To determine the outcome of pregnancy in women with rheumatoid arthritis (RA) in relation to disease activity and medication use during the pregnancy.
To determine the outcome of pregnancy in women with rheumatoid arthritis (RA) in relation to disease activity and medication use during the pregnancy.
In a prospective study, pregnant women with RA were evaluated before conception (when possible), during each trimester of the pregnancy, and postpartum. Clinical characteristics, disease activity, medication use, and pregnancy outcome were analyzed. To examine the independent influence of prednisone use and disease activity on birth weight, regression analyses were performed, with adjustments for gestational age of the child at delivery, the sex of the newborn, and the mother's smoking status, education level, parity, and use of an assisted reproduction technique. Kaplan-Meier curve analyses were performed to examine the association between medication use and gestational age at delivery.
Data from 152 Caucasian RA patients with singleton pregnancies were available. Both the mean ± SD birth weight (3,379 ± 564 gm) and the mean ± SD birth weight standard deviation score (SDS; +0.1 ± 1.1), which is the birth weight adjusted for the gestational age and sex of the newborn, were comparable with those in the general population. On multiple linear regression analyses of birth weight and birth weight SDS, both of which were adjusted for covariates, only disease activity was associated with lower birth weight (P = 0.025). The gestational age at delivery was significantly lower in women who were taking prednisone (38.8 versus 39.9 weeks; P = 0.001), and their delivery was more often premature (<37 weeks; P = 0.004).
Pregnancy outcome in women with well-controlled RA is comparable with that in the general population. The effect of prednisone on birth weight is mediated by a lower gestational age at delivery, whereas a higher level of disease activity independently influences birth weight negatively, suggesting an immune-mediated mechanism.
Rheumatoid arthritis (RA) is a systemic autoimmune disease that leads to a destructive arthropathy (1). Over the last several decades, the pharmacologic treatment of RA has improved considerably (2). Pregnancy is a special situation that is associated with spontaneous improvement in RA disease activity regardless of the medical treatment received (3–8).
The outcome of pregnancy, however, might be negatively influenced by RA, since unfavorable pregnancy outcome has been demonstrated in several rheumatic diseases (9–13). However, only a few studies have examined pregnancy outcomes in women with RA (5, 10, 11, 13–15). These studies were mostly retrospective, yielded conflicting findings, and the results were easily confounded by the patient selection method.
Lower birth weight, especially in newborns of women whose rheumatic disease was highly active during pregnancy, has previously been reported (11, 15). Low birth weight, particularly when categorized as small for gestational age, has not only been associated with developmental delay (16), but it has also been shown to be an important risk factor for perinatal complications, metabolic syndrome, and cardiovascular disease during childhood as well as (late) adulthood (17–19). In contrast to the reports that the mean birth weight of neonates was lower in RA patients than in healthy controls, the risk of low birth weight babies (defined as a weight of <2,500 gm) does not seem to be elevated in women with RA (14, 15). Higher rates of small for their gestational age newborns have also been reported in women with rheumatic diseases (11, 13). In addition, prednisone use may be associated with both lower birth weight and lower gestational age at delivery (20), but the only study conducted thus far lacked the power to elucidate the separate influences of disease activity and prednisone use on birth weight (15).
In our Pregnancy-Induced Amelioration of Rheumatoid Arthritis (PARA) study (8), a large prospective nationwide cohort of pregnant women with RA, we provided measurements of standardized disease activity scores determined during pregnancy, as well as well-documented data on medication use. These data enabled us to discriminate the independent effects of these variables on pregnancy outcome and to elucidate their mechanisms. In addition, new insights into pregnancy outcome and birth weight have resulted in the recent development of the birth weight standard deviation score (SDS), in which the birth weight is adjusted for gestational age at the time of delivery as well as for the sex of the newborn (21, 22).
In the present study, we sought to determine pregnancy outcomes in women with RA and to determine whether disease activity and medication use, especially prednisone use, during pregnancy were independent influences on birth weight and whether their effects on birth weight were mediated via gestational age at delivery.
