Preoperative Exclusive Enteral Nutrition Is Associated With Reduced Skin and Soft Tissue and Intra-abdominal Infections in Patients With Crohn’s Disease Undergoing Intestinal Surgery: Results from a Meta-Analysis

Abstract

Background

Malnutrition is an independent risk factor for adverse postoperative outcomes and is common among patients with Crohn’s disease (CD). The objective of this meta-analysis was to precisely quantify the association of preoperative exclusive enteral nutrition (EEN) and total parenteral nutrition (TPN) with surgical outcomes in patients undergoing intestinal surgery for CD.

Methods

PubMed, Embase, and Scopus were queried for comparative studies evaluating the impact of preoperative nutritional support on postoperative outcomes in patients undergoing surgery for CD. Random effects modeling was used to compute pooled estimates of risk difference. Heterogeneity was assessed using I².

Results

Fourteen studies, all nonrandomized cohort studies, met inclusion criteria for studying EEN. After pooling data from 14 studies (874 EEN treated and 1044 control patients), the relative risk of intra-abdominal septic complications was decreased 2.1-fold in patients receiving preoperative EEN (relative risk 0.47, 95% confidence interval [CI], 0.35-0.63, I² = 0.0%). After pooling data from 9 studies (638 EEN treated and 819 control patients), the risk of skin and soft tissue infection was decreased 1.6-fold (relative risk 0.63; 95% CI, 0.42-0.94, I² = 42.7%). No significant differences were identified in duration of surgery, length of bowel resected, or operative blood loss. Among the 9 studies investigating TPN, no significant differences were identified in infectious outcomes.

Conclusions

Preoperative nutritional optimization with EEN was associated with reduced risk of infectious complications in CD patients undergoing intestinal surgery. Preoperative nutritional support with EEN should be considered for optimizing outcomes in CD patients requiring bowel resection surgery.

Introduction

Poor nutritional status is relatively common among patients with Crohn’s disease (CD) in both the inpatient and outpatient setting.^1,2 Mechanisms for development of malnutrition in CD include restrictive food intake and malabsorption.^3,4 Poor nutritional status is associated with adverse outcomes following surgical intervention.^5,6 Although overall risk of requiring surgery in CD has decreased over recent decades, almost half of patients with CD will require surgery within 10 years of diagnosis.⁷

Perioperative management of surgical patients has increasingly emphasized multidisciplinary, evidence-based care within the Enhanced Recovery After Surgery (ERAS) paradigm.⁸ The ERAS aims to optimize quality through interventions in the preoperative, intraoperative, and postoperative setting. Preoperative ERAS protocols emphasize nutritional screening and, if needed, support to reduce complications.

The European Crohn’s and Colitis Organization (ECCO) 2020 guidelines on surgical therapy in CD recommend nutritional assessment prior to surgery and optimization in patients with identified deficiencies through enteral or parenteral support.⁹ Prior studies investigating the impact of preoperative nutritional support on clinically relevant outcomes in CD have been limited by small sample sizes, and no randomized controlled trials have been completed to date on this subject. Our study aimed to utilize meta-analysis to better elucidate the impact of preoperative exclusive enteral nutrition (EEN) or total parenteral nutrition (TPN) on surgical outcomes in CD patients undergoing intestinal surgery.

Methods

This review was conducted according to MOOSE guidelines¹⁰ and registered with The International Prospective Register of Systematic Reviews (PROSPERO): registration number CRD42022346627.

Search Strategy

PubMed, Embase, and Scopus were queried for primary literature evaluating surgical outcomes in patients receiving surgical intervention for inflammatory bowel disease with and without preoperative nutritional support (hereafter referred to as nutritional optimization). The search strategy was adapted from Grass et al¹¹ and organized with the following syntax: (Crohn OR Crohn’s disease OR inflammatory bowel disease OR ulcerative colitis) AND (nutrition OR conditioning OR nutritional support) AND (preoperative OR perioperative). The complete search strategy can be found in Supplemental Table 1. Searches were performed in July 2022, and Covidence software was used for screening, extraction, and data management.¹²

Inclusion and Exclusion Criteria

Included studies assessed surgical outcomes in both nutritionally optimized and nonoptimized IBD patients. Studies that had a control group receiving more than 7 days of preoperative nutritional optimization or more than 600 kcal/day of nutritional support were excluded. Studies without stratified outcome analysis by nutrition subgroup were also excluded. The search was restricted to articles written in English. No restrictions on date or country were applied.

