J Clin Epidemiol Vol. 51, No. 7, pp. 581–586, 1998 0895-4356/98/$19.

00
Copyright  1998 Elsevier Science Inc. All rights reserved. PII S0895-4356(98)00035-3

Mortality of Women with Polycystic Ovary Syndrome

at Long-term Follow-up
T. Pierpoint,1 P. M. McKeigue,1,* A. J. Isaacs,1,3 S. H. Wild,1 and H. S. Jacobs 2
1
Department of Epidemiology and Population Sciences, London School of Hygiene and Tropical Medicine;
2
Division of Endocrinology, University College London Medical School; and 3 Department of Academic
Therapeutics, Chelsea and Westminster Hospital, London

ABSTRACT. Metabolic disturbances associated with insulin resistance are present in most women with polycys-
tic ovary syndrome. This has led to suggestions that women with polycystic ovary syndrome may be at increased
risk of cardiovascular disease in later life. We undertook a long-term follow-up study to test whether cardiovascu-
lar mortality is increased in these women. A total of 786 women diagnosed with polycystic ovary syndrome in
the United Kingdom between 1930 and 1979 were traced from hospital records and followed for an average of
30 years. Standardized mortality ratios (SMRs) were calculated to compare the death rates of these women with
national rates. The SMR for all causes was 0.90 (95% CI, 0.69–1.17), based on 59 deaths. There were 15 deaths
from circulatory disease, yielding an SMR of 0.83 (95% CI, 0.46–1.37). Of these 15 deaths, 13 were from
ischemic heart disease (SMR 1.40; 95% CI, 0.75–2.40) and two were from other circulatory disease (SMR 0.23;
95% CI, 0.03–0.85). There were six deaths from diabetes mellitus as underlying or contributory cause, compared
with 1.7 expected (odds ratio 3.6; 95% CI, 1.5–8.4). Breast cancer was the commonest cause of death (SMR
1.48 based on 13 deaths; 95% CI, 0.79–2.54). We conclude that women with polycystic ovary syndrome do
not have markedly higher than average mortality from circulatory disease, even though the condition is strongly
associated with diabetes, lipid abnormalities, and other cardiovascular risk factors. The characteristic endocrine
profile of women with polycystic ovary syndrome may protect against circulatory disease in this condition. j
clin epidemiol 51;7:581–586, 1998.  1998 Elsevier Science Inc.

KEY WORDS. Polycystic ovary syndrome, mortality, insulin, resistance, cardiovascular disease

INTRODUCTION The objectives of this study were to determine whether

PCOS in young women is associated with increased cardio-
Polycystic ovary syndrome (PCOS) is now recognized to be
vascular mortality in later life, and more generally to estab-
very common; about 20% of women of reproductive age
lish whether the condition has long-term consequences for
have polycystic ovaries on ultrasound, and about half of
women’s health.
these have clinical or biochemical evidence of anovulation
and androgen excess [1–3]. In comparison with controls,
women with PCOS are resistant to insulin-mediated glu- METHODS
cose uptake [4] and have other metabolic disturbances asso- Sample Size
ciated with insulin resistance: glucose intolerance [5], hy-
From published studies of cardiovascular risk factors [8,9] in
perinsulinemia [5], high plasma triglyceride [6], low plasma
women with PCOS, and cohort studies in which these risk
high-density-lipoprotein (HDL) cholesterol [7,8], and cen-
factors had been measured in women at baseline [17,18], we
tral obesity [9]. This pattern corresponds to the insulin resis-
estimated that a three-fold relative risk of coronary heart
tance syndrome described in other groups at high risk of
disease in women with PCOS compared with the general
non-insulin-dependent diabetes and coronary heart disease
population was plausible. For 90% power to detect a three-
[10,11]. We [8,12] and others [6,13–15] have therefore sug-
fold relative risk at 5% significance, the expected number
gested that women with PCOS may be at increased risk of
of events must be at least five. At current mortality rates
coronary heart disease in later life, especially if the distur-
for England and Wales, five deaths from CHD would be
bances of lipid and carbohydrate metabolism persist after
expected if 1400 women were followed to age 55 years, or
the menopause as a study in Sweden has indicated [16].
