Postoperative complications following surgery for rectal cancer.

OBJECTIVE

This systematic review was designed to determine postoperative complication rates of radical surgery for rectal cancer (abdominal perineal resection and anterior resection).

SUMMARY OF BACKGROUND DATA

Lack of accepted complication rates for rectal cancer surgery may hinder quality improvement efforts and may impede the conception of future studies because of uncertainty regarding the expected event rates.

METHODS

All prospective studies of rectal cancer receiving radical surgery published between 1990 and August 2008 were obtained by searching Ovid MEDLINE, EMBASE, as well as ASCO GI, CAGS, and ASCRS meeting abstracts between 2004 and 2008. There was no language restriction. The outcomes extracted were anastomotic leak, pelvic sepsis, postoperative death, wound infection, and fecal incontinence. Summary complication rates were obtained using a random effects model; the Z-test was used to test for study heterogeneity.

RESULTS

Fifty-three prospective cohort studies and 45 randomized controlled studies with 36,315 patients (24,845 patients had an anastomosis) were eligible for inclusion. Most of the studies found were based in continental Europe (58%), followed by Asia (25%), United Kingdom (10%), North America (5%), and Australia/New Zealand. The anastomotic leak rate, reported in 84 studies, was 11% (95% CI: 10, 12); the pelvic sepsis rate, in 29 studies, was 12% (9, 16); the postoperative death rate, in 75 studies, was 2% (2, 3); and the wound infection rate, in 50 studies, was 7% (5, 8). Fecal incontinence rates were reported in too few studies and so heterogeneously that numerical summarization was inappropriate. Year of publication, use of preoperative radiation, use of laparoscopy, and use of protecting stoma were not significant variables, but average age, median tumor height, and method of detection (clinical vs. radiologic) showed significance to explain heterogeneity in anastomotic leak rates. Year of publication, study origin, average age, and use of laparoscopy were significant, but median tumor height and preoperative radiation use were not significant in explaining heterogeneity among observed postoperative death rates. With multivariable analysis, only average age for anastomotic leak and year of publication for postoperative death remained significant.

CONCLUSIONS

Benchmark complication rates for radical rectal cancer surgery were obtained for use in sample size calculations in future studies and for quality control purposes. Postoperative death rates showed improvement in recent years.