Original article Influence of hospital type on outcomes after oesophageal and gastric cancer surgery J. L. Dikken 1,4 , M. W. J. M. Wouters 1,5 , V. E. P. Lemmens 7 , H. Putter 2 , L. G. M. van der Geest 3 , M. Verheij 4 , A. Cats 6 , J. W. van Sandick 5 and C. J. H. van de Velde 1 Departments of 1 Surgery and 2 Medical Statistics, Leiden University Medical Center, and 3 Comprehensive Cancer Centre Leiden, The Netherlands, Departments of 4 Radiotherapy, 5 Surgery and 6 Gastroenterology and Hepatology, Netherlands Cancer Institute – Antoni van Leeuwenhoek Hospital, Amsterdam, and 7 Comprehensive Cancer Centre South, Eindhoven, The Netherlands Correspondence to: Professor C. J. H. van de Velde, Department of Surgery, K6-R, Leiden University Medical Centre, PO Box 9600, 2300 RC Leiden, The Netherlands (e-mail: [email protected]) Background: Outcomes after oesophagectomy and gastrectomy vary considerably between hospitals. Possible explanations include differences in case mix, hospital volume and hospital type. The present study examined the distribution of oesophagectomies and gastrectomies between hospital types in the Netherlands, and the relationship between hospital type and outcome. Methods: Data were obtained from the nationwide Netherlands Cancer Registry. Hospitals were categorized as university hospitals (UH), non-university teaching hospitals (NUTH) and non-university non-teaching hospitals (NUNTH). Hospital type–outcome relationships were analysed by Cox regression, adjusting for case mix, hospital volume, year of diagnosis and use of multimodal therapies. Results: Between 1989 and 2009, 10 025 oesophagectomies and 14 221 gastrectomies for cancer were performed in the Netherlands. The percentage of oesophagectomies and gastrectomies performed in UH increased from 17·6 and 6·4 per cent respectively in 1989 to 44·1 and 12·9 per cent in 2009. After oesophagectomy, the 3-month mortality rate was 2·5 per cent in UH, 4·4 per cent in NUTH and 4·1 per cent in NUNTH (P = 0·006 for UH versus NUTH). After gastrectomy, the 3-month mortality rate was 4·9 per cent in UH, 8·9 per cent in NUTH and 8·7 per cent in NUNTH (P < 0·001 for UH versus NUTH). Three-year survival was also higher in UH than in NUTH and NUNTH. Conclusion: Oesophagogastric resections performed in UH were associated with better outcomes but, owing to variation in outcomes within hospital types, centres of excellence cannot be designated solely on hospital type. Detailed information on case mix and outcomes is needed to identify centres of excellence. Presented to a meeting of the Dutch Surgical Society, Ede, The Netherlands, November 2011 Paper accepted 27 March 2012 Published online in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.8787 Introduction Long-term survival for patients with resectable oesophageal and gastric cancer is low in the Western world. The 5-year overall survival rate is below 25 per cent after oesophagec- tomy and less than 40 per cent after gastrectomy 1,2 . Both are high-risk operations with correspondingly high post- operative mortality rates 3,4 . Both postoperative mortality and long-term survival after oesophagogastric cancer surgery can be improved by performing these complex procedures in centres with sufficient experience and high annual volumes 3,5 . An exact cut-off value that defines high-volume surgery has not, however, been established. In a recent survey of all oesophagectomies and gastrectomies performed in the Netherlands between 1989 and 2009, oesophagectomies carried out in high-volume hospitals (more than 20 pro- cedures per year) were associated with lower postoperative mortality and improved survival rates compared with those performed in low-volume hospitals. No such relationship was found after gastrectomy, but the number of high- volume hospitals was small 6 . Although hospital volume can be used as a proxy for quality of care, another approach is to compare outcomes by type of hospital in which the surgery takes place 7 . University hospitals have been associated with better outcomes than non-university hospitals for a variety of procedures and diseases, including radical prostatectomy 8 , 2012 British Journal of Surgery Society Ltd British Journal of Surgery Published by John Wiley & Sons Ltd