600 women were followed to age 60 years. As the age distri-
*
Address for correspondence: Dr. Paul McKeigue, Department of Epidemi- bution of the cohort could not be accurately estimated in
ology and Population Health, London School of Hygiene and Tropical
Medicine, Keppel Street, London WC1E 7HT, UK. advance, a target of at least five expected deaths from CHD
Accepted for publication on 23 March 1998. was set for the sample size.
582
Case Ascertainment
To ensure that women diagnosed as young adults could be
followed at least until age 45 years, we restricted the search
to cases diagnosed before 1970. In the last few centers the
search was continued up to cases diagnosed in 1979. Al-
though ultrasound detection of polycystic ovaries was first
described in 1972 [19], ultrasound was not widely used to
diagnose PCOS until high-resolution real-time sector scan-
ners became available in the late 1970s [20]. Before that
time, the diagnosis of polycystic ovary syndrome could be
confirmed only by macroscopic or microscopic examination
of the ovaries. This was usually accompanied by wedge re-
section of the ovaries as a therapeutic measure [21]. In hos-
pitals where operating theater records and histopathology
records had been retained, complete ascertainment of cases
with a confirmed diagnosis was thus possible.
We enquired throughout the UK at all specialist women’s
hospitals and all general hospitals with at least 500 beds and
three consultant gynecologists on the staff in 1969. Permis-
sion to search the records was obtained at all except one
of the hospitals where records from before 1970 had been
retained. Histopathology records were the main means of
initial case ascertainment. Diagnostic indexes were used
where possible; otherwise records of specimens received
were searched to identify ovarian tissue specimens obtained
from wedge resection or biopsy. The original pathologists’
reports were examined, and in a few cases where these had
been destroyed the original slides were reviewed. The other
main source of cases was operating theater records, which
were searched for cases who had undergone wedge resection,
biopsy of ovaries, or culdoscopy. Other sources were hospital
admission and discharge records where these included a di-
agnosis, and indexes of patients by diagnosis or presenting
complaint which existed in a few centers. For all likely cases
identified by initial searches, case notes, histopathology rec-
ords, and other sources of identification data such as admis-
sion records were retrieved where they had not been de-
stroyed.
Case Definition
Each suspected case was scored positive, negative, or un-
known for the following three categories
1. Histological evidence of polycystic ovaries: at least two
of the following: thickened capsule, subcapsular follicu-
lar cysts, luteinization of the theca interna, increase in
ovarian stroma. Recent corpora lutea or corpora albican-
tia did not exclude the diagnosis.
2. Macroscopic evidence of polycystic ovaries: ovaries de-
scribed as enlarged, and fibrotic or pearly white when
examined at culdoscopy, laparoscopy, or laparotomy.
3. Clinical evidence of ovarian dysfunction: hirsutism, sec-
ondary amenorrhea, oligomenorrhea, clinical evidence
T. Pierpoint et al.
of chronic anovulation (oligomenorrhea or amenor-
rhea), and androgen excess (hirsutism).
The cases included in the cohort were classified as follows:
1. Definite PCOS: histological evidence with clinical evi-
dence of ovarian dysfunction.
2. Possible PCOS: histological evidence with clinical infor-
mation not available, macroscopic evidence with clini-
cal evidence of ovarian dysfunction, or clinical diagnosis
by an experienced consultant.
Mortality Follow-up
Identifying details were passed to the National Health Ser-
vice Central Registry, which holds a register of the entire
population of the UK dating back to 1939 that is continu-
ously updated with birth records, registrations with general
practitioners, emigrations, and death registrations. Identi-
fying details supplied to the Registry were matched with
records in the population register on the basis of name, date
of birth, National Health Service number, and address at
the time of diagnosis. Where a unique match was not
achieved, the usual reason was either that the full date of
birth was not available in the hospital records or that the
surname had changed on marriage. Where a successful
match was achieved the vital status was ascertained. Copies
of death certificates were obtained with the original Inter-
national Classification of Diseases coding assigned by the
Office of National Statistics. National mortality rates were
used to calculate expected deaths by underlying cause from
the number of woman-years at risk in each five-year age
group and five-year calendar period. For those who had been
diagnosed in Scotland or Northern Ireland, Scottish mortal-
ity rates were used (data for Northern Ireland are not avail-
able on disk); for all others, mortality rates for England and
Wales were used.
