Article Text

A multicentre cohort study assessing day of week effect and outcome from emergency appendicectomy
  1. Henry JM Ferguson1,
  2. Nigel J Hall2,
  3. Aneel Bhangu1,
  4. on behalf of the National Surgical Research Collaborative
  1. 1West Midlands Research Collaborative, Queen Elizabeth Hospital, Birmingham, UK
  2. 2University College London, London, UK
  1. Correspondence to Dr Aneel Bhangu, West Midlands Research Collaborative, Academic Department of Surgery, Queen Elizabeth Hospital, Birmingham, B15 2TH, UK; aneelbhangu{at}


Background There is evidence to suggest that patients undergoing treatment at weekends may be subject to different care processes and outcomes compared with weekdays. This study aimed to determine whether clinical outcomes from weekend appendicectomy are different from those performed on weekdays.

Method Multicentre cohort study during May–June 2012 from 95 centres (89 within the UK). The primary outcome was the 30-day adverse event rate. Multilevel modelling was used to account for clustering within hospitals while adjusting for case mix to produce adjusted ORs and 95% CIs.

Results When compared with Monday, there were no significant differences for other days of the week considering 30-day adverse events in adjusted models. On Sunday, rates of simple appendicitis were highest, and rates of normal (OR 0.62, 95% CI 0.42 to 0.90) and complex appendicitis (OR 0.65, 95% CI 0.46 to 0.93) lowest. This was accompanied by a 43% lower likelihood in use of laparoscopy on Sunday (OR 0.47, 95% CI 0.32 to 0.69), accompanied by the lowest level of consultant presence for the week. When pooling weekends and weekdays, laparoscopy use remained less likely at the weekend (OR 0.68, 95% CI 0.55 to 0.83), with no significant difference for 30-day adverse event rate (OR 1.01, 95% CI 0.80 to 1.29).

Conclusions This study found that weekend appendicectomy was not associated with increased 30-day adverse events. It cannot rule out smaller increases that may be shown by larger studies. It further illustrated that patients operated on at weekends were subject to different care processes, which may expose them to risk.

  • Surgery
  • Risk management
  • Audit and feedback
  • Performance measures
  • Patient safety
View Full Text

Statistics from


Recent publications have raised concern over surgical care provision at the weekend. Elective surgery from North America and England showed that mortality after elective surgery increased when procedures were performed later in the week or at the weekend.1 ,2 Similar relationships of this ‘weekend effect’ have been described relating to emergency admissions in general medical and aggregated surgical populations.3 ,4

Surgeons, hospital leaders and policy makers therefore require guidance as to the safety and suitability of weekend elective and emergency operating. There is a lack of data relating directly to surgical outcomes from weekend operating, especially for emergency admissions. Mortality is often an inadequate marker due to low frequency, whereas morbidity can be variable and high.5 With over 50 000 emergency appendicectomies performed in the UK every year, it is the most common emergency general surgical operation.6 It is a diverse clinical condition that has significant variations in presentation, management and outcome, and is therefore suitable as a high-variation quality target.

The potential reasons contributing to weekend effect so far identified include poorer quality of care and a difference in patient case mix admitted or operated on over the weekend.1 ,3 ,4 ,7–9 Current studies have used amalgamated data from several surgical procedures to test these hypotheses. Furthermore, there is little evidence surrounding the detailed quality of operative surgical provision at the weekend. Analysis in detail of a single surgical procedure allows these deficiencies to be addressed. This study aimed to determine whether surgically focused outcomes from weekend appendicectomy differed from those performed during the week.


Patients and data collection

The Multicentre Appendicectomy Audit collected data on prospective patients undergoing appendicectomy from 89 centres within the UK and 6 overseas centres. Complete methodology has been previously described.10 Briefly, design and data collection was trainee-led, protocol-driven, prospective and multicentred. The study period included May and June 2012, with 30-day follow-up for the last patient to the end of July 2012. Results for variation in the use of laparoscopy, normal appendicectomy rates and adverse events rates have been previously published without data related to day of week operating.10 Permission to perform the audit was granted from each individual site's Clinical Audit Department.

