Adverse perinatal outcomes after Roux-en-Y Gastric Bypass vs. Sleeve Gastrectomy: a systematic review

Obesity is a growing global health pandemic and it is estimated that approximately 39% of adults over the age of 18 suffer from the condition [1]. Among obese women, infertility is a common issue and studies show that women are strongly over-represented among patients undergoing bariatric surgery [2]. Significant weight loss following bariatric surgery often results in these women regaining fertility shortly after surgery [3]. Whilst maternal health may be improved, there is a large body of evidence that has found an increased risk of perinatal complications secondary to bariatric surgery like preterm birth, small for gestational age (SGA), large for gestational age (LGA), perinatal mortality among other adverse outcomes [3, 4]. However, results may vary markedly between studies and the type of bariatric surgery performed [3, 4]. While gastric bypass, also known as Roux-en-Y Gastric Bypass (RYGB), is a well-studied procedure, there are other procedures that are becoming more popular including Sleeve Gastrectomy (SG) and gastric banding (GB) [5].

RYGB is a restrictive-malabsorptive bariatric surgery that, up until recently, has been the most common bariatric procedure [6]. A small gastric pouch is created at the top of the ventricle and the lower end of the small intestine is connected to the gastric pouch. The small size of the pouch restricts food intake substantially, while the bypass past the small intestine, hinders absorption. RYGB can result in 60–70% loss of excess body weight [7]. SG is a restrictive procedure where most of the ventricle is removed, and much like RYGB, helps the patient to feel full with less food. SG is becoming increasingly popular as it has been associated with less adverse outcomes yet yields similar weight loss results as compared to RYGB [8]. However, there is a lack of long-term outcomes recorded in SG as compared to RYGB since RYGB has been performed over a longer period of time and in a larger group of patients [9].

As such, comparisons of adverse perinatal outcomes between RYGB and SG are important to enable women to make informed choices regarding type of surgery since fertility is an important reason for undergoing bariatric surgery in the first place. To the best of our knowledge, there are currently no systematic comparisons between the two techniques with respect to adverse perinatal outcomes. Therefore, the objective of this systematic review was to compare adverse perinatal outcomes following RYGB vs. SG in pregnant women. More specifically, we wanted to establish whether mode of delivery and maternal body mass index (BMI) differed in women after RYGB vs. SG surgery, and the incidence of adverse perinatal outcomes including low birthweight (in grams), gestational age (in days) and intrauterine fetal death after RYGB vs. SG surgery.

The search generated 3201 records and after duplicates were removed, there were 3143 records to screen against title and abstract. A total of 2969 records were excluded because of irrelevancy and thereafter 174 reports were sought for retrieval of which 170 went on to full-text screening. Reasons for exclusion are shown in Fig. 1. Characteristics of the studies included in the review are presented in Table 1 with 1100 post-RYGB and 209 post-SG participants. The overall risk of bias assessment, Table 2, for most articles was moderate or serious. The major problems of the included studies were bias due to confounding and bias in selection of participants.

In Table 3, maternal characteristics are presented. Adams et al. [18] and Kjaer et al. [19] were the only studies that presented OR. Among maternal outcomes, there were generally small differences between case and control groups, suggesting that maternal BMI, vaginal birth rates and number of Cesarean sections were homogenous between the groups. Time from surgery to conception varied greatly between the studies and was in general shorter in women that underwent SG surgery as compared to RYGB (Table 3).

In the study by Kjaer et al. [19], no CI or SD was reported for maternal BMI, only mean values were given. Furthermore, some results were reported as a “Bariatric surgery group”, meaning a population of both RYGB and Gastric Banding (GB) patients. Mode of delivery, time from surgery to conception and intrauterine death were the parameters reported in this way. There were 339 patients, 84% of which were RYGB patients and 16% were GB patients. Kjaer. [19] reported time from surgery to delivery instead of surgery to conception. The results were presented in days instead of months, and were thus converted to median time from surgery to delivery in months instead of days. In the study by Adams et al. [18], the parameter was “surgery to birth (4.2 ± 3.2 years)” instead of surgery to conception and results were presented as years, which were converted into months. Adam et al. [18] looked at different patient groups: group 1, 2 and 3. For this study, we extracted details with regards to group 2, since this was the only group that fitted the criteria. However, mode of delivery was not documented for all women in Group 2 and therefore only n = 295 were included in this systematic review (and not the entire sample size of n = 764).