Data Extraction

Two investigators (L.K. and A.W.) independently screened abstracts of search results and performed full-text reviews of included studies to evaluate eligibility. Discrepancies in voting were resolved by discussion and consensus with a third author (J.L.). Two investigators (L.K. and A.W.) independently extracted pertinent data from included studies. Extracted data included author, year, country, IBD classification, type of artificial nutrition utilized, description of nutritional support provided, number of surgeries, body mass index (BMI), age, sex, preoperative serum albumin, preoperative C-reactive protein (CRP), type of surgical resection, surgical approach, length of bowel resected, duration of procedure, duration of postoperative stay, postsurgical infection rate, readmission rate, estimated operative blood loss, and mortality.

Several studies noted that in circumstances where EEN alone failed to achieve nutritional improvement, parenteral supplementation would be initiated.^13–15 Based on an intention to treat analysis framework, these patients were included in the EEN group despite receiving additional forms of alimentation.

For studies with summary statistics reported as a median and interquartile range, the median was assumed to be equivalent to the mean for analysis purposes, and standard deviation was estimated by dividing the interquartile range by 1.35. If only a range was given, standard deviation was estimated as range divided by 4. Supplementary raw data from 2 of the included studies were obtained to perform analysis stratified by nutrition type in studies that grouped EEN and TPN together.^16,17

Statistical Analysis

All analyses were conducted in Stata (version 17). Outcomes of interest were compared across supplemental nutrition categories using a random-effects model to estimate pooled risk ratios (RRs) and 95% confidence intervals (CIs) for categorical outcomes and standardized mean difference (SMD) with 95% CI for continuous outcomes. Studies were weighted according to the variance estimates to determine their contribution to the final summary statistic. Statistical heterogeneity was assessed via I², with values greater than 50% indicating significant heterogeneity. Pooled estimates were derived using the metan command.

Quality Assessment and Publication Bias

The risk of bias in included studies was assessed using the CLARITY Risk of Bias in Cohort Studies Tool.¹⁸ The tool grades the risk of bias of cohort studies across 8 categories using a 4 point scale. Publication bias was assessed using the Egger test for all comparisons, in addition to funnel plots for the intra-abdominal septic complication and surgical site infection analyses.

Results

A total of 1193 studies were imported for screening, and following removal of 479 duplicates, 714 abstracts were screened. Of these, 56 were assessed in full-text review for eligibility, and 20 studies were ultimately included for extraction.^{13–17,19–33}Figure 1 is the PRISMA diagram for the screening process.

Study Characteristics

Characteristics of the 20 included studies can be found in Table 1. All included studies were published between 1982 and 2022. Six studies were conducted in China, 5 studies were conducted in the United States, 3 studies were conducted in the United Kingdom, and the remainder took place in Israel, France, Denmark, Japan, Portugal, or Sweden. Nineteen studies were retrospective cohort studies, and 1 was a prospective cohort study. Patients with CD were investigated in 19 studies, and patients with UC were investigated in 2. Since only 2 studies included patients with UC, no separate analyses were performed on UC patients. Among studies of patients with CD, 9 included patients who had received total parenteral nutrition. Data on the nutritional support given to control patients, and the proportion of patients who transitioned from EEN to TPN and were categorized as EEN under intention-to-treat framework are included in Supplemental Table 2.

Population Characteristics

On average, baseline CRP levels were higher in the nutritional optimization groups compared with controls, and baseline albumin levels were lower. The standardized mean difference between albumin levels in patients receiving EEN vs controls was 0.01 g/L preoptimization and 0.55 g/L postoptimization (Figure 2a). Among studies reporting both pre- and postoptimization data for EEN-treated patients, there was a significant increase in albumin with nutritional optimization (SMD 0.55; 95% CI, 0.34-0.77; Supplemental Figure 1). The standardized mean difference between CRP levels in patients receiving EEN vs controls was 0.84 mg/L preoptimization and −0.78 mg/L postoptimization (Figure 2b). Among studies reporting both pre- and postoptimization data for EEN-treated patients, there was a significant decrease in CRP with nutritional optimization (SMD −0.87; 95% CI, −1.21 to −0.53; Supplemental Figure 2). Data on pre- and postoptimization differences in albumin and CRP among the TPN cohort were largely unavailable. Patient selection for EEN was based on baseline nutritional status in 2 of 14 (14%) studies and was part of routine care in 7 of 14 (50%). For TPN, patient selection was based on baseline nutritional status in 5 of 8 (63%) studies and part of routine care in 0 studies. Remaining studies did not clarify rationale behind how patients were assigned to nutritional intervention (Supplemental Table 2).

