|Home | About | Journals | Submit | Contact Us | Français|
The objective of this meta‐analysis was to determine whether gestational use of hydroxychloroquine (HCQ) for autoimmune disorders leads to an increase in the risk for adverse pregnancy outcomes.
MEDLINE, EMBASE, Web of Science, and Cochrane Central Register of Controlled Trials databases were searched from inception to November 21 2014. Studies which reported the outcomes of pregnant women after exposure to HCQ during pregnancy and including a control (unexposed) group were included. Two independent reviewers carried out the review and the quality assessment using the Methodological Index for Non‐Randomized Studies (MINORS). A random effects method was used to calculate the odds ratios (OR) for the outcomes.
The meta‐analysis reported no significant increases in rates of major congenital (OR 1.13, 95% confidence interval (CI) 0.59, 2.17), craniofacial (OR 0.62, 95% CI 0.13, 3.03), cardiovascular (OR 1.06, 95% CI 0.29, 3.86), genitourinary (OR 1.38, 95% CI 0.42, 4.53), nervous system malformations (OR 1.81, 95% CI 0.31, 10.52), stillbirth (OR 0.69, 95% CI 0.35, 1.34), low birth weight (OR 0.69, 95% CI 0.21, 2.27) or prematurity (OR 1.75, 95% CI 0.95, 3.24). The rate of spontaneous abortions, however, was found to be significantly increased in HCQ exposed pregnancies (OR 1.85, 95% CI 1.10, 3.13). No significant heterogeneity was detected among the studies for the evaluated outcomes except prematurity.
Prenatal exposure to HCQ for autoimmune diseases does not appear to increase the risk of adverse pregnancy outcomes except spontaneous abortion rate, which may be associated with the underlying disease activity (bias by indication) and needs further investigation.
Connective tissue diseases predominantly affect women in their reproductive period 1. The prevalance of previously undiagnosed systemic rheumatic disease is reported as 1% in pregnant women 2. Although fertility is maintained in women with systemic lupus erythematosus (SLE) 3, SLE pregnancies have been associated with increased risk for adverse outcomes such as spontaneous abortion, stillbirth, pre‐term birth, low birth weight and pre‐eclampsia 4, 5, 6, 7.
Hydroxychloroquine (HCQ), the hydroxyl derivative of chloroquine, is an antimalarial agent which is widely used, either alone or in combination with other agents, in the treatment of SLE, rheumatoid arthritis (RA) and various other autoimmune diseases 8, 9, 10. Both HCQ and chloroquine decrease the frequency of disease flares and the risk of complications and are found to be beneficial for the maintanence of remission in SLE patients 11, 12, 13.
Various proposed mechanisms through which HCQ exerts its immunomodulating action have been suggested, such as: inhibition of lysosomal acidification, phagocytosis, proteolysis, antigen presentation and chemotaxis, decreasing the production of pro‐inflammatory cytokines and prostaglandins, inhibition of matrix metalloproteinases, blocking T and B‐cell receptor and toll‐like receptor signalling, DNA stabilization, absorption and preventing ultraviolet light cutaneous reactions 8, 9, 10, 14, 15. Additionally, HCQ has been suggested to have lipid lowering, anticoagulant and antidiabetic effects which may contribute to reduce the high cardiovascular risk in SLE patients 15, 16. A significant reduction in the frequency of infections in SLE and RA patients has been observed previously 14, 17. Recent studies also indicated a possible therapeutic benefit for pregnancies with antiphospholipid syndrome, since HCQ has been shown to reverse, at least partially, the antiphospholipid antibody‐induced inhibition of trophoblast migration and to restore the diminished trophoblast fusion and function 18, 19, 20, 21.
Almost 40 years before its transplacental passage was shown in humans 22, ototoxic and retinotoxic findings were shown in infants who were exposed to chloroquine in utero 23, 24, 25, which were similar to those seen in animal studies 12 and raised concerns regarding possible fetotoxicity of chloroquine and HCQ. These findings were not confirmed in three case series which conducted ocular assessment in children who were prenatally exposed to HCQ 26, 27, 28. However a recent study 29, in which the methodology was questioned 30, reported abnormalities in electroretinogram findings in six of 21 neonates whose mothers were treated with HCQ 29.
