Description of trihalomethane levels in three UK water suppliers - Nature

Journal of Exposure Analysis and Environmental Epidemiology (2003) 13, 17 – 23 # 2003 Nature Publishing Group All rights reserved 1053-4245/03/$25.00

www.nature.com/jea

Description of trihalomethane levels in three UK water suppliersy HEATHER WHITAKER,a MARK J. NIEUWENHUIJSEN,b NICOLA BEST,a JOHN FAWELL,c ALISON GOWERSd AND PAUL ELLIOT a a Small Area Health Statistics Unit, Department of Epidemiology and Public Health, Imperial College School of Medicine at St. Mary’s, Norfolk Place, London W2 1PG, UK b Department of Environmental Sciences and Technology, Royal School of Mines, Imperial College, Prince Consort Road, London SW7 2BP, UK c Environmental Division, Warren Associates, 8 Prince Maurice Court, Devizes, Wiltshire SN10 2RT, UK d National Centre for Environmental Toxicology, WRc - NSF Limited, Henley Road, Medmenham, Marlow, Bucks SL7 2HD, UK

Samples of drinking water are routinely analysed for four trihalomethanes ( THMs ), which are indicators of by - products of disinfection with chlorine, by UK water suppliers to demonstrate compliance with regulations. The THM data for 1992 – 1993 to 1997 – 1998 for three water suppliers in the north and midlands of England were made available for a UK epidemiological study of the association between disinfection by - products and adverse birth outcomes. This paper describes the THM levels in these three supply regions and discusses possible sources of variation. THM levels varied between different suppliers’ water, and average THM levels were within the regulatory limits. Chloroform was the predominant THM in all water types apart from the ground water of one supplier. The supplier that distributed more ground and lowland surface water had higher dibromochloromethane ( DBCM ) and bromoform levels and lower chloroform levels than the other two suppliers. In the water of two suppliers, seasonal fluctuations in bromodichloromethane ( BDCM ) and DBCM levels were found with levels peaking in the summer and autumn. In the other water supplier, chloroform levels followed a similar seasonal trend whereas BDCM and DBCM levels did not. For all three water suppliers, chloroform levels declined throughout 1995 when there was a drought period. There was a moderate positive correlation between the THMs most similar in their structure ( chloroform and BDCM, BDCM and DBCM, and DBCM and bromoform ) and a slight negative correlation between chloroform and bromoform levels. Journal of Exposure Analysis and Environmental Epidemiology ( 2003 ) 13, 17 – 23 doi:10.1038/sj.jea.7500252 Keywords: chlorination, disinfection by - products, drinking water, trihalomethanes.

Introduction Disinfection is one of the important barriers to prevent microbiological pathogens from entering drinking water, and in the UK, all public drinking water supplies are disinfected. There are a variety of drinking water disinfectants available including chlorine, ozone, chloramine, ultraviolet radiation, and chlorine dioxide, but in the UK and many other parts of the world, chlorine is by far the most widely used (Breach, 1996 ). The chlorine, or chlorine hydrolysed to hypochlorous acid, reacts with natural organic matter and bromide in the raw water via halogen substitution and oxidation reactions forming a wide range of compounds known as disinfection by- products (DBPs ). Typically, the most common DBPs are the trihalomethanes (THMs ); these

1. Address all correspondence to: Dr. Heather Whitaker, Department of Statistics, Faculty of Mathematics and Computing, The Open University, Walton Hall, Milton Keynes, MK7 6AA, UK. Tel.: + 44 - 1908 - 654370. E - mail: [email protected] Received 19 August 2002. y The views expressed in this publication are those of the authors and not necessarily of the funding departments.

