Human Reproduction vol.12 no.6 pp.

12751279, 1997

Risk factors for adenomyosis

Fabio Parazzini1,2,3, Paolo Vercellini1, Stefania

Panazza1, Liliane Chatenoud2, Sabina Oldani1 and
Pier Giorgio Crosignani1
1Centro

Medicina della Riproduzione, I Clinica Ostetrico

Ginecologica Universita` di Milano, via Commenda 12, 20122
Milano, and 2Istituto di Ricerche Farmacologiche Mario Negri,
via Eritrea, 62-20157 Milano, Italy
3To

whom correspondence should be addressed at:

Istituto di Ricerche Farmacologiche Mario Negri, via Eritrea,
62-20157 Milano, Italy

In order to analyse risk factors for adenomyosis, 707

consecutive women who underwent hysterectomy
between January 1993 and June 1994 at the Clinica
Luigi Mangiagalli, Milan, Italy, were interviewed before
surgery by trained physicians. Information on the
presence of adenomyosis was obtained from pathologic
records. Out of the 707 women, adenomyosis was
identified in 150 subjects (21.2%). Women who smoked
tended to be at decreased risk of the condition: in
comparison with women who had never smoked, the
risk for current smokers was 0.7 (0.31.3) and the risk
decreased with number of cigarettes smoked per day,
the odds ratios being 0.8 and 0.6 respectively in women
reporting fewer than 10 and more than 10 cigarette
smoked per day (2 trend 3.57, P J 0.06). The frequency
of adenomyosis was higher in parous women: in comparison with nulliparae, the odds ratio of the disease
were 1.8 [95% confidence interval (CI) 0.93.4] and 3.1
(95% CI 1.75.5) respectively in women reporting one
and two or more births (2 trend 20.71, P < 0.01).
Likewise, women reporting one or more spontaneous
abortions had an odds ratio of 1.7 (95% CI 1.12.6) for
adenomyosis, in comparison with those reporting no
spontaneous abortion.
Key words: adenomyosis/reproductive factors/risk factors

Introduction
Islands of adenomyosis in the uterine musculature are relatively
frequent, their prevalence ranging in series of hysterectomies
from ~570% (Bird et al., 1972; Lee et al., 1984; Young et al.,
1986; Gambone et al., 1990; Merrill and Creasman, 1990;
Chrysostomou et al. 1991). Despite its frequency, however,
epidemiological studies on adenomyosis are rare and little is
known about the epidemiological profile of women at risk.
Clinical series have suggested that the frequency of the
condition increases with age until the menopause and levels
off thereafter (Molitor, 1971). Multiparity has been associated
European Society for Human Reproduction and Embryology

with an increased frequency of adenomyosis (Molitor, 1971;

