Development and validation of a prognostic model for death 30 days after adult emergency laparotomy

The probability of death after emergency laparotomy varies greatly between patients. Accurate pre-operative risk prediction is fundamental to planning care and improving outcomes. We aimed to develop a model limited to a few pre-operative factors that performed well irrespective of surgical indication: obstruction; sepsis; ischaemia; bleeding; and other. We derived a model with data from the National Emergency Laparotomy Audit for patients who had emergency laparotomy between December 2016 and November 2018. We tested the model on patients who underwent emergency laparotomy between December 2018 and November 2019. There were 4077/40,816 (10%) deaths 30 days after surgery in the derivation cohort. The final model had 13 pre-operative variables: surgical indication; age; blood pressure; heart rate; respiratory history; urgency; biochemical markers; anticipated malignancy; anticipated peritoneal soiling; and ASA physical status. The predicted mortality probability deciles ranged from 0.1% to 47%. There were 1888/11,187 deaths in the test cohort. The scaled Brier score, integrated calibration index and concordance for the model were 20%, 0.006 and 0.86, respectively. Model metrics were similar for the five surgical indications. In conclusion, we think that this prognostic model is suitable to support decision-making before emergency laparotomy as well as for risk adjustment for comparing organisations.

Variation in Rates of Post-Mastectomy Radiotherapy Among Women with Early Invasive Breast Cancer in England and Wales: A Population-Based Cohort Study

AIMS

This study examined whether patterns of post-mastectomy radiotherapy (PMRT) among women with early invasive breast cancer (EIBC) varied within England and Wales and explored the role of different patient factors in explaining any variation.

MATERIALS AND METHODS

The study used national cancer data on women aged ≥50 years diagnosed with EIBC (stage I-IIIa) in England and Wales between January 2014 and December 2018 who had a mastectomy within 12 months of diagnosis. A multilevel mixed-effects logistic regression model was used to calculate risk-adjusted rates of PMRT for geographical regions and National Health Service acute care organisations. The study examined the variation in these rates within subgroups of women with different risks of recurrence (low: T1-2N0; intermediate: T3N0/T1-2N1; high: T1-2N2/T3N1-2) and investigated whether the variation was linked to patient case-mix within regions and organisations.

RESULTS

Among 26 228 women, use of PMRT increased with greater recurrence risk (low: 15.0%; intermediate: 59.4%; high: 85.1%). In all risk groups, use of PMRT was more common among women who had received chemotherapy and decreased among women aged ≥80 years. There was weak or no evidence of an association between use of PMRT and comorbidity or frailty, for each risk group. In women with an intermediate risk, unadjusted rates of PMRT varied substantially between geographical regions (range 40.3-77.3%), but varied less for the high-risk (range 77.1-91.6%) and low-risk groups (range 4.1-32.9%). Adjusting for patient case-mix reduced the variation in regional and organisational PMRT rates to a small degree.

CONCLUSIONS

Rates of PMRT are consistently high across England and Wales among women with high-risk EIBC, but variation exists across regions and organisations for women with intermediate-risk EIBC. Effort is required to reduce unwarranted variation in practice for intermediate-risk EIBC.

The Association Between Survival and Receipt of Post-mastectomy Radiotherapy According to Age at Diagnosis Among Women With Early Invasive Breast Cancer: A Population-Based Cohort Study

AIMS

Clinical trials of post-mastectomy radiotherapy (PMRT) for early invasive breast cancer (EIBC) have included few older women. This study examined whether the association between overall survival or breast cancer-specific survival (BCSS) and receipt of PMRT for EIBC altered with age.

MATERIALS AND METHODS

The study used patient-level linked cancer registration, routine hospital and radiotherapy data for England and Wales. It included 31 243 women aged ≥50 years diagnosed between 2014 and 2018 with low- (T1-2N0), intermediate- (T3N0/T1-2N1) or high-risk (T1-2N2/T3N1-2) EIBC who received a mastectomy within 12 months from diagnosis. Patterns of survival were analysed using a landmark approach. Associations between overall survival/BCSS and PMRT in each risk group were analysed with flexible parametric survival models, which included patient and tumour factors; whether the association between PMRT and overall survival/BCSS varied by age was assessed using interaction terms.

RESULTS

Among 4711 women with high-risk EIBC, 86% had PMRT. Five-year overall survival was 70.5% and BCSS was 79.3%. Receipt of PMRT was associated with improved overall survival [adjusted hazard ratio (aHR) 0.75, 95% confidence interval 0.64-0.87] and BCSS (aHR 0.78, 95% confidence interval 0.65-0.95) compared with women who did not have PMRT; associations did not vary by age (overall survival, P-value for interaction term = 0.141; BCSS, P = 0.077). Among 10 814 women with intermediate-risk EIBC, 59% had PMRT; 5-year overall survival was 78.4% and BCSS was 88.0%. No association was found between overall survival (aHR 1.01, 95% confidence interval 0.92-1.11) or BCSS (aHR 1.16, 95% confidence interval 1.01-1.32) and PMRT. There was statistical evidence of a small change in the association with age for overall survival (P = 0.007), although differences in relative survival were minimal, but not for BCSS (P = 0.362).

