Asbestos and drinking water in Canada

The Science o[ the "l'oto! Enuironment, 18 (1981) 77-B9

77

Elsevier Scientific Publishing Company, Amsterdam -- Printed in Th~ NethezlnndJ

ASBESTOSAND DRINKINGWAIER IN CANADA

P. TOFT, P. WIGLE, J.C. MERANGERand Y. MAO, Health Protection Branch, Department of National Health and Welfare, Ottawa, Ontario. Canada.

ABSTRACT Samples of raw, treated and distributed tap water were collected from 71 municipalities ~ross Canada and anal~ed for asbestos content by transmission electron microscopy, Chrysottle asbestos was identified as the major asbestos type present in drinking water with some 5% of public water supplies containing asbestos at concentrations greater than I0 million fibres per l l t r e . Improvement factors qf up to 300 were observed for the removal of chrysotlle fibres from drinking w a t e r during treatment, indicating that coagulation/filtration treatment is e f f i c i e n t for this purpose, In certain cases there is evidence to suggest that erosion of asbestos from plpe matertal is taking place. Age*standardlzed mortality rates for gastro-lntestlnal cancers were calculated for each city for ~he period of 1966 to 1976. Rates for the 2 localities with the highest (~10 ~/L) concentrations of asbestos fibres In treated drinking water were co~ared wlth the weighted average of the rates for the 52 localities with asbestos concentrations not significantly greater than zero, Eleven localities had intermediate concentrations of asbestos and six were too small for meanlngful statlstical analysls. Relatively high mortallty rates were apparent amongst males In clty I for cancer of the large intestine except rectum, and in both sexesln city I and males in city 2 for stomach cancer. I t is f e l t that these ftndlngs are probably related to occupatton~] exposure to asbestos. Further statistical analyses are required, however, before the significance of these observations can be f u l l y assessed. i

m

i

i

i

.

I

m

l

i

B

ii

i

_

I

.

_

INTRODUCTION Asbestos is an important mineral in Canada, which ts second only to the Soviet Union in world production, The mineral occurs tn Newfoundland, British Colu~ia and tho Yukon, but mtntng is ch!efly concentrated in Quebec (ref. 1 ) . In 1979 about 1,500,000 metric tons of asbestos were produced in Canada (ref. 2). About gsg of this production is exported - largely to the U.S.A. and Europe (ref, ] ) , More than 3,000 uses are known for this unique substance, Host of the • asbestos produced, however, is used i n the construction industry, especially In co~tnatton with cement in asbestos-cement piPeS. and sheets (ref, 3).

;8

OCCURRENCE IN WATER Asbestos can be Introduced tnto natural waters by the dissolution of asbestos-containing m|nerals ~nd ores, or iron1 dtsposal of tndustrta~ ~astes (ref. 4). Atmospheric pollution may also contribute to the asbestos content of natural waters, The use of asbestos-cement pipe in the dlstrlbutlon systems of public water supplies is a potentla3 source of asbestos In'drinklng water. I t is ~ l l established that exposure to airborne asbestos represents a hazard to health. Available evidence indicates that prolonged inhalation of asbestos leads to increased incidence of mesothelfoma and cancer of the G.I. tract (refs. 5-7), I t is not kno~ whether such an effect is systemic or results from swallowing of inhaled fibres. This possibility, together with the demonstrated presence of asbestos fibres in municipal water supplies (refs. 8, g), prompted us to conduct a national survey of Canadi~ drinklng water / supplies for asbestos content as the basis for an eptdemtology s~udy. CLASSIFICATION Asbestos ts a general term for a variety of fibrous silicate minerals which can be separated into soft, silky fibres. These minerals are divided into two main classes on the basis of their crystal structure -- the serpentine and amphibole groups. Each of the tsbestlform minerals has a non-fibrous counterpart of similar chemical co~osJttono Chrysottle is the only fibrous n~er of the serpentine group. (More than 99.9% of the asbestos fibre produced tn Canada is chrysottle). The ~ h t b o l e group includes droctdoltte, anthophylltte, tremoltte, actinoltte and amostte. These varieties are much less abundant than chrysottle even though the non-fibrous ~u~phibole minerals are quite common. Variation tn chemical composition of asbestos minerals can occur as a result of substitution, leaching ~nd association. Because of this, chemical analysis alone is not sufficient Lo positively identify a particular mineral fibre. ANALYTICAL MrTHODS There are two ways of reporting data on the degree of asbestos contamination: {a) the mass of asbestos per unit volume and (b) the nunt~er of fibres per unit volu~. If the data are reported as mass concentrations using, for example, x-ray diffraction or infrared spectroscopy, no information |s available about the dtmer~sions of individual particles. On the other hand, techniques which report number concentrations and nu~er distributions, such as microscopy, can give detailed Information aoout the size and shape of individual fibres. Studies of the transport of both a~htbole and chrysotlle fibres Into the l u n g s and gastrointestinal tract have shown that there is a significant relationship between the degree of penetration of the fibres, their physical dimensions (and especially the dta~ters of the fibres), and the incidence of associated diseases. I t i s therefore desirable that any investigation into asbestos in d~tnktng water and dl~ease incidence should

