8629 ©
Saturday 14 January 1989
GEOGRAPHICAL RELATION BETWEEN
ALZHEIMER’S DISEASE AND ALUMINIUM IN
DRINKING WATER
C. N. MARTYN1
C. OSMOND1
J. A. EDWARDSON2
D. J. P. BARKER1
E. C. HARRIS1
R. F. LACEY3
Medical Research Council Environmental Epidemiology Unit,1
Southampton General Hospital, Southampton SO 9 4XY; Medical
Research Council Neurochemical Pathology Unit,2 Newcastle General
Hospital, Newcastle upon Tyne NE4 6BE; and Water Research
Centre,3 Medmenham, Marlow SL7 2HD
Summary In a survey of eighty-eight county districts
within England and Wales, rates of
Alzheimer’s disease in people under the age of 70 years were
estimated from the records of the computerised
tomographic (CT) scanning units that served these districts.
Rates were adjusted to compensate for differences in
distance from the nearest CT scanning unit and for
differences in the size of the population served by the units.
Aluminium concentrations in water over the past 10 years
were obtained from water authorities and water companies.
The risk of Alzheimer’s disease was 1·5 times higher in
districts where the mean aluminium concentration exceeded
0·11 mg/l than in districts where concentrations were less
than 0·01 mg/l. There was no evidence ofa relation between
other causes of dementia, or epilepsy, and aluminium
concentrations in water.
Introduction
ALUMINIUM has been detected in both senile plaques!
and neurofibrillary tangle bearing neurons2 in brains of
patients with Alzheimer’s disease. These findings suggest
that exposure to aluminium may be important in causation
ofthe disease. Aluminium is ubiquitous in the environment
and in Britain some 5-10 mg is ingested each day.3 However,
only a very small proportion of this ingested aluminium is
absorbed.’
In many parts of Britain aluminium sulphate is used as a
coagulant in the treatment of water. Its purpose is to help
with the removal of suspended matter and highly coloured
humic substances, thus reducing the dose of chlorine later
required to ensure satisfactory microbiological quality.
Most ofthe aluminium added is also removed in the process
of clarification but residual amounts may pass into supply.
Water treatment plants should be operated so as to keep the
residual concentration as low as possible. For aesthetic
reasons the current European Community directive
specifies a maximum acceptable concentration of 0-2 mg/1.
Aluminium from drinking water forms only a small part of
the total daily intake but because the aluminium is largely
uncomplexed it may make a disproportionate contribution
to the total amount absorbed from the gastrointestinal tract.
To examine the possible relation between exposure to
aluminium in water and the development of Alzheimer’s
disease we conducted a survey in eighty-eight county
districts within England and Wales with differing
concentrations of aluminium in the water supply. The
incidence of Alzheimer’s disease in each district was
estimated from diagnostic rates in people aged under 70
years. Diagnostic rates were obtained by examining records
ofcomputerised tomographic (CT) scanning centres. Mean
concentrations of aluminium in water in each district over
the previous 10 years were calculated from information
supplied by water undertakings.
Methods
The county districts studied were within Northumberland,
Durham, and Tyne and Wear; Lancashire, Merseyside, Cheshire,
and Clwyd; Nottinghamshire and Derbyshire; Norfolk, Suffolk,
and Cambridgeshire; Dyfed, West Glamorgan, Mid Glamorgan
and South Glamorgan, and Gwent; Devon and Cornwall; and
Hampshire and the Isle of Wight. These places were chosen to
represent a wide range of water aluminium concentrations within
different parts of the country. Although there were exceptions, the
concentrations ofaluminium in water in the chosen county districts
within each area tended to be similar. The highest concentrations
were found in Northumberland, Tyne and Wear, and Durham, and
in Devon and Cornwall; the lowest in Norfolk, Suffolk, and
Cambridgeshire, in Nottinghamshire and Derbyshire, and in
Hampshire.
60
Measurements ofRates of Alzheimer’s disease
Cases of dementia were ascertained from the records of CT
scanning units. Two of us (C. N. M., E. C. H.) visited units within
the seven study areas and examined their records to identify all
patients aged 40-69 years scanned because ofthe presence ofclinical
features ofa dementing illness. Where possible, records of all scans
conducted over a 3-year period, either 1983-85 or 1984-86 were
examined. For 4 of the units this was not practicable either because
the CT scanner had not been in operation for 3 years or because
adequate records were not available; records for a shorter period
were then used.