Azziz, 1989). Further menstrual characteristics, smoking and
history of dilatation and currettage have been associated with
the risk of the conditions (Vercellini et al., 1995).
In order to reanalyse the role of these factors, in this paper
we present the results of a cross-sectional study on risk factors
for adenomyosis conducted on more than 700 women who
underwent hysterectomy at the Clinica Mangiagalli of Milan,
Northern Italy. An analysis of epidemiological determinants
of risk of adenomyosis may be useful in suggesting an
aetiological hypothesis for adenomyosis, a condition potentially
associated with pelvic pain and subfertility.
Materials and methods
All women who consecutively underwent hysterectomy between
January 1993 and June 1994 at the Clinica Luigi Mangiagalli of
Milan, in Northern Italy, were identified before surgery by trained
physicians. After gynaecological examination, information was
collected on socio-demographic characteristics, with an oriented
medical history and reproductive and menstrual factors. The
presence of dysmenorrhoea, intermenstrual pelvic pain and dyspareunia was self-reported. Information regarding previous dilatation and currettage for gynaecological indications was confirmed by
direct inspection of clinical records whenever available. Dilatation
and curettage performed during the diagnostic work-up for surgery
(i.e. within 6 months before interview) were not considered.
The choice of information to be collected in the study was based
on the risk factors for adenomyosis as reported in the literature. A
total of 707 women (median age 50 years, range 1984) were
interviewed before hysterectomy. Of those, 372 (52.6%) underwent
surgery for fibroids and/or menorrhagia, 140 (19.8%) for genital
prolapse, 100 (14.1%) for ovarian cysts, 81 (11.5%) for genital cancer
and 14 (2.0%) for miscellaneous conditions.
Information on the presence of adenomyosis, endometriosis and
endometrial hyperplasia was obtained from pathology records.
Adenomyosis was diagnosed when the distance between the lower
border of the endometrium and the affected myometrial area was
over one-half of a low-power field (~2.5 mm) (Zaloudek and Norris,
1987). Endometriosis was diagnosed in the presence of endometrialtype epithelium and stroma. Fewer than 2% of eligible cases refused
to participate in the study.
Statistical analysis
The odds ratios of adenomyosis, together with their corresponding
95% confidence intervals (CI), in relation to various considered
factors were estimated from data stratified for quinquennia of age by
the MantelHaenszel procedure (Mantel and Haenszel, 1959). Where
appropriate, the linear trend in risk for exposure was computed with
the test given by Mantel (1963).
To allow for the effect of several potential confounding factors
simultaneously, unconditional multiple logistic regression, fitted by

1275

F.Parazzini et al.

Table I. Distribution of 707 womena, and corresponding odds ratios, according to presence of adenomyosis
and selected characteristics
Adenomyosis
yes
Education (years)
,7
51
7
99
Smoking habits
Never
108
Ever
,10 cigarettes day
11
10 cigarettes day
15
Ex-smokers
11
2 trendd
Oral contraceptive use
Never
105
Ever
35
IUD use
Never
123
Ever
16
Endometriosis
No
136
Yes
14
Diagnosis of endometrial hyperplasia
No
135
Yes
15

Odds ratio (95% CI)

no

Age adjustedb

Multivariate estimatesc

117
440

1d
0.7 (0.41.0)

1d
0.8 (0.41.2)

344

1d

1d

0.8 (0.41.5)
0.6 (0.31.1)
0.8 (0.41.6)
3.57 (P 5 0.06)

0.8 (0.41.8)
0.5 (0.30.9)
1.0 (0.52.1)
3.92 (P ,0.05)

366
157

11
1.2 (0.82.0)

11
1.1 (0.71.9)

459
62

1d
1.2 (0.72.3)

1d
1.3 (0.72.3)

511
46

1d
1.5 (0.82.9)

1d
1.6 (0.83.3)

537
20

1d
2.5 (1.25.2)

1d
2.7 (1.35.8)

48
93
45

CI 5 confidence interval; NS5 not significant.

aIn some cases the sum does not add up to the total because of missing values.
bAdjusted for age by the MantelHaenszel procedure.
cMultivariate estimate including terms for age, indication for hysterectomy, education, smoking habits, parity,
spontaneous abortion, history of dilatation and curettage, intensity of flow.
dReference category.

the method of maximum likelihood, was used (Breslow and Day,

1980). Included in the regression equations were terms for age,
indication for surgery, education and factors found significantly
associated with the risk of adenomyosis in the age-adjusted analysis.
The significance of the difference in the mean age in women with
and without adenomyosis was computed using Students t-test.

Results
Adenomyosis was identified in 150 of the 707 women (21.2%).
The frequency of adenomyosis was similar in women with
indication for surgery for ovarian cysts (30.0%) and prolapse
(31.4%), but was lower in women with fibroids and
menorrhagia (14.8%) or genital cancer (24.7%). The mean
age (6 SD) at hysterectomy was 52.9 (6 11.3) years in
women with adenomyosis and 47.8 (6 13.9) years in those
without the condition (P ,0.05). The distribution of subjects
according to the presence of adenomyosis and selected characteristics is shown in Table I. The risk of adenomyosis tended
to be lower in more educated women; the odds ratio for
adenomyosis was 0.7 (95% CI 0.41.0) in those reporting 7
years of schooling in comparison with women reporting
,7 years of schooling, but the finding was not statistically
significant. Smokers were at a lower risk; in comparison with
those who had never smoked, the risk for current smokers was
0.7 (95% CI 0.31.3) and decreased with number of cigarettes
smoked per day, being 0.8 for women smoking fewer than 10
cigarettes per day and 0.6 in those smoking 10 or more
cigarettes per day. The age-adjusted trend in risk was of
1276

borderline statistical significance (2 trend 3.57, P 5 0.06).