are a group of compounds including chloroform ( CHCl3 ), bromodichloromethane (BDCM; CHBrCl2 ), dibromochloromethane ( DBCM; CHBr2Cl), and bromoform (CHBr3 ). Hundreds of other DBPs have also been identified including the haloacetic acids, haloacetonitriles, haloketones, aldehydes, and chloral hydrate (Richardson, 1998 ). However, in the UK, only THMs are regulated (and therefore measured to any great extent ) and are used as an indicator of total DBP concentration. There are a number of determinants in the raw water, during treatment and in distribution that impacts on the DBP levels reaching the tap. Factors that affect DBP production in the raw water include bromide levels (Dore´ et al., 1988; Symons et al., 1993 ) and the amount and nature of the natural organic matter (Reckhow and Singer, 1985; Krasner et al., 1996 ), which depends on water source type. Around 70% of UK drinking water is derived from surface sources such as upland reservoirs and lowland rivers and reservoirs. The amount of natural organic matter in these surface waters is very variable. The remaining 30% of UK drinking water is derived from groundwater sources, which tend to contain low levels of natural organic matter. Bromide levels are high in some ground and lowland surface waters, but are low in

Whitaker et al.

Trihalomethane levels in UK water suppliers

Table 1. Limits of detection. Water company

THM

Limit of detection ( g / l ) 1993 – 1995

North West chloroform 2.5 Water BDCM 2.5 1992 – 1997 DBCM 2.5 bromoform 2.5 Severn Trent chloroform 0.46 Water BDCM 0.74 1993 – 1998 DBCM 0.71 bromoform 0.96 Northumbrian chloroform 3 Water BDCM 3 1993 – 1998 DBCM 3 bromoform 3

1996 – 1998

2.5 2.5 2.5 2.5 1.59 0.57 0.43 0.46 6 4 3 2

Percent zones with all samples under the limit of detection in a given year 3.48 4.88 31.96 84.79 23.2 22.14 15.93 21.16 4.6 5.8 81.9 78.4

upland surface waters. High levels of bromide in UK ground waters tend to occur in aquifers subject to saline intrusion, although geology has some influence; other sources of excess bromide particularly affecting lowland sources include industrial and agricultural chemicals. During treatment, factors affecting the type and levels of DBPs formed include the treatment practices used, the water pH and temperature, the type and dose of disinfectants used, and the point in the treatment process

where the disinfectant is added (Krasner, 2001 ). In the UK, most surface waters are subject to coagulation and filtration processes. Many lowland surface waters are also subject to advanced treatments such as filtration through granular activated carbon (Breach, 1996 ), which contributes to the removal of a substantial amount of the natural organic matter. Ground waters are usually subject to minimal treatment before entering the distribution system, depending on the extent to which they are influenced by surface water and the consequent risk of contamination. In the distribution system, further DBP formation will depend on the water temperature, pH, level of free chlorine, and residence time. These factors affect each of the DBPs differently, but generally THM levels increase over time, and with higher pH values and temperatures ( Chen and Weisel, 1998 ). There has been concern over the potential adverse health effects of a number of DBPs (Nieuwenhuijsen et al., 2000 ). A large study to examine the possible association between DBP levels and adverse health outcomes in the UK is currently being carried out, using THMs as a marker for DBPs. In this paper, we present descriptive analyses of the exposure data for this study from three water suppliers — North West Water (now known as United Utilities), Severn Trent Water, and Northumbrian Water — which were collected for routine monitoring purposes. We present patterns of seasonal variation in the data, which are of importance to epidemiological studies

Figure 1. Distribution of annual zone mean THM levels ( g / l ) for North West Water ( 1992 – 1997 ), Severn Trent Water ( 1993 – 1998 ), and Northumbrian Water ( 1993 – 1998 ). The box is bounded by the lower and upper quartiles; the line across the middle is the median. The whiskers extend to the minimum and maximum values. 18

Journal of Exposure Analysis and Environmental Epidemiology (2003) 13(1)


Whitaker et al.

Table 2. Mean THM levels ( g / l ). Water supplier

Chloroform

BDCM

DBCM

Bromoform

North West Water 1992 – 1997 Severn Trent Water 1993 – 1998 Northumbrian Water 1993 – 1998

36.1

7.9

2.8

1.8

17.9

8.1

5.3

2.4

38.1

6.4

2.4

1.8

of birth defects for which the aetiologically relevant exposure period is 3 months; correlation between THM levels, which show whether one THM may be used as marker for another; and other descriptive analyses, which help us to determine how to treat our exposure data. Insight into the reasons behind the variation in THM levels and implications of our findings for our epidemiological study are discussed.

