close

Вход

Забыли?

вход по аккаунту

?

The pill parity and rheumatoid arthritis.

код для вставкиСкачать
THE PILL, PARITY, AND RHEUMATOID ARTHRITIS
TIM D. SPECTOR. EVE ROMAN, and ALAN J. SILMAN
We report on a case-control study investigating
the relationship of oral contraceptive pill (OCP) use and
parity to the development of rheumatoid arthritis (RA).
Women with RA were compared with 2 separate control
groups, women with osteoarthritis (OA) and women
randomly selected from a population-based electoral
register. Nulliparity was found to be a risk factor for the
development of RA, with age-adjusted odds ratios of
1.82 (95% confidence interval [CI] 1.09-3.03) versus the
OA control group and 1.83 (95% CI 1.03-3.06) versus
the population control group. Use of OCPs before the
age of 35 was negatively associated with RA (odds ratio
0.56, 95% CI 0.29-1.12 versus the OA control group;
odds ratio 0.6, 95% CI 0.30-1.17 versus the population
control group). Some evidence of a duration-response
effect was seen, although the numbers were small. The 2
variables were also multiplicative, with nulliparous
non-OCP users having a 4-fold risk of RA compared
with parous OCP users. These findings suggest that
pregnancy and OCP use have a “protective effect” on
the development of RA, although the mechanism remains unclear.
-.
From the Depanment of Epidemiology. The London Hospital Medical College. and The London School of Hygiene and
Tropical Medicine, London, United Kingdom.
Dr. Spector is a Wellcome Research Trust Fellow.
Tim D. Spector. MSc, MRCP (current address: Department
of Rheumatology. St. Bartholomew’s Hospital. London. UK); Eve
Roman. PhD; Alan J. Silman. MD, FFCM (current address: Arthritis and Rheumatism Council Epidemiology Research Unit,
Manchester, UK).
Address reprint requests to l i m D. Spector. MSc, MRCP.
Department of Rheumatology.
-. St. Banholomew’s Hospital. London
EClA 7BE. UK.
Submitted for publication April 3. 1989; accepted in revised
form December 6. 1989.
Arthritis and Rheumatism, Vol. 33. No. 6 (June 1990)
Epidemiologic and immunologic evidence suggests that hormonal factors may play a role in the
etiology and course of rheumatoid arthritis (RA). This
hypothesis is based on several observations, including
a 2-3-fold excess incidence in women at all ages ( l ) , an
amelioration of the disease in 80% of women during
pregnancy (2). and reported variations in disease severity during the menstrual cycle (3). It has also been
suggested that pregnancy may play a role in the later
development of RA. However, the evidence about the
direction of this association is contradictory. One large
population study indicated that parous women are
more likely to develop R A (4). whereas another indicated that nulliparous women are at increased risk ( 5 ) .
A number of European studies have indicated
that women who have used the oral contraceptive pill
(OCP) are less likely to develop RA than women who
have not used it (6-9). Studics from the United States,
however, have failed to show a similar relationship
(10.1 1). The reasons for these discrepancies are unclear, but differences in study design, patient selection,
and methods of analysis have been implicated (12).
We report herein the findings of an investigation
into the effects of panty and OCP use on the development of RA.
PATIENTS AND METHODS
Patients. The study had a case-control design. Two
hundred ninety female outpatients (age range 35-70 years)
who attended 1 of 6 rheumatology centers in east London
were studied. The diagnosis of RA was made by the attending physician, according to the criteria of the American
Rheumatism Association (13). The majority of the patients
(89%) had used a second-line agent at some time, and 76%
were seropositive. Women <35 years old were excluded
783
ORAL CONTRACEPTIVES, PREGNANCY, AND RA
from the study because they were less likely to have completed their families. and women >70 years old were excluded because of the difficulties they may have had recalling
past events.
Two different control groups were studied. One
group consisted of 320 women in the same age range with a
clinical diagnosis of osteoarthritis (OA) (made by a rheumatologist and confirmed by radiography) who were attending
the same rheumatology outpatient clinics as the R A patients.
The diagnoses of the RA patients and OA controls were
subsequently validated wherever possible from medical records. The other control group consisted of women in the
same age range who were selected randomly from electoral
registers of greater London. Data on the socioeconomic
status of the women studied were not collected directly, but
the catchment population of the referring clinics is wide and
likely to be socioeconomically similar to greater London as
a whole. In the UK. electoral registration is obligatory,
includes all individuals over the age of 18. and is updated
annually.