Transcript
Original article
Influence of hospital type on outcomes after oesophagealand gastric cancer surgery
J. L. Dikken1,4, M. W. J. M. Wouters1,5, V. E. P. Lemmens7, H. Putter2, L. G. M. van der Geest3,M. Verheij4, A. Cats6, J. W. van Sandick5 and C. J. H. van de Velde1
Departments of 1Surgery and 2Medical Statistics, Leiden University Medical Center, and 3Comprehensive Cancer Centre Leiden, The Netherlands,Departments of 4Radiotherapy, 5Surgery and 6Gastroenterology and Hepatology, Netherlands Cancer Institute – Antoni van Leeuwenhoek Hospital,Amsterdam, and 7Comprehensive Cancer Centre South, Eindhoven, The NetherlandsCorrespondence to: Professor C. J. H. van de Velde, Department of Surgery, K6-R, Leiden University Medical Centre, PO Box 9600, 2300 RC Leiden,The Netherlands (e-mail: [email protected])
Background: Outcomes after oesophagectomy and gastrectomy vary considerably between hospitals.Possible explanations include differences in case mix, hospital volume and hospital type. The presentstudy examined the distribution of oesophagectomies and gastrectomies between hospital types in theNetherlands, and the relationship between hospital type and outcome.Methods: Data were obtained from the nationwide Netherlands Cancer Registry. Hospitals werecategorized as university hospitals (UH), non-university teaching hospitals (NUTH) and non-universitynon-teaching hospitals (NUNTH). Hospital type–outcome relationships were analysed by Coxregression, adjusting for case mix, hospital volume, year of diagnosis and use of multimodal therapies.Results: Between 1989 and 2009, 10 025 oesophagectomies and 14 221 gastrectomies for cancer wereperformed in the Netherlands. The percentage of oesophagectomies and gastrectomies performedin UH increased from 17·6 and 6·4 per cent respectively in 1989 to 44·1 and 12·9 per cent in 2009.After oesophagectomy, the 3-month mortality rate was 2·5 per cent in UH, 4·4 per cent in NUTH and4·1 per cent in NUNTH (P = 0·006 for UH versus NUTH). After gastrectomy, the 3-month mortalityrate was 4·9 per cent in UH, 8·9 per cent in NUTH and 8·7 per cent in NUNTH (P < 0·001 for UHversus NUTH). Three-year survival was also higher in UH than in NUTH and NUNTH.Conclusion: Oesophagogastric resections performed in UH were associated with better outcomes but,owing to variation in outcomes within hospital types, centres of excellence cannot be designated solely onhospital type. Detailed information on case mix and outcomes is needed to identify centres of excellence.
Presented to a meeting of the Dutch Surgical Society, Ede, The Netherlands, November 2011
Paper accepted 27 March 2012Published online in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.8787
Introduction
Long-term survival for patients with resectable oesophagealand gastric cancer is low in the Western world. The 5-yearoverall survival rate is below 25 per cent after oesophagec-tomy and less than 40 per cent after gastrectomy1,2. Bothare high-risk operations with correspondingly high post-operative mortality rates3,4.
Both postoperative mortality and long-term survivalafter oesophagogastric cancer surgery can be improvedby performing these complex procedures in centres withsufficient experience and high annual volumes3,5. An exactcut-off value that defines high-volume surgery has not,however, been established. In a recent survey of all
oesophagectomies and gastrectomies performed in theNetherlands between 1989 and 2009, oesophagectomiescarried out in high-volume hospitals (more than 20 pro-cedures per year) were associated with lower postoperativemortality and improved survival rates compared with thoseperformed in low-volume hospitals. No such relationshipwas found after gastrectomy, but the number of high-volume hospitals was small6.
