Norway cDepartment of Biochemistry, McMaster University, Hamilton, Ontario, Canada L8N 32S Received 8th December 1998, Accepted 4th February 1999 The two objectives of this study were to compare urinary nickel excretion in pregnant women and their newborns living in the Murmansk and Arkhangelsk Counties of Russia with that in comparable Norwegian populations living in Finnmark and the city of Bergen and to assess the influence on pregnancy outcome of diVerent risk variables, specifically urinary nickel concentrations and questionnaire-based anamnestic information.Life-style information and urine samples were collected from 50 consecutive mother–infant pairs from hospital delivery departments in three Russian and three Norwegian communities.Pregnancy outcomes were verified from medical records. Urinary nickel excretion was significantly higher in the Russian communities, independent of the presence of a nickel refinery as a local environmental source. The birth weight and mean body mass index of the newborn children (BMIC) were significantly lower ( p<0.001) in the Russian groups, with or without adjustment for gestational age.A multivariate linear regression analysis indicated that maternal urinary nickel concentration had no impact on birth weight. The maternal body mass index (BMI) and maternal height were positive explanatory variables; maternal urinary creatinine is suggested as a weak negative factor. Smoking was shown to be a strong negative predictor only in the Norwegian group among whom there was a significantly higher smoking frequency ( p= 0.005).The significant contribution of a country factor in the predictive model is interpreted to indicate that a number of important risk factors for low birth weight were not identified. in a Russian nickel refinery.4,5 The evidence of teratogenic Introduction and developmental eVects of nickel in animal studies is diYcult Considerable attention is being paid to industrial pollution in to interpret with confidence.6–8 Compared with cadmium,9 the regions adjoining the Norwegian–Russian border, especially lead10 and mercury,11 there is clearly a paucity of data in relation to the substantial emissions of sulfur dioxide and concerning nickel exposure in relation to reproduction and metals by the nickel industry in the Kola Peninsula of Russia.pregnancy outcome. Further, the literature on nickel concen- Nearly half of the tonnage of metals released is nickel, with trations in body fluids during pregnancy is very scarce, and iron and copper other major contributors.1 Extensive environ- the suggested nickel urine reference values12,13 or the results mental damage has resulted from these pollutants and has reported in surveys14–16 do not include pregnant women or generated considerable concern about public health in Russia, newborn children.Norway and Finland. A Joint Norwegian–Russian Commission The present study had the following two specific objectives: on Environmental Cooperation was established in 1988 and (i) to compare urinary nickel excretion in pregnant women encouraged extensive investigation of the pollution of air, water, and their newborns living in the Murmansk and Arkhangelsk soil and biota.2,3 By contrast, very little has been done on the Counties of Russia with that in comparable Norwegian popuinvestigation of human health, even though there has been lations living in eastern Finnmark and the city of Bergen; (ii) considerable focus by the media on reproductive and develop- to assess the influence on pregnancy outcome of diVerent risk mental issues related to nickel exposure.Our interest in the variables, specifically urinary nickel concentrations and quespresent project derived from this public interest and had as its tionnaire-based anamnestic information. objective the determination of whether there is a scientific basis This study was approved by The Regional Ethical for this concern.We used a two-pronged approach: (i) an Committee, University of Tromsø, Norway, the Norwegian assessment of the exposure to nickel of pregnant women selected Data Inspectorate and the Regional Health Administrations from the general public and its relation, together with other of Murmansk and Arkhangelsk Counties.factors, to pregnancy outcome; (ii) an investigation of the reproductive health of female nickel workers. The results of the Materials and methods first component are reported in this study. The reproductive/developmental eVects of nickel have not Study population and procedures been firmly established in humans, although some concern exists about apparent increases in spontaneous abortions and Personal contacts with colleagues in the diVerent hospital structural malformations (especially cardiovascular and mus- delivery departments were established, and all procedures and protocols were provided in Norwegian, English and Russian.culoskeletal ) in newborn babies whose mothers were employed J. Environ. Monit., 1999, 1, 153–161 153The Russian geographic sites in our study were Nikel and in Nikel, 1 in Monchegorsk, 19 in Bergen and 4 in Kirkenes.Even though anamnestic information and maternal samples Monchegorsk in Murmansk County and Arkhangelsk in were collected, the mother and child were excluded from the Arkhangelsk County. Nikel (23 000 inhabitants) and study’s statistical analyses (see below). Monchegorsk (65 000 inhabitants) are communities with nickel In an interview, local midwives or gynaecologists adminis- refining operations.Arkhangelsk is the biggest city on the tered a questionnaire that addressed the following particulars: White Sea, with almost 450 000 inhabitants and five big pulp age, parity, height and weight of mother, ethnic background, and paper plants in the surrounding area, but no metal places of residence exceeding 6 months, schooling, occupation, producing industry.The Norwegian reference cities were smoking habits, alcohol consumption, medication, serious Kirkenes, Hammerfest and Bergen (Fig. 1). Kirkenes (approxidiseases and dietary habits related to local food intake. The mately 4500 inhabitants) is located near the Russian– following information was collected from the delivery depart- Norwegian border, 50 km from Nikel.The delivery department ment medical records about the mothers and births: maternal there receives women from the eastern part of Finnmark (total age, weight and height, Naegele term, date of birth, length population, 28 000), the area geographically adjacent to