Questionnaire. Information on pregnancies. OCP
use, and other reproduction variables was obtained by a
questionnaire accompanied by an explanatory letter mailed
between September 1986 and February 1987. The letter
explained that information was to be collected for a general
health survey in women. No hypotheses or particular factors
were mentioned. A second questionnaire was sent if no reply
had been received within 6 weeks. The questionnaire contained identical questions for all groups, on pregnancy,
marital status, OCP use (including ever use. age at first use.
and duration of use). and other gynecologic and obstetric
information. Women were asked to list all pregnancies in
chronological order, with each outcome. For the RA patients
and OA controls, additional questions were asked about
onset of rheumatic symptoms and date of first specialist
referral. The questions on pregnancy and the OCP in the
questionnaire were previously validated against information
collected at an interview as part of a previous study (14).
Two hundred seventy (93%) of the 290 questionnaires sent to women with RA and 292 (91%) of the 320
questionnaires sent to the OA group were returned. Of 1 ,OOO
questionnaires mailed to women randomly selected from the
electoral registers of greater London, 508 (50.8%) were
returned, and of these, 245 were from women age 35-69. The
proportion of women age 35-69 who b o t h received and
returned a questionnaire is difficult to determine. because as
electoral registers are compiled primarily for voting purposes, age is not recorded. In addition, addresses, especially
in the greater London area, are often incorrect or out of date.
Statistical analysis. Demographic and other variables
were compared between the K A and the control groups
initially b,y the Student's 1-test and the chi-square test. Thc
effects of parity and OCP use were further examined by
calculating crude odds ratios and 95% confidence .intervals
(CI) against both control groups. The logistic regression
package EGRET (epidemiological graphics, estimation, and
testing; Statistical and Epidemiology Research Corp. Seattle. WA) was used to adjust the odds ratios for the potential
confounding variables of year of birth, marital status, and
age at henarche. 'lhe results of logistic regression are
presented using categorical variables, i.e.. year of birth in
Table 1.
Baseline characteristics of the 3 study groups'
Total group
n
Age at survey
Nulliparous
Marital statub
Single
Married
Widowed
Divorced
Other
Age at menarche
Age at natural
menopause
Age 31 first symptoms
Age at first referral
Subgroup of women born
1926-1951 with
disease onset after
age 35
n
Age
OCP use before age 35
Age at first OCP use
Duration of OCP use
< I year
I
4 years
>4 years
RA
OA controls
270
58.1 ? 7.7
71 (26.3)
292
57.4 ? 8.9
52 (17.8)t
31 (11.5)
35 (12.0)
164 (60.7) 165 (56.5)
45 (16.7) 42 (14.4)
40 (13.7)
22 (8.1)
8 (3.0)
10 (3.4)
13.4 2 1.6 13.0 2 1.6t
48.4 2 4.5 48.5 2 5.2
47.9
49.5
2
2
9.9 49.2
9.8 53.2
f
2
10.3
9.9t
I57
150
52.9 2 5.9 50.8 2 6.6t
19 (12.7)
38 (24.2)t
28.2 2 3.5 26.6 2 4.3
3 (15.8)
I I (57.9)
5 (26.3)
2 (5.4)
18 (48.6)
17 (45.9)
Population
controls
245
50.7 ? 10.3+
33 (13.5)t
20 (8.2)
172 (70.2)
16 (6.5)
37 (15.1)
0 (0)
13.0 z 1.6t
49.0 2 4.4
NA
NA
I80
46.2 2 7 3
73 (40.6)t
25.3 2 4.2t
8 (11.1)
22 (30.6)
42 (58.3)
Values are the mean ? SD or the number We). RA = rheumatoid
arthritis; OA
osteoarthritis; NA = not applicable; OCP = oral
contraceptive pill.
t P < 0.05 versus RA group. by Student's !-test.
7
5-year period categories. In the parity analyses, OCP use
was taken into account. and in the OCP analyses. parity was
included in the model.
OCP use before the age of 35 was examined in the
subgroup of women born after 1926. Women born prior to
1926 were unlikely to have been exposed to OCP use. The 22
women with RA and I I women with OA whose disease, as
far as could be ascertained. began bcforc they were 35 years
old were excluded from this part of the analysis since the
effect of the OCP on disease onset could not be studied.
