Systemic lupus erythematosus (SLE) is an autoimmune disease that commonly affects the heart, skin, joints, kidneys and eyes, and can cause progressive, permanent damage to these organs [1, 2]. The disease is characterized by periods of quiescence, and periods of flare-ups where autoimmune functioning is highly active [3]. Individuals who will be diagnosed with SLE can exhibit autoimmune abnormalities three years before the clinical onset of SLE [4, 5].

Women are nine times as likely to be diagnosed with SLE compared to men, and among women, the incidence among African American women is 3 times the incidence compared to white women [6–8]. As SLE is primarily diagnosed among women, and is typically diagnosed during their reproductive years, understanding the effect of SLE on perinatal outcomes is important [9]. Studies suggest that preterm birth, small-for-gestational age, preeclampsia, fetal and maternal death are all more common among women with SLE than women without SLE [10–16].

While the risk of extreme outcomes such as fetal and maternal mortality are low among women with SLE, the high risk of more common outcomes, such as preterm birth (25-30% of SLE pregnancies among white women) remains concerning [13, 15–17]. Several studies have examined racial disparities in perinatal outcomes between African American and white women with SLE. These studies often had few African American participants and therefore examined composite adverse outcomes or used hospital discharge data to examine ICD-9 codes for preterm labor as a proxy for preterm birth [18–21]. Only one study was large enough to provide estimates for preterm birth, but not small-for-gestational age among African American women with SLE [22]. Small-for-gestational age (SGA) is used as a population-level proxy of fetal growth restriction, which is measured by successive ultrasounds in utero [23]. The estimates for small-for-gestational age are less consistent in SLE pregnancies, but some indicate the risk of small-for-gestational age is nearly 20% among SLE births to white women, while there are no estimates specifically among African American women [11, 13]. Infants born preterm or small-for-gestational age are vulnerable to infant respiratory issues, infection, slow weight gain, stunted growth and impaired cognitive development [23, 24]. Several studies have suggested that there is an elevated risk of adverse perinatal outcomes among women with SLE even before their clinical diagnosis [13, 17, 22, 25, 26].

Despite the interest in pregnancy outcomes among women with SLE, important gaps in the literature remain. There is limited research on preterm birth and SGA specifically among African American women with SLE, and only one study has examined perinatal outcomes before SLE diagnosis among African American women. In the present analysis, we examine the risk of preterm and small-for-gestational age birth among African American women with SLE compared to the general population of African American women in a large metropolitan area. We examine births both before and after the date of SLE diagnosis. We hypothesize that compared to the general African American population, the risk of preterm and SGA births is elevated among African American women with SLE before diagnosis, and elevated to an even greater extent after diagnosis.

Our final analytical sample included 583 births to African American women with SLE and 11,660 births to African American women in the general population from metropolitan Atlanta. Among women with SLE, the majority of births occurred before SLE diagnosis, with the greatest proportion (40.8%) occurring more than three years before diagnosis (Table 1). The distribution of maternal age at delivery was slightly younger among African American women with SLE compared to the general population of African American women, with 51.3% of SLE births to women under age 25, compared to 46.0% of births in the general population. Women with SLE were likely to initiate prenatal care in the first trimester (72.2% of births), which was similar to the general population (69.7%). About 40% of SLE births and of births in the general population were first births. A noticeably higher proportion of SLE births were delivered by cesarean section (33.1% vs. 24.7%).

Overall, 28.5% of births in our cohort of African American women with SLE were preterm, compared to 15.5% among African American women in the general population. The proportion of births that were preterm among women with SLE increased linearly by timing of the birth in relation to SLE diagnosis, ranging from 16.4% of births more than three years before diagnosis to 43.3% of births more than 3 years after diagnosis (Figure 1). The pattern was different when examining SGA births. The risk of SGA birth was 10.9% more than three years before diagnosis, but 27.9% within 3 years after diagnosis. By more than three years after SLE diagnosis, the proportion of births that were SGA was 21.3% of births. There also was a trend in the proportion of SLE births that were delivered by cesarean section. Births occurring before SLE diagnosis were about as likely to be delivered by cesarean section as the general population (24.7%), but births occurring within 3 years after diagnosis and more than 3 years after diagnosis were much more likely to be delivered by cesarean section (45.6% and 44.7%) (Figure 2). Preterm births to African American women with SLE were more likely to be delivered by cesarean section than preterm births to women without SLE, especially within 3 years after diagnosis where over 70% of preterm deliveries were by cesarean section, compared to 29% in the general population (Figure 3).

In models adjusted for maternal age, education and parity, births occurring to African American women with SLE more than 3 years before their diagnosis did not have an increased risk of preterm birth compared to the general population (RR: 1.03, 95% CI: 0.77, 1.38) (Table 2). Births occurring within 3 years before diagnosis did have an increased risk of preterm birth compared to the general population (RR: 1.71, 95% CI: 1.24, 2.35). For births to women with SLE occurring after their diagnosis, the risk of preterm birth was more than twice that of the general population. Births more than three years after diagnosis had the highest risk of preterm birth, with a risk ratio of 2.83 (95% CI: 2.36, 3.38). These estimates differed only slightly from the unadjusted estimates.