The standardized mortality ratio (SMR) is the ratio of
observed to expected deaths. Exact confidence intervals and
P-values for SMRs were calculated from the Poisson distri-
bution using the Stata package (Strata Corporation, TX).
Data for England and Wales in 1986 (the most recent year
for which multiple-cause coding is available) [22] were used
to compare the observed proportion of death certificates
which mentioned diabetes mellitus with the expected pro-
portion in each 10-year age group from 25–74 years. A con-
ditional maximum likelihood estimate was calculated for
the age-adjusted odds ratio for mention of diabetes mellitus
on the death certificates of women with PCOS.
RESULTS
We identified 1028 women treated for PCOS between 1930
and 1979 from hospital records in the UK: 440 in Greater
London, 430 elsewhere in England, 39 in Wales, 49 in Scot-
Mortality of Women with Polycystic Ovary Syndrome 583

TABLE 1. Deaths by underlying cause up to end of August 1997, compared with expected numbers derived from national
rates
Cause of death Observed Expected SMR 95% CI

All causes 59 65.3 0.90 0.69–1.17

All circulatory 15 18.1 0.83 0.46–1.37
Ischemic heart disease 13 9.3 1.40 0.75–2.40
Other circulatory 2 8.5 0.23 0.03–0.85
All neoplasms 27 29.7 0.91 0.60–1.32
Breast 13 8.8 1.48 0.79–2.54
Bronchus 1 4.1 0.24 0.01–1.35
Ovary 1 2.6 0.39 0.01–2.17
Other/unknown primary 12 14.2 0.85 0.44–1.48
External causes (injuries, poisoning, and ill-defined) 5 4.5 1.12 0.36–2.62
Diabetes mellitus 2 0.74 2.7 0.33–9.76
Other 8 12.3 0.65 0.28–1.28
No death certificate 2 — — —

land, and 70 in Northern Ireland. Of those, 556 cases were
found by searching through pathology records, 223 from op-
eration log books, 162 from admission books or discharge
summaries, and the remaining 87 from other record systems.
Case notes were retrieved for 64% of the cases identified.
For 77% (789 of 1028) of cases the date of birth was avail-
able from the original records, but for the others only the
year of birth could be obtained from the date of admission
and age on admission. We classified 641 cases as definite
and 387 as possible PCOS. Of the 387 cases classified as
possible PCOS, histological evidence of polycystic ovaries
was available on 192, and macroscopic evidence of polycys-
tic ovaries without histological examination was recorded
on 105. Only 90 cases had been diagnosed on clinical evi-
dence alone, and on 53 of these evidence of anovulation
(secondary amenorrhea or oligomenorrhea) and androgen
excess (hirsutism) was documented. Seventy-five percent
(770 of 1028) of cases had been treated by wedge resection
of the ovaries.
Of the 1028 cases identified, 786 (77%) were successfully
traced in the National Health Service Central Registry. For
the 789 women whose date of birth was available in the
original records, the trace rate was 85% (668 of 789). For
the 786 women who were traced, the mean age at diagnosis
was 26.4 years, and the average duration of follow-up was
30 years. The number of woman-years of follow-up in each
age group and calendar period was sufficient to yield 9.3
expected deaths from coronary heart disease, compared with
the original target of five expected events.
There were 59 deaths from all causes before age 75 years
between 1955 and the end of August 1997, giving an SMR
of 0.90 (Table 1). When this analysis was restricted to the
698 women whose date of birth had been available in the
original hospital records, the SMR did not change (45
deaths observed against 49.8 expected, SMR 0.90). Death
certificates were obtained for 57 of the 59 women who had
died. Breast cancer was the leading cause of death in the
cohort (13 deaths observed against 8.8 expected). Although
overall mortality from circulatory disease was not signifi-
cantly different from the national average, the distribution
of causes showed an unusual pattern: of 15 deaths from cir-
culatory disease, 13 were from ischemic heart disease and
only two (one from sub-arachnoid hemorrhage and the
other from mitral stenosis) were from other circulatory dis-
ease. The deficit of deaths from circulatory disease other
than ischemic heart disease was statistically significant (two
deaths observed against 8.5 expected, two-sided Poisson
P-value of 0.02). In women in this age range, cerebrovascu-
lar disease accounts for most deaths from circulatory disease
other than ischemic heart disease. The ratio of observed to
expected deaths from circulatory disease was lowest in the
age group 40–59 years (Table 2).
