Introduction Unlike other chronic inflammatory diseases, little is known about how ankylosing spondylitis (AS) affects outcomes of pregnancy and birth characteristics.
Methods In a nationwide population-based case–control study, 388 deliveries among women with AS (identified in the Swedish National Patient Register and Medical Birth Register) and deliveries among matched controls (n=1082) from the general population were included. Information regarding pregnancies after AS diagnosis, birth outcomes and possible confounders were retrieved from the national Swedish registers. ORs with 95% CIs were calculated with generalised estimating equations.
Results Emergency and elective Caesarean section (CS) were performed in 16.5% and 9.8% of deliveries among women with AS compared with 6.5% and 6.9%, respectively, in population controls, resulting in OR of 3.00 (95% CI 2.01 to 4.46) and 1.66 (95% CI 1.09 to 2.54), respectively. Offspring of women with AS were more often preterm (9.0% vs 4.9%) and small-for-gestational-age (SGA) (3.1% vs 1.5%), resulting in an OR of 1.92 (95% CI 1.17 to 3.15) and 2.12 (95% CI 1.00 to 4.50), respectively. Adjustment for smoking habits, age, educational level, parity and exclusion of women with comorbidities resulted in similar or only slightly lower point estimates of risk. Cases with a more extensive antirheumatic therapy exposures tended to have a higher risk for elective CS and being SGA.
Conclusions Women with AS had a higher prevalence for several adverse birth outcomes, with results suggesting an influence by both disease severity and comorbidities.
- Ankylosing Spondylitis
Statistics from Altmetric.com
Previous studies have shown that inflammatory diseases such as rheumatoid arthritis (RA) and inflammatory bowel disease (IBD) affects outcomes of pregnancy and birth characteristics, with an increased risk of premature birth and restricted infant growth.1–3
Ankylosing spondylitis (AS) is a chronic inflammatory rheumatic disease. Prevalence figures for AS vary, but recent Swedish studies suggest a national prevalence of 0.2%.4 Symptoms of AS usually starts between 15 and 30 years of age, but there is often a delay before the diagnosis is confirmed by the healthcare system, which in Scandinavia has been shown to be on average 10 years.5 ,6 Characteristic disease manifestations are pain and reduced flexibility of the axial skeleton and often peripheral arthritis, extra-articular manifestations or enthesitis, which all can cause activity limitations, impairment and reduced quality of life.7 ,8 AS may be associated with systemic inflammation and other inflammatory diseases such as psoriasis and IBD. Treatment in clinical practice consists of physiotherapy and pharmacological treatment with non-steroidal anti-inflammatory drugs (NSAIDs), tumour necrosis factor (TNF) inhibitors (TNFi) and disease-modifying antirheumatic drugs (DMARDs). All of the above factors could possibly contribute to adverse pregnancy outcomes in women with AS.
However, there are a few previous reports regarding birth outcomes and AS. Nationwide register-based cohort studies from Norway have demonstrated decreased fertility in women with chronic arthritis overall as well as adverse birth outcomes with increased risks for growth restriction, preterm birth and delivery with Caesarean section (CS).9 ,10 In these studies, AS constituted a minority of all patients and was not reported on separately. Thus, there is a need for larger studies evaluating the potential effect on pregnancy and fetal outcomes specifically for AS.
To study the occurrence of adverse pregnancy outcomes such as pre-eclampsia, preterm birth, low 5 min Apgar score (<7), small-for-gestational-age (SGA), large-for-gestational-age (LGA) birth, stillbirth and delivery outcomes such as use of CS and the use of epidural analgesia among women with AS compared with population-based controls.
Materials and methods
We conducted a population-based study using national registers in Sweden. Sweden has a population of just over 9.6 million, of which 22% are women in the age groups 15–50 years as of December 31, 2013.11 Cases were identified from the National Patient Register (NPR) and controls from the population register. The Swedish healthcare is public and tax funded and provides care to a limited cost to all residents. Using the individually unique civil registration number for Swedish citizens, we linked information from the NPR, the Medical Birth Register (MBR), the Prescribed Drug Register (PDR), Statistics Sweden and the Swedish biologics register.
The NPR (nationwide since 1987) holds information on healthcare visits in public care (inpatient since 1964 and outpatient since 2001), including the Swedish version for International Classification of Diseases (ICD) codes for diagnoses given at discharge from hospital or at a visit to physician. Since 1997, the 10th version of ICD (ICD-10) is used in Sweden. Since 2001, the register includes information of outpatient hospital visits in specialised care, which entail the large majority of patients with a diagnosis of AS and which is the source for identifying diagnosed healthcare visits in the present study.12 ,13 A national validation study of the NPR showed that over 79% of patients diagnosed with AS fulfilled the modified New York criteria for AS and 89% any of the commonly used criteria for various forms of axial spondyloarthropathy.6
From the PDR information was obtained about drug dispensation (NSAID, DMARDs, TNFi) 12 months prior to delivery (only for births after 1 January 2006). Patients with AS were classified into three groups according to the type of drug dispensation: (1) no drugs, (2) only NSAIDs and (3) TNF-α inhibitors, DMARDs or corticosteroids (with or without NSAIDs). Information regarding dispensation of medication for diabetes and hypertension was also retrieved from the PDR.
