The impact of surgical therapies for inflammatory bowel disease on female fertility

Summary of findings for the main comparison. Any previous surgery compared to no previous surgery for inflammatory bowel disease on female infertility

Risk of infertility and pregnancy outcomes among women who did and did not undergo previous IBD surgery
Patient or population: females with inflammatory bowel disease Setting: outpatient Intervention: previous IBD surgery Comparison: no previous IBD surgery
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№ of women/pregnancies (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with no previous surgery	Risk with any previous IBD surgery
Infertilty at 12 months	Study population		RR 5.45 (0.41 to 72.57)	114 women (2 observational studies)	⊕⊝⊝⊝ Very low^a,b
	47 per 1000	253 per 1000 (19 to 1000)
Infertility at 24 months	Study population		RR 3.59 (1.32 to 9.73)	106 women (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	83 per 1000	297 per 1000 (109 to 805)
Miscarriage	Study population		OR 2.03 (1.14 to 3.60)	776 pregnancies (5 observational studies)	⊕⊝⊝⊝ Very low^a
	69 per 1000^c	130 per 1000 (78 to 210)
Stillbirth	Study population		RR 1.96 (0.42 to 9.18)	246 pregnancies (3 observational studies)	⊕⊝⊝⊝ Very low^a,b
	20 per 1000	39 per 1000 (8 to 184)
Use of ART	Study population		RR 25.09 (1.56 to 403.76)	106 women (1 observational study)	⊕⊝⊝⊝ Very low^a,b	We were unable to calculate absolute effects: 20 out of 66 women with previous surgery used ART compared to 0 out of 40 women with no previous surgery
	See comment	See comment
Caesarean section	Study population		RR 2.23 (1.00 to 4.95)	20 pregnancies (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	308 per 1000	686 per 1000 (308 to 1000)
Preterm delivery	Study population		RR 1.91 (0.67 to 5.48)	194 pregnancies (3 observational studies)	⊕⊝⊝⊝ Very low^a,b
	52 per 1000	98 per 1000 (35 to 282)
Low birth weight	Study population		RR 0.61 (0.08 to 4.83)	19 pregnancies (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	154 per 1000	94 per 1000 (12 to 743)
Small for gestational age	Study population		RR 2.54 (0.80 to 8.01)	65 (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	125 per 1000	318 per 1000 (100 to 1000)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ART: assisted reproductive technology; CI: confidence interval; IBD: inflammatory bowel disease; OR: odds ratio; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded due to high overall risk of bias. ^bDowngraded due to imprecise estimates. ^cRisk of miscarriage in women with IBD who have not had a previous IBD surgery was based on data from 189 pregnancies (3 studies).

Summary of findings 2. Laparoscopic compared to open surgery for inflammatory bowel disease and female infertility

Risk of infertility among women undergoing surgery with a laparoscopic approach compared to an open approach
Patient or population: females with inflammatory bowel disease Setting: outpatient Intervention: laparoscopic surgery Comparison: open surgery
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with open surgery	Risk with laparoscopic surgery
Infertilty at 12 months	Study population		RR 0.70 (0.38 to 1.27)	19 (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	647 per 1000	453 per 1000 (246 to 822)
Infertility at 24 months	These outcomes were not reported
Miscarriage
Stillbirth
Use of ART
Caesarean section
Preterm delivery
Low birth weight
Small for gestational age
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ART: assisted reproductive technology; CI: confidence interval; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded due to high overall risk of bias. ^bDowngraded due to imprecise estimates.

Background

Description of the condition

Inflammatory bowel disease (IBD) is an immune‐mediated disorder characterized by chronic inflammation of the gastrointestinal tract (Baumgart 2007). The main subtypes of IBD are Crohn’s disease (CD), ulcerative colitis (UC), and IBD‐type unclassified (IBDU). UC affects the colonic mucosa, and CD can affect any segment of the gastrointestinal tract with transmural inflammation (Baumgart 2007). Patients with IBDU have colonic IBD and present with characteristics of both CD and UC (Silverberg 2005).

IBD is a remitting and relapsing disease (Rahimi 2008). During a period of active disease, affected individuals can experience symptoms of diarrhea, abdominal pain, nausea, malnutrition, and extreme fatigue (Carter 2004). Although genetic, environmental, and microbial factors all play a role in the pathogenesis of IBD, the exact cause of IBD remains to be determined (Loftus 2004).

Treatment options for IBD include pharmacotherapy, such as corticosteroids, immunomodulators, and biologic agents that alter immune function to minimize inflammatory burden (Kozuch 2008). Surgical intervention may be required when patients do not respond to medical therapy (Loewardi 2003). At present, no cure for IBD is known. The goals of both pharmacologic and surgical treatment are to reduce underlying inflammation and maintain remission.

IBD is most commonly diagnosed between the second and fourth decades of life, during a period of peak reproductive potential for females (Benchimol 2014). Infertility does not appear to differ between women without IBD and those with IBD before undergoing surgery (De Felice 2014). However, available surgical treatments may impact infertility. An increase in infertility is described in the literature among women who undergo surgery for UC. Specifically, colectomy with ileal pouch anal anastomosis (IPAA), a procedure that is commonly performed in people with medically refractory UC (Olsen 2002), was noted to increase infertility (Hudson 1997; Rajaratnam 2011; Waljee 2006). However, the degree to which IPAA may affect infertility remains unclear, and whether other surgical interventions for IBD result in measurable changes in infertility is also unclear.

Description of the intervention

Restorative proctocolectomy with IPAA is the standard surgical treatment for UC (Buskens 2014; Koh 2016), and approximately 15% of patients with UC undergo a colectomy within 10 years of diagnosis (Frolkis 2013). IPAA involves the construction of an internal pouch utilizing a distal portion of the remaining small intestine to create a reservoir for stool (McGuire 2007). These surgeries are frequently performed in multiple stages. An ileal pouch can be made and anastomosed to the anus immediately following proctocolectomy (one‐stage IPAA). Alternatively, the same ileal pouch can be protected from the fecal stream by a loop ileostomy, which is subsequently reversed in a second procedure (two‐stage IPAA). A three‐stage option exists as well, in which a total abdominal colectomy and ileostomy are first performed. This is followed by completion proctectomy with an IPAA and loop ileostomy as the second stage, and loop ileostomy reversal as the third stage. The indication for the colectomy (emergent or elective) often influences the choice of procedure.

Nearly 50% of patients with CD will require an intestinal resection within 10 years of diagnosis, and 25% undergo a second resection within five years of their first surgery (Frolkis 2013; Frolkis 2014). Patients with CD may require surgery for medically refractory disease or because of complications, including abscesses, fistulae, strictures, toxic megacolon, or intestinal perforation. Similar to patients with UC, patients with CD may undergo colectomy but typically would not proceed to an IPAA due to the likelihood of disease recurrence in the pouch resulting in pouch dysfunction. In addition, patients may require a small intestinal resection, most often involving removal of a portion of the ileum with primary anastomosis. Alternatively, patients with narrowed intestinal segment(s) resulting from an accumulation of scar tissue may undergo a strictureplasty. In this procedure, a lengthwise incision is made in the narrowed segment of intestine then is sutured together widthwise. The goals of strictureplasty are to preserve length and to minimize the risk of short bowel syndrome. Patients with perianal CD may undergo surgical procedures such as abscess drainage, seton placement, fistulotomy, fistulectomy, or the creation of an ostomy to divert fecal effluent (de Zoeten 2013).

Surgical procedures used in the management of IBD can be performed as conventional open surgeries or as minimally invasive techniques, such as the multiport laparoscopic approach. Open surgeries require a large incision along the abdomen, providing direct access to the small and/or large bowel. In contrast, a multiport laparoscopic procedure requires multiple small incisions made in the abdomen and insertion of specialized instruments into the openings to perform the operation. Compared to the traditional open approach, the laparoscopic procedure is less invasive and decreases overall mortality, reduces risk of ileus, and results in a faster return of bowel function (Gu 2014).

Patients with IBD not requiring surgery are typically prescribed medication that suppresses the immune system, where the goal of treatment is to heal the mucosal lining of the gastrointestinal tract. Available options include systemic and non‐systemic steroids, immunomodulators (azathioprine or methotrexate), and biologics (infliximab, adalimimab, certolizumab, golimumab, vedolizumab, and ustekinumab). Methotrexate is contraindicated in pregnancy due to its teratogenic effects on a developing fetus. Our review primarily compares infertility and pregnancy‐related outcomes in women undergoing IBD‐related surgery versus those with medically managed disease. We compare these same outcomes in women who have had open and laparoscopic procedures.

How the intervention might work

Surgical effects on infertility

Studies have reported increased infertility among women with IBD who have undergone an IPAA, with infertility rates ranging from 44% to 82% (Rajaratnam 2011; Waljee 2006). Surgical interventions may result in infertility due to scarring of the fallopian tubes. More recent research suggests that a laparoscopic approach may result in decreased rates of infertility when compared to open surgery (Bartels 2012; Beyer‐Berjot 2013).

Previous systematic reviews and meta‐analyses exclusively investigated the effects of IPAA on infertility (Rajaratnam 2011; Waljee 2006). This review provides a systematic evaluation of the effects of multiple surgical treatments for IBD on female infertility for women with both CD and UC. We reviewed colectomy, bowel resection, strictureplasty, and perianal surgery in patients with CD and UC, as well as the impact of open versus laparoscopic surgical techniques.

Why it is important to do this review

Our review expands the scope of previous systematic reviews on the impact of surgery on infertility in women with IBD (Rajaratnam 2011; Waljee 2006). Specifically, we addressed the complete range of surgeries for women with IBD, including both CD and UC, and the implications these different procedures have for infertility and pregnancy‐related outcomes. To adequately counsel women requiring surgical intervention for IBD, it is important to correctly establish the relationship between surgical interventions and risk of infertility. Additionally, knowledge of the effects of surgical techniques on infertility could inform practices and procedures with the goal of preserving fertility and improving outcomes for children born to women who have undergone surgery for IBD. We conducted a systematic review and meta‐analysis to establish whether women with IBD who have undergone surgery have greater risk of infertility and to determine the magnitude of the association and the rates of infertility among women undergoing different surgical procedures. We also characterized the outcomes of pregnancy in these women.

Objectives

Primary objective

To determine the effects of surgical interventions for IBD on female infertility.

Secondary objectives

To evaluate the impact of these surgical interventions on the need for assisted reproductive technology (ART), time to pregnancy, risk of miscarriage, risk of stillbirth, risk of prematurity, mode of delivery (spontaneous vaginal, instrumental vaginal, or Caesarean section), infant requirement for resuscitation and/or neonatal intensive care, low and very low birth weight, small for gestational age, antenatal and postpartum hemorrhage, retained placenta, postpartum depression, gestational diabetes, and gestational hypertension/preeclampsia.

Methods

Criteria for considering studies for this review

Types of studies

Interventional or observational studies (cohort and case‐control) of women of reproductive age (12 years and older) who underwent surgery for IBD were included in the review. Eligible studies provided a comparison of infertility and/or secondary outcomes between IBD patients undergoing different types of surgery or between patients with IBD undergoing surgery and those not undergoing surgery (i.e. treated medically). Studies comparing infertility and/or secondary outcomes in patients before and after they underwent surgery were included. We excluded cross‐over studies, case series, and case reports.

Types of participants

Studies with females of reproductive age with CD, UC, or IBDU were included. Studies comparing patients with IBD to non‐IBD control patients were excluded. In addition, we excluded studies that combined patients with IBD with those undergoing similar surgeries for other indications (e.g. familial adenomatous polyposis).

Types of interventions

For CD, UC, and IBDU, we determined the relative rate of infertility following partial colectomy, total colectomy with or without removal of the rectal stump, and IPAA. In addition, we determined the infertility risk following small bowel resection for CD with or without large bowel resection.

Where possible, we assessed infertility rates through the following comparisons.

Colectomy (laparoscopic) versus colectomy (open) (CD, UC, IBDU).
Small bowel resection versus pharmacologic treatment without surgery (CD only).
Colectomy versus pharmacologic treatment without surgery (CD, UC, IBDU).
Colectomy with IPAA versus colectomy and ostomy without IPAA (CD, UC, IBDU).
IPAA (laparoscopic) versus IPAA (open).
Surgical versus medical management of perianal disease (CD only).
Any type of intestinal resection versus non‐surgical management (CD, UC, IBDU).

Comparisons could reflect the primary objective of the study or a secondary objective. For example, we included studies if they primarily compared women with IBD and those without IBD and secondarily compared women with IBD based on the above comparisons. In addition, we included studies that compared women with IBD before and after any of the above surgical procedures.

Types of outcome measures

Primary outcomes

Primary outcomes included the following:

Risk of infertility, defined as inability to become pregnant after one year of regular unprotected sexual intercourse without the use of birth control; and
Infertility after 6 months, 18 months, and 24 months.

Secondary outcomes

Secondary outcomes included the following:

Risk of miscarriage (spontaneous abortion);
Risk of stillbirth;
Use of ART, including hormonal or mechanical treatments and in vitro fertilization;
Spontaneous vaginal delivery, instrumental vaginal delivery (forceps or vacuum at delivery), delivery by elective Caesarean section, and delivery by emergency Caesarean section;
Prematurity (categorized as birth at < 28 completed weeks' gestation, at 28 to < 32 completed weeks' gestation, at 32 to < 37 completed weeks' gestation);
Infant requirement for resuscitation and/or neonatal intensive care;
Low birth weight (< 2500 g);
Very low birth weight (< 1500 g);
Small for gestational age (birth weight below the 10th percentile for gestational age);
Antenatal hemorrhage;
Postpartum hemorrhage (≥ 500 mL blood loss following a vaginal birth or ≥ 1000 mL blood loss following a Caesarean section);
Retained placenta;
Postpartum depression;
Gestational diabetes;
Gestational hypertension/preeclampsia;
Time to pregnancy; and
Fertility rate (total number of children that would be born to each woman if she were to live to the end of her childbearing years).

All outcomes (primary and secondary), when available, were included in the summary of findings tables.

Search methods for identification of studies

Electronic searches

We searched the following databases from inception to September 27, 2018.

MEDLINE;
Embase + Embase Classic;
Cochrane Central Register of Controlled Trials (CENTRAL), in the Cochrane Library; and
Cochrane IBD Group Specialized Register.

