Introduction

Bowel dysfunction following ileal pouch-anal anastomosis (IPAA) for ulcerative colitis (UC) is a significant problem, having a negative impact on the quality of life for up to one fourth of patients [1]. It is generally agreed that postoperative infectious complications, especially pelvic sepsis, have the most detrimental effect on bowel function [2]. Identification of covariates associated with an increased risk of pelvic sepsis is therefore paramount. Many consider fecal diversion with a loop ileostomy at IPAA the standard approach, as the risk of pelvic sepsis is reduced [3]. Consequently, omission of fecal diversion at IPAA may therefore increase the risk of septic complications, and ultimately be causally linked to a poorer pouch function.

However, the short-term morbidity of a high stoma and potential risks associated with ileostomy reversal has led some surgeons to only divert ileal pouches selectively [4]. Further, exclusion of a bowel segment from the normal intestinal circuit leads to atrophy of the bowel musculature [5, 6], and depletion of luminal nutrition may impair the absorptive capacity of the epithelium [7, 8]. Although these structural changes to the defunctioned bowel are likely momentary and mended by the restoration of bowel continuity, the long-term effects of fecal diversion on bowel physiology remain unclear, and a potential negative effect on long-term bowel function cannot be dismissed [9, 10].

The aim of this study was therefore to explore the association between primary fecal diversion and pouch dysfunction following IPAA for UC.

Methods

Setting

This study was conducted using cross-sectional questionnaire data obtained from a national cohort of IPAA patients [11] in addition to prospectively registered nationwide health registry data. All individual-level data was linked by the civil registration number, which is a ten-digit personal identifier issued to all Danish citizens at birth or upon attainment of citizenship [12].

Cohort

From a source population of all patients who had IPAA surgery in the period 1980–2010 in Denmark [11], we identified those operated in 2000–2010 in order to have a cohort representing a contemporary patient population with minimal attrition. Data on pouch function was collected as part of a previous nationwide cross-sectional study [11], where questionnaires were sent to patients in July 2011. The questionnaire is validated and consists of 5 subdomains concerning bowel function within the preceding 14 days, with each item given a specific weight based on its impact on quality of life. A summarized score ranging from 0 to 7.5 is thus calculated, and a score below 2.5 is considered to have no negative impact on quality of life, while a score equal to or above 2.5 has a moderate or severe negative impact on quality of life (see Supplementary Table 1 for the score sheet).

Outcome

The primary outcome of interest was pouch dysfunction, i.e., a pouch dysfunction score of 2.5 or more. The bowel function subdomains contained in the score, i.e., incontinence, frequency of defecation, incomplete emptying, urgency, and use of anti-diarrheal medication, were chosen as secondary outcome measures, to infer causality to any possible associations.

Covariates

The presence of a diverting ileostomy at IPAA was included as an explanatory, dichotomous covariate, as were the variables gender (male or female), stage (completion proctectomy with IPAA or proctocolectomy with IPAA), and surgical approach (minimally invasive or open). Age at IPAA was included as an ordinal variable in 10-year intervals (≤ 30 years, 31–40, 41–50, 51–60, > 60).

For patients who had primary fecal diversion, we determined the number of weeks with fecal diversion, from IPAA to ileostomy reversal.

Data on surgical procedures with associated dates was collected from the Danish National Patient Registry [13], which is a nationwide register that contains information on all in-hospital health care contacts since 1977 and out-patient care since 1995. Surgical procedures are coded according to the Nordic Medico-Statistical Committee (NOMESCO) classification, which replaced the Danish Classification of Surgical Procedures and Therapies in 1996. The overall national IPAA cohort has previously been described in detail [11, 14].

From the National Patient Register, we also included information on diagnoses and procedures relating to pelvic complications. We defined that a code for intraabdominal abscess or intestinal fistula occurring within 1 year after pouch surgery or anastomotic leakage and extremely early-onset pouchitis occurring within 30 days of primary pouch surgery constituted a pelvic complication (see specific codes in Supplementary Table 2).

Statistical analysis

For descriptive analysis, mean and standard deviation (sd) describe normally distributed data, while median and interquartile ranges (IQR) describe non-normally distributed.