Although hospital volume can be used as a proxy forquality of care, another approach is to compare outcomesby type of hospital in which the surgery takes place7.University hospitals have been associated with betteroutcomes than non-university hospitals for a variety ofprocedures and diseases, including radical prostatectomy8,
2012 British Journal of Surgery Society Ltd British Journal of SurgeryPublished by John Wiley & Sons Ltd
J. L. Dikken, M. W. J. M. Wouters, V. E. P. Lemmens, H. Putter, L. G. M. van der Geest, M. Verheij et al.
Netherlands Cancer Registry 1989–2009Diagnosis of oesophageal or gastric cancer
n = 71 090
Excluded (no hospital in registry) n = 8
Resection of oesophageal or gastric cancern = 27 436
Excluded (in situ or metastatic disease)n = 3190
Resection of M0oesophageal or gastric cancer
n = 24 246
Excluded (no resection) n = 43 646
Calculation ofannual hospital volumes
Mortality and survivalanalyses
Fig. 1 Study flow chart
heart failure, myocardial infarction and stroke9,10. Ina previous study, no difference was found in survivalafter gastrectomy between university teaching, non-university teaching and non-teaching hospitals, althoughthe number of patients and hospitals was limited11. Theeffect of hospital type on outcomes after oesophagogastricresections remains unclear.
The present study aimed to describe the distributionof oesophagectomies and gastrectomies between hospitaltypes in the Netherlands between 1989 and 2009, and toanalyse the effect of hospital type on short- and long-termoutcomes after these operations.
Methods
Netherlands Cancer Registry
Data were obtained from the Netherlands Cancer Registry(NCR), in which information on all newly diagnosedmalignancies in the Netherlands, a country of 16·5 millioninhabitants, was collected. Patient, tumour and treatmentcharacteristics were collected routinely by trained registrarsfrom the hospital records 6–18 months after diagnosis.The quality and completeness of the data are known to bealmost 100 per cent12.
Topography and morphology were coded accord-ing to the International Classification of Diseases forOncology (ICD-O)13. ICD-O morphology codes wereused to classify tumours as adenocarcinoma (8140–8145,8190, 8201–8211, 8243, 8255–8401, 8453–8520, 8572,8573, 8576), squamous cell carcinoma (8032, 8033,
No.
of r
esec
tions
1200
1000
800
600
400
200
0
Year of diagnosis
200920052001199719931989
No.
of r
esec
tions
800
600
400
200
0
Year of diagnosis
a Oesophagectomies
b Gastrectomies
200920052001199719931989
University hospitalsNon-university teaching hospitalsNon-university non-teaching hospitals
Fig. 2 Number of a oesophagectomies and b gastrectomiesperformed in different hospital types, 1989–2009
2012 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of SurgeryPublished by John Wiley & Sons Ltd
Hospital type and outcomes after oesophagogastric cancer surgery
8051–8074, 8076–8123) and other or unknown histol-ogy (8000–8022, 8041–8046, 8075, 8147, 8153, 8200,8230–8242, 8244–8249, 8430, 8530, 8560, 8570, 8574,8575). Tumours were staged according to the Interna-tional Union Against Cancer tumour node metastasis(TNM) classification in use in the year of diagnosis.Vital status was obtained initially from municipal reg-istries, and from 1994 onwards from the nationwidepopulation registries network. These registries providecomplete coverage of all deceased Dutch citizens. Follow-up was complete for all patients until 31 December2009. The study was approved by the NCR ReviewBoard.
Oesophagectomy and gastrectomy were analysed sep-arately. As the NCR is a topography-based registry,oesophagectomies were defined as resections for cancers of
the oesophagus (C15.0–15.9) and gastric cardia (C16.0),whereas gastrectomies were defined as resections fornon-cardia gastric cancer (C16.1–16.9).