Searching other resources

Additional articles and trials were identified by:

reviewing references cited in the published literature (included studies and relevant review articles);
contacting study investigators to identify any ongoing studies;
searching published abstracts from conferences (American College of Gastroenterology, Digestive Disease Week, United European Gastroenterology Week) for the past five years;
conducting a search of the grey literature including the grey literature databases Open System for Information on Grey Literature (OpenSIGLE), European Association for Grey Literature Exploration (EAGLE), and National Technical Information Service (NTIS) (Higgins 2011); and
searching the clinical trial registries including ClinicalTrials.gov, the World Health Organization (WHO) Trials Portal, and the European Clinical Trials Register (EU‐CTR).

Data collection and analysis

Selection of studies

Two review authors (SL and MC) independently reviewed the titles and abstracts identified by the literature search. For those suspected to meet our inclusion criteria, we conducted a review of the full article to determine eligibility. We resolved disagreements by consensus with a third review author (CHS, EIB, or MEK). We used a PRISMA flowchart to document the process for selection of studies. We included studies published in any language following translation. We used DistillerSR (Evidence Partners, Ottawa, Canada) to facilitate abstract and full‐text reviews.

Data extraction and management

We developed a standardized data extraction form using DistillerSR (Evidence Partners, Ottawa, Canada) to electronically record and organize data as they were extracted. Two review authors (SL and MC) independently extracted data from all included studies. Agreement between extractors was determined, and disagreements were resolved by consensus with CHS, EIB, or MEK. We extracted the following data.

Details of the study including:
- year of publication;
- journal name;
- volume and issue;
- lead author; and
- full title of study.
Study methods including:
- study design ‐ randomized controlled trial (RCT) or observational study (and type, if applicable);
- method used to recruit patients for the study;
- randomization, blinding, and allocation concealment method (if RCT); and
- inclusion and exclusion criteria.
Characteristics of women included in the studies such as:
- number of women in the study;
- age of women at the time of surgery and at the time of the study;
- type of IBD: CD (with or without perianal disease), UC, or IBDU;
- disease activity and study authors’ definition of disease activity, when available; and
- concomitant and past medications.
Details of intervention and comparator arms including:
- type of surgery;
- number of previous surgeries;
- surgical approach used (e.g. open or laparoscopic); and
- approach to disease management when the comparator was not surgical.
Details of infertility and pregnancy‐related outcomes for intervention and comparator groups including:
- definition of infertility used in the study;
- infertility at 6, 12, 18, and 24 months; and
- measures of secondary outcomes including miscarriage, stillbirth, prematurity, mode of delivery, infant requirement for resuscitation and/or neonatal intensive care, low birth weight, very low birth weight, small for gestational age, time to pregnancy, fertility rate, antenatal or postpartum hemorrhage, retained placenta, postpartum depression, gestational diabetes, and gestational hypertension/preeclampsia.

Assessment of risk of bias in included studies

Two review authors (SL and MC) independently assessed the risk of bias of included studies (all observational), using the Newcastle‐Ottawa Scale (Wells 2011). These studies were assessed for the following domains: (1) selection of study participants; (2) comparability (of cases and controls for case‐control studies, and of exposed and non‐exposed for cohort studies); (3) outcome assessment (cohort studies only); and (4) exposure assessment (case‐control studies only). A maximum of one star was awarded per domain for the selection, exposure, and outcome domains, and a maximum of two stars was awarded for the comparability domain. Stars are indicative of low risk of bias (denoted with green circles in our summary figure), whereas no stars indicates that studies are at high risk of bias (denoted with red circles in our summary figure). Studies with unclear risk of bias are denoted with yellow circles in our summary figure.

We planned to assess the risk of bias of randomized controlled trials (RCTs) using the Cochrane "Risk of bias" tool. This tool assesses bias across six domains: (1) selection bias: random sequence generation and allocation concealment; (2) performance bias: blinding of participants and study personnel; (3) detection bias: blinding of outcome assessment; (4) attrition bias: incomplete outcome data; (5) reporting bias: selective reporting of study outcomes; and (6) other sources of bias (Higgins 2011). We would have rated each domain as having low, high, or unclear risk of bias.

We contacted study authors if insufficient information was provided in the publication to assess risk of bias. If study authors were unable to provide data and information was insufficient to assess the quality of a specific domain, we described the quality of this domain as unclear.

We utilized the GRADE approach to assess the overall quality of evidence, which we rated as (1) high quality, (2) moderate quality, (3) low quality, or (4) very low quality. Observational studies began with a low quality rating but could be upgraded based upon three potential factors: (1) a large magnitude of effect, (2) a judgment that all plausible confounding reduced a demonstrated effect or suggested a spurious effect when results showed no effect, or (3) a dose response gradient (Higgins 2011). We conducted GRADE analyses by using GRADEpro software.

Measures of treatment effect

We used Review Manager (RevMan) 5.3 (The Cochrane Collaboration) to conduct the analyses. For continuous outcomes, we would have calculated the mean difference (MD) and the corresponding 95% confidence interval (CI). For dichotomous variables, we calculated the risk ratio (RR) with 95% CI to compare outcomes for each comparison, where possible. When individual studies reported odds ratios (ORs) with corresponding CIs that were adjusted for confounding variables and did not provide raw contingency tables, we pooled ORs instead of RRs using the generic inverse variance method.

Unit of analysis issues

Cluster randomized trials

Cluster randomized trials can result in unit of analysis errors. Had we identified cluster randomized trials for inclusion in our study, we would have completed approximate analysis by calculating effective sample sizes for each trial.

Cross‐over studies

Studies in which participants undergo multiple interventions may result in unit of analysis issues. Given the permanency of surgical intervention, it was not possible to compare infertility in women undergoing surgery for IBD in the context of a cross‐over study. Therefore, we excluded cross‐over studies.

Repeated measures

For studies that reported multiple observations of the same patient (e.g. infertility measured before and after surgical intervention), we provided a qualitative synthesis of findings to minimize bias resulting from unmatched analyses (Higgins 2011).

Dealing with missing data

Missing variance

If the variance of effect estimates for relevant outcomes were not reported in identified studies, we calculated the standard deviation from the standard error, confidence intervals, P values, or t values. When this information was not reported, we contacted study authors to request these statistics. If study authors were unable to provide this information, we imputed the variance, using the technique recommended by Wiebe 2006, based on the availability of variance estimates from other studies.

Missing patient data

In cases where outcomes of some study participants were not reported (i.e. participants were lost to follow‐up), we contacted study authors to obtain this information. If study authors were unable to provide these data, we carried the most recent observation forward (e.g. if a woman was infertile at 12 months, it was assumed that she was also infertile at 18 months). We conducted a sensitivity analysis to compare the results of using available cases and using imputed values by carrying the last observation forward.

Missing outcome data

If identified studies did not report on all outcomes (primary and secondary), we contacted study authors to obtain these data. When these data were not available, we acknowledged this as a limitation of our study and identified it as an area for future research.

Assessment of heterogeneity

We assessed heterogeneity by (1) visually inspecting forest plots; (2) calculating the Chi² statistic with a P value of 0.10 or less indicating statistically significant heterogeneity; and (3) calculating the I² statistic to assess the magnitude of heterogeneity. We interpreted I² values as follows.

0% to 40%: might not be important.
30% to 60%: may represent moderate heterogeneity.
50% to 90%: may represent substantial heterogeneity.
75% to 100%: considerable heterogeneity (Higgins 2011).

Assessment of reporting biases

We did not construct a funnel plot nor test for funnel plot asymmetry as no analyses included more than ten studies (Higgins 2011).

Data synthesis

We pooled data for analysis if interventions, patient groups, and outcomes were sufficiently similar, as determined by consensus between SL, MC, CHS, EIB, and MEK. We did not pool data if there was a high degree of heterogeneity between studies (i.e. I² > 75%). In addition, we did not pool studies if review authors believed there were important clinical or methodological differences across studies. We performed a random‐effects meta‐analysis to account for expected heterogeneity between studies. We conducted all analyses using RevMan 5.3.

Subgroup analysis and investigation of heterogeneity

We conducted the following subgroup analyses.

Type of IBD: CD versus UC versus IBDU.
Age of women at surgery (< 18 years, 18 to 34 years, ≥ 35 years).

We had additionally planned the following subgroup analyses but could not complete them due to lack of data among the included studies.

Open versus laparoscopic surgical approach (when applicable).
Use of adhesion barrier agents.
Left‐sided ulcerative colitis versus extensive colitis versus ulcerative proctitis.
Stricturing CD versus penetrating CD versus non‐stricturing, non‐penetrating CD.

Sensitivity analysis

We conducted a sensitivity analysis separately to analyze population‐based studies versus studies recruiting patients from tertiary care centers. We additionally conducted a post‐hoc sensitivity analysis excluding pregnancy in women with active disease at conception or a diagnosis of IBD during pregnancy or postpartum. We had planned the following sensitivity analyses but could not complete them as no studies met the criteria and/or studies did not report the required data.

Excluding all studies with overall high risk of bias.
Excluding studies published only as an abstract.
Limiting analysis to randomized studies.
Limiting analysis to women who had a viable pregnancy before undergoing surgery for management of their IBD.

Results

Description of studies

Results of the search

Through our search of MEDLINE, Embase, and Cochrane CENTRAL, we identified 1092 records, with an additional three records obtained from review of references of included studies and relevant review articles (Figure 1). After duplicates were removed, 829 abstracts remained and underwent review; 151 full texts were assessed for eligibility. Sixteen observational studies were identified and included in the review. We identified no randomized studies. Ten studies included data on our specified comparisons and were synthesized quantitatively (i.e. were included in outcome‐ and comparison‐specific meta‐analyses). Seven studies compared the risk of infertility and/or secondary outcomes in women before and after surgery and were summarized qualitatively. One included study provided both a comparison of women with IPAA versus women without previous surgery and a comparison of women with IPAA before and after surgery; this study was included in both the meta‐analysis and the qualitative summary. We have described in Table 1 and Table 2, the sample size and outcomes of studies included in the meta‐analysis and in the qualitative synthesis.

Figure 1

Study flow diagram.

Table 1. Sample size and outcomes among studies included in the meta‐analysis

Infertility

Secondary outcomes

Study

Comparison

Type of IBD

Total sample

12 months

24 months

Miscarriage

Stillbirth

Use of ART

Caesarean section

Preterm delivery

Low birth weight

Small for gestational age

Banks 1957

Any previous surgery vs no previous surgery

56 women (87 pregnancies)

87 pregnancies

72 pregnancies

Bartels 2012

Open vs laparoscopic IPAA

37 women

Bortoli 2011

Any previous surgery vs no previous surgery

145 women

Hudson 1997

Any previous surgery vs no previous surgery

86 women^a

86 women

104 women^a

132 pregnancies

94 pregnancies

Johnson 2004

IPAA vs no previous surgery

106 women^a

106 women

Koivusalo 2009

Restorative proctocolectomy with ileoanal anastomosis vs no previous surgery

8 women^a

Moser 2000

Any previous surgery vs no previous surgery

65 women (76 pregnancies)

65 pregnancies^b

Naganuma 2011

Any previous surgery vs no previous surgery

80 women (69 pregnancies^c)

69 pregnancies^d

245 women (234 pregnancies^c)

234 pregnancies^d

Nielsen 1984

Any previous surgery vs no previous surgery

68 women (109 pregnancies)

109 pregnancies

80 pregnancies

Zavorova 2017

Any previous surgery vs no previous surgery

12 women (14 pregnancies)

14 pregnancies

13 pregnancies^e

IPAA vs no previous surgery

5 women (6 pregnancies)

6 pregnancies

^aExcludes women who were voluntarily infertile or were excluded from estimates of infertility for other reasons (e.g. not married or cohabitating).

^bFor women with multiple pregnancies, one was chosen at random.

^cOnly pregnancies occurring on or after the onset of IBD were included.

^dPregnancies among women with CD and UC were analyzed together.

^eBirthweight was missing for one infant

ART: assisted reproductive technology.
CD: Crohn's disease.
IBD: inflammatory bowel disease.
UC: ulcerative colitis.

See: Summary of findings for the main comparison Any previous surgery compared to no previous surgery for inflammatory bowel disease on female infertility; Summary of findings 2 Laparoscopic compared to open surgery for inflammatory bowel disease and female infertility

Table 2. Sample size and outcomes among studies comparing women before and after surgery

				Infertility			Secondary outcomes
Study	Type of surgery	Type of IBD	Total sample	6 months	12 months	24 months	Miscarriage	Stillbirth	Use of ART	Caesarean section	Time to pregnancy
Gorgun 2004	Restorative proctocolectomy with IPAA	UC	185 women		Before: 117 women After: 120 women Both before and after: 52 women				Before: 45 women^a After: 70 women^a
Hahnloser 2004	IPAA	UC	37 women				Before and after: 37 women (37 pregnancies)	Before: 27 women (27 pregnancies) After: 31 women (31 pregnancies)
Johnson 2004	IPAA	UC	153 women		Before: 95 women^a After: 66 women^a				Before: 95 women^a After: 66 women^a
Mortier 2006	Total colectomy with ileorectal anastomosis	UC	37 women	Before: 5 women^a After: 15 women^a	Before: 5 women^a After: 15 women^a	Before: 5 women^a After: 15 women^a					Before: 5 women^a After: 5 women^a
Olsen 2002	Restorative proctocolectomy with IPAA	UC	290 women (331 attempted pregnancies)		Before: 84 attempted pregnancies^a After: 149 attempted pregnancies^a	Before: 84 attempted pregnancies^a After: 149 attempted pregnancies^a
Ravid 2002	IPAA	UC	38 women^b				Before: 12 women (16 pregnancies) After: 38 women (65 pregnancies^c)	Before: 11 pregnancies After: 49 pregnancies^c		Before: 11 pregnancies After: 48 pregnancies (49 deliveries^d)
Tulchinksy 2013	Restorative proctocolectomy	UC	41 women		Before: 26 women^a After: 27 women^a		Before: 26 women (70 pregnancies) After: 17 women (32 pregnancies)		Before: 26 women (70 pregnancies)^a After: 27 women (32 pregnancies)^a	Before: 62 pregnancies After: 26 pregnancies	Before: 26 women (70 pregnancies) After: 17 women (32 pregnancies)

^aUnclear how many women were at risk of developing the outcome both before and after surgery.

^bAll women included in the study gave birth at least once after IPAA.

^cOngoing pregnancies excluded at the time of the study.

^dA set of twins included (one born vaginally and one born via Caesarean section).

ART: assisted reproductive technology.
CD: Crohn's disease.
IBD: inflammatory bowel disease.
IPAA: ileal pouch anal anastomosis.
UC: ulcerative colitis.