Methods The Data Water suppliers in the UK are required by government regulations to keep their total THM levels below 100 g/ l on a 3 -month rolling average. Each water supplier’s region is split into water supply zones, which are limited to a maximum population of 50,000. These water supply zones are established specifically for regulatory monitoring and reporting purposes and should represent areas supplied by water of the same quality. At any time, water supplied within a zone is from the same sources, although these sources may change over time. Tap water samples are routinely collected from a random or fixed point in each water zone. After allowing the tap to run, water samples are collected in vials containing a preservative and free

chlorine is removed. The vials are filled to the top to ensure that the volatile THMs are not lost into the headspace. Solvent or headspace extraction is followed by analysis by gas chromatography and mass spectrometry or electron capture detection. The UK Drinking Water Inspectorate operates an audit scheme to check the data collection, and analytical and archiving procedures. For most water supply zones, THM levels are monitored at least four times a year, though where THMs levels are known to be low, samples may be collected as little as once a year, or where THM levels are particularly high, water suppliers monitor THM levels more often. The routinely collected data on THM levels considered here cover the regions of three UK water suppliers, North West Water for the period 1992 –1997, Severn Trent Water for the period 1993 – 1998, and Northumbrian Water for the period 1993 – 1998. These three suppliers were selected for the epidemiological study ( Toledano et al., 2001 ) because they have a wide range of THM levels. Hence, they may not be representative of THM levels for the whole of the UK. Initial analyses of some aspects of the data for one of the water suppliers ( North West Water ) have been reported elsewhere (Keegan et al., 2001 ). North West Water supplies to approximately 6.8 million people spread over 310 water zones from 176 treatment works. The supply area covers the North West of England between Crewe and Carlisle, a diverse area covering the urban conurbations of Liverpool and Manchester as well as the rural Lake District. Sixty - seven percent of the water originates from reservoirs, 25% from rivers, and the remaining 8% is derived from boreholes. Severn Trent Water supplies drinking water to approximately 7.2 million people across approximately 300 water zones. The area covers the West Midlands including the cities of Birmingham, Coventry, Derby, Gloucester, Leicester, Nottingham, Stoke -on -Trent, and Worcester.

Table 3. Correlation between THMs. Water supplier

THM

Chloroform

North West Water 1992 – 1997

chloroform BDCM DBCM bromoform total THMs chloroform BDCM DBCM bromoform total THMs chloroform BDCM DBCM bromoform total THMs

1 0.40 0.28 0.20 0.98 1 0.61 0.01 0.34 0.87 1 0.34 0.12 0.16 0.98

Severn Trent Water 1993 – 1998

Northumbrian Water 1993 – 1998


BDCM

DBCM

Bromoform

Total THMs

1 0.54 0.17 0.56

1 0.17 0.11

1 0.18

1

1 0.68 0.02 0.89

1 0.43 0.48

1 0.06

1

1 0.37 0.05 0.49

1 0.3 0.01

1 0.11

1 19

Whitaker et al.

Ground water sources account for 35% of the supplied water; rivers and reservoirs account for 40% and 25%, respectively. Northumbrian Water supplies to 2.6 million customers in 125 water zones from 45 treatment works in the north east of England including the cities Newcastle upon Tyne and Middlesborough. The majority of the water (approximately 80% ) come from upland reservoirs and rivers; the remainder comes from underground aquifers. Analysis Annual zone mean concentrations of chloroform, BDCM, DBCM, and bromoform were obtained by averaging the measured concentrations of each THM in the sample collected for each water zone and year. The distribution of these annual zone means was then summarised for each THM and water supplier. The reason for summarising variation in the zone means rather than the raw data is that the latter distribution may be biased by the fact that more samples are deliberately collected from water zones with THM levels in the higher ranges. It was necessary to use the raw data to summarise the mean monthly THM levels. The correlation coefficients between THM levels were also calculated from individual samples. Samples under the limit of detection were arbitrarily set to two -thirds of the detection limit. This is approximately equal the expected value of the undetected sample if the data are log -normally distributed. Table 1 shows the limits of detection for each water supplier and the percentage of water zones with all samples in a given year under the limit of detection ( i.e., annual zone means are equal to two -thirds of the detection limit ). These percentages are particularly high for bromoform in North West Water and for DBCM and bromoform in Northumbrian Water, and hence, the distribution of annual zone means for these THMs and regions will be highly influenced by the value chosen to represent the censored observations.