If the hospital of surgery was not registered, the hospitalof diagnosis was assumed to be the hospital of surgery.Annual hospital volumes were defined as the number ofoesophagectomies or gastrectomies per hospital per year.Volume categories were defined as very low (1–5 peryear), low (6–10 per year), medium (11–20 per year) andhigh (at least 21 per year). Hospital types were definedas university hospitals, non-university teaching hospitalsand non-university non-teaching hospitals. Universityhospitals are attached to one of the eight universitiesin the Netherlands, and these hospitals collaborate closelywith the corresponding medical faculty. A hospital was
Table 1 Characteristics of all 10 025 patients with resected non-metastatic oesophageal cancer in the Netherlands between 1989 and2009
Non-university hospitals
Non-teaching(n = 2561)
Teaching(n = 3905)
University hospitals(n = 3559) P†
Sex ratio (M : F) 1952 : 609 3004 : 901 2694 : 865 0·454Age (years) < 0·001
Values in parentheses are percentages. *Chemotherapy with or without radiotherapy. SCC, squamous cell carcinoma; TNM, tumour node metastasis.†χ2 test.
2012 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of SurgeryPublished by John Wiley & Sons Ltd
J. L. Dikken, M. W. J. M. Wouters, V. E. P. Lemmens, H. Putter, L. G. M. van der Geest, M. Verheij et al.
considered a teaching hospital if it offered (part of) asurgical residency programme.
Statistical analysis
Changes in the distribution of operations between hospitaltypes over time and differences in patient characteristicsbetween hospital types were analysed by means of theχ2 test. Overall survival was calculated from the dayof the histological diagnosis until death, because thedate of surgery was not available before 2005. Three-month overall survival was calculated unconditionally,whereas 3-year overall survival was calculated conditionallyon surviving the first 3 months after diagnosis. Possiblerelationships between hospital type and outcomes wereanalysed by stratified Cox regression, adjusted for annual
hospital volume, year of diagnosis, sex, age, socioeconomicstatus14, tumour stage, morphology, preoperative therapyuse, postoperative therapy use (only for 3-year survival)and for clustering of deaths within hospitals15. A separateanalysis was performed including only patients diagnosedbetween 2005 and 2009. To assess potential referral bias,analyses were repeated for hospital of diagnosis insteadof hospital of surgery. Analyses were performed withSPSS version 17.0.2 (SPSS, Chicago, Illinois, USA) andR version 2.12.2 (R Project for Statistical Computing,Vienna, Austria).
Results
Between January 1989 and December 2009, 71 090 patientswith oesophageal or gastric cancer were diagnosed (Fig. 1).
Table 2 Characteristics of all 14 221 patients with resected non-metastatic gastric cancer in the Netherlands between 1989 and 2009
Non-university hospitals
Non-teaching(n = 7387)
Teaching(n = 5702)
University hospitals(n = 1132) P‡
Sex ratio (M : F) 4423 : 2964 3458 : 2244 683 : 449 0·669Age (years) < 0·001
Type of resection† < 0·001Total gastrectomy 266 (37·7) 479 (32·7) 143 (51·1)Subtotal gastrectomy 440 (62·3) 986 (67·3) 137 (48·9)
Values in parentheses are percentages. *Chemotherapy with or without radiotherapy. †Only available for 2005–2009. TNM, tumour node metastasis.‡χ2 test.
2012 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of SurgeryPublished by John Wiley & Sons Ltd
Hospital type and outcomes after oesophagogastric cancer surgery
Some 43 646 patients who did not undergo surgicaltreatment and eight without information on the hospitalof diagnosis or surgery were excluded, leaving 27 436resections for analysis.
Before 2005, the hospital where the resection wasperformed was registered in 53·3 per cent of cases, showing
a match with the hospital of diagnosis in 79·8 per centof patients. For the remaining 46·7 per cent of cases, thehospital of diagnosis was considered the hospital of surgery.
