A Retrospective, Observational Study of Enteral Nutrition in Patients with Enteroatmospheric Fistulas

Patients and setting. The medical records of all patients with an OA treated from October 2012 to December 2014 in the Department of General and Visceral Surgery of the University Hospital, a 50-bed department in a maximum care level hospital specialized in gastrointestinal, hepatobiliary, and transplantation surgery, were reviewed. The study protocol was approved by the Institutional Ethics Board (IRB-No 50/15), appropriate to the observational character of the study. Patient informed consent was obtained for retrospective, anonymized data processing.

Data collected. Demographic and morbidity-related characteristics recorded were age, gender, index operation, prior operations, and OA grading according to the classification of Björck et al¹⁴: 1A — clean OA without adherence between bowel and abdominal wall or fixity  (ie, rigid retraction and immobility) of the abdominal wall; 1B — contaminated OA without adherence or fixity; 2A — clean OA with developing adherence or fixity; 2B — contaminated OA with developing adherence or fixity; 3 — OA complicated by fistula formation; 4 — frozen OA with adherent or fixed bowel. Preexisting and underlying diseases, body mass index (BMI), serum albumin, and serum creatinin were obtained. The time between the index operation, installation of the OA, formation of the EAF, initiation and termination of EN, and length of time the patient received NPWT were recorded. 

The number and location of the EAFs were noted; the location of the first fistula was determined from surgical records and/or radiological examinations.

Fistula volume was obtained by summing the liquids collected by NPWT systems, stoma appliances, and dressing material to an accuracy of 10 mL.

Outcome parameters recorded included length of stay, follow-up operations, changes in wound management (eg, frequency of dressing changes, technical modifications, wound revision performed under general anesthesia), fistula volume, medication (maximum and median per 24 hours), weaning from parenteral nutrition (PN), return to PN, catheter-associated complications, wound and fistula situation at discharge (ie, number and localization of fistulas, technique of supply, frequency of changes for dressings or appliances), readmission rate, continuity of fistulas, operative fistula reconstruction, and survival.

Data were retrieved by the authors in close cooperation with nurses from the wound care department at UKF from the electronic patient files (ORBIS^®, Fa. Agfa, Belgium; and Metavision^®, Fa. iMDsoft, Germany) and collected in a database only accessible to the authors. Patients were anonymized for further workup.

Statistical analysis. SPSS software, version 20.0 (IBM, Chicago, IL) was employed for statistical analysis. Primary descriptive statistics were used. The Mann-Whitney test was used to evaluate the changes in fistula volumes and BMI. Categorical data were compared using Fisher’s exact test. A P value <0.05 was considered statistically significant. 

Patient nutrition and wound protocols of care. Patients were allowed to drink clear liquids 6 hours after the index operation or the formation of the OA with/without an EAF; ventilated patients and patients with impaired swallowing capabilities or vigilance received EN via nasogastric (NG) or nasojejunal (NJ) tubes, with the intent to achieve sufficient enteral intake (25 kcal/kg/day).¹⁵ Assimilation of EN was monitored by body weight and serum albumin. The starter formula was Fresubin^® Energy Fibre Drink (Fresenius Kabi, Germany). After establishing a solid food diet, patients were supplemented with Fresubin^® Original, Energy Fibre, Protein Energy, or 2 Kcl (Fresenius Kabi, Germany) according to their needs. Nutrison^® Advanced (Nutrica, Germany) was administered via NG or NJ. PN was provided using Smofkabiven^® central (Fresenius Kabi, Germany).

All patients received NPWT using the ABThera^® System (KCI, an Acelity company, San Antonio, TX). The NPWT was adapted after formation of a resistant granulation layer on the abdominal viscus using the V.A.C Ultra^® System (KCI, an Acelity company, San Antonio, TX), and Mepithel^® (Mölnycke Health Care, Düseeldorf, Germany), a soft silicone layer, was used to protect the viscus and avoid adherence of the NPWT sponge. NPWT was provided at 75 mm Hg to 100 mm Hg continuous negative pressure. NPWT dressings were changed every 3 days according to manufacturer recommendations without restrictions to the minimal or maximal duration of NPWT.¹⁶ Dressing changes were provided more frequently if the dressings leaked. 

Changes in wound management, especially treatment changes during the return to EN, were monitored. Local complications were graded light when resulting in additional changes of wound dressing; moderate when resulting in treatment changes (eg, discontinuation of NPWT, change of treatment regimen, or changes in medication); and severe when additional surgical interventions or dressing changes under general anesthesia had to be performed. Cessation of EN caused by local worsening was classified as a moderate complication of EN. 

