Fecundability and Parental Exposure to Ambient Sulfur Dioxide

Articles

Fecundability and Parental Exposure to Ambient Sulfur DioxideJan Deimek,1 Richard Jelinek,2 Ivo SolanskjQ Ivan Bene43 and Radim J. SrAm1

1Laboratory of Genetic Ecotoxicology, Regional Institute of Hygiene of Central Bohemia and Institute of Experimental Medicine,Academy of Sciences of Czech Republic, Prague, Vfdefska, Czech Republic; 2Charles University, Faculty of Medicine, Center ofBiomedical Sciences, Prague, Ruska, Czech Republic; 3District Institute of Hygiene, Teplice, Czech Republic

Recently it has been observed that birth rates in Teplice, a highly polluted district in NorthernBohemia, have been reduced during periods when sulfur dioxide levels were high. This study,which is based on data from 2,585 parental pairs in the same region, describes an analysis of theimpact of SO2 on fecundability in the first unprotected menstrual cycle (FUMC). We obtaineddeuiled personal data, induding time-to-pregnancy information, via maternal questionnaires atdelivery. We estimated individual exposures to S02 in each of the 4 months before conception onthe basis of continual central monitoring. Three concentration intervals were introduced: < 40pg/m3 (reference level); 40-80 pg/m3; and . 80 pg/m3. We estimated adjusted odds ratios(AORs) of conception in the FUMC using logistic regression models. Many variables werescreened. for confounding. AORs for conception in the FUMC were consistently reduced only forcouples exposed in the second month before conception to SO levels as follows: 40-80 pg/m3AOR 0.57 [95% confidence interval (CI), 0.37-0.88; p < 0.011]; > 80 pg/m3, AOR 0.49 (CI,0.29-0.81; p < 0.006). The association was weaker in the second 2 years of the study, probablydue to the gradual decrease of S02 levels in the region. The relationship between S02 andfecunldability was greater in couples living dose to the central monitoring station (within 3.5 kin).The timing of these effects is consistent with the period of sperm maturation. This is in agree-ment with recent findings; sperm abnormalities origina during spermatid maturation werefound in young men from Teplice region who were exposed to the increased levels of ambientS02' Alternative exphnati'ons of our results are also possible. Kq wordk: air pollution, environ-mental exposure, fecundability, human fertility, reproductive effects, S02, sperm maturation.Environ Healh Pepect 108:647-654 (2000). [Online 5 June 2000]http:/llebpnetlene iecs. nib.v/ldcs/2000/ 8p647-654dejmeklahs:ra./btm1

Several recently published papers have sug-gested that air pollution has detrimentaleffects on reproduction (1-5). Reduced birthrates in periods with high sulfur dioxide levelswere found in a heavily polluted region ofNorthern Bohemia during the late 1980s(6,i). Sram and colleagues (6,2) hypothesizedthat SO2 or some associated copollutants mayreduce reproductive success through adverseeffects on oocyte fertilization, an effect thathad been produced experimentally (8).Detailed information on the course and out-come of almost all pregnancies occurring inthe same region of the Czech Republic wascollected over the last 5 years; relevant pollu-tion data were also obtained. The purpose ofthe present paper is to verify the impact ofSO2 on fertility in the population.

The probability of conceiving during themenstrual cycle-fecundability-varies con-siderably, even for healthy and reasonablysexually active pairs in the human popula-tion. Some proportion of couples will con-ceive in the first unprotected menstrual cycle(FUMC), but others need more time tobecome pregnant. Little is known about theparticular causes of these differences (9).Obviously, this variation is mostly due to thebiologic and social differences amongparental pairs (10); however, fecundabilitymay be also influenced by environmental fac-tors such as temperature, photoperiodicity,

and food availability or quality (11). There iseven growing evidence that some occupa-tional (12-14) and lifestyle factors (15,16)and some environmental noxae can adverselyaffect human fertility (7,17).

Exogenous factors may influence thereproductive ability of parental pairs byaffecting different functions in various levelsand stages of the reproductive process (17);biologic as well as behavioral functions ofone or both partners may be impaired.Couples exposed to adverse factors takelonger time to achieve a clinically recogniz-able pregnancy. Therefore, the commonconsequence of such effects-irrespective ofthe particular mechanism-may be a con-ception delay. Thus, fecundability can besimply measured as the number of nonpro-tected menstrual cycles (MCs) required tobecome pregnant. This approach called timeto pregnancy (TTP) was introduced in thelate 1980s (13,18) and it has been useful innumerous epidemiologic studies (19-23).

A conception is usually not recognizedbefore the 5th week of gestation, and onlyapproximately two-thirds of pregnanciesreach this stage (24,25). Data about TTPobtained from parents can be related only torecognized pregnancies. These data reflect notonly the effects of factors that reduce theprobability of conception itself (interferingwith gametogenesis, transport of gametes, or

their fertilization ability); they also include theeffects of losses of conceived fetuses (loss dur-ing transport of zygotes, failure to successfullyimplant, and especially subdinical abortion).

The TTP approach does not allow fordifferentiating among these mechanisms.However, it may be sensitive enough todetermine relatively weak effects on humanfertility that could be expected for such com-mon toxins as air pollutants. Thus, weapplied a modified version of the TTPmethod in the present study to investigatethe impact of SO2 on human fertility.

Materials and MethodsThe present study was designed as a preva-lence population study with interviews atdelivery. The background sample includedall full-term singleton births in the district ofTeplice between April 1994 and March1998. We excluded all couples who admittedthat they "did something to prevent theindex pregnancy" or they "were treated forfertility disorders." Only spouses of Europeanorigin were enrolled to avoid additional vari-ability related to ethnic and cultural differ-ences. The sample was further restricted tothe mother's first delivery during the studyperiod. The completion of a writteninformed consent form was the final condi-tion for enrollment in the study.

Personal and lifestyle data were obtainedvia questionnaires and medical records. Self-administered maternal questionnaires werecompleted in the hospital after delivery, withthe assistance of a specially trained nurse.These data included occupational and other

Address correspondence to J. Dejmek, Laboratoryof Genetic Ecotoxicology, Institute of ExperimentalMedicine, Academy of Sciences of CR, 142 20Prague 4, VfdeiiskA 1083, Czech Republic.Telephone: 420 2 472 4756. Fax: 420 2 475 2785.E-mail: [email protected] thank I. Hertz-Picciotto for invaluable com-

ments and critical review of the manuscript. Wethank the many gynecologists and their staff mem-bers from the Departments of Obstetrics andGynecology in hospitals in Teplice and Duchcovfor their excellent collaboration. We also thank ourcolleagues from the District Institutes of Hygienein Teplice and Prachatice for their support and col-laboration.Supported by grants from the Czech Ministry of

Environment (Teplice Program II), U.S.Environmental Protection Agency/U.S. Agency forInternational Development, and CEC (PHARE II,EC/HEA-18/CZ).Received 3 December 1999; accepted 14 March

2000.

