Abstract. Nørgaard M, Larsson H, Pedersen L, Granath F, Askling J, Kieler H, Ekbom A, Sørensen HT, Stephansson O (Aarhus University Hospital, Denmark, Karolinska Institutet; Karolinska University Hospital, Solna; and Karolinska Institutet, Solna; Stockholm, Sweden). Rheumatoid arthritis and birth outcomes: a Danish and Swedish nationwide prevalence study. J Intern Med 2010; 268: 329–337.
Objectives. To examine the prevalence of preterm birth, infants with low Apgar score, small for gestational age (SGA) birth, stillbirth and congenital abnormalities in women with rheumatoid arthritis (RA) compared with women without RA.
Design. Prevalence study.
Setting. Combined Sweden and Denmark nationwide from 1994 to 2006.
Subjects. We included 871 579 women with a first-time singleton birth identified through population-based healthcare databases.
Main outcome measures. We compared the prevalence of preterm birth, low Apgar score (<7 at 5 min), SGA birth, stillbirth and congenital abnormalities amongst women with RA compared with women without RA using prevalence odds ratio (OR) with 95% confidence interval (95% CI), whilst controlling for maternal age, smoking, parental cohabitation and year. We stratified analyses by period of birth (1994–1997, 1998–2001 and 2002–2006).
Results. Amongst 1199 women with RA, 7.8% gave birth between 32 and 36 gestational weeks (adjusted OR, 1.44; 95% CI, 1.14–1.82), 1.4% gave birth before gestational week 32 (adjusted OR, 1.55; 95% CI, 0.97–2.47), 1.6% had an infant with a low Apgar score (OR, 0.99; 95% CI, 0.95–1.65), 5.9% had an SGA birth (adjusted OR, 1.56; 95% CI, 1.2–2.01), 0.9% experienced stillbirth (adjusted OR, 2.07; 95% CI, 0.98–4.35) and 4.3% gave birth to an infant with congenital abnormalities (adjusted OR,1.32; 95% CI, 0.98–1.79). The OR for congenital abnormalities decreased from 2.57 (95% CI, 1.59–4.16) in 1994–1997 to 1.00 (95% CI, 0.64–1.56) in 2002–2006.
Conclusions. Women with RA had a high prevalence of most adverse birth outcomes. This could be due to inflammatory activity, medical treatment or other factors not controlled for.
Rheumatoid arthritis (RA) is a chronic systemic autoimmune inflammatory disease that predominantly affects women . Although the peak age of onset is in the fourth or fifth decade of life, RA is common amongst younger women. In the UK, the estimated prevalence is 0.12% in women of childbearing age. Women with RA probably experience a longer time to conception than healthy women , yet they seem able to achieve pregnancy at similar rates .
Rheumatoid arthritis may adversely influence pregnancy through several mechanisms. The inflammatory process of RA  may affect the placenta, leading to adverse birth outcomes such as low birth weight . RA disease activity may thus necessitate drug treatment to ensure a successful pregnancy outcome. At the same time, exposure to anti-inflammatory, immunosuppressive or biological drugs in utero may adversely affect pregnancy outcome. In recent years, RA treatment regimens have included earlier and more aggressive use of disease-modifying drugs . In addition, lifestyle factors such as smoking, itself a risk factor for RA , can lead to preterm birth and small for gestational age (SGA) infants .
There have only been a few studies on birth outcomes amongst women with RA [9, 10], which have suggested a 1.5- to 2.2-fold increased risk of premature birth, SGA birth and low birth weight amongst RA patients compared with women in the general population. However, these studies included only a total of 604 women with RA. Yet, similar findings have been reported in studies amongst women with different types of inflammatory polyarthritis including RA [11–13].
We conducted a combined Swedish and Danish nationwide study of primiparous women who gave birth between 1994 and 2006, to examine the prevalence of preterm birth, SGA birth, infants with a low Apgar score at 5 min, stillbirth and congenital abnormalities amongst women with RA compared with women without RA.
Materials and methods
We conducted this prevalence study in Sweden and Denmark with a combined population of more than 14 million . Residents of both countries are provided with free tax-supported medical care. We included all female residents with a first-time singleton live birth or stillbirth between 1994 and 2006. The women were identified through medical birth registries.