To date, safety of HCQ use in pregnancy was reported in seven observational cohort studies 31, 32, 33, 34, 35, 36, 37 and one randomized controlled trial 38. A 2009 meta‐analysis, involving four of the observational cohort studies 31, 32, 33, 34 and excluding the randomized controlled trial 38, reported no increase in the risk of major congenital malformations, spontaneous abortions, fetal death and prematurity 39.
Given that half of all pregnancies are unplanned 40 and there are theoretical and established beneficial effects of using HCQ in the treatment of autoimmune diseases, investigating the safety of HCQ use and/or exposure in pregnancy is highly important.
The objectives of this systematic review were to assess whether prenatal exposure to HCQ for autoimmune diseases is associated with adverse pregnancy outcomes by conducting an updated meta‐analysis, involving the recent cohort studies 35, 36, 37. The main outcomes considered for this meta‐analysis were the rates of major congenital malformations and specific malformations such as craniofacial, cardiovascular, nervous system and genitourinary malformations, spontaneous abortion, stillbirth, low birth weight and prematurity. Cardiac neonatal lupus or congenital heart block was evaluated qualitatively, since there was insufficient data to perform a quantitative analysis.
Searches were conducted in MEDLINE, EMBASE, Web of Science and the Cochrane Central Register of Controlled Trials in November 2014. Both database subject heading fields (e.g. MeSH in MEDLINE, Emtree in EMBASE) and text word fields were searched using ‘hydroxychloroquine’ and other synonymous terms gleaned from the database scope notes. A combination of subject headings and text word fields were searched to capture the concepts of congenital malformations or pregnancy. These two concepts of HCQ and congenital malformations or pregnancy were combined with the Boolean ‘AND’. Database subject headings were exploded, when applicable, to include narrower headings. In all databases, truncation symbols and adjacency operators were used in text word searches when appropriate, to capture variations in spelling and phrasing. No language or date restrictions were applied. Hand searching was not performed. The study flow diagram is presented in Figure 1. Details of the search process are presented in the Appendix.
Randomized‐controlled, observational cohort and case–control studies with prenatal exposure to HCQ were included. Patients with any autoimmune condition/disease, treated with HCQ, were included. A study was considered eligible if it met the following criteria: 1) exposure to HCQ at least in the first trimester of pregnancy was reported, 2) a control (non‐exposed to HCQ) group was included, 3) the data reported were not overlapping with another study by the same authors and 4) study sample size larger than 10. The exclusion criteria for the meta‐analysis were case reports, case series, animal studies and editorials.
The rates of major congenital, craniofacial, cardiovascular, nervous system and genitourinary malformations, spontaneous abortions, stillbirth, low birth weight and prematurity were the main outcomes of interest for this meta‐analysis. Cardiac neonatal lupus or congenital heart block were evaluated qualitatively.
Two authors (YCK and JO) independently reviewed the titles and abstracts and data were extracted by using a standardized form. The references of the relevant studies were also reviewed to identify any eligible study which was not retrieved in the previous searches. Any disagreements were resolved by discussion and consulting with another author (GK). The Methodological Index for Non‐Randomized Studies (MINORS) scale was used for quality assessment of the observational study methodologies 41. The only randomized controlled trial 38 was assessed using the Jadad Sscale 43 and Cochrane Collaboration Risk of Bias tool 44, 45 in terms of methodological quality.
Data were extracted from eligible studies and arranged in a two by two table. The odds ratios (OR) and 95% confidence intervals (CI) for the dichotomous outcomes of interest were calculated. Outcome data were combined by using a random effects model with RevMan 5.3 (Review Manager 5.3; Cochrane Collaboration, Oxford, UK). Heterogeneity was assessed utilizing the Q and I‐square statistic. An I‐square value between 25%–50% signifies low heterogeneity, between 50%–75% moderate and >75% signified high heterogeneity 46 A funnel plot was not utilized to assess publication bias since it is suggested to have low power for detecting asymmetry with good accuracy if the number of included studies is below 10 47.