Included studies

Infertility

Meta‐analysis

Three studies were included in a meta‐analysis evaluating the impact of surgical procedures on female infertility in women with IBD at 12 months; two compared patients with and without prior surgery (Johnson 2004; Koivusalo 2009), and one compared laparoscopic and open IPAA (Bartels 2012). One study evaluated infertility at 24 months, comparing patients with and without prior surgery (Hudson 1997).

Qualitative summary

Five studies comparing female infertility before and after surgery at 12 months (Gorgun 2004; Johnson 2004; Mortier 2006; Olsen 2002; Tulchinksy 2013), one study comparing infertility before and after surgery at six months (Mortier 2006), and two studies on infertility at 24 months were summarized qualitatively (Mortier 2006; Olsen 2002). All studies included patients with UC only.

Pregnancy‐related secondary outcomes

Meta‐analysis

Eight studies were included in meta‐analyses evaluating the association between surgical procedures and the following secondary outcomes (Banks 1957; Bortoli 2011; Hudson 1997; Johnson 2004; Moser 2000; Naganuma 2011; Nielsen 1984; Zavorova 2017):

Miscarriage was evaluated in five studies. Three studies included patients with CD (Bortoli 2011; Hudson 1997; Nielsen 1984), one study included patients with UC (Banks 1957), and one study grouped patients with all types of IBD (Naganuma 2011).
Stillbirth was evaluated in three studies. Two studies included patients with CD (Hudson 1997; Nielsen 1984), and the third included patients with UC (Banks 1957).
Use of ART was evaluated in a single study of patients with UC (Johnson 2004).
One study included delivery via Caesarean section as an outcome, reporting findings separately for patients with CD and UC (Zavorova 2017).
Preterm birth was an outcome in three studies. Two studies included only patients with CD (Hudson 1997; Nielsen 1984), and one study included both patients with CD and patients with UC (Zavorova 2017).
One study evaluated low birth weight, reporting CD‐ and UC‐specific associations (Zavorova 2017).
One study, including only patients with CD, included small for gestational age (≤ 10th percentile) as an outcome (Moser 2000).

Qualitative summary

Five studies compared secondary pregnancy‐related outcomes before and after surgery (Gorgun 2004; Hahnloser 2004; Mortier 2006; Ravid 2002; Tulchinksy 2013), and all studies included only women with UC. These studies evaluated the following outcomes:

Miscarriage (Hahnloser 2004; Ravid 2002; Tulchinksy 2013);
Stillbirth (Hahnloser 2004; Ravid 2002);
Use of ART (Gorgun 2004; Johnson 2004; Tulchinksy 2013);
Caesarean section (Ravid 2002; Tulchinksy 2013);
Gestational diabetes (Hahnloser 2004);
Preeclampsia (Hahnloser 2004); and
Time to pregnancy (Mortier 2006; Tulchinksy 2013).

Descriptions of included studies

Banks 1957

This retrospective cohort study compared miscarriage and stillbirth in women with definitive previous and no definitive previous surgery for UC. For this study, medical charts of patients admitted for UC at Beth Israel Hospital in Boston, Massachussetts, USA, between 1931 and 1950 were reviewed. UC was broadly defined as colitis, UC, chronic colitis, superficial colitis, or hemorrhagic colitis. Definite surgery included ileostomy, ileostomy and subtotal colectomy, subtotal colectomy, total colectomy, abdominal‐perineal resection, colostomy, colostomy and subtotal colectomy, colostomy and local resection, local resection with end‐to‐end anastomosis, unconventional procedures, explorations, and abdominal‐perineal and ileostomy. Data stratified by type of surgery were not reported. A total of 245 patients with UC were included in the study, with one (0.03%) lost to follow‐up. Patients with obvious irritable colon syndrome or with inadequate data to support a diagnosis of non‐specific UC and ileocolitis were excluded (n = 13). Data were collected via personal interviews, patients' physician notes, and hospital records and through correspondence with internists or surgeons. Average patient follow‐up was 12.1 years; 73% were followed for ≥ 11 years after disease onset. Data on miscarriage and stillbirth were available for 56 women with 87 pregnancies. Among women included in this study, 28 women with 46 pregnancies were in remission at the time of conception, 14 women with 23 pregnancies had active disease at the time of conception, five women were diagnosed during pregnancy, four women were diagnosed in the postpartum period, and five women (nine pregnancies) had previous surgery. Subsequent pregnancies occurring among women diagnosed during pregnancy or in the postpartum period were excluded due to lack of detailed information about the outcomes of these pregnancies. One pregnancy occurring among women in remission and eight pregnancies among women with active disease at conception ended in therapeutic abortion; no therapeutic abortions were reported among women diagnosed with UC during pregnancy or in the postpartum period. Among 51 women without previous surgery (78 pregnancies), three (3.8%) pregnancies resulted in miscarriage. Among 46 pregnancies in 28 women with inactive disease at conception and no prior surgery for UC, there were two (4.3%) miscarriages. Of the remaining 43 pregnancies, two (4.5%) resulted in stillbirth. Five women (nine pregnancies) underwent prior surgery for UC; two (22.2%) had miscarriages and none had a stillbirth (0%).

Bartels 2012

This cross‐sectional study compared the risk of infertility in women with IPAA by a laparoscopic approach versus women with IPAA by an open approach. IPAA was conducted in a single‐stage or two‐stage procedure with and without the creation of loop ileostomy, with ovaries left unimpaired. Two‐stage IPAA consisted of subtotal colectomy followed by complete proctectomy and creation of IPAA. This study was conducted at three university hospitals in The Netherlands and Belgium between February 2010 and February 2011 and included women who underwent IPAA between 1993 and 2009. These patients were identified from hospital databases. Of 179 patients who were eligible for the study, five (2.8%) refused consent and 14 (7.8%) did not respond. An additional 110 patients (61.5%) were excluded because they did not wish to have a child after IPAA. The study included 50 female patients ≥ 18 years of age at the time of the study who had IPAA before 41 years of age (open: n = 23; laparoscopic: n = 27). Indications for IPAA surgery included UC (n = 37), familial adenomatous polyposis (n = 12), and others (n = 1). Subgroup analyses specific to women with UC were described (open: n = 20; laparoscopic: n = 17). Data were collected via a self‐reported questionnaire that was developed and mailed out to patients, with a reminder sent after three weeks. The questionnaire included questions on medical history, desire for children, infertility, obstetric history before and after IPAA, smoking at the time of conception attempts, history of pelvic inflammatory disease, and history of extrauterine gravidity. Patients with incomplete or unclear answers were contacted. Answers were then verified by a check of available medical records. Average follow‐up was 9.4 years (standard deviation [SD] = 5.0 years). Infertility was defined as not becoming pregnant within 12 months of trying to conceive. Infertility at 12 months was reported in nine (45%) women whose surgery was performed via a laparoscopic approach and in 11 (65%) women whose surgery was performed through an open approach.

Bortoli 2011

This prospective cohort study compared women with previous surgery for IBD versus those with no previous surgery for IBD. The surgeries these women underwent were not described. This study was conducted at 68 centers in 12 European countries between 2003 and 2006. The primary objective of this study was to compare women with IBD to those without IBD. Women with IBD who did and did not have prior surgery were also compared. Of 520 pregnant women with IBD enrolled consecutively (CD: n = 244; UC: n = 264; IBDU: n = 12), 332 were matched to non‐IBD pregnant controls (CD: n = 145; UC: n = 187). Women with IBD were excluded if they could not be matched to a woman without IBD (n = 147), if their matched control had missing data (n = 32), or if they were diagnosed with IBDU (n = 9). Each participating center received electronic case report forms for trained physicians to record the requested data prospectively at enrollment and every three months until the end of pregnancy. Information was collected by in‐person or telephone interview and by review of patients' medical records. All completed forms were sent electronically to the central database storage. Requested data included maternal variables, treatment for IBD, disease activity, pregnancy outcomes, and mode of delivery. The association between prior surgery and miscarriage was provided only for women with CD. Study authors also described the association between previous surgery and Caesarean section and preterm delivery (< 37 weeks' gestation) in women with CD and the association between previous surgery and Caesarean section in women with UC. However, we did not include these outcomes, as all pregnancies (including those ending in spontaneous and therapeutic abortion) were included as the denominator, and the true association between surgery and these outcomes could not be ascertained. As a result, we included only women with CD in our meta‐analysis. There were 47 (32.4%) pregnant women with CD who had a previous IBD‐related surgery, and there was no association between miscarriage and previous surgery among women with CD (OR 1.30, 95% CI 0.21 to 7.90).

Gorgun 2004

This cross‐sectional study compared infertility and use of ART before and after restorative proctocolectomy with IPAA in women with UC and familial adenomatous polyposis. Eligible study participants were identified from a prospectively maintained pelvic pouch database from a single center between 1983 and 2001. Of 500 women undergoing pelvic pouch operation between 15 and 44 years of age and eligible for the study, 300 (60.0%) responded to the mailed questionnaire. Of the 300 respondents, 206 (68.7%) attempted to conceive before and/or after restorative proctocolectomy with IPAA (before: n = 127; after: n = 135; before and after: n = 56). The study included 185 women with UC. Of these women, 117 attempted to conceive before surgery and 120 after surgery; 52 attempted to conceive both before and after pregnancy. Female patients self‐reported information on infertility before and/or after surgery. Additional information was obtained from the pelvic pouch database, including patients' demographic and disease history and technical characteristics of the surgery. Infertility was defined as 12 months of unprotected intercourse without conception among women trying to conceive. There were 45 (38.5%) women with UC who were infertile before restorative proctocolectomy with IPAA and 70 (58.3%) women with UC who were infertile after restorative proctocolectomy with IPAA. In a subgroup analysis of women trying to conceive both before and after restorative proctocolectomy, 24 of 52 (46.2%) were infertile before restorative proctocolectomy with IPAA and 36/52 (69.2%) were infertile afterward. Of 45 women who were infertile before restorative proctocolectomy with IPAA, 19 (42.2%) women used ART, whereas 36 of 70 (51.4%) women who were infertile after restorative proctocolectomy used ART.

Hahnloser 2004

This cross‐sectional study compared outcomes of pregnancy in women before and after IPAA. IPAA involved either excising the anal transition zone and hand‐sewing the pouch to the anal canal or preserving the anal transition zone and double‐stapling the pouch of the anal canal. All patients had a diverting ileostomy constructed, which was closed two to three months after IPAA construction. Eligible patients were identified from a prospectively maintained database of all individuals undergoing IPAA between 1981 and 1995. A total of 544 female patients underwent IPAA for UC before 40 years of age and underwent colectomy, complete mobilization of the small bowel mesentery, and proximal rectal mobilization via a close rectal dissection technique. All eligible patients identified in a prospective hospital database as having undergone IPAA for UC were sent a standardized questionnaire, which included information on pregnancy outcomes and mode of delivery before and after IPAA. The response rate was 82.7% (450/544), and the mean follow‐up after IPAA was 13 years. Outcomes included miscarriage, stillbirth, gestational diabetes, and preeclampsia. Of the 450 respondents, 141 were pregnant (236 pregnancies) after diagnosis with UC but before IPAA and 135 (30.0%) were pregnant (232 pregnancies) after IPAA. Eight (1.8%) women were excluded from the analyses because the time of pregnancy was not reported. Data on outcomes of pregnancy occurring before IPAA were reported with the number of women as the denominator, and the number of pregnancies was the denominator for outcomes of pregnancies occurring after IPAA. Of the 141 women who were pregnant before IPAA, 27 (19.1%) had a miscarriage, four (2.8%) had a stillbirth, 16 (11.3%) had preeclampsia, and 14 (10.6%) had gestational diabetes. Of the 232 pregnancies occurring after IPAA, 31 (14.4%) were miscarriages and two (0.9%) were stillbirths. Information on complications of pregnancy was missing for 16 of 232 pregnancies occurring after IPAA. Of the 216 pregnancies with data on complications, mothers developed gestational diabetes in five (2.3%) and preeclampsia in ten (4.6%). A total of 52 women were pregnant before and after IPAA, of whom 37 (71.2%) had data available for both time periods. Data are included for the first pregnancy following IPAA. Before IPAA, ten women (27.0%) had a miscarriage, two (5.4%) had a stillbirth, three (8.1%) had gestational diabetes, and two (5.4%) had preeclampsia. Of the same 37 female patients with UC who were pregnant after IPAA, six (16.2%) had a miscarriage, two (5.4%) had a stillbirth, and no women had gestational diabetes or preeclampsia. This study also reported information on mode of delivery; however, these results were not presented as miscarriages, and stillbirths were combined with the denominator for mode of delivery, resulting in biased estimates of the impact of surgery on the need for Caesarean section.

Hudson 1997

This cross‐sectional study compared infertility and pregnancy‐related outcomes in women with IBD with and without prior surgical intervention. Detailed information about the type(s) of surgical procedures and the approach to medical management was not provided. This study was conducted at one large teaching hospital (i.e. a single referral center for gastroenterology, obstetrics, and pathology) in the North East of Scotland, including the Orkeny and Shetland Isles, with a stable population of 600,000 over the course of the study. Between 1967 and 1986, 466 women between 16 and 45 years of age had an established diagnosis of CD or UC, based on accepted clinical, radiologic, endoscopic, and/or histologic criteria, who resided in the Grampian Region for six months or longer before the onset of symptoms. A four‐page anonymous questionnaire was sent with a prepaid return envelope to eligible women, which included information on menstruation, past and current contraceptive status, desire for pregnancy, date of each delivery, pregnancy outcomes, and any conception difficulty. Additional information was obtained through hospital, medical, or obstetric records and, if necessary, from general practitioners. Of 466 eligible women with CD or UC, 409 (87.8%; CD = 177, UC = 232) responded to the questionnaire. Infertility was defined as failing to achieve a desired spontaneous conception after 24 months of unprotected intercourse. Study authors also included data on miscarriage (spontaneous abortion), stillbirth, and premature delivery among women with CD. No information on pregnancy‐related outcomes among women with UC was presented. Of 261 (63.8%; CD n = 123, UC n = 138) women who responded and were potentially fertile, involuntary infertility was analyzed in 86 (69.9%) women with CD and 104 (75.4%) women with UC. The remaining women were voluntarily infertile. Of those with CD, 46 (53.5%) underwent previous surgical intervention and 40 (46.5%) had disease that was managed only medically. Among women with UC, 11 (10.6%) had prior surgery and UC was managed medically in the remaining 93 (89.4%) women attempting to conceive. Infertility at 24 months was observed in 7 of 46 (15.2%) females with CD who had prior surgery, 3 of 40 (7.5%) females with medically managed CD, 5 of 11 (45.5%) females with UC who had prior surgery, and 8 of 93 (8.6%) females with medically managed UC. A total of 132 pregnancies were observed among women with CD; 79 of these pregnancies occurred in women whose disease was managed medically and 53 pregnancies occurred in women with previous IBD‐related surgery. Evidence of recurrent disease based on endoscopy and/or radiology was noted at the time of conception in 14 (26.4%) pregnancies among women with previous surgery. Of the 39 cases that occurred after surgical intervention without objective evidence of disease recurrence, nine (23.1%) resulted in a miscarriage (spontaneous abortion) compared to 9 of 79 (23.1%) women with CD whose disease had been managed only medically and 5 of 14 (36%) with previous surgery and evidence of disease recurrence. Nine therapeutic abortions were reported among women with no previous surgery, four therapeutic abortions among those with past surgery but without evidence of disease recurrence, and two therapeutic abortions among those with previous surgery and recurrent disease. Among the remaining pregnancies, one stillbirth was observed. This occurred in a woman with previous IBD surgery and evidence of recurrent disease. Among pregnancies not ending in miscarriage or therapeutic abortion (n = 94), premature delivery was observed in 2 of 61 (3.3%) pregnancies among women without previous IBD surgery, 1 of 26 (3.8%) pregnancies among women with previous IBD surgery without disease recurrence, and one of seven (14.3%) pregnancies in women with previous surgery and recurrent CD.

