The routine use of drains for pelvic anastomosis is controversial. Thi
s study was undertaken to determine whether drainage alters leak rate,
aids in diagnosing a leak, and/or prevents the need for laparotomy wh
en a leak occurs. Records of 156 consecutive patients who underwent el
ective resection and pelvic anastomosis for cancer from 1986 to 1994 w
ere retrospectively reviewed. The patients were stratified into two pa
rent groups, I and II. One hundred and eleven Group I patients who rou
tinely had Jackson-Pratt 10-mm drains inserted were subdivided into su
bgroup I-A (n = 24) with proximal intraoperative diversion and subgrou
p I-B (n = 87), without diversion, Forty-five Group II patients routin
ely did not have drains inserted. They were also subdivided, into II-A
(n = 3) and II-B (n = 42), i.e., with and without simultaneous divers
ion, respectively. The overall leak rate was 5.1 per cent (8/156). Sub
groups I-A and II-A had leak rates of 8.33 per cent (2/24) and 0 per c
ent (0/3), respectively. The leak rates were 4.6 per cent (4/87) in su
bgroup I-B and 4.8 per cent (2/42) in subgroup II-B, with no significa
nt difference (P > 0.05). Also, no significant difference was noted wh
en the overall leak rate for diverted cases was compared with nondiver
ted ones. Of the four leaks in I-B, three occurred after the drains we
re removed; two of these required laparotomy, drainage, and diversion.
The fourth occurred with the drain in place, but failed to demonstrat
e feces or pus in the drainage fluid or prevent the need for laparotom
y. The presence of a drain did not affect the leak rate (P > 0.10). Dr
ain contents did not aid in diagnosing anastomotic dehiscence. Routine
prophylactic use of Jackson-Pratt drains seems unjustified because le
aks either occurred after the drains were removed or failed to prevent
the need for laparotomy.