Introduction Regional volume-outcome analysis of UK colorectal cancer registries has demonstrated survival benefit if major cancer resections are performed in high compared to low volume units. The Association of Coloproctology of Great Britain and Ireland (ACPGBI) and the National Bowel Cancer Audit (NBCA) have published NHS Trust and individual surgeon outcomes, allowing the hypothesis to be tested against a more representative sample.

Grouping individual hospital data under a Trust to reflect multidisciplinary team resources may over-estimate numbers of high volume colorectal units where surgeons operate in separate hospitals with different clinical teams. A hospital based volume-outcome analysis was conducted on the comprehensive dataset for patients treated by elective colorectal cancer resection with curative intent in England April 2010–March 2013.

Method Number of procedures, deaths, adjusted 90-day mortality rate (DMR) and Trust for each surgeon was extracted. The DMR had been adjusted for age, sex, ASA grade, TNM stage, admission mode, cancer site, co-morbidities, interaction between age and metastases using multivariate logistic regression. Trusts with multiple hospitals providing colorectal surgical services were contacted to confirm if consultant surgeons were site-specific, unless detailed in Trust literature. Trusts with site-specific services were sub-divided into individual hospitals. Institutions were categorised as low (0–209 procedures; n = 54), medium (210–309; n = 55) or high volume (310–708; n = 54).

Results 42,877 operations were performed in 163 centres at 143 NHS trusts. Procedures and surgeons per unit ranged from 30–708 (median 259) and 1–9 (median 4). 1,222 patients died within 90 days of resection, DMR 2.85. DMR for each volume group was: low 3.40, medium 2.93, high 2.61. 53.9% of surgeons experienced ≤1 death; 2.7% >5 deaths. 37% (20/54) of low volume units had lower mortality than the average of the high volume group, 2.61; 24% (13/54) of high volume units had higher mortality than the average of the low volume group.

Conclusion Unlike previous smaller studies our analysis fails to demonstrate a significance in correlation between low, medium or high volume units and DMR (r = –0.097). The majority of surgeons had good outcomes delivering major resectional elective cancer care. The argument for centralisation develops the idea that there is an unacceptable risk adjusted mortality rate; here expressed as 3.4%. The argument fails to recognise the relatively large proportion of low volume centres with better outcomes than high volume centres. 22% more perioperative deaths occur in 13 large volume units with ‘unacceptable’ mortality rates than in 24 small units with these same rates. These observations add insight to the reasons many surgeons fail to engage with centralisation agenda.

Disclosure of interest None Declared.