Retrospective analysis of 30‐day mortality for emergency general surgery admissions evaluating the weekend effect. BJS 2016; 103: 1557-1565.

Published: 12th August 2016

Authors: I. J. D. McCallum, R. C. McLean, S. Dixon, P. O'Loughlin

Background

The weekend effect describes excess mortality associated with hospital admission on Saturday or Sunday. This study assessed whether a weekend effect exists for patients admitted for emergency general surgery.

Method

Data for emergency general surgical admissions to National Health Service hospitals in the Northern Deanery in England between 2000 and 2014 were collected, including demographics, co‐morbidities, diagnoses, operations undertaken and outcomes. The primary outcome of interest was in‐hospital death within 30 days of admission. Cox regression analysis was undertaken with adjustment for co‐variables.

Results

There were 12 100 in‐hospital deaths within 30 days of admission (3·3 per cent). The overall 30‐day mortality rate reduced significantly during the 15‐year interval studied, from 5·4 per cent (2000–2004) to 4·0 per cent (2005–2009) and 2·9 per cent during 2010–2014 (P < 0·001). There was no significant mortality difference for patients admitted at the weekend in adjusted Cox models (hazard ratio (HR) 1·00 for Saturday and 0·90 for Sunday, versus Wednesday). There was a significantly higher mortality for operations undertaken at the weekend (HR 1·15 for Saturday and 1·40 for Sunday; P = 0·021 and P < 0·001 respectively). The significantly increased mortality that was evident for emergency surgery at the weekend compared with weekdays in 2000–2004 (HR 1·46 for Saturday and 1·55 for Sunday; both P < 0·001); had reduced by 2010–2014, when the adjusted mortality risk was not significant (HR 1·18 for Saturday and 1·12 for Sunday).

Conclusion

During the past 15 years there has been a weekend effect in patients undergoing emergency general surgery based on day of operation, but not day of admission. Overall mortality for emergency general surgery has improved significantly, and in the past 5 years the increased mortality risk of weekend surgery has reduced.

Full text

Your comments

2 Comments

BJS admin

3 years ago

Dear Sir,

I read with interest the analysis by McCallum et al. (1). With all the negative media attention focusing on the ‘weekend effect’ it is encouraging to see improvements in outcomes have been made.

There are, however, important weaknesses not considered within the current study. The risk of mortality varies between hospitals, highlighted by the latest NELA report (2). It is curious why a two-level hierarchical model was not used to analyse outcomes with patients at ‘level 1’, nested within hospitals at ‘level 2’. The effects of patient and disease variables on mortality would have been considered independently of hospital variables, which themselves could have been quantified in this type of analysis.

With regards to the disease variables considered, a novel ‘clinical risk grouping’ was generated based on the crude mortality rate for the primary diagnoses. This has resulted in extremely heterogeneous groups of operative codes being included with a spectrum of predicted mortalities including low-risk patients undergoing appendectomy and perianal sepsis drainage to high-risk patients undergoing small bowel and liver transplantation. The latter are probably not true emergency admissions. As the numbers of individual cases is not presented the effects of such extremely high-risk operations on the analysis is not clear.

Ravinder Vohra
Trent Oesophago-Gastric Unit
City Hospital Campus
Nottingham University Hospitals NHS Trust
Hucknall Road
Nottingham NG5 1PB
UK
ravinder.vohra@nuh.nhs.uk

References:
1. McCallum I. J. D., et al. Retrospective analysis of 30-day mortality for emergency general surgery admissions evaluating the weekend effect. BJS, 2016; DOI:10.1002/bjs.10261
2. NELA Project Team. First Patient Report of the National Emergency Laparotomy Audit. Royal College of Anaesthetists: London, 2014.

BJS admin

3 years ago

Dear Sir,

We thank Mr Vohra for his interest in our work and comments. Examining any large and multi-layered dataset will always generate debate as to the optimal analysis strategy. For our analysis we preferred the simplest protocol to achieve a fair and accurate answer to our primary question regarding the presence or absence of a ‘weekend effect’. Whilst hierarchical modelling has advantages of improved prediction and analysis between groups, in this case hospital, we wished to present data which were generalizable to UK healthcare, we did not feel hospital level analyses added significantly to our results.

With reference to our clinical risk grouping, such analysis has been undertaken in previous studies examining weekend mortality (1,2). Such adjustment is designed to account for some variations in case mix and illness severity and so abrogates rather than exacerbates potential variations described in Mr Vohra’s correspondence. We disagree that relatively minor procedures such as drainage of perianal sepsis and appendicectomy do not reflect non-elective/emergency surgical care and the exclusion of such low-risk cases would create unquantifiable bias to the outcomes and decrease the generalisability of the findings.

IJD McCallum
RC McLean
P O’Loughlin

Department of Colorectal Surgery
Queen Elizabeth Hospital
Gateshead NE9 6SX
UK
paul.oloughlin@ghnt.nhs.uk

References:
1. Ruix M, Bottle A, Aylin PP. Global Comparitors Project: international comparison of 30-day mortality by day of the week. BMJ Quality and Safety. 2015; 10.1136/bmjqs-2014-003467.
2. Aylin P, Alexandrescu R, Jen MH et al. day of week of procedure and 30-day mortality fro elective surgery: retrospective analysis of hospital episode statistics. BMJ 2013; 10.1136/bmjf2424