CONCLUSIONS

The association between PMRT and overall survival/BCSS does not appear to be modified by age among women with high- or intermediate-risk EIBC and, thus, treatment recommendations should not be modified on the basis of age alone.

Change in the Use of Fractionation in Radiotherapy Used for Early Breast Cancer at the Start of the COVID-19 Pandemic: A Population-Based Cohort Study of Older Women in England and Wales

Aims

Adjuvant radiotherapy is recommended for most patients with early breast cancer (EBC) receiving breast-conserving surgery and those at moderate/high risk of recurrence treated by mastectomy. During the first wave of COVID-19 in England and Wales, there was rapid dissemination of randomised controlled trial-based evidence showing non-inferiority for five-fraction ultra-hypofractionated radiotherapy (HFRT) regimens compared with standard moderate-HFRT, with guidance recommending the use of five-fraction HFRT for eligible patients. We evaluated the uptake of this recommendation in clinical practice as part of the National Audit of Breast Cancer in Older Patients (NABCOP).

Materials and methods

Women aged ≥50 years who underwent surgery for EBC from January 2019 to July 2020 were identified from the Rapid Cancer Registration Dataset for England and from Wales Cancer Network data. Radiotherapy details were from linked national Radiotherapy Datasets. Multivariate mixed-effects logistic regression models were used to assess characteristics influential in the use of ultra-HFRT.

Results

Among 35 561 women having surgery for EBC, 71% received postoperative radiotherapy. Receipt of 26 Gy in five fractions (26Gy5F) increased from <1% in February 2020 to 70% in April 2020. Regional variation in the use of 26Gy5F during April to July 2020 was similar by age, ranging from 49 to 87% among women aged ≥70 years. Use of 26Gy5F was characterised by no known nodal involvement, no comorbidities and initial breast-conserving surgery. Of those patients receiving radiotherapy to the breast/chest wall, 85% had 26Gy5F; 23% had 26Gy5F if radiotherapy included regional nodes. Among 5139 women receiving postoperative radiotherapy from April to July 2020, nodal involvement, overall stage, type of surgery, time from diagnosis to start of radiotherapy were independently associated with fractionation choice.

Conclusions

There was a striking increase in the use of 26Gy5F dose fractionation regimens for EBC, among women aged ≥50 years, within a month of guidance published at the start of the COVID-19 pandemic in England and Wales.

Prediction of long-term survival after gastrectomy using random survival forests

BACKGROUND

No well validated and contemporaneous tools for personalized prognostication of gastric adenocarcinoma exist. This study aimed to derive and validate a prognostic model for overall survival after surgery for gastric adenocarcinoma using a large national dataset.

METHODS

National audit data from England and Wales were used to identify patients who underwent a potentially curative gastrectomy for adenocarcinoma of the stomach. A total of 2931 patients were included and 29 clinical and pathological variables were considered for their impact on survival. A non-linear random survival forest methodology was then trained and validated internally using bootstrapping with calibration and discrimination (time-dependent area under the receiver operator curve (tAUC)) assessed.

RESULTS

The median survival of the cohort was 69 months, with a 5-year survival of 53.2 per cent. Ten variables were found to influence survival significantly and were included in the final model, with the most important being lymph node positivity, pT stage and achieving an R0 resection. Patient characteristics including ASA grade and age were also influential. On validation the model achieved excellent performance with a 5-year tAUC of 0.80 (95 per cent c.i. 0.78 to 0.82) and good agreement between observed and predicted survival probabilities. A wide spread of predictions for 3-year (14.8-98.3 (i.q.r. 43.2-84.4) per cent) and 5-year (9.4-96.1 (i.q.r. 31.7-73.8) per cent) survival were seen.

CONCLUSIONS

A prognostic model for survival after a potentially curative resection for gastric adenocarcinoma was derived and exhibited excellent discrimination and calibration of predictions.

O14: RANDOM FOREST MODELS FOR PREDICTING SURVIVAL AFTER OESOPHAGECTOMY

Introduction

For patients with oesophageal cancer, producing accurate prediction models for survival after oesophagectomy has proved challenging. We investigated whether Random Survival Forests (RSF), a novel machine learning method, could produce an accurate prognostic model for overall survival after oesophagectomy.