Johnson 2004

This cross‐sectional study compared the risk of infertility and the need for ART among women with UC who had an IPAA versus medically managed UC. Eligible patients were identified from a prospectively maintained IBD database at Mount Sinai Hospital, in Toronto, Canada. Details of IPAA and the approach to medical management in these patients were not described. There were 560 female patients with UC who were registered in the IBD database (IPAA: n = 323; medical treatment: n = 237). Information in the IBD database at Mount Sinai Hospital was prospectively collected and entered. Identified female patients were sent questionnaires by mail and were called, to encourage a reply or to clarify any missing data. A total of 397 (70.9%) women completed and returned the questionnaire, including 254 with IPAA and 143 with medically managed disease. A total of 213 (IPAA: n = 153; medical treatment: n = 60) women met the eligibility criteria: ≥ 18 years of age at time of study, < 44 years of age at time of IPAA or diagnosis of UC, and married or cohabitating for at least 12 months after surgery or after diagnosis. Patients with incomplete data regarding relationship history and reproductive history were excluded. The mean (SD) duration of follow‐up after IPAA was 11.6 (4) years, and the mean (SD) duration of follow‐up after diagnosis for women with medically managed UC was 15.6 (10.7) years. Infertility was defined as failure to become pregnant during 12 months of unprotected intercourse. Among the 153 women with an IPAA included in the study, 95 attempted to become pregnant before surgery and 66 attempted to become pregnant after surgery. Among women with an IPAA, infertility was observed in 4 of 95 (4.2%) before surgery and in 29 of 66 (43.9%) after surgery. Among the 40 women who did not have surgery, 1 of 40 (2.5%) was infertile. ART was used in 5 of 95 (5.3%) women who would later have an IPAA, 20 of 66 (30.3%) women after undergoing IPAA, and 0 of 40 (0%) women without prior surgery for UC.

Koivusalo 2009

This cross‐sectional study compared infertility among women with UC who underwent restorative proctocolectomy with IPAA and those without prior surgery. Women in the group with restorative proctocolectomy with ileoanal anastomosis had J‐pouch or straight ileoanal anastomosis. Indications for surgery were UC recalcitrant to medical treatment and fulminant colitis. In the group receiving medical management, patients were on 5‐ASA and azathioprine, 5‐ASA, mercaptopurine, prednisone, or no medication. This study included patients treated at two major university hospitals (i.e. Helsinki and Tampere) in Finland between 1985 and 2005. Data were collected by self‐reported questionnaire during a follow‐up outpatient visit. The questionnaire collected information on sexual function, bowel function, stool frequency, fecal incontinence, and medication use. Out of 81 eligible patients with prior surgery, 79 (97.5%) were traced and invited for a follow‐up outpatient visit and 36 (45.6%) attended. This included 25 (69.4%; 17 female) with pediatric‐onset UC (diagnosed < 16 years of age) who underwent restorative proctocolectomy with ileoanal anastomosis, were sexually active, and were > 16 years of age at the time of the study. These patients were compared to 38 (19 female) controls who were sexually active, were medically treated for UC, and were > 16 years of age at the time of the study. Infertility was defined as not becoming pregnant within 12 months after trying to conceive. Among 17 women who underwent restorative proctocolectomy with ileoanal anastomosis, five (29.4%) attempted pregnancy and infertility was observed in three (60.0%) women. Among 19 women who were medically treated for UC, three attempted pregnancy, and infertility was observed in one of three (33.3%) women.

Mortier 2006

This cross‐sectional study compared infertility in women with UC before and after total colectomy with ileorectal anastomosis. Details of the surgical procedure were not provided. Data were collected between May and September 2003. Eligible patients were treated at a single center (Department of Abdominal Surgery, Centre Hospitalier Régional Universitaire de Lille, in Lille, France) between 1962 and 1999. Data were collected via structured interviews conducted via telephone by instructed and trained female interviewers. The same set of questions was asked to gather information about three time periods: before diagnosis of UC, from diagnosis of UC to ileorectal anastomosis, and from ileorectal anastomosis to the date of the interview. No patient was lost to follow‐up, and the median follow‐up was 16 years (range 4 to 35 years). A total of 37 female patients underwent ileorectal anastomosis for UC. Patients were excluded if they died from postoperative complications or alcoholic cirrhosis (n = 2), had a confirmed diagnosis of CD (n = 5), or underwent proctectomy for rectal cancer (n = 1). Infertility was defined as the proportion of those not pregnant with unprotected intercourse at 6, 12, and 24 months. Five (13.5%) women desired to become pregnant before surgery, and 15 (40.5%) women attempted pregnancy afterward. Infertility at six months was observed in one of five (20.0%) females before colectomy with ileorectal anastomosis and in 9 of 15 (60.0%) females afterward. Infertility at 12 months was observed in one of five (20.0%) females before total colectomy with ileorectal anastomosis and in 8 of 15 (53.3%) females after total colectomy with ileorectal anastomosis. Infertility at 24 months was observed in one of five (20.0%) women before surgery and in 7 of 15 (46.7%) women afterward. Before surgery, the median (range) time to pregnancy was two (one to three) months compared to five (2 to 36) months after surgery.

Moser 2000

This retrospective cohort study evaluated the risk of having a baby that was small for gestational age among women with CD who did and did not have a prior bowel resection. The details of surgical intervention were not provided. Patients whose medical charts included a code for pregnancy and IBD or CD between January 1993 and December 1997 were examined for eligibility. This study was conducted in Edmonton, Canada. Retrospective chart reviews were conducted by reviewing the data on prenatal sheets, the discharge data summary, and the neonate's history and physical examination findings. Women were eligible for the study if there was documentation in their charts of pregnancy and IBD, CD, or gastrointestinal consultation (n = 91). Patients were excluded if they were admitted for missed therapeutic abortion, abortion, or spontaneous abortion before 20 weeks (n = 23) or had active disease at the time of conception (n = 5). The study included 65 women with CD whose pregnancy had gone beyond 32 weeks (76 pregnancies). Only one pregnancy was selected at random for women with multiple pregnancies in the study period. Thus, 65 women with CD with 65 pregnancies were matched to healthy controls selected by a computer from the Northern Alberta Perinatal Outreach Program database by age, parity, smoking status, approximate date of delivery, and gestational age. The primary objective of this study was to compare women with CD versus healthy controls. However, study authors assessed predictors of giving birth to an infant who was small for gestational age, defined as < 10th percentile, among women with CD. Of 65 included pregnancies, 41 (63.1%) occurred after previous bowel resection and 24 (36.9%) pregnancies occurred in women with no previous bowel resection. Infants were small for gestational age in 13 (31.7%) and three (12.5%) pregnancies that occurred in women with and without previous bowel resection, respectively. With adjustments for smoking during pregnancy, ileal disease location, and poor weight gain (< 15 pounds throughout pregnancy or < 10 pounds by the 28th week of gestation), the OR (95% CI) for the association was 4.16 (0.92 to 18.92).

Naganuma 2011

This cross‐sectional study compared several pregnancy‐related outcomes in women with and without previous surgery for IBD. Details of the surgical procedures these women underwent were not provided. Study participants were women with IBD treated at six large IBD centers in Japan who conceived between 1989 and 2008. A questionnaire was given to participants that collected information on type of disease, marital status, number of children, smoking habits, age at each conception, history of surgery for IBD, history of surgery for perianal lesions, abscesses and fissuring ulcerations in CD patients, history of therapy for infertility, outcomes of conceptions, mode of delivery, birth weight, and presence of congenital abnormalities. Female patients, when pregnant, were given a notebook to report information regarding conception and outcomes of pregnancy. Information on the course of pregnancy and its outcomes (including birth weight) was completed by physicians or nurses. Medical records were reviewed to confirm the diagnosis of disease, continuation or discontinuation of medications at the time of conception, outcomes, mode of delivery, and congenital malformation. Of 589 patients enrolled in the study, 11 (1.9%) refused to participate and six (1.0%) patients with indeterminate colitis were excluded. In total, there were 325 female patients (CD: n = 80; UC: n = 245) between the ages of 17 and 46 years who experienced conception at least once and were eligible to participate in the study. A total of 534 conceptions were reported among women participating in the study, of which 303 (CD: n = 69; UC: n = 234) occurred during or after IBD onset. Miscarriage, Caesarean section, and low birth weight were included as outcomes. However, we included only miscarriage data, as the true association between previous surgery and other outcomes could not be determined because study authors included pregnancies ending in spontaneous and therapeutic abortion among those at risk. Among 303 conceptions occurring during or after IBD onset, 60 (19.8%; CD: n = 44; UC: n = 16) occurred in women with previous surgery. CD and UC were combined for all analyses reported in this paper. Previous surgery was not associated with miscarriage in pregnant women with IBD (OR 1.4, 95% CI 0.71 to 2.90).

Nielsen 1984

This retrospective cohort study evaluated the association between pregnancy outcomes in women with CD who did and did not have a previous intestinal resection. Details of the surgical intervention were not provided. This study included patients treated in outpatient clinics or admitted to hospital at a single center in Copenhagen, Denmark, between January 1968 and December 1980. Information was obtained through a review of medical records, and women with insufficient data were contacted by telephone. Participants included 68 women (149 pregnancies) with CD under the age of 37 years at the time of conception. Of 149 pregnancies, 40 (36.7%) occurred before the beginning of the study period or ended six months before the onset of disease and were excluded. Of 68 women, 51 (75.0%) had previous bowel resection. These 51 women had a total of 77 pregnancies, and 17 women without prior surgery had 32 pregnancies. Of the 109 pregnancies, 57 occurred among women with quiescent disease at conception, 43 occurred among women with active disease at conception, three occurred among women who were diagnosed during pregnancy, and six occurred in women who were diagnosed postpartum. Based on the information provided, we were not able to exclude patients with active disease at conception. Eleven and eight therapeutic abortions were reported among women with and without previous surgery, respectively. With exclusion of pregnancies ending in miscarriage or therapeutic abortion, three (5.3%) pregnancies in women with previous bowel resection resulted in stillbirth compared to one (4.3%) pregnancy in women without prior surgery. There were 13 (22.8%) pregnancies resulting in preterm birth among women with previous surgery compared to two (8.7%) in women with no prior bowel resection. Nine (11.7%) miscarriages and one (3.1%) miscarriage occurred in women with and without prior surgery, respectively.

Olsen 2002

This cross‐sectional study compared infertility before and after restorative proctocolectomy with IPAA. Details of the surgical procedure were not provided. Eligible patients were treated at four centers in Denmark and Sweden between November 1982 and January 1998; they were contacted by letter to obtain consent or refusal. Non‐responders whose telephone numbers could be found in the directory were contacted by telephone. Data were collected by telephone interview via a highly structured interview form conducted thoroughly by trained female interviewers. Information about the following time periods was collected during the interview: before the diagnosis of UC, from diagnosis until colectomy, and from stoma closure until the date of the interview (after IPAA). Participants were asked once about reproductive history, starting and ending dates, duration of the first interval of unprotected intercourse in each time window, intent to become pregnant or pregnancy planning, and exposures around the starting date of this first‐time interval. Study participants were 343 women with UC who underwent restorative proctocolectomy with IPAA; they were ≤ 40 years old at the time of stoma closure and > 18 years of age at the time of the interview. Of 343 women with UC who were eligible for the study, 290 (84.5%) agreed to participate and provided information. These women had a total of 331 attempted pregnancies (before diagnosis: n = 98, before colectomy: n = 84, after IPAA: n = 149). Infertility was defined as the inverse of the probability of becoming pregnant within 12 and 24 months of unprotected intercourse. Infertility at 12 months was observed in 18 of 84 (21.4%) attempted pregnancies before colectomy and in 122 of 149 (81.9%) attempted pregnancies after IPAA. Infertility at 24 months was observed in 13 of 84 (15.5%) attempted pregnancies before surgery and in 108 of 149 (72.5%) attempted pregnancies after surgery.

Ravid 2002

This cross‐sectional study compared infertility and outcomes of pregnancy (miscarriage and Caesarean section) before and after IPAA. A total of 26 female patients had a stapled J‐pouch and 12 female patients had an S‐pouch. In all, 18 female patients had a three‐stage operation, 18 female patients had a two‐stage operation, and two female patients had a one‐stage operation. Seventeen had a hand‐sewn ileoanal anastomosis, and the anastomosis was stapled in 21. There were nine patients without loop ileostomy. Eligible patients underwent IPAA at the Inflammatory Bowel Disease Unit of Mount Sinai Hospital, in Toronto, Ontario, Canada, between 1982 and 1998. Questionnaires were sent to eligible patients to collect information on infertility, pregnancy, type of delivery, birth weight of the baby, and complications of pregnancy and pouch. Female patients who had become pregnant after IPAA construction and gave birth at least once after IPAA were further contacted by telephone. Pouch function before, during, and after pregnancy was assessed with questions related to stool frequency, incontinence during the day and night, pad usage, skin irritation, and medication usage. Of a total of 1088 IPAAs performed, 481 (44.2%) female patients were identified and 400 (83.2%) female patients responded to the questionnaire, including 38 (9.5%) who became pregnant a total of 67 times after IPAA. At the time of the study, 2 of the 67 pregnancies occurring after IPAA were ongoing; these pregnancies were excluded from the analysis. Of the 38 women who gave birth after IPAA, 12 (31.6%) had also given birth before IPAA (16 pregnancies). One of 16 (6.3%) pregnancies resulted in a miscarriage before IPAA, and 9 of 65 (13.8%) pregnancies resulted in a miscarriage after IPAA. Four therapeutic abortions were performed before IPAA and seven therapeutic abortions after IPAA. Before IPAA, none of the remaining 11 pregnancies resulted in a stillbirth or were delivered by Caesarean section. After IPAA, one of the 49 (2.0%) remaining pregnancies resulted in a stillbirth; 24 of 49 (49.0%) were delivered by Caesarean section.