The current study was embedded in the PARA study, a prospective nationwide study in The Netherlands (8). Women with RA were visited in their homes before pregnancy (if possible), during each trimester of the pregnancy, and 3 times after delivery (6, 12, and 26 weeks postpartum). The PARA study has been described in detail previously (8). Briefly, rheumatologists in The Netherlands recruited women with RA who fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) 1987 revised criteria for RA (23), had a desire to become pregnant or were already pregnant, and had a good understanding of the Dutch language. Women who were not visited prior to conception were visited from the first trimester after application to our study (either first, second, or third) and thereafter.
In the present analysis on pregnancy outcome in women with RA, only singleton pregnancies in Caucasian women who delivered a baby without congenital deformities were included. No women contributed twice; only the results of a woman's first pregnancy for which we collected data were included. These specific inclusion criteria were used because birth weight is different in non-Caucasians (24), in twin pregnancies, and in newborns with congenital deformities. Inclusion of a second or subsequent pregnancy might introduce bias, since it may represent a selection bias for women who previously had a good experience with their RA course during and after pregnancy and/or a good experience with the outcome of the pregnancy.
The PARA study was approved by the Medical Ethics Committee, Erasmus MC (University Medical Center Rotterdam, Rotterdam, The Netherlands) in June 2002. Written informed consent was obtained from all participants.
Disease activity was measured with the Disease Activity Score in 28 joints (DAS28), using 3 variables: the number of swollen joints, the number of tender joints, and the C-reactive protein (CRP) level (25). Disease activity during pregnancy can be measured most reliably with this variant of the DAS28 (26). To minimize interobserver variability, all joint counts were performed at all visits by the same research team, the members of which participated in training sessions for comparison of joint counts at least twice a year. CRP levels were measured in blood samples by Tina-Quant CRP immunologic test system (Roche Diagnostics, Mannheim, Germany).
Clinical characteristics, including maternal age at delivery, rheumatologic history, obstetric history, smoking habits, education level, and medication use, were collected by interview. The rheumatology history was ascertained by obtaining information from the patient's own rheumatologist. Data included the duration of RA, erosion status, presence of rheumatoid factor (RF), and presence of anti–cyclic citrullinated peptide (anti-CCP) antibodies. The presence of these autoantibodies was also determined in our laboratory. Smoking habits were obtained by interview. Only women who were smokers during the first trimester were defined as those who “smoked during pregnancy.” The woman's education level was also recorded; however, most data were obtained retrospectively, after the last visit. The education level was categorized according to the highest level attained, as follows: low (≤9 years of primary school, lower and intermediate secondary school, or lower vocational training), intermediate (10–14 years of higher secondary school or intermediate vocational training), and high (≥15 years of higher vocational training or university education). Medication use was monitored at all visits.
The method of conception was obtained by interview and was categorized as spontaneous or assisted reproductive endocrinology techniques. Assisted techniques included ovulation induction, intrauterine insemination, in vitro fertilization, or intracytoplasmic sperm injection.
Data on pregnancy outcome included spontaneous abortions, intrauterine fetal death, birth weight of the newborn, gestational age at delivery, sex of the child, mode of delivery, and location of delivery. Complications of pregnancy were also recorded, including the incidence of premature birth, of newborns that were small for their gestational age, and of hypertensive disorders (gestational hypertension and preeclampsia). The data were collected at various visits (by interview and analysis of medical charts). Information on the mode of delivery was categorized as spontaneous birth, instrumental vaginal birth, or cesarean section. Instrumental vaginal birth was defined as the use of forceps or a ventouse extractor during labor, and the reason was documented as “failure to progress” or “fetal distress.” Cesarean section procedures were documented as either “elective” or “emergency.”