Environmental Health Perspectives * VOLUME 108 1 NUMBER 7 July 2000 647

Articles * Dejmek et al.

risks, smoking, and consumption of alcohol,health status and medication and detailedinformation on reproductive history andhabits. The main questions concerning TTPfrom the standard short questionnaire recom-mended by Baird et al. (26) were included:

1) Were you (or your partner) doing somethingto prevent pregnancy at the time you mostrecently got pregnant? If no: 2) Did you getpregnant during the first menstrual cycle ofunprotected intercourse? 3) the second 4) thethird 5) other? Please specify which cycle ...

Mothers who answered "yes" to question 1were excluded from the study.

Because the fathers were not interviewed,the women were proxy reporters for theirpartners' information; more data are there-fore missing for paternal characteristics.Hospital staff abstracted medical and healthcare data on the course and outcome of theindex pregnancy from the clinical records.

We estimated the gestational age inweeks using each woman's prenatal historylog (her maternity card), which included herreported last menstrual period (LMP), plusdata on prenatal visits, ultrasound measure-ments, etc. We calculated the estimated dateof conception (EDC) using the gestationalage and correcting from the LMP (decreas-ing the gestational age by 2 weeks) (5).

In the original version of the TTPmethod, an effect on fecundability is mea-sured by comparing the distribution inexposed and nonexposed couples of the num-ber of menstrual cycles required to becomepregnant. In this arrangement, a singleexposure estimate is not meaningful for per-manently changing exposures such as air pol-lution. On the other hand, the exposure andits timing can be easily defined for parentswho become pregnant in the FUMC.Moreover, according to Baird et al. (5), theproportion conceiving in the first cycle givesan unbiased estimate of the mean fecund-ability in the cohort.

Based on these considerations, we devel-oped a simplified version of the TTP method:only couples achieving a dinically recognizedpregnancy in the FUMC were categorized assuccessfully "conceived"; others were classifiedas "nonconceived." This approach examineswhether the proportion of women whobecome pregnant in the FUMC may differaccording to the exposure of the parents toair pollution before conception. We estimat-ed the exposure of each parental pair usingmean 30-day averages of SO2 levels in eachof 4 months before the EDC. We usedthree empirically chosen concentrationintervals for S02 according to exposure dis-tribution in population (low, < 40 pg/mi3;medium, 40 to > 80 pg/m3; and high, 2 80pg/m3). The cutoffs were near tertiles in thefirst 2 years of study (27); they fell to 55, 25,

and 20%, respectively, during the last 2years as a consequence of the downgradingof S02 in ambient air. We sorted parentalpairs by exposure to S02 levels in the partic-ular month before pregnancy and by con-ception success in the FUMC. Conceptionrates in the FUMC were estimated for eachconcentration interval and each monthbefore pregnancy.

Centralized air pollution monitoring inthe Teplice district was organized in cooper-ation with the U.S. EnvironmentalProtection Agency (U.S. EPA) (28). Themonitoring station was located in the centerof town of Teplice. Concentrations of SO2were measured continuously by the pulsefluorescence method using model 43A(Thermo Environmental Instruments, Inc.,Franklin, MA). Several other pollutants werealso measured, including nitrogen oxide,particulate matter, and polycyclic aromatichydrocarbons.

Individual exposures to SO2 were derivedfrom centralized monitoring of ambient air.To reduce inevitable misclassification andimprove the exposure estimate, we used someadditional variables. A unified system of pro-tection for inhabitants has been in place inthe Teplice District since 1993. In periodswith extremely high air pollution levels (forexample, meteorologic inversion) districtauthorities broadcast a special signal via localmedia and establish a special phone service[3-hr average SO2 > 400 pg/m3 or the sumof 24-hr average SO2 + (2 times total sus-pended particles) > 750 pg/m3]. Inhabitantsare encouraged to reduce outdoor activities.They are also encouraged to delay or reduceairing of their homes. Information about thetiming of such signals was obtained fromdistrict authorities. A variable signal wasdefined as "yes" for each 30-day period thatincluded a high-pollution episode, under theassumption that behavior would be alteredduring those times, resulting in a reductionof personal exposure.

Two other approaches were applied totake into account the influence of nonstan-dard behavior of inhabitants during inver-sions: we excluded all parental pairs who wereconsidered exposed to extreme levels of pollu-tion before conception. Two versions of thissample restriction were a) the exclusion ofcouples who were considered to be exposed tomonthly mean concentrations > 110 pg/m3,which were observed only in rare inversionepisodes, and b) the exclusion of couples whowere exposed to inversion situations duringthe 4 months before conception.

We expected that the authenticity of theexposure estimates would be higher for cou-ples living nearer to the monitor irrespectiveof the distances of sources or the wind direc-tion. To test this presupposition, we analyzed

the relationship between estimated exposureto pollution and fecundability separately forparents living at different distances from thecentralized monitor. For this purpose,parental pairs were classified into two groups(in and out) according to the distance ofmaternal permanent residence from themonitoring station in the year before deliv-ery. Couples living within 3.5 km of themonitoring station were classified as in; allothers were classified as out (Figure 1).

Many characteristics are related tohuman fertility (17,26) and to SO2 expo-sure. We initially examined the relationshipsbetween fecundability in the FUMC and thecharacteristics of the parents using t-test andchi-square analyses. These results were usedto construct logistic regression models.We estimated adjusted odds ratios

(AORs) and their confidence intervals (CIs)and Wald's chi-squares using logistic regres-sion procedures (29). Conception in theFUMC resulting in a live birth was entered asthe outcome measure in logistic regressionmodels. A spectrum of parental characteristicsassociated with the outcome (fertility) andsome parameters associated with the exposure(pollution) were tested for inclusion in thefinal model. These characteristics were mater-nal age (< 19, 19-34, 2 34 years), maternaland paternal education (basic, high school,university), marital status (currently married,other), parity (first, second or third, higher),spontaneous abortion (< 2, > 1), inducedabortion (< 2, > 1), alcohol habits of motherand father (< 1 drinks in a week, other),maternal smoking before conception (0, 1-9,> 9/day) passive smoking (0, 1-9, > 9/day)and paternal smoking (0, 1-9, > 9/day),employment of mother and father (employedyes or no), and occupational risk of motherand father (yes or no). Eight types of occupa-tional risk (as radiation, chemicals, dust,infection, etc.) were screened in the models.

