Objectives Exploring the associations between disease activity and medications with offspring birth weight, pre-eclampsia and preterm birth in systemic lupus erythematosus (SLE).
Methods Data from the Medical Birth Registry of Norway (MBRN) were linked with data from RevNatus, a nationwide observational register recruiting women with inflammatory rheumatic diseases. Singleton births in women with SLE included in RevNatus 2006–2015 were cases (n=180). All other singleton births registered in MBRN during this time (n=498 849) served as population controls. Z-score for birth weight adjusted for gestational age and gender was calculated. Disease activity was assessed using Lupus Activity Index in Pregnancy. We compared z-scores for birth weight, pre-eclampsia and preterm birth in cases with inactive disease, cases with active disease and population controls.
Results Z-scores for birth weight in offspring were lower in inactive (−0.64) and active (−0.53) diseases than population controls (−0.11). Inactive disease did not predict pre-eclampsia while active disease yielded OR 5.33 and OR 3.38 compared with population controls and inactive disease, respectively. Preterm birth occurred more often in inactive (OR 2.57) and active (OR 8.66) diseases compared with population controls, and in active compared with inactive disease (OR 3.36).
Conclusions SLE has an increased odds for low birth weight and preterm birth, amplified by active disease. The odds for pre-eclampsia is elevated in active, but not inactive disease. This calls for tight follow-up targeting inactive disease before and throughout pregnancy.
- systemic lupus erythematosus
- disease activity
Statistics from Altmetric.com
Systemic lupus erythematosus (SLE) is a chronic rheumatic disease often affecting women in fertile age. In SLE, there is an increased risk of unfavourable pregnancy outcomes including low birth weight and preterm birth and complications like pre-eclampsia, even though there is evidence for less elevated risk over the last decades.1 Pre-eclampsia is one of the risk factors for preterm birth. The increased risk of pre-eclampsia including early-onset pre-eclampsia (before 34 weeks) in SLE may be independent of traditional risk factors such as pregestational hypertension and body mass index (BMI).2 High disease activity and flare shortly before or during pregnancy are factors predictive for complications,3–5 whereas no or low disease activity is favourable.6 A recent population-based study reported lower mean birth weight and gestational age in both first and subsequent births in women with SLE compared with references.7 Low birth weight caused by intrauterine growth restriction is associated with an increased risk of cardiovascular disease and diabetes in the offspring.8 Pre-eclampsia, preterm birth and low offspring birth weight are events associated with a future higher risk of maternal cardiovascular disease9 and death.10 Prednisolone use in pregnancy has been associated with preterm birth and a lower birth weight.11 In SLE, prednisolone is commonly used to treat disease flares, and confounding by indication may explain this finding. Most studies investigating the influence of disease activity on pregnancy outcomes have found high disease activity or disease flare to be risk factors for adverse pregnancy outcomes. To our knowledge, prospective studies comparing pregnancy outcomes in SLE women with inactive disease, SLE women with active disease and population controls have not been reported. We regard birth weight adjusted for gestational age and gender (z-score) as the relevant birth weight outcome in this context. The aim of this study was to explore the possible associations of disease activity and medications with offspring birth weight z-score and the occurrence of pre-eclampsia and preterm birth in women with SLE.
Patients and methods
In this population-based cohort, we linked data from the Medical Birth Registry of Norway (MBRN) with data from RevNatus. MBRN is a national health registry with mandatory registration of variables on all births in Norway. It includes information about maternal health before and during pregnancy as well as maternal and neonatal complications during pregnancy and birth. The variables were decided by consensus among obstetricians, neonatologists and epidemiologists. Since December 1998,12 pre-pregnant maternal diseases including rheumatic diseases have been coded according to the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10).
RevNatus is a nationwide Norwegian multicentre, prospective observational register recruiting women with inflammatory rheumatic diseases who plan pregnancy or are pregnant. Women 18 years or older are included, with follow-up in each trimester and at 6 weeks, 6 and 12 months after birth. All women are diagnosed by a specialist in rheumatology prior to inclusion. Information on obstetric history, disease activity and medications as well as maternal and neonatal outcomes including complications is registered. In the present study, singleton births recorded in MBRN 2006–2015 were eligible for inclusion. Births among women with the diagnosis of SLE recorded in MBRN (ICD-10 codes M32.1, M32.8 and M32.9) and included in RevNatus formed the patient group (n=180). Population controls were all other singleton births registered in MBRN during the same period (n=498 849), but excluding births among women with any rheumatic inflammatory disease (n=2492) according to ICD-10 diagnoses (online supplementary figure S1). The 2015 age cohort was excluded from the population controls as the registration of ICD diagnoses of maternal pre-pregnant disease was not completed. One woman could have several births during the study period. This applied to 28 (15.6 %) of 180 women in the patient group and an unknown proportion of the women in the control population.