79 ,attempt to determine fibre size and concentration. This requirement immediately limits t h e a n a l y t l c a l techniques available to electron or optical microscopy. Light microscopes provide a relatively simple and fast method for determining airborne asbestos d u s t concentrations. However, the limit of resolution is such that only fibres longer than five micrometers and larger than 2.2 micrometers in diameter can be counted reproducibly. Asbestos fibres encountered in water sables are usually less than I micrometer in length, and therefore the optical technique is not suitable, The limit of resolution of the scanning electron microscope is of the order of I 0 - ~ nanometers, which is not sufficient to characterize the smallest fibres observed. Tran~nlsslon electron microscopy has a resolution limit of 0.5 nanometers and is therefore well suited for the analysis of even the smallest fibres and eas the technique selected for our survey. The major disadvantage of this technique is that an elaborate sample preparation technique is required. NATIONAl. SURVEY .Sampling} Samples of raw, treated and distribution water were collected during August and Septe~er 1977 from 71 1ocatlons across Canada, With this type of sampllng scheme the impact of the water treatment and distribution system on the fibre concentrations can be evaluated. All samples were collected in one-lltre polypropylene bottles which had been, ultrasonlcally cleaned. No preservative was added to these samples, since i t has been found that samples for asbestos analysis do not degrade i f they are not exposed to light during storage (ref. ZO).

Sampleprep~atlon The samples were analyzed using the U.S, E.P.A. interim procedure (refs. 11, 1El. Followlng treatment in an ultrasonlc bath appropriate allquots of the water samples were filtered through a "Nuclepore" f i l t e r (47 mm diameter, O~l um pore size), carbon coate~ under vacuum (5-10 nm thickness and mounted on a copper suppa,t grid (80 um ). The f i l t e r material was then dissolved by placing the sample in a Oaffe washer for 15 hours. The sample then consists of a carbon extraction replica of the f i l t e r surface with the particulate ~edded in i t . In order to obtain a good representation of the deposit over the whole area of the . f i l t e r , three- grids were prepared from different areas of the f iI ter, Fibre count Inl~

The fibres present in 10 grid openings sel~tedfrom all spec..!men-grlds were counted:, In this study a flbre-l!; defined as a mineral ~inwhlch the length to diameter ratio is equal or greater than 3:1, For, economic reasons the.counting of fibres is usually,limlted to I0 grid openingssince'the time involved is I/2 day per.sample. With ,this technique a preClslonof ~ 30% can be obtained .

80

In most cases. The precision of the fibre count is limited largely by the counting statistics. Fibres were identified as either amphibole or chrysotile by morphology as the primary criterion wlth supporting evidence from selected area electron diffraction (SAED) and energy dispersive X-ray analysis (EDXA).