Seven observational cohorts and one randomized controlled trial met the inclusion criteria including a total of 740 infants born to 800 HCQ‐exposed and 1080 infants born to 1130 control (unexposed) pregnancies, respectively 31, 32, 33, 34, 35, 36, 37, 38. MINORS score of the included observational studies ranged from 14 to 22 over a total of 24 41. The risk of bias in the randomized controlled study was detected as low for five domains (random sequence generation, blinding of outcome assessment, incomplete outcome data, selective reporting and other bias) and unclear for two domains (allocation concealments and blinding of participants). The Jadad score was 3, which suggested a good methodological quality. Two of the studies were presented as abstracts 33, 37. The most common indication for HCQ use was SLE. The HCQ dose used was only reported in three of eight studies, as 200–400 mg day–1 31, 32, 35. Duration of HCQ use varied widely among studies from within 3 months before or throughout the pregnancy. Additional medications used by the pregnant women were as follows: corticosteroids, non‐steroidal anti‐inflammatory drugs (NSAIDs), azathioprine (AZA), low molecular weight heparin (LMWH), aspirin (ASA), intraveous immunoglobulin (IVIG), sulfasalazine and mesalamine. Detailed descriptive characteristics of included studies are presented in Table 1.
Eight studies were included for the major congenital malformation meta‐analysis 31, 32, 33, 34, 35, 36, 37, 38. However the presence of zero events in the exposed and control groups did not permit us to calculate ORs for two of the studies 31, 38. Therefore, this meta‐analysis included a total of 698 exposed and 1026 control (unexposed) infants from six studies. The major congenital malformation rate in the exposed group was not significantly higher than those in the control group (OR 1.13, 95% CI 0.59, 2.17). No significant heterogeneity was detected among the studies (P = 0.31, I‐square = 16%) (Figure 2A).
For craniofacial, cardiovascular and genitourinary malformations six studies were found eligible 31, 32, 33, 35, 36, 38. Numbers of specific malformations were obtained through personal communication with the author for the study by Diav‐Citrin et al. 35 In two of the studies, 31, 38 ORs were not available since zero events were reported for the exposed and the control groups. Therefore four studies with 476 exposed and 743 control infants were included in the meta‐analysis of each specific malformation 32, 33, 35, 36. The craniofacial (OR 0.62; 95% CI 0.13, 3.03), cardiovascular (OR 1.06, 95% CI 0.29, 3.86) and genitourinary malformation rates (OR 1.38, 95% CI 0.42, 4.53) were not significantly different from the control, respectively. No significant heterogeneity was present among the studies (for craniofacial malformations; P = 0.87, I‐square = 0%, for cardiovascular malformations P = 0.59, I‐square = 0%, for genitourinary malformations P = 0.61, I‐square = 0%) (Figure 2B).
For the meta‐analysis of nervous system malformations, six studies were found eligible 31, 32, 33, 35, 36, 38. However, ORs were not calculated for three of them since they reported zero events for the exposed and the control groups 31, 32, 38. Therefore the remaining three studies were included with a total of 359 exposed and 684 control infants 33, 35, 36. There was no significant difference in nervous system malformation rate between the exposed and control groups (OR 1.81, 95% CI 0.31, 10.52). No significant heterogeneity was found to be existing between studies (P = 0.47, I‐square= 0%) (Figure 2C).
Five studies comprising a total of 373 exposed and 717 control pregnancies were found eligible for the meta‐analysis of spontaneous abortion rates 31, 32, 34, 35, 38. The study by Levy et al. was not included in the calculation of ORs since it reported zero events in exposed and control groups 38. Meta‐analysis yielded a significantly higher spontaneous abortion rate in the exposed group compared with control (OR 1.85, 95% CI 1.10, 3.13). No significant heterogeneity was present among the studies (P = 0.33, I‐square = 13%) (Figure 3A). As a subgroup analysis, we repeated this meta‐analysis excluding the pregnant women with antiphospholipid syndrome (368 exposed vs. 687 unexposed pregnancies), and the significance in the spontaneous abortion rate persisted (OR 1.77, 95% CI 1.09, 2.88), with no significant heterogeneity (P = 0.44, I‐square = 0%) (Figure 3B). In the sensitivity analysis, excluding the studies either by Clowse et al. 34 or Diav‐Citrin et al. 35 yielded non‐significant ORs with no significant heterogeneity (OR 1.59, 95% CI 0.88, 2.89, OR 1.56, 95% CI 0.69, 3.50, respectively) (P = 0.32, I‐square = 12%, P = 0.24, I‐square = 31%) (Figure 3C).
For stillbirth rates, six studies including a total of 571 exposed and 916 control pregnancies were included 31, 32, 34, 35, 36, 38. The stillbirth rate in the exposed group was not significantly higher than the control group (OR 0.69, 95% CI 0.35, 1.34). No significant heterogeneity was detected between the studies (P = 0.91, I‐square = 0%) (Figure 3D).