Patients with dementia were classified into four categories
accordingto the clinical information supplied on a request form and
the radiologist’s report ofthe CT scan. Criteria for inclusion in each
category are given below:
Probable Alzheimer’s disease.-The clinical information
supplied on the request form specified Alzheimer’s disease or
dementia or contained at least two of the following clinical
features: deterioration of memory, intellectual deterioration,
persistent confusion, dysphasia in the absence of hemiparesis.
There was no history of stroke, systemic hypertension, alcohol
abuse, or epilepsy, nor was there any other indication of any
specific cause for dementia. The CT scan was reported either as
normal or as showing only cerebral atrophy.
Possible Alzheimer’s disease.-The clinical information
supplied indicated a history of intellectual deterioration but
without additional details. The CT scan was reported either as
normal or as showing only cerebral atrophy.
Cerebrovascular dementia.-The clinical information indicated
a history of stroke, transient ischaemic attacks, or recent epilepsy
as well as features of a dementing illness, or the CT scan was
reported as showing evidence of cerebral infarction including
lacunar infarction or Binswanger’s syndrome.
Dementia from other causes.-Patients were placed in this
category when either the clinical information on the request form
or the report ofthe CT scan provided evidence ofa specific cause
for the dementing illness. The commonest diagnoses in this
group were Huntington’s disease, multiple head injury, and
chronic alcohol abuse.
A study of 52 consecutive cases of presenile dementia scanned at
the Wessex Neurological Centre, Southampton General Hospital,
showed that classification of the type of dementia from the clinical
information supplied on the request form for the CT scan and from
the radiologist’s report was in good agreement with the final clinical
diagnosis recorded in the patients’ case-notes. Of 24 cases classified
as probable Alzheimer’s disease 19 had Alzheimer’s disease
recorded as the final clinical diagnosis.
To examine the effect of differences in referral practices between
areas we also identified patients within the same age range who were
investigated by CT scanning because of epilepsy.
Rates of disease for each diagnosis in each county district were
directly age-standardised to the 40-69-year-old population of
England and Wales for 1983 by 5-year bands. Analysis was
restricted to eighty-eight county districts served exclusively by CT
scanning centres within the seven areas.
Estirnation of Water Aluminium Concentrations
From the water undertakings responsible for supplying the
eighty-eight county districts we obtained details concerning type of
source, water treatment processes in use, and distribution of supply
for all water sources contributing more than 5 % ofthe total supply.
We also obtained data about residual aluminium concentrations for
each water source over the previous 10 years. Analyticalmethods for
measuring aluminium in water have changed over the past 10 years
and different water undertakings may have used different methods
at different times. Estimates of mean water aluminium
concentrations were therefore divided into five groups. The first
group contained county districts with concentrations of 001 mg/1 or
less. None of these districts had used aluminium coagulation in
water treatment during the previous decade and aluminium in their
waters was usually undetectable. The other four groups, in which
aluminium concentrations ranged from 0 02 to 0-20 mg/l, were
divided so that they contained approximately equal numbers of
people at risk.
The relation between age standardised rates of Alzheimer’s
disease, other categories of dementia and epilepsy, and mean
concentration of aluminium in water was investigated by use of log
linear and logistic regression analysis.
Results
We identified 1203 patients with one ofthe four categories
of dementia. 18 had to be excluded because details of their
age or place of residence could not be found. 445 of these
patients (37%) were classified as having probable
Alzheimer’s disease. Table i gives the numbers ofpatients in
each of the diagnostic categories. There were 2997 patients
with epilepsy, of whom 61 had to be excluded because of
inadequate information.
In calculating the rates for dementia and epilepsy it was
necessary to take account of the effects of distance of the
county district from the nearest CT scanning centre. Table
11 shows that rates for all diagnostic categories of disease
were lower in county districts only a few miles from the
scanner than those for the districts in which the scanner was
located. Further increase in distance had little effect on rates.
Because the size ofthe population served by an individual
CT scanning unit varied between areas, we examined how
the total annual number of brain scans performed per 1000
population influenced our estimates of rates of disease.
Scanning rates varied threefold between different areas but
the corresponding variation in rates of dementia and
epilepsy was less than 1 ’4.
The relation between aluminium and rates of disease in
the different diagnostic categories is summarised in table ill.