Cases were included and considered at risk in the present analyses from the date they had been given a diagnoses of AS (ICD-10: M45) in outpatient care (January 2001–2009) in the NPR. This time period was chosen to represent patients with contemporary disease. Patients who had received a diagnosis of systemic lupus erythematosus were excluded. All subsequent deliveries (2001–2009) in the MBR after first date of AS diagnosis were included with the exception of twin births.
Controls were selected from the Swedish Population Register. For each included woman with AS up to five general population comparators (alive and without AS at the date of the index patients’ first AS diagnosis in the NPR) were identified.
For all deliveries among cases and controls, information was gathered from the MBR on 5 min Apgar score, stillbirth, birth weight, gestational week of delivery, use of epidural analgesia and delivery with elective or emergency CS. Preterm birth was defined as a live birth less than week 37 and stratified into groups with moderate preterm birth (week 32–36) and very preterm birth (less than week 32), respectively. Low 5 min Apgar score was defined as a score <7. SGA and LGA were defined as a birth weight >2 SDs below or above the mean for gestational age, based on the Swedish reference curve of estimated fetal growth.14 Information regarding pre-eclampsia was retrieved from the MBR (see online supplementary table S1 for ICD-codes). Pre-eclampsia was defined as a rise in blood pressure (≥140/90 mm Hg) combined with proteinuria (≥0.3 g/24 h or two urinary protein dip sticks of at least +1) occurring after 20 weeks of gestation.
For both cases and controls information regarding covariates and possible confounders were retrieved from the MBR, Statistics Sweden, the NPR, PDR and the Swedish biologics register. From the MBR, information collected by midwifes at the first and subsequent antenatal visits regarding maternal smoking was retrieved and defined categorically (yes or no). Parity was retrieved from the MBR. From Statistics Sweden maternal education was retrieved and dichotomised as either <12 or >12 years of education. From the NPR, information for AS-related comorbidities and renal disease were identified based on being given a corresponding ICD-10 code in the NPR at an outpatient visit prior to delivery (for ICD codes, see online supplementary table S1).
Since information from the PDR is only available from July 2005, information regarding other comorbidities and AS-related medication was only defined in births after 1 January 2006. Diabetes and hypertension are often managed in primary care, from which we had no linked information. We, therefore, added data from the PDR when defining these comorbidities, in order to maximise their detection rate. Diabetes was thus defined as either receiving an ICD-10 code in the NPR or a diabetes drug dispensation prior to delivery (see online supplementary table S1 for ICD and Anatomical Therapeutic Chemical (ATC) codes). Hypertension was defined as having a dispensed prescription of possible antihypertensive medication, such as β blockers, calcium channel blockers, ACE inhibitors and diuretics 1 year prior to delivery (ATC codes, see online supplementary table S1). Information regarding exposure to medication related to AS was also retrieved from the PDR, where data on dispensed prescriptions for NSAIDs, DMARDs, corticosteroids and subcutaneously administered TNFi within 1 year before delivery was retrieved (for ATC codes, see supplementary table S1). Additional information regarding exposure to TNFi with intravenous route of administration was retrieved from the Swedish biologics register 1 year prior to delivery.
Generalised estimating equation analyses were used to compute crude and adjusted ORs with related 95% CIs, to control for confounding, and taking the correlated data structure (more than one birth per woman was allowed) into account. Adjustments were made for maternal age, maternal age squared, maternal smoking, maternal education level and parity. All analyses were conducted using SAS software V.9.3 (SAS Institute, Cary, North Carolina, USA).
A total of 1470 singleton births were included in the analysis. Of these, 388 deliveries occurred in 301 women with AS and 1082 deliveries in 698 women within the control group. Maternal characteristics within this group showed that women with AS demonstrated a higher age at delivery (p<0.0001), higher BMI (p=0.44), higher frequencies for smoking (p=0.83), lower educational level (p=0.10) and parity (p=0.64) compared with controls (table 1). There were, as expected, higher occurrence for AS-associated comorbidities in the AS group. In addition, some comorbidities that may affect birth outcomes were more in patients with AS, such as diabetes (p=0.043), renal disease (p<0.0001) and hypertension (p=0.15).