Tulchinksy 2013

This cross‐sectional cohort study compared infertility, use of ART, time to pregnancy, and pregnancy‐related outcomes before and after restorative proctocolectomy. Details of the surgical interventions were not provided. Study participants were identified from a prospectively collected database of patients from a comprehensive pouch clinic at a tertiary referral center in Israel. Participants were contacted during visits to the clinic or by telephone. Eligible participants filled out a questionnaire on infertility, pregnancy, delivery, newborn status, issues on time to conceive, need for infertility therapies, medical treatment before and during pregnancy, miscarriages, duration of pregnancy, and pregnancy‐related complications. Additional information, such as patient demographics, past medical history, and history of operations, was collected from the pouch clinic database and during patient visits. Among 55 women with UC who underwent restorative proctocolectomy before the age of 45 and who were ≥ 21 years of age at the time of the study, 47 (85.5%) met the inclusion criteria and 41 (87.2%) returned the questionnaire. Before surgery, 26 women had 70 pregnancies. Following restorative proctocolectomy, 27 women attempted pregnancy; 32 pregnancies were reported among 17 women. Infertility was defined as the inability to achieve conception after one year of unprotected sex with a fertile male partner. None of the 26 women were infertile before surgery, and 10 of 27 (37.0%) were infertile afterward. Among 11 women attempting pregnancy both before and after surgery, none were infertile before surgery, and six (54.5%) were infertile after surgery. Eight of 70 (11.4%) pregnancies resulted in a miscarriage before surgery compared to 6 of 32 (18.8%) pregnancies after surgery. A total of 6 of 70 (8.6%) pregnancies used ART before restorative proctocolectomy and 11 of 32 (34.4%) afterward. Eight of 62 (12.9%) pregnancies were delivered by Caesarean section before restorative proctocolectomy compared to 12 of 26 (46.2%) delivered by Caesarean section after restorative proctocolectomy. The mean (SD) time to pregnancy was 5.0 (11.6) months before surgery and 16.3 (25.1) months afterward.

Zavorova 2017

This retrospective cohort study compared pregnancy‐related outcomes in women with and without previous surgery for IBD. Surgery for women with UC included IPAA. Surgery for women with CD included a mix of colectomy with end ileostomy, ileocecal resection with anastomosis, and other procedures. Medical treatment in 10 female patients with CD included mesalazine (n = 2), a combination of mesalazine and budesonide (n = 1), azathioprine (n = 3), azathioprine and adalimumab (n = 1), infliximab (n = 1), and adalimumab (n = 2). This study was conducted in the Department of Obstetrics and Gynecology at the Hospital Horovice in the Czech Republic, between October 2014 and March 2016. Data on infertility, therapy during pregnancy, and mode of delivery were obtained from the hospital database. Loss to follow‐up was not reported. Participants included 17 female patients (CD: n = 12; UC: n = 5). Seven women with CD who had no prior surgery gave birth nine times. Of these infants, two (22.2%) were born by Caesarean section, one (11.1%) had low birth weight, and none were born prematurely (< 37 weeks' gestation). Five women with previous surgery for CD delivered five babies, four of whom were delivered by Caesarean section. None of the infants born to women with CD who had prior surgery were low birth weight or were born prematurely. The birth weight of one infant born to a mother with CD who had previous surgery was missing. Three women with UC with no previous surgery had four deliveries, and two women with UC who had prior IPAA delivered two babies. Half (2/4) of the deliveries in women with UC without IPAA and both (2/2) of the deliveries in women with UC and an IPAA were performed via Caesarean section. One (25.0%) low birth weight baby was born and one (25.0%) infant was born prematurely to a mother without prior surgery for UC, compared to no low birth weight or premature births among women with UC who had an IPAA.

Excluded studies

Of the 151 studies identified for full‐text review, 135 were excluded (see Characteristics of excluded studies). Excluded studies were review articles and studies that did not include infertility or one of our secondary outcomes, were not limited to women with IBD, and did not include an eligible comparison group (Figure 1).

Risk of bias in included studies

The risk of bias in the 16 included observational studies is summarized in Figure 2. Overall, we found high risk of selection bias in terms of the representativeness of the exposed cohort and selection of the non‐exposed cohort, comparability of cohorts, and outcome assessment. Only one study was at low risk of selection bias based on the representativeness of the cohort and selection of the non‐exposed cohort (Hudson 1997); the risk of bias due to selection of exposed and non‐exposed individuals was unclear in one study (Hahnloser 2004). Three studies were at high risk of bias based on methods used to ascertain surgical status (Bortoli 2011; Hudson 1997; Olsen 2002), and the method used to determine surgical status was unclear in two studies (Johnson 2004; Koivusalo 2009). The outcome of interest was not present at the start of each study, resulting in low risk of selection bias for this domain. Only three studies presented adjusted analyses and were at low risk of bias for this domain (Bortoli 2011; Moser 2000; Naganuma 2011). In five studies (Bartels 2012; Moser 2000; Naganuma 2011; Nielsen 1984; Zavorova 2017), outcomes were obtained from medical records or from an existing database; therefore, these studies were at low risk of bias for outcome assessment. The remaining studies were at high risk of bias because outcomes were ascertained from patient self‐report. Generally, follow‐up was long enough for outcomes to occur, but risk of bias due to length of follow‐up was unclear in two studies (Koivusalo 2009; Zavorova 2017). Four studies had high risk of bias due to poor response rates (Gorgun 2004; Johnson 2004; Koivusalo 2009; Olsen 2002), and the risk of bias for this domain was unclear in one study (Zavorova 2017).

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Effects of interventions

Any previous surgery vs no previous surgery

Infertility at 12 months

Two studies compared the risk of infertility at 12 months in women with UC who did and did not have previous surgery (see Analysis 1.1; Johnson 2004; Koivusalo 2009) (RR 5.45, 95% CI 0.41 to 72.57; 114 participants). Heterogeneity between studies was substantial (I² = 74%). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias and imprecise estimates. Johnson 2004 compared women with previous IPAA to those with no previous surgery, and Koivusalo 2009 compared women with restorative proctocolectomy with ileoanal anastomosis to those with no previous surgery.

Infertility at 24 months

The risk of infertility at 24 months among women with and without previous surgery for IBD was compared in a single study, which reported findings separately for CD and UC (see Analysis 1.2; Hudson 1997). The following associations between previous surgery and 24‐month infertility were observed:

CD: RR 2.03, 95% CI 0.56 to 7.33, 86 participants;
UC: RR 5.28, 95% CI 2.09 to 13.34, 104 participants; and
Combining women with CD and UC: RR 3.59, 95% CI 1.32 to 9.73, 190 participants.

There was no notable heterogeneity (I² = 39%). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias and imprecise estimates. Hudson 1997 compared women with and without previous IBD‐related surgery but provided no specific details on the types of surgical procedures performed.

Miscarriage

Five studies compared the association between previous surgery for IBD and miscarriage (see Analysis 1.3; Banks 1957; Bortoli 2011; Hudson 1997; Naganuma 2011; Nielsen 1984) (OR 2.03, 95% CI 1.14 to 3.60; 776 pregnancies; I² = 0%). Three studies evaluated this association among pregnancies in women with CD (OR 2.56, 95% CI 1.19 to 5.51; 386 pregnancies), and one study evaluated this association among pregnancies in women with UC (OR 7.14, 95% CI 1.02 to 50.18; 87 pregnancies). Findings were similar in a single population‐based study (OR 2.79, 95% CI 1.11 to 7.04; 132 pregnancies) and in four studies recruiting patients from tertiary care centers (OR 1.84, 95% CI 0.84 to 4.03; 644 pregnancies) (see Analysis 1.4). Results were also similar in a sensitivity analysis of two studies excluding patients with active disease at the time of conception (RR 2.38, 95% CI 1.11 to 5.11; 173 pregnancies; I² = 0%) (see Analysis 1.5; Banks 1957; Hudson 1997). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias. All studies compared women with and without previous IBD‐related surgery, and none provided detailed information on the types of surgical procedures the women required.

Stillbirth

Three studies evaluated the association between previous surgery and stillbirth in women with IBD (RR 1.96, 95% CI 0.42 to 9.18; 246 pregnancies; I² = 0%), two evaluating the risk in women with CD (RR 1.98, 95% CI 0.32 to 12.16; 174 pregnancies; I² = 0%) and the other in women with UC (RR 1.91, 95% CI 0.10 to 36.02; 72 pregnancies; I² = 0%) (see Analysis 1.6; Hudson 1997; Nielsen 1984; Banks 1957). One study was population based (Hudson 1997), and the other two included women treated at tertiary care centers (Banks 1957; Nielsen 1984), reporting the following results (see Analysis 1.7):

Population‐based: RR 5.47, 95% CI 0.23 to 130.65; 94 pregnancies; and
Tertiary care: RR 1.43, 95% CI 0.24 to 8.35; 152 pregnancies; I² = 0%.

Two studies were included in a sensitivity analysis excluding pregnancies in women with active disease at conception (see Analysis 1.8; Banks 1957; Hudson 1997). No stillbirths were reported in Hudson 1997 among women in remission at conception, so no estimate of the association between previous surgery and stillbirth could be obtained from this study. In Banks 1957, the association between previous surgery and stillbirth in women with inactive UC at conception was RR 1.26 (95% CI 0.07 to 23.54; 72 pregnancies). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias and imprecise estimates. Studies compared women with and without previous IBD‐related surgery but provided no specific details on the types of surgical procedures the women required.

Use of ART

The association between previous surgery and use of ART was evaluated in a single study of 106 women with UC (see Analysis 1.9): RR 25.09, 95% CI 1.56 to 403.76 (Johnson 2004). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias and imprecise estimates.

Caesarean section

The association between previous surgery and delivery via Caesarean section among women with IBD was reported in a single study, with stratification by type of IBD (see Analysis 1.10; Zavorova 2017):

CD: RR 3.60, 95% CI 0.98 to 13.19; 14 pregnancies;
UC: RR 1.67, 95% CI 0.61 to 4.59; 6 pregnancies; and
Combining CD and UC: RR 2.23, 95% CI 1.00 to 4.95; 20 pregnancies; I² = 0%.

Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias. All women with UC in Zavorova 2017 requiring surgery had IPAA and were compared to those without previous surgery; the women with CD underwent a variety of procedures.

Prematurity

Three studies determined the association between preterm birth among women with and without previous surgery (RR 1.91, 95% CI 0.67 to 5.48; 194 pregnancies; I² = 0%) (see Analysis 1.11; Hudson 1997; Nielsen 1984; Zavorova 2017). All three studies evaluated this association in women with CD (RR 2.32, 95% CI 0.75 to 7.21; 188 pregnancies; I² = 0%), and one study also evaluated this association in women with UC (RR 0.56, 95% CI 0.03 to 9.73; 6 pregnancies). Findings were similar in a single population‐based study (RR 1.85, 95% CI 0.27 to 12.53; 94 pregnancies) and in two studies recruiting patients from tertiary care centers (RR 2.02, 95% CI 0.56 to 7.28; 100 pregnancies) (see Analysis 1.12). One study was included in a sensitivity analysis limited to women in remission at the time of conception (RR 1.17, 95% CI 0.11 to 12.38; 87 pregnancies) (see Analysis 1.13; Hudson 1997). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias and imprecise estimates. Bortoli 2011, Hudson 1997, and Nielsen 1984 all compared women with and without previous IBD‐related surgery without providing specific details of the surgical procedures. All women with UC in Zavorova 2017 requiring surgery had IPAA and were compared to those without previous surgery; the women with CD had a variety of surgical procedures.

Low birth weight

The association between previous IBD surgery and giving birth to a low birth weight infant was evaluated in a single study (Zavorova 2017), which reported the association in women with CD (RR 0.67, 95% CI 0.03 to 13.60; 13 pregnancies) and UC (RR 0.56, 95% CI 0.03 to 9.73; 6 pregnancies). When CD was combined with UC, this association was RR 0.61 (95% CI 0.08 to 4.83; 19 pregnancies) (see Analysis 1.14). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias and imprecise estimates. All women with UC requiring surgery had IPAA and were compared to those without previous surgery; the women with CD underwent a variety of procedures.

Very low birth weight

One study allowed for the association between surgery and giving birth to a very low birth weight baby (Zavorova 2017). Of 14 pregnancies in women with CD and six pregnancies in women with UC, none resulted in the birth of a very low birth weight infant.

Small for gestational age

The association between previous surgery and giving birth to a small for gestational age infant was evaluated in a single study that included only women with CD: RR 2.54, 95% CI 0.80 to 8.01; 65 pregnancies (see Analysis 1.15; Moser 2000). The association was similar when adjusting for poor weight gain during pregnancy, ileal disease location, and tobacco use during pregnancy (OR 4.16, 95% CI 0.92 to 18.92). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias and imprecise estimates. No details were provided about the types of surgery women required.

Laparoscopic vs open surgery

Infertility at 12 months

One study compared a laparoscopic approach and an open approach to restorative proctocolectomy with regard to female infertility at 12 months in 37 women with UC: RR 0.70, 95% CI 0.38 to 1.27 (see Analysis 2.1; Bartels 2012). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias and imprecise estimates.