RESULTS AND DISCUSSION Analytlcal Reproducibility The analytical reproduclbllity can be appreciated by considering Table 2 In which analyses of a number of dupllcate samples are reported. I t can be seen that the reproducibility is bettor than a factor of two between palrs of results where the values are greater than 106 f i b r e s / l l t r e . Most of these duplicated results were analyses of two separate samples co1|ected fro~ the same source at the same time. The reproducibIIlty was also shown to be satisfactory even at the very high fibre concentration levels, where d i l u t i o n of the s~unple was necessary prior to f i l t r a t i o n and analysis. Result~ Amphlbole asbestos is not a significant contaminant of Canadian drinking water supplies (ref. 13). Only 7% of the 336 samples showed detectable l~vels of amphibole fibres. These levels ranged up to a maximum of 13 x 10f l b r e s / l l t r e , found in a raw water sample from Thompson, Manitoba. These higher values were generally associated with high chrysotile levels. Chrysottle asbestos ~as identified as the major asbestos type present in drinking water wtth some 5% of the population receiving water with asbestos concentrations greater than x 106 flbres/lftre (ref. 13)~ Correspo~dlngly about 0.6% recelve water having more than 100 x 10_ flbres/lltre. The locations with fibre concentrations exceeding 5 x 106 f i b r e s / l l t r e can be seen in Table ~. The highest result for a distribution sa~le in this survey was 1800 x 10: flbres/lftre which was recorded for Bale Verte in Newfoundland. In this particular case the provincial government has initiated measures to reduce the levels;

I0

Typical fibre length distributions for a location having a fibre concentration in the distribution water in excess of 5 x 106 f i b r e s / l t t r e ts shown in reference 13. In general, the median lengths observed l i e between 0.5 and 0.8 micrometres. In some locations very to~9 fibres (up to 50 urn) were also observed, but these did not form a majority of the fibres present. Remgval of Fibres by Filtration Although pilot plant tests have been made to establish methods of asbestos fibre removal from drlnklng water supplies (ref. 14) no reliable data have to-date been available concerning the efficiency of extsting'J;~lant operations in this respect. For a reliable measurement, the input concentration should be high enough so that analytical sensitivity problem are'not a factor. Table I summarizes the data from a l l . suttable locations haying a combination of

t n ~'t oa ~ 0

0

m

(::) r .i~

I

0

L n ~ ~1) ~l) ~J 0

0

0

0

o

o

o

0

kq 4~ 0

o

o

0

ul G) 0

ml

~ I

r - - r-~ 0 I

!

I

0 I

r - , ~ ) r - - r ' ~ r d ~-!

I

I

¢

I

I

!

¢~1 (~1 r " I

i

~ ) U') , - - ~ J r I

I

t

L

I

0 i

C) ri

~

~

x m

~

~2

~

"<'-I

~

N m.

N~N

u'l,

N

~~

° °

,-oo

~

o

o

"1 r,*

I;-"~"

0

~

0

"~ 0 ~ -

v~ ~

~-"~D~ -

3[.~,.~ ~ , 1 - . )