Since 1973, virtually all births have been registered in the medical birth registry in each country . Maternal data, including self-reported smoking habits and cohabitation with the baby’s father, are collected by midwives during the first antenatal visit. In Sweden, height and weight are measured at the first antenatal visit, allowing calculation of body mass index (BMI), and since 2004 this information has also been reported in Denmark. We retrieved information from the birth registries on maternal age, gestational age, birthweight, parity, stillbirth, mode of delivery and civil registration number of the mother and child (a unique identifier, which encodes sex and date of birth, and is assigned to all residents in Denmark  and Sweden ).
Using civil registration numbers, we linked birth registry data with national patient registries to identify those with a diagnosis of RA registered before childbirth. The Swedish Patient Registry (established in parts of Sweden in 1964; nationwide since 1987) contains information on dates of hospital admission and discharge, and diagnoses on discharge classified according to the International Classification of Diseases (ICD) revisions 7–10. Since 2001, this registry has also included information about outpatient hospital visits. In Denmark, the National Registry of Patients (established in 1977) includes dates of hospital admission and discharge, surgical procedures and up to 20 diagnoses coded by physicians at discharge according to ICD-8 until the end of 1993 and ICD-10 thereafter . ICD-9 has never been used in Denmark. Diagnoses from hospital outpatient visits have been included in the Danish registry since 1995.
We obtained information regarding all hospital admissions and outpatient visits associated with an RA diagnosis prior to childbirth from 1987 in Sweden and from 1977 in Denmark (see Appendix for ICD codes). As we lacked specific biochemical information on disease severity, we used RA-related orthopaedic surgery as a surrogate measure. We thus retrieved information on all RA-related orthopaedic operations recorded in the patient registries (i.e. RA-related surgical intervention codes indicating an orthopaedic upper or lower limb surgery, excluding fracture- or infection-related codes) [19, 20].
Birth outcome data and potential confounders
From the birth registries we retrieved information on Apgar score at 5 min, stillbirth (defined as delivery of a dead foetus at >27 completed weeks of pregnancy) and birthweight. We defined moderately preterm birth as a live birth between 32 and 36 completed weeks of gestation and very preterm birth as a live birth before 32 completed weeks of gestation. We were able to further categorize Swedish births into spontaneous and induced preterm births. Low Apgar score was defined as a score at 5 min <7. SGA was defined as a birthweight more than two standard deviations below the mean for neonates of similar gestational age, based on the Swedish reference curve of estimated foetal growth .
For liveborn children, diagnoses of congenital abnormalities during the first year of life were obtained from the birth and hospital registries (see Appendix for codes). Diagnoses of congenital dislocation of the hip and undescended testis were excluded because of their poor expected validity . Also excluded were diagnoses of pre-auricular appendage or tag, tongue tie, congenital non-neoplastic naevus and chromosomal abnormalities.
We also retrieved information about caesarean delivery and diagnoses of pre-eclampsia (see Appendix for codes). Information on pregestational diabetes was retrieved from the Swedish Medical Birth Registry and from the Danish National Patient Registry. For descriptive purposes, we also retrieved information on the use of assisted reproductive techniques from the Swedish Medical Birth Registry (information included since 1995).
We used logistic regression analysis to control for confounding and computed crude and adjusted prevalence odds ratio (OR) as an estimate of relative risk, with associated 95% confidence interval (CI), for the following variables (overall and stratified by country): moderate and very preterm birth, low Apgar score, SGA, stillbirth and congenital abnormalities overall and by specific type. We adjusted for maternal age (≤19, 20–24, 25–29, 30–34 and ≥35 years), smoking (yes/no), parental cohabitation (yes/no) and year of delivery (1994–1997, 1998–2001 and 2002–2006). For Swedish births, we further conducted an analysis in which we categorized preterm births into spontaneous and induced preterm birth (before 37 gestational weeks).
To look for temporal trends, we stratified the analyses by three calendar periods (1994–1997, 1998–2001, 2002–2006). In a second analysis, we included pregestational diabetes (yes/no) as a covariate. In an analysis restricted to Swedish data for 1997–2006 and Danish data for 2004–2006, we further adjusted for BMI.