After analysing hospital type distributions and theirrelation with annual hospital volume, 288 patients withcarcinoma in situ and 2902 with distant metastases were
Table 3 Multivariable Cox regression analysis of the relationship between hospital type and outcomes after oesophagectomy andgastrectomy, 1989–2009
Values in parentheses are 95 per cent confidence intervals. NUTH, non-university teaching hospitals; NUNTH, non-university non-teaching hospitals;UH, university hospitals; TNM, tumour node metastasis; SCC, squamous cell carcinoma.
2012 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of SurgeryPublished by John Wiley & Sons Ltd
J. L. Dikken, M. W. J. M. Wouters, V. E. P. Lemmens, H. Putter, L. G. M. van der Geest, M. Verheij et al.
excluded, leaving 24 246 patients with non-metastaticinvasive carcinoma available for hospital type–outcomeanalyses.
Hospital types over time
There are eight university hospitals in the Netherlandsand one specialized cancer centre that was analysedas a university hospital. The number of non-universityhospitals where oesophagectomies and gastrectomies wereperformed decreased, from 120 in 1989 to 82 in 2009.
The annual number of oesophagectomies increased overthe years, from 352 in 1989 to 723 in 2009 (Fig. 2a). Thepercentage of oesophagectomies performed in universityhospitals increased from 17·6 per cent (62 of 352) in 1989to 44·1 per cent (319 of 723) in 2009 (P < 0·001).
The annual number of gastrectomies decreased from1107 in 1989 to 495 in 2009 (Fig. 2b). The percentage ofgastrectomies performed in university hospitals increasedfrom 6·4 per cent (71 of 1107) in 1989 to 12·9 per cent(64 of 495) in 2009 (P < 0·001). Most gastrectomies arecurrently performed in non-university teaching hospitals.
Patient, tumour and treatment characteristics
Between 1989 and 2009, 10 025 patients underwentoesophagectomy and 14 221 underwent gastrectomy forcancer (Tables 1 and 2). The median age of patients whounderwent oesophagectomy in university hospitals was63 years, compared with 64 and 65 in non-universityteaching and non-teaching hospitals respectively. Patientsin university hospitals were more likely to have a squamouscell carcinoma (26·1 per cent versus 20·9 and 19·9 per centin non-university teaching and non-teaching hospitalsrespectively) and a higher tumour stage (stage III diseasein 39·0 per cent (1388 of 3559), 33·4 per cent (1306 of3905) and 34·4 per cent (881 of 2561) respectively). Ahigher proportion of patients in university hospitalsreceived multimodal therapy. Annual hospital volumeswere higher in university hospitals: 69·9 per cent ofoesophagectomies (2488 of 3559) in such hospitals wereperformed in centres with an annual volume of at least21, compared with 11·1 per cent (434 of 3905) in non-university teaching hospitals and no oesophagectomies innon-teaching hospitals.
Patients who underwent a gastrectomy in universityhospitals had a median age of 67 years, compared with71 years in both types of non-university hospital. Patientsin university hospitals also received more preoperativeand postoperative multimodal therapy. Annual hospitalvolumes were highest in non-university teaching hospitals:
43·9 per cent of gastrectomies (2503 of 5702) in non-university teaching hospitals were performed in centreswith an annual volume of at least 11, compared with34·1 per cent (386 of 1132) in university hospitals and24·7 per cent (1822 of 7387) in non-teaching non-university hospitals.
