Gut, 1978, 19, 729-734

Factors which influenced postoperative complicationsin patients with ulcerative colitis or Crohn's diseaseof the colon on corticosteroidsJ. R. ALLSOP AND EMANOEL C. G. LEE

From the Department of Surgery, Massachusetts General Hospital, Boston, USA, and the GastroenterologyUnit, Radcliffe Infirmary, Oxford

SUMMARY A retrospective analysis was undertaken of the records of 107 patients with Crohn'sdisease of the colon or with ulcerative colitis who underwent 162 operations under steroid cover.The study revealed no correlation between steroid dosage and postoperative morbidity or mor-tality. The incidence of wound dehiscence and incisional hernia compared favourably with thereports of other unselected series of similar patients. Contamination did significantly influenceresults. Septic complications were more frequent when the operative field was contaminated andboth delayed wound healing and mortality were related to this sepsis. A 'clean and dirty' techniquewas effective in controlling contamination during elective bowel division but preoperative bowelperforation and accidental entry into the lumen of the bowel during dissection were potentiallyavoidable sources of contamination. Primary healing of the perineal wound after proctocolectomywas seldom achieved in contaminated patients where a drain tube was brought out through the mainperineal incision. When perineal sinuses or fistulae followed a proctocolectomy, patients withCrohn's disease had a significantly slower rate of healing than did patients with ulcerative colitis.However, there was no difference in the healing of abdominal wounds in relation to the primarypathology. Even abdominal incisions which were used on more than one occasion healed as well asthose which were used for the first time. A prophylactic antibiotic regime of either ampicillin ortetracycline offered little protection against postoperative sepsis. The organisms which caused suchinfections were often insensitive to the two antibiotics.

Corticosteroids are employed widely and oftensuccessfully to treat acute attacks of ulcerativecolitis. They are also effective in improving thesymptoms of patients with Crohn's disease of thecolon, but in both diseases their use is often con-sidered to be a major disadvantage if surgerybecomes necessary, because they may increase thepatients' susceptibility to infection (Fuenfer et al.,1973); they may retard healing (Erlich et al., 1973);and they may be the cause of other serious compli-cations such as peptic ulceration (Glenn and Grafe,1967). This paper reports the incidence and cause ofsuch post-operative complications in a group of 107patients who had surgery for ulcerative or Crohn'scolitis under corticosteroid cover between January1969 and December 1974.

Received for publication 2 March 1978

Mlethods

SUBJECTSA retrospective study of the case notes of coliticpatients who were treated surgically in Oxfordduring this time was undertaken. If the followingcriteria were met:

1. The prinary diagnosis was ulcerative colitis orCrohn's colitis.

2. Prednisolone in a dose of 20 mg or more perday was administered during the operation and hadbeen continued for at least five days postoperatively.

3. All patients had undergone a singlestageprocto-colectomy, or split ileostomy.

4. A standard technique had been employed foreach operation and the preoperative preparation andpostoperative management had been standardised.Some of the patients had undergone more than oneoperation, and subsequicnt abdominal operations in

729

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.19.8.729 on 1 August 1978. D

ownloaded from

730

these patients were also analysed provided thatsteroids had again been administered. Table 1 setsout the 162 operations in the 107 patients acceptedon these criteria. Variations of operative techniquewere minimal. The following points of surgicaltechnique were employed.

Table 1 Diagnoses and operations performedNumber Panprocto- Split Closure Other Totalof colectomy ileos- ileos- operationspatients tomy tomy

Ulcerativecolitis 56 48 18 3 4 73Crohn'sdisease 51 15 46 19 9 89Total 107 63 64 22 13 162

SURGICAL TECHNIQUEOne-hundred-and-fifty-nine of the 162 abdominalincisions were of the standard paramedian type, andthese were closed in the following manner. Theperitoneum and posterior rectus sheath was re-paired with a continuous suture of chromic catgutwith interrupted locking stitches of the same materialinserted every 8 to 10 cm. A continuous mono-filament nylon stitch was used for the anterior rectussheath. When there was a deep layer of subcutaneousfat it was approximated with chromic catgut. Theskin was closed with interrupted black silk sutures.In addition, in 14% of the patients, monofilamentnylon tension sutures were inserted through alllayers of the abdominal wall except the posteriorsheath and peritoneum.