All women signed a written informed consent before inclusion. Access to data from MBRN was granted in September 2016 (MBRN assignment 15-1819).
For both patients and population controls, data on maternal age, parity, smoking and BMI were derived from MBRN, as were data on newborns and complications including pre-eclampsia,13 14 preterm birth (<37 gestational weeks) and very preterm birth (<34 gestational weeks). BMI was included as a variable in MBRN in 2012 and reported by 40% of the birth institutions, resulting in high missing numbers. For the patient group, educational status, prior obstetric history and disease-specific information were retrieved from RevNatus. Fulfilment of the 1997 American College of Rheumatology criteria for classification of SLE required ≥4 criteria.15 A positive test for lupus anticoagulant, anticardiolipin antibody IgG and antibeta2 glycoprotein I IgG was defined according to thresholds for positivity at the time of the test.
Assessment of disease activity
Disease activity was assessed by the Lupus Activity Index in Pregnancy (LAI-P), a modification of the Lupus Activity Index (LAI) validated for use in pregnancy.16 LAI-P is described in detail elsewhere.17 Briefly, disease activity is assessed on a scale from 0 (inactive disease) to 2.6 (very high disease activity), with a score above 0.5 considered moderate disease activity. It is a composite score, including items describing general and organ-specific clinical manifestations, current medication and certain laboratory findings. LAI-P was assessed in each trimester and at 6 weeks after birth, and dichotomised to inactive disease (LAI-P=0) and active disease of any severity (LAI-P>0). There were missing data on disease activity at all visits, and most frequently among the preterm (39% missing) and very preterm (50% missing) births in the third trimester. The data were not missing completely at random, as many of these women did not attend the third trimester visit due to birth before scheduled visit. Data on disease activity were more complete in the second trimester (missing in 20% of term and 6% of preterm outcomes).
Calculation of birth weight z-score adjusted for gestational age and gender
Recorded pregnancy outcomes in MBRN included birth weight (grams), gender and gestational age at delivery in days based on a mid-trimester ultrasound examination. Birth weight is influenced by gestational age and gender, differs from country to country and has secular changes. Accordingly, z-score for birth weight was calculated using Norwegian birth weight by gestational age standards covering 20–44 completed weeks, separately for males and females.18 The z-scores were calculated using gestational age in days, with linear interpolation between weeks.
Group comparisons were performed using independent t-test for continuous variables and the Pearson χ2 test or the unconditional z-pooled test for categorical variables.19 We used linear regression with z-score as dependent variable, and logistic regression for dichotomous-dependent variables (pre-eclampsia and preterm birth). As covariates, we compared population controls with cases with inactive disease (LAI-P=0) and cases with active disease (LAI-P>0) in the second trimester. We carried out the analyses unadjusted, and adjusted for maternal age (<35 years/≥35 years), parity (no birth/≥1 birth) and smoking in pregnancy (yes/no). We also carried out analyses for first and subsequent births separately. Separate analyses were performed concerning use of prednisolone (yes/no) in the second trimester, and adjusting for hydroxychloroquine (yes/no) and azathioprine (yes/no). Missing values were handled by available case analysis. Two-sided P values less than 0.05 were considered statistically significant, and 95% CIs are reported where relevant. The statistical analyses were performed using SPSS V.22.
During 2006–2015, 237 inclusions among 203 women diagnosed with SLE were registered in RevNatus. Of known outcomes (n=223), 5% did not become pregnant, miscarriage was reported in 12%, and 83% resulted in live birth. There were 180 singleton and 6 twin deliveries. Among the singleton births, 26 women had two deliveries and 2 women had three deliveries. The majority (141/180) were included in the first trimester, and the remaining in the second trimester. A total of 498 849 singleton births registered in MBRN during 2006–2014 served as population controls. Maternal mean age among patients was significantly higher compared with population controls (mean difference 1.09 years), a lower proportion smoked, and parity and BMI were similar (table 1).
The cases were grouped according to inactive disease (LAI-P=0) and active disease (LAI-P>0) of any severity in the second trimester. In 32 patients, disease activity was not registered. Clinical characteristics of the disease activity groups and the above group are presented in table 2. The disease activity groups showed no differences of statistical significance except smoking, which was more common in women with active disease.
Between 56.6% and 59.9% of women with SLE had inactive disease during pregnancy and 6 weeks after birth, and less than 10% experienced moderate disease activity or higher (LAI-P>0.5) (table 3). Women delivering preterm mainly had active disease (LAI-P>0) on the four scheduled visits (60.0%, 64.5%, 70.0% and 64.7%, respectively). Active disease in the first or second trimester resulted in very preterm birth in 15.4% and 12.9%, respectively, whereas inactive disease resulted in very preterm birth in 6.0% in both groups. The most common disease manifestations in the first and second trimesters were skin (36.0% and 26.3%), joint (26.0% and 17.5%) and haematologic (17.4% and 14.8%). Only 4.2% and 3.6% had active kidney disease, respectively.