Cesarean section was either recorded separately as before or during labor [14] or as Yes/No [15, 17–20]. There was therefore no information on whether these Cesareans were elective or medically motivated.

Perinatal outcomes including birth weight, SGA, LGA, gestational age and intrauterine death are shown in Table 4. Adams et al. [18], Kjaer et al. [19], Santulli et al. [20] and Rottenstreich et al. [14] showed significantly lower birth weight in the bariatric surgery operated women compared to control groups (P < 0.0001, P < 0.001, P < 0.0001, P = 0.001). Patel et al. [19] did not have significant results for the difference in birth weight between the groups (P = 0.361).

In Adams et al. [18], the RYGB group (group 2) had a significantly lower incidence of LGA (P < 0.0001, OR = 0.33 [0.21–0.51]) and higher incidence of SGA (P = 0.0003, OR = 2.16 [1.43–3.32]) compared to controls. In Kjaer et al. [19], the RYGB group had a significantly lower frequency of LGA (P = 0.001, OR = 0.09 [0.02–0.36]) and higher frequency of SGA (P = 0.001, OR = 2.78 [1.56–4.96]) compared to controls. Patel et al. [21] and Santulli et al. [20] showed no significant results in incidences of SGA between the RYGB group and the control group. In group A of Karadag et al. [15], the SG group had a significantly lower incidence of LGA (P < 0.05) and a higher incidence of SGA (P < 0.05). In group B of Karadag et al. [15], the SG group had a significantly lower incidence of LGA (P < 0.05), although not significantly higher incidence of SGA compared to the control group. The SG group in Rottenstreich et al. [14] had significantly lower rates of LGA (P = 0.001) and higher rates of SGA (P = 0.01) compared to controls.

For Kjaer et al. [19], the mean gestational age was significantly lower compared to the control group (P < 0.001). In the other studies, no significant results according to gestational age were observed between the groups. There were no significant results in any of the studies for intrauterine death.

This systematic review aimed to compare perinatal outcomes after two major types of bariatric surgery, namely RYGB and SG. Although our review did not demonstrate any differences in BMI or mode of delivery between RYGB operated women and SG operated women, our results showed similar results of birthweight and gestational age, as well as rates of intrauterine death between the groups. We noticed higher rates of both SGA (22.9%, 11.9%, 14.2%) and LGA (4.2%, 4.8%, 1.7%) in SG compared to Roux-en-Y group (SGA: 8.8%, 7.7%, 11.5%, 8.3% and LGA: 3.4%, 0.7%). However, there was also a tendency toward shorter surgery to conception intervals in these studies, which might be associated with higher rates of perinatal complications. As shown in group A in Karadag et al. [15], the shortest surgery to conception interval (7.8 ± 3.4 months) was for SG which was also associated with the highest rate of SGA infants (22.9%). The adverse impact of a shorter interval was strengthened by a systematic review by Akhter et al. [4] who concluded that SGA occurs more frequently in malabsorptive procedures as compared to restrictive procedures, thus suggesting conflicting results from our study. However, the higher rate of LGA in the SG group did correlate with Akhter et al. [4] and important maternal characteristics, such as advanced maternal age, BMI > 30, alcohol, and smoking may act as confounders to adverse outcomes in the post-surgical group. Similarly, in our study the post-surgical groups often had a higher mean age than the control non-surgical groups [4, 17, 20].