The birth weight of the newborn was recorded (in gm), and the birth weight SDS was determined. The use of this reference standard facilitates evaluation of size deviation from the standard at birth (21). The birth weight SDS was calculated using the following sex-specific formulas. For boys, the formula was as follows:
For girls, the formula was as follows:
where bweight represents the birth weight (in kg) and ga represents the gestational age at delivery (in weeks). Small for gestational age was defined as a birth weight SDS that was at least 2 SD below the mean (less than –2 SD) (27).
Pregnancy complications were defined according to the following criteria. Spontaneous miscarriage was defined as a gestational age of <16 weeks. Prematurity was defined as delivery at <37 weeks (259 days) of the gestational age. Gestational hypertension was defined as a systolic blood pressure ≥140 mm Hg and/or diastolic blood pressure ≥90 mm Hg without proteinuria after 20 weeks of gestation in previously normotensive women, according to the criteria of the International Society for the Study of Hypertension in Pregnancy (28). Preeclampsia was defined as gestational hypertension with concurrent proteinuria (>2 gm/liter, or 2+ as measured by dipstick) (28, 29).
Descriptive statistics on pregnancy outcome were presented as numbers and percentages. Two reference groups were used to compare pregnancy outcomes in the present RA cohort with the general population in The Netherlands. The first reference group contains the data from the general obstetric population registered in The Netherlands Perinatal Registry (PRN). This registry covers almost all deliveries in The Netherlands in 2004 (∼179,000) by linking the database of midwives and general practitioners with the database of obstetricians (30). The vast majority of the PRN (82%) consists of pregnancy outcomes in Caucasian women. The second reference group contains the data from the Generation R study, a population-based prospective cohort study of 10,000 pregnancies beginning at fetal life and continuing until young adulthood that was conducted in Rotterdam, an urban area of The Netherlands, with children born between 2002 and 2006 (31). No statistical analyses were performed between the present cohort and these 2 registries because the inclusion criteria were different, and the registries therefore cannot be used as formal control groups.
To describe the association between the mother's RA disease activity during the third trimester and the baby's birth weight or birth weight SDS, the following covariates were analyzed: maternal age, parity, smoking status during pregnancy, education level, gestational age of the newborn at delivery, and the sex of newborn (the latter 2 are not applicable for analyzing birth weight SDS, since these are incorporated in the score). First, simple linear regression analyses were performed. Second, multiple linear regression analyses were performed using all possible explanatory variables that were associated with birth weight or birth weight SDS and those that had a P value <0.2 in the simple linear regression analyses. Smoking was addressed because of its known negative correlation with birth weight (32). Multiple linear regression analyses were also performed with disease activity and prednisone use.
A Kaplan-Meier curve was constructed to visualize whether gestational age at delivery was dependent on medication use. Differences in gestational age at delivery in terms of prednisone use and prednisone dosage, as well as sulfasalazine use and disease activity in those not taking prednisone, were tested for significance by the Wilcoxon-Gehan statistic. The difference between the number of women taking prednisone and the number not taking prednisone who had a gestational age at delivery of <37 weeks was examined with Fisher's exact test.
P values (2-sided) less than or equal to 0.05 were considered significant. All statistical analyses were performed using SPSS software (version 15.0 for Windows; SPSS, Chicago, IL).
From May 1, 2002 until August 31, 2007, 388 patients were recruited into the PARA study. Figure 1 presents a flow chart showing the inclusion of study patients. After applying the inclusion criteria, only 1 additional patient was excluded from the study; she was diagnosed as having breast cancer during her pregnancy. Therefore, 152 Caucasian women with RA and with singleton pregnancies were eligible for the present study.
