Our study based on data from the Norwegian quality register RevNatus, showed that breastfeeding rates in 101 women with SLE six weeks, six and 12 months after delivery, were 78%, 54% and 30%, respectively. Factors positively associated with breastfeeding were younger maternal age and multiparity. Only delivery by emergency caesarean section was associated with non-lactation.

The present study is currently the largest prospective study on breastfeeding in SLE. According to a nationwide survey performed by the Norwegian Central Bureau of Statistics (SSB) in 2013, 97% of Norwegian mothers initiated breastfeeding in the general population [13]. This is slightly higher than in our cohort (91%).

Our study showed an association between duration of lactation and parity. This has been shown in other studies as well, and may be explained by an association between prior positive breastfeeding experience and breastfeeding duration [20, 21]. Primiparity has been identified as a risk factor for delayed lactogenesis [22]. Breastfeeding initiation and duration have been found to increase with the age of the mother [23]. However, in our cohort, we found that women older than 34 years discontinued breastfeeding earlier than younger women.

Although knowledge on SLE and breastfeeding is sparse, a few studies on the topic have been published. The analysis of Noviani et al., assessing 51 pregnancies in women from the Duke Autoimmunity in Pregnancy Registry during the period of 2008 to 2013, showed that approximately 50% initiated breastfeeding [24]. Factors associated with breastfeeding were low disease activity after delivery, full term delivery and planning of breastfeeding early in the pregnancy. However, the American study included a larger proportion of participants with high disease activity compared to our study population.

Also, a study from Argentina showed that among 31 mothers with SLE and 31 non-SLE mothers, 19% of the patients did not initiate breastfeeding compared to 6% of the controls [6]. The average duration of breastfeeding in SLE mothers was six months, and 12 months in non-SLE mothers. The Argentinian study was retrospective and based on self-reported data. The patients had pregnancies subsequent to their diagnosis of SLE. Time from delivery to reporting data was in average 5 years for patients and controls.

Opposite, an Italian study including 43 SLE patients with 57 pregnancies, and with self-reported data on breastfeeding during 2008 to 2019, showed no significant difference in initiation of breastfeeding between SLE mothers and women in the general Italian population [12]. This is comparable to our results. However, the median time of breastfeeding was three months among the patients which is substantially shorter than in the Norwegian study.

The disease should have stayed inflammatory inactive for at least six months in women with SLE planning pregnancy [7]. Although pregnancy is a well-known trigger for increased SLE disease activity [7], a previous Norwegian study showed low or no disease activity at conception and during pregnancy in most SLE women [9]. This correlated well with our results, as all participants had low disease activity scores at inclusion of RevNatus. Skorpen et al. found that disease activity was significantly higher six and 12 months postpartum [9] compared to time of conception. In contrast, our study showed an overall low postpartum disease activity, with a mean LAI-P score of 0.15 and a mean SLEDAI score of 1.4 The extensive use of HCQ among our patients (75%) may be one explanation for the very few flares observed postpartum. HCQ has been recommended for use in pregnancy and during lactation by the ACR (American Collegue of Rheumatology) organization [25].

A higher proportion of elective and emergent caesarian deliveries in non-breastfeeding women with SLE has previously been reported [12]. In our study, only delivery by emergency caesarean section was positively associated with non-breastfeeding at six weeks after delivery. Positive associations between initiation and duration of breastfeeding and term delivery have also been demonstrated in a previous study [6]. We did not find a significant association between prematurity and non-breastfeeding or cessation of breastfeeding. However, analyzing the pregnancies ending in emergency caesarean delivery (n = 10), 50% were born preterm. A previous study has shown that up to 80% of preterm mothers have difficulties establishing sufficient milk supplies and initiating lactation compared to mothers delivering at term [26]. This corresponds to our data, showing that 80% of mothers of preterm infants delivered by emergency caesarean section had discontinued breastfeeding within six months after birth. Acute surgical delivery may lead to significant maternal and fetal stress, which are found to reduce oxytocin release. This in turn may compromise lactogenesis [22, 27]. Not being able to sufficiently supply your child with breastmilk, could possibly worsen the stress response thus reducing lactogenesis even further. These factors may lead to early cessation of breastfeeding.

A limitation of our study was that we were not able to distinguish between full and partial breastfeeding. However, traditionally, Norwegian women do full breastfeeding, also women with rheumatic diseases, during the first three to six months after delivery. In Norwegian maternity wards, all mothers are encouraged to initiate breastfeeding. However, the length of stay in maternity wards are substantially shorter than previous years. This results in shorter follow-up time in hospital, which may lead to early cessation of breastfeeding in some women. Another limitation was the switch of preferred disease activity scoring index, from LAI-P to SLEDAI. At inclusion 79% of the pregnancies were assessed using LAI-P. The scoring systems emphasize and account for different elements, thus, the proportion of exacerbations at different time points could be incorrectly estimated. A third limitation is related to the amount of missing data. This applies especially to the six-month visit which had the lowest attendance (61%), leading to large amounts of missing data at this point. Also, participants did not always complete the digital self-registration prior to each visit, resulting in missing information. Large quantities of missing data pose a risk of type 2 errors. Non-significant data should therefore be interpreted with caution.

The main advantage of our study is the prospective design, allowing us to monitor the participants over time in a public health care system available to all the population, securing more accurate and unbiased registrations of disease activity and quality of life. The diagnoses of SLE were confirmed by an experienced rheumatologist and according to the ACR/EULAR classification criteria. Additionally, by including missing data from subjects who refrained from visits at certain time points, we have exploited all available data in our analyses.