Method

The study used data from the National Oesophago-Gastric Cancer Audit and included patients diagnosed with oesophageal adenocarcinoma or squamous cell carcinoma between 2012 and 2018 in England and Wales and who underwent a curative oesophagectomy with adequate lymphadenectomy (15 LN) and survived to discharge (n=6198). Missing data was handled using multiple imputation and the data was split into training and validation cohorts. 13 variables were selected for inclusion using Random Forest variable importance and used to train the final model. The same variables were used to develop a traditional Cox regression model.

Result

Median survival was 53 months in both cohorts. The final RSF model had good discrimination in the validation cohort with a C-index of 0.757(0.755-0.759), exceeding the Cox model; 0.748(0.746-0.750). At 3 years post-surgery, overall survival was 56.2%. The RSF yielded a mean predicted survival of 55.8%(IQR 29.5%-81.7%) compared to 55.4%(40.0%-77.7%) for the Cox model. The most important variables were lymph node involvement and pT/ypT stage, however other variables including neoadjuvant treatment completion and surgical complications were also found to be important.

Conculsion

A Random Forest survival model provided better performance in predicting survival after curative oesophagectomy. This will allow more personalised predictions to be delivered clinicians and patients.

Take-home message

Random Forest survival models can accurately predict post-operative prognosis after oesophagectomy.

Surgical decisions in older women with early breast cancer: patient and disease factors

BACKGROUND

Studies reporting lower rates of surgery for older women with early invasive breast cancer have focused on women with oestrogen receptor (ER)-positive tumours. This study examined the factors that influence receipt of breast surgery in older women with ER-positive and ER-negative early invasive breast cancer .

METHODS

Women aged 50 years or above with unilateral stage 1-3A early invasive breast cancer diagnosed in 2014-2017 were identified from linked English and Welsh cancer registration and routine hospital data sets. Logistic regression analysis was used to evaluate the influence of tumour and patient factors on receipt of surgery.

RESULTS

Among 83 188 women, 86.8 per cent had ER-positive and 13.2 per cent had ER-negative early invasive breast cancer. These proportions were unaffected by age at diagnosis. Compared with women with ER-negative breast cancer, a higher proportion of women with ER-positive breast cancer presented with low risk tumour characteristics: G1 (20.0 versus 1.5 per cent), T1 (60.8 versus 44.2 per cent) and N0 (73.9 versus 68.8 per cent). The proportions of women with any recorded co-morbidity (13.7 versus 14.3 per cent) or degree of frailty (25 versus 25.8 per cent) were similar among women with ER-positive and ER-negative disease respectively. In women with ER-positive early invasive breast cancer aged 70-74, 75-79 and 80 years or above, the rate of no surgery was 5.6, 11.0 and 41.9 per cent respectively. Among women with ER-negative early invasive breast cancer, the corresponding rates were 3.8, 3.7 and 12.3 per cent. The relatively lower rate of surgery for ER-positive breast cancer persisted in women with good fitness.

CONCLUSION

The reasons for the observer differences should be further explored to ensure consistency in treatment decisions.

Long-term survival after endovascular and open repair of unruptured abdominal aortic aneurysm

Background

The aim of this study was to examine patterns of 10-year survival after elective repair of unruptured abdominal aortic aneurysms (AAAs) in different patient groups.

Methods

Patients having open repair or endovascular aneurysm repair (EVAR) in the English National Health Service between January 2006 and December 2015 were identified from Hospital Episode Statistics data. Postoperative survival among patients of different age and Royal College of Surgeons of England (RCS) modified Charlson co-morbidity score profiles were analysed using flexible parametric survival models. The relationship between patient characteristics and risk of rupture after repair was also analysed.

Results

Some 37 138 patients underwent elective AAA repair, of which 15 523 were open and 21 615 were endovascular. The 10-year mortality rate was 38·1 per cent for patients aged under 70 years, and the survival trajectories for open repair and EVAR were similar when patients had no RCS-modified Charlson co-morbidity. Among older patients or those with co-morbidity, the 10-year mortality rate rose, exceeding 70 per cent for patients aged 80 years. Mean survival times over 10 years for open repair and EVAR were often similar in subgroups of older and more co-morbid patients, but their survival trajectories became increasingly dissimilar, with open repair showing greater short-term risk within 6 months but lower 10-year mortality rates. The risk of rupture over 9 years was 3·4 per cent for EVAR and 0·9 per cent for open repair, and was weakly associated with patient factors.

Conclusion

Long-term survival patterns after elective open repair and EVAR for unruptured AAA vary markedly across patients with different age and co-morbidity profiles.

Effectiveness of a quality improvement collaborative in reducing time to surgery for patients requiring emergency cholecystectomy

Background

Acute gallstone disease is a high‐volume emergency general surgery presentation with wide variations in the quality of care provided across the UK. This controlled cohort evaluation assessed whether participation in a quality improvement collaborative approach reduced time to surgery for patients with acute gallstone disease to fewer than 8 days from presentation, in line with national guidance.