To maximize the quality of the study, the Cochrane standardized protocol for systematic reviews was followed. The study protocol was published in PROSPERO before the start of the review [10] and a Risk of Bias assessment via the ROBINS-I tool [16] was performed which helped minimize the influence of bias in our study. A thorough literature search was also performed so as to make sure no relevant articles would be missed. In our Risk of Bias assessment, most articles were assessed to have a moderate or serious overall risk of bias. This did not mean that these articles were of low quality, but rather that risk of biases are always a factor to consider with cohort and case-control studies. Due to the nature of our research questions, randomized control trials (RCTs) were not available since no studies were found that compared the interventions solely to compare perinatal outcomes. A meta-analysis was not conducted mainly due to the handful of studies that fit the search criteria and heterogeneity in the comparisons being made by the primary studies. Critical differences, for example, were found for the ‘time from surgery to conception’, which varied greatly between the studies and was in general shorter in women that underwent SG surgery as compared to RYGB. The risk of bias assessment was also considered in this regard, as we anticipated that it would simply compound possible errors and produce results that may be interpreted incorrectly.

BMI at conception were similar in all study groups but due to the confounding effect of surgery to conception interval, we reported separate results for different groups from Karadag et al. [15], as we wanted to highlight the influence of this interval. Another weakness was the differences in total populations in the two intervention groups compared. As Cochrane protocol were strictly followed, the exclusion criteria were decided before performing the database search. This unfortunately resulted in a relatively small selection of SG studies leading to a smaller population of post-SG mothers included. This limited the possibilities to draw conclusions based on the studies included in the review. Similarly, there were differences in time interval from surgery to conception between the different studies. Conception during the malnutrition period after bariatric surgery is known to affect perinatal outcomes and therefore it is not representable to compare groups with greater differences in interval [21].

As mentioned, deviations from the study protocol were also made and an early detection of the very small sample with due to strict surgery to conception interval led to a collective decision to change this exclusion criteria from a maximum of 3 years to a mean smaller than 4 years by the study group. This was decided early on, and no articles were excluded by the 3-year interval criteria before the deviation from the protocol.

There were several aberrations in type of values reported in the different studies. Kjaer et al. [19] did not report a SD or CI for maternal BMI at conception and then used median instead of mean values. The study also reported some outcomes for RYGB and some for RYGB and Gastric banding, resulting in data that could not be added to the current review. Some articles did not report on all outcomes described in the introduction and method. For example, Karadag et al. [15] did not report birthweight even though they reported SGA and LGA which requires birthweight to be calculated.

Systematic reviews are meant to develop and improve patient care and can form the basis for clinical practice. Our goal with this review was to identify which type of bariatric was associated with the least adverse perinatal outcomes. This would have enabled clinicians to provide the best potential care for their patients and give the patients the opportunity to be involved in their course of treatment for the best possible perinatal outcome.

In this systematic review, we demonstrated similar results between pre-pregnancy BMI, mode of delivery, birthweight, gestational age and rates of intrauterine death in both RYGB and SG operated women. Higher rates of SGA and LGA were found in the SG group, however, a possible major confounding factor was the time interval from surgery to conception which was found to be shorter in the SG group. This, together with other confounding factors such as maternal age, parity, smoking and alcohol; may strongly affect the risk of maternal-fetal and neonatal complications. Future studies are indicated to establish which type of bariatric surgery is associated with the least adverse perinatal outcomes.

Below is the link to the electronic supplementary material.

The authors thank Krister Aronsson, who performed the database search and Dr. Matteo Bruschettini for his valuable insight and tips.

Conceptualization: M.Z. Protocol writing: M.Z. Formal analysis: A.K. and A.W, Methodology: M.Z., A.K. and A.W, Project administration: M.Z. Supervision: M.Z. Validation: M.Z. Writing – original draft: A.K. and A.W, Writing – review & editing: M.Z. All authors reviewed the manuscript.

Open Access funding enabled and organized by Projekt DEAL.

The study was financed by grants from the Swedish state under the agreement between the Swedish government and the county councils, the ALF-agreement (YF0054). The funding organization had no role in the design of the study; collection, management, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Open access funding provided by Lund University.

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.