Association between SLE disease activity and birth weight z-score, pre-eclampsia and preterm birth
The birth weight z-score was significantly lower in offspring of women with SLE than of population controls (mean difference 0.47). We found significantly lower birth weight z-scores in both disease activity groups compared with population controls, but no significant difference between disease groups (table 4). There was a significantly higher odds of small for gestational age (SGA, ≤10 percentiles) in inactive as well as active diseases compared with population controls (OR 2.45, 95% CI 1.47 to 4.08, P=0.001 and OR 2.66, 95% CI 1.49 to 4.75, P=0.001, respectively). We found no significant differences between disease groups.
Women with SLE had a statistically significantly higher odds of pre-eclampsia and preterm birth compared with population controls, OR 2.70 (95% CI 1.56 to 4.65), P<0.001 and OR 4.03 (95% CI 2.78 to 6.59), P<0.001, respectively. Regarding pre-eclampsia, we found no statistically significant difference between population controls and women with inactive disease, but statistically significantly higher odds when women had active disease. There was substantially higher odds for pre-eclampsia in women with active compared with inactive disease (table 5). Concerning preterm birth, there was a statistically significantly higher odds compared with population controls, both in women with inactive and active diseases. Active disease had a more than twofold increased odds compared with inactive disease. In table 5, OR and P value for pre-eclampsia and preterm birth are shown for inactive disease compared with population controls, for active disease compared with population controls, and for active compared with inactive disease.
We adjusted for factors known to influence outcomes.7 20 21 The results presented in tables 4 and 5 were substantially unchanged after adjusting for maternal age (<35 years/≥35 years), parity (no birth/≥1 birth) and smoking in pregnancy (yes/no) (data not shown). In separate analyses for first and subsequent births, the observed association was greater for subsequent than for first births for z-score and pre-eclampsia, while this was not the case for preterm birth (online supplementary tables S1 and S2). The P values for the interaction between parity and disease activity were 0.78, 0.24 and 0.51, respectively. Although not considered statistically significant, we find it noteworthy that these effects of parity are observed for both disease activity groups.
Supplementary data 2
Influence of medications on birth weight z-score, preterm birth and pre-eclampsia
Prednisolone was used significantly more often in the second and third trimesters among women with active (58.1% and 57.9%) compared with inactive disease (38.1% and 37.5%). There were no significant differences in the use of hydroxychloroquine or azathioprine between the groups in any of the trimesters, or of prednisolone in the first trimester (51.0% and 38.8%). There was similar use of acetylsalicylic acid (online supplementary table S3). Birth weight z-score was statistically significantly lower in offspring of women using prednisolone (mean difference 0.33). There was a substantially higher odds of pre-eclampsia when using prednisolone (OR=2.33), and we found a statistically significant threefold increase in preterm birth (table 6). Results were substantially unchanged after adjusting for hydroxychloroquine (yes/no) and azathioprine (yes/no) (data not shown).
Supplementary data 3
We found a lower birth weight z-score in offspring of the disease group compared with population controls, both in inactive and active diseases. The occurrence of SGA was also increased in the disease groups. Our observations of lower birth weight and restricted fetal growth are in accordance with previous studies.3 7 22 23 There was no evidence of lower birth weight z-score in offspring of women with active compared with inactive disease. There may be several explanations. Our patients had mainly mild disease, with only 4.7% experiencing moderate to high disease activity in the second trimester (LAI-P>0.5). Antiphospholipid syndrome (APS) is a factor independent of disease activity that increases the risk of intrauterine growth restriction and lower birth weight.24 There were positive anticardiolipin antibodies of similar occurrence in the two disease groups, even though we do not know the occurrence of APS, representing an increased risk for lower birth weight. We found prednisolone use to be a risk factor for a lower birth weight z-score, contributing in both disease activity groups.