We conducted the analyses using different categorizations of RA as indicators of disease severity. First, we used rheumatoid factor status (seropositive or seronegative). Second, we categorized RA patients into those with at least one hospital admission for RA and those treated as outpatients only. Finally, we categorized RA according to occurrence of RA-related orthopaedic surgery within 5 years before delivery (yes/no). In these analyses we adjusted for maternal age (≤19, 20–24, 25–29, 30–34 and ≥35 years), smoking (yes/no), parental cohabitation (yes/no) and year of delivery (1994–1997, 1998–2001 and 2002–2006).
All analyses were conducted using sas software, version 9.2 (SAS Institute Inc., Cary, NC, USA). The study was approved by the Danish Data Protection Agency (record no. 2005-41-6044) and by a Regional Ethical Review Board in Stockholm, Sweden.
A total of 871 579 first-time singleton births were included in the analysis. Of these, 1199 (0.14%) children were born to mothers with RA [696 births (0.13%) out of 523 941 total births in Sweden and 503 births (0.15%) out of 346 439 total births in Denmark]. Birth characteristics amongst women with and without RA are shown in Table 1.
Table 1. Descriptive data of births amongst 1199 women with rheumatoid arthritis (RA) and 870 952 women without RA
|Country of residence|
| Sweden||696 (58.0)||523 941 (60.2)|
| Denmark||503 (42.0)||346 439 (39.8)|
|Calendar period of birth|
| 1994–1997||269 (22.4)||273 250 (31.4)|
| 1998–2001||267 (22.3)||250 393 (28.8)|
| 2002–2006||663 (55.3)||346 737 (39.8)|
|Maternal age at delivery|
| ≤19 years||19 (1.6)||33 716 (3.9)|
| 20–24 years||162 (13.5)||193 337 (22.2)|
| 25–29 years||444 (37.0)||352 456 (40.5)|
| 30–34 years||411 (34.3)||219 103 (25.2)|
| ≥35 years||163 (13.6)||71 174 (8.2)|
|Cigarette smoking||165 (13.8)||129 251 (14.8)|
|Parents living together||1059 (88.3)||766 429 (88.1)|
|Caesarean section||312 (26.0)||143 897 (16.5)|
|Pre-eclampsia||60 (5.0)||29 616 (3.4)|
|Maternal diabetes||11 (0.9)||2899 (0.3)|
Women with RA were slightly older than women without RA at the time of first childbirth. Median age was 31 years (interquartile range, 28–34 years) for RA patients and 29 years (interquartile range, 26–33 years) for women without RA. Women with RA were more likely to have a caesarean section than unaffected women (26.0% vs. 16.5%) and they had a higher prevalence of pre-eclampsia (5.0% vs. 3.4%).
Amongst 660 Swedish women with RA who gave birth between 1995 and 2006, assisted reproductive techniques had been used in 37 (5.6%) compared with 11 571 (2.4%) in the 480 206 Swedish women without RA who gave birth during the same period.
With the exception of having a child with a low Apgar score, women with RA had an increased prevalence of all adverse birth outcomes. Amongst the 1199 women with RA, the prevalence of moderately preterm birth was 7.8% and the prevalence of very preterm birth was 1.4%, compared with 5.3% and 0.9%, respectively, amongst women without RA (Table 2). The corresponding crude OR values were 1.53 (95% CI, 1.23–1.90) for moderately preterm birth and 1.59 (95% CI, 1.04–2.43) for very preterm birth. Adjustment did not substantially change these estimates (Table 2); inclusion of diabetes as a covariate also had no effect (data not shown). When restricting analysis to the Swedish data, we found the relative risk of induced preterm birth (adjusted OR 2.27; 95% CI, 1.50–3.46) was higher than that of spontaneous preterm birth (adjusted OR, 1.31; 95% CI, 0.92–1.86).