the and weight of baby, weight of placenta, Apgar score (clinical Russian border.Hammerfest (12 000 inhabitants) is a coastal estimate of the conditions of an infant 1–5 min after birth), city of Finnmark; the hospital delivery department there congenital malformations, gestational age and individual com- receives women from the western part of Finnmark (a total ments by the doctor or midwife.The informed consent form of 45 000 residents), including the main native Saami centres and collection of anamnestic information were completed in Finnmark County. Bergen is the second biggest city of before the delivery process started in order to minimize stress. Norway in the southwest part of the country (total population, Collection of the neonatal urine specimens was performed in 220 000), with no major metal producing industry.It was a manner that avoided interruption of the delivery situation. included because it represents a non-Arctic urban community. Information by questionnaire and urine samples were col- Sample collection and analysis lected from at least 50 consecutive women presenting themselves to the hospital delivery departments in each location; The urine of mothers was sampled at two stages: the first time the first-voided urine from their newborn babies was also at week 20 of pregnancy (only in Kirkenes and Nikel ) and obtained. The registration and sampling were performed in the second time together with blood sampling at 1–2 days the following time periods: Arkhangelsk in April–May 1993; postpartum.Maternal urine was collected directly into a Kirkenes in November 1993–January 1994; Hammerfest in disposable plastic cup for transfer to containers (NØD-0438 December 1993–January 1994; Bergen in June 1994; and Nikel CERBO Norge A/S, Oslo, Norway; volume, 20 mL); both the and Monchegorsk in March–June 1994. Urines were collected cup and the container were tested and found not to contamifrom a total of 265 children; of these 137 were Russian (Nikel, nate the urine samples with detectable amounts of nickel N=42; Monchegorsk, N=49; and Arkhangelsk, N=46) and (<10 nmol L-1).For sampling from the neonate immediately 128 were Norwegian (Kirkenes, N=46, Hammerfest, N=51; postpartum, a uridome (126–0004, Hollister Norge, Oslo, and Bergen, N=31). The women were asked to join the study Norway) was attached by plasters to the child’s genitalia until by means of completing a consent form.the first void was produced. For ethical reasons, we had to Very few cases interrupted the consecutive enrolment. stop the sampling if the uridome irritated the very sensitive Pregnant women in Kirkenes with prepartum complications skin of the newborns. The urine samples were immediately or suspected delivery problems are sent to the Regional frozen at -20 °C.Within 3 months, all the materials collected Hospital in Tromsø. In the sampling period, two transfers were transported frozen to the city of Bodø in Norway for were registered. Hammerfest hospital has a neonatal intensive storage in a -70 °C freezer. care department and no cases were sent to Tromsø before After thawing, the urine samples were heated for 1 h at delivery during the sampling period.None of the delivering 95 °C in an oven to redissolve the urine precipitates and to women refused to join the study. However, the complicated ensure sterility. Urine was analysed without further preurine sampling procedure for neonates sometimes resulted in treatment. Nickel in urine was measured by electrothermal skin irritation, especially if the period before the first voiding atomic absorption spectrometry employing Zeeman-based was long.Consequently, neonatal urine samples could not be Perkin-Elmer Model 5100 PC/HGA-600 and Perkin-Elmer obtained from some of the infants: 4 cases in Arkhangelsk, 8 SIMAA 6000/THGA graphite atomizer systems (Bodenseewerk Perkin-Elmer GmbH, U� berlingen, Germany) calibrated with urine matched standard solutions.17 The accuracy and precision of the measurements were assessed routinely by using human urine quality materials obtained from NycoMed Ltd., Oslo, Norway (Seronorm STE 101021 and 403125).The day-to-day variation of the nickel measurements in these reference materials was typically 10%.The average nickel concentrations of STE 101021 and 403125 human urine measured during the analysis period were 41±5 nmol L-1 and 661±56 nmol L-1 respectively. This is in good agreement with the manufacturer’s recommended values of 43 and 681 nmol L-1. The detection limit of the method used was 10 nmol L-1 of nickel. The creatinine content was measured by a Beckman Creatinine Analyser (Beckman Instruments, Brea, CA, USA) based on JaVe’s reaction.Only in Nikel and Monchegorsk did the study group include women who worked in departments of the local refinery with potential exposure to nickel. It should be noted that, today, maternity leave starts 70 days before the anticipated date of delivery, while it was 56 days during the study period.Fig. 1 Schematic map of the Russian Barents region. Bergen is located Consequently, since urine samples in Monchegorsk were col- in the southwest of Norway, and is not shown on the map. (Courtesy of Elin Hanssen, NILU, Tromsø, Norway.) lected shortly after birth, the employees included were not 154 J. Environ. Monit., 1999, 1, 153–161recently occupationally exposed.In Nikel, a comparison is 0.07). Mean maternal weights at term were not diVerent ( p= possible between samples collected while still at work (week 0.11), while mothers’ mean height was significantly lower in 20) and postpartum. the Russian group ( p<0.001). However, this did not result in Tap water samples (10 or 12) were collected in the following diVerent maternal body mass index (BMI) values (p=0.87).Russian communities: Nikel, Zapolyarniy, Monchegorsk, Pregnancy outcomes Arkhangelsk and Umba (the latter is located in the Kola Peninsula, south of Apatity on the White Sea). Homes were Compared to the Norwegian results, the mean birth weight selected at random, the taps were flushed for a few minutes and body mass index of the newborn child (BMIC) were and the samples were collected directly into the same containers significantly lower in the Russian group (p<0.001, Table 2), as used for the urine samples.The shipping, storage and a diVerence that was retained after adjustment for gestational analysis of the tap water were carried out in the same manner age or gender. Babies were somewhat longer in the Russian as for