A possible confounding effect due toother pollutants was also taken into account.The findings of several preliminary studiesshowed that simultaneous inclusion of high-ly correlated variables in one model givesrather misleading results (27,30). The mutu-al correlations between levels of SO2 andsuch pollutants as particulate matter or NOXwere high in both regions studied; therefore,we did not include other pollutants in themodels. Although pollutant levels were asso-ciated with season, season by itself mightalso be a surrogate for other changes such asweather patterns and the consumption offruits and vegetables. We defined summer asthe months from April through Septemberand winter as the months from Octoberthrough March. We also tested other defini-tions of the seasons using four 3-monthannual periods.

VOLUME 108 1 NUMBER 7 July 2000 * Environmental Health Perspectives648

Air pollution and fecundability

Secular and/or seasonal rhythms of con-ceptions in Czech population were estimatedon the basis of 10-year data from official sta-tistics (31); a parameter rhythm was intro-duced, weighing each calendar month by itsrelative contribution in long-time year totals.

Temperature is an important factor thatmay influence fertility (11). For that reason,we introduced temperature in logistic modelsas expressed in four variables: avg (monthlyaverage temperature), avgmaX (monthly aver-age of daily maximum temperatures), max(dummy variable introduced for months thatincluded > 10 days that were warmer than80% of the days in a year), and avgmj,,(monthly average of daily minimum temper-atures). Common respiratiory infectionssuch as influenza are more frequent in thewinter. The sexual behavior of couples(coital frequency) may change during suchperiods. We defined a weekly incidenceabove 1,500/100,000 of acute respiratorydiseases in the Teplice District as an epidem-ic situation. Using a dummy variable epi-demic situation, couples were differentiatedaccording to their exposure to this situationduring the particular month before concep-tion; we tested the variable for inclusion intologistic models.

Because air pollution levels have changedin Teplice District over time, the potentialfor secular changes was evaluated by compar-ing results from the two 2-year periods (peri-od I, from April 1994 to March 1996 andperiod II, from April 1996 to April 1998).

Each month before pregnancy was ana-lyzed separately, allowing some factors tovary over time (e.g., pollution levels, temper-ature, and season). Some factors consideredfor analysis were highly correlated (e.g.,mother's active and passive smoking andpaternal smoking; temperature variables; sea-son and conception rhythms). Therefore,development of the final model used a step-wise approach to select the most appropriatefactors. After this initial stage, we examinedother factors for potential confounding byexamining AORs of SO2 exposures to seehow much change results from inclusionversus exclusion of the potential confounder.Change by > 15% (> 0. 15) for beta was usedas a reasonable criterion for inclusion (29).We used SAS program 6.12 for statisticalanalysis (32).

ResultsThere were 3,651 singleton live births toparental pairs between April 1994 and March1998 in Teplice District. More than onequarter of the parental pairs [964 (26.4%)]admitted they did something to prevent theindex pregnancy and were therefore excludedfrom the study; another 102 were excludedbecause of previous consultation or treatment

for infertility. Of the 2,585 parental pairsincluded, 587 (22.7%) conceived in theFUMC.

Monthly mean levels of SO2 during the4 years of the study are shown in Figure 2.Average concentrations were generally lowestduring summer and highest during winter.Annual mean levels of SO2 decreased con-secutively during the 4 years of the study:1994-1995 = 54.3 ,ug/m3, 1995-1996=50.5 pg/m3, 1996-1997 = 48.9 pg/m3, and1997-1998 = 38.0 pg/m3. The overalldecrease would have been greater had it notbeen for an inversion episode (20 days ofextreme S02 levels from 20 December 1996through 20 January 1997) (Figure 2). Themean levels of SO2 in 2-year periodsdecreased from 53.6 ± 27.6 pg/m3 in periodI to 44.6 ± 35.7 pg/m3 in period II.

Results of descriptive analysis are pre-sented in Table 1. They tend to suggest thatonly a few characteristics of parents whoconceived in the FUMC differ significantlyfrom those of less successful pairs. Couples

who became pregnant in the FUMC weresignificantly more frequently single (32.2 vs.26.7%; p < 0.01) and conceived more oftenduring the summer months (55.2 vs.50.2%; p < 0.05) compared to the others.Mothers from successful pairs were morefrequently unemployed or employed with-out occupational risk.

Table 2 presents the results of multivari-ate analysis of the impact of SO2 exposure indifferent months before conception onfecundability in the FUMC. AORs for con-ception in the FUMC were consistentlyreduced, with higher levels of SO2 in thesecond month (30-60 days) before concep-tion. The AOR for the medium SO2 levelwas 0.57 (CI, 0.37-0.88; p < 0.011); it was0.49 (CI 0.29-0.8 1; p < 0.006) for the highlevel. No significant association wasobserved in any other period before concep-tion (Table 2).

We conducted a similar analysis forthose parental pairs delivering in the firstand the second 2-year periods (n = 1,527

Figure 1. (A) The Czech Republic showing Teplice District. (B) The Teplice District. The dotted circle indi-cates the "in" area (up to 3.5 km away from the monitoring station in the center of the city). Approximatelyone-half of the couples (n = 1,297) were living in this area (according to the permanent residence of themother).

o 150-

;0E. 120a

.CE.90E

xASK'WE O0

60E

30c.

=0

150

eJ40 sU

e

.E

30 t

I._U0

20 oU

10.!Sa.

,Ua CD CD CDQ0 0 0 0

Figure 2. 30-day running averages of S02 levels (in micrograms per cubic meter) and 30-day maximal dailytemperatures (0C) compared to percent conception in the FUMC by the second month before conception.

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

.1

649

Articles * Dejmek et al.

and 1,058 couples, respectively). The onlyclear-cut relationship between fecundabilityand SO2 was observed for pairs delivering inperiod I in the second month before concep-

tion: for medium SO2 levels the AOR =

0.49 (CI, 0.25-0.96; p < 0.037) and for highlevels the AOR = 0.43 (CI, 0.20-0.93; p <

0.033). In period II there was a consistentbut nonsignificant tendency (Table 3).

The variable signal was associated withS02 exposure. Inclusion of this covariateinto logistic models tended to decreaseslightly the AORs for the fourth and first-months before conception; we did not see an

influence on the AORs for the third and sec-

ond preconceptional months. Another wayto reduce possible distortion of the associa-tion between exposure and effect during

inversion episodes was to exclude cases

exposed to inversion from the sample. Afterthe exclusion of 63 parental pairs exposed tomonthly means > 1 10 pg/m3 from the wholesample, the AORs for both medium andhigh S02 levels were further reduced (0.51and 0.42 instead of 0.57 and 0.49), and sig-nificance levels were higher (Table 4). Thisapproach was also applied to the analysis ofsamples for periods I and II. The exclusionof exposures > 110 pg/m3 did not changethe results in periodfor period II changeexposures are exclud(AOR for medium e

0.28-1.07; p < 0.08exposures was 0.4< 0.08). Thus, after

Table 1. Background characteristics of study groups.