Among the 388 deliveries in women with AS, 6%, 4% and 10% have also received diagnoses at an outpatient visit in the NPR corresponding to IBD, psoriasis and peripheral arthritis, respectively, prior to delivery. Among the 388 deliveries, 21%, 10%, 13% and 4% had dispensed prescriptions of NSAIDs, DMARDs (26/29 sulfasalazine, 2/29 methotrexate and 1/29 others), corticosteroids or TNFi within 12 months prior to delivery. For the controls, the corresponding frequencies were 4.2%, 0.2%, 0.2% and 0%, respectively.
Regarding fetal outcomes, women with AS had a higher frequency of preterm birth versus the controls, 9.0% vs 4.9%, resulting in a crude OR of 1.92 (95% CI 1.17 to 3.15). Estimates were statistically significant for both moderate (crude OR=1.74, 95% CI 1.02 to 2.96) and very preterm birth (OR=3.38, 95% CI 1.03 to 11.1). Among those with preterm births, a diagnosis of pre-eclampsia was more common among women with AS (17%, n=6) than in the control group (9%, n=5). The frequency of SGA was 3.1% and 1.5% for women with AS and the controls, respectively (crude OR=2.12, 95% CI 1.00 to 4.50). The proportion of stillbirth was similar in women with AS and controls accounting for 0.3% and 0.1% of the total number of deliveries, respectively. Adjustments did not substantially change these point estimates (table 2).
The mean birth weight was 3463 g vs 3570 g for the cases and controls, respectively, resulting in an unadjusted difference of 107 g (95% CI 30.8 to 183.6 g). The unadjusted gestational age in weeks was 39.3 and 39.8 for the cases and controls, respectively, resulting in an unadjusted difference of 0.54 week (95% CI 0.26 to 0.81 weeks).
The ORs for both elective and emergency CS were increased in patients with AS compared with controls (table 2). Among those with elective CS, normal fetal presentation (72% vs 53%) but not pre-eclampsia was more common in AS versus control births, suggesting that other factors (than type of fetal presentation) was the reason for choosing elective CS as delivery method (see online supplementary table S2). Among those with emergency CS, pre-eclampsia (24% vs 8%) but not abnormal fetal presentation was more common in AS-group births versus control births, suggesting that an increase in pre-eclampsia may largely explain the increased risk for emergency CS among AS deliveries (see online supplementary table S2).
Dispensation of NSAID, DMARDs, TNFi or corticosteroids increased the risk of having SGA birth or delivery with either elective or emergency CS. Adjustments for maternal age did not change point estimates (table 3).
In a first sensitivity analyses, limiting the analysis to first delivery in each mother after diagnosis of AS (see online supplementary table S3) with adjustment for confounders, we found similar point estimates as in the main analyses (table 2).
In a second sensitivity analysis, all patients and controls with IBD or psoriasis prior to delivery were excluded. These restricted analyses revealed similar risk estimates when comparing outcomes in women with AS to controls for having preterm birth (crude OR=2.02, 95% CI 1.22 to 3.35), SGA fetus (crude OR=2.11, 95% CI 0.98 to 4.58) and the mother being delivered with emergency (crude OR=1.56, 95% CI 0.99 to 2.45) or elective (crude OR=3.08, 95% CI 2.05 to 4.65) CS (see online supplementary table S4).
In a third sensitivity analysis, all patients with diabetes, renal disease or patients on hypertension medication during 1 year prior to delivery were excluded. In these analyses, the point estimates for preterm birth (crude OR=2.17, 95% CI 1.12 to 4.19) and SGA (crude OR=3.02, 95% CI 1.07 to 8.53) in AS versus control births were similar, whereas those for elective (crude OR=1.63, 95% CI 0.97 to 2.74) and emergency CS (crude OR=1.28, 95% CI 0.73 to 2.26) were slightly lower, all compared with the results of the main analyses (see online supplementary table S4).
In a fourth set of sensitivity analyses, adjustment for healthcare region (for the delivery, in total six regions), and type of delivery hospital (university hospital, ≥1500 deliveries a year, <1500 deliveries a year) did not change point estimates (data not shown).
In this large population-based study, we found that pregnancies among mothers with AS were associated with a more frequent use of CS, with preterm birth and offspring being SGA. Subset analyses indicate that mechanisms behind these findings are complex, differ for the various outcomes and may include a more severe phenotype of AS, an increased occurrence of comorbidities, pre-eclampsia and/or undetermined factors.