Methods

Patients admitted to hospital with acute biliary conditions in England and Wales between 1 April 2014 and 31 December 2017 were identified from Hospital Episode Statistics data. Time series of quarterly activity were produced for the Cholecystectomy Quality Improvement Collaborative (Chole‐QuIC) and all other acute National Health Service hospitals (control group). A negative binomial regression model was used to compare the proportion of patients having surgery within 8 days in the baseline and intervention periods.

Results

Of 13 sites invited to join Chole‐QuIC, 12 participated throughout the collaborative, which ran from October 2016 to January 2018. Of 7944 admissions, 1160 patients had a cholecystectomy within 8 days of admission, a significant improvement (P < 0·050) from baseline performance. This represented a relative change of 1·56 (95 per cent c.i. 1·38 to 1·75), compared with 1·08 for the control group. At the individual site level, eight of the 12 Chole‐QuIC sites showed a significant improvement (P < 0·050), with four sites increasing their 8‐day surgery rate to over 20 per cent of all emergency admissions, well above the mean of 15·3 per cent for control hospitals.

Conclusion

A surgeon‐led quality improvement collaborative approach improved care for patients requiring emergency cholecystectomy.

A step-wise approach to developing indicators to compare the performance of maternity units using hospital administrative data

Hospital administrative data are attractive for comparing performance of maternity units because of their often large sample sizes, lack of selection bias and the relatively low costs of accessing these data compared with conducting primary data collection. However, using administrative data to develop indicators can also present challenges including varying data quality, the limited detail on clinical risk factors and a lack of structural and user experience measures. This review illustrates how to develop performance indicators for maternity units using hospital administrative data, including methods to address the challenges that administrative data pose.

TWEETABLE ABSTRACT

How to develop maternity indicators from administrative data.

Population-based study of mortality and major amputation following lower limb revascularization

Background

The aim of this study was to estimate separate risks of major lower limb amputation and death following revascularization for peripheral artery disease (PAD) using competing risks analysis.

Methods

Routinely collected data from Hospital Episode Statistics (HES) were used to identify patients who underwent endovascular or open lower limb revascularization for PAD in England from 2005 to 2015. The primary outcomes were major lower limb amputation and death within 5 years of revascularization. Cox proportional hazards and Fine–Gray competing risks regression were used to examine the competing risks of these outcomes.

Results

Some 164 845 patients underwent their first lower limb revascularization for PAD during the study interval. Most were men (64·6 per cent) and the median age was 71 (i.q.r. 62–78) years. Following endovascular revascularization, the 5-year cumulative incidence of amputation was 4·2 per cent in patients with intermittent claudication and 18·0 per cent in those with a record of tissue loss. The corresponding rates were 10·8 and 25·3 per cent respectively after open revascularization, and 8·1 and 25·0 per cent after combined procedures. The 5-year cumulative incidence of death varied from 24·5 to 39·8 per cent, depending on procedure type. Competing risks methods consistently produced lower estimates than standard methods.

Conclusion

The 5-year risk of major amputation following lower limb revascularization for PAD appears lower than estimated previously. Patients undergoing revascularization for tissue loss and those who require an open procedure are at highest risk of limb loss.

Predicting 30-day mortality after hip fracture surgery: Evaluation of the National Hip Fracture Database case-mix adjustment model

OBJECTIVES

The National Hip Fracture Database (NHFD) publishes hospital-level risk-adjusted mortality rates following hip fracture surgery in England, Wales and Northern Ireland. The performance of the risk model used by the NHFD was compared with the widely-used Nottingham Hip Fracture Score.

METHODS

Data from 94 hospitals on patients aged 60 to 110 who had hip fracture surgery between May 2013 and July 2013 were analysed. Data were linked to the Office for National Statistics (ONS) death register to calculate the 30-day mortality rate. Risk of death was predicted for each patient using the NHFD and Nottingham models in a development dataset using logistic regression to define the models' coefficients. This was followed by testing the performance of these refined models in a second validation dataset.

RESULTS

The 30-day mortality rate was 5.36% in the validation dataset (n = 3861), slightly lower than the 6.40% in the development dataset (n = 4044). The NHFD and Nottingham models showed a slightly lower discrimination in the validation dataset compared with the development dataset, but both still displayed moderate discriminative power (c-statistic for NHFD = 0.71, 95% confidence interval (CI) 0.67 to 0.74; Nottingham model = 0.70, 95% CI 0.68 to 0.75). Both models defined similar ranges of predicted mortality risk (1% to 18%) in assessment of calibration.

CONCLUSIONS

Both models have limitations in predicting mortality for individual patients after hip fracture surgery, but the NHFD risk adjustment model performed as well as the widely-used Nottingham prognostic tool and is therefore a reasonable alternative for risk adjustment in the United Kingdom hip fracture population.Cite this article: Bone Joint Res 2017;6:550-556.