Rates have been adjusted to take account of distance of
county districts from the nearest CT scanning centre and
differences in scanning rates between centres. Rates of
disease in districts where concentrations of aluminium in
water were less than 0-01 mg/1 were 5-4 per 100 000 for
probable Alzheimer’s disease, 3-0 per 100 000 for possible
Alzheimer’s disease, and 6-3 per 100 000 for dementia from
other causes. The rate for all causes of dementia combined
was 14-7 per 100 000 and for epilepsy it was 37-6 per
100 000. These rates have been taken as a baseline and rates
ofdisease in the other four groups ofdistricts with increasing
water aluminium concentrations have been expressed as a
risk relative to this baseline.
The risk for all causes of dementia combined exceeded
unity in all groups of districts where concentrations of
TABLE I-NUMBERS OF DEMENTED PATIENTS
TABLE II-EFFECT OF DISTANCE FROM COUNTY DISTRICT OF
RESIDENCE TO CT SCANNING UNIT ON DIAGNOSTIC RATES OF
DISEASE (PER 100 000 POPULATION)
61
TABLE III-RELATIVE RISKS (95% CI) OF ALZHEIMER’S DISEASE,
DEMENTIA FROM OTHER CAUSES, AND EPILEPSY IN PATIENTS
AGED 40-69 IN COUNTY DISTRICTS GROUPED ACCORDING TO
WATER ALUMINIUM CONCENTRATION (RISKS ADJUSTED FOR
DISTANCE FROM CT SCANNING UNIT AND CT SCANNING RATE)
aluminium were greater than 0.01 mg/1. Separate
examination of the different diagnostic categories of
dementia showed that the increase in risk was restricted to
one diagnostic category-probable Alzheimer’s disease. In
this category, relative risks for groups of districts where
aluminium concentrations exceeded 0-01 mg/1 ranged from
13 to 1-5 (see table in). After exclusion of patients aged 65
years or older, the relative risk in the highest aluminium
group reached 1-7 (95% confidence interval 1-1-2-7) and a
gradient of risk was present with increasing concentrations
ofaluminium (table IV). A relation between aluminium and
probable Alzheimer’s disease was also present when sexes
were examined separately although the effect did not reach
the 5% level of statistical significance for women.
Relative risks for possible Alzheimer’s disease,
cerebrovascular dementia, and other causes of dementia
tended to be slightly greater than unity in districts where
water aluminium concentrations were greater than 0-01 mg/1
but the relative risk never exceeded 1-2 and none was
statistically significant. In the tables, the diagnostic
categories cerebrovascular dementia and other causes of
dementia are shown combined.
No relation between water aluminium concentrations and
epilepsy was found. Relative risks were either 1 or 0-9 in all
ranges of aluminium concentrations.
TABLE IV-RELATIVE RISKS (95% CI) OF ALZHEIMER’DISEASE,
DEMENTIA FROM OTHER CAUSES, AND EPILEPSY IN PATIENTS
AGED 40-64 (RISKS ADJUSTED FOR DISTANCE FROM CT SCANNING
UNIT AND CT SCANNING RATE)
I I I
TABLE V-RELATIVE RISKS (95% CI) OF ALZHEIMER’S DISEASE AND
DEMENTIA FROM OTHER CAUSES ESTIMATED FROM
PROPORTIONAL RATES IN PATIENTS AGED 40-69 IN COUNTY
DISTRICTS GROUPED ACCORDING TO WATER ALUMINIUM
CONCENTRATIONS
I I I
As another method of examining the effect on diagnostic
rates of distance from the CT scanner and the size of the
population served by a CT scanner, we related numbers of
cases in each category of dementia to the total number of
cases of epilepsy and dementia for each district. In this way
we hoped to take account of differences between areas in
their practices of referral and CT scanning. Logistic
regression analysis revealed that this ratio varied little with
changes in distance from the scanner or with population size.
The relation between water aluminium concentrations and
probable Alzheimer’s disease was still present. These results
are summarised in table v.
Discussion
This survey, conducted in eighty-eight county districts
within England and Wales, shows that rates of Alzheimer’s
disease in people under the age of 70 years are related to the
average aluminium concentrations present in drinking water
supplies over the previous decade. The relation to water
aluminium concentrations was specific to the group of
patients in whom the diagnosis of Alzheimer’s disease was
most certain; no such relation with any of the other
diagnostic categories ofdementia or to epilepsy was present.
We used diagnostic rates of dementia in patients under
the age of 70 years to estimate the incidence of Alzheimer’s
disease because we judged that most patients in this age
group who presented with symptoms of cognitive
impairment would be referred to neurologists, psychiatrists,
and other hospital specialists with expertise in dementia and
that their subsequent investigation would include a CT
brain scan. Younger patients with dementia are generally
investigated more aggressively and case ascertainment is
likely to be more complete. In the analysis we examined
separately a subgroup of patients under the age of 65 years
and found a closer relation between aluminium
concentrations and Alzheimer’s disease than in patients
under the age of 70 years.