For low birth weight rate, two studies with a total number of 110 exposed and 189 unexposed infants were included 31, 34. There was no significant difference in low birth weight rate between the exposed and control groups (OR 0.69, 95% CI 0.21, 2.27). A non‐significant, yet moderate heterogeneity was detected between the studies (P = 0.54, I‐square = 72%) (Figure 4A).
In assessing the rate of prematurity, five studies with a total number of 515 exposed and 828 control infants were included 31, 32, 34, 35, 36. No significant difference in terms of prematurity rates between the exposed and control groups was detected (OR 1.75, 95% CI 0.39, 3.24). A significant moderate heterogeneity was present among the studies (P = 0.007, I‐square = 72%) (Figure 4B). A sensitivity analysis demonstrated the cause of this significant heterogeneity was the study by Diav‐Citrin et al. 35 The OR for prematurity remained non‐significant (OR 1.27, 95% CI 0.88, 1.85) in a further meta‐analysis after excluding this study (P = 0.91, I‐square = 0%).
The literature search retrieved three cohort and one case–control study which reported cardiac neonatal lupus and/or congenital heart block as the primary or one of the investigated outcomes 32, 37, 45, 46. Costedoat‐Chalumeau et al. reported no ocurrence of congenital heart block in HCQ‐exposed and control groups 31. Gayed et al. detected three congenital heart block cases out of 149 infants exposed to HCQ (2.0%) and four congenital heart block cases out of 138 infants in the control group (2.9%) 37. This difference was not statistically significant (P = 0.96). Izmirly et al., with a case–control approach, showed that overall HCQ exposure in mothers with SLE who had anti‐Sjögren's‐syndrome‐related antigen A (anti‐SSA/Ro) and/or anti‐Sjögren's‐syndrome‐related antigen B (anti‐SSB/La) antibodies was significantly associated with diminished risk of cardiac neonatal lupus of which a great majority was presented as congenital heart block (OR 0.28, 95% CI 0.12, 0.63). Although this significance was lost in multivariable analysis (OR 0.46, 95% CI 0.18, 1.18), in which the data adjusted for birth period, race/ethnicity, antibody status, non‐fluorinated steroid use and prior cardiac‐NL risk, the authors suggested that the OR estimate remained in the direction of a protective effect 48. In a subsequent historical multinational cohort, the maternal use of HCQ was associated with a significant decrease in the risk of recurrent cardiac neonatal lupus (7.5% vs. 21.2%, P = 0.050) and this significance remained unaltered in multivariable and propensity score analysis (OR 0.23, 95% CI 0.06, 0.92), which considered the maternal age at time of birth, study source, maternal diagnosis, race/ethnicity, non‐fluorinated steroids, IVIG, antibody status, gender of child and birth year 49. Thirty‐three cases included in this study 49 were also included in the previous case–control evaluation 48.
Our systematic review and meta‐analysis evaluated all available epidemiologic data with respective control (unexposed) groups and showed no significant increases in the rates of major congenital, craniofacial, cardiovascular, nervous system and genitourinary malformations in the infants of HCQ‐exposed pregnant women, respectively. Additionally, no significant increase in the rates of stillbirth, low birth weight and prematurity were detected. However, a significant increase in spontaneous abortion rate of HCQ‐exposed pregnancies was demonstrated. Qualitative review on HCQ use during pregnancy and occurence of cardiac neonatal lupus or congenital heart block suggested a possible protective effect regarding reccurence. However more data are needed before reaching a conclusion.