These exclusions left 150 RA patients. 157 OA controls. and
180 population controls for analysis of OCP use. The few
women who stated first OCP use at or after the age of 35
were designated as "never used" in all groups, as were those
whose reported use was < I month.
RESULTS
The baseline characteristics of the 3 groups are
compared in Table 1. The characteristics of all women
are given in the upper part of the table. and the
characteristics of women born between 1926 and 1951
in the lower part. The mean -t SD age and corresponding year of birth were similar in the RA and OA
784
SPECTOR ET AL
Table 2. Adjusted odds ratios for nulliparity and oral contraceptive pill use in the 3 study groups’
Adjusted odds ratio
(95% confidence interval)
Variable
Nulliparityt
OCP use before age 35t
RA vs. OA
RA vs. population
controls
1.82 (I.09-3.03)
0.56 (0.29-1.12)
1.83 (1.03-3.26)
0.60 (0.30-1.17)
~~
* See Table I for definitions.
t Adjusted for year of birth, marital status, OCP use, and age at
menarche.
t Adjusted for year of birth, marital status, parity, and age at
menarche. Calculations were based on women born 1926-1951 with
disease onset after age 35.
groups, but the population controls were significantly
younger ( P < 0.05). The percentage of nulliparous
women in the RA group was significantly greater than
in the OA control group or the population control
group (26.3% versus 17.8% and 13.5%, respectively; P
< 0.05). The percentage of single women was similar
in the RA and OA groups ( I 1.5% and 12.0. respectively). The mean ? SD age at menarche was later in the
RA group compared with the 2 control groups. No
major differences were noted in the ages at natural
menopause between the groups. The ages at first
symptoms and referral were similar in the RA and OA
groups, although the RA patients were referred sooner
(mean 1.6 years after the onset of symptoms) than the
OA group (Table 1).
For women born between 1926 and 1951 with
disease onset after age 35, OCP use before the age of
35 was significantly lower in the RA group (12.7%)
than in the OA group (24.2%) or the population controls (40.6%). and fewer women used OCPs for more
than 4 years. The effects of nulliparity and OCP were
further examined, and the adjusted odds ratios for this
analysis are presented in Table 2. The differences in
adjusted odds ratios for nulliparity for all women were
similar to the crude ratio, and reached conventional
statistical significance for comparisons of the RA
group versus the other groups. Thus. nulliparity appeared to be more frequent in women with KA. The
effect of increasing parity was also examined. No
significant differences were seen for those women
having 4 or more pregnancies, although the numbers
involved were small.
The odds ratios for OCP use before the age of
35, adjusted for year of birth, age at menarche, marital
status, and parity are shown in Table 2. Reclassifying
OCP use after age 35 as “exclusions” rather than
“never used” produced similar results, as did analysis
using year of birth and menarche either as continuous
or factored variables.
Initial differences in the crude rates of past
and/or present OCP use (Table I ) , which showed a
significant reduction in the rate of use among women
with RA and a higher rate in the population control
group than in the OA controls, became smaller after
adjustment for year of birth. The adjusted odds ratios
were 0.56 for R A versus OA and 0.60 for RA versus
controls (Table 2). For both comparisons, the upper
95% confidence limit exceeded unity. Analysis of OCP
use for different duration categories produced odds
ratios, for the comparison with both control groups,
that were lower in the long duration users (Table 3).
OCP use for <I2 months produced odds ratios above
unity, although only 13 women were in this category.
The odds ratio for the R A group versus both control
groups, for OCP use longer than 48 months compared
with “never used,” was 0.45 (95% CI 0.214.95).
Although the sample sizes in each group were small
and the confidence intervals were wide, the chi-square
test for trend showed a significant difference ( P <
0.01). These findings were compatible with a durationresponse relationship.
To examine the combined effect of parity and
OCP use. women born between 1926 and 1951 were
assigned to 1 of 4 parity/OCP use groups: parous OCP
users, nulliparous OCP users, parous non-OCP users,
and nulliparous non-OCP users. Parous OCP users
were used as the baseline group against which odds
ratios of the other categories were compared. The
numbers of women in these 4 categories are given in
Table 4. There were no nulliparous OCP users in the
RA group and consequently, the adjusted odds ratio
could not be calculated for this category. Adjusted
odds ratios for nulliparous non-OCP users compared
Table 3. Adjusted odds ratios for oral contraceptive pill use and
duration of use in the 3 study groups’
Adjusted odds ratio (Y5% Confidence interval)
Duration of
use (years)
0
<I
1 4
>4
RA vs. OA
RA vs. population
controls
I (referent)
I (referent)
1.70 (0.48-5.98)
0.88 (0.39-1.99)
0.49 (0.21-1.13)t
1.80 (0.54-6.05)
1.08 (0.47-2.47)
0.45 (0.20-1.03)t
* Odds ratios were adjusted for year of birth, marital status. age at
menarche. and parity. See Table I for definitions.
t P < 0.01 versus “never used” group, by chi-square test for trend.