~

~L~ ~ . 0

~

~-j~

I

82

TABLE 2

Analytlcal Roproduclblllty

Analysis of

Locdtton

Chrysottle __

Bate V e r t e

D u p l | c a t e WAter Samples

__

asbestos concentretton, . _gsZ_conf]dencelJmlts

1200 - 2500 310 -

-

10 6 f t b r e s / 1 1 t r e _

_

___

1100 - 1800

460

160 -

240

11

10 -

14

La S c t e

6.0

T h u n d e r Bay

1.3-

4.3

0.7-

4.0

0.4 -

3.6

0.6 -

2.1

_

B3

{elatlvely htgh raw water fibre concentrations and f i l t r a t i o n plants. Improvement factors of 18 - 300 are sho~ for 7 locations indicating that the coagulation and f i l t r a t i o n treatments remove chrysottle asbestos fibres very efficiently. From the table, i t is evident that at those locations which do not have water f l l t r a t l o n plants It would be possible to reduce the fibre levels In the treated water using the same procedures. At the tlme of sampllng there was a large variation In the ablllty of the various f i l t r a t i o n plants to remove chrysotlle fibres. T~e fibre removal efficiency, however, may be related to the tlme since the f l l t e r bed was back-washed, and s ~ l l n g over the entire cycle would be required to completely characterize a given situation. :.COL ntrtbution.of Asbestos-Cement Pipe Table I also shows data from those Iocatlons where water samples collected from the distribution network had slgnlflcantlyhlgher fibre counts than those from the treated source. In the cases of Kamloops, Beaulac, Sherbrooke and WhJtehorse, there is a real posslbillty that the variations were due to random flu:~uations of the actual fibre content at dlfferent parts of the water system. Examination of distribution sample values for Bale Verte and Disraeli confirms that s u c h large variations do occur, especially when the concentrations are high. I t is tn~ortant to recognize that these are real variations at different points of the distribution system; they are not a consequence of zqalytlca] variability. Manyof the high values were checked by repeat analyses, and duplicate measurements from the same bottle dtd not dtsplab, thts order of variation, Thts Is an indication that grab samples of the type collected are not always sufficient to con~letely characterize a fibre-contaminated potable water source, and that pooled samples over a period of time may be ~ecessar~. Wtnntpeg is apparently a classtc case where the fibre levels progressively increase from the raw -source through to the, faucet in the home. Raw water travels 100 mlles by aqueduct. Dupllcate raw water values of 0.3 x106 and 0.4 x 106 f!bres/Iitre were ~talned. The nw~anvalues after treatment ~ere 0.5 x 106 and 1;3 x 10 f i b r ~ s / l t t r e , ~htch ts not a • " 's t a t i s t i c a l l y significant increase over the ~aw wa~er values. A t points in the Winnipeg distribution, values as high as 6.5 X 10 were found. These data provide a statistlcally valld indication that erosion of asbestos-cement.ptbes Is taking place. Vancouver ts a more complex situation, Tworeservoirs provide the majority of the. water for metropolitan.Vancouver; Treated but6unftltered waters from two different reservoirs had concentrations of 0.9 x l O : and 5.4 x 106. ftbres/lttre respectively. These two sources with'. -different fibre concentrations mix in the same d l s ~ l b u t l o n netxork. However, the fibre concentrations, observed t n the distribution network. varied wiclelyfrom less than 0.7 x '106.to about'12 x 'i06 f l b r e s / l i t r e , lAPart from t h e fact that one of t ~ treated source ftl~re concentrations was higher, the "p~ttern"

84

is rather similar to that of Winnipeg, Support for this Interpretation is given by the fact that aggressive water ts knom tr erode asbestos-cement composition water pipe (refs. 15, 16). STUDY OF HORTALXTY AND ASBESTOS IN DRINKIN~ WATER In order to study the possible i~pajt on health of asbestos I~ drinking water, the relationship between the mortality rates and asbestos levels was examined, The 71 cities surveyed were divided into t~o groups -- those with drinking water concentrations greater than I00 million fibres per Hire and those wlth less thon 5 million fibres per lltre. Those cities with intermediate concentratl(~s and also those with small populations {less than I0 000) were excluded fro~ thls study, There remained 2 cities {Sherbrooke and Thetford Hines) in the relatively high group and 52 cities In the low group. The death hates from various causes ln- these two groups ~ere then compared in the following w~y. (t) Heth.~...~ The localities ~ r e defined according to 1976 Census geographic boundaries. All deaths among usual residents of the localities were retrieved from the computerized n)tlonal mortality files. The population at risk was estimated using datafrom the censuses of 1966, 1971 and 1976 adjusted to 1976 geographic boundaries, Person-years at risk during the period 1966 to 1976 were calculated by sex and age for each locality. Age-speclflc mortality rates were calculated by dividing the number of deaths in a given category defined by locality, cause of death, sex and ageby the person-years at rlsk in the s~me locality, sex and age group. Age-standardlzed mortality rates {ASflRs) were then calculated by the direct method usingthe 1971 Canadian population as standard. The standard errors of ASHRs ~ere ~alculated by the .~ethod of Chlang (ref. 17). AS~Rs are expressed as deaths ~er I00,000 person-years. ASHRs for the age range 26 to 6g were selected for presentation as it is generally believed that the reliability of the certified cause of death in this age range is greater than fcr older persons. {if) R e s u l ASMRs and their 951 confidence intervals for major causes of death by sex and region are presented in Table 3. The all-cause and all-cancer mortality rates for males i n Thetford Hlnes were significantly higher than thosefor Sherbrooke or the 52 comparison localities but the rates fop females were similar In all 3 regions. Mortality rates for ~on-neoplasttc diseases of the respiratory system were slgnlflcantly hlgh for males in Sherbrooke and lhetford Hines; the rate for females in Thetford Hines was significantly low. The ASHR for males in

B~

P,,U~

O4

0

g ~ in

IJ'l r " "

~'=l

~

~.,..

