This national rheumatologist survey focused on knowledge of medication safety, primarily csDMARDs and biologics/small molecules, and respondents’ practice in planned and unplanned pregnancies. There was consensus on discontinuation of 4 csDMARDs (cyclophosphamide, leflunomide, methorexate, and mycophenolate mofetil) in planned pregnancies but limited knowledge on when to stop them prior to pregnancy and no consensus for unplanned pregnancies. Respondents agreed that 3 csDMARDs (azathioprine, hydroxychloroquine, and sulfasalazine) were safe in all pregnancies. There was consensus for using 4 biologics/small molecules (adalimumab, etanercept, certolizumab, and infliximab) in planned pregnancies but uncertainty on when to discontinue them and no consensus in unplanned pregnancies. There was consensus with prednisone use in planned and unplanned pregnancies but uncertainty regarding NSAIDs.
Our survey informs rheumatologists’ practice patterns for IA treatment in pregnancy with implications for identifying gaps in knowledge and research. Guidance for the use of antirheumatic drugs before and during pregnancy from the European League Against Rheumatism (EULAR) [8] and from the British Society for Rheumatology (BSR) and British Health Professionals in Rheumatology (BHPR) on prescribing csDMARDs and biologics in pregnancy [9], were both published in 2016, at the time of our survey administration. Consequently, survey responses may reflect understanding of these guidelines or, given the brief period between their publication and our survey’s administration, baseline practice patterns for future evaluation of the impacts of these guidelines. Nonetheless, survey responses align with EULAR points for discontinuation of methotrexate, cyclophosphamide, and mycophenolate mofetil before pregnancy and continuation of antimalarials, azathioprine, and sulfasalazine. Regarding biologics/small molecules, the EULAR guidance support use of anti-TNFs during the start of pregnancy, with certolizumab and etanercept use being acceptable throughout pregnancy as they have little transplacental passage. While survey responses regarding biologics/small molecules continuation in planned pregnancies largely aligned with these guidelines, medication discontinuation during pregnancy is not fully addressed by the EULAR points. Furthermore, the EULAR points do not address actions in unplanned pregnancies for which we observed uncertainty for 9 of 12 csDMARDs and all 12 biologics/small molecules queried.
The BSR BHPR guidelines address gaps in the EULAR points, particularly when medications should be discontinued. For csDMARDs incompatible with pregnancy, cyclosporine should not be used peri-conception, methotrexate should be stopped 3 months before conception, and mycophenolate mofetil 6 weeks in advance [9]. Leflunomide was suggested to only be compatible during peri-conception if accompanied by a cholestyramine washout regimen [9]. Survey responses for the discontinuation of methotrexate aligned with these generally more conservative guidelines. Further, responses on azathioprine, hydroxychloroquine and sulfasalazine use align with the BSR BHPR guidelines for their continuation during pregnancy. Among TNF-alpha inhibitor biologics, only certolizumab was recommended for use throughout pregnancy, golimumab had no data to support any recommendations, and all others were recommended for first trimester (infliximab) or first and second trimester use only (etanercept and adalimumab) [9]. The BSR BHPR guidelines acknowledged the lack of evidence to support recommendations for non-TNF alpha inhibitor biologics/small molecules but stated unintentional first trimester exposure to these drugs would unlikely be harmful [9]. While these guidelines address gaps observed in our survey for biologics/small molecules use in planned pregnancies, the BSR BHPR guidelines largely do not address medication use in unplanned pregnancies, only touching on methotrexate and leflunomide in ‘accidental’ pregnancies. Given that up to 50% of pregnancies are unplanned [12], our survey points to potential areas needing attention in future guideline work.
Multiple factors may explain the variability of survey responses. When treating maternal illness in pregnancy, disease severity and potential disease effect on pregnancy outcomes need to be balanced against potential treatment risks to developing fetus. Active IA, particularly RA, psoriatic arthritis and ankylosing spondylitis, can damage joints and lead to maternal disability if untreated [13, 14]. Furthermore, active IA, particularly RA, can have adverse effects on pregnancy outcome (small for gestational age infants and premature delivery) [13, 14]. Thus, it is important that maternal disease activity be controlled before and during pregnancy with medications that are safe for the fetus. If maternal disease is in remission or has low disease activity, no or little treatment may be required. Each rheumatologist and patient must decide together the severity of maternal disease activity and balance it against the risk of adverse pregnancy outcomes and potential fetal harm. Patient comfort levels taking medication during pregnancy also needs addressing. Additionally, response variability may be due to relative lack of data regarding medication use in pregnancy compared to the non-pregnant state. It is unethical to perform prospective, double-blind, randomized controlled trials of medications safety in pregnancy. There may be animal data on the effect of some drugs which may or may not be applicable to humans. As such, the only method to obtain drug data in human pregnancies is when a mother becomes pregnant while taking a medication or takes one during pregnancy and the outcomes of the pregnancy and fetus are documented. Further, to collect a sufficient number of case reports of pregnancies exposed to a particular medication requires it be on the market for many years. For new medications, there will be almost no data available on that drug in human pregnancies. Lack of data from prospective randomized trials leaves a gap in knowledge regarding the medication safety in pregnancy, especially newer medications. Some clinicians are satisfied with less observational data and others require more before prescribing drugs in pregnancy; as such, there are variable levels of comfort among prescribers. These aforementioned reasons may be more relevant to explaining survey responses for planned pregnancies. With unplanned pregnancies, such reasons apply along with considerations on values and beliefs around termination of pregnancy.
Limitations of our survey study deserve comment. First, despite centralized invitation from the CRA, the sample size was small and participation rate (20%) was modest. However, the aforementioned survey of UK rheumatologists on DMARDs use during pregnancy similarly yielded a 20% response rate [7]. Additionally, prior surveys to CRA membership using the same methods have yielded similar response rates [15, 16]. With 69% of respondents representing rheumatologists practicing in academic or teaching hospital settings, findings may be subject to potential selection bias. Despite querying 12 csDMARDs, 12 biologics/small molecules, 3 NSAIDs, and prednisone, our survey omitted some medications covered by the EULAR points (e.g. tacrolimus) and BSR BHPR (e.g. belimumab). In addition, survey scenarios with planned and unplanned pregnancies in IA patients queried did not incorporate disease activity or severity, which are important considerations. Finally, our survey did not ask rheumatologists about recommendations made to patients after delivery, specifically, regarding breastfeeding.