As shown in Table 1, the patients had a mean age at delivery of 32.5 years, which is slightly older than the general population, and they had a median disease duration of 70 months (range 8–356 months) at delivery. Only 3.3% of RA patients smoked during pregnancy; this percentage is much lower than that documented in the Generation R reference group. Artificial reproduction techniques were used much more frequently in our study cohort (17.8%) than in the PRN reference group (3%).
|Present RA cohort (n = 152)||Reference groups†|
|PRN (n = 175,498)||Generation R (n = 3,659)|
|Age at delivery, mean ± SD years||32.5 ± 3.7||30.5 ± 4.9||31.2 ± 4.5|
|Duration of RA at delivery, median (range) months||70 (8–356)||NA||NA|
|Missing data||18.4 (28)||NA||0.8|
|Smoked during pregnancy||3.3 (5)||NA||17.0|
|Assisted reproduction technique||17.8 (27)||3||NA|
|Ovulation induction||6.0 (9)||NA||NA|
|Ovulation induction plus IUI||3.3 (5)||NA||NA|
|In vitro fertilization||5.2 (8)||NA||NA|
|Intracytoplasmic sperm injection||3.3 (5)||NA||NA|
|During the third trimester (n = 148)|
|<3.2 (low)||47 (69)||NA||NA|
|3.2–5.1 (intermediate)||45 (66)||NA||NA|
|>5.1 (high)||9 (13)||NA||NA|
|Median (IQR) scores over time|
|Before conception (n = 68)||3.9 (1.4)||NA||NA|
|First trimester (n = 125)||3.8 (1.8)||NA||NA|
|Second trimester (n = 140)||3.6 (1.6)||NA||NA|
|Third trimester (n = 148)||3.3 (1.7)||NA||NA|
|Before conception (n = 68)||32 (22)||NA||NA|
|First trimester (n = 127)||31 (39)||NA||NA|
|Second trimester (n = 140)||29 (41)||NA||NA|
|Third trimester (n = 150)||27 (40)||NA||NA|
|Before conception (n = 68)||40 (27)||NA||NA|
|First trimester (n = 127)||36 (46)||NA||NA|
|Second trimester (n = 140)||36 (50)||NA||NA|
|Third trimester (n = 150)||35 (52)||NA||NA|
Because some patients were included during their second or third trimester of pregnancy, the DAS28-CRP scores are lacking in 27 patients during the first trimester, and in 12 patients during the second trimester. The median disease activity throughout pregnancy (DAS28-CRP) decreased from 3.8 during the first trimester to 3.3 during the third trimester, indicating that disease activity was well-controlled. The DAS28-CRP values, as measured during the first and second trimesters and as measured during the first and third trimesters, were strongly correlated (r = 0.715 and r = 0.647, respectively, by Spearman's correlation coefficient).
Medication use during pregnancy was restricted to prednisone (∼50 patients), hydroxychloroquine (2 patients), and sulfasalazine (∼40 patients). All women in the present study stopped taking methotrexate at least 3 months before conception (77 women had ever used methotrexate before pregnancy), and none of the data presented here concern women who had used biologic agents during pregnancy (19 women had ever used biologic agents before pregnancy). The mean daily dose of prednisone during pregnancy was similar during each trimester: 7.7 mg, 8.0 mg, and 7.7 mg during the first, second, and third trimester, respectively. Only 8 women took more than 10 mg of prednisone per day, and only 10 patients changed their prednisone usage status and/or their dosage throughout their pregnancy. Data on prednisone use during the first and second trimesters and during the first and third trimesters were strongly correlated (r = 0.880 and r = 0.969, respectively).
Pregnancy outcome is shown in Table 2. The mean birth weight of neonates in our cohort was within the normal range (21). The mean ± SD birth weight SDS was +0.1 ± 1.1, which is comparable with that in the standard population (21).