the urine specimens.In some cases, the water samples group (p<0.001), while the diVerences in mean placenta were not frozen and were analysed soon after their arrival at weight and head circumference were not statistically significant. the laboratory in Oslo. Twenty-four hours before measure- The children in Arkhangelsk had the lowest mean birth weight, ment, 0.5 mL 65% ultrapure nitric acid was added to each with the highest values being reported in Kirkenes. The mean 20 mL sample in order to recover any surface-adsorbed nickel.gestational age was significantly lower in Russia (p<0.001), Nickel measurements were performed in the same way as for being the lowest in Arkhangelsk. These findings are consistent urine, but calibration was against aqueous standard solutions. with a prevalence of 15% of children with birth weights of less Surface water CSPS-SW1 (Spectrapure Standards Ltd., Oslo, than 2500 g in Arkhangelsk, compared to 3.5% in the two Norway) trace-metal quality-control materials were used other Russian groups.Importantly, the diVerence between the routinely. The day-to-day variation of the nickel measuremean birth weights for Russia and Norway remained signifi- ments in these reference materials was typically 5%.The mean cant (p<0.001) when omitting the Arkhangelsk group in the nickel concentration measured in CSPS-SW1, batch 102, comparison. As might be expected from the small number of was 164±5 nmolL-1 (certified value 170±9 nmolL-1 of nickel ). subjects in our study,18 only four children were born with registered malformations, two in Norway and two in Russia.Statistical analysis One perinatal death (a child with hydrocephalus in In the statistical assessment univariate analysis, analysis of Arkhangelsk) was identified. variance and multiple linear regression analyses were In the Russian group, a total of 15 women were employed employed. An association was accepted when the 95% confi- in the nickel industry, six in Nikel, who apparently worked dence interval (CI) of the regression coeYcient did not include in departments with little nickel exposure, and nine in zero.The Bartlett’s test for homogeneity of variance and the Monchegorsk, who were potentially exposed in the electrore- Mantel–Haenszel and Fischer exact tests for comparison of fining department.However, as already indicated, due to proportions were used. In all groups, the urinary nickel results their maternity leave none of these women were recently were highly skewed, and the non-parametric Wilcoxon rank occupationally exposed. The nine women working in the sum test or the Kruskal–Wallis test for two groups were Monchegorsk electrorefinery had babies with an average birth selected.Concentrations below the detection limit (DL) were weht of 3342 g (2540–4200), which is higher than the mean arbitrarily assigned the value of 1/2DL. Information on the birth weight in the total Russian study population; one baby years of education, defined as years at school, was missing in had a registered malformation (a minor limb defect).Seven the Russian questionnaire responses and could therefore not of these nine nickel workers were daily smokers (77.8%), be included in the regression analysis. Neither could the which exceeds the mean smoking frequency reported in Table 1; answers about alcohol consumption be used, since 49 of the the delivery frequency was identical to that for the total Russian respondents were reluctant to answer this section of Russian group (mean of 1.2).the questionnaire. Since only one child was lighter than 2500 g among the Norwegian subjects, it was decided to use the birth Urinary nickel concentrations weight outcome as a continuous variable instead of categorical The urinary nickel levels for the children and their mothers (i.e. by defining low birth weight as <2500 g).Birth weights are provided in Table 3. Both the neonatal and maternal corresponding to gestational ages of 30–38 weeks were adjusted medians of the unadjusted urinary nickel concentrations were using the observed rate of increase of 166 g per week. considerably higher among the Russian subjects compared to the Norwegians ( p<0.001). Values reported for the Norwegian Results group fall within the baseline reference interval of 9–100 Population characteristics nmol L-1;12,13,19 while the Russian concentrations are mostly outside of it (maternal range at term, 5–2108 nmol L-1).The mean age of the mothers was significantly higher in Creatinine adjustment did not aVect these comparisons. Norway than in Russia (27.7, range 17–40; 25.1, range 14–44; Focusing on the towns of Nikel and Monchegorsk and the respectively; p<0.001; Table 1).The percentage of the city of Arkhangelsk, we found that, for all three Russian Norwegian mothers who were registered as smokers was centres, the median nickel concentrations were higher than for 35.9%, of whom 13.3% might be regarded as heavy smokers the Norwegian populations studied.Intercommunity compari- (>10 cigarettes per day). By comparison, the Russian group sons for the three Russian populations revealed that, relative had a smoking prevalence of 17.4%, none of them in the heavy to Arkhangelsk, the median creatinine-adjusted nickel concen- smokers’ category (Table 1). Complications related to pretrations for Nikel and Monchegorsk were significantly higher eclamptic conditions, specifically hypertension, oedema, pro- ( p<0.001).This diVerence was not apparent for the unad- teinuria or anaemia, were significantly higher in the Russian justed nickel concentrations. In the Nikel and Kirkenes groups, group (p=0.03). Local food consumption was categorized on additional urines were also collected in the 20th week of the basis of whether the consumption of locally produced fish, pregnancy and at term.Neither the Nikel or the Kirkenes meat, vegetables, mushrooms and berries occurred daily or group demonstrated diVerences between the two stages of not. The Monchegorsk group was not adequately examined pregnancy (p>0.5, N=42 and p>0.5, N=41 respectively). in this component of the questionnaire to be considered in the Occupationally exposed women in Monchegorsk (N=9, who interpretation of the data.The number of previous deliveries had worked in the electrolysis tank house, but with no recent was significantly lower for the Russian subjects compared to the Norwegian group (means of 1.2 and 1.4 respectively; p= industrial exposure) had a median urinary nickel concentration J.Environ. Monit., 