The authors concluded that anterior resection for rectal tumours improved survival and decreased recurrence, compared with abdominoperineal excision. These conclusions do not seem to reflect the presented evidence and are unlikely to be reliable, as the search was limited, the review processes were unclear, the full quality assessment results were not reported, and the synthesis was limited.

To compare the outcomes of total mesorectal excision by anterior resection or by abdominoperineal excision for rectal cancer.

MEDLINE, EMBASE, The Cochrane Library, and the Internet were searched. Search terms were reported. The dates of the search were not reported and no language restrictions were reported.

To be eligible for inclusion, studies had to be conducted after 1994, when total mesorectal excision became widely practised, or explicitly mention the use of total mesorectal excision. They had to include both anterior resection and abdominoperineal excision, have a minimum of 100 patients, and report at least one outcome measure for both types of surgery. The outcomes of interest were circumferential resection margin status, tumour perforation rates, specimen quality, local recurrence, overall survival, cancer-specific survival, and disease-free survival. If two studies were reported by the same institution, the authors selected either the better quality study or the more recent publication.

All included studies were published between 1993 and 2010. Patient age and gender were not reported, but there were significant differences in age, gender split, and tumour height, diameter, and stage between groups within several studies.

The number of reviewers who selected studies was not reported.

The modified Newcastle-Ottawa Scale for non-randomised studies was used to assess their quality. The number of assessors was not reported.

The study and patient characteristics, outcome data and details on neoadjuvant and adjuvant therapies were extracted from the included publications. Details of how the data were extracted, such as how many reviewers, were not provided.

Studies were summarised in a narrative synthesis.

Twenty-four studies were included in the review, with 11,621 participants. Sample sizes ranged from 106 to 2,136. The authors judged all studies to be of good quality, with Newcastle-Ottawa Scale scores of six or more. Details for individual studies were not provided.

There was a positive association between abdominoperineal excision and an increased tumour perforation rate (five studies). The circumferential resection margin involvement rate was significantly higher following abdominoperineal excision, compared with anterior resection (six studies). For neoadjuvant and adjuvant therapies (10 studies) and local recurrence rates (19 studies) the results for the two types of surgery were inconsistent.

There was no increased risk of distant recurrence associated with either choice of surgical procedure (three studies). Four of seven studies reporting on disease-free survival, four of six studies reporting on overall survival, and three of seven studies reporting on cancer-specific survival favoured anterior resection; the other studies found no significant difference between the two procedures.

Anterior resection improved survival and decreased the recurrence rates, compared with abdominoperineal excision, but the tumours treated by abdominoperineal excision were lower and more locally advanced.

The research question and inclusion criteria were appropriate, but the decision to only include one study where multiple studies were conducted by the same institution, might mean that relevant data were excluded. Relevant sources were searched, but studies might have been missed if language restrictions were applied and if unpublished material was not sought. The patient characteristics were not reported, making it impossible to assess if they were appropriate and representative. Reviewer error and bias might have been present, as the details of the review processes were not reported.

A standardised scale was used to assess study quality, but the results for individual studies were not reported. There was variability between the studies and none of them were randomised, so a narrative synthesis, rather than a quantitative meta-analysis, seems reasonable. No numerical results were presented and the magnitude of any effects could not be assessed. The results table did not contain all the relevant information and the statements in the text could not be confirmed.

Studies might have been missed in the search; the review processes were unclear; the full quality assessment results were not reported; and the synthesis was limited. The authors' conclusions do not seem to reflect the presented evidence and are unlikely to be reliable.

The authors did not report any implications for practice and research.

Not stated.