Conceived in the FUMC (n = 587) CcVariable Mean ± SD No. (%) MeE

First 2 years of study (period 1)Second 2 years of study (period 11)Ambient levelsa of SO2PeriodPeriod II

Summer at conceptionMaternal characteristicsAge at conception (years)<1920-29.30

Body mass<2222 to <27.27

ParityFirstSecond-third2 Fourth

Previous abortionSpontaneousInduced

EmployedOccupational riskEducationBasicbHigh schoolcUniversity

Currently marriedCurrently single/otherSmoking before conceptionActivePassive

Cigarettes/day among smokersAlcohol before conception 2 1/week

Paternal characteristicsEmployedOccupational riskEducationBasicbHigh schoolcUniversity

SmokingCigarettes/day among smokersAlcohol 2 1/week

49.6 ± 27.243.4 ± 34.5

334 (56.9)253 (44.1)

324 (55.2)*

24.6 ± 4.8 38 (6.5)469 (79.9)

- 80 (13.6)

- 40 (6.8)- 491 (83.6)- 56 (9.6)

1.70 ± 0.97 304 (51.7)- 258 (44.0)

25 (4.3)

0.15± 0.430.49 ± 1.00

11.9 ± 6.1

14.3 ± 7.3

330 (56.1)209 (35.5)

113 (19.3)446 (76.3)26 (4.4)

397 (68.0)187 (32.0)**

226 (38.5)318 (54.2)

106 (20.9)

524 (89.1)302 (51.4)

76 (13.4)463 (81.6)28 (4.9)

312 (55.0)

251 (55.9)

high exposures, the results in the second 2years of the study are similar to those fromthe first 2 years, though with the marginalsignificance only. The results of the alterna-tive analysis of the samples after exclusion ofthe 92 couples exposed to inversions beforeconception yielded similar results (Table 4).Approximately one-half of the couples(1,297) lived up to 3.5 km away from thecentral monitor (classified as "in") (Figure1). Results in Table 5 show that the associa-

I. In contrast, the results tion between conception success in theconsiderably when high FUMC and SO2 exposure in the second pre-

ed (Tables 3 and 4): the conception month is greater for the group

xposures was 0.54 (CI, who lived closer to monitor ("in"). In the) and the AOR for high "in" group, the AOR of conception in the14 (CI, 0. 18-1.09; p FUMC was consistently decreased for medi-exclusion of extremely um (0.56; CI, 0.31-1.00) and high (0.36;

Cl, 0.17-0.73) SO2 exposure. The fecund-ability/SO2 association in the "out" group

onceived later In = 1,998) was weaker and nonsignificant (Table 5).an± SD No. (%) Excluding both the "out" group and the

1,193 (59.7) inversion cases magnified even more the- 805 (40.31) effect of SO2 on fecundability in the second

month before conception (Table 5).54.7 ± 27.748.7 ± 38.9

1,003 (50.2)

24.8 ± 4.6 108 (5.4)- 1,606 (80.4)- 384 (14.2)

- 123 (6.2)- 1,687 (84.3)- 188 (9.5)

304 (51.7) 1,095 (54.8)833 (41.7)70 (3.5)

Anl n.ARu.0.3E

11.

14.

0 T U.14+±0.72 -

- 1,552 (59.5)- 1,020 (39.1)

- 368 (18.4)1,547 (77.5)

- 81(4.1)- 1,460 (73.3)- 531 (26.7)

- 768 (38.4)- 1,110 (55.6)

.7 ±7.0- 360 (20.4)

- 2,310 (88.6)- 1,354 (51.9)

- 275 (14.2)- 1,569 (80.8)

97 (5.0)- 1,081 (56.1)

.8 9.8- 880 (57.4)

DiscussionTeplice District lies in a highly industrializedmining area of Northern Bohemia withheavy industry and many large power plantsusing low-energy-quality brown coal withhigh sulfur content. Pollution reaches itshighest levels during meteorologic inver-sions, which are not infrequent events in thismountainous area. Ambient air monitoringtends to suggest that SO2 levels were fallingduring the 4 years of this study. This trendseems mainly to be due to changes in indus-try profiles, technological improvement oflarge power plants, and a rapid conversion oflocal heating systems from coal to gas in theTeplice area.

The longitudinal version of the TTPmethod presents a problem for evaluating theimpact of air pollution on fecundability. Allpotential parents were continually exposed invarious periods before a particular concep-

tion and the levels of pollutants are continu-ally changing. Appropriate comparisons forcouples conceiving in the FUMC are easierto achieve. Baird et al. (9) suggested thatdata about the proportion conceiving in thefirst cycle give an unbiased estimate of themean fecundability in the cohort, providinga rationale for the design of the present

study. In this approach, information con-

cerning the distribution of later conceptionsis lost. On the other hand, a bivariate out-

come measure and yes or no responses makeit possible to apply logistic regression (J$, a

powerful device for controlling potentiallyconfounding covariates. It should be empha-sized that the approach used is based on dataobtained at delivery; therefore, all early losses

VOLUME 108 1 NUMBER 7 July 2000 * Environmental Health Perspectives

aAmbient levels of pollutants during the second month before conception. bBasic education = approximately 9 years.cHigh school (with maturity exam) = approximately 11-12 years in the Czech Republic during the relevant period. *p <0.05. **p < 0.01.

650

Air pollution and fecundability

were included within the nonconceivedgroup; on the other hand, all conceptionsthat resulted in clinical spontaneous orinduced abortion were omitted.

We used the EDC as a reference date foranalysis. Stolwijk et al. (33) recently showedthat this approach might involve substantialbias arising from a seasonal pattern of preg-nancy planning. The authors instead recom-mended using the date of onset of TTP,which is not biased in this manner. This rec-ommendation cannot be followed in the pre-sent study, as data about TTP onset cannotbe obtained from couples with longer con-ception delay. Therefore, we included annualconception rhythms observed longitudinallyin the Czech population in logistic models;thus a possible influence of seasonality inpregnancy planning was reduced. Accordingto Stolwijk et al. (33), a residual effect of thisbias can cause an underestimation of thestrength of the relation, but not a change inthe direction of the effect estimators.