|Present RA cohort (n = 152)||Reference groups†|
|PRN (n = 179,457)||Generation R (n = 3,659)|
|Birth weight, mean ± SD gm||3,379 ± 564||3,363||3,485 ± 555|
|Gestational age at delivery, mean ± SD days||276 ± 11.2||275 ± 15.7||278 ± 11.9|
|Birth weight SDS, mean ± SD||+0.1 ± 1.1||–||–|
|Sex of child|
|Mode of delivery (n = 146)|
|Spontaneous vaginal delivery||65 (96)||75.3||–|
|Instrumental vaginal delivery||17 (25)||10.5||–|
|Failure of labor to progress||56 (14)||–||–|
|Fetal distress||36 (9)||–||–|
|Combined reason||8 (2)||–||–|
|Cesarean section||17 (25)||14.2||–|
|Location of delivery|
|Prematurity (<37 weeks)||8.6 (13)||6.2||–|
|Intrauterine fetal death||0 (0)||–||–|
|Birth weight SDS less than −2 (SGA)||3.3 (5)||–||–|
|Hypertensive disorders (verified)||7.2 (11)||8.5||–|
|Gestational hypertension||6.6 (10)||7.9||–|
|Preeclampsia with admission to hospital||0.7 (1)||0.4||–|
Twenty-five patients required a cesarean section for delivery of their child. In about half of these patients, an elective cesarean section was required because of problems with the fetus (n = 10; breech presentation), the mother (n = 1), or both (n = 1). The remaining patients required an emergency cesarean section because of failure of labor to progress (n = 7), fetal distress (n = 4), or combined reasons (n = 2). Another 25 patients (17%) required an instrumental vaginal delivery. This occurred more frequently in our RA cohort than in the PRN reference group (10.5%).
When the RA patients were categorized according to their disease activity during the third trimester (“low” group versus “intermediate or high” group), cesarean sections were performed significantly more often in the group with intermediate or high disease activity (22% versus 10% in the low group; P = 0.04 by chi-square test), while instrumental vaginal births were performed at equal frequency in both disease activity groups (25% versus 19%; P = 0.43 by chi-square test). In this study, no cesarean section was performed because of hip involvement or hip prosthesis (only 1 patient).
The incidence of premature births was 8.6%, and the incidence of small for gestational age newborns was 3.3%. The incidences of hypertensive disorders in women with RA were 6.6% for gestational hypertension (n = 10) and 0.7% for preeclampsia with admission to the hospital (n = 1). Although the number of patients in the present study cohort was too small to properly examine the incidence of these fairly rare complications, the incidence was relatively comparable with that in the general population.
Table 3 shows the association between birth weight and disease activity or prednisone use in the RA cohort. Disease activity during the third trimester (as determined by the DAS28-CRP) was significantly negatively associated with the actual birth weight (in gm) of the newborn, as well as with the birth weight SDS (P < 0.009 and P = 0.025, respectively). Similar results were noted for disease activity during the first and second trimesters as well as when subgroup analyses were performed only in women who had not taken prednisone during the pregnancy. Furthermore, prednisone use during pregnancy was significantly associated only with the actual birth weight (in gm), and not with the birth weight SDS (P = 0.002 and P = 0.256, respectively). This may indicate that prednisone indirectly influences birth weight via gestational age at delivery.
|Simple linear regression||Actual birth weight (n = 152)||Birth weight SDS (n = 152)|
|Standardized coefficient (95% CI)||P||Standardized coefficient (95% CI)||P|
|DAS28-CRP third trimester||−108.9 (−190.6, −27.2)||0.009||−0.18 (−0.340, −0.023)||0.025|
|Prednisone use (yes)||−296.5 (−479.3, −113.7)||0.002||−0.21 (−0.574, 0.154)||0.256|
|Gestational age at delivery (days)||187.7 (142.6, 232.7)||<0.001||–||–|
|Parity (multiparous)||383.1 (210.44, 555.8)||<0.001||0.77 (0.438, 1.104)||<0.001|
|Age of mother (years)||18.5 (−6.0, 43.1)||0.137||0.049 (0.002, 0.096)||0.041|
|Smoked during pregnancy (yes)||−15.0 (−522.5, 494.6)||0.954||0.41 (−0.570, 1.389)||0.410|
|Sex of child (male)||99.3 (−83.8, 282.3)||0.286||–||–|
|Education level (low versus high)†||−171.0 (−512.5, 169.3)||0.321||−0.39 (−1.073, 0.292)||0.260|
|Conception type (assisted reproduction)||−5.8 (−243.1, 234.6)||0.831||0.05 (−0.409, 0.509)||0.831|
To detect potential confounders for the association between birth weight or birth weight SDS and disease activity or prednisone use, simple linear regression analyses were performed with independent variables (Table 3). In simple linear regression analyses of birth weight, parity and gestational age at delivery were the most predictive. Multiparous women gave birth to heavier newborns, and smoking was not associated with lower birth weight in this cohort. In simple linear regression analyses of birth weight SDS, only parity and maternal age were both statistically significantly associated with birth weight SDS. All other potential confounders were not significantly associated with birth weight SDS.