Management variables

Appendicectomy is the most commonly performed emergency surgical procedure. However, it is performed for heterogeneous presentations, and surgical approach is not standardised across the UK.10 Appendicitis is a diverse condition, ranging from mild inflammation to perforation and potentially fatal peritonitis with an associated systemic inflammatory response. Diagnosis is predominantly clinical, with preoperative radiology reserved for equivocal cases. In mild or early cases of appendicitis, conventional teaching is that antibiotics should only be given following the decision to operate, but there is increasing evidence that, in established sepsis, early antibiotics improve outcome.11 Decision on the timing of surgery is dependent on the perceived severity of sepsis by the decision-making surgeon and the patient's comorbidities. The ASA (American Society of Anaesthesiologists) grade quantifies comorbidities on a scale of 1 (fit and well) to 4 (a patient with comorbidities that are a constant threat to life). The decision between open or laparoscopic appendicectomy is dependent on the competence of the operating and responsible surgeons. Outcomes are at least equivalent between the two techniques, but recent evidence suggests laparoscopy is preferential and is therefore used as a quality marker. Conversion from laparoscopic surgery to conventional open surgery is dependent on a combination of the experience of the operating surgeon and the pathology encountered, with perforated or retrocolic appendices more likely to result in conversion.12

Intra-abdominal wash is important to reduce peritoneal contamination and the risk of postoperative intra-abdominal collections in cases where pus is present, but the necessity of its routine use is not agreed upon in mild cases.13 Postoperative histology is first used to determine whether appendicitis is present, and then to quantify the inflammation in the context of perforation or peritonitis (ie, complex pathology). Normal histology following appendicectomy is a recognised consequence of appendicitis and occurs in the order of 5–20% of cases.10 ,14

Outcome measures

The primary outcome measure for this study was the 30-day adverse event rate. Thirty-day adverse events were assessed using a composite outcome including wound infection, intra-abdominal abscess, readmission, unscheduled postoperative ultrasound scan or CT, further surgical or radiological intervention within 30 days, mortality and other postoperative adverse events. Wound infection was defined according to that provided by the Centers for Disease Control (definitions for Surgical Site Infection (SSI)).15 The secondary outcome measures were rates of laparoscopy, laparoscopic conversion and normal/complex histopathology (recorded from the original histopathology report). Within the limits of this audit of current practice, these events were detected through patient review prior to discharge, by clinical review if normal outpatient follow-up occurred by day 30 and/or administrative review of hospital systems (electronic and/or paper notes) for evidence of adverse events, readmission, reimaging and correspondence from external sites (eg, other hospitals).

Statistical analysis

The main exploratory variable was the specific day of week (Monday to Sunday). A second variable comparing weekday (Monday–Friday) to weekend (Saturday and Sunday) was also created. Data were analysed using SPSS V.19 (SPSS Inc, Chicago, Illinois, USA). Baseline characteristics were compared between groups using the χ2 test with a gamma correction for trend. To test the impact of day of week while taking into account case mix, binary logistic regression models were built for the outcome measures. These were presented in unadjusted (ie, univariable) and adjusted (ie, multivariable form). To account for clustering of patients within hospitals, a hierarchical two-level generalised linear mixed model was constructed, with patients being at the first level and the hospital at the second. Covariables included in the adjusted model included those relevant to the individual patient and that contained <5% missing data. These were age, gender and ASA) score. The results of logistic regression are presented as adjusted OR and 95% CI. If the 95% CI for the OR does not cross 1 and if p<0.05, the finding was considered to be significant.


Baseline characteristics

Overall, 73.5% (n=2444) of appendicectomies were performed during weekdays and 26.2% (872) over the weekend; in three cases the day was unknown. The spread of cases per day ranged from a lowest of 12.6% (404) on Monday to a highest of 16.2% (537) on Wednesday. Tables 1 and 2 show the baseline characteristics of the study population. Summaries of missing data are shown in the online supplementary table S1. The lowest rates of consultant presence in theatre (17.6%), CT (10.9%), night-time operating (16.8%), initial laparoscopy (60.0%), normal appendicectomy (15.5%) and complex histopathology (19.2%) were on Sunday. The lowest rate of preoperative/induction antibiotics (65.7%) and the highest rate of intraoperative/postoperative antibiotics (27.9%) were also on Sunday.

Table 1

Baseline characteristics of study population by day of procedure

Table 2.