Our findings concerning the occurrence of pre-eclampsia and preterm birth support our hypotheses that disease activity of any severity increases the risk of adverse events. A higher risk of pre-eclampsia in women with SLE is well known.2 7 25 To our knowledge, it has not been demonstrated earlier that women with inactive disease do not have increased risk compared with population controls. The two disease groups were similar concerning risk factors for pre-eclampsia like maternal age, parity, BMI, diabetes, hypertension, prior kidney disease, positive anticardiolipin antibodies and multiple pregnancies. We believe that a threefold higher odds in active versus inactive disease is clinically relevant, even though it did not reach statistical significance. The odds of preterm birth was elevated both in active and inactive diseases compared with population controls, and in active compared with inactive disease. The most vulnerable, very preterm children were also most commonly delivered in women with active disease. In our cohort, we found a lower proportion of women with active kidney disease than reported in other studies.6 26 Active kidney disease is an important predictor of pre-eclampsia and preterm birth.22 23 27 Our results showed similar occurrence of these events to other studies,6 28 which implies that even less serious disease is an important contributor. There was a twofold increase in the odds of pre-eclampsia in women using prednisolone, and a statistically significant threefold increased odds for preterm birth. The assessment of disease activity (LAI-P) includes medication as one of four groups, contributing to the score if medication is increased. However, there was a stable use of medication in our cohort, indicating that it did not influence the score. We therefore do not believe this to be a confounding factor. It is difficult to delineate prednisolone use from active disease as prednisolone is the medication of choice to treat flares in pregnancy. Treatment with prednisolone does in itself indicate more severe disease. However, we cannot exclude the independent effect of prednisolone use. In clinical practice, this finding emphasises the importance of stable disease-modifying treatment with hydroxychloroquine and azathioprine, minimising the need for prednisolone when the disease is not active.
A limitation of this study is a possible selection of patients. Women with more severe disease may choose not to become pregnant, and adverse events can discourage later pregnancies. Another limitation is missing data on disease activity scores. We knew from our recent longitudinal study on disease activity in this patient group that disease activity was not higher in third than in second trimester,29 and used this registration. The 32 women with missing scores had similar outcomes to the inactive disease group (online supplementary table S4). Since there were lacking data on antiphospholipid antibody status in many patients, we cannot exclude a role for these antibodies concerning our outcomes. Another limitation is that we could not account for dependent observations due to multiple births from the same woman, since this information was unavailable for the population controls. Hence, the precision may be effectively smaller than reported.
Supplementary data 4
Strengths include the utilisation of two nationwide registers. MBRN has existed for more than 40 years. The validity of information on gestational age including birth weight, preterm birth and pregnancy-related hypertensive complications is very good.30 According to Norwegian guidelines,31 women with SLE are offered a multidisciplinary follow-up in pregnancy. We therefore believe there are few women who are not followed up closely and included in RevNatus. The tight follow-up through RevNatus contributes to better controlled disease and improved outcomes. Due to the linkage of registers, we could also confirm a good compliance concerning diagnoses. Of 180 women in RevNatus with the diagnosis of SLE, only 10 (5.6%) did not have this diagnosis in MBRN. This is a lower misclassification rate than earlier reported for pre-pregnant rheumatic diseases in MBRN.32 Furthermore, the diagnosis in RevNatus had to be confirmed by a rheumatologist prior to inclusion, securing the correct diagnosis. An additional strength is the utilisation of a disease activity score validated for use in pregnancy, avoiding pregnancy-related symptoms to be interpreted as active disease. Finally, the birth weight z-score was based on Norwegian standards and gives a more precise estimate for difference in birth weight. However, birth weights in Scandinavian populations cannot be generalised to all ethnic populations.18
In conclusion, we found that offspring of women with SLE have lower birth weight than offspring of population controls without rheumatic diseases. Preterm birth is more common in SLE than population controls, and the risk is amplified by active disease. The risk of pre-eclampsia is elevated in active, but not inactive disease. This calls for tight follow-up targeting inactive disease before and throughout pregnancy.
The authors thank the Medical Birth Registry of Norway (MBRN) for providing data, and the participating departments of rheumatology at the following hospitals for including patients in RevNatus: Betanien Hospital, Skien; Diakonhjemmet Hospital, Oslo; Haugesund Sanitetsforenings Rheumatism Hospital, Haugesund; Haukeland University Hospital, Bergen; Helse Førde, Førde Hospital, Førde; Helse Møre og Romsdal, Ålesund Hospital, Ålesund; Lillehammer Hospital for Rheumatic Diseases, Lillehammer; Nordland Hospital, Bodø; Oslo University Hospital Rikshospitalet, Oslo; Private practice Anne N Bendvold, Kristiansand; St Olavs Hospital, Trondheim University Hospital, Trondheim; Sørlandet Hospital Kristiansand, Kristiansand; University Hospital of North Norway, Tromsø; Vestre Viken Hospital, Drammen; Østfold Hospital, Moss.
Handling editor Tore K Kvien
Contributors CGS, IMG, JFS, KÅS and MW planned the study. CGS, IMG, ØP, HSSK, BJ and MW provided the data. CGS, SL and MW performed the analysis and drafted the paper. All authors contributed to editing the draft for content and approved the final version. CGS and MW had full access to all the data in the study and take responsibility for the integrity of the data and accuracy of the data analysis.
Funding The actual work was supported by the Liaison Committee for Education, Research and Innovation in Central Norway.
Competing interests None declared.
Ethics approval RevNatus was approved by the Regional Committee for Medical and Health Research Ethics (REK Mid-Norway). The present study and linking with MBRN was approved by REK Mid-Norway (2012/1905).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.