Skip navigation

Antibiotic prophylaxis for colorectal surgery

This HTA review [1] has many interesting things in it for those interested in clinical trial design. Mostly it says that serious wound infections are reduced by the use of prophylactic antibiotics (no real surprise there), but despite having found over 147 trials it was hard to say which was best.


Four trials with 293 patients were comparisons of antibiotic prophylaxis with no antibiotic. Overall, 40% of patients had a serious wound infection without antibiotic compared with 13% who had a prophylactic antibiotic (Figure 1), giving a relative risk of 0.3 (0.2 to 0.5) and a NNT of 3.7 (2.7 to 5.8).
None of the trials was large, but infection rates with no treatment in the trials ranged from 32 to 58%.

Change in patterns of trials

The authors examined the trials they found in four-year periods from 1984 to 1995. Over that period the reporting of true randomisation doubled, their definition of outcomes improved, and the proportion following up patients for at least 28 days nearly doubled. Most interesting was trial size. In the early period well over half of all trials had fewer than 100 patients, while in the later period nearly 40% of trials included over 300 patients (Figure 2).


This review is interesting and well done. What will surprise many who glance at its well-presented tables listing the trials and odds ratios won't be the odds ratios because they almost always include 1. But looking at the incidence of serious wound infections, one can find, for the same antibiotic, infection rates as low as 2% and as high as 30%. Actually most studies with prophylactic antibiotics have quite low rates of serious wound infection, so it would have been interesting to have extracted particular points from those studies with higher rates.

There is a particularly interesting discussion which pulls together important aspects of effectiveness and policy, and the authors point out that for a reasonably moderate beneficial effect of one antibiotic regimen over another, a trial would need at least 400 patients. Anyone engaged in postoperative care should take note of this review, but it also has more general interest for those who think about clinical trials and meta-analyses.


  1. F Song, AM Glenny. Antimicrobial prophylaxis in colorectal surgery: a systematic review of randomised controlled trials. Health Technology Assessment 1998 2 (7).

previous or next story in this issue