Baseline characteristics of study population by weekday (Monday to Friday) versus (Saturday and Sunday)

Statistically significant differences in treatment variables (antibiotic use, imaging, timing of surgery, use of laparoscopy, consultant presence in theatre, skin closure) and outcome variables (histopathology, unplanned imaging, laparoscopic conversion) when stratified by day are shown in tables 1 and 2. In some cases, these differences were small, despite statistical significance, and non-linear. There were no statistically significant differences by day of the week for patient-related variables (age, gender, ASA grade), although when pooled, there were fewer of the youngest patients over the weekend (table 2).

Outcome measures

Unadjusted and adjusted models for individual day (vs Monday, table 3) and weekend (vs weekday, table 4) showed no significant difference with any day of week and rate of 30-day adverse events. There were no 30-day mortalities.

Table 3

Unadjusted and adjusted ORs of outcome measures for day of week of operation

Table 4

Unadjusted and adjusted ORs of outcome measures by weekday versus weekend

On Sunday, there was a 53% lower likelihood of undergoing a laparoscopic operation compared with Monday (adjusted OR 0.47, 95% CI 0.32 to 0.69). When pooling weekdays and weekends, there was significantly less use of laparoscopy during weekends (adjusted OR 0.68, 95% CI 0.55 to 0.83). There was no significant difference in the rate of laparoscopic conversion.

There was a significantly lower likelihood of normal appendicectomy (adjusted OR 0.62, 95% CI 0.42 to 0.90) and complex histopathology (adjusted OR 0.65, 95% CI 0.46 to 0.93) on Sunday. There were no significant differences in likelihood of normal appendicectomy (OR 0.83, 95% CI 0.67 to 1.02) or complex histopathology (OR 0.88, 95% CI 0.72 to 1.07) rates over the pooled weekend.


This study found no increase in the rate of 30-day adverse events for appendicectomies performed at weekends compared with weekdays. This finding is likely to be attributable to a preselection of less advanced/complex cases identified over the weekend, especially on Sunday. There were, however, treatment disparities, including reduced use of laparoscopy and consultant presence. Although these did not affect outcome in the present study, they have the potential to contribute to risk. Trends towards increased laparoscopic conversion on Sunday and over the pooled weekend that did not reach significance (OR 1.79, p=0.080 and OR 1.40, p=0.064) may still be clinically relevant. It is reasonable that these findings could be extrapolated to wider emergency general surgical practice, although validation is needed. It is also possible that larger studies may find smaller increases in weekend morbidity that were not detected by this study.

Previous mortality-based analyses of population-level data rely on grouping of multiple procedures to provide a cluster of operations with a large enough mortality rate to be included in the analysis. A study published in 2001 of 3 789 917 patients from Ontario, Canada, included 18.3% of patients undergoing surgery following weekday admission and 17.5% following weekend admission.4 For some serious conditions, a greater likelihood of death was seen following weekend admission, including ruptured aortic aneurysm (adjusted OR 1.28, 95% CI 1.13 to 1.46), but not acute hip fracture (OR 0.95, 95% CI 0.90 to 1.04). In 2007, a study of 188 212 patients undergoing surgery at 124 Veterans Affairs hospitals from the USA included specialties of general surgery, urology, orthopaedic surgery, vascular surgery, otolaryngology, neurosurgery, thoracic surgery and plastic surgery.1 This study showed that patients admitted postoperatively to regular wards on Friday were at higher risk of mortality compared with Monday through Wednesday (OR 1.17, 95% CI 1.05 to 1.26, p=0.003); a similar relationship to intensive care unit admission or discharge home was not shown. The debate has been continued with English hospital administrative data. In 2009, from a total of 4 317 866 emergency admissions, the adjusted OR of death was 10% higher (OR 1.10, 95% CI 1.08 to 1.11) in patients admitted at the weekend compared with those admitted during a weekday (p<0.001).3 In a similar analysis of 4 133 346 elective surgical admissions, the adjusted ORs of death were 44% and 82% higher in procedures carried out on Friday (OR 1.44, 95% CI 1.39 to 1.50) or a weekend (1.82, 95% CI 1.71 to 1.94) compared with Monday.2

The present study adds to this literature by analysing a single, commonly performed emergency procedure without the need to pool procedures or extrapolate outcomes. There are likely to be other factors relevant to explain the differences seen, especially when compared with elective surgery. These may include increased access to emergency theatres on Sunday, developed acute care pathways compared with weekend elective care or an adverse selection bias of elective cases for the end of the week. The inability to fully adjust models to take into account unobserved patient risk factors from previous elective studies has been noted.9 The present study shows that cases performed on Sunday are likely to be surgically easier. This is indicated by the higher rates of simple histopathology with consequently lower rates of challenging complex cases and diagnostically confusing normal cases. The reason for selection of weekend emergency cases in this study is uncertain, especially on Sundays, and may have been surgeon driven or due to patterns in patient presentation over the weekend. The association between lower normal rates and higher rates of perforated or gangrenous appendicitis has recently been challenged, suggesting the two states of appendicitis are independent pathophysiologies.16 However, in the present study, the finding of lower normal and complex rates on Sunday was likely due to a selection bias.