Apart from age, there are several factors that may affect
whether patients with dementia receive a CT brain scan and
hence influence our estimates of disease rates. Where
differences in diagnostic rates were in part dependent upon
differences in the size of the population served by the CT
scanner or the distance of a county district from the nearest
CT scanning centre, it was possible to make adjustment
during the analysis. As an alternative method of controlling
for these and other possible influences on the diagnostic rate
of dementia we took another diagnosis, epilepsy, and
analysed the rate of dementia as a proportion of the rates of
dementia and epilepsy combined. In this analysis no
adjustment for the effect of either distance of the county
district from the CT scanning unit or the local CT scanning
rate was necessary. A positive relation between rates of
Alzheimer’s disease and water aluminium concentrations
was present whichever way the data were analysed. Neither
method of analysis suggested a relation between water
aluminium and other categories of dementia. We conclude
that the relation is not likely to be a result of variation in
referral patterns or scanning practices between areas.
The current view of Alzheimer’s disease is that
neuropathological confirmation is required for definite
diagnosis.5 However, several studies have demonstrated that
in about 80% of cases a diagnosis made on clinical features
alone is in accord with the neuropathological fmdings6.7 and
that CT scanning is helpful in the differential diagnosis of
dementia.8 Our categorisation of dementia, based only on
62
information from the CT scan request form and the CT scan
report, conformed closely to the clinical diagnosis. The
majority of patients in our diagnostic category of probable
Alzheimer’s disease could be expected to show
neuropathological features of Alzheimer’s disease. We
should point out that the likely effect of misclassification of
type of dementia and incompleteness of case ascertainment
would be to obscure any relation between aluminium
concentrations and Alzheimer’s disease and to reduce the
magnitude of the estimate of relative risk.
Although detailed information about the concentrations
of aluminium in the water supply to individual county
districts did not become available until identification and
categorisation of patients had been completed, the
investigators collecting data from CT scanning units were
aware, in broad terms, of the aluminium concentrations
present in each area. This could have produced a bias in the
allocation of patients to different disease categories.
However, criteria for inclusion in each category were strictly
defined at the beginning of the study and a relation with
aluminium is still present when all categories ofdementia are
combined, or when the categories of probable and possible
Alzheimer’s disease are analysed together. Bias in the
classification of dementia is unlikely to explain our results.
Because aluminium in drinking water makes only a small
contribution to the total intake of aluminium a relation
between concentrations of aluminium in water and
Alzheimer’s disease is, at first sight, surprising. However,
urinary excretion of aluminium is only 0-02 to 0-05 mg per
day.4 This represents less than 1% of the 5-10 mg ingested
daily, so most dietary sources ofaluminium are presumably
not absorbed. Aluminium in drinking water is either
dissolved or readily brought into solution and its
bioavailability may therefore be much higher than
aluminium from other sources. Moreover, if other sources
ofdietary aluminium also contribute to Alzheimer’s disease,
this will tend to diminish the strength of any relation with
aluminium in drinking water.
Epidemiological evidence from Norway has previously
suggested a link between the concentration of aluminium in
water and dementia.9,1O In these studies there was no attempt
to distinguish between different causes of dementia and the
use of mortality data has also been criticised.11 Mortality
data for dementia have been shown to contain serious biases
in England and Wales12 and in Australia.13 Although
Norwegian data may be more reliable, the relevance ofthese
studies to Alzheimer’s disease is hard to assess in the absence
of evidence that mortality from dementia reflects the
incidence of Alzheimer’s disease.
The results of the present survey provide evidence of a
causal relation between aluminium and Alzheimer’s disease.
However, care is needed in interpretation because, as in all
epidemiological surveys, the possibility exists that the
relation observed is due to the operation of some unknown
confounding variable. Further studies in different
populations are required to confirm these results and we are
now conducting a case-control study to investigate the
relation between dietary aluminium and Alzheimer’s disease
at an individual level.
This work would not have been possible without the willing cooperation of
the water authorities and water companies supplymg the areas in the survey.
We are most grateful to them for their assistance. We are also indebted to Dr
A. Appleby, Dr L. Brock, Dr E. H. Burrows, Dr A. Carty, Dr P. Cook, Dr K.
Grant, Dr T. D. Hawkins, Dr I Holland, Dr M. Hourihan, Dr V.