The odds ratio for ex-smokers was 0.8 (95% CI 0.41.6).
No relationship was found between risk of adenomyosis and
the use of oral contraceptives or an intrauterine device (IUD).
Endometrial hyperplasia in the pathology records was more
frequent among cases with adenomyosis: the estimated odds
ratio for adenomyosis was 2.5 (95% CI 1.25.2) in women
with endometrial hyperplasia in comparison with those without.
No significant association emerged between adenomyosis and
presence of endometriosis.
The frequency of adenomyosis was higher in parous women:
in comparison with nulliparae, the odds ratio for the disease
was 1.8 (95% CI 0.93.4) and 3.1 (95% CI 1.75.5) respectively
in women reporting one and two or more births (2 trend
20.71, P ,0.01, Table II). Likewise, women reporting one or
more spontaneous abortions had an odds ratio for adenomyosis
of 1.7 (95% CI 1.12.6), in comparison with those reporting
no spontaneous abortion. No relation emerged with age at first
birth and history of induced abortions (Table II).
Menstrual history and risk of adenomyosis is considered in
Table III. The risk of adenomyosis was higher in women selfreporting heavy flows, in comparison with those reporting
regular flows (odds ratio 1.7, 95% CI 1.12.6). In comparison
with women who did not report dilatation and curettage, those
reporting it had an odds ratio for adenomyosis of 2.2 (95% CI
1.44.0).
Age at menarche, menopausal status, lifelong menstrual
pattern, and duration of flows were not associated with the

Epidemiology of adenomyosis

Table II. Distribution of 707 womena, and corresponding odds ratios, according to presence of adenomyosis
and reproductive history
Adenomyosis
no

Age adjustedb

Multivariate estimatesc

19
32
99

202
131
224

1d
1.8 (0.93.4)
3.1 (1.75.5)
20.71 (P,0.01)

1d
1.5 (0.83.0)
2.6 (1.44.6)
13.81 (P,0.01)

50
49
25

131
128
87

1d
0.9 (0.61.5)
0.7 (0.41.2)
3.42 (P 5 0.06)

1d
1.1 (0.71.8)
0.9 (0.51.6)
0.33 (P 5 NS)

108
42

458
99

1d
1.7 (1.12.6)

1d
1.6 (1.02.4)

130
20

492
65

11
1.2 (0.72.0)

11
1.3 (0.72.2)

yes
Parity
0
1
2
2 trend
Age at first birth (years)
24
2427
.27
2 trend
Spontaneous abortions
0
1
Induced abortions
0
1

Odds ratio (95% CI)

CI 5 confidence interval; NS 5 not significant.

aIn some cases the sum does not add up to the total because of missing values.
bAdjusted for age by the MantelHaenszel procedure.
cMultivariate estimate including term for age, indication for hysterectomy, education, smoking habits, parity,
spontaneous abortion, history of dilatation and curettage, intensity of flow.
dReference category.