Results of descriptive analysis suggestedthat parental pairs conceiving in the FUMCdid not differ from less successful couples inmost of the characteristics listed in Table 1.Mothers conceiving in the FUMC weremore frequently single and unemployed.The employed women from this group wereless likely to be exposed to occupational risk.After the inclusion of particular occupationalrisks into logistic models, theAOR of concep-tion in the FUMC was significantly reducedfor mothers exposed to any risk (0.75; CI,0.58-0.97), and to radiation (0.30; CI,0.69-1.12); the same is true for fathersexposed to dust (0.64; CI, 0.46-0.89).Inclusion of these variables in the models didnot influence the final association betweenSO2 and fecundability in the FUMC.Success in the FUMC is more frequent dur-ing the summer (24.4%) than during thewinter (20.9%). This could be due to betteropportunities for intercourse during vacationtime and to other supporting influences ofsummer (10). Moreover, couples successfulin the FUMC were exposed to lower levelsof SO2 during the second month before con-ception (Table 1). Thus, association betweenSO2 and fecundability may also contributeto higher conception success in the FUMCduring summer.

We used monthly means of SO2 to char-acterize the exposure. We examined shorteras well as longer periods in preliminaryinvestigations to find an optimal measure.Misclassification of exposure estimates maybe frequent, and the accuracy of the EDC islimited in the present study. An intervalshorter than 30 days is inadequate to ensurethe reliability of data. On the other hand,the biologic sensitivity window (e.g., in thecase of influence on spermatogenesis) can be

relatively narrow, on the order of weeks. Tohit this small target using such an inaccuratedevice requires choosing the optimal interval,such as 1 month, because only analyses basedon 30-day periods yielded consistent results.

Ambient SO2 is only one of many com-ponents of a complex mixture. The possibleeffects of SO2 and four other noxae, namelyNOx, particulate matter < 10 pm in aerody-namic diameter (PM10), particulate matter< 2.5 pm in aerodynamic diameter, and poly-cyclic aromatic hydrocarbons, on fecundabili-ty were examined in two preliminary studiesusing the same approach (27,30). Levels of all

five noxae were mutually highly correlated ina range of 0.55-0.83. Analyzing each pollu-tant in a separate model, we observed theonly consistent relationship to fecundabilitywas that for SO2. A much weaker associationwith PM1O could be explained by the highcorrelation between SO2 and PM1O levels (r= 0.83; p < 0.0001). An analysis of thesimultaneous effects of the two pollutants inone model yielded an increased SO2 effect(AORs for SO2 40-80 pg/m3 = 0.53; CI,0.39-81 and AORs for SO2 > 80 pg/m3 -

0.41; CI, 0.25-0.70) and eliminated anysuggestion of PM1O association (AORs for

Table 2. AORs of the fecundability in the FUMC by exposure to SO2 before conception.

Mediuma HighaMonth AORb Cl p-Value AORb Cl p-Value- 4 1.32 0.90-1.91 0.15 0.93 0.57-1.51 0.78- 3 0.95 0.63-1.48 0.75 0.90 0.55-1.48 0.68- 2 0.57 0.37-0.88 0.011 0.49 0.29-0.81 0.006- 1 1.01 0.68-1.51 0.95 0.96 0.58-1.58 0.86

Four years: April 1994 to March 1998, European births in Teplice (n = 2,585)."Low < 40 pg/mr3 (reference level); medium 40 to < 80 pg/M3; and high 2 80 pg/M3. bAdjusted for maternal age, parity, con-ception seasonality, currently married, temperature average, temperature maxima, signal, year and season, and epidemicsituation.

Table 3. AORs of the fecundability in the FUMC by exposure to S02 before conception.

Mediuma HighaMonth AORb Cl p-Value AORb Cl p-ValueFirst 2 years of studyc

- 4 1.58 0.85-2.74 0.16 1.26 0.58-2.71 0.56- 3 0.88 0.49-1.57 0.66 0.86 0.41-1.82 0.70- 2 0.49 0.25-0.96 0.037 0.43 0.20-0.93 0.033- 1 1.14 0.67-1.97 0.62 1.20 0.58-2.48 0.62

Second 2 years of studyd- 4 0.90 0.51-1.61 0.74 0.88 0.41-1.85 0.73- 3 0.85 0.45-1.57 0.59 0.96 0.45-2.03 0.91- 2 0.67 0.36-1.28 0.22 0.59 0.38-1.28 0.20- 1 1.16 0.59-2.29 0.66 1.15 0.59-3.59 0.31

'Low < 40 pg/m3 (reference level); medium 40 to < 80 pg/M3; and high 2 80 pg/M3. bAdjusted for maternal age, parity, con-ception seasonality, currently married, temperature average, temperature maxima, signal, year and season, and epidemicsituation. cApril 1994 to March 1996, European births in Teplice (n = 1,527). dApril 1996 to March 1998, European births inTeplice (n= 1,058).

Table 4. Effect of exclusion of extremely high exposures to S02 in the second month before conception.

SO2 exposure Medium" High"in the second month AORb Cl p-Value AORb Cl p-ValueAll exposedc

0.49 0.25-0.96 0.037 0.43 0.20-0.93 0.03311 0.67 0.36-1.28 0.22 0.59 0.35-1.28 0.20Total 0.57 0.37-0.88 0.011 0.49 0.29-0.81 0.006

Exposed to < 110 plm3d0.49 0.25-0.96 0.038 0.43 0.20-0.93 0.033

11 0.54 0.28-1.07 0.08 0.44 0.18-1.09 0.08Total 0.51 0.33-0.80 0.004 0.42 0.25-0.71 0.001

Exposed to inversion situatione0.49 0.25-0.96 0.037 0.43 0.19-0.93 0.033

11 0.54 0.27-1.09 0.08 0.41 0.17-0.99 0.049Total 0.51 0.33-0.81 0.004 0.42 0.25-0.72 0.002

First and second 2-year periods of study.kLow < 40 pg/mi3 (reference level); medium 40 to < 80 pg/M3; and high 2 80 pg/M3. bAdjusted for maternal age, parity, con-ception seasonality, currently married, temperature average, temperature maxima, signal, year and season, and epidem-ic situation. Cn = 2,585. dn= 2,522. All mean month SO2 exposures> 110 pg/m3 occurred during the inversion episode from15 December 1996 through 20 January 1997; thus, only results for the second 2-year period 11 (April 1996 to March 1998)were influenced. On = 2,493. Results of analysis after exclusion of couples exposed to the inversion situation did not differfrom those calculated after exclusion of extremely exposed parents. Either extremely high exposures or altered behaviorof people during the inversion episode 15 December 1996 through 20 January 1997 might distort the S0O/fecundabilitydetected in this study.

Environmental Health Perspectives * VOLUME 1081 NUMBER 7 July 2000 651

Articles * Dejmek et al.