To investigate the causes of the increase in spontaneous abortion rate, we first tried to rerun the meta‐analysis after excluding the pregnant women with antiphospholipid syndrome (APS), since it may be associated with an increased rate of fetal loss 7. However, we were only able to exclude pregnant women with APS in the study by Clowse et al. 34 since the other studies did not report the numbers in a way making it possible to actually identify the number of pregnant women with APS. This exclusion slightly reduced both the OR and lower limit of CI (OR 1.85, 95% CI 1.10, 3.13 vs. OR 1.77, 95% CI 1.09, 2.88), although the significance persisted. Secondly, we conducted a sensitivity analysis which showed that the significance may stem either from the study by Clowse et al. 34 or Diav‐Citrin et al. 35. Clowse et al. 34 used three groups in their comparison, a control (unexposed with disease), a HCQ‐exposed (throughout the pregnancy) and a group in which HCQ was stopped (anytime or just before pregnancy). While conducting this meta‐analysis, we combined the latter two in defining the exposed group. Originally, no statistically significant increase in the spontaneous abortion rate was detected between the groups (4%, 13% and 11%, respectively) in this study. However, the authors mentioned a significant increase in lupus activity and flares after cessation of treatment regardless of prior lupus activity in the HCQ‐stopped group. Our results may have been affected by the increase in disease activity since flares in SLE pregnancies, depending on the severity, are suggested to be associated with fetal loss 50. In the other study by Diav‐Citrin et al. 35 two groups of pregnant women, an exposed and a non‐disease control, were compared and spontaneous abortion rate was found to be significantly elevated in the exposed group. After adjustment for predictors such as smoking status, maternal age, miscarriage history, gestational age at initial contact and type of exposure, an increased rate for spontaneous abortions was significantly associated with an earlier age at the initial contact and a higher maternal age. Although basal characteristics between the exposed and control groups were similar for the majority of the parameters in this study, gestational age at initial contact was lower in the HCQ‐exposed group (median 7 weeks, interquartile range 5–12) compared with control (median 10 weeks, interquartile range 6–17), which may have led to a significant increase in the spontaneous abortion rate in HCQ‐exposed group. These findings from two studies suggest that the significant increase in spontaneous abortion rate in our meta‐analysis may be either associated with the pooled effect of the underlying disease activity in the study by Clowse et al. 34 or earlier gestational age of the exposed group in the study by Diav‐Citrin et al. 35 or both. Nevertheless, the increase in spontaneous abortion rate lacks biological plausibility, since recent evidence, although based on in vitro studies, implies a beneficial effect of HCQ on trophoblast fusion and migration 18, 19, 20, 21. However, this issue deserves further investigation.
The findings in our meta‐analysis, except the increase in spontaneous abortion rate, were comparable with the findings in the previous meta‐analysis by Sperber et al. 39.
MINORS scale 41 was chosen as the quality assessment tool for the observational cohort studies included in our study. Although it was originally designed to assess surgical trials, the questions included in MINORS are not specific for surgery. We believe that MINORS is more comprehensive for assessing studies investigating the outcomes after medication exposure during pregnancy than the other scales used for cohort studies such as the Newcastle‐Ottawa scale 42. For instance, MINORS asks about the baseline equivalence of groups separately which is particularly important for this meta‐analysis since HCQ‐exposed pregnant women had autoimmune disorders and disease activity between groups might differ and lead to bias. MINORS also assesses prospective calculation of the sample size and determines a threshold of 5% for the lost to follow‐up rate, the latter of which is not assessed in the Newcastle‐Ottawa scale.
Some important strengths of our meta‐analysis deserve to be mentioned. Observational cohort or case–control studies are the primary sources of data for investigating the outcome of medication exposure in pregnancy, since ethical barriers exist for randomized controlled trials. In our meta‐analysis the majority of the data were prospectively collected from diverse geographical locations. In seven of the eight studies included, HCQ‐exposed pregnant women were compared with unexposed pregnant women who were diagnosed with similar autoimmune diseases which decreases the risk of confounding by indication. Additionally, a majority of the studies tried to rule out possible confounders by comparing maternal characteristics, disease type, severity and concomitant medication characteristics at the beginning of the study. Moreover, some studies also tried to measure the changes in disease activity, concomitant medication characteristics and maternal outcome during and at the end of the pregnancy 31, 34, 37, 38. We report no significant heterogeneity for the great majority of our findings, except for the prematurity rates which also remained non‐significant in further assessment.
Some limitations of our meta‐analysis should also be discussed. Our sample size was relatively small and CIs of some outcomes, such as nervous system malformations, were wide. However, for most of the outcomes, our meta‐analysis was able to reject a 2 to 3‐fold increase in risk. Another limitation is that the medication exposure was not exclusively HCQ, but other immunosupressants such as AZA, prednisone and other adjunct medications such as LMWH, aspirin, IVIG, sulfasalazine and mesalamine. However, this issue was taken into account by some of the studies which reported similar basal characteristics of concomitant medications for the exposed and control groups at the time of the enrolment. A few studies also showed that concomitant medication characteristics differed among the groups during the study period. For instance, women who continued HCQ treatment required significantly lower doses of prednisone at term in the study by Levy et al. 38. Similarly, Clowse et al. 34 reported that the numbers of the pregnant women who required prednisone and high dose corticosteroids were lower in the group who used HCQ throughout pregnancy, although the latter finding was not significant.