The observed proportion of death certificates on which
diabetes mellitus was mentioned as underlying or contribu-
tory cause of death was compared with the expected propor-
tion calculated from national data on deaths coded by mul-
tiple causes. Diabetes mellitus was mentioned on 6 of the
57 death certificates, compared with 1.7 expected (odds ra-
tio 3.6; 95% CI, 1.5–8.4; two-sided P 5 0.002).
TABLE 2. Deaths from all causes and from circulatory dis-
ease by age group, up to end of August 1997
Age group
Total deaths Circulatory deaths
(years) Observed Expected Observed Expected
20–39 6 8.1 3 1.3
40–59 39 38.0 3 9.1
60–74 14 19.3 9 7.7
Total 59 65.4 15 18.1
584
DISCUSSION
As far as we can tell, our cohort is representative of women
diagnosed with PCOS in the UK before 1970. Until ultra-
sound examination came into wide use in the late 1970s
the diagnosis could be confirmed only by macroscopic or
histological examination of the ovaries, requiring invasive
procedures such as culdoscopy, laparoscopy, or laparotomy.
Thus ascertainment from operating theater records and his-
topathology records should have yielded all cases with a
confirmed diagnosis in each center. Our search included all
UK hospitals where women with PCOS were likely to have
been treated in significant numbers and records were avail-
able. Although we have no data on the socioeconomic sta-
tus of women in the cohort, we have no reason to believe
that the selection of the cohort was biased by socioeco-
nomic factors, as access to National Health Service hospi-
tals is not dependent on income and there were few private
hospitals in the UK before 1979. The comparison with all
UK women as a control group is therefore appropriate.
For 91% of the women in this cohort, the diagnosis of
PCOS had been confirmed by histological or macroscopic
examination of the ovaries. As these women underwent in-
vasive procedures for the diagnosis or treatment of PCOS,
it is likely that they had more severe clinical manifestations
than are now typical of women with PCOS who are diag-
nosed on the basis of clinical, biochemical, and ultrasound
findings. As the severity of clinical manifestations of PCOS
is correlated with the severity of endocrine and metabolic
disturbances [23], selection bias in this cohort which led to
inclusion of cases with more severe clinical manifestations
would exaggerate any increased cardiovascular risk associ-
ated with such endocrine and metabolic disturbances.
The ability to match identifying details from hospital rec-
ords with a record in the national population register de-
pends upon the completeness of the identifying details, but
does not depend upon whether death has occurred. Thus
although only 77% of the original cohort of 1028 could be
traced, we have no reason to believe that this has biased the
mortality rates. When the analysis was restricted to women
whose date of birth was available in the case records, for
whom the trace rate was 85%, the SMR for all-cause mortal-
ity did not change.
As data on smoking and socioeconomic status were not
recorded in any standardized form in the hospital records,
we are unable to adjust for possible confounding by these
factors. This does not invalidate our main conclusion that
women with PCOS are not at higher risk of dying from
circulatory disease than the general population, but it does
introduce further uncertainty into any estimate of the effect
of PCOS as an independent risk factor for circulatory dis-
ease. We can however estimate the maximum upward revi-
sion of the risk ratio for CHD mortality in women with
PCOS compared with the general population that could re-
T. Pierpoint et al.
sult from adjustment for socioeconomic status and smoking
if confounding by these factors were present. Data on socio-
economic gradients in CHD mortality are available for a
1% sample of women aged 15–74 years in England and
Wales during 1976–81 [24]. In the highest of the four socio-
economic groups defined for this study, the standardized
mortality ratio for CHD was 0.66 (1 5 all women in En-
gland and Wales). Even if we make the unlikely assumption
that all the women with PCOS in this cohort were in the
highest socioeconomic group, and multiply the CHD mor-
tality rates for the standard population by 0.66 to adjust for
this confounding, the SMR for CHD among women in this
cohort would increase only from 1.4 to 2.1. Similarly we can
estimate whether failure to adjust for differences in smoking
rates could plausibly account for our failure to detect a
three-fold excess of CHD in women with polycystic ovary
syndrome. The prevalence of smoking among women in the
UK has remained around 30% for the last two decades [25].