In multivariate analyses, higher disease activity during the third trimester was significantly associated with lower actual birth weight and with the birth weight SDS after adjustment for all covariates (Table 4).
|No.||Standardized coefficient (95% CI)||P||No.||Standardized coefficient (95% CI)||P|
|Birth weight||148†||−111.9 (−193.6, −30.2)||0.008||117‡||−82.6 (−154.5, −10.6)||0.025|
|141§||−74.9 (−141.7, −8.0)||0.029|
|Birth weight SDS||148||−0.181 (−0.340, −0.023)||0.025||117¶||−0.193 (−0.362, −0.037)||0.025|
|141#||−0.174 (−0.331, −0.016)||0.031|
Prednisone use was associated with lower actual birth weight (in gm), but not with the birth weight SDS, on simple linear regression analyses (Table 3). Multiple linear regression analyses of prednisone use during pregnancy showed a significant association with shorter gestational age (P < 0.001) after correction for potential confounders, such as the DAS28-CRP during the third trimester, the mode of delivery (spontaneous versus cesarean section), parity (1 versus >1), and the age of the mother. This shows that prednisone is independently associated with lower birth weight via shorter gestational age.
A statistically significant difference in gestational age of infants born to RA patients who did and did not take prednisone during pregnancy was confirmed through Kaplan-Meier curve analyses (Figure 2A). However, no effect of the prednisone dosage was observed (Figure 2B). Women taking prednisone during pregnancy delivered 1 week earlier, on average, than did women who were not taking prednisone during pregnancy (38.8 weeks [95% confidence interval 38.2, 39.3] versus 39.9 weeks [95% confidence interval 39.6, 40.1]). This could not be explained by differences in the frequency of induction of labor (14% in both groups) or in the frequency of elective cesarean sections. Women taking prednisone more often delivered at <37 weeks of gestational age, as compared with women who were not taking prednisone (P = 0.004 by Fisher's exact test).
No influence of sulfasalazine treatment on gestational age was observed (Figure 2C). The additional analysis of only women who were not taking prednisone during pregnancy showed that gestational age was not influenced by the level of disease activity during the third trimester (Figure 2D).
In the present prospective study of pregnancy and RA, we found that women with well-controlled RA have a pregnancy outcome that is comparable with that in the general obstetric population. In addition, we demonstrated that disease activity and prednisone use during pregnancy were both negatively associated with birth weight. Only higher disease activity was associated with a lower birth weight SDS, whereas the effect of prednisone on birth weight was mediated by a shortening of the gestational age at delivery. This resulted in significantly more women who were taking prednisone delivering at <37 weeks of gestation.
This study is the first to use a birth weight standard deviation score (21, 22) to evaluate the birth weight of newborns of RA patients. Because we used this analytical method, we can certify that in the present study, the weight of the newborns was, on average, appropriate for their gestational age at birth, since the mean ± SD birth weight SDS was +0.1 ± 1.1. Furthermore, use of the birth weight SDS enabled us to determine that the effect of disease activity on birth weight was a direct effect (both the birth weight SDS and the actual birth weight were affected), whereas the effect of prednisone was mediated through the gestational age (only the actual birth weight, not the birth weight SDS, was affected).