Closure of the perineal wound after a procto-colectomy always involved approximation of thelevators with interrupted chromic catgut suturesand repair of the pelvic peritoneum with continuouschromic catgut. It was always possible to obtainthis closure without tension because a technique ofperimuscular rectal excision was used (Lee andDowling, 1972). Below the level of the levators,fascia covering fat of the ischiorectal fossa wasapproximated with chromic catgut and the skinclosed with interrupted silk sutures. In only fourpatients who had perianal infections or fistulae wasthe wound superficial to the levators left open andpacked. In all cases a tube drain with gentle suctionwas used in the perineum; before January 1972 thiswas inserted through the main perineal incision butafter this date the drain was introduced through aseparate stab incision in the ischiorectal fossa.Whenever it was necessary to divide or anastomose

bowel, a 'clean and dirty' technique was carried outin the following way. A separate instrument tray

J. R. Allsop and Emanoel C. G. Lee

with drapes was used to isolate the segment of bowelconcerned from the wound edges and from otherloops of intestine. After dividing the bowel betweenligatures, the ends were invaginated with a pursestring suture and then painted with a 10% povidone-iodine solution. In addition bowel ends destined tobe part of an operative specimen were enclosed in asurgical glove. At the end of the procedure the dirtvinstruments and drapes were discarded and allmembers of the operating team changed glovesbefore proceeding with the operation. To preventperitoneal or wound contamination when splitileostomies were being fashioned, the closed ends ofthe ileum were brought through the abdominal walland the final fashioning of stomas was left until themain incision had been closed and sealed with awaterproof dressing (Lee, 1J975).

Similarly, the closure of a split ileostomy wasperformed in several stages as follows: the bowelwas mobilised from surrounding skin and abdominalwall, closed with a catgut suture, invaginated with apurse string suture, and painted with 10% povidone-iodine. After this the abdomen was opened afterre-preparing and draping. The bowel ends were thendissected free of the remaining layers of the ab-dominal wall in preparation for the re-anastomosis,which was perfomed with separate instruments. Oncompleting the anastomosis instruments and gloveswere changed before closing the wound.

DEFINITIONS FOR PURPOSES OF THIS STUDYDelayed healing was defined as a failure to heal bythree weeks. Wounds reopened for any reason otherthan infection during the first three weeks wereassessed for healing from the date of the secondoperation and for the purposes of overall figures,the two operations were regarded as a single unit.Wound contamination was said to have occurred

if there had been accidental entry into the lumen ofthe bowel or if an abscess cavity or a fistulous tractwere entered. Preoperative bowel perforation wasalso defined as contamination. Elective division ofthe bowel using the 'clean and dirty' technique wasregarded as not having caused contamination.Wound infection was defined as the spontaneous

discharge or release of pus from a wound. Minorcellulitis or small stitch abscesses were not classifiedas wound infection.

Intra-abdominal sepsis was defined as a collectionof pus deep to the parietal peritoneum. A pelvicabscess, one form of intra-abdominal sepsis, onlywas diagnosed if it needed to be drained by ab-dominal incision. This was to distinguish it from themore common, though sometimes deeply situated,perineal sepsis which sometimes followed procto-colectomy.

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.19.8.729 on 1 August 1978. D

ownloaded from

Factors which influenced postoperative complications in patients with ulcerative colitis

Results

STEROID DOSEDuring the first 14 postoperative days patientsreceived between 170mg and 870mg prednisolone.The mean steroid dose given over this period was413mg (SD 125mg); therefore the average daily doseper patient was 30mg, although most patients hadhigher doses for the first few postoperative dayswhich were then progressively reduced.

MORTALITYThe overall mortality was 5 5% with the figures foreach type of opposition being proctocolectomy 8 %,split ileostomy 4 7 %, 'other' operations 7 7 %, andclosure of a split ileostomy nil.

In six patients the principal cause of death wasuncontrolled sepsis, with an associated small bowelfistula being a contributing factor in three of these.The remaining three patients died of pulmonaryembolism. In view of the causes of death it wasworth noting the high incidence of wound con-tamination amongst the patients who died. (Table 2).

