Improvements in survival of gynaecological cancer in the Anglia region of England: are these an effect of centralisation of care and use of multidisciplinary management?

Authors


Dr RA Crawford, Consultant Gynaecological Oncologist, Department of Gynaecological Oncology, Addenbrookes Hospital, PO Box 242, Cambridge University Hospitals NHS Foundation Trust, Hills Road, Cambridge, CB2 0QQ, UK. Email robin.crawford@addenbrookes.nhs.uk

Abstract

Please cite this paper as: Crawford R, Greenberg D. Improvements in survival of gynaecological cancer in the Anglia region of England: are these an effect of centralisation of care and use of multidisciplinary management? BJOG 2012;119:160–165.

Objective  Our hypothesis is that the adoption of Department of Health (DH) guidance has led to an improvement in outcome in gynaecological cancer survival.

Setting  In 1999 the DH in England introduced the Improving Outcomes in Gynaecological Cancer guidance, advising case management by multidisciplinary teams with surgical concentration in specialist hospitals. This guidance was rapidly adopted in the East of England, with a population of 2.5 million.

Population  The population of the Anglia Cancer Network was approximately 2.3 million.

Methods  From 1996 to 2003, details of 3406 cases of gynaecological cancer were identified in the Anglia region of England. Survival analysis was performed by Cox proportional hazards regression, relative to cases diagnosed in 1996.

Main outcome measure  Primary endpoint was survival.

Results  The survival rates for cases diagnosed between 1996 and 1999 were broadly the same across the time period, with a marked improvement taking place in 2000, and continuing to 2003 (HR 0.71, 95% CI 0.64–0.79, comparing 2000–03 with 1996–99 diagnoses), for all gynaecological sites combined. Adjustment for treatments or method of case follow-up did not attenuate these improvements. There was a concurrent change towards major surgery being performed in specialist centres from 2000.

Conclusions  The adoption of the 1999 guidance on gynaecological cancer, which included multidisciplinary case management and centralisation of surgery, resulted in a marked step-change improvement in survival of gynaecological cancer in an area of eastern England in 2000.

Introduction

Cancer care in the UK has been under intense scrutiny for more than a decade. The Calman Hine report (1995) and the NHS Cancer Plan (2000) attempted to provide a strategy for improvement,1,2 as the data from Eurocare 4 had shown poorer rates of survival in England and Wales compared with better outcomes in mainland Europe.3 In 1999, the Department of Health (DH) produced good-practice guidance on commissioning cancer services for gynaecology, entitled Improving Outcomes in Gynaecological Cancer.4 The guidance was service based, requiring a reorganisation of services, and dealt with moderately rare cancers (ovarian, uterine, cervical and vulval) with marked survival differences. Best practice would include multidisciplinary team management and a concentration of surgical expertise in one hospital serving a population in excess of one million. It was felt that this centralisation was likely to improve survival and quality of life, although the evidence for this was limited.4 Better care and increased expertise are suggested to be related to a higher caseload,5,6 although this has not been defined in gynaecological cancer.

In the area now defined by the Anglia Cancer Network (2007), comprising the counties of Bedfordshire, Cambridgeshire, Norfolk and Suffolk, the guidance was rapidly implemented by the year 2000, with the centralisation of care predominantly shifting to Addenbrookes Hospital, serving a population of 1.5 million (receiving referrals from five other hospitals – Bedford, Hinchingbrooke, King’s Lynn, Peterborough and West Suffolk – as well as its own catchment area), and the Norfolk and Norwich Hospital, managing a population of approximately 800 000 (and serving an additional hospital, Great Yarmouth). Ovarian and endometrial cancers constitute 77% of the gynaecological cancer workload in England (http://www.statistics.gov.uk/downloads/theme_health/MB1-37/MB1_37_2006.pdf). In addition to the guidance given about commissioning services, in 2000 the DH published referral guidance for primary care for symptoms that were suggestive of cancer (the 2-week wait initiative).7 Hospital providers were expected to respond to the referral in a timely fashion, with a target of 2 weeks between the date of referral and the first appointment. By encouraging a more rapid referral, it was hoped to improve outcomes and to address the variation in access to diagnosis and treatment.

In this paper, we examine the survival data relating to gynaecological cancer in the Anglia region of the UK. Our hypothesis is that the centralisation of surgical care within a multidisciplinary team has led to an improvement in outcome.