1999, 1, 153–161 155Table 1. Population characteristics Russia Norway Characteristic (N=137) (N=128) p Values Mean maternal agea/years 25.1 27.7 <0.001e (s, range) (5.9, 14–44) (5.3, 17–40) Mean number of deliveriesa 1.2 1.4 0.07e (range) (0–4) (0–5) Mean maternal weighta/kg 70.9 73.5 0.11e (range) (41–101) (46–123) Mean maternal heighta/cm 163 166 <0.001e (range) (150–174) (155–182) Mean body mass index (BMI)/kg m-2 26.7 26.6 0.87e (range) (17.3–39.6) (18.4–38.0) Pre-eclamptic conditionac 26 12 0.03f (19.0%) (9.4%) Smoking habitsb (%) Non-smokers 82.6 64.1 0.005f 1–10 cigarettes per day 17.4 22.7 >10 cigarettes per day 0 13.3 Smoking frequency 17.4 35.9 Local food intake (%)bd 86.1 61.0 <0.001f aBased on medical records; for maternal height only, N=121 (Russia) and N=93 (Norway).bBased on questionnaire. cAt least two of the parameters: hypertension, oedema, proteinuria. dRegular use of locally produced vegetables, potatoes, berries and/or locally produced fish or meat; the Monchegorsk group is not included (see text). et-test. fChi-squared test. Table 2 Comparison between selected pregnancy outcomes in Russia and Norway Russia Norway Outcomea (N=137) (N=128) p Values Mean birth weight/g 3195 3590 <0.001b (s, range) (579, 1400–5100) (502, 2200–4960) Mean length of baby/cm 51.8 50.7 <0.001b (range) (41–58) (45–58) BMIC/kg m-2 11.9 13.9 <0.001b (s, range) (1.7, 7.5–18.2) (1.4, 10.1–19.4) Mean placenta weight/g 582 621 0.06b (range) (300–900) (350–1050) Mean gestational age/weeks 38.7 39.8 <0.001b (range) (31–42) (36–42) Mean head 35.0 35.0 0.92b circumference/cm (range) (32–38) (30–39) aAll information is derived from medical records.bComparison of country means by t-test. Table 3 Median nickel urine concentrations of delivering women and their babies in Arctic areas of Russia and Norway (nmol L-1)a Group studied Russia Nikel Monchegorsk Arkhangelsk Norway Kirkenes Hammerfest Bergen (N=137) (N=42) (N=49) (N=46) (N=128) (N=46) (N=51) (N=31) p Valueb Maternal urine nickel 85 90 83 85 5 5 14 15 <0.001 at term/nmol l-1 5–2108 5–694 5–2108 19–1258 5–85 5–26 5–82 5–85 Maternal urine nickel 9 13 10 6 1 1 3 2 <0.001 at term, adjusted for 1–285 1–139 3–285 1–49 0.2–41 0.2–8 0.2–41 0.3–36 creatinine/nmol Ni per mmol creat Nickel in baby’s first-voided 34 41 37 24 5 5 5 5 <0.001 urine/nmol L-1 5–561 5–561 5–374 5–260 5–48 5–20 5–48 5–24 Baby’s urine nickel, 11 31 12 5 2 3 2 2 <0.001 adjusted for 1–510 3–510 2–45 1–325 0.4–187 0.4–34 0.5–187 0.5–13 creatinine/nmol Ni per mmol creat aMedian value and range are given. For statistical purposes, values below the DL of 10 nmol L-1 were set at cDL. bFor the comparison of the total Norwegian and Russian data sets, non-parametric statistics were used (see text).of 66 nmol L-1, somewhat less than that in the total Russian variable. The regression coeYcient corresponding to the change in the baby’s weight in grams per unit of the explanatory group (median value, 85 nmol L-1). variable was first examined. For both the Russian and Norwegian subjects, maternal urinary creatinine is suggested Univariate linear regression as a weak negative predictor (not significant).For the total The results of the univariate analysis are provided in Table 4. Russian group, there were no other variables significantly associated or even nearly so with the two outcomes. For the The birth weight or the BMIC was selected as the dependent 156 J.Environ. Monit., 1999, 1, 153–161Table 4 Linear regression analysis of birth weight Russia/Norway Russia Norway (adjusted for country) Weight Weight Weight Variable change (CI)a p Valueb change (CI)a p Valueb change (CI )a p Valueb Maternal Age (years) 8.5 (-8, 25) >0.05 10 (-7, 27) >0.05 9 (-3, 21) >0.05 Smoking (categorical, 116 (-86, 318) >0.05 -193 (-312, -73) <0.005 -91 (-200, 18) >0.05 amount smoked)c Local food consumption -82 (-366, 202) >0.05 -181 (-359, -3) <0.05 -146 (-302, 10) >0.05 (categorical, yes/no)c Pre-eclamptic conditions 1.8 (-249, 252) >0.05 -5 (-307, 297) >0.05 0 (-190, 189) >0.05 (categorical, yes/no)c Number of deliveries 58 (-45, 161) >0.05 23 (55, 101) >0.05 38 (-25, 100) >0.05 Gender of baby, M/F 64(- 136, 263) >0.05 -126 (-316, 65) >0.05 -19 (-158, 120) >0.05 Maternal weight/kg 6 (-3, 15) >0.05 12 (5, 19) <0.001 9 (3–15) <0.005 Maternal height/cm -2 (-22, 18) >0.05 21 (6, 36) <0.005 10 (-3, 22) >0.05 BMI/kg m-2 16 (-7, 39) >0.05 31 (11, 52) <0.005 22 (6, 38) <0.01 Maternal urine -0.1 (-0.5, 0.3) >0.05 1 (-4, 6) >0.05 -0.1 (-0.5, 0.3) >0.05 nickel/nmol L-1 Maternal urine -8 (-23, 6) >0.05 -13 (-32, 7) >0.05 -10 (-21, 2) >0.05 creatinine/mmol L-1 Child’s urine -0.6 (-2.0, 0.8) >0.05 -3 (-14, 8) >0.05 -0.6 (-1.9, 0.7) >0.05 nickel/nmol L-1 Child’s urine creatinine/mmol L-1 12 (-15, 39) >0.05 10 (-13, 33) >0.05 11 (-6, 29) >0.05 Maternal adjusted urine -0.9 (-4, 2) >0.05 9 (-8, 27) >0.05 -3 (-6, 0.1) >0.05 nickel/nmol Ni per mmol creat Child’s adjusted urine -0.6 (-2, 0.6) >0.05 1 (-4, 6) >0.05 -0.6 (-1.7, 0.6) >0.05 nickel/nmol Ni per mmol creat aWeight change in grams per unit of explanatory variable; CI, 95% confidence interval.bBased on the partial F-statistic. cSee Table 1 for classification.Table 5 Multivariate linear regression analysis model to predict birth combined Norwegian subjects, the following associations weight for the combined Russian/Norwegian populationab reached significance: maternal weight ( p<0.001), maternal height ( p<0.005), BMI (p<0.005), smoking ( p<0.005) and Change in birth local food consumption ( p<0.05). On combining the Russian weight/g and Norwegian study groups and adjusting for country, the Variable (95% CI) p Value associations with maternal weight ( p<0.005) and BMI Maternal urinary -1 (-6, 5) >0.05 ( p<0.01) remained significant, as did the near significance of nickel/nmol L-1 maternal urinary creatinine.Adjustment of birth weight for BMI/kg m-2 25 (8, 42) <0.005 gestational age strengthened the maternal weight and BMI Maternal height/cm 14 (0, 27) <0.05 dependences (both p<0.001), but did not alter the remaining Maternal urinary -10 (-22, 3) >0.05 associations substantially.creatinine/mmol L-1 Smoking (0, 1–10 -25 (-148, 99) >0.05 For BMIC, there were no significant associations for the or >10 cigarettes) Russian subjects; among the Norwegian group, smoking was Country (Russia/Norway) 315 (143, 487) <0.001 negatively correlated (p<0.025), while maternal weight ( p<0.025) and BMI (p<0.025) were positively so.Only the aSince 52 maternal heights