Table 2 Mortality in relation to contamination

Deaths Survivors Total

Contamination 7 32 39No contamination 2 121 123Total 9 153 162

x2= 1209; n = 1; p <0-001.

WOUND DEHISCENCEThere were four cases of wound dehiscence in 162abdominal incisions. Three of these patients sub-sequently died while the sole survivor healedprimarily after resuture and did not develop a lateincisional hernia.

INCISIONAL HERNIAThere was one hernia through the main abdominalincision (0 7 %); one para-ileostomy hernia (0 6% ofstoma sites) and one hernia through a stoma sitewhich was closed at the time of restoring bowelcontinuity (1 6% of closed stoma sites).

POST-OPERATIVE ADHESIONSIt was the impression of the surgeons involved thatadhesions encountered at subsequent abdominalsurgery were less than might be anticipated by thetype of primary surgery which had been undertaken.Often adhesions were few in number and filmy innature. Small bowel obstruction from adhesionsoccurred in only three of the patients, twice after aproctolectomy and once after a split ileostomy. No

patients required reoperation for adhesions withinthe first four postoperative weeks.

SEPSISThirty-five per cent of the patients had septiccomplications at one or more sites in the post-operative period. The abdominal wound infectionrate was 19% and the rate for perineal wounds was57%. An intra-abdominal abscess occurred in 15%of the patients.

WOUND HEALINGEighty-two per cent of patients acheived primaryhealing of the abdominal wound and 27% ofperineal wounds healed primarily.

OTHER COMPLICATIONSFistulae of various types were not uncommon.These were: six small bowel to abdominal wallfistulae, five skin level fistulae in association with theileostomy stoma, two skin-to-skin fistulae, onevaginoperineal fistula, and one urethroperinealfistula which resulted from damage to the bulbousurethra during the perineal dissection of the rectum.

Mechanical small bowel obstruction followingtechnical errors occurred twice, once from failure toclose the 'lateral space' and the other from reversedsiting of the ileostomy and mucous fistula. Adendritic corneal ulcer and a haematemesis from apeptic ulcer were other complications and arementioned because of their possible relationship tosteroid administration.

FACTORS WHICH INFLUENCED MORBIDITYContaminationThis was associated with an increased risk of sepsisat all sites as shown in Table 3. The figures were allhighly significant.

Table 3 Incidence of sepsis in relation tocontamination

Abdominal Perineal Intra- Overall*wounds* woundst abdominal*

Contaminated 46% 81% 41% 77%Non-contaminated 10% 39% 6-5% 21%Significance x2 = 23-6 x' = 9-76 x2 = 25-3 x' - 38

p < 0-001 p < 0-01 p < 0-001 P<0-001

*39 contaminated patients. 123 non-contaminated patients.t27 contaminated patients. 36 non-contaminated patients.

Operation performedThe overall sepsis rate varied from 11 % for splitileostomy, to 59% for panproctocolectomy. It wasapparent that this reflected the incidence of con-tamination for the various operations.

731

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.19.8.729 on 1 August 1978. D

ownloaded from

J. R. Allsop and Emanoel C. G. Lee

Presence ofpreoperative stomaThere was a higher incidence of infection in non-contaminated patients when there was a stomapresent at the time of operation, 18% comparedwith 7% for patients without a stoma at the startof the operation.Primary diagnosisUlcerative colitis and Crohn's disease patients had thesame rate of abdominal wound infection when onlynon-contaminated cases were considered. Thissuggests that primary diagnosis per se does notinfluence liability to infection. However, ulcerativecolitis patients overall did have a higher infectionrate because they were the majority of patientshaving a panproctocolectomy; the operation withthe highest incidence of contamination. Primarydiagnosis, did, however, seem to influence the rateof perineal wound healing in the 58 panprocto-colectomy patients surviving longer than 12 months.As shown in Table 4 patients with Crohn's diseasefared less well than those with ulcerative colitis.The difference in healing after 12 months is signifi-cant at the 5% level x2 = 4X66, n = 1, P < 0 05).