Methods

Women diagnosed between 1996 and 2003 with invasive gynaecological cancer (International Statistical Classification of Diseases and Related Health Problems 10th revision, ICD10, site codes C51–C58) at the nine main NHS hospitals in the current Anglia Cancer Network were identified by ECRIC (the Eastern Region Cancer Registration and Information Centre). This represented a population of approximately 2.5 million people in the counties of Bedfordshire, Cambridgeshire, Norfolk and Suffolk, in the UK, and 5724 cases were identified, of which 97.3% were histologically confirmed. In order to make valid comparisons between outcomes, any cases that did not have tumour node metastasis (TNM) staging or histological grading were excluded. There were 3406 cases of known stage and grade selected for further analysis. At ECRIC, the TNM v5 stage was determined by a resident consultant oncologist based on pathological and clinical reports. TNM staging is used because it is applicable to most cancer sites; TNM and the International Federation of Gynecology and Obstetrics (FIGO) staging are identical for endometrial and ovarian cancers, which make up the majority of this cohort. The vital status of patients was determined at the end of February 2007, and was censored to 31 August 2006, 6 months earlier, to allow for any reporting delay. ECRIC actively followed up the vital status of each individual patient in this study in early 2007 by querying the NHS Strategic Tracing Service, so it is expected that these data are substantially complete and reliable. In order to ensure that any change in survival in more recent years was not an artefact of any long-tem change in registry follow-up methods, survival was also analysed with vital status censored 2 years after diagnosis. A similar assessment of survival was performed for colorectal cancer and bladder cancer over the same time period. This allowed a comparison with other cancers that were subject to a similar improved primary care referral pathway after 2000. The national guidance for colorectal cancer had been released earlier than that for gynaecology, and did not predicate a centralised service. Bladder cancer was recognised as benefiting from centralisation for major surgery, but this service reconfiguration was not implemented until after the study period.

Data elements recorded by ECRIC include hospital of diagnosis, age at diagnosis, pathological tumour size, numbers of nodes excised and positive nodes, condensed TNM v5 stage,8 histological grade, treatment type (surgery, radiotherapy, chemotherapy, use of taxanes in chemotherapy for ovarian cancer and hormone therapy) and index of multiple deprivation, based on the patients’ electoral ward of residence.

Statistical analysis

Data were analysed by Cox regression, which accounts for censoring in the data for those not dead at last follow-up.9 The end point was overall survival, regardless of cause of death. The range of analysis gave us 3 years prior to the changes and 3 years following the adoption of the policy. We first estimated survival differences for each of the eight years of diagnosis adjusted for age, index of multiple deprivation, and tumour stage and grade. Further analyses were adjusted for treatment factors (surgery, radiotherapy, chemotherapy, taxanes for ovarian cancer and hormone therapy), with the aim of identifying any change in treatment practices that might be responsible for differences in survival between years. This survival graph is shown for gynaecological cancer overall in Figure 1, and separately for endometrial cancer and ovarian cancer in Figure 2.

Figure 1.

 Survival function for all gynaecological cancers (ICD10 C51–58) by year of diagnosis, 1996–99 prior to IOG versus 2000–03 post IOG enactment in Anglia; adjusted for age, stage and grade.

Figure 2.

 (A) Survival function for endometrial cancer (ICD10 C54) by year of diagnosis: 1996–99 prior to IOG versus 2000–03 post-IOG enactment in Anglia; adjusted for age, stage and grade. (B) Survival function for ovarian cancer (ICD10 C56) by year of diagnosis: 1996–99 prior to IOG versus 2000–03 post-IOG enactment in Anglia; adjusted for age, stage and grade.

Results

Sixty percent of the cases had known stage and grade, and, to ensure that any comparisons were case-matched, only these were selected for the subsequent survival analysis. Although 85% of endometrial cancers were staged and graded, only 47% of ovarian cancers carried this information. These results are similar to those reported previously.5 Analysis of the entire group or the subset with accurate stage and histology data both showed a significant survival advantage following the reorganisation in 1999–2000, suggesting no selection.

The survival of cancers diagnosed between 1997 and 1999 was effectively identical to the survival of cancers diagnosed in the baseline year, 1996 (Table 1). Survival of cases diagnosed between 2000 and 2003 increased dramatically, and highly significantly, with a hazard ratio of 0.7 when compared with cancers diagnosed in the four earlier years. Figure 1 clearly shows the marked difference in survival between the two diagnosis periods. The differences in survival when vital status was censored 2 years after diagnosis were at least as marked, indicating that the improvements occurred soon after treatment. This effect can be seen in the rapid divergence of the survival curves before and after 2000. Further adjustment for surgery (yes/no), radiotherapy (yes/no), chemotherapy (yes/no) and hormone therapy (yes/no) had no effect on the differences between years. Figure 2A, B shows the relevant data for endometrial and ovarian cancers, respectively.