Association between age and access to immediate breast reconstruction in women undergoing mastectomy for breast cancer

BACKGROUND

National guidelines state that patients with breast cancer undergoing mastectomy in England should be offered immediate breast reconstruction (IR), unless precluded by their fitness for surgery or the need for adjuvant therapies.

METHODS

A national study investigated factors that influenced clinicians' decision to offer IR, and collected data on case mix, operative procedures and reconstructive decision-making among women with breast cancer having a mastectomy with or without IR in the English National Health Service between 1 January 2008 and 31 March 2009. Multivariable logistic regression was used to examine the relationship between whether or not women were offered IR and their characteristics (tumour burden, functional status, planned radiotherapy, planned chemotherapy, perioperative fitness, obesity, smoking status and age).

RESULTS

Of 13 225 women, 6458 (48·8 per cent) were offered IR. Among factors the guidelines highlighted as relevant to decision-making, the three most strongly associated with the likelihood of an offer were tumour burden, planned radiotherapy and performance status. Depending on the combination of their values, the probability of an IR offer ranged from 7·4 to 85·1 per cent. A regression model that included all available factors discriminated well between whether or not women were offered IR (c-statistic 0·773), but revealed that increasing age was associated with a fall in the probability of an IR offer beyond that expected from older patients' tumour and co-morbidity characteristics.

CONCLUSION

Clinicians are broadly following guidance on the offer of IR, except with respect to patients' age.

National trends in immediate and delayed post-mastectomy reconstruction procedures in England: A seven-year population-based cohort study

INTRODUCTION

Little is known about post-mastectomy reconstruction procedural trends in women diagnosed with breast cancer in England. Our aim was to examine patterns of immediate and delayed reconstruction procedures over time and within regions.

METHODS

Women with breast cancer who underwent unilateral index immediate or delayed post-mastectomy reconstruction between 2007 and 2014 were identified using the National Hospital Episode Statistics database. Women were grouped into categories based on the type of reconstruction procedure. Adjusted rates of implant and free flap reconstructions were then calculated across regional Cancer Networks using a regression model to adjust for age, disease, comorbidities, ethnicity, and deprivation.

RESULTS

Between 2007 and 2014, 21 862 women underwent immediate reconstruction and 8653 delayed reconstruction. Immediate implant reconstruction increased from 30% to 54%, and immediate free flap reconstruction from 17% to 21%. Adjusted immediate implant and free flap proportions ranged from 17 to 68% and 9-63%, respectively, across regions. Free flaps became more common in the delayed setting, rising from 25% to 42%. However, adjusted rates ranged from 23% to 74% across regions. Networks with high/low rates of free flaps for immediate tended to have high/low rates for delayed reconstruction.

CONCLUSION

There has been a substantial increase in the use of immediate implant reconstruction in England. In comparison, there has been an increasing use of autologous free flap reconstruction for delayed procedures. Significant regional variation exists in the type of reconstruction performed, and these patterns need to be examined to determine if variation is related to service provision and/or capacity barriers.

National trends and regional variation in immediate breast reconstruction rates

BACKGROUND

Previous studies have identified variation in immediate reconstruction (IR) rates following mastectomy for breast cancer across English regions during a period of service reorganization, a national audit and changing guidelines. This study analysed current variations in regional rates of IR in England.

METHODS

Patient-level data from Hospital Episode Statistics were used to define a cohort of women who underwent primary mastectomy for invasive or in situ breast carcinoma in English National Health Service (NHS) hospitals between April 2000 and March 2014. A time series of IR rates was calculated nationally and within regions in 28 cancer networks. Regional IR rates before and after the national audit were compared, using logistic regression to adjust for patient demographics, tumour type, co-morbidity and year of mastectomy.

RESULTS

Between 2000 and 2014, a total of 167 343 women had a mastectomy. The national IR rate was stable at around 10 per cent until 2005; it then increased to 23·3 per cent by 2013-2014. Preaudit (before January 2008), adjusted cancer network-level IR rates ranged from 4·3 to 22·6 per cent. Postaudit (after April 2009) adjusted IR rates ranged from 13·1 to 36·7 per cent, with 20 networks having IR rates between 15 and 24 per cent. The degree of change was not greatest amongst those that started with the lowest IR rates, with four networks with the largest absolute increase also starting with relatively high IR rates.

CONCLUSION

The national IR rate increased throughout the study period. Substantial regional variation remains, although considerable time has elapsed since a period of service reorganization, guideline revision and a national audit.