Table 4 Healing ofperineal wound in relation toprimary diagnosis

Percentage of wounds healed at:

3 2 3 4 6 12weeks months months months months months

Ulcerativecolitis (45) 33 58 66 73 80 91Crohn's (13)disease 15 31 31 31 46 62

Perineal drain tubeThe siting of the perineal drain tube was changedfrom midline through the perineal wound to lateralin January 1972. Table 5 indicates the significantimprovement in primary healing of the perinealwound which followed the introduction of thelaterally placed drain. This finding occurred despitethe post-January 1972 group having double the rateof contamination of the earlier group with midlinedrains. In other respects, though, the two groupswere remarkably well matched.

Table 5 Influence of drain tube site on perinealwound healing

Primary Delayed Totalhealing healing

Midline drain 2 20 22Lateral drain 15 25 40Total 17 45 62

xI = 4 42, n = 1, P < 0-05.

Repeated use of abdominal incisionThe experience with incisions used on more than oneoccasion is set out in Table 6. There is no significantdifference between the results for incisions beingused for the first time and those being used for asecond, third, or fourth time.

Table 6 Wound healing in relation to number oftimes incision was used

No. of times No. of Burst Inc.sional Delayedincision used patients abdomens hernia healing

1 106 3 1 172 44 1 - 113 10 - - -4 2 - - -Total 162 4 1 28

Tension suturesThese were employed at surgeon's discretion and itwas therefore difficult to draw valid conclusionsabout their merits. There was, however, no tendencyfor them to be used more often in wounds beingclosed for a second, third, or fourth time.Prophylactic anitibioticsEighty-five per cent of patients were given anti-biotics at the-time of operation and for five to 10days after. Tetracycline or ampicillin was the anti-biotic chosen in all but two of these patients. Therate of sepsis in patients receiving prophylacticantibiotics was 37%, while the rate in patients notreceiving antibiotics was 23 ,. It was also noted thatbacteria isolated from infected wounds were almostinvariably resistant to the antibiotic previously given.Steroid doseThe total steroid dosage given to patients over thefirst 14 postoperative days did not vary significantlybetween patients with any given complication andthose free of that problem. Nor did the administra-tion of steroids over the days immediately precedingsurgery correlate with postoperative infection.Other factorsThe patients' sex, age, duration of symptoms beforeoperation, the fact that the patient had to have anemergency as compared with an elective operation,the presence of anaemia, and the presence of a lowserum albumin were all computed against results interms of sepsis and delayed wound healing. None ofthese factors had any demonstrable effect on theresults.

Discussion

CONTAMINATIONThis was the most important factor predisposing toinfection and, in turn, there was a significant

732

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.19.8.729 on 1 August 1978. D

ownloaded from

Factors which influenced postoperative complications in patients with ulcerative colitis

association with mortality and with failure to achieveprimary healing. The pathology was often such thatcontamination was unavoidable. However, con-tamination was potentially avoidable in three out ofthe seven contaminated patients who subsequentlydied. Similarly, accidental entry into the lumen of thelarge bowel during dissection accounted for 10 outof the 27 contaminated proctocolectomy patients.Our immediate concern therefore was whetherpermuscular dissection predisposed to accidentalopening of the rectum. Unfortu.nately, few writershave recorded their incidence of this problem, thoughBroader stated that the bowel was 'ruptured' duringdissection in 15 out of 54 panproctocolectomies whenthe traditional type of synchronous rectal dissectionwas being used (Broader, 1974). Thus on presentevidence it appears justifiable to persist with theperimuscular technique and thus obtain the benefitsof maximal preservation of pelvic parasympatheticnerves.The other source of potentially avoidable con-

tamination was preoperative large bowel per-foration. With diffuse soiling of the peritoneal cavityand localised or sealed perforations, there were fivesuch patients in this series of proctocolectomies. Inthe Oxford unit physician and surgeon make a jointdecision about the timing of surgical intervention inpatients with an acute attack of colitis. This dependson the clinical response to a five day course ofprednisolone given intravenously (Truelove andJewell, 1975).