The results are in line with studies of other chronic inflammatory diseases. Women with IBD and RA have increased risks for both preterm delivery and having an SGA offspring.1 ,3 Furthermore, in a meta-analysis by Cornish et al15 on pregnancies in women with IBD an increased use of CS was demonstrated. The results of the present study, which is the first study of these issues in relation to AS, are in line with the before mentioned studies of RA and IBD. Furthermore, our point estimates with regard to preterm birth and SGA remained unchanged when restricting the analyses to cases and controls without comorbidities. Pre-eclampsia tended to be over-represented in AS cases with preterm birth, but not to an extent suggesting it to be the sole explanation for the increased rate of preterm birth among women with AS. Based on small numbers, we also found that the risk for SGA (but not preterm birth) increased with more extensive pharmacological dispensation, which supports that SGA may be associated with disease severity in AS.
Our findings could potentially be explained by a number of confounders, that is, maternal smoking, education level, age at delivery and parity. Adjusting for these factors did, however, not notably change our point estimates of the selected outcomes. This suggests that the increased risk for CS, preterm and SGA births are related to AS or closely related factors such as co-occurrence of IBD or psoriasis, which occur at increased frequencies in AS or other comorbidities. However, restricting the analyses to women without IBD and psoriasis did not materially affect the observed associations, and restricting it to those without other comorbidities only moderately reduced the point estimates for elective and emergency CS. On the other hand, the demonstrated gradual increase of CS with more extensive pharmacological dispensation supports that CS may be associated with disease severity in AS, although we cannot formally exclude effects mediated through these treatments per se. Since AS is not an indication for CS according to clinical practice in Sweden, the increased frequency of CS, may be explained by either complications in relation to delivery or that women's symptoms or concerns for pain in relation to delivery. The latter assumption may be supported by that normal (crown) fetal presentation (among those with elective CS) was indeed more common in deliveries among women with AS than in controls. Although pre-eclampsia was not statistically increased in pregnancies among women with AS, subanalyses indicated that a substantial proportion of the excess rate of emergency CS in deliveries among women with AS was due to pre-eclampsia.
For some outcomes, such as the risk for stillbirth, low 5 min Apgar score, very preterm birth, LGA and pre-eclampsia the frequencies were too low to give reliable estimates or did not suggest an increased risk. For epidural analgesia, where we found no difference between AS cases and controls, results may be difficult to interpret, since for example, obstetricians may have concerns to perform such procedure in patients with back problems.
Some limitations of present study should be acknowledged. First, a diagnosis of AS may not be valid. However, we have previously in the NPR validated such diagnoses and showed that 89% of patients with an AS diagnosis fulfilled at least one of the most commonly used criteria for axial spondyloarthropathy. Second, we lacked information regarding disease activity and severity during pregnancy and conception. As a surrogate marker for disease severity, we used data on drug dispensation the year prior to delivery. It should be noted that these data does not take into consideration the exact timing of drug exposure in relation to conception and pregnancy. Third, we may not have included patients treated exclusively by a rheumatologist in private practice, which according to previous estimates constitute a minority of the patients with AS.13 Fourth, complete information regarding congenital abnormalities was not included in the linkage from the NPR and thus not included in the analysis. Finally, since this is a register-based study, we do not have complete data on some factors, for example, cephalopelvic disproportions or fetal distress, which could have affected the outcomes.
There are also strengths of the present study. First, the study was conducted in the Swedish healthcare system with complete registrations of all births. Second, our AS sample group is practically population based, since we included all women with a diagnosis at an outpatient visit from the NPR, where all hospital visit to specialists are registered. Third, the Swedish system with a personal identification number enabled us to link information from a number of national registers and thus to some extent explore the mechanisms behind our findings.
In conclusion, deliveries among mothers with AS are more often performed with CS and the offspring are more often than expected born SGA and preterm. Exploratory analyses suggested that several factors may explain our findings including a more severe AS phenotype as reflected by a more extensive medication, certain comorbidities and pre-eclampsia, although effects of residual or unmeasured confounding cannot be excluded.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Handling editor Tore K Kvien
Contributors `Study concept and design: LTHJ, JA and OS. Acquisition of data: GLJ, LTHJ, JA and OS. Statistical analyses: GLJ and LTHJ. Drafting of the manuscript: GLJ and LTHJ. Critical revision of the manuscript for important intellectual content: GLJ, LTHJ, JA and OS.
Funding OS was supported by grants from the Swedish Research Council (2013–2429) and by grants provided by the Stockholm County Council (ALF project 20130156). JA was supported by grants from the Swedish Research Council, the Swedish Foundation for Strategic Research and the Stockholm County Council. LTHJ was supported by grants from the Gothenburg University and the Swedish Rheumatism Association.
Competing interests None declared.
Ethics approval Granted by the Regional Ethics Committee, Karolinska Institutet, Stockholm, Sweden (ethical approval number: 2011/29-31/1).
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Additional data and information about the study can be accessed by contacting the corresponding author.
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.