Almost all studies of pregnant women with RA that have been conducted so far have reported lower birth weights, yet still within the normal range, of newborns of RA patients. However, almost all of these studies reported the birth weight without correction for gestational age and without prospective measurements of disease activity during the pregnancy (11, 13, 14, 33). The discrepancy between our results and previous reports can be explained by several factors. First, RA is treated more aggressively nowadays, which may have resulted in low levels of disease activity in our cohort. Second, in our RA cohort, only a few patients smoked and a higher percentage of patients had a higher education level than in the general population. Both of these factors are associated with higher birth weight babies. The low percentage of women who smoked during pregnancy and the higher education level suggest a healthy cohort effect in this study. On the other hand, our RA patient cohort was older and had a higher percentage of patients who were nulliparous, both of which are associated with a lower birth weight.
RA disease activity during pregnancy was associated with a lower actual birth weight and a lower birth weight SDS. We demonstrated that disease activity during the third trimester was a negative influence on birth weight and was independent of many covariates, such as prednisone use, parity, smoking, sex of the child, duration of pregnancy, maternal age, education level, and assisted reproduction. In multiple linear regression analyses, we focused on the role of disease activity and prednisone use during the third trimester, since at that time point, our cohort included the most patients. Because levels of disease activity during the first, second, and third trimesters were highly correlated, our data are not suitable for determining whether a certain critical period exists for the effect of disease activity on birth weight.
As a general rule of thumb, only a difference of 0.5 in the birth weight SDS is considered to be of clinical relevance. Since the estimated mean decrease in the SDS was 0.18 for a 1.0 increase in the DAS28-CRP, our findings may be clinically relevant only for women with high levels of disease activity during pregnancy. Although this small difference in birth weight is unlikely to result in more perinatal complications, a small decrease in birth weight, even if the birth weight is still within the normal range, has been associated with future developmental delay (16, 34) and is a risk factor for the metabolic syndrome and cardiovascular disease in children and adolescents (17–19). This will be the subject of further research.
This study is also the first to show that among the medications taken by RA patients during pregnancy, only prednisone had an indirect negative effect on birth weight via a shortening of the gestational age at delivery. This influence of prednisone has already been shown in a study of women taking prednisone for a wide spectrum of indications (20), and the present study confirms this effect solely in RA patients. This is of clinical importance for obstetricians, since women with RA who were taking prednisone more frequently delivered at <37 weeks of gestational age, as compared with women who were not taking prednisone, regardless of induction of labor or elective cesarean sections.
Another 4 of our observations in the present cohort need some further discussion. The first is the relatively high number of pregnancies that were accomplished with assisted reproduction techniques. This may be explained in several ways: the time to conceive in women with RA may be longer, as previously suggested (35); women with RA may be more readily referred for artificial reproduction techniques; or a referral bias may be operant, since rheumatologists will be more aware that a patient wants to become pregnant because it takes longer to conceive, and patients with difficulties in conceiving may be more willing to participate in a study.
The second observation for discussion is the high rate of instrumental vaginal delivery as compared with that in the general population. However, this should be interpreted against the higher number of nulliparous women, since in general, nulliparous women need instrumental assistance with vaginal delivery in ∼20% of cases. We could not explain this by higher disease activity during the third trimester.
The third observation concerns cesarean sections. It has previously been documented that cesarean sections are more prevalent among RA patients than among healthy controls (11, 13, 14). In the present study, however, the number of cesarean sections that were conducted and the subclassification of those procedures as being elective or emergency seem to be comparable with the data in the general population. The higher number of cesarean sections in previous studies may be explained by the fact that the mothers examined in previous studies may have had higher levels of disease activity. As we showed in our cohort, cesarean sections were significantly more frequent in women with high levels of disease activity.