Population-based cohort study of the management and survival of patients with early-stage oesophageal adenocarcinoma in England

BACKGROUND

Until recently, oesophagectomy was the treatment of choice for early oesophageal cancer. Endoscopic treatment has been introduced relatively recently. This observational national database study aimed to describe how endoscopic therapy has been introduced in England and to examine the safety of this approach.

METHODS

A population-based cohort study was undertaken of patients diagnosed with oesophageal adenocarcinoma between October 2007 and June 2009 using three linked national databases. Patients with early-stage disease (T1 tumours with no evidence of spread) were identified, along with the primary treatment modality where treatment intent was curative. Short-term outcomes after treatment and 5-year survival were evaluated.

RESULTS

Of 5192 patients diagnosed with oesophageal adenocarcinoma, 306 (5·9 per cent) were considered to have early-stage disease before any treatment, of whom 239 (79·9 per cent of 299 patients with data on treatment intent) were managed with curative intent. Of 175 patients who had an oesophagectomy, 114 (65·1 (95 per cent c.i. 57·6 to 72·7) per cent) survived for 5 years. Among these, 47 (30·3 per cent of 155 patients with tissue results available) had their disease upstaged after pathological staging; this occurred more often in patients who did not have staging endoscopic ultrasonography before surgery. Of 41 patients who had an endoscopic resection, 27 (66 (95 per cent c.i. 49 to 80) per cent) survived for 5 years. Repeat endoscopic therapy was required by 23 (56 per cent) of these 41 patients.

CONCLUSION

Between 2007 and 2009, oesophagectomy remained the initial treatment of choice (73·2 per cent) among patients with early-stage oesophageal cancer treated with curative intent; one in five patients were managed endoscopically, and this treatment was more common in elderly patients. Although the groups had different patient characteristics, 5-year survival rates were similar.

Risk adjustment models for short-term outcomes after surgical resection for oesophagogastric cancer

BACKGROUND

Outcomes for oesophagogastric cancer surgery are compared with the aim of benchmarking quality of care. Adjusting for patient characteristics is crucial to avoid biased comparisons between providers. The study objective was to develop a case-mix adjustment model for comparing 30- and 90-day mortality and anastomotic leakage rates after oesophagogastric cancer resections.

METHODS

The study reviewed existing models, considered expert opinion and examined audit data in order to select predictors that were consequently used to develop a case-mix adjustment model for the National Oesophago-Gastric Cancer Audit, covering England and Wales. Models were developed on patients undergoing surgical resection between April 2011 and March 2013 using logistic regression. Model calibration and discrimination was quantified using a bootstrap procedure.

RESULTS

Most existing risk models for oesophagogastric resections were methodologically weak, outdated or based on detailed laboratory data that are not generally available. In 4882 patients with oesophagogastric cancer used for model development, 30- and 90-day mortality rates were 2·3 and 4·4 per cent respectively, and 6·2 per cent of patients developed an anastomotic leak. The internally validated models, based on predictors selected from the literature, showed moderate discrimination (area under the receiver operating characteristic (ROC) curve 0·646 for 30-day mortality, 0·664 for 90-day mortality and 0·587 for anastomotic leakage) and good calibration.

CONCLUSION

Based on available data, three case-mix adjustment models for postoperative outcomes in patients undergoing curative surgery for oesophagogastric cancer were developed. These models should be used for risk adjustment when assessing hospital performance in the National Health Service, and tested in other large health systems.

Socioeconomic deprivation and inpatient complication rates following mastectomy and breast reconstruction surgery

BACKGROUND

Socioeconomic deprivation is known to influence the presentation of patients with breast cancer and their subsequent treatments, but its relationship with surgical outcomes has not been investigated. A national prospective cohort study was undertaken to examine the effect of deprivation on the outcomes of mastectomy with or without immediate breast reconstruction.

METHODS

Data were collected on patient case mix, operative procedures and inpatient complications following mastectomy with or without immediate breast reconstruction in the English National Health Service between 1 January 2008 and 31 March 2009. Multivariable logistic regression was used to examine the relationship between patients' level of (regional) deprivation and the likelihood of local (mastectomy site, flap, flap donor and implant) and distant or systemic complications, after adjusting for potential confounding factors.

RESULTS

Of 13,689 patients who had a mastectomy, 2849 (20.8 per cent) underwent immediate reconstruction. In total, 1819 women (13.3 per cent) experienced inpatient complications. The proportion with complications increased from 11.2 per cent among the least deprived quintile (Q1) to 16.1 per cent in the most deprived (Q5). Complication rates were higher among smokers, the obese and those with poorer performance status, but were not affected by age, tumour type or Nottingham Prognostic Index. Adjustment for patient-related factors only marginally reduced the association between deprivation and complication incidence, to 11.4 per cent in Q1 and 15.4 per cent in Q5. Further adjustment for length of hospital stay, hospital case volume and immediate reconstruction rate had minimal effect.