bromoform ) for each water supplier. There was a moderate positive correlation between the THMs that are the most similar in their structure (chloroform and BDCM, BDCM and DBCM, and DBCM and bromoform ). Also there was a slight negative correlation between chloroform and bromoform levels. These associations were slightly stronger for Severn Trent Water. The chloroform levels were strongly correlated with the total THM level. The correlation between the BDCM level and total THM level was moderate for North West and Northumbrian Water; for Severn Trent Water, this correlation was strong, and there was also a moderate correlation between the DBCM level and total THM level.

Results Figure 1 and Table 2 show the distribution of the annual zone means for North West over the period 1992 – 1997, Severn Trent for 1993 – 1998, and Northumbrian Water for 1993 –1998. Clearly, for all three suppliers, chloroform exists in the highest levels, followed by BDCM, then DBCM, and lastly bromoform. The levels of each of the THMs are similar between North West and Northumbrian Waters, whereas Severn Trent Water has lower chloroform levels, wider variation in levels of the brominated THMs ( BDCM, DBCM, and bromoform ), and higher DBCM and bromoform levels. Table 3 shows the correlation between the concentrations of each of the THMs and the total THM level (i.e., the sum of concentrations of chloroform, BDCM, DBCM, and 20

Figure 2. Mean THM levels ( g / l ) by water source, ( i ) North West Water 1997. ( ii ) Severn Trent Water 1993 – 1998. ( iii ) Northumbrian Water 1993 – 1997. Given in parentheses are the percentages of water zones supplied with water from that source. Journal of Exposure Analysis and Environmental Epidemiology (2003) 13(1)


Whitaker et al.

Figure 2i, ii, and iii shows the mean of the annual zone mean THM levels for water zones supplied with water originating from ground, mixed, or surface water sources for North West, Severn Trent, and Northumbrian Waters, respectively. Only 1 year (1997) of data on water sources was available from North West water. Ground waters clearly had very low chloroform levels in comparison to surface and mixed waters. However, the proportion of the brominated THMs was much higher, and in North West Water and Severn Trent Water, ground waters have the highest bromoform levels. Surface waters had the highest chloroform levels. Mixed waters are comprised of some mixture of ground and surface

(i)

70

average monthly THM level (ug/l)

60

50

40 chloroform BDCM DBCM bromoform

30

20

10

(ii)

Jan-98

Jan-97

Jan-96

Jan-95

Jan-94

Jan-93

Jan-92

0

70

chloroform BDCM DBCM bromoform

average monthly THM level

60

50

40

30

20

waters and had the highest levels of BDCM and DBCM (the two THMs that incorporate both chlorine and bromine ). Figure 3i, ii, and iii shows the monthly mean THM levels for North West, Severn Trent, and Northumbrian Waters, respectively. In North West Water, the chloroform levels did not appear to follow any particular seasonal trend; however, seasonal trends did arise in the BDCM and DBCM data. BDCM levels peaked from August to September and were lowest February to March; DBCM levels peaked from July to August and were lowest from November to February. There was wide variation in chloroform levels in Severn Trent Water in 1993, 1997, and 1998 with levels at their lowest between January and April and highest in August or September. No particular trends in chloroform levels were apparent in 1994 and 1996 and chloroform levels decreased throughout 1995. Seasonal trends were evident for BDCM, DBCM, and bromoform, with concentrations peaking from July to September and dipping from December to February. Variation in chloroform levels in Northumbrian Water was large, and with the exception of 1995, levels peaked from September to October and were consistently low from January to March. There were no clear seasonal trends for the brominated THMs. There appeared to be some annual trends in chloroform levels across the three water suppliers. Chloroform levels were high in 1993 and peaked in September; levels were lower in 1994. Throughout 1995, chloroform levels decreased ( for Northumbrian water, there was no same high peak as in previous years ). Chloroform levels were low in 1996, although North West Water differs in that levels increased throughout the year. Severn Trent Water differed from the other two companies in that the chloroform and BDCM levels were lower at the end of the 6 -year period than at the beginning.