Relationship between hospital type and outcomes
In multivariable regression analysis adjusting for casemix, annual hospital volume, year of diagnosis and
0NUTH NUNTH UH
2
4
3-m
onth
mor
talit
y (%
)
6
8
10
0
No. at riskUH 3384
35022268
313431772088
258424781679
208919241351
170815101090
14241237936
12271049
809NUTHNUNTH
6 12 18
Time after diagnosis (months)
24 30 36
20
40UHNUTHNUNTH
Ove
rall
surv
ival
(%
)
60
80
100
a 3-month mortality
b Survival
Fig. 3 Relationship between hospital type and a 3-monthmortality after oesophagectomy in 10 025 patients and b 3-yearsurvival in 9154 patients, conditional on surviving the first3 months after oesophagectomy. Adjustments were made for yearof diagnosis, sex, age, socioeconomic status, stage, morphology,preoperative therapy use, and annual hospital volume (mortalityand survival analyses) and postoperative therapy use (survivalanalysis only). UH, university hospitals; NUTH, non-universityteaching hospitals; NUNTH, non-university non-teachinghospitals. a *P = 0·006 versus NUTH (Cox regression);b P = 0·027, UH versus NUTH (Cox regression)
2012 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of SurgeryPublished by John Wiley & Sons Ltd
Hospital type and outcomes after oesophagogastric cancer surgery
use of multimodal therapy, both oesophagectomies andgastrectomies in university hospitals were associated withlower 3-month mortality and higher 3-year survival(Table 3).
The adjusted 3-month mortality rate after oesophagec-tomy was 2·5 (95 per cent confidence interval 1·8 to 3·2)per cent in university hospitals, 4·4 (3·5 to 5·2) per centin non-university teaching hospitals and 4·1 (3·2 to 5·0)per cent in non-university non-teaching hospitals (Fig. 3a).Corresponding 3-year survival rates were 46 (44 to 49), 42(40 to 44) and 43 (40 to 59) per cent (Fig. 3b).
0NUTH NUNTH UH
2
4
3-m
onth
mor
talit
y (%
)
6
8
10
0
No. at riskUH 1055
50106471
98446336009
83138675034
72432384283
60827463740
54023633308
47720802983
NUTHNUNTH
6 12 18
Time after diagnosis (months)
24 30 36
20
40UHNUTHNUNTH
Ove
rall
surv
ival
(%
)
60
80
100
a 3-month mortality
b Survival
Fig. 4 Relationship between hospital type and a 3-monthmortality after gastrectomy in 14 221 patients and b 3-yearsurvival in 12 536 patients, conditional on surviving the first3 months after oesophagectomy. Adjustments were made for yearof diagnosis, sex, age, socioeconomic status, stage, morphology,preoperative therapy use, and annual hospital volume (mortalityand survival analyses) and postoperative therapy use (survivalanalysis only). UH, university hospitals; NUTH, non-universityteaching hospitals; NUNTH, non-university non-teachinghospitals. a *P < 0·001 versus NUTH (Cox regression);b P < 0·001, UH versus NUTH (Cox regression)
60
50
40
30
20
10
0
Non-university hospitalsUniversity hospitals
3-m
onth
mor
talit
y (%
)
Hospitals ranked by 3-month mortality
60
50
40
30
20
10
0
3-m
onth
mor
talit
y (%
)
Hospitals ranked by 3-month mortality
a Oesophagectomy
b Gastrectomy
Fig. 5 Three-month mortality rates after a oesophagectomy andb gastrectomy analysed at hospital level. Dots represent Coxregression estimates, bars represent 95 per cent confidenceintervals. Adjustments were made for year of diagnosis, sex, age,socioeconomic status, stage, morphology, preoperative therapyuse and annual hospital volume
Adjusted 3-month mortality rates after gastrectomy were4·9 (3·7 to 6·1) per cent in university hospitals, 8·9 (8·1to 9·7) per cent in non-university teaching hospitals and8·7 (8·0 to 9·4) per cent in non-university non-teachinghospitals (Fig. 4a). Respective 3-year survival rates were 58(55 to 61), 52 (51 to 54) and 52 (51 to 54) per cent (Fig. 4b).