Table 1.   Results of multivariate Cox regression analysis, mutually adjusted for age, stage, grade and year of diagnosis in patients with gynaecological cancer (ICD10: C51–C58)
FactorCategoryRelative hazard and 95% confidence interval
Censored to 31 August 2006Censored to 24 months from diagnosis
Hazard ratioPHazard ratioP
AgeTrend per year1.04 (1.03–1.04)<0.0011.04 (1.03–1.04)<0.001
TNM stage11.00 (–)1.00 (–)
22.70 (2.27–3.22)<0.0013.48 (2.72–4.43)<0.001
35.26 (4.61–6.01)<0.0016.56 (5.42–7.94)<0.001
49.50 (8.01–11.27)<0.00113.21 (10.60–16.47)<0.001
Histological grade11.00 (–)1.00 (–)
21.67 (1.38–2.02)<0.0011.73 (1.30–2.31)<0.001
32.24 (1.84–2.72)<0.0012.23 (1.67–2.97)<0.001
44.28 (3.03–6.05)<0.0014.76 (3.08–7.37)<0.001
Year of diagnosis19961.00 (–)1.00 (–)
19971.07 (0.88–1.30)0.5121.09 (0.84–1.42)0.522
19980.92(0.76–1.12)0.4050.89 (0.69–1.16)0.392
19990.92 (0.75–1.14)0.4430.98 (0.75–1.29)0.887
20000.70 (0.57–0.86)0.0010.64 (0.49–0.85)0.002
20010.71 (0.57–0.87)0.0010.65 (0.49–0.85)0.002
20020.66 (0.53–0.82)<0.0010.60 (0.45–0.79)<0.001
20030.69 (0.56–0.86)0.0010.61 (0.47–0.80)<0.001
2000–03 compared with 1996–990.71 (0.64–0.79)<0.0010.63 (0.54–0.74)<0.001 

The number of cases that should have been treated in the specialist centre, as defined by the Improving Outcomes guidance (IOG), is compared with those treated in non-specialist centres, and is shown in Figure 3. Although the proportion of cases treated in specialist centres has increased each year, by far the largest change occurred in 2000, the year in which the IOG was introduced.

Figure 3.

 Hospital of surgery for cases of cervical, endometrial and ovarian cancer with stage and grade appropriate for treatment at specialist centres, as determined by the IOG.

Looking at the gynaecological service changes in isolation may have missed other drivers that affected cancer care across the spectrum. In 2000, the DH introduced advice for primary care about referral of patients with symptoms suspicious for any cancer.7 Gynaecological, colorectal and bladder cancers were all similarly subject to this more rapid primary care referral pathway. Although there was a general positive trend in survival year-on-year for colorectal cancer, there was little change in survival in bladder cancer: neither of these cancers showed any sign of the step-change improvement seen in gynaecology around the year 2000.

Discussion

The data shows that there has been a significant increase in survival of gynaecological cancers over the time period reviewed. In addition to the minor year-on-year improvements, there was a significant stepwise improvement in survival in 2000, which coincided with a major reorganisation in the provision of health care in the region.

ECRIC is known for the quality of its staging and follow-up data.10 Another indicator of the quality of these data is that over 97% of cases in this study were histologically confirmed. Taking into account cancer follow-up recording methodology changes in 2005, the survival improvement did not involve any methodological artefacts, and was a real finding. The registry used the TNM staging system, which is not typical for the gynaecological oncology community. However, this was applied by the registry personnel using the same criteria throughout the study, and did not bias the study.

It appears that the data support our hypothesis that the adoption of the guidance has led to a significant increase in survival. The majority of gynaecological cancer care relates to endometrial and ovarian cancer (77.4%). Prior to the organisational change, referral for possible adjuvant therapy would occur after surgery, depending on the gynaecologist’s view. Referral to the centre for high-risk endometrial cancer allowed for central pathology review at the tumour board, the use of state of the art imaging,11 and more extensive specialist surgery, all of which would contribute to better staging and more tailored adjuvant therapy. During the last decade we have seen an increase in staging with magnetic resonance imaging (MRI) in the referring hospital, although this was not routinely available in the study period. Suspected ovarian cancer cases were referred after triage with the ‘risk of malignancy index’. This has led to most of the advanced-stage cases of ovarian cancer being referred. In addition, cases of early cancer operated on at the unit hospital were discussed early, and consideration was given to restaging procedures and adjuvant chemotherapy.12 This allowed for specialist surgery by trained gynaecological oncologists at the centre, and review by the multidisciplinary team for those having non-specialist surgery. When we analysed the data, we saw that there has been no significant change in cervical cancer outcome. During the study period, the effect of the reorganisation only applied to a very small number of cases undergoing radical surgery in the periphery, as all the radiotherapy treatment was already centralised.