10

(iii)

Jan-98

Jan-97

Jan-96

Jan-95

Jan-94

Jan-93

0

70

chloroform BDCM DBCM bromoform

average monthly THM level

60

50

40

30

20

10

Jan-98

Jan-97

Jan-96

Jan-95

Jan-94

Jan-93

0

Figure 3. Monthly average THM levels, North West Water ( i ) 1992 – 1997. ( ii ) Severn Trent Water 1993 – 1998. ( iii ) Northumbrian Water 1993 – 1998. Journal of Exposure Analysis and Environmental Epidemiology (2003) 13(1)

Discussion The levels of the four THMs differed for each of the three water suppliers. Chloroform levels were highest in North West Water and Northumbrian Water, whereas levels of the brominated THMs ( i.e., bromoform, BDCM, and DBCM) were higher and more variable in Severn Trent Water. Results showed that there were clear differences in the levels of each of the THMs formed in ground and surface waters. Therefore, much of the variation in THM levels among water zones in these three water suppliers can be attributed to the source of the supplied water, although differences in treatment processes may also be a contributory factor. Lower levels of natural organic matter and higher levels of bromide occurring in ground waters give a high bromide:organic matter ratio; also, ground waters typically require lower chlorine doses, leading to a high bromide:chlorine ratio. These ratios lead to a THM species 21

Whitaker et al.

shift towards the more brominated THMs (Amy et al., 1991 ). Different water sources can also account for the slight negative correlation found between chloroform and bromoform levels; relatively high chloroform levels and low bromoform levels are found in upland surface waters and vice versa for ground waters. A negative correlation between chloroform and bromoform levels was also observed by Krasner et al. (1990 ) in the US. There is a shift in formation from chloroform to BDCM to DBCM to bromoform as the level of bromide in the chlorinated water increases ( Krasner et al., 1994 ). This accounts for the strong correlation found between the most similar THMs. The chloroform level was found to be strongly correlated with the total THM level for all three water suppliers. This was to be expected since chloroform makes up most of the total THM level. This correlation was a little weaker for Severn Trent Water where chloroform was not the dominant THM in as many water zones (33% in comparison with < 3% in North West Water and Northumbrian Water ). The correlation between BDCM and DBCM levels and the total THM level were stronger for Severn Trent Water where the brominated THMs constitute a larger proportion of the total THM level than the other two water suppliers. In Northumbrian Water, seasonal fluctuations were found in the chloroform levels, whereas in the other two water suppliers, clear seasonal trends were found in the BDCM and DBCM levels, while chloroform levels did not appear to follow any particular trends. Where seasonal fluctuations have been observed, they tend to peak during the summer and autumn and dip in the winter. Similar trends have been found in several other studies (e.g., McGuire and Meadow, 1988; Krasner et al., 1989; Chen and Weisel, 1998 ). Water temperatures and raw water quality would be expected to vary according to season and may explain this trend. Higher water temperatures favour THM production; thus, it is expected that THM levels will tend to be higher during the summer months. Seasonal variation also occurs in the levels and nature of the natural organic matter. Seasonal factors that lead to an increase in natural organic matter are increased biological activity in raw water in the warmer months, leaf fall in the autumn, and higher levels of rainfall during the winter. Chloroform levels were found to decline throughout 1995 in all three water suppliers, during which there was a period of drought. Rainfall gives rise to runoff, which is an important source of natural organic matter ( Aiken and Cotsaris, 1995 ). Bromide levels may also increase during drought (Krasner et al., 1994 ), though no obvious increase in brominated THM levels was observed here. There was wide variation in chloroform levels from one month to the next month whereas levels of the brominated THMs ( BDCM, DBCM, and bromoform ) were considerably more stable. Brominated THM levels may be more 22