CONCLUSION

Rates of postoperative complications after mastectomy and breast reconstruction surgery were higher among women from more deprived backgrounds.

Rates of subsequent surgery following endometrial ablation among English women with menorrhagia: population-based cohort study

OBJECTIVE

To assess the risk of further surgery amongst women who had an initial endometrial ablation (EA) for the treatment of heavy menstrual bleeding (HMB).

DESIGN

A retrospective cohort study using a national administrative database.

SETTING

Population-based study of hospital care in the English National Health Service.

POPULATION

A cohort of 114,910 women who had EA for HMB between January 2000 and December 2011.

METHODS

Multiple Cox regressions were performed to identify the risks of a further procedure, adjusted for age, social deprivation, year and type of initial EA, and presence of fibroids/polyps.

MAIN OUTCOME MEASURES

Time to repeat EA or hysterectomy after initial surgery.

RESULTS

Of 114,910 women undergoing EA, 16.7% had at least one subsequent procedure within 5 years. Higher rates of subsequent surgery were associated with younger age at initial EA, with women aged under 35 years having an adjusted hazard ratio of 2.83 (95% CI 2.67-2.99), compared with women aged over 45 years. Women who had radiofrequency ablation were less likely to have subsequent surgery as compared with first-generation techniques (HR 0.69, 95% CI 0.63-0.76). The rate of a subsequent hysterectomy within 5 years was 13.5%. Younger women (OR 0.59, 95% CI 0.51-0.69) and those who had balloon, microwave, or radiofrequency ablation were less likely to have a second EA procedure, rather than a hysterectomy.

CONCLUSIONS

One in six women have further surgery after EA for HMB, which is a higher rate than reported in clinical trials. This risk of further surgery decreases with age.

Impact of Caesarean section on subsequent fertility: a systematic review and meta-analysis

STUDY QUESTION

Is there an association between a Caesarean section and subsequent fertility?

SUMMARY ANSWER

Most studies report that fertility is reduced after Caesarean section compared with vaginal delivery. However, studies with a more robust design show smaller effects and it is uncertain whether the association is causal.

WHAT IS KNOWN ALREADY

A previous systematic review published in 1996 summarizing six studies including 85 728 women suggested that Caesarean section reduces subsequent fertility. The included studies suffer from severe methodological limitations.

STUDY DESIGN, SIZE, DURATION

Systematic review and meta-analysis of cohort studies comparing subsequent reproductive outcomes of women who had a Caesarean section with those who delivered vaginally.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Searches of Cochrane Library, Medline, Embase, CINAHL Plus and Maternity and Infant Care databases were conducted in December 2011 to identify randomized and non-randomized studies that compared the subsequent fertility outcomes after a Caesarean section and after a vaginal delivery. Eighteen cohort studies including 591 850 women matched the inclusion criteria. Risk of bias was assessed by the Newcastle-Ottawa scale (NOS). Data extraction was done independently by two reviewers. The meta-analysis was based on a random-effects model. Subgroup analyses were performed to assess whether the estimated effect was influenced by parity, risk adjustment, maternal choice, cohort period, and study quality and size.

MAIN RESULTS AND THE ROLE OF CHANCE

The impact of Caesarean section on subsequent pregnancies could be analysed in 10 studies and on subsequent births in 16 studies. A meta-analysis suggests that patients who had undergone a Caesarean section had a 9% lower subsequent pregnancy rate [risk ratio (RR) 0.91, 95% confidence interval (CI) (0.87, 0.95)] and 11% lower birth rate [RR 0.89, 95% CI (0.87, 0.92)], compared with patients who had delivered vaginally. Studies that controlled for maternal age or specifically analysed primary elective Caesarean section for breech delivery, and those that were least prone to bias according to the NOS reported smaller effects.

LIMITATIONS, REASONS FOR CAUTION

There is significant variation in the design and methods of included studies. Residual bias in the adjusted results is likely as no study was able to control for a number of important maternal characteristics, such as a history of infertility or maternal obesity.

WIDER IMPLICATIONS OF THE FINDINGS

Further research is needed to reduce the impact of selection bias by indication through creating more comparable patient groups and applying risk adjustment.

Reoperation rates after breast conserving surgery for breast cancer among women in England: retrospective study of hospital episode statistics

OBJECTIVES

To examine whether rate of reoperation after breast conserving surgery is associated with patients' characteristics and investigate whether reoperation rates vary among English NHS trusts.

DESIGN

Cohort study using patient level data from hospital episode statistics.

SETTING

English NHS trusts.

PARTICIPANTS

Adult women who had breast conserving surgery between 1 April 2005 and 31 March 2008.

MAIN OUTCOME MEASURE

Reoperation rates after primary breast conserving surgery within 3 months, adjusted using logistic regression for tumour type, age, comorbidity, and socioeconomic deprivation. Tumours were grouped by whether a carcinoma in situ component was coded at the time of the primary breast conserving surgery.