785
ORAL CONTRACEPTIVES, PREGNANCY, AND RA
Table 4.
Adjusted odds ratios for parity and OCP use in women born between 1926 and 1951, according to parity/OCP groups*
Adjusted odds ratio
(95% confidence interval)t
~
Group
Parous OCP users
Nulliparous OCP users
Parous non-OCP users
Nulliparous non-OCP
users
RA (n
=
150)
19 (12.7)
0 (0)
95 (63.3)
36 (24)
OA controls
(n = 157)
Population controls ( n =: 180)
RA vs.
OA
RA vs.
population
controls
32 (20.4)
6 (3.8)
102 (65)
17 (10.8)
64 (35.5)
9 (5)
95 (52.8)
12 (6.7)
I (referent)
1 (referent)
NA
1.25 (0.62-2.5)
4.65 (1.68-12.91)
NA
1.39 (0.71-2.75)
4.13 (1.67-13.37)
-
* Values are the number (%I. See Table 1 for definitions.
t Adjusted for year of birth, marital status. and age at menarche.
with parous OCP users were 4.65 (95% CI I .68-12.91)
for RA versus O A and 4.73 (95% CI 1.67-13.37) for RA
versus population controls. The odds ratio for parous
non-OCP users was lower, and the confidence intervals included unity. Only 14 women had their first
pregnancy after age 35, and reclassification of these
women as “nulliparous” did not alter the results.
DISCUSSION
Three major conclusions can be drawn from our
findings in this study: I ) Nulliparity may be a risk
factor or marker for the development of R A in women.
2) OCP use before the age of 35 was associated with a
reduction in the risk of developing R A at a later age,
and OCP use of more than 48 months may confer
“greater protection” against RA. 3) The combined
effect of nulliparity and non-OCP use before age 35
results in a 4-fold increase in the subsequent risk of
R A , compared with parous OCP-using women.
Before discussing these findings in detail, the
possible biases in this study will be considered. The
quality of the exposure data may be questioned, since
no validation procedures were carried out in this
retrospective study, which was based on subject recall. However, random error in exposure data, if
present, would lead to a dilution of any observed effect
and would not have an important effect on the results.
Several studies have shown good agreement between
questionnaire data and physicians’ records of basic
data on OCP use (15-17). The questions on pregnancy
had been validated by personal interview in a previous
study (14). Selective recall bias, especially between
the 2 disease groups, although potentially important,
was unlikely to be a problem since the R A and O A
groups were questioned in identical ways, and the
letter explaining the reasons for the study was the
same.
The poorer response rates in the women from
the electoral register were to be expected (18) and
could have affected the results if responders in that
group were more likely than nonresponders to have
been pregnant and/or to have taken the OCP. Findings
of methodologic studies of the effects of nonresponse
on results of similar investigations are often contradictory and difficult to interpret. However, 1 study of
response rates and estrogen use found that nonresponders were less likely to have used estrogens (191,
and another mail survey found no correlation between
any estrogen use or gynecologic information and response rates (20). A recent. large, casecontrol study
in the UK found no differences in OCP use between
control responders and nonresponders, although nonresponders were more likely to be nulliparous (211.
However, in the present study, the odds ratios were
similar for panty and OCP use in both the patient and
the population control groups.
Disease onset was taken as the date of first
referral. based on the patients’ medical records. In our
hospitals. the median delay from first symptoms to
referral was 21 months. Using the date of onset as the
date of first symptoms (when available), however, did
not alter the results. Protopathic bias has been proposed as a potential distorting factor in OCP studies,
whereby women who have subclinical disease alter
their behavior and are less likely to subsequently use
OCPs. This bias is impossible to discount given the
uncertainties of defining R A onset. However, since the
mean interval between the start age of OCP use and
the start of clinical symptoms of RA was 17 years, we
believe it unlikely that this bias occurred to any extent
in this study. Only 22 of 439 parous women (5.0%)
786
SPECTOR ET AL
Table 5. Summary of findings in studies of oral contraceptive pill (OCP) use and rheumatoid arthritis.