stable because these THMs typically form fairly rapidly while chloroform formation increases over time (Krasner, 2001 ). Chloroform levels may, therefore, be more likely to be affected by residence time and chlorine residual in the distribution system and show wider spatial variation. Chloroform levels decreased significantly in Severn Trent Water between the beginning and end of the study period. This is likely to be a result of the changes Severn Trent Water have made to their surface water treatment processes throughout the 6 -year period, which involves the introduction of filtration through granular activated carbon and the removal of prechlorination. Implications for Use of Routinely Collected THM Levels as Exposure Measures in Epidemiological Studies The relevant period for the possible association of chemical exposures with adverse developmental and birth outcomes is believed to be only 3 months ( the first trimester of pregnancy for congenital malformations and last trimester of pregnancy for birth weight and stillbirth ). The clear seasonal variation in the data means that it is essential to the accuracy of our exposure assessment to gain estimates of average water zone THM levels over this period. Where seasonal fluctuations are large, estimation of average THM levels over longer periods ( e.g., annual average THM levels ) will be inaccurate. With few samples taken in each water zone throughout the year, reliable estimation of a mean water zone THM level for such a short period is difficult. Thus, formal statistical modelling of zone -specific THM levels is required to improve the reliability and precision of estimates of exposure to THMs. Since patterns of seasonal variation differ each year, these models should not borrow information across years. A paper discussing statistical modelling of THMs has also been submitted (Whitaker et al., 2002). The epidemiological study for which the present data were collected (Toledano et al., 2001 ) has so far only explored the association between total THM levels and some adverse birth outcomes. We expect the relationship between chloroform levels and adverse birth outcomes to be similar given that chloroform accounts for the majority of the total THM level. The lack of correlation between the brominated THMs and the total THM level implies that the total THM level may not provide a good marker for these THMs and other bromine containing DBPs. Therefore, the association between the individual THMs and adverse birth outcomes will be addressed. Two studies in the US found an association between BDCM exposure and adverse reproductive outcomes. Waller et al. ( 1998 ) found that the strongest association between any THM and spontaneous abortion was with high BDCM exposure ( 18 g /l and 5 glasses of cold tap water per day ). King et al. ( 2000 ) observed that the strongest risk for stillbirth was between high ( 20 g/ l) and low ( < 5 g / l ) Journal of Exposure Analysis and Environmental Epidemiology (2003) 13(1)


BDCM levels. Our analyses show that BDCM levels within the three water suppliers regions are, on average, much below these levels. The BDCM exposure category cut -offs will need to be lower in our study.

Acknowledgments The Small Area Health Statistics Unit is funded by a grant from the Department of Health; Department of the Environment, Transport, and the Regions; Health and Safety Executive; Scottish Executive; National Assembly for Wales; and Northern Ireland Assembly. J.F. and A.G. were funded by a contract from UK Water Industry Research. H.W. was supported by a CASE award studentship funded by Medical Research Council and UK Water Industry Research. We are grateful to North West Water, Severn Trent Water, and Northumbrian Water for providing the data and for their assistance and advice during the study.

References Aiken G., and Cotsaris E. Soil and hydrology: their effect on NOM. J Am Water Works Assoc 1995: 87(1): 36. Amy G.L., Tan L., and Davis M.K. The effect of ozonation and activated carbon absorption on trihalomethane spceciation. Water Res 1991: 25(2): 191. Breach R.A. Severn Trent Water Disinfection and disinfection by products in Great Britain: final report to ISPRA. In: An Assessment of the Presence of Trihalomethanes ( THMs ) in Water Intended for Human Consumption and the Practical Means to Reduce Their Concentrations Without Compromising Disinfection Efficiency. EU Joint Research Centre ( JRC ) at ISPRA on Behalf of the EC. European Commission Joint Research Centre, Ispra, Italy, June 1996. Chen W.J., and Weisel C.P. Halogenated DBP concentrations in a distribution system. J Am Water Works Assoc 1998: 90: 151 – 163. Dore´ M., Merlet N., Legube B., and Croue´ J. - P. Interactions between ozone, halogens and organic compounds. Ozone Sci Eng 1988: 10: 153. Keegan T., Whitaker H., Nieuwenhuijsen M.J., Toledano M.B., Elliot P., Fawell J., Wilkinson M., and Best N. The use of routinely collected trihalomethane drinking water data for epidemiological purposes. Occup Environ Med 2001: 58(7): 447 – 452.