Infectious Outcomes

After pooling data from 14 studies^{13–17,20–23,27,28,31–33} consisting of 874 patients who received EEN and 1044 who did not, patients who received EEN had a significant, 2.1-fold lower risk of intra-abdominal septic complication (IASC) after surgery (RR, 0.47; 95% CI, 0.35-0.63, I² = 0.0%; Figure 3a., Table 2). After pooling data from 9 studies (638 EEN treated and 819 control patients),^{14,16,17,21–23,28,31,32} the risk of skin and soft tissue infection (SSI) was 1.6-fold lower in the EEN patients (RR, 0.63; 95% CI, 0.42-0.94, I² = 42.7%; Figure 3b., Table 2). No significant difference was found in IASC rates between patients who received TPN and those who did not (8 studies, 247 TPN patients and 876 controls; Supplemental Figure 3).^{13,16,17,19,24,26,29,30} Similarly, no significant difference was found in SSI rates between patients who received TPN and those who did not (6 studies, 222 TPN patients and 799 controls; Supplemental gure 4).^{16,17,19,24,26,30}

Given the observed reduced rates of infectious complications, increased albumin, and lower CRP concentration in the patients receiving EEN, we used meta regression to assess whether the relative rates of infectious outcomes were correlated with change in albumin and CRP following EEN. The anticipated linear trends were observed but did not reach statistical significance for change in CRP (P = .15, Supplemental Figure 5) or albumin (P = .64, Supplemental Figure 6).

Operative and Postoperative Outcomes

The rates of laparoscopic surgery were 1.42-fold (95% CI, 1.15-1.75) greater than laparotomy in patients optimized with EEN than controls (Supplemental Figure 7). The mean duration of surgery was shorter in the EEN group than in the control group; however, the pooled estimate was not statistically significant (SMD -0.96; 95% CI, −2.01-0.09; I² = 96.7%; Supplemental Figure 8, Table 2). No difference between control and EEN patients was observed in postoperative blood loss or total length of stay (Supplemental Figure 9, Supplemental Figure 10, Table 2). No difference in the length of resected bowel or laparoscopic-to-open surgery conversion rate was observed between EEN-treated vs control patients (Supplemental Figure 11, Supplemental Figure 12, Table 2) or TPN-treated vs control patients (Supplemental Figure 13, Supplemental Figure 14).

Bias Assessment

No statistically significant publication bias was detected for any of the comparisons using the Egger test (P > .05 for all comparisons, (Table 2). Some qualitative asymmetry was observed in the funnel plots for IASCs and SSIs analyses, indicating the possibility of publication bias (Supplemental Figure 15). Study-level risk of bias assessment is summarized in Figure 4. The majority of studies are at moderate-high risk of bias for incomplete matching of possible prognostic variables, moderate risk of bias for dissimilarities in co-interventions between groups, and some risk of bias for the assessment of the presence of prognostic variables.

Discussion

This meta-analysis aimed to evaluate the impact of preoperative EEN in patients with CD undergoing intestinal surgery. Utilizing pooled data from observational cohort studies, our study identified a significant reduction in important infectious complications, both superficial and deep, among the cohort optimized with EEN. In addition to the reductions in skin/soft tissue infections and intra-abdominal septic complications observed in the EEN treated patients, operative time was on average less in patients receiving EEN compared with non-nutritionally optimized patients, a notable difference that approached but did not ultimately reach statistical significance. Differences were not identified in other clinically relevant outcomes based on receipt of preoperative EEN, including operative blood loss, length of bowel resected, rate of conversion to open surgery, and hospital length of stay.

The findings of our study are particularly notable when considering the baseline health of the CD patients that comprised the EEN-optimized pooled cohort. Despite the rationale behind the choice of patient selection for optimization with EEN not being provided in one-third of the studies, analysis of baseline CRP and albumin levels suggests that the nutritional status of patients chosen to receive EEN was worse than for controls. The finding that these patients were more clinically unwell at baseline renders their improved postoperative outcomes even more impressive.

There are several possible mechanisms by which preoperative EEN may reduce postoperative infectious complications. More severe malnutrition and higher dose corticosteroid exposure are both associated with infectious complications after surgery in a dose-dependent manner.^6,34–38 As demonstrated by the observed increases in albumin and reduction in CRP, EEN may work by correcting nutritional deficiency and reducing inflammation and corticosteroid dose before surgery. Several studies noted that part of their EEN protocol was an attempt to wean corticosteroids.^{13,14,17,20,21,23,28} In exploratory analyses, meta-regression showed that the magnitude of reduction in CRP, more so than elevation in albumin, generally correlated with the effect size of the reduction in IASCs, though this trend was not statistically significant. It is also possible that the observed benefit of preoperative EEN is indirectly mediated by surgical route. Specifically, it is notable that Dreznik 2018 is the only study where the EEN-optimized cohort did not have a lower rate of laparotomy (vs laparoscopic surgery) than controls. Likewise, Dreznik 2018 is also the main study with IASC and SSI data more in line with the null hypothesis. Ideally, more consistent data on surgical approach would help in evaluating the primary factors through which improved preoperative nutritional status mediates reductions in certain postoperative complications.