95%
Reference
Cases
Controls
Cohort number
Odds ratio
for "ever
used OCPs"
Wingrave and Kay (6)
Vandenbroucke et a1 (7)
Linos et al (10)
Allebeck et al (8)
Del Junco et al ( 1 I )
Vandenbroucke et al(9)
Vessey et a l (25)
Present study*
94
228
229
76
182
246
78
150
NA
302
458
152
182
232
NA
337
46.000
NA
NA
NA
NA
NA
17.034
NA
NA
0.40
I .70
0.70
1.1
0.55
NA
0.55
Relative risk
for "ever
used OCPs"
confidence
interval
0.6U
0.45-1.03
0.22-0.72
0.8-0.35
0.40-1.24
0.7-1.7
0.3 1-0.97
0.79-1.79
0.30-1.01
~~
NA
NA
NA
NA
NA
I . 12t
NA
* N A = not applicable.
t Estimated results; actual relative risk not given.
$ Pooled results from the osteoarthritis and population control groups.
from the patient groups had their disease onset prior to
their first pregnancy, and exclusion of this group did
not affect the results.
Large social. ethnic, or religious differences
between the groups could have occurred by chance,
although this is unlikely to have occurred in the RA
and OA groups since they were drawn from the same
local populations and more than 95% were white.
Whether the randomly selected women from the electoral register were similar to the women with OA is
difficult to judge. In general, the women attending the
clinics were drawn from a wide area that included a
cross-section of social class groups, similar to that
expected in greater London. Two percent of the
electoral register women would have been expected to
be RA cases, but this would not have altered any
results.
Based on these results, there was undoubtedly
a large cohort effect of year of birth on the use of
OCPs, with adjustment for year of birth reducing the
apparent protective effect of OCP use-the former
being a powerful predictor of the latter. However, year
of birth adjustment ultimately produced similar results
in the 2 control groups. If current trends continue,
studies of OCP exposure will no longer be possible: A
recent study found that only 5 . I% of young females in
the U K had never been exposed to OCP use (22).
There have been previous studies on the effects
of parity and fertility in RA. Hargreaves, in a study of
65 married women with RA from West Cornwall, UK.
made the observation that they appeared to have fewer
children than control women (23). A case-control
study of family size and RA concluded that the women
with RA had an overall reduced fertility both before
and after disease onset (24). Further analysis of those
data, however, showed that most of this difference
was due to increased rates of nulliparity prior to
disease onset (21.7% in married women with RA
versus 16% in women with OA), with an odds ratio of
1.97, similar to that found in the present study. Two
large, cross-sectional population studies of RA have
shown opposite results regarding the effect of parity.
As part of the large U S National Health Examination
survey of 1960-62, Engel found that nulliparous
women above the age of 45 had a reduced prevalence
of RA. although the prevalence was slightly increased
between ages 25 and 44 (4). Lawrence, however. in
population studies in northern England, found that the
prevalence of RA among 25-65-year-old women was
increased in nulliparous individuals (5). Reasons for
the discrepancies in these 2 population studies are
unclear, although regional differences in diagnosis and
the diagnostic uncertainties that surrounded RA in the
1960s may play a part.
From the results of the study by Vandenbroucke et al (9), we were able to calculate a crude
odds ratio for nulliparity of 1.84 for women with RA
versus controls, a figure similar to that found in our
study. Thus, previous studies offer some support for
the notion that nulliparity is a risk factor for RA.
Although the upper confidence limits included
unity, our finding of a protective effect of OCP use on
the order of 40% concurs with the findings of other
published studies (6-9), but not with 2 studies from the
Mayo Clinic, which failed to show any effect of OCP
use (10.1 1) (Table 5 ) . A preliminary report of a prospective study of married US nurses has indicated no
difference in OCP use among those who have developed RA since 1976 (25). A recent report from the
Oxford Family Planning Study has also shown no
association between RA and OCP use (26). Although
excellent exposure data were available in that study,
ORAL CONTRACEPTIVES, PREGNANCY, AND RA
the authors admit that case ascertainment relying on
self-reporting and general practitioners' referrals may
have been inaccurate.
Methodologic differences between the studies
have formed the basis of considerable controversy,
which has yet to be resolved (12.27). A meta-analysis
of the available studies suggested that the protective
effect was demonstrated only in those studies using
hospital cases (28). and a recent consensus conference
of key researchers in the field supported this view (29).