When we restricted the preterm birth models to first births only (and adjusted for maternal age and education), the estimates showed a similar pattern. SLE was not associated with preterm birth more than 3 years before diagnosis (RR: 1.20, 95% CI: 0.77, 1.86), but the relative risk increased for each diagnosis timing category (within 3 years before, within 3 years after, more than 3 years after), with the greatest relative risk occurring among births more than 3 years after diagnosis (RR: 3.56, 95% CI: 2.74, 4.62). The results were also similar when we restricted to vaginal births only. The estimates from the fully adjusted model for preterm birth did not significantly change after adjusting for time period (results not shown).

In models adjusted for maternal age, education and parity examining the risk of SGA birth, among births to women with SLE more than three years before diagnosis, there was no evidence of an increased risk of SGA birth compared to the general population of African American women (RR: 1.05, 95%: 0.72, 1.53) (Table 3). Among births to women with SLE occurring within 3 years before diagnosis, the relative risk increased to 2.38 (95% CI: 1.69, 3.36), and to a maximum of 2.89 (95% CI: 2.03, 4.13) when births occurred within 3 years after SLE diagnosis. Although still greater than the general population, the risk of SGA birth was lower among births occurring more than 3 years after diagnosis compared to those occurring closer to the time of SLE diagnosis (RR: 2.28, 95% CI: 1.69, 3.07). The unadjusted estimates were of a similar magnitude and showed the same pattern, with the greatest increased risk among SLE births occurring within 3 years after diagnosis.

We also restricted the adjusted model for SGA to first births only. There was a similar pattern among first births. There does not appear to be an association among women more than 3 years before their SLE diagnosis (RR: 0.85, 95%: 0.47, 1.55), but an increase within 3 years before diagnosis (RR: 2.60, 95%: 1.70, 3.99), the greatest increase in risk observed among births occurring within 3 years after SLE diagnosis (RR: 2.96, 95%: 1.74, 5.03), which then decreased more than three years after diagnosis (RR: 2.15, 95%: 1.37, 3.36). The estimates showed a similar pattern when we restricted the model to vaginal births only. The estimates from the fully adjusted model for SGA birth also did not significantly change after adjusting for time period (results not shown).

The results of the simulation analysis presented in the Supplementary Appendix S1 suggest that the precision of the estimated confidence intervals may have been slightly overestimated, by 3-4%, by not accounting for the covariance of births occurring to the same woman (Supplementary Figure S1, Supplementary Figure S2).

The births to African American women with SLE in our study population occurred to slightly younger women than births to the general population of African American women. Births to women with SLE were roughly equally likely to have been to women with a college degree and to be first births compared to the general population of African American women. In our analysis, we found an increased risk of preterm birth and SGA birth among African American women with SLE, both in the years immediately before diagnosis of SLE and in the years after diagnosis. However, we saw a different pattern for preterm births compared to SGA births. The risk of preterm birth was elevated among births occurring within 3 years before diagnosis, 3 years after diagnosis and among births occurring more than three years after diagnosis. The greatest increased risk of SGA occurred in the three years immediately after SLE diagnosis.

Individuals with SLE demonstrate immune abnormalities prior to SLE diagnosis, but perinatal outcomes before diagnosis are not often considered. We could only find five studies that examined perinatal outcomes both before and after SLE diagnosis that also included a comparison group [13, 17, 22, 25, 26]. Two of these studies are from the early 1990’s, and may not reflect current practices in counseling SLE pregnancies. Only two used statistical modeling to control for potential confounders. With the exception of Barnado et al., none of these studies provide estimates specifically among African American women. We provide an important replication of Barnado et al.’s findings in a population of African American women in a major metropolitan setting as well as extend their findings to include SGA estimates among African American women with SLE.

All five studies that examined the risk of preterm and/or SGA birth before diagnosis found an increased risk of preterm birth and SGA birth before diagnosis when comparing to a non-SLE cohort. Only one of these studies, Arkema et al., distinguished between births occurring 2-5 years before diagnosis and 0-2 years before diagnosis in a Swedish cohort. This study found the risks of preterm and SGA births were especially elevated 0-2 years before diagnosis. Our results also provide support for an association between a pre-diagnosis state of SLE and adverse perinatal outcomes among African American women who will eventually be diagnosed with SLE. This could be due to immune abnormalities experienced in the years before a clinical diagnosis. Antiphospholipid antibodies and anti-Ro antibodies can both be elevated years before a clinical SLE diagnosis is possible, and have also been shown to be associated with preterm birth [5, 16, 20]. Conversely, the observed association could be due to the symptoms of active SLE, which have been shown to be associated with adverse perinatal outcomes [31–33]. Diagnosis of SLE is often delayed even after the presentation of clinical symptoms [34].