It has been further suggested that differences in elective and emergency weekend outcome are potentially due to poor care, as a result of fewer or inexperienced staff,1 ,3 ,7 ,8 or that the patient cohort is different at the weekend in terms of disease complexity.3 ,4 There is additional literature relating admission of elective surgery patients to intensive care units on weekends with higher mortality rates compared with weekdays despite accepted adequate staffing levels of these units, suggesting that staffing alone is not the sole issue.9 ,17 The lower weekend supervision rates shown in the present study may explain some of the differing weekend treatment variables seen, including reduced laparoscopy and increased intraoperative/postoperative antibiotics. However, in absolute terms, these differences were low; the lowest rate of consultant presence in theatre was 17.6% on Sunday compared with 23.2% on Monday (absolute difference 5.6%) and the highest level of 28.3% on Thursday (10.7% difference to Monday); the difference was 4.8% lower on weekend versus weekday. A previous analysis of the present cohort showed that hospital-related risk factors for reduced laparoscopy, increased laparoscopic conversion and higher adverse event rates include a higher rate of night-time operating, lower hospital volume and reduced consultant presence in theatre.10 It further showed from a questionnaire of included hospitals that only 68.3% of centres provided weekend out-of-hours laparoscopy if requested by trainees in any patient. This partly explains the reduced laparoscopy rates seen here. Thus, a reduction in weekday and weekend night-time operating may improve laparoscopy rates and reduce negative appendectomy rates, without increasing perforation rates.10 ,18 Any change in service should be prospectively analysed to ensure safety maintenance and data accuracy.

The strengths and limitations of this study's design, particularly in terms of data collection and patient identification, have been previously discussed.10 Briefly, the strengths lie in the multicentre nature of the study and the collection of detailed clinically related data, which is more relevant to practice than administrative data. Weaknesses of methodology include some adverse events being missed and patients presenting to other hospitals with complications. Within the confines of audit, this study emphasised proactive attempts to identify all adverse events, although active research-based follow-up of all patients was not performed. Although this may have underestimated true rates, the 12.5% overall 30-day adverse event rate, 3.5% wound infection rate and 2.7% pelvic abscess rate are in keeping with those found by a Cochrane review of 50 randomised controlled trials (including wound infection rate 5.4% (IQR 2.5–7.6%) and pelvic abscess rate 1.4% (0–2.3%)19), suggesting that the results are valid. The consequences of underreporting are likely to be spread across all days rather than focused on a particular day, evening out any remaining risk from this approach when considering the primary aim. The second key limitation is that this study may have lacked power to detect small differences in increases in adverse event rates. A posthoc power calculation revealed that 872 weekend patients matched 2:1 to weekday patients provides 87% power to detect a 36% increase in adverse events (12.5–17%). This study can claim that a rise of above this did not occur, but it is feasible that a rise below this may not have been detected. However, the similarity of the overall weekend/weekday results (12.5% and 12.6%, respectively) may indicate that the underlying rates will remain similar. Future large numbered studies derived from clinical data with complete clinical outcomes should be sought to confirm this. The third key limitation was the short ‘snapshot’ nature of the study period, which could potentially have resulted in a lack of generalisability to other seasons or years.

Additional weaknesses lie in the inability to ensure no patients were missed during identification and the presence of missing data. Missing data were not replaced, although ensuring minimum missing data tolerances for included covariables reduced its impact. Furthermore, the day of surgery was missing in only three patients. Subsequent multilevel modelling was used to take into account clustering of patients and subsequent treatment differences between hospitals.

This study found that weekend appendicectomy was not associated with increased 30-day adverse events compared with weekdays. This suggests that 7-day emergency cover within appendicitis care is consistent. It further illustrated that patients operated on at weekends were subject to different care processes, which may expose them to risk. This study cannot rule out smaller increases that may be shown by subsequent larger studies.