Table 5. Influence of the distance from the monitor on AOR of fecundability in the second month beforeconception.

Mediuma HighaDistance AORb Cl p-Value AORb Cl p-Value< 3.5 km 0.56 0.31-1.00 0.05 0.36 0.17-0.73 0.005> 3.5 km 0.58 0.31-1.08 0.09 0.70 0.34-1.45 0.34Total 0.57 0.37-0.88 0.011 0.49 0.29-0.81 0.006Combined effectc 0.51 0.27-0.95 0.034 0.28 0.13-0.61 0.0012

European births in Teplice: < 3.5 km, n= 1,297; > 3.5 km, n = 1,288."Low < 40 pg/rM3 (reference level); medium 40 to < 80 pg/M3; and high 2 80 pg/M3. bAdjusted for maternal age, parity, con-ception seasonality, currently married, temperature average, temperature maxima, signal, year and season, and epidem-ic situation. cCombined effect of the limitation of the distance of residence from the monitoring station (< 3.5 km) and thelimitation of considered exposure to S02 (< 1 10 pg/M3).

PM1O 40-50 pg/m3 = 1.18; CI 0.90-1.56and AORs for PMIO 2 50 pg/m3=1.30; CI,0.92-1.93) (30). These results would be mis-leading in view of the high correlation of bothvariables. Therefore, the analysis was concen-trated on the effect of SO2 alone in the pre-sent communication. The rather complexquestion of the simultaneous effects of copol-lutants will be discussed in a separate study.

Multivariate analysis of data from all 4years showed a relatively strong inverse asso-ciation between the concentration of SO2during the second month before conceptionand conception success in the FUMC (Table2). The AOR of conception in the FUMCwas reduced to 0.57 for couples exposed toSO2 levels of 40-80 pg/m3 in the secondmonth before conception. This value falls to0.49 for those exposed to levels > 80 pg/m3in the same preconceptional stage. A similarrelationship was also observed when data weredivided into two subsets corresponding to thefirst (period I) and second (period II) 2-yearperiods of the study (Table 3). However, theresults show unequivocally that the associa-tion is much stronger in period I. The par-ticular AOR values were 0.49 and 0.43 formedium and high SO2 exposures; lower thanfor the total 4-year sample (Table 2). On theother hand, the association in period II wasmuch weaker; in fact, it was no longer signifi-cant. This tendency may be rooted in thepermanent decline ofSO2 levels in the regionduring the past several years.

Misclassification in exposure estimates isa usual weakness in studies based on central-ized monitoring. To reduce the deleteriouseffects of misclassification, we conducted asensitivity analysis. The variable signal shouldcorrect possible exposure-protective behaviorsof people to the warning system duringinversion episodes. However, an introductionof this variable into logistic models did notaffect the results in any preconception peri-od. Another method used to control theinfluence of possible protective behavior ofinhabitants during inversion episodes was theexclusion of inversion-exposed cases from theanalysis. This approach seemed to clarify theexposure/outcome associations. An exclusionof parental pairs exposed to monthly means

of SO2 > 110 pg/m3 (63 couples) had thesame effect as the exclusion of 92 coupleswho were exposed to an inversion episodebefore pregnancy (Table 4). The reductionofAORs for medium and high exposures inthe second month before conception wasstronger and more significant. It may be par-ticularly important that the differencebetween the effect of SO2 on fecundabilityin periods I and II was reduced after control-ling for the influence of inversion situations.

We derived the mean monthly exposureof couples to SO2 from daily measurementsby a central monitor. Exposure estimatesshould be more accurate for persons livingnear the monitoring station than for others,irrespective of the distance of pollutionsources or wind direction. If the relationshipbetween SO2 and fertility is real, the statisti-cal association between estimated levels ofSO2 and fecundability derived from logisticmodels should be stronger for the "in" sam-ple. The present results show unequivocallythat the relationship between SO2 and fertil-ity for couples living < 3.5 km from themonitor station is much stronger than forother pairs (Table 5). Approximately one-half of the inhabitants of the district wereliving inside and the other half outside thisarea (1,297 and 1,288, respectively): thismade the cutoff optimal for comparison ofboth subsamples. This result strengthens thehypothesis that some part of the variation infecundability may be explained due to thechanges in SO2 levels.

It has been suggested that temperaturemay affect human fertility (11). Temperaturemay influence hormonal levels (34), frequen-cy of intercourse (11), and spermatogenesis(35). In our study, average as well as maximaltemperatures in the second preconceptionalmonth showed significant association withfecundability in the FUMC in models with-out pollutants (p < 0.003 and p < 0.03); noconsistent association of fecundability withtemperature during other months wasobserved. In models containing SO2, bothmaximal and average temperatures signifi-cantly influenced AORs, especially in thesecond month before conception. On theother hand, the influence of SO2 in models

without temperature reduced the fecundabil-ity significantly although the AORs werelower than in the complete model [mediumSO2 level AOR = 0.67 (CI, 0.44-0.96; p< 0.03); high SO2 level AOR = 0.60 (CI,0.38-0.97; p < 0.03)]. Minimal tempera-tures did not affect the S02/fecundabilityassociation and were not included into finalmodels. Success in the FUMC was observedsignificantly more frequently in the summer(from April through September) than in thewinter (Table 1). Conception rates in theFUMC were also positively related to thewarmest periods in the logistic models used.These observations are not surprisingbecause the conditions for fertilization aregenerally more favorable during the warmermonths in temperate latitudes (36). Loosesummer clothing enables better scrotal cool-ing, which optimizes spermatogenesis andsperm quality. Other influences such ashigher coital frequency in the summer canalso contribute to this seasonal fluctuation(37). It seems that seasonal epidemics ofrespiratory diseases may influence fecund-ability. The observed fecundability/S02association in the second month before con-ception decreased slightly after controllingfor the influence of epidemic situations. Onthe other hand, AORs for other periodsbefore conception tended to decline to unityas if those epidemics may explain a part ofthe variability in fertility. Some changes insexual behavior during such epidemics canbe hypothesized. Thus, fecundability mayvary due to many seasonal factors other thanpollution.

Currently, few papers have examinedpossible associations between human fertilityand air pollution. Recently, reduced birthrates were observed during periods of highSO2 concentrations in the Teplice region(6,7). It has been suggested that high SO2 orsome associated pollutants may reduce theability of oocytes to be fertilized. Jagiello (8)previously observed this effect experimental-ly. Srim et al. (7) hypothesized that an effectof other environmental mutagens, inducingmutations in gametes or early embryos, maybe potentiated by SO2 due to the suppres-sion ofDNA repair mechanisms.