Table III. Distribution of 707 womena, and corresponding odds ratios, according to presence of
adenomyosis, menstrual history and dilatation and curettage
Adenomyosis
no

Age adjustedb

Multivariate estimatesc

27
76
47

128
293
136

1d
1.2 (0.71.9)
1.3 (0.72.2)
0.46 (P 5 NS)

1d
1.0 (0.61.7)
1.2 (0.72.1)
1.36 (P 5 NS)

78
72
(days)
21
125
4

370
187

1d
0.9 (0.41.9)

1d
0.9 (0.41.9)

63
470
24

1d
0.6 (0.31.1)
0.5 (0.11.6)

1d
0.7 (0.41.3)
0.6 (0.22.0)

98
43

371
164

1d
1.0 (0.71.6)

1d
0.9 (0.61.4)

88
58

353
194

1d
1.7 (1.12.6)

1d
1.4 (0.92.2)

129
21

520
37

1d
2.2 (1.24.0)

1d
2.1 (1.13.8)

yes
Age at menarche (years)
11
1213
14
2 trend
Menopausal status
Pre/in menopause
Post-menopause
Lifelong menstrual patterns
25
2630
31
Flow (days/month)
5
.5
Intensity of flow
Regular
Heavy
Dilatation and curettagee
Never
Ever

Odds ratio (95% CI)

CI 5 confidence interval; NS 5 not significant.

aIn some cases the sum does not add up to the total because of missing values.
bAdjusted for age by the MantelHaenszel procedure.
cMultivariate estimate including term for age, indication for hysterectomy, education, smoking habits, parity,
spontaneous abortion, history of dilatation and curettage, intensity of flow.
dReference category.
eFor gynaecological indication only.

frequency of adenomyosis. Presence of dysmenorrhoea, intermenstrual pelvic pain, dyspareunia and risk of adenomyosis
are considered in Table IV. Women with adenomyosis tended
more frequently to report intermenstrual pelvic pain (odds

ratio 1.6, 95% CI 0.73.4) and dyspareunia (odds ratio 1.4,

95% 0.63.0), but these findings were not significant.
Dysmenorrhoea was not related to the frequency of adenomyosis.
1277

F.Parazzini et al.

Table IV. Distribution of 707 subjectsa, and corresponding odds ratios, according to presence of adenomyosis
and selected symptoms
Adenomyosis
yes
Dysmenorrhoea
No
114
Yes
34
Intermenstrual pelvic pain
No
138
Yes
10
Dyspareuniae
No
134
Yes
13

Odds ratio (95% CI)

no

Age adjustedb

Multivariate estimatesc

408
142

1d
1.2 (0.71.8)

1d
1.0 (0.61.5)

515
36

1d
1.6 (0.73.4)

1d
1.3 (0.62.9)

506
43

1d
1.4 (0.63.0)

1d
1.3 (0.62.7)

CI 5 confidence interval; NS 5 not significant.

aIn some cases the sum does not add up to the total because of missing values.
bAdjusted for age by the MantelHaenszel procedure.
cMultivariate estimate including term for age, indication for hysterectomy, education, smoking habits, parity,
spontaneous abortion, history of dilatation and currettage, intensity of flow.
dReference category.
eOnly women reporting lifelong sexual intercourse.

Multivariate estimates of odds ratios were generally similar

to the age-adjusted ones, but the multivariate estimates for
spontaneous abortions were only of borderline significance
and the estimate for the intensity of flow was not statistically
significant. Conversely, the multivariate trend in risk for
cigarettes smoked per day was statistically significant.

Discussion
The population studied was women who had undergone
hysterectomy. Thus, these results should only refer to this
group of women with adenomyosis, and cannot be generalized
to all women with adenomyosis. The diagnosis of adenomyosis
is related to the pathologists awareness of the condition (Bird
et al., 1972). The pathologist in this study was not aware of
the study goal, and if anything, underdiagnosis of the disease
should tend to reduce a potential association. All patients who
underwent hysterectomy during the study period were included
in the analysis and diagnostic procedures did not change during
the study. Although information on menstrual symptoms and
intensity of flow could be partially misreported, it is unlikely
that information bias substantially differs in patients with or
without adenomyosis, since information was collected before
hysterectomy. Similar considerations may be made in relation
to the potential misreporting of induced abortions or misclassification of induced and spontaneous abortions.
In this series of hysterectomized women the prevalence of
adenomyosis was 21%. The reported frequencies range from
570% (Bird et al., 1972; Lee et al., 1984; Young et al.,
1986; Gambone et al., 1990; Merrill and Creasman, 1990,
Chrysostomou et al., 1991). Most of these differences are at
least partly attributable to different diagnostic criteria and
populations, e.g. autopsy or hysterectomy (Azziz, 1989). We
followed criteria which excluded sub-basal lesions (Bird et al.,
1972). A prevalence of ~20% emerged from a previous study
conducted with similar criteria in the same clinical setting
(Vercellini et al., 1995) and consistent findings have been
1278