 Patient characteristics. A total of 10 patients (8 women, 2 men, median age 55.4 [range 44–71] years) with an EAF were treated. Indications for OA treatment included severe abdominal sepsis due to intestinal perforation, anastomotic dehiscence, and ileus (see Table 1). Two (2) patients were transferred from other hospitals after the index operation; 1 patient was treated before the transfer in a different department of UKF. Five (5) patients developed the fistula in the direct course of a malignant disease, 3 patients had a history of cancer, and the remaining 2 patients had no malignant disease. Comorbidities included diabetes (1 patient), hypothyreosis (3 patients), and Factor V Leiden mutation (1 patient). 

Surgical aspects and fistula characteristics. Among the 10 patients, 4 developed fistulas after small bowel resections, 2 underwent anterior rectum resection (1 for cancer, 1 for rectal prolapse), 1 underwent pylorus-preserving pancreatoduodenectomy, 1 had a right hemicolectomy, and 1 had a relaparotomy and revision of a colostomy and a multivisceral resection due to ovarian cancer. The median time from the index operation to the OA treatment was 11.5 (range 3–25) days, in which patients underwent repetitive reoperations, resulting in the OA. Two (2) patients were transferred with an OA to the authors’ department. In 4 patients, the OA was installed during reoperation because of persisting abdominal sepsis; in 2, this was a result of fascia and skin retraction. The median time from the index operation to the development of a fistula was 12.5 (range 3–25) days.

Seven (7) patients developed a complete fixed bowel package (referred to as a frozen abdomen) with grade 4 fistulas according to the Björck classification system.¹⁴ The remaining 3 patients had a grade 3 OA. Nine (9) patients had high-output fistulas (>500 mL/24 hours).¹⁷ 

Eight (8) patients developed multiple fistulas, leading to a median fistula rate of 2.5 (range 1–6 fistulas). In 7 patients, the first fistula was located between the flexura duodenojejunalis and the first 60 cm of jejunum (upper jejunal fistula [UJF]). No gastroatmospherical fistulas occurred. All patients with an UJF had more than 1 fistula. In these patients, the exact locations of the further fistulas could not be estimated via x-ray enteroclysis. 

All patients were treated for OA using NPWT as described for a median of 104 (range 14–337) days. In 2 patients, NPWT was continued until reconstruction of the gastrointestinal passage and in 3 patients until reconstruction of the abdominal wall. In the remaining 7 patients, NPWT was discontinued after the establishment of a surface that allowed a safe treatment with laparostomy or stoma pouches (SenSura^®, Assura^®, Coloplast^®, ClickXPro^® [Coloplast, Hamburg, Germany] or Wound Pouches^™ [EAKIN GmbH, Munich, Germany]).

Nutritional aspects. All patients with fistulas received EN during the hospital stay with a median time from the first appearance of a fistula to the first administration of EN of 11.5 (range 5–105) days and 32.5 days to a full-calorie enteral intake, respectively. One (1) patient successfully implemented a 25 kcal/kg/day oral diet within 5 days after the appearance of the OA with EAF and received no PN, whereas 9 patients received additional PN equivalent to their calculated needs. Septic patients with hypovolemia and severe hypoalbuminemia below 1.9 g/dL were substituted with human albumin 20% in the course of their treatment, making it impossible to use serum albumin levels for nutrition monitoring during the initial phase of fistula development. 

Initial weaning from PN was achieved in 6 of the 9 patients; of these, 4 with UJF required additional PN. PN eventually was reduced in volume or frequency in 3 of these patients. Patients without an UJF were successfully weaned from PN and remained on an oral diet (see Table 2). 

Outcome and complications. The median hospital stay following the index operation and formation of the EAF was 107.5 (range 42–371) days. A comparison between patient BMI at admission (BMI-A) and discharge (BMI-D) revealed a significant loss of weight in all patients (mean BMI-A 27.7 [range 23.4–39.9] versus BMI-D 21.4 [range 14.4–26.8]; P <0.001).

The wound and fistula condition improved as rated during the daily rounds and documented in the patient file (including photo documentation) in all but 1 patient after initiating EN. Median fistula volume as well as maximum volume per day did not change significantly after initiating EN (median volume: 970 mL [range 230–1820 mL] versus 720 mL [range 200–1370 mL]); maximum volume per 24 hours: 1880 mL [range 400–4600 mL] versus 2520 mL [range 700–6900 mL]; P = 0.25) (see Table 3). The maximum fistula volume per 24 hours increased in 8 of the 10 patients after initiating EN, whereas the median fistula volume decreased in all patients. No association between peak fistula volume and the initiation of EN could be observed.

In 6 of the 8 patients provided NPWT at the initiation of EN, the frequency of NPWT dressing changed from the original schedule of every third day. Five (5) needed NPWT changes less frequently than before EN (mean PN 2.35 [1.5–2.75] days versus EN 2.96 [2.7–3.3] days; P = 0.19). Only 1 patient (12.5%) needed the NPWT dressing changed more often (3 (EN) versus 2.3 (PN) days between dressing changes). 