were not reported and the data sets for the other variables had some omissions, N=200 (see Table 1); all values maternal weight (p<0.025) and BMI (p<0.01) retained sigare mutually adjusted.bThe F-statistic for the model is 7.3, with d.f. nificance for the combined countries. As with birth weight, 6 and 194; p<0.001. adjustment for gestational age strengthened the relationship with maternal weight and BMI (both p<0.001). Log transformation or square transformation of the nickel concentrations added no additional information; neither did urinary creatinine is negative (nearly significant); a very strong the grouping of birth weight by 500 g increments.unspecified country factor is evident ( p<0.001). The model was optimized20 by removal of local food consumption and Multivariate linear regression neonatal urinary creatinine. Judging by the partial F-statistic, the explanatory contributions of both of these variables were The variables for which significant or near-significant associations with birth weight were observed in either of the two low and not statistically significant; that of maternal urinary nickel was zero.Introducing gestational age-adjusted birth countries (see Table 4) were tested in a multivariate model. Because a major focus of this study was a comparison of weight as the dependent variable strengthened the positive predictive value of BMI (p<0.001) and maternal height urinary nickel excretion, maternal urinary nickel was also carried forward.Maternal body weight was not included since (p<0.025) and weakened the country factor slightly (p<0.01), leaving the other predictors unchanged. In the corresponding it is related to BMI.By contrast, the exclusion of maternal height as an independent explanatory variable resulted in a multivariate model with BMIC as the dependent variable (Table 6), BMI (p<0.01 or 0.001) and the country factor significant loss of information. The model summarized in Table 5 shows BMI (p<0.005) and maternal height (p<0.05) ( p<0.001) were the only notable variables (with or without adjustment for gestational age).as positive predictors, while the contribution of maternal J. Environ. Monit., 1999, 1, 153–161 157Table 6 Multivariate linear regression analysis model to predict child’s body mass index for the combined Russian/Norwegian populationab Change in birth weight/g Variable (95% CI) p Value Maternal urinary nickel/nmol L-1 0 (-0.03, 0.004) >0.05 BMI/kg m-2 0.07 (0.02, 0.12) <0.01 Maternal height/cm 0.01 (-0.03, 0.05) >0.05 Maternal urinary creatinine/mmol L-1 0.002 (-0.03, 0.04) >0.05 Smoking (0, 1–10 or >10 cigarettes per day) -0.06 (-0.42, 0.31) >0.05 Country (Russia/Norway) 1.9 (1.4, 2.4) <0.001 aSince 52 maternal heights were not reported and the data sets for the other variables had some omissions, N=199 (see Table 1); all values are mutually adjusted.bThe F-statistic for the model is 15.8, with d.f. 6 and 193; p<0.001. Birth weight in relation to neonatal urinary nickel concentrations Group analysis of neonatal urinary nickel concentrations (group 1, <15 nmol L-1; group 2, 15–170 nmol L-1; group 3, >170 nmol L-1) demonstrated no evidence for a dependence of birth weight (or BMIC; data not reported) on neonatal urinary nickel concentrations (Norwegian group, p= 0.5; Russian group, p=0.3 or 0.2; all relative to group 1 by ttest).Neither was a risk for reduced birth weight apparent when the weights were compared for urinary nickel levels below and above 34 nmol L-1, corresponding to the mean reference background concentration reported for nonpregnant adults.19 Nickel in tap water Russian communities with local nickel refineries, namely Nikel Fig. 2 A comparison of the frequency distribution of the logand Zapolyarniy (see Fig. 1), had significantly higher nickel transformed urinary nickel data for the Russian groups surveyed in levels in the drinking water (median values of 1224 and the present study (N=179; includes the results from week 20 of 578 nmol L-1, respectively) compared to the four other pregnancy in Nikel ) with that (N=52) reported for female workers employed in the Electrorefinery Department of the Severonickel Russian and two Norwegian locations (medians in the range operation in Monchegorsk.8 The sample collection and analytical 9–85 nmol l-1, p<0.001). In the Russian communities with procedures were identical to those described in the text; urine samples no local point sources of nickel, specifically Kirovsk, Apatity were collected in May 1996 and correspond to the first morning void and Umba, values were closer to the Norwegian levels (median collected at home to minimize inadvertent contamination.The median values of 9, 85 and 14 nmol L-1 respectively).Ten out of 12 and range of the nickel concentrations for the nickel workers were tap water samples from Arkhangelsk were at the DL or below 1496 (289–14 620) nmol L-1, compared to 99 (5–2108) nmol L-1 in the total Russian population in the current study. Ambient air nickel (all from private kitchens); the two detectable values were 12 levels in the Monchegorsk Electrorefinery Department8 were in the and 20 nmol L-1 (collected at the Institute of Physiology, range 60–1200 mg m-3, which appear to be higher by about a factor Ural Branch of Russian Academy of Science, Arkhangelsk, of 1000 to 10 000 or more than reported air levels for the town of and a private kitchen of a colleague).Nikel and the surrounding area, including the Norwegian–Russian border area.14,15 Discussion during parturition and postpartum suVered from severe analyt- Urinary nickel concentrations ical deficiencies and therefore can only be interpreted with considerable caution.6 Birth-associated trauma may have con- The comparison of the urinary nickel concentrations reported in the present study with published background reference tributed to the slightly higher urinary nickel concentrations in the present pregnant females, when compared to other adults.intervals12,13,19 (<100 nmol L-1) is beneficial. Clearly, the results for the Russian study groups exceed this reference Making comparisons for Nikel, the median urinary nickel value for the pregnant women of the present study was 90 upper limit, while the values for the Norwegian communities are well below it.The slightly higher creatinine-adjusted con- nmol L-1, compared to 58 nmol L-1 for other residents (N= 371, 70% female).15 centrations observed for the Nikel and Monchegorsk populations relative to Arkhangelsk (p<0.001) suggest