Before and after surgery

Infertility at 12 months

Five studies assessed infertility at 12 months before and after surgery (see Table 3; Gorgun 2004; Johnson 2004; Mortier 2006; Olsen 2002; Tulchinksy 2013). All studies were limited to women with UC. Surgeries included restorative proctocolectomy with IPAA, restorative proctocolectomy, and total colectomy with ileorectal anastomosis. Before surgery, 68 of 327 (20.8%) females reported that they were infertile following 12 months of unprotected intercourse. Following surgery, 239 of 377 (63.4%) females were infertile after 12 months. Gorgun 2004 additionally reported on a subgroup of 52 women with UC who attempted pregnancy both before and after IPAA; 24 of 52 (46.2%) women were infertile before surgery and 36 of 52 (69.2%) were infertile after surgery. This study also reported infertility before and after surgery stratified by age at surgery. Age‐specific findings are summarized in Table 4. Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias.

Table 3. Infertility at 12 months in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Infertile, n (%)	Total, n	Infertile, n (%)	Total, n
Restorative proctocolectomy with IPAA	Gorgun 2004	45 (38.5)	117	70 (58.3)	120
Restorative proctocolectomy with IPAA	Olsen 2002	18 (21.4)	84	122 (81.9)	149
IPAA	Johnson 2004	4 (4.2)	95	29 (43.9)	66
Total colectomy with ileorectal anastomosis	Mortier 2006	1 (20.0)	5	8 (53.3)	15
Restorative proctocolectomy	Tulchinksy 2013	0 (0.0)	26	10 (37.0)	27
Total		68 (20.8)	327	239 (63.4)	377

IPAA: ileal pouch anal anastomosis.

Table 4. Infertility at 12 months in female patients with UC before and after surgery as reported in Gorgun 2004

	Before surgery		After surgery
Age at surgery	Infertile, n (%)	Total, n	Infertile, n (%)	Total, n
≤ 19 years	0 (0)	1	4 (50)	8
20 to 34 years	38 (40.9)	93	61 (57.0)	107
≥ 35 years	7 (30.4)	23	5 (100)	5

UC: ulcerative colitis.

Infertility at six months

One study assessed infertility at six months before and after colectomy with ileorectal anastomosis (Mortier 2006). Among five women with attempted pregnancy before surgery, one (20.0%) was infertile. Infertility was reported by 9 of 15 (60%) of women attempting pregnancy after surgery. Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias.

Infertility at 24 months

Two studies observed infertility at 24 months before and after surgery in female patients with UC (see Table 5; Mortier 2006; Olsen 2002). Overall, 14 of 89 (15.7%) females with UC were infertile before surgery and 116 of 164 (70.7%) were infertile after surgery. Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias. Surgeries included restorative proctocolectomy with IPAA and total colectomy with ileorectal anastomosis.

Table 5. Infertility at 24 months in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Infertile, n (%)	Total, n	Infertile, n (%)	Total, n
Restorative proctocolectomy with IPAA	Olsen 2002	13 (15.5)	84	108 (72.5)	149
Restorative proctocolectomy	Mortier 2006	1 (20.0)	5	7 (46.7)	15
Total		14 (15.7)	89	115 (70.1)	164

IPAA: ileal pouch anal anastomosis.
UC: ulcerative colitis.

Miscarriage

Three studies observed miscarriage before and after surgery in females with UC (see Table 6; Hahnloser 2004; Ravid 2002; Tulchinksy 2013). There were 19 of 123 (15.4%) and 21 of 134 (15.7%) pregnancies before and after surgery, respectively, that resulted in miscarriage. Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias.

Table 6. Miscarriage in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Miscarriage, n (%)	Total, n	Miscarriage, n (%)	Total, n
IPAA	Hahnloser 2004	10 (27.0)	37	6 (16.2)	37
IPAA	Ravid 2002	1 (6.3)	16	9 (13.8)	65
Restorative proctocolectomy	Tulchinksy 2013	8 (11.4)	70	6 (18.8)	32
Total		19 (15.4)	123	21 (15.7)	134

IPAA: ileal pouch anal anastomosis.
UC: ulcerative colitis.

Stillbirth

Two studies examined stillbirth in women with UC before and after IPAA (see Table 7; Hahnloser 2004; Ravid 2002). Before surgery, 2 of 38 (5.3%) resulted in a stillbirth compared to 3 of 80 (3.8%) pregnancies occurring after surgery. Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias.

Table 7. Stillbirth in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Stillbirth, n (%)	Total, n	Stillbirth, n (%)	Total, n
IPAA	Hahnloser 2004	2 (7.4)	27	2 (6.5)	31
Restorative proctocolectomy	Ravid 2002	0 (0)	11	1 (2.0)	49
Total		2 (5.3)	38	3 (3.8)	80

IPAA: ileal pouch anal anastomosis.
UC: ulcerative colitis.

Use of ART

Three studies observed use of ART before and after surgery in female patients with UC (see Table 8; Gorgun 2004; Johnson 2004; Tulchinksy 2013). Due to differences in the denominator for each study, results are not combined. Gorgun 2004 reported that 19 of 45 (42.2%) women who were infertile at 12 months used ART before restorative proctocolectomy with IPAA and 36 of 70 (51.4%) used ART after surgery. Johnson 2004 reported the proportion of women using ART among all women attempting pregnancy, finding that 5 of 95 (5.3%) used ART before IPAA compared with 20 of 66 (30.3%) after IPAA. Tulchinksy 2013 reported that ART was used in 6 of 70 (8.6%) pregnancies occurring before restorative proctocolectomy compared with 11 of 32 (34.3%) pregnancies after surgery. Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias.

Table 8. Use of ART in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Use of ART, n (%)	Total, n	Use of ART, n (%)	Total, n
IPAA	Johnson 2004	5 (5.3)	95^a	20 (30.3)	66^a
Restorative proctocolectomy with IPAA	Gorgun 2004	19 (42.2)	45^b	36 (51.4)	70^b
Restorative proctocolectomy	Tulchinksy 2013	6 (8.6)	70^c	11 (34.3)	32^c

^aNumbers of women attempting pregnancy before and after surgery.

^bNumber of women infertile following 12 months of attempting pregnancy.

^cNumber of pregnancies in which women used ART.

Note: Totals not calculated due to differing denominators among included studies.

ART: assisted reproductive technology.
IPAA: ileal pouch anal anastomosis.
UC: ulcerative colitis.

Caesarean section

Two studies described delivery by Caesarean section before and after surgery in females with UC (Ravid 2002; Tulchinksy 2013). Eight of 73 (11.0%) deliveries occurred by Caesarean section before surgery and 36 of 75 (48.0%) deliveries occurred by Caesarean section after surgery (see Table 9). Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias. Surgeries included IPAA and restorative proctocolectomy.

Table 9. Caesarean section in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Caesarean section, n (%)	Total, n	Caesarean section, n (%)	Total, n
IPAA	Ravid 2002	0 (0)	11	24 (49.0)	49^a
Restorative proctocolectomy	Tulchinksy 2013	8 (12.9)	62	12 (46.2)	26
Total		8 (11.0)	73	36 (48.0)	75

^aIncludes a set of twins (one born vaginally and one born via Caesarean section).

IPAA: ileal pouch anal anastomosis.
UC: ulcerative colitis.

Gestational diabetes

One study observed gestational diabetes before and after surgery in female patients with UC (Hahnloser 2004). Among the 37 women with pregnancies before and after IPAA, three (8.1%) women developed gestational diabetes before surgery and no women experienced gestational diabetes in a pregnancy after surgery. Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias.

Preeclampsia

One study observed preeclampsia before and after IPAA among women with UC (Hahnloser 2004). Of the 37 women with pregnancies before and after IPAA, two (5.4%) had preeclampsia before surgery compared to 0 after surgery. Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias.

Time to pregnancy

Two studies observed time to pregnancy before and after surgery among women with UC (see Table 10; Mortier 2006; Tulchinksy 2013). Mortier 2006 reported that the median (range) time to pregnancy among five women attempting pregnancy before colectomy with ileorectal anastomosis was two months (one to three) compared to five months (2 to 36) among 15 women attempting pregnancy after surgery. Tulchinksy 2013 reported that the mean (SD) time to pregnancy before restorative proctocolectomy was 5.0 (11.6) months among 26 women attempting pregnancy, with conception reported as occurring "immediately" in 32 pregnancies. Among 27 women attempting to conceive after surgery, the mean (SD) time to conception was 16.3 (25.1) months. Based on GRADE analyses, the overall certainty of evidence was downgraded from low (observational study) to very low due to high overall risk of bias.

Table 10. Time to pregnancy in female patients with UC before and after surgery, measured in months

			Before surgery		After surgery
Type of surgery	Study	Measure	Time to pregnancy	Number of women attempting pregnancy	Time to pregnancy	Number
Total colectomy with ileorectal anastomosis	Mortier 2006	Median (range)	2 (1‐3)	5	5 (2‐36)	15
Restorative proctocolectomy	Tulchinksy 2013	Mean (SD)	5.0 (11.6)	26 (70 pregnancies^a)	16.3 (25.1)	27^b (32 pregnancies)

^aPregnancy occurred "immediately" in 32 pregnancies.

^b17/27 women successfully conceived.

SD: standard deviation.
UC: ulcerative colitis.

Discussion

Summary of main results

Based on available data, we are uncertain about the association between previous IBD‐related surgery and risk of infertility in women with IBD due to very low‐quality evidence. Data on the association between previous surgery and infertility at 12 months were limited to women with UC, while data on infertility at 24 months were available for both CD and UC. It is uncertain if there are any differences in infertility among women with UC who underwent surgery via a laparoscopic or open approach. Further, although no associations were observed between IBD‐related surgery and risk of stillbirth, preterm delivery, or delivery of a low birth weight or small for gestational age infant, and associations were observed with miscarriage, use of ART, and Caesarean section delivery, the evidence supporting these associations is uncertain. Our conclusions are based on very low‐quality evidence due to the observational nature of the data, the limited number of studies, the small sample sizes of the studies, and high risk of bias. Thus, our findings should be interpreted with caution.

Overall completeness and applicability of evidence

We lacked data on several of the comparisons and outcomes we sought to investigate. Specifically, we were able to present data only from studies comparing colectomies conducted via open and laparoscopic approaches and from studies comparing women who had previous surgery versus those who did not. Although some studies defined the types of surgical procedures that women underwent, no two studies evaluating infertility and providing specific details of the type of surgical procedure included women who had the same surgical procedure. We additionally reported data on studies comparing infertility and pregnancy‐related outcomes in women before and after colectomy and/or IPAA. However, as outlined below, we are concerned about the validity of these studies. We had initially aimed to describe the impact of specific surgeries (i.e. IPAA, small bowel resection, perianal surgery) on infertility and secondary outcomes, but many of the included studies grouped multiple surgery types into a single comparison and did not provide any details of the type(s) of surgery the women included in their studies required. Given that variation across types of surgery is likely, the applicability of our findings for women considering an IBD‐related intestinal resection is limited. For example, the creation of an IPAA is a pelvic surgery that is more likely to result in damage to the fallopian tubes when compared with more limited intestinal resections (e.g. small bowel resection) (Asztély 1991; Oresland 1994). Tubal factors were found to be a more common reason for infertility among women with an IPAA compared to women without an IPAA and may be less likely to result from other procedures (Pabby 2015). Further, variations may exist across types of pouches and techniques used for their construction. Some studies included women with J, S, or Kock pouches. One small study reported a numeric increase in adhesions among women who underwent three‐stage IPAA construction compared to women who had a two‐stage procedure, which may subsequently impact the risk of infertility (Hull 2012). Another study reported numeric increases in infertility among those with a stapled anastomosis compared to those with a hand‐sewn anastomosis (Harnoy 2016).

In addition to missing data on several specific comparisons, we did not identify any studies evaluating several of our specified secondary outcomes, including requirement for instrumental delivery, infant resuscitation, antenatal hemorrhage, postpartum hemorrhage, retained placenta, postpartum depression, and fertility rate. Thus, we are unable to draw conclusions with regard to how these pregnancy outcomes may be impacted by previous IBD‐related surgery.

Quality of the evidence

All studies were identified as having high risk of bias for at least one domain. All but one of the included studies recruited patients from tertiary care centers and were at risk of selection bias (Hudson 1997). For example, at smaller hospitals, less experienced surgeons may perform surgery in patients with IBD, which may result in higher infertility rates. Alternatively, patients treated at tertiary care centers may have more severe or more complicated disease, resulting in more complicated surgery. However, results of sensitivity analyses (when possible) demonstrated that results were similar in population‐based studies and in studies recruiting patients from tertiary care centers. Further, most studies did not adjust for important confounding factors. Several studies relied on self‐reported data for outcome assessment and/or exposure ascertainment. The use of self‐reported data could result in reporting biases, whereby women with previous IBD‐related surgery may be more likely to recall having trouble becoming pregnant or experiencing a negative outcome of pregnancy. Additionally, these women may be more likely to notice an early miscarriage compared to women without prior surgery. Most studies included participants for a sufficient length of time for outcomes to develop, and studies had high response and retention rates.

The presence of selection bias and lack of control for important confounding factors (e.g. maternal age, disease activity or severity, corticosteroid use, presence of perianal disease) prevented us from drawing conclusions about the impact of surgery on infertility and pregnancy outcomes in women with IBD. Specifically, active IBD and corticosteroid use during pregnancy are associated with negative outcomes, including giving birth to a premature or low birth weight infant and miscarriage, and methotrexate is teratogenic (Boyd 2015; Bröms 2014; Mahadevan 2007; Mahadevan 2017; Nguyen 2016). Although some studies presented adjusted results, one study was limited to patients in remission at conception and adjusted for ileal disease location (Moser 2000). Two additional studies presented data stratified by disease activity at conception, enabling a sensitivity analysis of pregnancies occurring among women in remission at conception. None of the studies evaluating infertility accounted for maternal age. Fertility declines with age and failure to account for maternal age particularly impacts studies comparing infertility in women before and after undergoing surgery (Crawford 2015). Additionally, delivery via Caesarean section is contraindicated in women with active perianal disease (Nguyen 2016); thus, findings may have been confounded by disease phenotype. Further, studies did not provide information on the time between surgery and attempts at pregnancy, which may also have confounded the associations investigated in this review. Last, several studies included multiple pregnancies per mother. Thus, information on several pregnancy outcomes may be biased on the basis that women who are more likely to achieve and sustain a healthy pregnancy are likely over represented. Further, the precision of the estimate is biased by failing to account for clustering of data within mothers (Vittinghoff 2005).

In addition to several concerns about the risk of bias among individual studies, we are concerned that our findings are exclusively drawn from observational studies. However, a randomized controlled clinical trial could not be ethically implemented to evaluate this research question. Further, most analyses were based on five or fewer studies, and these studies had small sample sizes, resulting in imprecise estimates. Additionally, we noted high heterogeneity for some comparisons. As a result of high risk of bias among included studies, as well as high levels of imprecision, all conclusions made in this review are based on very low‐quality evidence and, as such, should be interpreted with caution.