We thank Professor Dion Morton for his support and academic input.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Collaborators Local investigators (alphabetically by centre) of the National Surgical Research Collaborative were: I G Panagiotopoulou, N Chatzizacharias, M Rana, K Rollins, F Ejtehadi, B Jha (Addenbrooke’s Hospital, Cambridge); Y W Tan, N Fanous (Addenbrooke’s Hospital, Cambridge—Paediatric Surgery); G Markides, A Tan, C Marshal, S Akhtar (Airedale NHS Foundation Trust, Keighley); D Mullassery (Alder Hey Children’s Hospital Foundation Trust, Liverpool); A Ismail (Alexandra Hospital, Redditch); C Hitchins (Ashford and St Peter’s Hospitals); S Sharif, L Osborne (Barts Health NHS Trust, London); N Sengupta, C Challand, D Pournaras, K Bevan (Bedford Hospital NHS Trust, Bedford); J King (Birmingham Children's Hospital); J Massey, I Sandhu (Bradford Teaching Hospitals NHS Foundation Trust); J M Wells, D A Teichmann (Bristol Royal Hospital for Children); A Peckham-Cooper, M Sellers (Burton Hospitals NHS Foundation Trust); S E Folaranmi, B Davies (Central Manchester University Hospitals NHS Foundation Trust); S Potter, D Egbeare, C Kallaway, S Parsons, E Upchurch (Cheltenham General Hospital); A Lazaridis, D Cocker, D King, N Behar (Chelsea and Westminster Teaching Hospital, London); S P Loukogeorgakis (Chelsea and Westminster Hospital NHS Foundation Trust, London—Paediatric Surgery); R Kalaiselvan (Countess of Chester NHS Foundation Trust); S Marzouk, E J H Turner, S Kaptanis, V Kaur (Croydon University Hospital); G Shingler, A Bennett (Royal Glamorgan Hospital, Llantrisant); S Shaikh (Dewsbury District Hospital); M Aly, J Coad, T Khong, Z Nouman, J Crawford (Diana, Princess of Wales Hospital, Grimsby); P Szatmary (East Cheshire NHS Trust, Macclesfield); H West (Eastbourne District General Hospital); A MacDonald (Evelina Children's Hospital, Guy's and St Thomas’ NHS Foundation Trust, London); J Lambert, K Gash (Frenchay Hospital, North Bristol NHS Trust); K A Hanks, E Griggs, L Humphreys (Gloucester Royal Hospital); A Torrance, J Hardman, L Taylor (Good Hope Hospital, Sutton Coldfield); D Rex (Great Ormond Street Hospital NHS Foundation Trust; J Bennett, N Crowther (Great Western Hospital, Swindon); B McAree, S Flexer (Harrogate and District NHS Foundation Trust); P Mistry, P Jain, M Hwang (Heartlands Hospital, Birmingham); N Oswald, A Wells, H Newsome (Hinchingbrooke Health Care NHS Trust, Cambridgeshire); P Martinez (Hospital de San Bernabe, Berga, Spain); C A B Alvarez, J León (Hospital Santos Reyes, Aranda de Duero, Spain); D Carradice, R Gohil, M Mount (Hull and East Yorkshire Hospitals NHS Trust); A Campbell (Hull and East Yorkshire Hospitals NHS Trust—Paediatric Surgery); S Iype, E Dyson, T Groot-Wassink (Ipswich Hospital NHS Trust); A R Ross, C Jones (King George Hospital, Ilford); P Charlesworth (King's College Hospital, London); N Baylem, J Voll, T Sian, L Creedon (King's Mill Hospital, Mansfield); G Hicks (Leeds Teaching Hospitals NHS Trust); J Goring, V Ng (Leeds Teaching Hospitals NHS Trust—Paediatric Surgery); S Tiboni (Leicester Royal Infirmary); T Palser, B Rees, P Ravindra, C Neophytou (Lincoln County Hospital); H Dent, T Lo (Maidstone and Tunbridge Wells NHS Trust); L Broom, M O'Connell (Manor Hospital, Walsall); R Foulkes, D Griffith (Morriston, Wales); K Butcher, O Mclaren, A Tai (Musgrove Park Hospital, Taunton); H Yano (National Centre for Global Health and Medicine, Tokyo, Japan); H D T Torrance (Newham University Hospital); O Moussa, D Mittapalli, D Watt (Ninewells Hospital and Medical School, Dundee); S Basson (Norfolk and Norwich University