~

f#

q-

0

/

~t

I

r.-

B6

Th~ttford M,nes (84.g) was almost double that for the comparison group (43.4). v

ASMRs for selected gastrointestinal cancer sites by sex and region are pnl~ented in Table A. The only s t a t i s t i c a l l y stgntft¢;antly high ASMR was that ston.~,ch cancer among males in Thetford Hines; the mortality rate was 38.7, mo~le than double the rate In the comparison group (16,1l).

fo,j

The A'.HR for cancer of the large intestine minus re:tum was higher for males $n]Thetfnrd Hines (29.0) than in the comparison group (17,5) but the diCference wa,I not quite significant at the 5% level. Furthermore, the ASMRs for this catlcer were normal for both sexes in Sherbrooke and females in Thetford Hines. Th~ ASI.~(~ for all gastrointestinal sites combined v,as significantly high for males tI~ Thetford Mines (104 versus 66.5 in the comparison group). ASMR= for selected non-gastrointesttonal cancer s~tes by sex and region are presented in l)ble 5. The only statistlc()lly significant result in the high ASMR For lung cancer among males in Thetford Hines (139 versus 67.0 in the co~par;sow~ grout,), Ingested asbestos has been shown to accumulate in the brain, kidney and reticuloe.ndothellal system but the ASMRs for cancers in these sites were rot unusual (ref. 18). CONCLUSIONS 1. "he mortality rates for persons living in 2 localities in which asbestos was ;resent at high concentrations in water samples from the distribution syste~ were analyzed, Although there are serious llmitatluns to this type of study~ no consistent increase of mortality rates for any cancer was demonstrated, This finding is consistent with a previous, but more limited, epl~miology stuc(y carried out in Quebec (ref. Ig).

2. The high ASMRs for lung cancer and non-neoplastic respiratory diseases amo~.g male~ but not females in Thetford Hines are probably due to occupational asbHstos exposure. I t ts not possible, howeverp to conclusively separate the possible effects of the two types of e~posure (occupation versus drinking water). I'; might be assumed that females in Thetford Hines would not receive much occupational exposure although they would have been exposed to airborne as~)estos in the ambient air. 3. The present study has shown that amphibole asbestos Is currently not a significant contaminant of Canadian drinking water supplies, Chrysottle fibre concentrations on the other hand can reach levels up to ZOO0 x 106 fibres/lttre. However, the fibre concentration of drinking water ts generally, b~low 10 x 106 f t b r e s / l l t r e . 4.

It has

also

been demonstrated that filtration systems can efflclently

~move asbestos fibres from drinking water.

87

'c~

.....

~.~

'~,

"

•

~"7,

o~. I0 V

~e-.

L~,--

.'--r~

.e.-, r - .

00~"

QD~I"

°~

"r-

E -1-.

u

~'.'-.

"~

~

r=~ .4"

r-,¢~

¢"~ i

~.,~.

o',~.'~

I ' ~ 0,,.-

r,,~ eio

,

,C~

• I

,-.-, r--r-.

~,=.

~

I 4,--

U

¢ Crl

~J q-,

,o,,',

oo0

-,

~,-,.. kid

,~,, ~ l r

•

u

4~

U 0 o~.-

0

!i i:" .

:i.

.

•

i,a

~ .

.

0 U

~

0 ,L

,I,,a

~L

o l.

:=p

.,-~ . ~

...I

~

U

,l~

~ ~

I

X

~

g E e-

U

~,.

L

.~ 0 U

I. e,,,,

r.,.. o

.I'-"

la,.I

)g ~

..I

ii~

~ i~-

._ II

II

~

U'I,

,~"

#l

~

Z

8B

I;. , r '

I

I ~rt

0

C~

%,

'.0 ~ "

~

t.~ r , I

U'I r'.,

r-, ~4-

£3. 4.,.) rIJ °l,q01

pI U .tJ 14

E

V) ul

u 0

.