Hospital type–outcome analyses including only patientsdiagnosed between 2005 and 2009 produced no majorchanges in the results, except that the difference in 3-year survival after gastrectomy between hospital typesbecame non-significant (Fig. S1, supporting information).When analyses for 1989–2009 were repeated with thehospital of diagnosis instead of the hospital of surgery,again no major changes were found, although 3-monthmortality after oesophagectomy lost significance (Fig. S2,supporting information). When the analyses were repeated
2012 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of SurgeryPublished by John Wiley & Sons Ltd
J. L. Dikken, M. W. J. M. Wouters, V. E. P. Lemmens, H. Putter, L. G. M. van der Geest, M. Verheij et al.
with university hospitals as the reference category, thesehospitals were found to be associated with a significantlylower 3-month mortality rate after both oesophagectomyand gastrectomy, and significantly better 3-year survivalafter gastrectomy, compared with non-university non-teaching hospitals (Fig. S3, supporting information).
Performance of individual hospitals
Analysis of 3-month mortality rates at the level of individualhospitals indicated that most university hospitals hadgood outcomes (Fig. 5a,b). There were, nevertheless, non-university hospitals with outcomes similar to, or better thanthose of all university hospitals. There were also universityhospitals with average outcomes. The number of patientsper hospital was too small for statistical assessment ofdifferences in outcomes between hospitals.
Discussion
The effect of hospital type on outcomes after oesophagec-tomy or gastrectomy has been studied in a limited waybefore in the Netherlands11,16. In a large American study,postoperative mortality after oesophagectomy and gas-trectomy in National Cancer Institute (NCI)-designatedhospitals was lower than in non-NCI hospitals, even afteradjustment for hospital volume7. Most of these NCIcentres are university hospitals.
In the present study, the increasing number ofoesophagectomies in the Netherlands reflects the increas-ing incidence of oesophageal cancer. This increase hasbeen taken up by university and non-university teachinghospitals. University hospitals have high annual volumes,whereas non-university hospitals operate in lower volumes.
In contrast, the incidence of gastric cancer is declining,leading to a smaller number of gastrectomies over theyears17. Although the absolute number of gastrectomiesin university hospitals (approximately100 per year) andnon-university teaching hospitals (about 300 per year)has remained stable, the number performed in non-university non-teaching hospitals has decreased. Mostcentres, even university hospitals, performed fewer than11 gastrectomies annually. In 2012, gastrectomy will becentralized in the Netherlands to hospitals with a minimumannual volume of 20 per year, mainly towards those centrescurrently performing oesophagectomy.
In the present study, outcomes after oesophagectomyand gastrectomy were better in university hospitalsthan in non-university hospitals, but there were nosignificant differences between non-university teachinghospitals and non-teaching hospitals. Despite differences
of approximately 10 per cent between university and non-university hospitals, 3-year survival rates after gastrectomyin the Netherlands remain low compared with Asianoutcomes18. This difference might be explained bydifferences in tumour stage at presentation, stage migrationowing to more extended lymph node retrieval, and intrinsicbiological differences between Western and Asian patientswith gastric cancer19.
Studies comparing outcomes between hospitals are vul-nerable to various types of bias. The present methodologywas chosen to limit some of these factors. Most oesophagec-tomies performed in recent years were performed inuniversity and non-university teaching hospitals. As qual-ity of care in general is likely to have improved over theyears, better outcomes for operations performed in uni-versity and non-university teaching hospitals might reflectimprovements in perioperative care over the years, ratherthan a true difference between hospital types. Adjustmentfor year of diagnosis was used to eliminate this effect.
Adjustments were also made for annual hospital volume,reducing the effect of hospital volume on outcome whenexamining hospital types. Referral bias was assessed byrepeating the analyses with the hospital of diagnosis insteadof the hospital of surgery. No major differences in theresults were found, indicating that the better outcomesin university hospitals were not the result of selectivereferral of healthier patients from non-university touniversity hospitals. A third of all oesophagectomies wereperformed in university hospitals, but only 8·0 per centof gastrectomies. This tends to reduce the impact of theobservation that university hospitals had better outcomesafter gastrectomy.