Table 2. Prevalence and crude and adjusted prevalence odds ratios (ORs) for adverse birth outcomes amongst 1199 women with rheumatoid arthritis (RA) and 870 380 women without RA
| Gestational week 32–36||94 (7.8)||45 754 (5.3)||1.53 (1.23–1.90)||1.44 (1.14–1.82)|
| Gestational week <32||17 (1.4)||8090 (0.9)||1.59 (1.04–2.43)||1.55 (0.97–2.47)|
|Apgar score at 5 min below 7||19 (1.6)||12 246 (1.4)||1.13 (0.72–1.78)||0.99 (0.59–1.65)|
|Small for gestational age||71 (5.9)||32 939 (3.8)||1.60 (1.25–2.03)||1.56 (1.20–2.01)|
|Stillbirth||11 (0.9)||3862 (0.4)||2.14 (1.18–3.88)||2.07 (0.98–4.35)|
|Congenital abnormalities||51 (4.3)||26 519 (3.0)||1.42 (1.07–1.87)||1.32 (0.98–1.79)|
The prevalence of SGA was 5.9% amongst infants born to mothers with RA (adjusted OR, 1.56; 95% CI, 1.20–2.01). Only 11 (0.9%) women with RA experienced a stillbirth, but this was higher than the 0.4% prevalence of stillbirths amongst women without RA (adjusted OR, 2.05; 95% CI, 0.98–4.32). A total of 51 (4.3%) women with RA gave birth to a child with a congenital abnormality (adjusted OR, 1.32; 95% CI, 0.98–1.79). We found no major differences in OR for these outcomes in analyses stratified by country. Including BMI as a covariate did not substantially change the estimates (data not shown).
Stratification by calendar period of delivery (Table 3) showed that the increased prevalence of abnormalities was most pronounced during the 1994–1997 period (OR, 2.57; 95% CI, 1.59–4.16). Of note, there was no increased risk of congenital abnormality overall in the 2002–2006 period (adjusted OR, 1.0; 95% CI, 0.64–1.56).
Table 3. Adjusted prevalence odds ratios (ORs) for adverse birth outcomes amongst 1199 women with rheumatoid arthritis (RA) and 870 952 women without RA, stratified by period of delivery
|Moderately preterm birth||1.87 (1.19–2.92)||1.08 (0.60–1.93)||1.42 (1.04–1.94)|
|Very preterm birth||1.12 (0.36–3.50)||2.15 (0.89–5.22)||1.51 (0.81–2.83)|
|Small for gestational age||1.96 (1.24–3.11)||2.13 (1.33–3.43)||1.12 (0.73–1.69)|
|Congenital abnormalities||2.57 (1.59–4.16)||0.95 (0.42–2.15)||1.00 (0.64–1.56)|
Women with RA gave birth to a total of 51 children with congenital abnormalities. Thus, the absolute numbers of specific birth defects were low (ranging from 1 to 11 cases). The highest ORs for women with RA compared to those without were for limb-reduction defects (unadjusted OR, 3.68; 95% CI, 0.92–14.78) and cleft lip/palate (unadjusted OR, 3.42; 95% CI, 1.42–8.25); the increased rates were seen in both Swedish and Danish populations. Yet, these estimates were based on two infants with limb reduction and five with cleft lip/palate born to mothers with RA. We found no substantially increased prevalence of cardiac or urinary tract defects amongst children born to mothers with RA.
The association of different categorizations of RA with adverse birth outcomes is shown in Table 4. In contrast to women with seropositive RA, ORs were not substantially increased in those with seronegative RA. The prevalence of seropositive RA differed substantially between Sweden and Denmark (62% and 39% of maternal RA cases in each country, respectively). In addition, analyses restricted to Swedish data showed increased risks only amongst women with seropositive RA for moderately preterm birth (OR, 1.79; 95% CI, 1.27–1.81), very preterm birth (OR, 2.36; 95% CI, 1.33–4.19) and congenital abnormalities (OR, 1.43; 95% CI, 0.93–2.21). Women who underwent orthopaedic surgery for RA within 5 years of childbirth had higher ORs for preterm birth and SGA than those who did not. Although imprecise, our estimates suggest a lower risk of preterm birth amongst women who were treated as outpatients only compared with women with at least one hospitalization for RA, even after adjustment for year of delivery (Table 4).