some The higher urinary nickel excretion among the Russian groups does not have an obvious explanation.Nickel in tap environmental contribution due to the presence of the local nickel smelters. Such an impact of nickel refining on non- water, perhaps in combination with enhanced air levels, probably accounts for the higher creatinine-adjusted urinary nickel occupationally exposed community members has been reported previously for the city of Sudbury, Ontario.21,22 It is levels observed for Nikel and Monchegorsk, compared to Arkhangelsk.Since there are no nickel refineries or any other interesting to compare the Russian data for the pregnant women surveyed with recently obtained results for Russian obvious nickel point source in Arkhangelsk, other explanations are required to account for the major portion of the Russian/ females working in a Kola Peninsula nickel electrorefinery.This is done in Fig. 2. It is evident from this figure that the Norwegian urinary nickel anomaly. It is known that certain foodstuVs are relatively enriched in nickel, such as cocoa, nuts, two groups compared have separate frequency distribution patterns. The overlap between them is minimal. dried legumes and certain grains.7,24 Special dietary sources in the Russian community might therefore be suspect.Leaching There has been some suggestion that nickel excretion is increased during pregnancy, parturition and postpartum.23 into drinking water from nickel-plated pipes or from cooking utensils constitutes another explanation. Oral prostheses are Earlier reports of drastic increases in serum or urinary nickel 158 J. Environ. Monit., 1999, 1, 153–161in use in Russia that apparently have significant nickel con- of the explanatory variables should be consulted before acceptance of the suggested associations.tents, and this may well be an unexpected source of nickel.8 To our knowledge, nickel in first-voided urines of neonates The univariate linear regression data and the multivariate model reinforce that smoking is a predictor of low birth has not been reported previously. Because so many urinary nickel concentrations in both Norwegian mothers and neonates weight.The data presented suggest it reaches significance only if the smoking rate is moderately high, such as seen among were below or at the DL, it is diYcult to compare their relative magnitudes. This is not the case for the Russian population the Norwegian subjects.The retention of BMI ( p<0.005) and maternal height (p<0.05) in the multivariate model conforms studied. A comparison of maternal and neonatal creatininewith the general acceptance of maternal weight and height as adjusted nickel levels shows the latter to be lower (p<0.001). positive determinants. Similarly, urinary creatinine as a near From the perspective of the glomerular filtration rate (GFR), significant variable in the linear and multiple regressions this is not unexpected.At birth, the neonatal GFR appears to suggests that this parameter might be more sensitive in be about 20% of adult values.25 explaining birth weight than clinical diagnosis of the preeclamptic condition as a categorical variable.It is obvious from the magnitude of the ‘country’ slope factor in Table 5 Birth weight as a pregnancy outcome that we have failed to identify important predictors of birth Birth weight is recognized as an important indicator of the weight. The replacement of birth weight by BMIC did not health status of neonates.26 Temporal changes in this parameter improve the model, although adjusting for gestational age may serve as an index to socioeconomic conditions that impact strengthened the BMI influence and left the unspecified country on reproductive and developmental health.It is also known factor unchanged. Presumably, these results reflect unidentified to be sensitive to adverse environmental conditions, as illus- diVerences in economic conditions, cultural practices or trated by cigarette smoking.As a pregnancy outcome, birth nutritional factors. weight permits an outcome comparison unaVected by cultural diVerences and medical uncertainties such as outcome defi- Limitations of the study nitions. Congenital malformations and perinatal morbidity are examples. However, birth weight has some disadvantages as a A number of information biases can be identified.Answers dependent variable for pregnancy outcome, due to genetic concerning the consumption of local food were not provided diVerences.27,28 On the other hand, the BMIC has been in the Russian community of Monchegorsk. The reason for suggested to have some advantages as an index to the this may reflect the decision by the local interviewer not to nutritional status of newborn children.29–31 It would appear emphasize this component.The refusal by 36% of the Russian that the BMIC is somewhat less sensitive to the explanatory respondents to provide information about alcohol consumpvariables examined in our study. tion suggests that this topic is culturally sensitive. Even fewer Among the statistically significant associations reported in individuals responded to educational background questions Table 4 for the linear regression between birth weight and a there.Inconsistency between interviewers is also of concern. number of explanatory variables are those recognized for However, this was unavoidable since so many diVerent comhaving a known positive impact on birth weight (i.e. maternal munities were involved.Further, the questionnaire was transweight, maternal height, BMI, number of deliveries)32 or a lated from Norwegian into Russian and this may have had a negative eVect (i.e. smoking, hypertensive pregnancy compli- steering eVect on some of the questions. Of course, inherent cations).32–34 BMI is considered a measure of body fatness. It ambiguities in the questions may have influenced the consistis strongly influenced by age and sex, but not by ethnicity.35 ency of the answers.In our estimation, the discrepancies Since we are dealing with females of comparable age, BMI between Norwegian and Russian physicians in classifying may thus be taken as a suitable index of adipose tissue mass. pregnancy complications and outcomes were minimal. An Although not included in the analysis, alcohol consumption example is the diagnosis of the pre-eclamptic condition.during pregnancy is also well known to reduce birth weight. Selection bias might have occurred for two reasons, namely It is of interest that the 35.9% smoking frequency found