Analysing individual cancers, we found that this stepwise improvement remained for ovarian cancer, which presents late (more than 70% at FIGO stage 3 or 4) and endometrial cancer, which presents early, with 90% of women with post-menopausal bleeding and over two-thirds at FIGO stage 1. This is interesting as ovarian cancer presents later with a multitude of non-specific symptoms,13 whereas endometrial cancer has a significant symptom of postmenopausal bleeding that is well known and is usually seen early in the disease process. Thus, early referral resulting from a better recognition of symptoms was not important in the explanation of the improvement in outcome. We examined this in more detail by comparing other cancers. The change in survival for gynaecological cancer seen in 2000 may have been related to the increased global awareness of cancer symptoms, and to the introduction of guidance for the primary care team with respect to rapid referral pathways in the local hospital for symptoms suggestive of cancer. This guidance led to the establishment of rapid-access clinics for the speedy diagnosis of cancer. We examined colorectal and bladder cancer outcomes in the same time period. Colorectal cancer was selected as the service had already been reviewed and was a common cancer with a predominantly district-based service for surgery. Bladder cancer was chosen as subsequent guidance advised centralisation of surgical services for bladder cystectomy, although this was only introduced after 2005. There was no comparable stepwise change seen either in colorectal or bladder cancer, both of which were subject to the similar rapid referral guidance as gynaecology.

Over the time period from 1996 to 2003 there were changes to management that may have contributed to the changes in outcome observed. In ovarian cancer, the introduction of paclitaxel was considered an important step, although the ICON 3 study showed no difference in survival between single agent carboplatin and a combination of carboplatin and paclitaxel.14 We controlled the data for the type of chemotherapy received and found that the stepwise change persisted. This was despite the fact that many of the patients received paclitaxel in their management. The aim of surgery during this time period was to achieve maximal debulking of the tumour, either at the initial diagnosis or as an interval procedure after neo-adjuvant chemotherapy.15

In uterine cancer, the IOG suggested treatment at the district hospital for early stage disease (well or moderately differentiated endometrioid adenocarcinoma on biopsy, and no deep invasion on imaging). The Medical Research Council ASTEC (A Study in the Treatment of Endometrial Cancer) trial has shown that extended surgery does not improve survival.16 Patients for whom preoperative imaging suggested more extensive disease underwent more tailored extensive surgery at the centre, leading to upstaging.11 The reduction of the use of adjuvant radiotherapy for early low-risk cases, as described by the PORTEC (Postoperative Radiation Therapy in Endometrial Carcinoma) study,17 reflected the value of the multidisciplinary team using evidence-based protocols.

Advanced-stage cervical cancer (stages 2–4) had always been referred to the centre for radiotherapy, so there was no change in referral practice. In 2000 combination chemoradiotherapy was introduced as standard care for advanced cervical cancer, which accounts for a 3-year survival benefit of 5–8% (although this slight effect was not significant in the Anglia region because of the small numbers involved).18 Thus the effect of the IOG was only apparent for patients managed by radical surgery, and together with vulval cancer, these groups are too small for statistical assessment.

In conclusion, there was a highly significant step-change increase in survival in gynaecological cancers associated with the adoption of the 1999 national policy change. In Anglia, this was enacted swiftly, and women have benefitted from management within a multidisciplinary team and from access to specialist surgery. Although Rachet et al. include some of this data in their review of 1-year mortality, their conclusions are affected by considering heterogeneous data from cancer networks with patchy implementation of the IOG.19 This centralisation of care resulted in an improvement in 5-year survival from 58.6 to 68.6% for all gynaecological cancers that could be staged and graded (Figure 1). These changes have been most marked within endometrial and ovarian cancers.

Disclosure of interests

There are no conflicts of interest for either author.

Contribution to authorship

RC: concept, design, compilation and scrutiny of data, interpretation of data and writing the article. DG: concept, design, collection, compilation and scrutiny of data, analysis and interpretation of data, and writing the article.

Details of ethics approval

Ethics approval was not required for this paper.

Funding

There was no funding associated with this study.

Acknowledgements

The authors also wish to acknowledge the staff of the Eastern Cancer Registration and Information Centre for their work in collecting and evaluating the detailed information, which made this study possible.

Ancillary