.-,.-,

~,,.-.

~..-,

• ~.

-~.,

~,~.

~.~.

~.~

,-:~.

~.Z

";?

~.~"

~.~,"

0 ~h

.£

~.~.

0 ta

0

5 e ~1

~r k"

~'~-

=::: ~1"

:i:: ['-

:E ~"

::i: .~"

~ r L'-

S.~

~,,

.

L

3

U

~

0

C

,I~ ~

L

,, ~ ~ ~.,.. :;~

k

",-

k

-

89

REFERENCES 1 G.O. Vogt, "Asbestos" - in 'Ca,adian H:nerals Yearbook, Mtneral Resources Branch, Oept. of Energy, Hines ~nd Resources, Ottawa, 1975. 2 Statistics C~n=d~: Canada's mineral production, preliminary estimates, catalogue number Z6-20Z; Jan., 1980, 3 International Agency for Research on Cancer. IARCmonographs on the evaluation of carcinogenic rlsk of chemicals to man, ~Jme 14 - Asbestos, World Health Organization, 1977. 4 J.R. Kramer and O. Murdoch, Canadian Research and Development, (1974) 31. 5 M.L. Newhouse, Cancer among vcrkers In tile asbe~tos textile industry° In: Blologlcal effects of asbestos; P. Bogovski ed., Lyon, IARC, (1973) 203. 6 I.O. SeIikoff, E.C. Hammond, and H. Seidman, Cancer risk of insulation workers in the United States. Ibld., 209. 7 M.A. Sct,~lderman, Digestive system cancer among persons subjected to occupational Inhalation of asbestos particles: A literature revlewwtth emphasis on dose response. Environmental Health Perspectives. g (1974) 307. 8 H.M. Cunningham and R. Pontefract, Asbestos f4bres In beverages and drinking water. Nature 232 (1971) 332. 9 l.J. Mason, F.W. McKay and R.W. Miller, Asbestos-llke fibers in Duluth water supply: relation to cancer mortality, j.A.M.A. 228, (]974) 1019. 10 E.J. Chatfield and R.W. Glass, Improved methodology for determination of asbestos as a water pollutant. Ontario Research Foundation Report (1976). 11 C.H. Anderson and J.M. Long, Pre)Imlnary interim procedure for fibrous asbestos, U,S. E,P.A. July 31, (1976). 12 E.J. Chatfleld, R.W. Glass and M.O. Dillon, Preparation of water samples for asbestos fibre counting by electron microscopy, U,S, E.P.A. Report Number'EPA-600/4-78-011, January (1978). 13 A Natlonal Survey for Asbestos Fibres in Canadi~ Drinking Water Supplles; Health and Welfare Canada, Envlron~ntal Health Dlrectorate 79-EHD-34 (1979). 14 R.B. Hunslnger, J. Lawrence and K.J. Roberts, Pilot plant studies to effect chrysotlle asbestos fibre reduction during potable water treatment, Report No. 67, Ontario Ministry of the Environment May (1977). 15 M. Mah and E.S. Boatman, Scanning and transml~slon electron microscopy qf new and used asbestos-cement pipe utilized In the distribution of water supplies. Scanning Electron Microscopy I (1978) 85. 16 R.W. Buelow, J.R. Mlliette and E.F. HcFarren, Fteld investigation of the performance of asbestos-cement pipe under various water quality conditions. J. Am. ~ater Works Assoc. (in press). 17 C.L. Chfang, Standard error of the age-adjusted death rate. Vital Statistics Special Reports. 47 (1961) 275-285. 18 H.M. Cunnlngham, C.A. Noodle, G.A. Lawrence, R.D. Pontrefact, Chronic effects of Ingested asbestos tn rats. Arch, Environ. Contam. Toxicol. 6 (1977) 507-513. Ig D.T. Wlgle, Cancer mortality in relation to asbestos in municipal water supplies. Arch. Env. Health (1977) 185-189. COPYRIGHT This article was prepared by employees of the Canadian Government, \ Accordingly; copyright is retatnedby the Government of Canada,