Potential biases in the review process

We systematically reviewed several databases, conference abstracts, and references of identified review articles and included studies to identify all studies comparing the risk of infertility and several pregnancy‐related outcomes in women with IBD who did and did not have previous IBD‐related surgery. All screening of abstracts, reviews of full texts, and data extraction were done in duplicate by two independent review authors. Thus, bias in identifying studies was minimized. However, bias may have resulted from incomplete outcome reporting. For example, we were unable to identify relevant studies for several of our outcomes. Further, some outcomes were included in the identified studies, but these studies provided insufficient information to allow for inclusion in the systematic review. Last, we may have introduced bias through our calculation of the absolute risk of some secondary outcomes. Specifically, for studies reporting only an adjusted odds ratio and corresponding confidence intervals, numerator and denominator data used to calculate the absolute risk for each intervention arm were not available. Instead, these data were calculated from the data available in studies included in the review that provided contingency tables for relevant associations.

Agreements and disagreements with other studies or reviews

Infertility

Two previous systematic reviews specifically evaluated the impact of IPAA on infertility. One was limited to patients with UC (Waljee 2006), and the other included patients with UC and familial adenomatous polyposis (Rajaratnam 2011). Both concluded that infertility occurred more often in women with an IPAA compared to women without an IPAA. Our review differed from these previous reviews in that we included only studies that used a rigorous definition of infertility (i.e. an inability to become pregnant within a specified length of time such as 12 months of regular unprotected sexual intercourse without the use of birth control). This was an important distinction, as not using such a definition may result in biased conclusions, particularly if women without previous IBD‐related surgery are followed longer than patients with previous IBD‐related surgery. For example, if two women received the diagnosis of IBD at 23 years of age (one without surgery and one with surgery at 26 years of age) and were followed until age 30, the woman without previous surgery would have seven years of follow‐up data, whereas the woman with previous surgery would have only four years of follow‐up data. With longer follow‐up, the first woman is more likely to conceive, regardless of previous surgery. Consequently, lower infertility rates among patients without previous surgery may result from a longer period during which to conceive. Further, women with IBD who undergo surgery may be systematically different from those who do not undergo surgery. They may have more severe disease, long‐term disability, or systemic effects of chronic inflammation or steroid use. Therefore, they may be less likely to want to become pregnant, resulting in perceived higher infertility rates.

Another difference from other systematic reviews is that we did not meta‐analyze studies assessing infertility in women before and after IPAA construction. These studies were not pooled due to concerns about the statistical validity of failing to account for repeated observations in the analysis, as well as the clinical factors that may affect women’s choice to become pregnant, as described above. Studies comparing infertility before and after IPAA construction were biased due to confounding by age. Specifically, women were older in the postoperative period, which may have also contributed to postoperative infertility. Age‐specific estimates of infertility before and after IPAA were reported in one study (Gorgun 2004). No differences in infertility were reported among women undergoing surgery before the age of 30, and rates of infertility were higher following IPAA after the age of 30 ‐ an age after which fertility begins to decline more rapidly than at previous ages (Crawford 2015), likely biasing study authors’ overall findings.

Because of our decision not to pool studies comparing infertility before and after surgery, we were not able to meta‐analyze any studies evaluating the impact of IPAA on infertility. Although studies included in our qualitative summary did report an increased proportion of women being infertile following IPAA and ileorectal anastomosis procedures (Gorgun 2004; Mortier 2006; Olsen 2002), we are unable to make any statistical inferences about the impact of IPAA on infertility in women with IBD given the concerns with available data described above.

Definition of infertility

We specifically defined infertility as an inability to become pregnant after 12 months of unprotected intercourse and included additional time points (i.e. 6, 18, and 24 months) as secondary outcomes. Thus, we included only studies that reported the proportion of women who were infertile among those who were attempting to become pregnant. Other studies have used alternative definitions of infertility (e.g. live births, success of in vitro fertilization) or have included women who were married or cohabitating, regardless of whether or not they were trying to conceive. Although our definition of infertility was aimed at estimating the risk of infertility among women attempting to conceive, some studies excluded from this review have taken a broader approach to defining infertility, grouping voluntary infertility (i.e. choosing not to get pregnant) with involuntary infertility. That is, these studies have mixed women who choose not to become pregnant for disease‐related or other reasons with those who are unable to become pregnant despite trying to conceive. Both approaches may lead to biased conclusions with regard to the impact of IBD surgeries on female infertility. Further, estimates of infertility can vary drastically across definitions, and the choice of definition can lead to different conclusions when infertility is compared across groups of women (Jacobson 2017).

A prior systematic review identified rates of voluntary infertility among women with IBD ranging from 17% to 38% (Purewal 2018). In a survey of women of childbearing age from the United Kingdom, women who were voluntarily childless had more prior surgeries than women who had children following their IBD diagnosis (Selinger 2016). Voluntary childlessness among women with IBD is related to poorer pregnancy‐related IBD knowledge, as well as anxiety with regard to the heritability of IBD, the ability to raise a child, and the impact of pregnancy on IBD disease activity. It is unclear whether voluntary childlessness in women with IBD, particularly those with prior surgery, results from disease‐ and surgery‐related factors, lack of knowledge, or other psychosocial factors.

Sexual dysfunction may also contribute to infertility among women with IBD. Prior studies have suggested that sexual dysfunction is more common in women with IBD than in the general population and may be further increased in women who have had previous IBD surgery, although study findings have not been consistent (Eluri 2018; Sanders 2016). Thus, this impaired sexual function may also contribute to involuntary infertility, and studies evaluating infertility in women with IBD must consider the relative contributions of voluntary and involuntary infertility.

Comparison to other medical conditions

A single study suggests that IPAA increases infertility in women with familial adenomatous polyposis (Olsen 2003). However, this study also compared women before and after surgery and is consequently subject to the same residual confounding resulting from increased age in the postoperative period that resulted in bias in some research involving women with IBD.

Two other studies evaluated the impact of IPAA on infertility by comparing women with IBD who had undergone the surgical procedure to women without IBD who underwent appendectomy. One study was limited to patients with ulcerative colitis and noted higher rates of infertility among women with an IPAA compared to those who underwent appendectomy (Lepisto 2007). The other study included women with IBD and women with familial adenomatous polyposis and found no difference in infertility among the two groups of women (Beyer‐Berjot 2013). However, small sample sizes and choice of study design (e.g. definition of infertility) may have resulted in biased results. Notably, Lepisto 2007 defined infertility at 24 months using the same definition that we required for inclusion in our study, and Beyer‐Berjot 2013 did not define infertility within a specific time frame. Further, it is unclear if the reduction in infertility observed in these studies was the result of IPAA or IBD, as infertility is more common among women with IBD than in the general population (Tavernier 2013).

Mode of delivery

Based on very low‐quality evidence, our review suggests that women with prior IBD‐related surgery may be more likely to give birth via Caesarean section compared to women for whom IBD had been managed medically. However, insufficient information is presented among the included studies to determine if this association reflects physician and patient preference or obstetric indications for Caesarean section. Among the studies describing mode of delivery, three studies described reasons for Caesarean section (Hahnloser 2004; Ravid 2002; Zavorova 2017). In Zavorova 2017, Caesarean sections were planned for all women with surgically managed IBD (both CD and UC), whereas Caesarean section was performed only in the presence of obstetric complications among women who had not previously had IBD surgery. In the other two studies, less than a quarter of all deliveries were performed via Caesarean section specifically because the mother had an IPAA, with remaining Caesarean sections occurring for obstetric reasons or because of patient preference (Hahnloser 2004; Ravid 2002). Insufficient information was provided in studies to determine if active perianal disease contributed to decisions with regard to mode of delivery among included studies. A survey of Canadian gastroenterologists reported very little consensus regarding the impact of mode of delivery among women with an IPAA (Kuenzig 2018). More recent clinical practice guidelines indicate that Caesarean section should be recommended for women with IPAA due to concerns about pouch function and sphincter injury after delivery (Nguyen 2016), but this recommendation is based on limited data. Women with active perianal disease are also recommended to deliver via Caesarean section. For all other women with IBD, regardless of surgical history, vaginal delivery is recommended unless there are obstetric indications for Caesarean section (Nguyen 2016). Increasing evidence points to the safety of vaginal delivery for these women, despite earlier concerns about vaginal delivery increasing the risk of or worsening perianal disease (Foulon 2017).

Figure 1

Study flow diagram.

Figure 2

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Analysis 1.1

Comparison 1 Any surgery versus no surgery, Outcome 1 Infertiltiy at 12 Months.

Analysis 1.2

Comparison 1 Any surgery versus no surgery, Outcome 2 Infertility at 24 Months.

Analysis 1.3

Comparison 1 Any surgery versus no surgery, Outcome 3 Miscarriage.

Analysis 1.4

Comparison 1 Any surgery versus no surgery, Outcome 4 Miscarriage ‐ Sensitivity Analysis Based on Source of Participants.

Analysis 1.5

Comparison 1 Any surgery versus no surgery, Outcome 5 Miscarriage ‐ Sensitivity Analysis Including Only Women in Remisson at Conception.

Analysis 1.6

Comparison 1 Any surgery versus no surgery, Outcome 6 Stillbirth.

Analysis 1.7

Comparison 1 Any surgery versus no surgery, Outcome 7 Stillbirth ‐ Sensitivity Analysis Based on Source of Participants.

Analysis 1.8

Comparison 1 Any surgery versus no surgery, Outcome 8 Stillbirth ‐ Sensitivity Analysis Including Only Women in Remission at Conception.

Analysis 1.9

Comparison 1 Any surgery versus no surgery, Outcome 9 Use of ART.

Analysis 1.10

Comparison 1 Any surgery versus no surgery, Outcome 10 Caesarean Section.

Analysis 1.11

Comparison 1 Any surgery versus no surgery, Outcome 11 Preterm Delivery.

Analysis 1.12

Comparison 1 Any surgery versus no surgery, Outcome 12 Preterm Delivery ‐ Sensitivity Analysis Based on Source of Participants.

Analysis 1.13

Comparison 1 Any surgery versus no surgery, Outcome 13 Preterm Delivery ‐ Sensitivity Analysis Including Only Women in Remisson at Conception.

Analysis 1.14

Comparison 1 Any surgery versus no surgery, Outcome 14 Low Birth Weight.

Analysis 1.15

Comparison 1 Any surgery versus no surgery, Outcome 15 Small for Gestational Age.

Analysis 2.1

Comparison 2 Laparoscopic versus open surgery, Outcome 1 Infertiltiy at 12 Months.

Summary of findings for the main comparison. Any previous surgery compared to no previous surgery for inflammatory bowel disease on female infertility

Risk of infertility and pregnancy outcomes among women who did and did not undergo previous IBD surgery
Patient or population: females with inflammatory bowel disease Setting: outpatient Intervention: previous IBD surgery Comparison: no previous IBD surgery
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№ of women/pregnancies (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with no previous surgery	Risk with any previous IBD surgery
Infertilty at 12 months	Study population		RR 5.45 (0.41 to 72.57)	114 women (2 observational studies)	⊕⊝⊝⊝ Very low^a,b
	47 per 1000	253 per 1000 (19 to 1000)
Infertility at 24 months	Study population		RR 3.59 (1.32 to 9.73)	106 women (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	83 per 1000	297 per 1000 (109 to 805)
Miscarriage	Study population		OR 2.03 (1.14 to 3.60)	776 pregnancies (5 observational studies)	⊕⊝⊝⊝ Very low^a
	69 per 1000^c	130 per 1000 (78 to 210)
Stillbirth	Study population		RR 1.96 (0.42 to 9.18)	246 pregnancies (3 observational studies)	⊕⊝⊝⊝ Very low^a,b
	20 per 1000	39 per 1000 (8 to 184)
Use of ART	Study population		RR 25.09 (1.56 to 403.76)	106 women (1 observational study)	⊕⊝⊝⊝ Very low^a,b	We were unable to calculate absolute effects: 20 out of 66 women with previous surgery used ART compared to 0 out of 40 women with no previous surgery
	See comment	See comment
Caesarean section	Study population		RR 2.23 (1.00 to 4.95)	20 pregnancies (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	308 per 1000	686 per 1000 (308 to 1000)
Preterm delivery	Study population		RR 1.91 (0.67 to 5.48)	194 pregnancies (3 observational studies)	⊕⊝⊝⊝ Very low^a,b
	52 per 1000	98 per 1000 (35 to 282)
Low birth weight	Study population		RR 0.61 (0.08 to 4.83)	19 pregnancies (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	154 per 1000	94 per 1000 (12 to 743)
Small for gestational age	Study population		RR 2.54 (0.80 to 8.01)	65 (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	125 per 1000	318 per 1000 (100 to 1000)
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ART: assisted reproductive technology; CI: confidence interval; IBD: inflammatory bowel disease; OR: odds ratio; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded due to high overall risk of bias. ^bDowngraded due to imprecise estimates. ^cRisk of miscarriage in women with IBD who have not had a previous IBD surgery was based on data from 189 pregnancies (3 studies).

Summary of findings for the main comparison. Any previous surgery compared to no previous surgery for inflammatory bowel disease on female infertility

Summary of findings 2. Laparoscopic compared to open surgery for inflammatory bowel disease and female infertility

Risk of infertility among women undergoing surgery with a laparoscopic approach compared to an open approach
Patient or population: females with inflammatory bowel disease Setting: outpatient Intervention: laparoscopic surgery Comparison: open surgery
Outcomes	*Anticipated absolute effects (95% CI)**		Relative effect (95% CI)	№ of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with open surgery	Risk with laparoscopic surgery
Infertilty at 12 months	Study population		RR 0.70 (0.38 to 1.27)	19 (1 observational study)	⊕⊝⊝⊝ Very low^a,b
	647 per 1000	453 per 1000 (246 to 822)
Infertility at 24 months	These outcomes were not reported
Miscarriage
Stillbirth
Use of ART
Caesarean section
Preterm delivery
Low birth weight
Small for gestational age
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). ART: assisted reproductive technology; CI: confidence interval; RR: risk ratio.
GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.
^aDowngraded due to high overall risk of bias. ^bDowngraded due to imprecise estimates.