Whether SO2 affects fertility throughgenetic mechanisms remains an open ques-tion. There are conflicting data about thegenotoxic effects ofSO2 in humans. Schneiderand Calkins (38), Nordensen et al. (39) andYadav and Kaushik (40) observed dastogeniceffects in workers exposed to high concentra-tions of SO2. Chromosomal aberrations andsister chromatid exchanges (SCEs) in exposedworkers were observed by Meng and Zhang(41). On the other hand, Sorsa et al. (42observed neither chromosomal aberrationsnor SCEs.

VOLUME 108 1 NUMBER 7 1 July 2000 * Environmental Health Perspectives652

Air pollution and fecundabilit

Little is known about the backgroundvariability in fecundability in the normalhuman population (10,17,26). It is clear thatdelays in conception may result from a spec-trum of pathogenetic processes in one orboth sexes (17). The present approach can-not differentiate among the processesinvolved; therefore, we can only speculateabout mechanisms of the observed adverseeffects of SO2. However, the timing of theacute effect coincides with some importantstages of the reproductive process in men,namely sperm maturation. This finding isinteresting in light of the recent results ofSelevan et al. (43), in which semen qualitywas repeatedly analyzed in healthy youngmen from Teplice. Highly significant adversebut transient effects of increased S02 levelson sperm morphology and motility duringone spermatogenic cycle were observed inmultivariate analysis. The results tended tosuggest an effect on spermatogenesis ratherthan an acute influence on epididymalsperm function. These effects should operateduring transformation of the round sper-matids into differentiated sperm cells (43).These results support our findings withregard to the timing of conception: it isnoteworthy that in our results a significantreduction of the conception rate in theFUMC was associated with S02 levels onlyin the second month before fertilization.This is the same period described by Selevanet al. (43) as the most probable stage ofsperm maturation damage associated withS02 exposure. In addition, these authors'results contribute to the substantive questionof the relationship between S02 and fertility:there are indications that the type of spermdamage observed by Selevan et al. (43) mayactually reduce fertility (44).

In spite of this late preconception peri-od, an interference of SO2 or some copollu-tant with the respective stages of oogenesiscannot be excluded. Damage to gametesmight reduce the efficiency of reproduction,subsequently affecting fertilization, implan-tation, and early embryogenesis. All of thoseevents could increase the risk of early losses(subclinical abortion). On the other hand,toxic pollutants may also interfere withthese processes directly.

We are aware that our results are basedon some data of unequal reliability. Somequestionnaire data (parental ages, date ofdelivery, and other personal data) were cross-checked using other information sources.Pollution data were measured using the reg-ularly calibrated equipment and standard-ized methods developed by the U.S. EPA;these data should be reliable. However, wemade an important assumption in theassignment of exposure to different timeperiods: the measured exposure level at the

central monitoring station may not be entire-ly representative of an individual woman's orman's exposure. Wide variations in exposuremay result from individuals living at variousdistances from the monitor, varying windconditions, varying personal habits, and dif-ferences in daily routine. Exposure estimateswere more relevant for couples living near themonitoring station (Table 4). The expo-sure-effect relationship may be distorted inperiods with extremely high levels of pollu-tion (inversions).

Another possible source of error in expo-sure estimation may be incorrect determina-tion of the EDC. Errors in EDC could bluror wholly dissolve the exposure-period rela-tionship; systematic error may shift theimportant exposure to another time period.To prevent this, we made the EDC determi-nation using maternal prenatal recordsobtained in early pregnancy. Thus, systemat-ic errors in the EDC are less likely.

ConclusionsThe results of the present study suggest thatAORs of conception in the FUMC may bereduced in couples exposed to mean S02levels > 40 pg/m3 in the second monthbefore conception. No consistent relation-ship was observed in any other period duringthe 4 months before conception. The expo-sure-effect association tends to be strength-ened by the exclusion of couples living largerdistances from the monitor station and/orcouples exposed to extreme inversion situa-tions, when behavior may alter exposures.The timing of the effect coincides with thesperm maturation period. These results arein agreement with the findings of Selevan etal. (43), who observed spermatogenesis dam-age in the same stage in young healthy menexposed to ambient SO2; the observed typesofdamage may reduce fertility.

The impact of preconceptional exposureto other pollutants will be evaluated in thefuture using the same methods.

REFERENCES AND NOTES

1. Bobak M, Leon DA. Pregnancy outcomes and outdoorlevels of air pollution: an ecological study in districts ofCzech Republic 1986-1988. Occup Environ Med56:539-543 (1999).

2. Xu X, Ding H, Wang X. Acute effects of total suspendedparticles and sulfur dioxides on preterm delivery; a com-munity based cohort study. Arch Environ Health50:407-415 (1995).

3. Wang X, Ding H, Ryan L, Xu X. Association between airpollution and low birth weight: a community-based study.Environ Health Perspect 105:514-520 (1997).

4. Woodruff TJ, Grillo J, Schoendorf KC. The relationshipbetween selected causes of postneonatal infant mortali-ty and particulate air pollution in the United States.Environ Health Perspect 105:608-612 (1997).

5. Dejmek J, Selevan SG, Benes I, Solansky I, tram RJ.Fetal growth and parental exposure to particulate matterduring gestation. Environ Health Perspect 107:475-480(1999).

6. Sr6m R. New ethical problems related to environmentalpollution and behavioral changes in human population.In: Ethical Issues of Molecular Genetics in Psychiatry.(Sr6m RJ, Bulyzhenkov V, Prilipko 1, Christen Y, eds).Berlin, Heidelberg:Springer Verlag, 1991;94-105.

7. Sram RJ, Roinickova I, Albrecht V, Ber6nkova A,Machovska E. Monitoring congenital anomalies in popu-lations exposed to environmental mutagens. In:Mechanisms of Environmental Mutagenesis-carcino-genesis (Kappas A, ed). New York:Plenum Press,1991;255-266.

8. Jagiello GM. SO2 and its metabolite: effect on mam-malian egg chromosomes. Environ Res 9:84-93 (1975).

9. Baird DD, Wilcox AJ, Weinberg D. Use of time to preg-nancy to study environmental exposures. Am J Epidemiol124:470-480 (1986).

10. Steiner M, Dominik R, Trussel J, Hertz-Picciotto I.Measuring contraceptive effectiveness: a conceptualframework. Obstet Gynecol(3 suppl):24S-30S (1996).

11. Bronson FH. Seasonal variation in human reproduction:environmental factors. Q Rev Biol 70:141-164(1995).

12. Rowland A, Baird DD, Wilcox AJ. Reduced fertilityamong women employed as dental assistants exposed tohigh levels of nitrous oxide. N EngI J Med 327:993-997(1992).