These findings could be explained if prior use of OCPs
modified the disease process, producing a milder form
of RA. In these cases of RA. patterns of OCP use
would be similar to those in controls. Furthermore, in
the present study, inpatients with definite RA, usually
of long duration (mean 9 years), were chosen. In
addition, the majority of patients (76%) were seropositive, and nearly 90% had received disease-modifying
drugs. As with other similar studies, our data could be
explained by OCP use protecting against severe and
continuing disease, rather than disease susceptibility
per se. Differences between countries may thus have
occurred due to differences in patient selection and
referral patterns.
Another possible explanation for the divergent
results is a difference in the composition of the estrogens. The potency of these agents has decreased
markedly in the past 21 years. as has that of the
various types of progestogens. whose androgenic effects differ markedly (30). These secular changes
should be considered in any intercountry comparisons, as the changes in estrogen dosages and types of
progestogens have probably occurred over different
time periods. The finding of a duration-response effect
of OCP use in this study would be consistent with an
etiologic role of OCP use. but we cannot explain w h y
other studies have failed to find such a trend (7). As far
as we were aware, no other study has looked specifically at the relationship between OCP use. pregnancy
status, and RA, and several of the early studies failed
to obtain any information about pregnancy status.
If nulliparity and non-OCP use are both risk
factors in RA. how might they act? There are 2
possibilities. The first is that women developed RA
because they .had not taken the O C P ' o r become
pregnant, or both. The second is that non-OCP use
and nulliparity are markers of either early (subclinical)
disease or another risk factor. In support of the latter
is the observation that there was no increase in the
number of single women in the RA group, which might
have been expected if nulliparity was an independent
787
risk factor. What this other risk factor may be is
speculative, but it could be related to certain patterns
of sexual behavior, such as abstinence, or certain
personality traits. Interestingly, none of the 36 nulliparous women with R A used the pill, compared with 6
of 23 (26%) in the OA group and 9 of 21 (43%) in the
population control group. Why none of the nulliparous
women with RA took OCPs is difficult to explain.
Although we lack the necessary additional information
from these women themselves, 2 main possibilities
exist; either the OCPs were not needed or they were
not wanted. Infertility before the age of 35 is unlikely
to be the major factor in deciding on OCP use. Thus,
these findings may reflect differences in social or
sexual attitudes, such as sexual activity or other
differences in lifestyle.
Could the increase in nulliparity have been due
to the presence of subclinical disease 20 or 30 years
earlier. which caused some immunologic dysfunction?
This explanation is difficult to disprove. although one
might expect fertility in the parous women to have
been diminished by the same mechanism, producing a
smaller family size, and this was not found (31).
It is difficult to propose a simple biologic basis
for the association between pregnancy, OCP use, and
prevention of RA. However, several observations
point to a relationship. namely, the effect of pregnancy
(2) and the menstrual cycle (3) on the disease and the
possible beneficial effects of postmenopausal estrogens both on the disease process (32) and in prevention
(9). Modulation of the immune system by androgens
and estrogens has been demonstrated for some time in
animals (33), and alterations in the composition of
immunoglobulins may occur during pregnancy (34).
Both the OCP and pregnancy have been found to
produce a marked elevation of specific immunosuppressant factors in humans, such as pregnancyassociated alpha-2 glycoprotein (35,36). However,
these factors are not increased in all women, and the
increases do not appear to have any long-term effects.
Recently, large population studies have demonstrated
small but significant differences in sex hormonebinding globulin, which regulates estrogen levels, in
women who have ever been pregnant or ever received
the OCP (37,381. This implies a permanent alteration in
hormonal status. Another possibility is that the lifestyle changes associated with either OCP use or pregnancy alter the rates of exposure to potential pathogenic agents such as viruses.
Analogies with other diseases reveal that some
neoplastic diseases are known to be modified by
788
SPECTOR ET AL
hormonal influences and resemble RA in a number of
features. These include carcinoma of the ovary and
uterus. Both of these are more common in nulliparous
and single women and rarer in those taking the OCP.
occurring generally after menopause. Although a
crude oversimplification, it has been suggested that
OCPs have an effect rather like a state of pregnancy,
with both leading t o inhibition of ovulation. Recently,
it has been observed that the process of ovulation itself
stimulates interleukin-l production (39). and so it is
possible that OCP use and pregnancy act similarly in
regard t o the immune system, leading t o some longlasting immunoprotection. Nevertheless, although a
firm biologic explanation seems elusive, the observation that nulliparous non-OCP users have a 4-fold
increased risk of RA should stimulate continued research in this area.