As the risk ratio for death from CHD in smokers compared
with non-smokers is approximately two-fold among women
in the UK in middle age [26], we can calculate that the
CHD mortality rate for non-smoking women is approxi-
mately 0.77 times the rate for women in the general popula-
tion. Even if we assume that all the women with PCOS
in this cohort were non-smokers, and multiply the CHD
mortality rates for the standard population by 0.77 to adjust
for this confounding, the SMR for CHD among women in
this cohort would increase only from 1.4 to 1.8. We can
thus be confident that failure to adjust for possible con-
founding by smoking and socioeconomic status is not a
plausible explanation for our failure to find the risk ratio of
three or more for CHD mortality in women with PCOS
that our study was designed to detect. Other cardiovascular
risk factors such as obesity and diabetes are likely to be more
common in women with PCOS than in the general popula-
tion, but as these risk factors are features of the polycystic
ovary syndrome and may lie in the causal pathway between
PCOS and CHD, it is inappropriate to treat them as con-
founders.
Wedge resection of the ovaries, which was the standard
treatment for PCOS until the mid-1970s [21], was usually
successful in restoring ovulation but had only short-term
effects on androgen levels [27]. Oestrogen/progestagen ther-
apy was also used to lower androgen levels [27] but this does
not alleviate the central obesity, insulin resistance, and dys-
lipidemias associated with PCOS [9]. The distribution of
cardiovascular risk factors in women who had undergone
wedge resection for PCOS was examined in a clinical fol-
low-up study in Gothenburg, Sweden [6]. In comparison
with age-matched controls, the 33 women who had under-
gone wedge resection 22–31 years earlier had seven-fold
higher prevalence of diagnosed diabetes, three-fold higher
prevalence of treated hypertension, and average waist–hip
ratio higher by half a standard deviation [14]. These results
Mortality of Women with Polycystic Ovary Syndrome
suggest that the adverse distribution of cardiovascular risk
factors in women with PCOS persists after the menopause
and after wedge resection. Thus although we are not able
to look at the outcome of untreated PCOS in this cohort
where 75% of the women were treated by wedge resection,
this procedure does not appear to reverse the adverse distri-
bution of cardiovascular risk factors that has been described
in younger women with PCOS.
Mortality associated with diabetes is underestimated
when only deaths coded to diabetes as underlying cause are
included; including deaths from diabetes as contributory
cause partially overcomes this [28]. The significant excess
of deaths with mention of diabetes in women with polycys-
tic ovary syndrome, although based on small numbers, is
consistent with other reports that non-insulin-dependent
diabetes and impaired glucose tolerance are common in
women with PCOS after the age of 30 years [16,29]. This
is presumably a consequence of insulin resistance: women
with PCOS have higher insulin levels [6,9,30] and lower
insulin sensitivity [4] than weight-matched controls, except
when the comparison is restricted to lean women [31].
Applying the data on cardiovascular risk factors in Go-
thenburg women treated for PCOS to a predictive model
derived from a cohort study in the general population, Dahl-
gren and colleagues estimated that myocardial infarction
would be seven times commoner in women with PCOS
than in the general population [14]. This estimate of a
seven-fold relative risk is not consistent with the 95% con-
fidence interval of 0.8 to 2.4 that we obtain for the SMR
for CHD in this study. Since diabetes and hypertension are
risk factors for both CHD and stroke, we would expect to
see an excess risk of all circulatory disease in women with
PCOS; for this broader category the 95% confidence inter-
val for the SMR is narrow enough to exclude a relative risk
higher than 1.4, and there is no evidence of excess mortality
in any specific age group when observed and expected
deaths are tabulated by age group. It is possible that women
with PCOS are protected from circulatory disease by the
action of unopposed estrogen secreted during anovulatory
cycles. This explanation would be consistent with the ap-
parent protective effect of estrogen therapy against cardio-
vascular disease in post-menopausal women [32].