13. Joffe M. Male and female-mediated reproductive effectson occupation: the use of questionnaire method. JOccup Med 268:74-81 (1989).

14. Wilcox A, Weinberg C, Baird D. Caffeinated beveragesand decreased fertility. Lancet 31:1453-1455 (1988).

15. Olsen J. Cigarette smoking, tea and coffee drinking, andsubfecundity. Am J Epidemiol 133:734-739 (1991).

16. Ratclife JM, Gladden BC, Wilcox AJ, Herbert AL. Doesearly exposure of maternal smoking affect future fertilityin adult males? Reprod Toxicol 6:297-307 (1992).

17. Joffe M. Time to pregnancy: a measure of reproductivefunction in either sex. Occup Environ Med 54:289-295(1997).

18. Baird DD, Wilcox AJ. Cigarette smoking associated withdelayed conception. JAMA 253:2979-2983 (1985).

19. Weinberg CR. Pitfalls inherent in retrospective time-to-event studies: the example of time to pregnancy. StatMed 12:867-879 (1993).

20. Joffe M, U Z. Association of time to pregnancy and theoutcome of pregnancy. Fertil Steril 62:71-75 (1994).

21. Joffe M, Villard L, Li Z, Plowman R, Vessey M. A time topregnancy questionnaire designed for long term recall:validity in Oxford, England. J Epidemiol Commun Health49:314-319 (1995).

22. Weinberg CR, Hertz-Picciotto I, Baird DD, Wilcox AJ.Efficiency and bias in studies of early pregnancy loss.Epidemiology 3:17-22 (1992).

23. Weinberg CR, Baird DD, Wilcox AJ. Sources of bias instudies of time to pregnancy. Stat Med 13:674-681(1994).

24. Witschi E. Overripeness of the egg as a possible causein mental and physical disorders. Soc Biol 18:S9-S15(1971).

25. Wilcox AJ. Surveillance of pregnancy loss in humanpopulation. Am J Ind Med 4:285-291 (1983).

26. Baird DD, Weinberg CR, Rowland AS. Reporting errors intime-to-pregnancy data collected with a short question-naire. Am J Epidemiol 133:1282-1290 (1991).

27. Dejmek J, Jelinek R, Benes I, Sram RJ. Impact of air pol-lution on human fertility: accute effects of SO2, NOX andPM,, [Abstract]. Epidemiology 9(4):S148 (1998).

28. Pinto JP, Stevens RK, Willis RD, Kellogg R, Mamane Y,NovAk J, Santroch J, Benes I, Lenitek J, Buret V. CzechAir Quality Monitoring and Receptor Modeling Study.Environ Sci Technol 32:843-854(1998).

29. Rothman KJ, Greenland S. Modern Epidemiology, 2nd ed.Philadelphia,PA:Lippincott-Raven Publishers, 1998;256-257.

30. Deimek J, Jelinek R, Benes I, Solansky I, Sram RJ. Airpollution exposure before conception and human fertility[Abstract]. Pharmacol Toxicol 85(suppl 1):37 (1999).

31. Bureau of Statistics of Czech Republic, ed. Moving ofInhabitants. Births [in Czech]. Prague:State Bureau ofStatistics of Czech Republic, 1988-1997.

32. SAS Institute Inc. SAS/STAT Software. Changes andEnhancements for Release 6.12. Cary, NC:SAS InstituteInc., 1997.

33. Stolwijk AM, Straatman H, Zielhuis GA, Jongbloet PH.Seasonal variation in the time to pregnancy: avoidingbias by using the date of onset. Epidemiology 7:156-159(1996).

Environmental Health Perspectives * VOLUME 1081 NUMBER 71 July 2000 653

Articles * Dejmek et al.

34. Smals AGH, Kloppenborg WC, Bernaad TJ. Circannualcycle in plasma testosterone levels in man. J ClinEndocrinol Metab 42:979-982 (1976).

35. Levine R. Male factors contributing to the seasonality ofhuman reproduction. Ann NY Acad Sci 709:29-45 (1994).

36. James WH. Seasonal variation in human births. J BiosocSci 22:113-119 (1990).

37. Reinberg A, Lagoguey M. Circadian and circannualrhythms in sexual activity and plasma hormones (FSH,LH, testosterone) of five human males. Arch Sex Behav7:13-22 (1978).

38. Schneider LK, Calkins CA. Sulfur dioxide induced lym-phocyte defects in human peripheral blood cultures.Environ Res 3:473-483 (1970).

39. Nordensen IG, Beckman L, Rosenhall L, Stjenberg N. Isexposure to sulfur dioxide clastogenic? Hereditas93:161-164 (1980).

40. Yadav JS, Kaushik VK. Effect of sulphur dioxide exposureon human chromosomes. Mutat Res 359:25-29 (1996).

41. Meng Z, Zhang L. Chromosomal aberrations and sisterchromatide exchanges in lymphocytes of workersexposed to sulfur dioxide. Mutat Res 241:15-20 (1990).

42. Sorsa M, Hedman BK, Jarventaus H. No effect of sulfurdioxide exposure, in aluminium industry, on chromoso-mal aberrations and sister chromatide exchanges.Hereditas 95:159-161 (1982).

43. Selevan SG, Borkovec L, Slot VL, Zudova Z, Rubes J,Evenson DP, Perreault SD. Unpublished data.

44. Ayala C, Steinberger E, Smith DP. The influence ofsemen analysis parameters on the fertility potential ofinfertile couples. J Androl 17:718-725 (1996).

Excellencein basic research at the

National Institute of Environmental Health Sciences

http:Ilw"wwniehs.nih.gov/

N IEHS scientists and grantees are performing basic studies of our susceptibility to environment-related disease: demonstrating that a carcinogen in cigarette smoke (benzo(a)pyrene) alters part of agene to cause lung cancer ... showing the effects of fetal exposure to PCBs ... developing a strain ofmouse that lacks functional estrogen receptors and that helps evaluate how some pesticides and otherestrogen-like compounds might affect development and reproduction . . . discovering the genes forbreast, ovarian, and prostate cancers ... identifying women's optimal days of fertility ... seeking to reversethe damage from lead exposure ... finding alternatives to traditional animal tests ... pinpointing thefunctions of specific genes by eliminating them from specially bred mouse lines ... discovering a way, usingordinary yeast cells, to isolate and clone genes and other fragments of genetic material more quickly ...showing the effects of urban air on lung function ...

A part of the National

Institues of Health,

the Nadonal Instiute

ofEnvironmental Health

Sciences Is located In

Research Triangle Park,North Carolina.

654 VOLUME 1081 NUMBER 7 1 July 2000 * Environmental Health Perspectives