ACKNOWLEDGMENTS
We thank the rheumatology consultants and staff of
St. Bartholomew's Hospital. The London Hospital, Homerton Hospital. Whipps Cross Hospital, Wanstead Hospital,
and St. Andrew's Hospital for their help with this study.
10.
11.
12.
13.
14.
15.
16.
REFERENCES
1. Linos A, Worthing JW, O'Fallon WM, Kurland LT: The
2.
3.
4.
5.
6.
7.
8.
9.
epidemiology of rheumatoid arthritis in Rochester, Minnesota: a study of its incidence, prevalence and mortality. Am J Epidemiol I I1:87-98, 1989
Persellin RH: The effect of pregnancy on rheumatoid
arthritis. Bull Rheum Dis 27:9226, 1977
Rudge SR, Kowanko IC, Drury PL: Menstrual cyclicity
of finger joint size and grip strength in patients with
rheumatoid arthritis. Ann Rheum Dis 42:425430, 1983
Engel A: RA in US adults, Population Studies of the
Rheumatic Diseases. Edited by PH Bennett, PHN
Wood. Amsterdam, Excerpta Medica, 1%8
Lawrence JS: Rheumatism in Populations. London,
Heinemann, 1977
Wingrave SJ, Kay CR: Reduction in incidence of rheumatoid arthritis associated with oral contraceptives.
Lancet 1569-571, 1978
Vandenbroucke JP, Valkenburg HA, Boersma JW, Cats
S. Fester JJM. Huber-Bmning 0, Rasker JJ: Oral contraceptives and rheumatoid arthritis: further evidence
for a protective effect. Lancet I1:839-842, 1982
Allebeck P, Ahlbolm A, Ljungstrom K, Allander E: Do
oral contraceptives reduce the incidence of rheumatoid
arthritis? Scand J Rheumatol 13: 140-146, 1984
Vandenbroucke JP, Witteman JCM. Valkenburg HA,
Boersma JW, Cats A. Festen JJM, Hartman AP, Huber-
17.
18.
19.
20.
21.
22.
23.
24.
25.
Bruning 0, Rasker JJ, Weber J: Noncontraceptive hormones and rheumatoid arthritis in perimenopausal and
postmenopausal women. JAMA 255: 1299-1303, 1986
Linos A. Worthington JW. O'Fallon WM, Kurland LT:
Case-control study of rheumatoid arthritis and prior use
of oral contraceptives. Lancet 1: 1299-1300. 1983
Del Junco DJ, Annegers JF, Luthra HS, Coulam CB,
Kurland LT: Do oral contraceptives prevent rheumatoid
arthritis? JAMA 254: 1938-1941, 1985
Esdaile JM, Honvitz I: Observational studies of causeeffect relationships: an analysis of methodologic problems as illustrated by conflicting data for the role of oral
contraceptives in the etiology of rheumatoid arthritis. J
Chronic Dis 39:841-852. 1986
Ropes MW. Bennett GA. Cobb S, Jacox R, Jessar RA:
1958 revision of diagnostic criteria for rheumatoid arthritis. Bull Rheum 13s 9:175-176. 1958
Silman AJ, Roman E. Beral V. Brown A: Adverse
reproductive outcomes in women who subsequently
develop rheumatoid arthritis. Ann Rheum Dis 47:
979-98 I , 1988
Glass R. Johnson B, Vessey M: Accuracy of recall of
histories of oral contraceptive use. Br J Prev SOCMed
28:273-275, 1974
Stolley PD, Tonascia JA. Sartwell PE. Tockman MS,
Tonascia S. Rutlcdge A, Schinnar R: Agreement rates
between oral contraceptives users and prescribers in
relation to drug use histories. Am J Epidemiol 107:
226-235. 1978
Rosenberg MJ, Layde PM. Ory HW, Strauss LT, Rooks
JB, Rubin GL: Agreement between women's histories of
oral contraceptive use and physician records. Int J
Epidemiol 12:84-87, 1983
Norell SE. Ahlbolm A: Hospital versus population referents in two case-referent studies. Scand J Work Environ Health 13:62-66, 1987
Honvitz RI. Feinstein AR. Stremlan JR: Alternative
data sources and discrepant results in case-control studies of oestrogens and endometrial cancer. Am J Epidemiol 1 I1:38%394, 1980
Cartwright A: Who responds to postal questionnaires? J
Epidemiol Community Health 40:267-273. 1986
U K National Case Control Study Group: Oral contraceptive use and breast cancer risk in young women.