Table 4. Adjusted prevalence odds ratios (ORs) and corresponding 95% confidence intervals for birth outcomes in women with RA according to presence of rheumatoid factor, surgery within 5 years before giving birth and previous hospital admissions for RA, compared with women without RA
|Moderately preterm birth||1.65 (1.23–2.23)||1.19 (0.82–1.74)||2.27 (1.35–3.81)||1.31 (1.01–1.71)||1.64 (1.25–2.16)||1.07 (0.68–1.68)|
|Very preterm birth||2.04 (1.18–3.53)||0.96 (0.40–2.31)||2.40 (0.89–6.50)||1.41 (0.83–2.38)||1.83 (1.08–3.11)||1.01 (0.38–2.70)|
|Small for gestational age||1.92 (1.39–2.64)||1.16 (0.75–1.78)||3.68 (2.27–5.98)||1.24 (0.91–1.68)||1.61 (1.18–2.20)||1.45 (0.91–2.30)|
|Congenital abnormalities||1.66 (1.15–2.40)||0.92 (0.54–1.57)||1.06 (0.43–2.58)||1.37 (0.99–1.88)||1.43 (1.00–2.05)||1.12 (0.65–1.95)|
In this large population-based study including more than 1000 first-time singleton births amongst women with RA, we found that women with this condition had a 40–50% higher risk of SGA and preterm birth and a nearly two-fold increased risk of stillbirth compared with women without RA. We also found that RA was associated with a 30% higher prevalence of congenital abnormalities, although the elevated risk decreased over the study period. However, congenital malformations occurred in <5% of children of women with RA.
This study supports the few previous studies reporting an increased risk of adverse birth outcomes amongst women with RA [9, 10] or inflammatory polyarthritis including RA [11–13]. Our estimates are similar to those of Chakravarty et al. who found a higher age-adjusted risk of intrauterine growth restriction (OR, 2.2; 95% CI, 1.2–4.1) amongst 360 women with RA experiencing an obstetric hospitalization in 2002, compared with women in the general US population . Reed et al. similarly found an association between RA and adverse birth outcomes in an American cohort of 243 women with RA and 2559 healthy women who gave birth between 1987 and 2001. Their estimated relative risks ranged between 1.51 and 1.78 for SGA, low birth weight and prematurity .
Although we found an increased prevalence of congenital abnormalities during the full study period, we found no increased prevalence between 2002 and 2006. Reed et al. similarly found no increased risk of congenital abnormalities amongst women with RA who gave birth in 2002 (OR, 1.08; 95% CI, 0.4–2.66). Since 1994, improvements in ultrasound methods have increased prenatal recognition of foetal abnormalities. Although we lacked data regarding induced abortions, it is possible that their increased use may explain at least some of the observed decrease in the risk of congenital abnormalities.
The biochemical events involved in parturition, both at term and preterm, include the activation of pro-inflammatory mediators . The resulting local and systemic inflammation stimulates myometrial contractility and facilitates cervical ripening. Despite the role of inflammation in RA, our findings from the Swedish data suggest that the relative risk of spontaneous preterm delivery was not substantially higher amongst women with RA compared with women without RA. Chronic villitis of the placenta is a common finding in preterm delivery . However, there are few, if any, data on the prevalence of chronic villitis in pregnant women with RA. Although our data need to be interpreted with caution, the finding of a relatively higher prevalence of preterm delivery in women with RA who had orthopaedic surgery at least once within 5 years before giving birth compared with women with RA who had not had surgery suggests that the association could be related to RA severity or the consequences of inflammation.
The mechanisms leading to growth restriction are poorly understood, but seem to result from a complex interplay amongst the mother, the foetus and the environment . A Dutch study of 152 pregnant RA patients found that greater disease activity was associated with lower birth weight . The influence of corticosteroids on intrauterine growth is, however, controversial. According to the Dutch study, use of prednisone was associated with shorter gestational age even after adjustment for disease activity, and an apparent association between use of prednisone and lower birthweight was mediated by a lower gestational age . Maternal factors such as increased age, cigarette smoking and pregnancy-related disorders have also been associated with foetal growth restriction. In our study, however, women with RA were less likely to be cigarette smokers. Moreover, adjustment for maternal age and self-reported smoking did not substantially change the estimates. We also found that women with RA had a higher prevalence of pre-eclampsia, which may explain in part the increased risk for SGA birth and also the increased risk of stillbirth. The higher prevalence of stillbirth amongst women with RA may on the other hand stem from impaired foetal growth, as growth restriction and stillbirth are strongly correlated . Other risk factors for stillbirth such as placental thrombosis and antepartum haemorrhage  may also be more prevalent amongst women with RA. It is reassuring that we found no increase in the prevalence of low Apgar score, suggesting that RA was not associated with severe foetal distress during delivery.