Existing reviews analyzing the impacts of preoperative nutritional optimization on postoperative outcomes in IBD have also found that EEN or TPN reduce certain postoperative complications. Grass et al summarize that both EEN and TPN are efficient in decreasing postoperative morbidity but were unable to perform meta-analysis due to lack of matched control groups.¹¹ They conclude that nutritional support should be recommended to all high-risk patients with severe malnutrition, as well as low-risk patients with a nutritional risk score ≥3. Rocha et al remark in their systematic review that EEN was well tolerated by patients in all included studies, and that the 2 largest included studies found preoperative EEN to be an independent protective factor against infectious and noninfectious complications, as well as against anastomotic leaks and abscesses.³⁹ Finally, Brennan et al conducted a meta-analysis on 5 studies and found that with EEN and TPN data pooled together, the number needed to treat for preventing a single postoperative complication was 2.⁴⁰ They also determined that the rate of postoperative complications was significantly reduced in the nutritionally optimized cohort compared with the control (20% vs 61.3%); however, this was largely driven by contributions from the EEN data, and complications were only considered as an aggregate without stratification by complication type. The present study is the largest meta-analysis on the subject to date and includes pooled assessment of the impact of preoperative nutrition on specific complications and preoperative markers.

Further investigation will be important for determining the optimal patient selection protocol for nutritional support in patients with CD pending intestinal surgery. Gaps in understanding exist when considering which CD patient is most likely to benefit from preoperative nutritional optimization and which would most appropriately proceed more directly to surgery. Similarly, the optimum or minimum duration of nutritional support necessary for improved clinical outcomes cannot be elucidated from this meta-analysis. Most studies provided approximately 4 to 6 weeks of nutritional support. It is unknown whether shorter duration EEN than was administered in these studies may be adequate to derive the same clinical benefit. Understanding the optimal patient and duration for preoperative nutritional support is key, as the benefit of days or weeks of EEN may need to be weighed against the potential risks of delaying surgery. Furthermore, in CD patients for whom enteral nutritional support is contraindicated, such as obstruction or high output fistula, it remains to be determined whether parenteral nutrition can offer similar benefits to those we observed in our study.

There are several limitations to this meta-analysis that should be considered when interpreting the results. First, no randomized controlled trials were included, which limits the ability to assess causality given that observational studies are at risk of confounding and selection bias. Moreover, nearly all of the cohort studies were retrospective. Our risk of bias assessment for this meta-analysis identified moderate to high risk for bias in the many of the included studies with respect to matching between exposed and control groups. Medication exposure for the perioperative period, particularly with regards to corticosteroids, may have had an impact on certain postoperative outcomes. As noted, several of the studies weaned steroids as part of their EEN protocol.^{13,14,17,20,21,23,28} Although we could not adjust for this, it is likely part of the causal pathway rather than a confounder of the association between EEN and improved outcomes. Likewise, we could not adjust for biologic medication use, although given the long half-life, discontinuation of biologics during EEN would not be expected to meaningfully affect results. Similarly, we did not assess comorbidities such as diabetes that may have impacted the risk of surgical complications. The CD cohorts included people with different anatomic extent and behavior. Such heterogeneity in the patient population would be expected to possibly obscure a clinically important effect of EEN. Likewise, there were differences in the formulations of EEN used, as well as the duration of nutritional support, but our analysis showed very little between-study heterogeneity, suggesting that choice of formula and duration of preoperative EEN may be less important. Lastly, a few studies included patients in the control arm who received short attempts at nutrition optimization. This would likely have biased the results to the null and as such is unlikely to have changed the interpretation of the analyses related to infectious outcomes. Many of these limitations in the CD studies could be addressed in future prospective studies, particularly randomized trials.

Due to a lack of control-matched data for patients with UC, our study did not assess how preoperative nutrition affects postoperative outcomes in other forms of IBD. Such analysis will be important to perform as more data become available.

In conclusion, this meta-analysis represents the most current evaluation of the available data regarding preoperative optimization in CD. Pooled data from observational studies of preoperative EEN demonstrate strong and consistent associations with reduced intra-abdominal and skin infections during the postoperative period. Increased use of this strategy may improve outcomes for patients with CD who require bowel resection surgery; however, further research, particularly with randomized controlled trials, is indicated to more definitively test the efficacy of preoperative EEN and to determine the optimal patient and duration for therapy.

Supplementary Data

Supplementary data is available at Inflammatory Bowel Diseases online.

Funding

This work was supported by the Penn Center for Nutritional Science and Medicine and the Biomedical Data Science Core of the Center for Molecular Studies in Digestive and Liver Diseases (NIH grant P30DK050306).

Conflicts of Interest

J.D.L. has received research funding, served as a consultant, and participated in medical education programs from Nestle Health Science. The remaining authors have no conflict of interest to disclose.

References

Author notes