Lancet 1:973-982, 1989
Royal College of General Practitioners, Manchester
Research Unit: New oral contraceptive study: pilot trial
report. J R Coll (;en Pract 36:293. 545. 1986
Hargreaves ER: A survey of rheumatoid arthritis in
West Cornwall: a report to the Empire Rheumatism
Council. Ann Rheum Dis 17:61-67. 1958
Kay A, Bach F: Subfertility before and after the development of rheumatoid arthritis in women. Ann Rheum
Dis 24:169-173, 1965
Vessey MP, Villard-Mackintosh L, Yeates D: Oral
ORAL CONTRACEPTIVES, PREGNANCY, AND RA
26.
27.
28.
29,
30.
31.
32.
contraceptives, cigarette smoking and other factors in
relation to arthritis. Contraception 35:457464. 1988
Hernandez-Avila M. I.iang M. Willett WC. Stampfer
MJ, Colditz G , Rosner B, Hennekens CH, Spcizer FE:
Oral contraceptives and postmenopausal hormone use
and the risk of rheumatoid arthritis (abstract). Arthritis
Rheum 31 (suppl 4):S36. 1988
Spector TD. Silman AJ: Oral contraceptives and rheumatoid arthritis: bias in observational studies of cause
effect relationships. J Chronic Dis 40: 3063-3064. 1987
Spector TD, Hochberg MC: The protective effect of the
oral contraceptive pill on rheumatoid arthritis: an overview of the analytical epidemiological studies using
meta-analysis. Br J Rheumatol 28(suppl 1):11-12. 1989
Silman AJ, Vandenbroucke JP. editors: Female Sex
Hormones and Rheumatoid Arthritis: Proceedings of an
International Workshop, Leiden. The Netherlands,
March 20-21, 1989. Br J Rheumatol 28 (suppl I). 1989
Piper JM, Kennedy DL: Oral contraceptives in the
United States: trends in content and potency. Int J
Epidemiol 16:215-221. 1987
Spector TD. Silman AJ: Is adverse reproductive outcome a risk factor for rheumatoid arthritis? Ann Kheum
Dis 49:12-14. 1990
Bijlsma JWJ. Huber-Rruning 0, Thijssen JHH: Effect of
oestrogen treatment on clinical and laboratory manifestations of rheumatoid arthritis. Ann Rheum Dis 46:
777-779, 1987
789
33. Ahmed SA. Penhale WJ, Tala1 N: Sex hormones. immune responses. and autoimmune diseases: mechanisms
of sex hormone action. Am J Pathol 121:531-551, 1985
34. Stanworth DR: A possible immunochemical explanation
for pregnancy associated remissions in rheumatoid arthritis. Ann Rheum Dis 4739-90. 1988
35. Ungar A. Kay A. Griffin JA. Panayi GS: Disease activity
in rheumatoid arthritis during pregnancy. Br Med J
286:75&752. 1983
36. Von Schoultz B, Sligbrand T: Pregnancy zone protein:
chemistry. biology and clinical studies, Pregnancy Proteins: Biology, Chemistry and Clinical Application. Edited by JG Grudzinskas, I3 Teisner. M Seppala. London,
Academic Press. 1982
37. Moore JW, Key TJA, Bulbrook RD. Clark GMG. Allen
DS. Wang DY Pike MC: Sex hormone binding globulin
and risk factors for breast cancer in a population of
normal women who had never used exogenous sex
hormones. Br J Cancer 56:661-667, 1987
38. Key TJA. Pike MC. Moore JW. Bulbrook RD. Clark
GMG. Allen DS. Wong I>Y: The relationships of SHBG
with current and previous use of oral contraceptives and
estrogen replacement therapy. Contraception 39: 179186, 1989
39. Cannon JG. Dinarello CA: Increased plasma interleukinI activity in women after ovulation. Science 227:
1247- 1249, I985
Документ
Категория
Без категории
Просмотров
10
Размер файла
696 Кб
Теги
parity, arthritis, pills, rheumatoid
1/--страниц
Пожаловаться на содержимое документа