With longer follow-up and ascertainment of nonfatal
events in this cohort we will be able to define the long-
term consequences of PCOS more fully. For the moment it
is possible to reassure women with PCOS that, despite an
apparently unfavorable profile of cardiovascular risk factors,
their chance of dying before age 75 years is no higher than
average and they need not be specially targeted for interven-
tions to lower cardiovascular risk.
This study was supported by the UK Medical Research Council. We
are grateful to the hospital staff who helped us to retrieve case records, to
Sharon Rimmer and colleagues at the National Health Service Central
585
Registry for tracing the cases, and to Sheelagh Kerr and Joanie Wilkin-
son for programming help.
References
1. Polson DW, Adams J, Wadsworth J, Franks S. Polycystic ova-
ries—A common finding in normal women. Lancet 1988; 1:
870–872.
2. Clayton RN, Ogden V, Hodgkinson J, Worswick L, Rodin
DA, Dyer S, Meade TW. How common are polycystic ovaries
in normal women and what is their significance for the fertility
of the population? Clin Endocrinol 1992; 37: 127–134.
3. Farquhar CM, Birdsall M, Manning P, Mitchell JM, France
JT. The prevalence of polycystic ovaries on ultrasound scan-
ning in a population of randomly selected women. Aust NZ
J Obstet Gynaecol 1994; 34: 67–72.
4. Dunaif A, Segal KR, Futterwet W, Dobrjansky A. Profound
peripheral insulin resistance, independent of obesity, in poly-
cystic ovary syndrome. Diabetes 1989; 38: 1165–1174.
5. Burghen GA, Givens JR, Kitabchi AE. Correlation of hyper-
androgenism with hyperinsulinism in polycystic ovarian dis-
ease. J Clin Endocrinol Metab 1980; 50: 113–116.
6. Mattson LA, Cullberg G, Hamberger L, Samsioe G, Silfer-
stope G. Lipid metabolism in women with polycystic ovary
syndrome: Possible implications for an increased risk of coro-
nary heart disease. Fertility and Sterility 1984; 42: 579–584.
7. Lithell H, Nillius SJ, Bergh T, Selinus I. Metabolic profile in
obese women with the polycystic ovary syndrome. Int J Obes
1987; 11: 1–9.
8. Conway GS, Agrawal R, Betteridge DJ, Jacobs HS. Risk fac-
tors for coronary artery disease in lean and obese women with
the polycystic ovary syndrome. Clin Endocrinol 1992; 37:
119–125.
9. Rebuffe-Scrive M, Cullberg G, Lundberg PA, Lindstedt G,
Bjorntorp P. Anthropometric variables and metabolism in po-
lycystic ovarian disease. Horm Metab Res 1989; 21: 391–397.
10. Reaven GM. Role of insulin resistance in human disease. Dia-
betes 1988; 37: 1595–1607.
11. McKeigue PM, Shah B, Marmot MG. Relation of central obe-
sity and insulin resistance with high diabetes prevalence and
cardiovascular risk in South Asians. Lancet 1991; 337: 382–
386.
12. McKeigue P. Are women with polycystic ovary syndrome at
special risk of coronary heart disease? Clin Endocrinol 1992;
37: 117–118.
13. Wild RA, Painter PC, Coulson PB, Carruth KB, Ranney GB.
Lipoprotein lipid concentrations and cardiovascular risk in
women with polycystic ovary syndrome. J Clin Endocrinol
Metab 1985; 61: 946–951.
14. Dahlgren E, Janson PO, Johansson S, Lapidus L, Oden A. Po-
lycystic ovary syndrome and risk for myocardial infarction:
Evaluated from a risk factor model based on a prospective pop-
ulation study of women. Acta Obstet Gynecol Scand 1992;
71: 599–604.