Based on the current systematic review and meta‐analysis, HCQ use during pregnancy is not associated with a major increase in the rates of major congenital, craniofacial and cardiovascular, nervous system and genitourinary malformations in the infants. Our meta‐analysis also suggests that HCQ use during pregnancy is not associated with other adverse pregnancy outcomes such as stillbirth, low birth weight and prematurity. The significant increase in the spontaneous abortion rate in this meta‐analysis may have rooted from the aforementioned confounders inherent to included studies, and needs to be further studied. Our qualitative review suggested a possible protective effect of HCQ use on the occurrence and recurrence of cardiac neonatal lupus and congenital heart block. However available data are too limited to draw conclusions.
All authors have completed the Unified Competing Interest Form (available on request from the corresponding author) and declare no support from any organization for the submitted work, no financial relationships with any organizations that might have an interest in the submitted work in the previous 3 years and no other relationships or activities that could appear to have influenced the submitted work.
We would like to express our sincere thanks to Dr William O. Cooper and Dr Orna Diav‐Citrin for providing us detailed information regarding organ‐specific malformation rates in their studies. Preliminary findings of this study were presented orally in 2015 at the OTIS Annual Meeting (28th Annual Education Meeting for Organization of Teratology Information Specialists Members and MotherToBaby Affiliates) on June 28, 2015 in Montreal and an abstract was published in the conference book.
Database: EmbaseClassic+ Embase <1947 to 2014 Week 46>.
1 hydroxychloroquine/(14 912).
2 (“sn 8137” or chloroquinol or ercoquin or hydrochloroquine or hydrocloroquine or hydroxychlorochin or hydroxychloroquine or oxychlorochin or oxychloroquine or plaquenil or quensyl).mp. (15 503).
3 1 or 2 (15 503).
4 exp congenital disorder/(969 057).
5 prenatal drug exposure/or prenatal exposure/(24 278).
6 exp prenatal development/(195 358).
7 exp embryonic structures/(101 366).
8 exp teratogenesis/or exp teratogenic agent/(25 872).
9 exp pregnancy/(645 218).
10 exp prenatal disorder/(98 661).
11 prenatal period/(7163).
12 (pregnan* or prenatal* or antenatal*).mp. (944 293).
13 (teratogen* or embryo* or foetus* or fetus* or foetal or fetal).mp. (869 930).
14 ((birth or congenital) adj4 (defect* or malform* or anomal* or abnormal*)).mp. (235 974).
15 (maternal adj4 expos*).mp. (6895).
16 or/4–15 (2 396 434).
17 3 and 16 (1371).
Database: Ovid MEDLINE(R) 1946 to Present with Daily Update.
2 (“sn 8137” or chloroquinol or ercoquin or hydrochloroquine or hydrocloroquine or hydroxychlorochin or hydroxychloroquine or oxychlorochin or oxychloroquine or plaquenil or quensyl).mp. (2923).
3 1 or 2 (2923).
4 exp “congenital, hereditary, and neonatal diseases and abnormalities”/(999 851).
5 prenatal injuries/or prenatal exposure delayed effects/(22 499).
6 exp “Embryonic and Fetal Development”/(221 469).
7 exp Embryonic Structures/(385 041).
8 exp Teratogens/(24 434).
10 exp Pregnancy/(738 436).
11 (pregnan* or prenatal* or antenatal*).mp. (821 385).
12 (teratogen* or embryo* or foetus* or fetus* or foetal or fetal).mp. (668 381).
13 ((birth or congenital) adj4 (defect* or malform* or anomal* or abnormal*)).mp. (118 746).
14 (maternal adj4 expos*).mp. (9986).
15 or/4–14 (2 216 997).
16 3 and 15 (286).
Kaplan Y. C., Ozsarfati J., Nickel C., and Koren G. (2016) Reproductive outcomes following hydroxychloroquine use for autoimmune diseases: a systematic review and meta‐analysis. Br J Clin Pharmacol, 81: 835–848. doi: 10.1111/bcp.12872.