The fourth observation we would like to discuss concerns the increased risk of premature labor that has been reported in women with RA (10, 12, 14); however, in the present cohort, the risk in our RA patients was comparable with that in women in the PRN reference group. In the present cohort, only 5 newborns were small for their gestational age (3.3%), which is consistent with the frequency in the general population. Higher risks of hypertensive disorders (gestational hypertension and preeclampsia) in pregnant RA patients have been described (11, 13), while in other studies, this was not the case (14, 15, 33). The incidence of hypertensive disorders during pregnancy that could be obtained from the PRN reference group (all hypertensive disorders combined) and from the literature was 6–10% (28, 29, 36, 37). The incidence of hypertensive disorders among patients in the present study seemed to be comparable with that in the reference groups; however, this should be interpreted with caution because of the low number of patients in our cohort.
The present study has some limitations. We showed beforehand that patient-reported pregnancy characteristics are generally of good validity and high reliability as compared with the obstetric records (data not shown). Since the study was not designed as a case–control study, comparison with healthy controls can only be made indirectly, and the results should therefore be interpreted with caution. An important limitation of the present study is the lack of valid control data for all pregnancy outcomes. Finally, a healthy cohort effect may have influenced the results and may have even underestimated the results we found. In our opinion, the results of the present study can be generalized to Caucasian RA patients with singleton pregnancies who are seen in rheumatology clinics and are treated with the currently available therapies, resulting in well-controlled disease. The opinion of most rheumatologists that disease activity should be low before the patient becomes pregnant in order to have a better pregnancy outcome seems to hold true.
The mechanism of how high levels of disease activity lead to lower birth weight has yet to be elucidated; however, some pathophysiologic hypotheses might explain this. First, vasculopathy resulting from endothelial dysfunction may contribute to maldevelopment of the placenta. Endothelial dysfunction is a common complication of active RA and is thought to be the initial stage of atherosclerosis. Maldevelopment of the placenta is associated with an unfavorable pregnancy outcome (e.g., lower birth weight, hypertension) (38–40). Second, proper fetal development requires lower levels of fetal cortisol than maternal cortisol. In the placenta, the enzyme 11β-hydroxysteroid dehydrogenase type 2 (11β-HSD2) inactivates maternal cortisol. High cortisol levels in the mother (chronic stress) and low 11β-HSD2 levels in the placenta (famine) are both associated with decreased birth weight (41). Additionally, the presence of high levels of proinflammatory cytokines, such as tumor necrosis factor α, interleukin-1β, and interleukin-6, all of which are associated with active RA, can down-regulate placental levels of 11β-HSD2 (42), and may therefore result in lower birth weight. An examination of this possibility will be the subject of future studies.
In conclusion, this is the first prospective study describing pregnancy outcome and extensively linking disease activity and medication use throughout pregnancy to birth weight in babies born to women with RA. Although the incidence of pregnancy complications is comparable with that in the Dutch general obstetric population, among patients with high levels of disease activity, more cesarean sections were performed. Furthermore, we showed that newborns of women with higher levels of disease activity who were also taking prednisone had a lower birth weight. The effect of prednisone on birth weight is mediated indirectly by a shortened gestational age at delivery, whereas disease activity directly influences birth weight. For the obstetrician, our findings might imply that he or she should be aware of a higher incidence of cesarean sections in RA patients with high levels of disease activity, as well as an increased risk of preterm delivery in RA patients taking prednisone. For the rheumatologist, our findings might indicate that he or she should strive for low levels of disease activity, both before and during pregnancy, for a better pregnancy outcome for their RA patients.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. De Man had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. De Man, Hazes, Willemsen, de Groot, Steegers, Dohlain.
Acquisition of data. De Man, Hazes, Dolhain.
Analysis and interpretation of data. De Man, Hazes, van der Heide, Willemsen, de Groot, Steegers, Dolhain.
We gratefully acknowledge all of the study patients and the Dutch rheumatologists and obstetricians for their voluntary contribution to the PARA study. We would also like to thank Fleur van de Geijn, Esther Gasthuis, Tineke Krommenhoek, Eske te Boveldt, and the other research assistants of the PARA study for data collection.