transfer of cases with complications to regional hospitals and among the Norwegian group concurs almost exactly with the the exclusion from the statistical treatment of mothers and proportion of subjects identified as daily smokers prior to their neonates because of the unavailability of neonatal urine pregnancy in a recent survey of 4766 pregnant Norwegian samples.Patient transfers were few and were registered. women.36 Inclusion of the neonatal weights for those without urinary It is tempting to interpret the apparent but weak negative nickel samples only reduces the diVerence between the Russian impact of maternal urinary creatinine as indicative of subclin- and Norwegian birth weight means by 4 g, which does not ical pre-eclamptic manifestations.In support, it is noted that aVect the significance test (p<0.001). Clearly, the significant pre-eclampsia was found to be more prevalent in the Russian Russian/Norwegian dissimilarity in birth weights is not communities (p=0.03).Interestingly, past pre-eclampsia explained by gestational age diVerences. appears to increase the risk of microalbuminuria,37 and Although many explanatory variables were examined, it is increased urinary creatinine levels have been noted as early as likely that others exist, as well as unidentified confounders.week 19 of gestation in patients who subsequently developed However, there is evidence that this may not be too critical. pregnancy-induced hypertension.38 The positive association In two overview reports on our work, one on essential trace suggested for neonatal urinary creatinine and birth weight elements (copper, zinc, iron and selenium)40 and the other on (not significant) also seems inherently reasonable.One might metal pollutants (cadmium, lead and mercury)1 in the periexpect, by analogy to adults, that body mass or surface area pheral blood compartment of many of the individuals in the determines the rate of urinary creatinine excretion. The latter present study, no obvious concentration diVerences that might has indeed been correlated with birth weight.39 Local food be expected to influence birth weight appeared to exist between intake may constitute a nutritional factor or environmental the Norwegian and Russian groups.For example, the iron contaminant issue. However, caution must be practised in status of pregnant women was somewhat better in Russia than interpreting at face value the associations suggested by single in Norway, and serum zinc levels were low in both countries regressions, even when statistically significant, because of inter- relative to established reference intervals.40 Whole blood cadpredictor influences and possible correlations among variables.mium and mercury concentrations were higher in the Norwegian communities than in Russia, while it was the Consequently, the outcome of a multiple regression analysis J.Environ. Monit., 1999, 1, 153–161 159reverse order for lead. The mean Russian lead level of extended to Knut Dalaker, Ka°re Augensen, Babill Stray- 0.18 mmol L-1 in whole blood was within the international Pedersen, Alexander Duriagin, Elvira Khotova, Leonid background reference interval of <0.20 mmol L-1.Zhivakov, Irina Perminova, Jevgenij Bojko, Anatoli Tkatchev, Due to logistical and technical limitations, as well as econ- Tone Smith-Sivertsen, Gunhild Sand, Per Einar Fiskebeck omic restrictions, the sample size for each country was too and, especially, midwife Marie Hallonen for their kind support small to detect diVerences in the incidence of congenital in diVerent phases of the project.defects. Much larger populations need to be examined to identify such outcomes with adequate statistical power, as they are relatively rare.18 The small number of cases identified in References the present study is consistent with this. 1 J. Ø. Odland, N. Romanova, G. Sand, Y. Thomassen, B. Salbu, Although we made every eVort to minimize inadvertent E.Lund and E. Nieboer, in Environmental Biomonitoring, contamination of the samples, the lack of opportunity to be Exposure Assessment and Specimen Banking, ACS Symposium personally involved in the early stages of the sample history Series No. 654, ed. K. S. Subramanian and G. V. Iyengar, (i.e. collection, handling, storage and transport to the labora- American Chemical Society, Washington, 1997, pp. 135–150. tory) limited our ability to control the contamination risk. 2 B. Sivertsen, T. Makarova, L. O. Hagen and A. A. Baklanov, Air Pollution in the Border Areas of Norway and Russia, Summary This may account for some of the outliers corresponding to Report 1990–91, Report no NILU OR 8/92, Norwegian Institute the highest urinary nickel concentrations. for Air Research, Lillestrøm, Norway, 1992. 3 L. O. Hagen and B. Sivertsen, Overva°king av Luft- og Concluding remarks Nedbørkvalitet i Grenseomra° dene i Norge og Russland, Oktober 1991-Mars 1992, Report no 505/92, TA 897/1992, NILU OR It is interesting that two physical parameters (BMI and height), 82/92, Norwegian Institute for Air Research, Lillestrøm, Norway, a metabolic measure (maternal urinary creatinine excretion) 1992.and a life-style factor (smoking) were the strongest predictors 4 T. Norseth, Sci. Total Environ., 1994, 148, 103. 5 V. P. Chashschin, G. P. Artunina and T. Norseth, Sci. Total of neonatal birth weight in this combined Russian/Norwegian Environ., 1994, 148, 287. study population. The unimportance of urinary nickel 6 E. Nieboer, F. E. Rossetto and R.Menon, in Metal Ions in excretion, and thus nickel exposure, as an explanatory variable Biological Systems, Nickel and its Role in Biology, ed. H. Sigel and for birth weight is consistent with the recent conclusion of the A. Sigel, Marcel Dekker, New York, 1988, pp. 359–402. Norwegian–Russian Health Study that no major health eVects 7 International Programme on Chemical Safety, Environmental can be assigned to nickel as an air pollutant in the vicinity of Health Criteria 108, Nickel, World Health Organization, Geneva, 1991.the nickel refining operations in Nikel and Zapolyarniy.16 This 8 E. Nieboer, V. P. Tchachtchine, J. Ø. Odland and Y. Thomassen, is not surprising. The local ambient air nickel concentrations Reproductive and Developmental Health in Relation to there are low, although above background (54 ng m-3 com- Occupational Exposure to Nickel in the Kola Peninsula of Russia: a pared to 23 ng m-3 respectively; measured as the respirable Feasibility Study, McMaster University, Hamilton, ON, Canada, fraction, which seems to be comparable to the inhalable July 24, 1997.fraction in this instance15,16). It may be concluded from the 9 L. Jarup, M.Berglund, C. G. Elinder, G. Nordberg and M. Nikel and Monchegorsk urinary nickel data that the nickel Vahter, Scand. J. Work Environ. Health, 1998, 24, 1. 