RESULTS

55,297 women had primary breast conserving surgery in 156 NHS trusts during the three year period. 11,032 (20.0%, 95% confidence interval 19.6% to 20.3%) women had at least one reoperation. 10,212 (18.5%, 18.2% to 18.8%) had one reoperation only; of these, 5943 (10.7%, 10.5% to 11.0%) had another breast conserving procedure and 4269 (7.7%, 7.5% to 7.9%) had a mastectomy. Of the 45,793 women with isolated invasive disease, 8229 (18.0%) had at least one reoperation. In comparison, 2803 (29.5%) of the 9504 women with carcinoma in situ had at least one reoperation (adjusted odds ratio 1.9, 95% confidence interval 1.8 to 2.0). Substantial differences were found in the adjusted reoperation rates among the NHS trusts (10th and 90th centiles 12.2% and 30.2%).

CONCLUSION

One in five women who had breast conserving surgery in England had a reoperation. Reoperation was nearly twice as likely when the tumour had a carcinoma in situ component coded. Women should be informed of this reoperation risk when deciding on the type of surgical treatment of their breast cancer.

Regional variation in use of immediate breast reconstruction after mastectomy for breast cancer in England

AIMS

English national guidelines recommend that breast reconstruction is made available to women with breast cancer undergoing mastectomy. We examined the use of immediate reconstruction (IR) across English Cancer Networks, who are responsible for the regional organisation of cancer services and ensuring equitable access to treatment.

METHODS

We analysed Hospital Episodes Statistics data for all women with breast cancer who underwent mastectomy in the English NHS between April 2006 and February 2009. IR rates were calculated for the 30 Networks. Multivariable logistic regression was used to adjust the rates for patient age, comorbidity, ethnicity and socioeconomic deprivation.

RESULTS

Of 44 837 mastectomy patients, 7375 (16.5%) underwent IR. The IR rate was highest in women under 50 years (32.7%) and lowest in women aged 70 years or over (1.5%), and was lower in women with more comorbidities. Unadjusted IR rates varied from 8.4% to 31.9% among the 30 Networks (p<0.001). Adjusting for their patient characteristics did not appreciably reduce Network-level variation, with adjusted IR rates still ranging from 8.0% to 29.4% (p<0.001). The risk-model also suggested that non-white women and those from more deprived areas were less likely to undergo immediate reconstruction.

CONCLUSIONS

There is substantial regional variation in immediate reconstruction use in England that is not explained by the characteristics of the local patient population. English Cancer Networks should act to reduce this variation. They should also examine why rates of reconstruction differ between particular patient groups.

Linking measures of health gain to explicit priority setting by an area health service in Australia

A demonstration project was undertaken to develop an integer programming model that could help a regional health authority to take into account data on service effectiveness when allocating resources to acute inpatient services. The model was designed to find the mix of services that would maximise health gain from the available resources, and so provide information that could be used to encourage hospitals to change their patient mix. It was developed in collaboration with an Area Health Service in New South Wales, Australia, with the aim of assessing its potential as a decision support tool. Acute inpatient services were categorised in the model using classes derived from the Australian National Diagnosis Related Groups (AN-DRG) classification and the classes developed by the Oregon Health Services Commission. Estimates for the effectiveness of each service was derived from the Oregon benefit data. Estimates of resource use were derived from AN-DRG data. The expected demand for each service was derived from local activity data. Various scenarios were developed to assess the potential of the model to support decision makers. These mimicked plausible policy options and tested the sensitivity of the results to changes in the data. The scenarios demonstrated the model could reveal the consequences of different policy options, but also suggested that the difference in the cost-effectiveness of services close to the margin would be small and so a rigid approach to priority setting is undesirable. Difficulties in developing the model also demonstrate that incorporating health gain data into resource allocation decisions will not be straight-forward for health planners.

Using simulation to educate hospital staff about casemix

When the Australian government funded a casemix development program, few hospital clinicians or staff knew much about casemix classifications like Diagnosis Related Groups (DRGs). Although the concepts behind casemix are essentially simple, it is not a trivial task to explain the logic used to assign patients to classes, or the use of casemix data for management or funding. Therefore, as part of a project to create educational material, a computer-based management game, built around a simulation model of a hospital, was developed. The game was designed for use in a workshop setting, to allow participants to test their understanding of the casemix information presented to them. The simulation mimicked the operation of a hospital, with a player taking the role of a hospital manager. It aimed to demonstrate how AN-DRGs might be used for funding; how patient costs are influenced by hospital activity; and how casemix data can assist in monitoring the use of resources. The game, called Dragon, proved to be very successful, and is now distributed as part of the National Casemix Education series.