The pouch function questionnaires were sent out during the summer of 2011, and so, July 1st, 2011 defined the end of the observation period for all study subjects. To explore whether patients responding to the questionnaires could be considered representative of an unselected IPAA population, we assessed differences in demographic and clinical covariates between questionnaire responders and eligible non-responders (i.e., primary non-response or patients under research protection). The unpaired Students’ t test or Mann-Whitney U were used as appropriate for comparison of continuous data and Spearman’s rank for categorical variables. Among all patients operated in 2000–2010, the proportion of patients diagnosed with a pelvic complication in the abovementioned interval was compared between patients with and without primary fecal diversion, calculating an odds ratio with 95% confidence interval as a measure of relative risk.

For patients who responded to the questionnaire, the association between primary fecal diversion and pouch dysfunction was calculated on logistic regression as odds ratios (OR) with 95% confidence intervals (95% CI) on univariate analysis, and on multivariate analyses including age group, gender, laparoscopy, completion proctectomy, and pelvic complications as explanatory variables. The effect of number of weeks with fecal diversion on risk of pouch dysfunction was explored on logistic regression analysis, adjusting for the same potential confounders as described above. Differences in the respective subdomains of pouch function between patients with and without fecal diversion were compared using the unpaired Chi-squared test for binary variables and Spearman’s rank correlation for ordinal variables.

We also performed a sensitivity analyses on the risk of pouch dysfunction, stratifying patients by whether they had taken part in the development or validation part of the study. All analyses were performed using Intercooled Stata version 13.1 (College Station, TX: StataCorp LP).

Results

From the source population of 1757 patients having primary IPAA surgery in Denmark in the period 1980–2010, we included 871 patients from the period January 2000 to July 2010. Of these 871 patients, 80 had experienced pouch failure and thus had a permanent ileostomy, while 29 had died or emigrated. This left 762 eligible patients with a functioning pouch, of which 173 were under research protection (i.e., they refuse participation in research and can thus not be contacted). The remaining 589 were sent the pouch dysfunction questionnaire, and 504 (85.6%) returned it, and these patients constituted the final cohort. Figure 1 describes the flow of patients from the initial nationwide cohort to the final contemporary cohort, while demographic and clinical characteristics of both responders and eligible non-responders are listed in Table 1. Overall, there were no clinically significant differences between responders and non-responders; responders were slightly older at the questionnaire date, fewer had undergone minimally invasive IPAA, and the median time from IPAA to the questionnaire date was 5.8 years for responders, compared to 5.3 years for non-responders. Eleven percent of eligible non-responders had experienced a pelvic complication, compared to 7% among responders (OR = 1.58 (95% CI: 0.94–2.66)).

Fig. 1
figure 1

Patient flowchart

Full size image
Table 1 Characteristics between questionnaire responders and eligible non-responders
Full size table

In the source cohort of all 871 patients, the risk of pelvic complications occurring within the first year after primary pouch surgery was slightly higher for patients without primary fecal diversion (12.6%) compared to patients with primary fecal diversion (10.1%, OR = 1.29 (95% CI: 0.68–2.48)).

Pouch failure and death or emigration prior to the questionnaire date was more common for patients without primary fecal diversion (OR = 2.05 (95% CI: 1.12–3.76) and OR 2.21 (95% CI: 0.93–5.22), respectively), while non-response in otherwise eligible patients was equally common for those with and without primary fecal diversion (34 vs. 32%, OR = 1.10 (95% CI: 0.66–1.83)).

Of the 504 questionnaire responders, 51 (10%) had IPAA surgery without temporary fecal diversion, while 453 (90%) had fecal diversion. The median time from IPAA surgery to the questionnaire date was 5.8 years (interquartile range 3.4–8.3); 5.9 years for patients with diversion at IPAA compared to 5.2 years for patients without diversion (Mann-Whitney U p = 0.02).

Pouch dysfunction

The mean pouch dysfunction scores were 1.62 (95% CI: 1.47–1.78) and 2.44 (95% CI: 1.88–3.00) (median 1 (IQR 0–2.5) and 2 (IQR 1–4) for patients with and without fecal diversion, respectively. The risk of pouch dysfunction was higher for patients without primary fecal diversion (48% (95% CI: 34–62%)) compared to those with primary fecal diversion (30% (95% CI: 26–35%)) with a corresponding OR of 2.12 (95% CI: 1.17–3.84) (Table 2). The risk of dysfunction increased with age (OR for each 5-year increment = 1.25 (95% CI: 1.06–1.48)). Females had a higher risk of pouch dysfunction (OR = 1.33 (95% CI: 0.90–1.95)), while no differences were found for patients undergoing a proctocolectomy with IPAA versus completion proctectomy with IPAA (OR = 0.96 (0.59–1.57)). Patients undergoing a minimally invasive procedure had a higher risk of dysfunction, compared to patients undergoing open IPAA (OR = 1.59 (0.35–7.21)). The risk was also higher for those experiencing a pelvic complication (OR = 1.77 (95% CI: 0.89–3.52)).