reported in recent series of hysterectomized women (Lee et al.,

1984; Gambone et al., 1990).
Clinical observations have suggested that adenomyosis is
found more often in the uteri of multiparous than nulliparous
women (Azziz, 1986; Merrill and Creasman, 1990). In this
series the risk of adenomyosis was about twice in parous
women and increased with number of births. Similar findings
had emerged in the few epidemiological studies on the issue
(Vercellini et al., 1995). Foci of adenomyosis may be included
in the myometrium due to the aggressive action of the
trophoblast on the extension of myometrial fibres during
pregnancy. Otherwise the hormonal milieu of pregnancy may
favour the development of islands of adenomyosis.
In this study no association emerged between contraceptive methods (oral contraceptive and IUD use) and risk of
adenomyosis, however we have not analysed the role of
duration of oral contraceptive and IUD use, age at first use
and type of oral contraceptive used.
Hormonally-mediated mechanisms may be suggested to
interpret the reduced risk of adenomyosis in smokers. Lower
concentrations of oestrogens have been reported in smokers
(Baron et al., 1990) and adenomyosis has been suggested to
be oestrogen-dependent. High concentrations of oestrogens are
also related to endometrial hyperplasia (Kreiger et al., 1986),
and this condition was related to the risk of adenomyosis in
this series.
Further smoking has also been reported to reduce the risk
of endometrial cancer and fibroids (Parazzini et al., 1995,
1996); two conditions related to unbalanced oestrogen
exposure.
The increased risk of adenomyosis in women with a history
of dilatation and curettage may be interpreted in terms of
indication bias; women who required dilatation and curettage
may more frequently have a diagnosis of endometrial hyperplasia. However, the reciprocal adjustment for the two
covariates did not markedly change the estimated odds ratio.

Epidemiology of adenomyosis

Otherwise, the trauma of curettage may favour the inclusion

of islands of endometriosis in the myometrium.
Clinical observations have related adenomyosis with dysmenorrhoea, pelvic pain and uterine bleeding (Merrill and
Creasman, 1990; Owolabi and Strickler, 1977; Vercellini
et al., 1993). In this study no relationship was found between
adenomyosis and several menstrual characteristics including
polymenorrhoea and pain. The relationship with heavy
menstrual cycles disappeared in this analysis after adjustment
for potential covariates.
Finally, the presence of endometriosis was not associated
with adenomyosis. Endometriosis and adenomyosis have
different epidemiological characteristics. For example, parous
women are at a lower risk of endometriosis and a higher risk
of adenomyosis (Mangtani and Booth, 1993; Parazzini et al.,
1995a). These findings suggest that the two conditions are
different clinical and nosological entities with no shared
aetiological mechanisms.

reproductive and menstrual risk factors at different stages in Lombardy,

northern Italy. J. Epidemiol. Comm. Health, 49, 6164.
Parazzini, F., La Vecchia, C. and Negri E. (1995b) Smoking and risk of
endometrial cancer: results from an Italian case-control study. Gynecol.
Oncol., 56, 195199.
Parazzini, F., Negri, E., La Vecchia, C. et al. (1996) Uterine myomas and
smoking: results from an Italian Study. J. Reprod. Med., 41, 316320.
Vercellini, P., Ragni, G., Trespidi, L. et al. (1993) Adenomyosis: a de ja` vu?
Obstet. Gynecol. Surv., 48, 789794.
Vercellini, P., Parazzini, F., Oldani, S. et al. (1995) Adenomyosis at
hysterectomy: a study on frequency distribution and patient characteristics.
Hum. Reprod., 10, 11601162.
Young, R.H., Treger, T. and Scully, R.E. (1986) Atypical polypoid adenomyoma
of the uterus: a report of 27 cases. Am. J. Clin. Pathol., 139, 8690.
Zaloudek, C. and Norris, H.J. (1987) Mesenchymal tumours of the uterus. In
Kurman, R.J. (ed.), Blaunsteins Pathology of the Female Genital Tract.
3rd edn. Springer Verlag, New York, pp. 374.
Received on October 3, 1996; accepted on April 2, 1997