McAllister, Dr P. Norman, Dr R. Paxton, Dr J. Pilling, Dr R. Sellwood, and
Dr E. Wyn Jones for letting us examine the records of their CT scanning units.
C N MARTYN AND OTHERS: REFERENCES
1 Candy JM, Oakley AE, Klinowski J, et al Aluminosilicates and senile plaque
formation in Alzheimer’s disease Lancet 1986; i: 354-57.
2. Perl DP, Brody AR. Alzheimer’s disease X-ray spectrometric evidence ofaluminium
accumulation in neurofibrillary tangle-bearing neurones. Science 1980, 208:
297-99
3. Survey of aluminium, antimony, chromium, indium, nickel, thallium and tin in food.
MAFF—the fifteenth report of the steering group on food surveillance The
working party on the monitoring of food stuffs for heavy metals. London. HM
Stationery Office, 1985.
4 Ganrot PO. Metabolism and possible health effects of aluminium. Envir Health Persp
1986; 65: 363-441
5. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan E Clinical
diagnosis of Alzheimer’s disease. report of the NINCDS-ADRDA Work Group
under the auspices of Department of Health and Human Services Task Force on
Alzheimer’s disease Neurology 1984; 34: 939-44.
6 Sulkava R, Haltia M, Paetau A, Wikstrom J, Palo J Accuracy of clinical diagnosis in
primary degenerative dementia: correlation with neuropathological findings
J Neurol Neurosurg Psychiatry 1983, 46: 9-13
7. Molsa P, Paljarvi L, Rinne J, Rinne U, Sako E. Validity of clinical diagnosis in
dementia a prospective clinicopathological study J Neurol Neurosurg Psychiatry
1985; 48: 1085-90
8. Erkinjuntti T, Ketonen L, Sulkava R, Vuonalho M, Palo J CT in the differential
diagnosis between Alzheimer’s disease and vascular dementia Acta Neurol Scand
1987, 75: 262-70
9. Flaten TP An investigation of the chemical composition of Norwegian drinking water
and its possible relationships with the epidemiology of some diseases Thesis no 51,
Institutt for Norganisk Kjemi, Norges Tekniske Hogskole, Trondheim, 1986
10 Vogt T. Water quality and health-a study of a possible relationship between
aluminium in drinking water and dementia. Sosiale og okonomiske studier 61
1-99 Oslo Central Bureau of Statistics of Norway, 1986.
11. Martyn CN, Pippard EC Facts and fallacies in dementia epidemiology. Paper read at
Second International Symposium on Geochemistry and Health, London, April
1987
12 Martyn CN, Pippard EC Usefulness of mortality data in determining the geography
and time trends of dementia. JEpidemiol Community Health 1988; 42: 134-37,
13. Jorm A, Henderson A, Jacomb P Regional differences in mortality from dementia in
Australia. an analysis of death certificate data Acta Psychiatr Scand (in press).
COLD-INDUCED PULMONARY OEDEMA IN
SCUBA DIVERS AND SWIMMERS AND
SUBSEQUENT DEVELOPMENT OF
HYPERTENSION
P. T. WILMSHURST
A. CROWTHER
M. NURI*
M. M. WEBB-PEPLOE
Department of Cardiology, St. Thomas’ Hospital,
London SE1 7EH
Summary The effect of cold and/or a raised partial
pressure ofoxygen was examined in eleven
people with no demonstrable cardiac abnormality but who
had pulmonary oedema when scuba diving or surface
swimming, and in ten normal divers. These stimuli induced
pathological vasoconstriction in the pulmonary oedema
group, nine of whom also showed signs of cardiac
decompensation when so stimulated. The pulmonary
oedema patients have been followed-up for an average of 8
years. Seven have become hypertensive. Except for the
onset of lone atrial fibrillation in one normotensive female
diver and development of Raynaud’s phenomenon in a
normotensive man, there have been no cardiovascular
events and no deaths.
Introduction
ACUTE pulmonary oedema occurs when the pulmonary
capillary permeability is increased (non-cardiogenic
pulmonary oedema) or when the pulmonary capillary
hydrostatic pressure exceeds the plasma oncotic pressure
(cardiogenic pulmonary oedema). The term cardiogenic
pulmonary oedema implies that the primary defect causing
the oedema is within the left heart. Although this may be the
case in such conditions as myocardial infarction, there is
*Present address: Heart Clinic, Rawalpindi, Pakistan.