In discussing potentially avoidable contamination,it should be recalled that the significant differencesin sepsis between contaminated and non-con-taminated groups was demonstrable despite electivebowel division being classed as no contamination.We believe that this implies the success of 'clean anddirty' technique in avoiding significant contamina-tion, especially as all patients in the series had theterminal ileum divided or reanastomosed. Thedifferent bacterial counts between small and largebowel contents would also offer an explanation.However, there is no agreement about what consti-titutes significant contamination at surgery. Itseemes appropriate therefore to take all possibleprecautions at operation including a meticulous'clean and dirty' technique in all cases.

ANTIBIOTICSThere was no significant difference in sepsis ratesbetween the patients who received antibiotics andthose who did not, but it was not unexpected to findthe higher sepsis rate in the former group. Animportant finding was that the majority of orga-nisms recovered from sites of infection were resistantto the antibiotic given prophylactically. Because of

these findings, we have altered our antibioticregimen to meet the need for adequate tissue levelsof the appropriate antibiotic at the time con-tamination is likely to occur (Burke, 1961). Theantibiotics we now use are metronidazole (Flagyl),lincomycin, and gentamicin which are bacteriocidalto the majority of gastrointestinal organisms, andthe period of their administration has been restrictedto minimise the emergence of resistant strains aswell as to reduce to a minimum the chance of thepatient developing a complication from the drugs.The results of surgery carried out under the newregimen will be reported later.

WOUND DEHISCENCEThe 2 5% incidence in this series compares favour-ably with the 3% incidence for an unselected seriesof laparotomies, excluding McBurney's incision forappendicectomy, reported by Efron (1965).

INCISIONAL HERNIABlomstedt and Welin-Berger (1972) prospectivelystudied the incidence of incisional hernia in a varietyof incisions used for cholecystectomy. Absorbablesutures were used in 90% of their patients and therewas an overall incisional hernia rate of 9-6%, theparamedian rate being 9 5%. We believe the verylow incidence of incisional hernias in this series(0.7% of paramedians) indicates that closure withmonofilament nylon will keep tissues suitablyapproximated for a sufficient period for soundhealing to occur, even when the healing process ispotentially impaired by steroid administration.Monofilament nylon provea once again to be aparticularly acceptable non-absorbable suture be-cause wound infections almost invariably settledcompletely without the foreign material having to beremoved.

DELAYED WOUND HEALINGWith the abdominal wound, delayed healing wasinvariably a consequence of infection, and therelationship of this to contamination has beendiscussed. The primary diagnosis did not influencethe rate of abdominal wound healing.

Perinealwound healing however is a more complexmatter. In all the proctocolectomy patients, closureof the pelvic and perineal wound was possible usingthe technique of perimuscular dissection (Lee andDowling, 1962). In the majority an inmersphinctericplane of dissection was used as reported sub-sequently by Lyttle and Parks (1977); the pre-requisite for primary healing was again the pre-vention of contamination and infection. Anotherfactor which influenced healing was the siting of theperineal drain, a laterally placed drain giving better

733

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.19.8.729 on 1 August 1978. D

ownloaded from

734 J. R. Allsop and Emanoel C. G. Lee

results than a midline drain. As a result of this studythe use of a perineal drain has been abandoned.Instead the pelvis is drained by one or two Redivacnegative pressure drains brought out through theanterior abdominal wall as suggested by Broaderet al. (1974). Their reported 59% incidence of pri-mary healing of the perineal wound with thistechnique in their colitis patients was better than theprimary healing rate in even the latter part of thisseries (35-5 %). Our recent results support this viewand will be reported later.

Finally, the primary diagnosis was a factor whichinfluenced the perineal healing. Patients withulcerative colitis fared better than those withCrohn's disease in two ways. Firstly, those withCrohn's disease had a higher incidence of contamina-tion from existing perineal pathology and thisprejudiced the chances of achieving primary healing.Secondly, for reasons unknown and unrelated towhether or not there was contamination at operation,patients with Crohn's disease had persistent perinealsinuses more commonly and for longer periods thatthose with ulcerative colitis.Although there was a trend for our younger

patients to have slower perineal healing, we were notable to confirm findings of Broader et al. (1974) thatpatients under the age of 35 years had a significantlygreater incidence of delayed perineal healing whencompared with those over the age of 35 years. One ofthe reasons why this occurs may be because theyoung patients who require a proctocolectomy oftenhave very severe disease.