Summary of findings 2. Laparoscopic compared to open surgery for inflammatory bowel disease and female infertility

Table 1. Sample size and outcomes among studies included in the meta‐analysis

Infertility

Secondary outcomes

Study

Comparison

Type of IBD

Total sample

12 months

24 months

Miscarriage

Stillbirth

Use of ART

Caesarean section

Preterm delivery

Low birth weight

Small for gestational age

Banks 1957

Any previous surgery vs no previous surgery

56 women (87 pregnancies)

87 pregnancies

72 pregnancies

Bartels 2012

Open vs laparoscopic IPAA

37 women

Bortoli 2011

Any previous surgery vs no previous surgery

145 women

Hudson 1997

Any previous surgery vs no previous surgery

86 women^a

86 women

104 women^a

132 pregnancies

94 pregnancies

Johnson 2004

IPAA vs no previous surgery

106 women^a

106 women

Koivusalo 2009

Restorative proctocolectomy with ileoanal anastomosis vs no previous surgery

8 women^a

Moser 2000

Any previous surgery vs no previous surgery

65 women (76 pregnancies)

65 pregnancies^b

Naganuma 2011

Any previous surgery vs no previous surgery

80 women (69 pregnancies^c)

69 pregnancies^d

245 women (234 pregnancies^c)

234 pregnancies^d

Nielsen 1984

Any previous surgery vs no previous surgery

68 women (109 pregnancies)

109 pregnancies

80 pregnancies

Zavorova 2017

Any previous surgery vs no previous surgery

12 women (14 pregnancies)

14 pregnancies

13 pregnancies^e

IPAA vs no previous surgery

5 women (6 pregnancies)

6 pregnancies

^aExcludes women who were voluntarily infertile or were excluded from estimates of infertility for other reasons (e.g. not married or cohabitating).

^bFor women with multiple pregnancies, one was chosen at random.

^cOnly pregnancies occurring on or after the onset of IBD were included.

^dPregnancies among women with CD and UC were analyzed together.

^eBirthweight was missing for one infant

ART: assisted reproductive technology.
CD: Crohn's disease.
IBD: inflammatory bowel disease.
UC: ulcerative colitis.

Table 1. Sample size and outcomes among studies included in the meta‐analysis

Table 2. Sample size and outcomes among studies comparing women before and after surgery

				Infertility			Secondary outcomes
Study	Type of surgery	Type of IBD	Total sample	6 months	12 months	24 months	Miscarriage	Stillbirth	Use of ART	Caesarean section	Time to pregnancy
Gorgun 2004	Restorative proctocolectomy with IPAA	UC	185 women		Before: 117 women After: 120 women Both before and after: 52 women				Before: 45 women^a After: 70 women^a
Hahnloser 2004	IPAA	UC	37 women				Before and after: 37 women (37 pregnancies)	Before: 27 women (27 pregnancies) After: 31 women (31 pregnancies)
Johnson 2004	IPAA	UC	153 women		Before: 95 women^a After: 66 women^a				Before: 95 women^a After: 66 women^a
Mortier 2006	Total colectomy with ileorectal anastomosis	UC	37 women	Before: 5 women^a After: 15 women^a	Before: 5 women^a After: 15 women^a	Before: 5 women^a After: 15 women^a					Before: 5 women^a After: 5 women^a
Olsen 2002	Restorative proctocolectomy with IPAA	UC	290 women (331 attempted pregnancies)		Before: 84 attempted pregnancies^a After: 149 attempted pregnancies^a	Before: 84 attempted pregnancies^a After: 149 attempted pregnancies^a
Ravid 2002	IPAA	UC	38 women^b				Before: 12 women (16 pregnancies) After: 38 women (65 pregnancies^c)	Before: 11 pregnancies After: 49 pregnancies^c		Before: 11 pregnancies After: 48 pregnancies (49 deliveries^d)
Tulchinksy 2013	Restorative proctocolectomy	UC	41 women		Before: 26 women^a After: 27 women^a		Before: 26 women (70 pregnancies) After: 17 women (32 pregnancies)		Before: 26 women (70 pregnancies)^a After: 27 women (32 pregnancies)^a	Before: 62 pregnancies After: 26 pregnancies	Before: 26 women (70 pregnancies) After: 17 women (32 pregnancies)
^aUnclear how many women were at risk of developing the outcome both before and after surgery. ^bAll women included in the study gave birth at least once after IPAA. ^cOngoing pregnancies excluded at the time of the study. ^dA set of twins included (one born vaginally and one born via Caesarean section). ART: assisted reproductive technology. CD: Crohn's disease. IBD: inflammatory bowel disease. IPAA: ileal pouch anal anastomosis. UC: ulcerative colitis.

Table 2. Sample size and outcomes among studies comparing women before and after surgery

Table 3. Infertility at 12 months in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Infertile, n (%)	Total, n	Infertile, n (%)	Total, n
Restorative proctocolectomy with IPAA	Gorgun 2004	45 (38.5)	117	70 (58.3)	120
Restorative proctocolectomy with IPAA	Olsen 2002	18 (21.4)	84	122 (81.9)	149
IPAA	Johnson 2004	4 (4.2)	95	29 (43.9)	66
Total colectomy with ileorectal anastomosis	Mortier 2006	1 (20.0)	5	8 (53.3)	15
Restorative proctocolectomy	Tulchinksy 2013	0 (0.0)	26	10 (37.0)	27
Total		68 (20.8)	327	239 (63.4)	377
IPAA: ileal pouch anal anastomosis.

Table 3. Infertility at 12 months in female patients with UC before and after surgery

Table 4. Infertility at 12 months in female patients with UC before and after surgery as reported in Gorgun 2004

	Before surgery		After surgery
Age at surgery	Infertile, n (%)	Total, n	Infertile, n (%)	Total, n
≤ 19 years	0 (0)	1	4 (50)	8
20 to 34 years	38 (40.9)	93	61 (57.0)	107
≥ 35 years	7 (30.4)	23	5 (100)	5
UC: ulcerative colitis.

Table 4. Infertility at 12 months in female patients with UC before and after surgery as reported in Gorgun 2004

Table 5. Infertility at 24 months in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Infertile, n (%)	Total, n	Infertile, n (%)	Total, n
Restorative proctocolectomy with IPAA	Olsen 2002	13 (15.5)	84	108 (72.5)	149
Restorative proctocolectomy	Mortier 2006	1 (20.0)	5	7 (46.7)	15
Total		14 (15.7)	89	115 (70.1)	164
IPAA: ileal pouch anal anastomosis. UC: ulcerative colitis.

Table 5. Infertility at 24 months in female patients with UC before and after surgery

Table 6. Miscarriage in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Miscarriage, n (%)	Total, n	Miscarriage, n (%)	Total, n
IPAA	Hahnloser 2004	10 (27.0)	37	6 (16.2)	37
IPAA	Ravid 2002	1 (6.3)	16	9 (13.8)	65
Restorative proctocolectomy	Tulchinksy 2013	8 (11.4)	70	6 (18.8)	32
Total		19 (15.4)	123	21 (15.7)	134
IPAA: ileal pouch anal anastomosis. UC: ulcerative colitis.

Table 6. Miscarriage in female patients with UC before and after surgery

Table 7. Stillbirth in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Stillbirth, n (%)	Total, n	Stillbirth, n (%)	Total, n
IPAA	Hahnloser 2004	2 (7.4)	27	2 (6.5)	31
Restorative proctocolectomy	Ravid 2002	0 (0)	11	1 (2.0)	49
Total		2 (5.3)	38	3 (3.8)	80
IPAA: ileal pouch anal anastomosis. UC: ulcerative colitis.

Table 7. Stillbirth in female patients with UC before and after surgery

Table 8. Use of ART in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Use of ART, n (%)	Total, n	Use of ART, n (%)	Total, n
IPAA	Johnson 2004	5 (5.3)	95^a	20 (30.3)	66^a
Restorative proctocolectomy with IPAA	Gorgun 2004	19 (42.2)	45^b	36 (51.4)	70^b
Restorative proctocolectomy	Tulchinksy 2013	6 (8.6)	70^c	11 (34.3)	32^c
^aNumbers of women attempting pregnancy before and after surgery. ^bNumber of women infertile following 12 months of attempting pregnancy. ^cNumber of pregnancies in which women used ART. Note: Totals not calculated due to differing denominators among included studies. ART: assisted reproductive technology. IPAA: ileal pouch anal anastomosis. UC: ulcerative colitis.

Table 8. Use of ART in female patients with UC before and after surgery

Table 9. Caesarean section in female patients with UC before and after surgery

		Before surgery		After surgery
Type of surgery	Study	Caesarean section, n (%)	Total, n	Caesarean section, n (%)	Total, n
IPAA	Ravid 2002	0 (0)	11	24 (49.0)	49^a
Restorative proctocolectomy	Tulchinksy 2013	8 (12.9)	62	12 (46.2)	26
Total		8 (11.0)	73	36 (48.0)	75
^aIncludes a set of twins (one born vaginally and one born via Caesarean section). IPAA: ileal pouch anal anastomosis. UC: ulcerative colitis.

Table 9. Caesarean section in female patients with UC before and after surgery

Table 10. Time to pregnancy in female patients with UC before and after surgery, measured in months

			Before surgery		After surgery
Type of surgery	Study	Measure	Time to pregnancy	Number of women attempting pregnancy	Time to pregnancy	Number
Total colectomy with ileorectal anastomosis	Mortier 2006	Median (range)	2 (1‐3)	5	5 (2‐36)	15
Restorative proctocolectomy	Tulchinksy 2013	Mean (SD)	5.0 (11.6)	26 (70 pregnancies^a)	16.3 (25.1)	27^b (32 pregnancies)
^aPregnancy occurred "immediately" in 32 pregnancies. ^b17/27 women successfully conceived. SD: standard deviation. UC: ulcerative colitis.

Table 10. Time to pregnancy in female patients with UC before and after surgery, measured in months

Comparison 1. Any surgery versus no surgery

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Infertiltiy at 12 Months Show forest plot	2	114	Risk Ratio (M‐H, Random, 95% CI)	5.45 [0.41, 72.57]

2 Infertility at 24 Months Show forest plot	1	190	Risk Ratio (M‐H, Random, 95% CI)	3.59 [1.32, 9.73]

2.1 Crohn's Disease	1	86	Risk Ratio (M‐H, Random, 95% CI)	2.03 [0.56, 7.33]
2.2 Ulcerative Colitis	1	104	Risk Ratio (M‐H, Random, 95% CI)	5.28 [2.09, 13.34]
3 Miscarriage Show forest plot	5	776	Odds Ratio (Random, 95% CI)	2.03 [1.14, 3.60]

3.1 Crohn's Disease	3	386	Odds Ratio (Random, 95% CI)	2.56 [1.19, 5.51]
3.2 Ulcerative Colitis	1	87	Odds Ratio (Random, 95% CI)	7.14 [1.02, 50.18]
3.3 IBD	1	303	Odds Ratio (Random, 95% CI)	1.20 [0.54, 2.67]
4 Miscarriage ‐ Sensitivity Analysis Based on Source of Participants Show forest plot	5		Odds Ratio (Random, 95% CI)	Subtotals only

4.1 Population‐Based	1	132	Odds Ratio (Random, 95% CI)	2.79 [1.11, 7.04]
4.2 Tertiary Care	4	644	Odds Ratio (Random, 95% CI)	1.84 [0.84, 4.03]
5 Miscarriage ‐ Sensitivity Analysis Including Only Women in Remisson at Conception Show forest plot	2	173	Risk Ratio (M‐H, Random, 95% CI)	2.38 [1.11, 5.11]

6 Stillbirth Show forest plot	3	246	Risk Ratio (M‐H, Random, 95% CI)	1.96 [0.42, 9.18]

6.1 Crohn's Disease	2	174	Risk Ratio (M‐H, Random, 95% CI)	1.98 [0.32, 12.16]
6.2 Ulcerative Colitis	1	72	Risk Ratio (M‐H, Random, 95% CI)	1.91 [0.10, 36.02]
7 Stillbirth ‐ Sensitivity Analysis Based on Source of Participants Show forest plot	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

7.1 Population‐Based	1	94	Risk Ratio (M‐H, Random, 95% CI)	5.47 [0.23, 130.65]
7.2 Tertiary Care	2	152	Risk Ratio (M‐H, Random, 95% CI)	1.43 [0.24, 8.35]
8 Stillbirth ‐ Sensitivity Analysis Including Only Women in Remission at Conception Show forest plot	2	136	Risk Ratio (M‐H, Random, 95% CI)	1.26 [0.07, 23.54]

9 Use of ART Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

10 Caesarean Section Show forest plot	1	20	Risk Ratio (M‐H, Random, 95% CI)	2.23 [1.00, 4.95]

10.1 Crohn's Disease	1	14	Risk Ratio (M‐H, Random, 95% CI)	3.60 [0.98, 13.19]
10.2 Ulcerative Colitis	1	6	Risk Ratio (M‐H, Random, 95% CI)	1.67 [0.61, 4.59]
11 Preterm Delivery Show forest plot	3	194	Risk Ratio (M‐H, Random, 95% CI)	1.91 [0.67, 5.48]

11.1 Crohn's Disease	3	188	Risk Ratio (M‐H, Random, 95% CI)	2.32 [0.75, 7.21]
11.2 Ulcerative Colitis	1	6	Risk Ratio (M‐H, Random, 95% CI)	0.56 [0.03, 9.73]
12 Preterm Delivery ‐ Sensitivity Analysis Based on Source of Participants Show forest plot	3		Risk Ratio (M‐H, Random, 95% CI)	Subtotals only

12.1 Population‐Based	1	94	Risk Ratio (M‐H, Random, 95% CI)	1.85 [0.27, 12.53]
12.2 Tertiary Care	2	100	Risk Ratio (M‐H, Random, 95% CI)	2.02 [0.56, 7.28]
13 Preterm Delivery ‐ Sensitivity Analysis Including Only Women in Remisson at Conception Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

14 Low Birth Weight Show forest plot	1	19	Risk Ratio (M‐H, Random, 95% CI)	0.61 [0.08, 4.83]

14.1 Crohn's Disease	1	13	Risk Ratio (M‐H, Random, 95% CI)	0.67 [0.03, 13.60]
14.2 Ulcerative Colitis	1	6	Risk Ratio (M‐H, Random, 95% CI)	0.56 [0.03, 9.73]
15 Small for Gestational Age Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

Comparison 1. Any surgery versus no surgery

Comparison 2. Laparoscopic versus open surgery

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Infertiltiy at 12 Months Show forest plot	1		Risk Ratio (M‐H, Random, 95% CI)	Totals not selected

Comparison 2. Laparoscopic versus open surgery