Six (6) patients experienced local complications before initiating EN: 1 experienced light skin irritation, 4 had moderate complications (severe skin irritation, failure of NPWT, need for additional medication), and 1 had severe complications (additional operative intervention, dressing changes under general anesthesia). Of those, local complications improved (4 had light skin irritation with additional dressing changes and 2 had moderate to severe skin irritation and failure of NPWT) after initiating EN. One (1) patient (the same patient that required more frequent dressing changes) without local complications before EN developed moderate complications (severe skin irritation) after EN was started. 

Electrolyte disorders were noted in 6 patients, reversible liver failure in 3, and severe disturbance of blood sugar levels in 1 patient. Only in 1 case were severe electrolyte disorders clearly linked to PN. Central venous catheter thrombosis occurred in 2 patients, and 1 patient experienced catheter-associated problems requiring additional surgery that led to further complications.

Six (6) patients developed feeding intolerance following EN: diarrhea occurred in 3, vomiting in 4, and 2 suffered from both, but it was not necessary to stop EN. All 8 patients capable of oral food intake with additional PN reported a lack of appetite. No patient aspirated or had GI bleeding or ileus during the hospital stay. 

After initiating EN, 2 patients had an increase of serum creatinine linked to dehydration. Both patients needed parenteral fluids initially; 1 patient increased oral fluid intake and did not need additional parental fluids at discharge.

Seven (7) patients underwent reconstruction of the gastrointestinal tract and the abdominal cavity, 3 within the initial stay. One (1) patient refused a reconstructive operation; another was scheduled for 6 months after the index operation.

Two (2) patients died during the initial stay: 1 of a severe apoplexy, possibly linked to a thrombosis of an intraport catheter implanted for PN, and the other of septic complications leading to renal, kidney, and respiratory failure during the initial stay (hospital mortality rate of 20%). One (1) patient, initially discharged, died following the reconstruction operation due to thromboembolic complications caused by a Factor V Leiden thrombophilia and septic complications. During follow-up, another patient died after successful reconstruction and discharge due to hepatic, pulmonary, and peritoneal metastasis of her pancreatic cancer.

Although EAFs are a challenging and complex entity, publications on patients with such phenomena are limited to a few case reports or small series. Formation of an EAF represents 1 endpoint of an abdominal catastrophe caused by different index operations and diseases after variable timespans. This heterogeneity hinders a systematic, prospective scientific approach. Nutrition and wound management are interrelated problems in the critically ill EAF patient; multiple additional aspects influence outcomes.

Although a review by Lloyd et al¹⁸ outlines advantages of EN in patients with gastrointestinal fistula even if supplemental PN is necessary, the patient with an EAF and an OA has not been sufficiently addressed to date. Yin et al¹³ published their experience of successful EN in 9 patients with EAFs in OA after abdominal trauma, but in this series 3 patients had a gastroatmospheric fistula that could be bypassed by feeding tubes. Furthermore, in the series by Yin et al,¹³ patients with small bowel EAFs suffered from only 1 or 2 fistulas, allowing fistuloclysis in 3 patients, and the authors did not report the amount of residual jejunum. In addition, Yin et al¹³ reported a zero mortality rate; all of the patients analyzed developed EAF after traumatic abdominal injury, which might explain the low mortality rate compared to patients with ECFs and EAFs after abdominal sepsis. For the latter group of patients, a considerably higher mortality rate is described: Yuan et al¹² and de Aguilar-Nascimento et al¹⁹ reported an overall mortality rate of 31.7% and 46.4%, respectively. Long-term mortality in the current series was 40% (4/10), even though 1 patient died during a second stay for an elective reconstruction of the digestive tract and another after discharge due to recurrence of a pancreatic carcinoma.

Yin et al¹³ achieved complete EN in all patients. A small retrospective series by Teubner et al²⁰ reported a 91.7% rate of successful EN; fistuloclysis or bypass of the fistula was possible in these patients. Due to the number and location of the fistulas, the current authors could establish an effective EN in only 3 of 10 patients; 3 additional patients achieved sufficient nutrition using EN plus supplementary PN. In the current study, 70% of patients had an UJF and 80% had 2 or more (up to 6) fistulas, resulting in functional short bowel syndrome. As the amount of functional surface area of the small bowel could not be specified by radiological findings and/or intraoperative measurement, EN alone was attempted, even in patients with an UJF, but changed to combined EN and PN in cases of decreasing body weight or hypoalbuminemia. 