Whitaker et al.

King W.D., Dodds L., and Allen A.C. Relation between stillbirth and specific chlorination by - products in public water supplies. Environ Health Perspect 2000: 108(9): 883 – 886. Krasner S.W., Croue´ J. - P., Buffle J., and Perdue E.M. Three approaches to characterising NOM. J Am Water Works Assoc 1996: 88(6): 66. Krasner S.W., McGuire M.J., Jacangelo J.G., Patania N.L., Reagan K.M., and Aieta E.M. The occurrence of disinfection by - products in US drinking water. J Am Water Works Assoc 1989: 81: 41 – 53. Krasner S.W., Reagan K.M., Jacangelo J.G., Patania N.L., Aieta E.M., and Gramith K.M. Relationships between disinfection by - products and water quality parameters: implications for formation and control. In:. Proceedings from the 1990 American Water Works Association Conference, Cincinnati, OH AWWA, Denver, CO. 1990: 1803 – 1830 ( Part 2). Krasner S.W., Sclimenti M.J., and Means E.G. Quality degradation: implications for DBP formation. J Am Water Works Assoc 1994: 86(6): 34. Krasner S.W. Chemistry and occurrence of disinfection by - products. In: Craun G.F., Hauchman F.S., and Robinson D.E. ( Eds. ), Microbial Pathogens and Disinfection By - Products in Drinking Water: Health Effects and Management Risks. International Life Sciences Institute ( ISLI ) Press, Washington, DC, 2001, pp. 197 – 209. McGuire M.J., and Meadow R.G. AWWARF trihalomethane survey. J Am Water Works Assoc 1988: 80: 61 – 68. Nieuwenhuijsen M.J., Toledano M.B., Eaton N.E., Elliott P., and Fawell J. Chlorination disinfection by - products in water and their association with adverse reproductive outcomes: a review. Occup Environ Med 2000: 57: 73 – 85. Reckhow D.A., and Singer P.C. Mechanism of organic halide formation during fulvic acid chlorination and implications with respect to preozonation. In: Jolley R.L. ( Ed. ), Water Chlorination: Chemistry, Environmental Impact and Health Effects, Vol. 5. Lewis Publications, Chelsea, MI, 1985, pp. 1229 – 1257. Richardson S.D. Drinking water disinfection by - products. In: Meyers R.A. ( Ed. ), Encyclopedia of Environmental Analysis and Remediation, Vol. 3. Wiley, New York, 1998, pp. 1398 – 1421. Symons J.M., Krasner S.W., Simms L.A., and Sclimenti M.J. Measurement of THM and precursor concentrations revisited: the effect of bromide ion. J Am Water Works Assoc 1993: 85(1): 51. Toledano M.B., Nieuwenhuijsen M.J., Bennett J., Best N., Whitaker H., Cockings S., Fawell J., Jarup L., Briggs D., and Elliott P. Chlorination disinfection by - products and adverse birth outcomes in Great Britain: birthweight and still birth. Technical Report, Department of Epidemiology and Public Health, Imperial College, London. Epidemiology 2001 (submitted for publication). Waller K., Swan S.H., Delorenze G., and Hopkins B. Trihalomethanes in drinking water and spontaneous abortion. Epidemiology 1998: 9(2): 134 – 140. Whitaker H., Best N., Nieuwenhuijsen M., and Elliott P. Assessment of ecological exposure to disinfection by - products in drinking water. Epidemiology 2002 (submitted for publication).

23