Hospital—Paediatric Surgery); J Gilliland, S Pilgrim (Norfolk and Norwich University Hospitals NHS Trust); A Wilkins, J Yee (North Devon District Hospital, Barnstaple); H Cain, M Wilson, J Pearson, E Turnbull (North Tyneside General Hospital); A Brigic, N A Yassin, J Clarke, S Mallappa (North West London Hospitals NHS Trust, Northwick Park); P Jackson (Nottingham University Hospitals Trust—Paediatric Surgery); C Jones, B Lakshminarayanan (Oxford University Hospitals NHS Trust—Paediatric Surgery); A Sharma (Palmerston North Hospital, MidCentral District Health Board); R Velineni (Perth Royal Infirmary); K Fareed, G Yip (Peterborough City Hospital); A Brown, N Patel, M Ghisel (Plymouth Hospitals NHS Trust); N Tanner (Prince Charles Hospital, Merthyr Tydfil); H Jones, J Witherspoon, M Phillips (Princess of Wales Hospital, Bridgend); M F Ho, S Ng, T Mak (Prince of Wales Hospital, Hong Kong); N Campain, D Mukhey (Nottingham University hospitals NHS Trust); W K Mitchell, F Amawi, E Dickson, S Aggarwal (Royal Derby Hospital); L K Satherley, F Asprou (Royal Gwent Hospital, Newport); C Keys (Royal Hospital for Sick Children, Edinburgh); M Steven (Royal Hospital for Sick Children, Glasgow); M Johnstone (Royal Liverpool and Broadgreen University Hospital Trust); J Muhlschlegel (Royal United Hospital Bath NHS Trust); E Hamilton, J Yin (Royal Wolverhampton NHS Trust); M Dilworth, A Wright (Russells Hall Hospital, Dudley); P Spreadborough, M Singh (Sandwell and West Birmingham Hospitals NHS Trust Hospital); K Mockford, J Morgan (Scunthorpe General Hospital); W Ball, J Royle, J Lacy-Colson (Royal Shrewsbury Hospital); W Lai, S Griffiths, S Mitchell (South Devon Healthcare NHS Foundation Trust); C Parsons (Southampton University Hospitals NHS Trust—Paediatric Surgery); A S Joel, P F Mason, G J Harrison (Southport and Ormskirk NHS Trust); J Steinke, H Rafique (St George’s Hospital NHS Trust, London); C Battersby (St Helens and Knowsley Hospitals Trust, Merseyside); W Hawkins, D Gurram (The St George Hospital, South Eastern Sydney Illawarra Area Health Service, Australia); C A Hateley, A Penkethman, C Lambden (St Mary’s Hospital, Imperial Healthcare NHS Trust, London); A Conway, P Dent, D Yacob (Surrey and Sussex Healthcare NHS Trust); O A Oshin, A Hargreaves, G Gossedge (University Hospital Aintree, Liverpool); J Long, M Walls, K Futaba, T Pinkney, S Puig (University Hospital Birmingham Queen Elizabeth NHS Trust); A Boddy, A Jones (University Hospitals Bristol, NHS Foundation Trust); C Tennuci, N Battersby, R Wilkin, C Lloyd, E Sein (University Hospital of North Staffordshire, Stoke-on-Trent); K McEvoy, L Whisker, S Austin (Warwick Hospital, South Warwickshire NHS Foundation Trust); A Colori, P Sinclair, M Loughran, A Lawrence (Watford General Hospital); J Horsnell, J Bagenal (Weston Area Health NHS Trust); A Pisesky, S Mastoridis, K Solanki, I Siddiq (Whipps Cross University Hospital); L Merker, P Sarmah, C Richardson (Worcestershire Royal Hospital); D Hanratty, L Evans, M Mortimer (University Hospital of Wales, Cardiff and Vale University Health Board) Collaborative chairs: A Bhalla, D Bartlett, D Beral, N S Blencowe, J Cornish, J B Haddow, N J Hall, M Johnstone, S Pilgrim, S Strong, R Velineni West Midlands Research Collaborative Committee: P Marriot, R Vohra, A Patel, H Poon, E Hepburn.

  • Contributors The present study was devised by AB, who performed the statistical analysis. AB, HJMF and NH cowrote the manuscript. The local investigators identified patients and collected data at 95 local sites during the initial audit phase.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.