Control of intestinal fluids is crucial for wound management in patients with EAFs. As discussed in a review by Arebi and Forbes,²¹ fistula volume can increase through EN and the fluid might even become more corrosive and harmful to the skin and surrounding tissue. Because spontaneous closure of an EAF cannot be expected, short-term therapeutic approaches known from ECF treatment such as octreotide or “bowel rest” under total PN are not applicable.²² Therefore, the possible harm or benefits of EN in patients with EAFs needs to be further addressed. 

After initiating EN, an increase of the maximum daily fistula volume was observed and the median daily fistula volume decreased in all patients, although neither change was significant. Initiating EN was accompanied by an improvement of the wound situation (ie, fewer moderate and severe wound complications) in all but 1 patient and led to a decreased NPWT dressing change frequency. EN did not appear to negatively affect wounds in patients with complex EAF and OA. However, decreased dressing change frequency and severity of local complications also may be attributed to the progressive consolidation of the fistula and the OA related to changes in intestinal fluids from EN. As with other case series, the ability to draw firm conclusions about the current observations is limited. 

Two (2) factors are important for evaluating EN: food tolerance and adequate nutrition. A calculative, complete EN could be offered to all of the patients in this study, and 9 of 10 were able to take a solid diet orally. A high rate of food intolerance was observed after initiating EN as noted by nausea (40%) and vomiting and diarrhea (30%). These rates are high compared to data from the observational study by Reintam Blaser et al²³ involving 1712 critically ill patients (25% versus 3.3%, respectively), but they are in line with data published by Yin et al.¹³ In current patients, these symptoms were of only minor relevance for the recovery of the patients: interruption of EN was not necessary and diarrhea and nausea could be managed with medication.   

A conserved small bowel length of 60 cm to 100 cm is assumed to be necessary for the resorption of EN.^24,25 Five (5) patients in this study had a small bowel length not suitable for solely EN (<100 cm), classified as an UJF. These patients needed additional PN after discharge. On the other hand, evidence about the benefit of EN is clear, even if the surface for resorption is inadequate for full and sufficient EN: as described in a review by Suderaman et al,²⁶ EN is beneficial as noted by the preservation of the intestinal mucosal barrier and its immunologic function and to stimulate intestinal trophic hormones (eg, growth hormone, epidermal growth factor, enteroglucagon, and insulin-like growth factor I). These observations led to guideline recommendations of EN in critically ill patients as well as in patients with impaired small bowel length, insufficient for solely EN.²⁷ Independent of evidence-based guidelines, clinicians should consider prohibition or restriction of oral food intake may be a severe psychological burden for the patient. 

Catheter-associated complications were seen in 22.2% of the patients that received PN, twice as many as reported from a retrospective analysis²⁸ of patients receiving PN without EAFs and OA. A long-term observational study²⁹ reported 0.70 catheter occlusions per 1000 central venous catheters occurred with PN; the current data show 1.3 events per 1000 days. Because 1 patient with recurrent catheter occlusions suffered from Factor V Leiden mutation, it was not possible to determine whether this is a coincidence based on the occlusion rate or if increased thrombophilia may have occurred due to prolonged inflammation. 

Electrolyte disorder, liver and kidney dysfunction, and blood sugar disturbance (ie, symptomatic hyper- or hypoglycemia) were assessed in patients with PN in the initial phase of the OA and EAF formation. In this timespan, patients were complex and critically ill with impaired liver, kidney, and in some cases neurological and cardiopulmonary function, so it was not possible to establish a relationship between nutrition and these comorbidities. Only 1 patient had electrolyte disorders linked to PN (hyponatriemia) requiring additional intensive care.  

The overall mortality in the current study, including 1 patient who died from recurrence of a pancreas carcinoma, was 40%, similar to 44% in the largest published series³⁰ that included 18 patients with EAF. If limited to the initial stay following the index operation and formation of the EAF, the current authors observed a mortality rate of 20%. The high mortality and the long hospitalization reported in all EAF series underscores the severity of this disease pattern.

A strict differentiation between patients with ECFs and EAFs accompanied by an OA is necessary because the clinical course of these patients differs dramatically.^7,30 The incidence of EAF is low, but because the concept of OA treatment becomes more common among general surgeons as an accepted protocol of care for severe abdominal sepsis, an increase of EAFs is to be expected.^2,31 Data on these challenging patients is rare and hence valuable, despite the methodical deficits inherent in small, retrospective studies.

Data from this case series suggest EN does not exacerbate the wound condition of patients with OA/EAF. Median fistula volume did not increase significantly, and a potential change in fluid composition did not cause more or severe problems after initiation of EN. In this select group of patients, a sufficient EN was achieved only in patients with a lower jejunal fistula. No severe negative side effects of EN were observed. Although it may not provide fully effectual alimentation, the potential risk of implementing EN in patients with an EAF appears to be justifiable. n