Acknowledgements
This work was conducted within the framework of the CNR (Italian
National Research Council) applied project Fattori di Rischio di
Malattia (sottoprogetto Fattori di Malattia nella Patologia Materno
Infantile).
The authors thank Ivana Garimoldi, Judy Baggott, and the GA
Pfeiffer Memorial Library Staff for editorial assistance and Carmela
Mezzanotte for help with data analysis.

References
Azziz, R. (1986) Adenomyosis in pregnancy. J. Reprod. Med., 31, 224227.
Azziz, R. (1989) Adenomyosis: current perspectives. Obstet. Gynecol. Clin.
N. Am., 16, 221235.
Baron, J.A., La Vecchia, C. and Levi, F. (1990) The antiestrogenic effect of
cigarette smoking in women. Am. J. Obstet. Gynecol., 162, 502514.
Bird, C.C., McElin, T. and Manalo-Estrella, P. (1972) The elusive adenomyosis
of the uterus revisited. Am. J. Obstet. Gynecol., 112, 583593.
Breslow, N.E. and Day, N.E. (1980) Statistical Methods in Cancer Research.
The Analysis of Case-Control Studies. Vol. 1. IARC Scientific Publication
no. 32, IARC, Lyon, France.
Chrysostomou, M., Akalestos, G., Kallistros, S. et al. (1991) Incidence of
adenomyosis uteri in a Greek population. Acta Obstet. Gynecol. Scand.,
70, 441444.
Gambone, J.C., Reiter, R.C., Lerich J.B. et al. (1990) The impact of a quality
assurance process on the frequency and confirmation rate of hysterectomy.
Am. J. Obstet. Gynecol., 163, 545.
Kreiger, N., Marrett, L.D., Clarke, E.A. et al. (1986) Risk factors for
adenomatous endometrial hyperplasia: a case-control study. Am. J.
Epidemiol., 123, 291301.
Lee, N.C., Dicker, R.C., Rubin, G.L. and Ory, H.W. (1984) Confirmation of
the preoperative diagnoses for hysterectomy. Am. J. Obstet. Gynecol., 150,
283287.
Mangtani, P. and Booth, M. (1993) Epidemiology of endometriosis. J.
Epidemiol. Comm. Health, 47, 8488.
Mantel, N. (1963) Chi-square tests with one degree of freedom, extension of
the MantelHaenszel procedure. J. Am. Stat. Assoc., 58, 690700.
Mantel, N. and Haenszel, W. (1959) Statistical aspects of the analysis of data
from retrospective studies of disease. J. Natl. Cancer Inst., 22, 719748.
Merrill, J.A. and Creasman, W.T. (1990) Disorders of the uterine corpus. In
Scott, J.R., Di Saia, P.J., Hammond, C.B. et al. (eds), Danforths Obstetrics
and Gynecology. 6th edn. J.B.Lippincott, Philadelphia, p. 1023.
Molitor, J.J. (1971) Adenomyosis: a clinical and pathologic appraisal. Am. J.
Obstet. Gynecol., 110, 275284.
Owolabi, T.O. and Strickler, R.C. (1977) Adenomyosis. A neglected diagnosis.
Obstet. Gynecol., 50, 424427.
Parazzini, F., Ferraroni, M., Fedele, L. et al. (1995a) Pelvic endometriosis:

1279