SEARCH

SEARCH BY CITATION

Keywords:

  • barium enema;
  • colorectal cancer;
  • medical error

Background:  The aim of the present study was to determine the cause and clinicopathological factors associated with the failure of barium enemas to detect colorectal cancers.

Methods:  A histopathological database was used to identify all patients with a diagnosis of colorectal cancer between 1991 and 1995. These records were matched with the records from patients who underwent barium enema examinations between 1990 and 1995. Those patients who had a colorectal cancer histologically diagnosed within 24 months of a barium enema in which no car­cinoma was seen, were identified. Where possible the radiology was reviewed. Failure to identify a carcinoma was then attributed to either simple failure, technical, interpretive or perceptive difficulties.

Results:  There were 967 patients with colorectal cancers treated in Christchurch Hospital during the study period 1991–1995. Matching of these patient details with all barium enema records revealed 313 patients who had barium enemas and histologically proven colorectal cancer. There were 21 (6.7%) patients in whom a carcinoma was missed. Of these, 18 had a barium enema within 8 months of surgery, and three were performed outside this timespan (15, 18 and 28 months, respectively). On review, 11 carcinomas could not be identified (nine due to technical error: poor coating (n = 1), overlapping loops (n = 3), single-contrast enema (n = 4), faecal residue (n = 1)); and seven could be seen on review of the films (two interpretation errors, one technical and perceptive error, and four perceptive errors). In three cases films could not be found for review. In 16 of the 21 missed lesions the patient had a double-contrast barium enema (DCBE) while five patients had single-contrast barium enema (SCBE). The site and stage of missed tumours is presented.

Conclusions:  The most common reason for missed tumours was technical. The percentage of missed tumours in each region of the bowel correlates with the known incidence of tumours in each region and with a normal Dukes stage distribution, except in the caecum where the number of missed lesions was higher than expected.