POSTOPERATIVE ADHESIONSFrom the clinical point of view, the 3 2% incidenceof small bowel obstruction from adhesions, afterproctocolectomy, in this series, with a follow-upperiod ranging from 10 months to over six years issubstantially lower than that reported by Watts of3 to 4% per year for the first three years after opera-tion. A percentage of their patients also received pre-and postoperative steroids but dosage scheduleswere not indicated in their paper (Watts et al., 1966).None of our patients developed an intestinalobstruction because the pelvis had given way, arelatively common complication of abdomino-perineal resection of the rectum for carcinoma.

STEROID DOSEThe dose of steroids within the range of 170-870 mgof prednisolone used over the first 14 postoperativedays, had no demonstrable effect on morbidity ormortality. Unfortunately, a control group of patientshaving comparable surgery without any steroidadministration was not available for comparison.Our findings agree with those of Knudsen et al.

(1976) except that we did not note a higher morbid-ity and mortality in the patients who underwentemergency operations. The subject is an interestingone and our results of emergency treatment ofulcerative colitis will be reported at a later date.

Conclusion

Our findings suggest that the advantages of pred-nisolone administration for colitis are not offset byserious postoperative disadvantages if surgerybecomes necessary. The study indicated a number ofpoints where changes of operative technique mightbe advantageous and where the prophylactic use ofdifferent antiobiotics might reduce postoperativeinfection. These changes were introduced in 1975and the results are currently being analysed.

References

Blomstedt, B., and Welin-Berger, T. (1972). Incisionalhernias. Acta Chirurgica Scandinavica, 138, 275-278.

Broader, J. H., Masselink, B. A., Oates, G. D., and AlexanderWilliams, J. (1974). Management of the pelvic space afterproctectomy. British Journal of Surgery, 61, 94-97.

Burke, J. F. (1961). The effective period of preventive anti-biotic action in experimental incisions and dermal lesions.Surgery, 50, 161-168.

Efron, G. (1965). Abdominal wound disruption. Lancet, 1,1287-1291.

Ehrlich, H. P., and Hunt, T. K. (1968). Effects of cortisoneand vitamin A on wound healing. Annals of Surgery, 167,324-328.

Ehrlich, H. P., Tarver H., and Hunt, T. K. (1973). Effects ofvitamin A and glucocorticoids upon inflammation andcollagen synthesis. Annals of Surgery, 177, 222-227.

Fuenfer, M. M., Olson, G. E., and Polk, H. C. Jr (1975).Effect of various corticosteroids upon the phagocyticbactericidal activity of neutrophils. Surgery, 78, 27-33.

Glenn, F., and Grafe, W. R. Jr (1967). Surgical complica-tions of adrenal steroid therapy. Annals of Surgery, 165,1023-1034.

Knudsen, L., Christiansen, L., and Jarnum, S. (1976).Early complications in patients previously treated withcorticosteroids. Scandinavian Journal of Gastroenterology,11, suppl. 37, 123-128.

Lee, E. (1975). Split ileostomy in the treatment of Crohn'sdisease of the colon. Annals of the Royal College ofSurgeons of England, 56, 94-102.

Lee, E. C. G., and Dowling, B. L. (1962). Perimuscularexcision of the rectum for Crohn's disease and ulcerativecolitis. British Journal of Surgery, 59, 29-32.

Lyttle, J. A., and Parks, A. G. (1977). Intersphinctericexcision of the rectum. British Journal of Surgery, 64, 413-416.

Truelove, S. C., and Jewell, D. P. (1974). Intensive intra-venous regimen for severe attacks of ulcerative colitis.Lancet 1, 1067-1070.

Watts, J. M., De Dombal, F. T., and Goligher, J. C. (1966).Long-term complications and prognosis following majorsurgery for ulcerative colitis. British Journal ofSurgery, 53,1014-1023.

on Novem

ber 6, 2020 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.19.8.729 on 1 August 1978. D

ownloaded from