We could only find one other study that examined the risk of preterm birth before diagnosis specifically among African American women with SLE [22]. Our results were of a similar magnitude as Barnado et al. for births before SLE diagnosis, where they found an odds ratio of 1.87 (95% CI: 1.07, 3.28) for preterm birth, but were less extreme than their results for births after SLE diagnosis, where they found an odds ratio of 4.70 (95% CI: 2.42, 9.13) for preterm birth. Barnado et al. was conducted among Gullah American Americans (220 women with SLE, 217 women without SLE) in the Sea Islands of South Carolina, and the large odds ratio for preterm birth after diagnosis was driven by the low risk of preterm birth in their comparison population, where less than 5% of births were preterm. This low risk of preterm birth is not reflective of the risk of preterm birth among African American women in metropolitan Atlanta (16%), or African American women nationally (14%) [35]. Barnado et al. did not examine SGA births.

We found no other studies that separated births after SLE diagnosis into births occurring in the years immediately following diagnosis and births later after diagnosis, so we cannot compare the differing trend that we saw where the risk of SGA births eventually decreased in the 3 years after SLE diagnosis while preterm births continued to increase. The differing trends may be due to random variability of the data. Conversely, the difference may be due to real factors. While estimates vary, a large proportion of preterm births among women with SLE appear to be medically-indicated, with preeclampsia being a major indicator for preterm delivery [36–38]. Although preeclampsia may result in medically indicated preterm birth, it does not necessarily affect fetal growth. Other aspects of SLE disease that could affect fetal growth may be more likely to be stabilized more than 3 years after diagnosis, resulting in a lower risk of SGA. We could not distinguish between medically-indicated and spontaneous preterm births in our data, but we did have information on method of delivery from the birth certificate. Studies suggest that medically indicated preterm births are more likely to be delivered via cesarean section, while spontaneous preterm births are more likely to be delivered vaginally [36, 39]. In our sample, from within 3 years after diagnosis to 3 or more years after diagnosis, the proportion of preterm births delivered by cesarean section, while decreasing slightly, remained high, going from 71% to 59%. While method of delivery is only a rough proxy for the medical necessity of a preterm delivery, this suggests that spontaneous and medically-indicated preterm births remain common, even after a stable treatment regimen for SLE may be established in the early years after diagnosis.

Some limitations should be noted. In Georgia insurance status and maternal receipt of WIC were only available on the birth certificate beginning in 2009 which would have rendered our sample size too small to conduct a multivariable analysis. Data on hypertension and diabetes available on birth certificates is known to be of poor quality, and substantially underestimates the proportion of births to women affected by these conditions [40, 41]. We therefore chose to not include these variables in our analysis. We also did not have data on SLE disease activity or medications during pregnancy. We used birth certificates from the general population of African American women in Georgia as our comparison group. This set of comparison birth certificates potentially contained births to women with SLE, and we have no way to exclude these births using information from the birth certificate. The estimated prevalence of SLE among African American women in Georgia is 196.2 per 100,000 [6]. With the age distribution of the 11,660 African American women included in our comparison group, the expected prevalence of SLE in our comparison group is approximately 0.20%, or 22 women. We do not expect this to have influenced our results. We also did not have a unique identifier among the comparison births to identify births to the same woman. Our simulation analysis (Supplementary Appendix S1) demonstrated that not accounting for births to the same women likely generated estimated confidence intervals that were only 3-4% narrower than what would have been estimated had we been able to account for the covariance of multiple births to the same woman. Finally, as previously mentioned, we were unable to differentiate between medically-indicated and spontaneous preterm births.

Our analysis has several strengths. We included only validated cases of SLE. Studies that draw patients from administrative databases identify women with SLE by the presence of the International Classification of Diseases (ICD)-9 code 710.0 in discharge records [10, 12, 21, 38, 42, 43]. In studies that have validated the use of the ICD code against medical record abstraction in the general population, the positive predictive value of the ICD code alone was approximately 60%, suggesting that a number of SLE cases identified by the code are not true cases [44]. We also included births to women with SLE both before and after clinical diagnosis and were able to demonstrate that the risks for both preterm birth and SGA are elevated in the years before diagnosis. Finally, our study included a large sample of self-identified African American women with SLE, who have been understudied with respect to perinatal outcomes, yet represent a group at a high risk of SLE.

As the prognosis for women diagnosed with SLE improves, more women with SLE will likely pursue pregnancy and childbearing. Our results suggest that African American women with SLE are at greater risk for the adverse outcomes preterm birth and SGA than the general population of African American women, even before a clinical diagnosis of SLE. In general, African American women have a higher risk of preterm birth and SGA, a greater risk of pregnancy complications that may lead to these outcomes, including preeclampsia and gestational hypertension, and are more likely to die in childbirth than white women [45–48]. More work remains to be done to characterize the additional risks around pregnancy and childbirth that African American women with SLE face. Healthcare providers, especially those in communities of color should be better educated about SLE, and have a lower threshold to suspect SLE in the peripartum period.