15. Talbott E, Guzick D, Clerici A, Berga S, Detre K, Weimer K,
Kuller L. Coronary heart disease risk factors in women with
polycystic ovary syndrome. Arterioscler Thromb 1995; 15:
821–826.
16. Dahlgren E, Johansson S, Lindstedt G, Knutsson F, Oden A,
Janson PO, Mattson LA, Crona N, Lundberg PA. Women
with polycystic ovary syndrome wedge resected in 1956
to 1965: A long-term follow-up focusing on natural history
and circulating hormones. Fertility and Sterility 1992; 57:
505–513.
586
17. Lerner DJ, Kannel WB. Patterns of coronary heart disease
morbidity and mortality in the sexes: A 26-year follow-up of
the Framingham population. Am Heart J 1986; 111: 383–
390.
18. Lapidus L, Bengtsson C, Larsson B, Pennert K, Rybo E, Sjos-
trom L. Distribution of adipose tissue and risk of cardiovascu-
lar disease and death: A 12 year follow up of participants in
the population study of women in Gothenburg, Sweden. BMJ
1984; 289: 1257–1261.
19. Kratochwil A, Urban G, Friedrich F. Ultrasonic tomography
of the ovaries. Ann Chir Gynaecol Fenn 1972; 61: 211–214.
20. Swanson M, Sauerbrei EE, Cooperberg PL. Medical implica-
tions of ultrasonically detected polycystic ovaries. J Clin Ul-
trasound 1981; 9: 219–222.
21. Donesky BW, Adashi EY. Surgically induced ovulation in the
polycystic ovary syndrome: Wedge resection in the age of
laparoscopy. Fertility and Sterility 1995; 63: 439–463.
22. Office of Population Censuses and Surveys. Mortality statis-
tics: Cause. Review of the Registrar General on Deaths by
Cause, Sex and Age in England and Wales, 1986. Series
DH2 no. 13. London: HM Stationery Office; 1988.
23. Conway GS, Honour JW, Jacobs HS. Heterogeneity of poly-
cystic ovary syndrome: Clinical, endocrine and ultrasound fea-
tures in 556 cases. Clin Endocrinol 1989; 30: 459–470.
24. Goldblatt P. Mortality and alternative social classifications.
In: Goldblatt P, Ed. Office of Population Censuses and Sur-
veys Longitudinal Study: Mortality and Social Organization.
Series LS no. 6, London: HM Stationery Office; 1994: 163–
192.
T. Pierpoint et al.
25. Office of Population Censuses and Surveys. Health Survey
for England 1991. London: HM Stationery Office; 1993.
26. Isles CG, Hole DJ, Hawthorne VM, Lever AF. Relation be-
tween coronary risk and coronary mortality in women of the
Renfrew and Paisley survey: Comparison with men. Lancet
1992; 339: 702–706.
27. Barnes R, Rosenfield RL. The polycystic ovary syndrome:
Pathogenesis and treatment. Ann Intern Med 1989; 110:
386–399.
28. Fuller JH, Elford J, Goldblatt P, Adelstein AM. Diabetes mor-
tality: New light on an underestimated public health problem.
Diabetologia 1983; 24: 336–341.
29. Dunaif A, Graf M, Mandeli J, Laumas V, Dobrjansky A. Char-
acterization of groups of obese women with acanthosis nigri-
cans, impaired glucose tolerance and/or hyperinsulinemia. J
Clin Endocrinol Metab 1987; 65: 499–507.
30. Graf MJ, Richards CJ, Brown V, Meissner L, Dunaif A. The
independent effects of hyperandrogenaemia, hyperinsulin-
aemia and obesity in lipid and lipoprotein profiles in women.
Clin Endocrinol 1990; 33: 119–131.
31. Holte J, Bergh T, Berne C, Berglund L, Lithell H. Enhanced
early insulin response to glucose in relation to insulin resis-
tance in women with polycystic ovary syndrome and normal
glucose tolerance. J Clin Endocrinol Metab 1994; 78: 1052–
1058.
32. Meade TW, Berra A. Hormone replacement therapy and car-
diovascular disease. Br Med Bull 1992; 48: 276–308.