10 K. W. Andrews, D. A. Savitz and I. Hertz-Picciotto, Am. J. Ind. refineries as local point sources only minimally aVect the body Med. 1994, 26, 13. burden of nickel. 11 J. C. Hansen, U. Tarp and J.Bohm, Arch. Environ. Health, 1990, The above conclusion should not be extended to females 45, 355. occupationally exposed to nickel for the following reasons. 12 C. Minoia, E. Sabbioni, P. Apostoli, R. Pietra, L. Pozzoli, First, nickel is known to be transferred readily across the M. Gallorini, G. Nicolaou, L. Alesio and E. Capodaglio, Sci. placenta.41,42 Since nickel in urine is proportional to nickel in Total Environ., 1990, 95, 89. 13 D. M. Templeton, F. W. Sunderman Jr and R. F. M Herber, Sci. the blood plasma compartment,22 it may be inferred that Total Environ., 1994, 148, 243. serum nickel also constitutes an index to embryonic or foetal 14 T. Smith-Sivertsen, E. Lund, Y. Thomassen and T. Norseth, Arch. exposure. Because of the high occupational nickel exposures Environ.Health, 1997, 52, 464. experienced by these workers (see Fig. 28,43,44), in utero expo- 15 T. Smith-Sivertsen, V. P. Tchachtchine, E. Lund, V. Bykov, sure cannot be dismissed. Second, concern has been expressed Y. Thomassen and T. Norseth, Environ. Health Perspect., 1998, about increases in spontaneous abortions among females who 106, 503. 16 T. Smith-Sivertsen, V.P. Tchachtchine, E. Lund, T. Norseth and work in the nickel electrorefinery at Monchegorsk.5 An appar- V. Bykov, The Norwegian–Russian Health Study 1994/95. A Cross- ent increase in selected congenital malformations was also sectional Study of Pollution and Health in the Border Area, ISM noted. Third, mechanistically speaking,26 nickel compounds Skriftserie Nr. 42, University of Tromsø, Tromsø, Norway, 1997.are potentially teratogenic and embryotoxic because they are 17 B. Welz, G. Schlemmer and J. R. Mudakav, J. Anal. At. genotoxic. Respiratory tract cancers have been associated with Spectrom., 1992, 7, 1257. occupational exposures to water-soluble and particulate 18 Medical Birth Registry of Norway, Annual Report 1995, (mostly oxides and sulfides) forms.43,45,46 Clearly, follow-up University of Bergen, Bergen, Norway, 1996. 19 F. W. Sunderman Jr, Scand. J. Work Environ. Health, 1993, 19 studies are needed. We have recently illustrated that a compre- (Suppl. 1), 34. hensive epidemiological assessment of reproductive and devel- 20 D. G. Altman, Practical Statistics for Medical Research, Chapman opmental health among female nickel refinery workers in the & Hall, London, 1991, pp. 336–351. Kola Peninsula is technically feasible.8 21 S. M. Hopfer, W. P. Fay and F. W. Sunderman Jr., Ann. Clin. Lab. Sci., 1989, 19, 161. 22 E. Nieboer, W. E. Sanford and B. C. Stace, in Nickel and Human Acknowledgements Health: Current Perspectives (Advances in Environmental Science This work was supported by the University of Tromsø, Steering and Technology, Vol. 25), ed. E. Nieboer and J. O. Nriagu, John Wiley, New York, 1992, pp. 49–68. Group of Medical Research in Finnmark and Nordland, and 23 S. Nomoto, T. Hirabayashi and T. Fukuda, in Chemical the Royal Norwegian Department of Foreign AVairs, East- Toxicology and Clinical Chemistry of Metals, ed. S. S. Brown and European Secretariat. The authors wish to thank the staV at J. Savory, Academic Press, New York, 1983, pp. 351–352. the obstetric departments of hospitals in Bergen, Kirkenes, 24 G. D. Nielsen, in Nickel and Human Health: Current Perspectives Hammerfest, Nikel, Monchegorsk and Arkhangelsk for their (Advances in Environmental Science and Technology, Vol. 25), ed. excellent cooperation in the administration of the question- E. Nieboer and J. O. Nriagu, John Wiley, New York, 1992, pp. 201–210. naires and collection of specimens. Acknowledgement is 160 J. Environ. Monit., 1999, 1, 153–16125 I. Ichikawa, Pediatric Textbook of Fluids and Electrolytes, 37 R. A. North, D. Simmons, D. Barnfather and M. Upjohn, Aust. NZ J. Obstet. Gynaecol., 1996, 36, 233. Williams & Wilkins, Baltimore, 1990, pp. 494–495. 38 P. N. Baker and G. A. Hackett, Obstet. Gynecol., 1994, 83, 745. 26 H. Taskinen, Scand. J. Work Environ. Health, 1990, 16, 297. 39 J. L. Sutphen, Pediatrics, 1982, 69, 719. 27 T. J. Cole, J. V. Freeman and M. A. Preece, Stat. Med., 1998, 40 J. Ø. Odland, N. Romanova, G. Sand, Y. Thomassen, J. Brox, 17, 407. E. Khotova, A. Duriagin, E. Lund and E. Nieboer, Arct. Med. 28 T. Chard, A. Soe and K. Costeloe, J. Perinat. Med., 1997, 25, 111. Res., 1996, 55 (Suppl 1), 38. 29 K. Wandja, P. J. Hooft and H. P. van de Voorde, J. Gynecol. 41 F. W. Sunderman Jr., Ann. Clin. Lab. Sci., 1977, 7, 377. Obstet. Biol. Reprod. (Paris), 1995, 24, 444. 42 T. W. Clarkson, G. F. Nordberg and P. R. Sager, Scand. J. Work 30 T. J. Cole, G. L. Henson, J. M. Tremble and N. V. Colley, Ann. Environ. Health, 1985, 11, 145. Hum. Biol., 1997, 24, 289. 43 Y. Thomassen, E. Nieboer, D. Ellingsen, S. Hetland, T. Norseth, 31 L. Raman, G. Vasanthi, K. V. Rao, C. Parvathi, N. Balakrishna, J. Ø. Odland, N. Romanova, S. Chernova and V. P. Tchachtchine, N. Vasumathi, A. Raval and K. Adinarayana, Indian Pediatr., J. Environ. Monit., 1999, 1, 15. 1989, 26, 630. 44 E. Nieboer, in Nickel and Human Health: Current Perspectives 32 M. A. Wilcox and I. R. Johnson, Curr. Obstet. Gynaecol., 1990, (Advances in Environmental Sciences and Technology, Vol 25), ed. 2, 100. E. Nieboer and J. O. Nriagu, John Wiley, New York, 1992, 33 B. Backe, Acta Obstet. Gynecol. Scand., 1993, 72, 172. pp. 37–47. 34 P. Oian and J. M. Maltau, Tidsskr. Nor Laegeforen, 1995, 115, 45 R. Doll (ed.-in-chief ), Scand. J.Work Environ. Health, 1990, 16, 1. 698. 46 International Agency for Research on Cancer, Evaluation of 35 D. Gallagher M. Visser, D. Sepulveda, R. N. Pierson, T. Harris Carcinogenic Risks to Humans, Vol. 49, Chromium, Nickel and and S. B. Heymsfield, Am. J. Epidemiol., 1996, 143, 228. Welding, World Health Organization, Geneva, 1990. 36 K. Eriksson, K. Haug, K. A° . Salvesen, B-E Nesheim, G. Nylander, S. Rasmussen, K. Andersen, J. O. Nakling and S. H. Eik-Nes, Acta Obstet. Gynecol. Scand., 1998, 77, 159. Paper 8/09577I J. Environ. Monit., 1999, 1, 153–161 161