Congenital abnormalities may result from inflammatory activity related to RA or from medical treatment necessary for controlling disease activity. Confounding by unmeasured risk factors could also affect our results. Methotrexate, the mainstay treatment for RA, is considered a teratogen even in once-weekly doses of 20 mg or less . Moreover, methotrexate induces folate depletion and lower intake of folic acid around the time of conception has been associated with an increased risk of facial clefts . Corticosteroid use has also been suggested as a risk factor for oral clefts . Amongst congenital abnormalities, limb reduction and cleft lip/palate consistently had the highest prevalence in the two countries. Nevertheless, our estimates are based on very few outcomes.
The study was conducted in the setting of two uniform healthcare systems with complete registration of births and follow-up of congenital abnormalities diagnosed during the first year of life. We identified patients with RA based on a diagnosis from either a hospitalization or a hospital outpatient visit, which may be considered a study weakness as such diagnoses are known to be not entirely accurate. Although a Swedish study including 424 women diagnosed with RA between 1964 and1995 found that more than 90% fulfilled criteria for RA , a Danish study found that only 59% of 217 RA diagnoses registered between 1977 and 2001 could be confirmed by medical records . However, low validity of the RA diagnosis would most probably lead to underestimation of the OR values and thus cannot explain our findings of an increased risk of most adverse birth outcomes. Another concern is that only 39% of the pregnant Danish women with RA were registered as rheumatoid factor positive compared with 62% of the Swedish women, suggesting that this information is under-reported in the Danish National Patient Registry. However, when we restricted our analysis to Swedish data, the conclusion that the ORs were generally higher amongst children born to mothers with rheumatoid factor-positive RA compared to those with seronegative RA did not change. Information regarding congenital abnormalities is known to be of high quality in both Denmark [22, 33] and Sweden .
It is possible that we may have missed patients who were treated by rheumatologists in private practice. Yet, only a minority of patients with RA are treated by private practitioners and in both countries, most pregnant women with RA are likely to have records in the patient registries [34, 35].
Although we controlled for some possible confounding factors, such as maternal age and self-reported smoking, we did not include information on other factors, such as social status and years of education, which could confound our estimates. A comparative study including births in Nordic countries from 1981 to 2000 found that women with <10 years of education had increased risks of very preterm birth and moderately preterm birth when compared with women with >12 years of education . However, this study did not present adjusted estimates. Recently, a study from the Danish National Birth Cohort found that maternal smoking and BMI are important intermediates in the association between educational level and low birthweight at term . We controlled for maternal smoking and, in a subanalysis, also included BMI as a potential confounder and, therefore, controlled for factors that are likely to mediate the effect of low level of education. For that reason, we find it unlikely that level of education and socio-economic status will entirely explain our increased risks. Moreover, we adjusted for parental cohabitation, which is related to socio-economic status and has previously been shown to be associated with low birthweight .
We lacked clinical information on disease activity and treatment for RA during pregnancy. We, therefore, could not determine whether the increased risks of adverse birth outcomes were confined to women with high disease activity or perhaps more prevalent amongst women who received particular types of antirheumatic treatment. Whether the reason for our finding that the risk of abnormalities in children born to women with RA was not elevated in the most recent study period was due to increased use of diagnostic prenatal testing, to improved anti-inflammatory treatment, or to inclusion over time of more women who were treated exclusively as outpatients need to be evaluated in future studies.
In conclusion, we found an increased risk of adverse birth outcomes amongst women with RA compared to women without this condition. It remains unclear whether this increased risk is related to disease activity and thereby indicates a need for improved anti-inflammatory treatment, whether it is related to medical treatment itself or whether it is an artefact of uncontrolled confounding factor(s).
Conflict of interest statement
The Department of Clinical Epidemiology, Aarhus University Hospital and the Clinical Epidemiology Unit, Karolinska Institutet are involved in pharmacoepidemiological studies, including studies on drugs used for treatment of rheumatoid arthritis. These studies are funded by research grants from several pharmaceutical companies (Amgen, Pfizer, Glaxo and Centocor).