Table 2 Associations between baseline covariates and pouch dysfunction
Full size table

On multivariate regression analysis, the risk of pouch dysfunction was higher for patients without fecal diversion (adjusted OR = 2.23 (95% CI: 1.20–4.14)) and for females (adjusted OR = 1.41 (95% CI: 0.94–2.11)). The associations between pouch dysfunction and the variables age, completion proctectomy, a minimally invasive approach, and pelvic complications remained largely unchanged (Table 2).

For patients with primary fecal diversion, there was no association between number of weeks with diversion and major pouch dysfunction (adjusted OR for every week with diversion = 1.02 (95% CI: 0.99–1.05)).

Bowel function subdomains

Patients without diversion experienced ‘incomplete emptying’ (Spearman’s rho − 0.11, p = 0.02) and ‘urgency’ (Spearman’s rho − 0.10, p = 0.03)) more often than patients with diversion, and they also reported more bowel movements per day (median 8 (IQR 6–10) vs. 7 (IQR 5–9), Mann-Whitney U p = 0.04, Table 3).

Table 3 Differences in bowel function subdomains between patients with and without primary fecal diversion at IPAA
Full size table

No differences were found for degree of incontinence (Spearman’s rho 0.002, p = 0.96) or use of antidiarrheal medication (OR = 0.85 (0.46–1.57), p = 0.61).

Sensitivity analyses

When analyzing only patients who took part in the validation of the pouch dysfunction score, the estimates from the regression analyses and comparisons of bowel function subdomains were not affected materially (data not shown), although the low number of patients without fecal diversion in this subgroup (n = 22) made the comparisons of especially bowel function subdomains more uncertain.

Discussion

Temporary diversion of the fecal stream is by many considered standard practice when performing low colorectal or anal anastomoses, as the risk of symptomatic anastomotic leakage may be reduced [3, 15].

With this study, we found that more patients who did not have primary fecal diversion at IPAA for UC suffered from major pouch dysfunction compared to those who had diversion. Importantly, we used a validated bowel function questionnaire [11], developed for this specific patient population, to examine clinically meaningful differences in bowel function between those with and without primary fecal diversion at IPAA.

The higher occurrence of pouch failure prior to the questionnaire date for patients without fecal diversion suggests that these patients, who presumably had the worst pouch function, were excluded by default. This would in turn have biased our results towards a seemingly better function for those without fecal diversion, so the true difference in functional outcome between patients with and without primary diversion may be even more pronounced than reported here. The questionnaire response rate was remarkably high, and there were no clinically significant differences in demography or clinical variables between responders and non-responders. The minor difference in time passed from IPAA to the cross-sectional questionnaire date between responders and eligible non-responders is probably a chance finding with no clinical relevance, since pouch function remains relatively stable after the first few postoperative years [16]. It therefore seems likely that the responders were representative of a general IPAA population, although the higher risk of pelvic complications among eligible non-responders may theoretically underestimate the true prevalence of pouch dysfunction.

Despite recurring controversies regarding the effect of temporary fecal diversion in IPAA surgery, the only randomized controlled trial published 25 years ago with only 45 patients was too small to detect clinically significant differences in leak rates [17]. Diverging results from observational studies [3, 4, 14, 18,19,20,21] are likely to be a result of confounding by indication; some perform diversion as standard in primary IPAA surgery, and only refrain in cases of a short mesentery, truncal adiposity, or patient request. Others only divert when the a priori risk of septic complications is increased, e.g., in cases of malnutrition, emergency surgery, hand-sewn anastomosis, tension on the pouch-anal anastomosis, staple misfire, or when tissue rings from the circular stapling are incomplete. The rationale for this approach being, that the morbidity associated with a loop ileostomy, including risks associated with its closure, outweighs the advantages of diverting the ileostomy in low-risk patients. These two very different approaches to diverting an IPAA will inevitably lead to conflicting results in terms of the effect of temporary fecal diversion on complication rates and ultimately functional outcome, especially when the confounding factors are not adjusted for [4, 14, 18].

A meta-analysis based on data from 17 studies of varying methodology found that primary diversion was associated with a lower risk of pelvic sepsis following IPAA [3], and we speculate that this association in turn could explain the lower risk of pouch failure found in our national cohort study for patients with fecal diversion at IPAA and the better long-term functional outcome. Although the odds ratio of early pelvic complications was marginally higher for patients without primary fecal diversion compared to those with diversion, the low number of patients impairs the robustness of this estimate. The pattern of pouch dysfunction found in the present study does, however, bear some resemblance to what has been found in studies comparing functional outcome for patients with and without symptomatic anastomotic leakage following both IPAA [2, 22] and low anterior resection (LAR) for rectal cancer [23, 24]. The long-term effects of fecal diversion on postoperative bowel function may, however, differ between patients with IPAA and patients with LAR; a recently published post hoc analysis of a randomized trial investigating the effect of primary diversion in LAR for rectal cancer found that patients randomized to ileostomy were more likely to have incontinence for flatus and liquid stools and a worse low anterior resection syndrome score, compared to patients without ileostomies, 12 years after operation [9]. These findings are in line with a large prospective study performed in Denmark and the UK that found primary fecal diversion to be a key predictive factor for a high low anterior resection syndrome score [10]. Profound differences in physiology, structure, and microbiotic environment between the small bowel and rectum may offer a hypothetical explanation for differences in the effect of diverting ileostomy on long-term bowel function for patients with IPAA and LAR. The scarcity of literature on the physiological effects of fecal diversion in IPAA does, however, make this proposed differential effect of fecal diversion on small bowel and rectal function speculative at best.

Among patients who had primary fecal diversion, we found no association between major dysfunction and number of weeks with diversion. This argues against a substantial impact of possible small bowel atrophy on long-term functional outcome, although we are limited by a lack of information on this histological parameter. In addition, the indications to either accelerate reversal of the ileostomy (e.g., because of small bowel obstruction or severe dehydration) or to postpone it from the usual few months (e.g., because of pelvic complications) were unknown, and both scenarios may theoretically have an adverse effect on subsequent bowel function.

Of other limitations to our study, the most important is the cross-sectional evaluation of bowel function, which prevents any conclusions regarding causality, and since the median time from IPAA was 5.8 years, our results may not be transferable beyond the sixth postoperative year. Also, only a relatively low number of patients did not have primary fecal diversion, and this impairs the robustness of our estimates, including the comparison of pelvic complications.

Although we included data on pelvic complications to infer biological causality to our findings, the use of administrative data to ascertain the occurrence of surgical complications is likely associated with some misclassification. Nonetheless, any potential misclassification of procedures and diagnoses would be non-differential (i.e., unrelated to the presence of primary fecal diversion), biasing the estimate towards null.

We are also limited by an absence of information on rectal dissection method, type of anastomosis, and pouch design, although the choice to omit fecal diversion is unlikely driven by any of these factors. We therefore consider it unlikely, albeit not irrefutable, that there are differences in these covariates between diverted and non-diverted cases to introduce substantial bias or confound the associations. A more probable source of bias lies in the absence of data concerning body mass index (BMI) [25, 26]. At our department, the mean BMI of patients undergoing IPAA in the period 2004–2014 was 24.2 (unpublished data from previous study [27]), and patients without diverting ileostomy had a higher BMI (27.2 versus 23.8) amounting to a mean difference of 3.4 kg/m2 (p < 0.05). The low absolute BMI value of 27.2 for patients without diversion is nonetheless an unlikely explanation for any possible differences in pelvic complications [25, 28].

Adequately powered, prospective studies that account for potentially confounding factors in the causal relationship between fecal diversion and postoperative outcome seem warranted to overcome the limitations of existing studies. A randomized study on highly selected patients with a low a priori risk of postoperative pelvic sepsis also seems justified to determine if fecal diversion could be safely omitted, thereby avoiding the morbidity associated with an ileostomy.

In conclusion, in a cross-sectional evaluation of bowel function following IPAA for UC in a national cohort, we found a higher proportion of patients experiencing pouch failure and suffering from pouch dysfunction among those who did not have primary fecal diversion, compared to those who did. Causality remains unclear, but a higher risk of pelvic complications associated with omission of fecal diversion seems likely.