The differences in outcomes between university andnon-university hospitals may not be simply explained bytype of hospital, regardless of any other factors. Rather,hospital type might act as a proxy for differences in infras-tructure and processes of care between different typesof hospital. In the Netherlands, university hospitals havehigher staff-to-patient ratios, more financial resources perpatient, more specialized treatments20, and higher-levelintensive care units than non-university hospitals21. Fur-thermore, individual hospitals may differ in quality ofthe diagnostic process, patient selection, administration ofmultimodal therapy, perioperative care, quality of surgeryand ability to deal with complications. Excellent perfor-mance in all parts of this multidisciplinary care pathwaycontributes to a high standard of care and favourableoutcome22. Identification of centres of excellence shouldbe based on robust and case mix-adjusted data provided byhigh-quality clinical audits, where detailed information onthe performance of individual hospitals is collected.
2012 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of SurgeryPublished by John Wiley & Sons Ltd
Hospital type and outcomes after oesophagogastric cancer surgery
Acknowledgements
This study was funded by the Signalling Committee onCancer of the Dutch Cancer Society. The funder hadno involvement in the study design, data collection, dataanalysis, manuscript preparation or decisions regardingpublication.Disclosure: The authors declare no conflict of interest.
References
1 Allum WH, Stenning SP, Bancewicz J, Clark PI,Langley RE. Long-term results of a randomized trial ofsurgery with or without preoperative chemotherapy inesophageal cancer. J Clin Oncol 2009; 27: 5062–5067.
2 Cunningham D, Allum WH, Stenning SP, Thompson JN,Van de Velde CJ, Nicolson M et al.; MAGIC TrialParticipants. Perioperative chemotherapy versus surgeryalone for resectable gastroesophageal cancer. N Engl J Med2006; 355: 11–20.
3 Birkmeyer JD, Siewers AE, Finlayson EV, Stukel TA,Lucas FL, Batista I et al. Hospital volume and surgicalmortality in the United States. N Engl J Med 2002; 346:1128–1137.
4 Gruen RL, Pitt V, Green S, Parkhill A, Campbell D,Jolley D. The effect of provider case volume on cancermortality: systematic review and meta-analysis. CA Cancer JClin 2009; 59: 192–211.
5 Birkmeyer JD, Sun Y, Wong SL, Stukel TA. Hospitalvolume and late survival after cancer surgery. Ann Surg 2007;245: 777–783.
6 Dikken JL, Dassen AE, Lemmens VE, Putter H, Krijnen P,van der Geest L et al. Effect of hospital volume onpostoperative mortality and survival after oesophageal andgastric cancer surgery in the Netherlands between 1989 and2009. Eur J Cancer 2012; 48: 1004–1013.
7 Birkmeyer NJ, Goodney PP, Stukel TA, Hillner BE,Birkmeyer JD. Do cancer centers designated by the NationalCancer Institute have better surgical outcomes? Cancer 2005;103: 435–441.
8 Trinh QD, Schmitges J, Sun M, Shariat SF, Sukumar S,Bianchi M et al. Radical prostatectomy at academic versusnonacademic institutions: a population based analysis. J Urol2011; 186: 1849–1854.
9 Polanczyk CA, Lane A, Coburn M, Philbin EF, Dec GW,DiSalvo TG. Hospital outcomes in major teaching, minorteaching, and nonteaching hospitals in New York state. Am JMed 2002; 112: 255–261.
10 Allison JJ, Kiefe CI, Weissman NW, Person SD,Rousculp M, Canto JG et al. Relationship of hospital teachingstatus with quality of care and mortality for Medicare patientswith acute MI. JAMA 2000; 284: 1256–1262.
11 Siemerink EJ, Schaapveld M, Plukker JT, Mulder NH,Hospers GA. Effect of hospital characteristics on outcome ofpatients with gastric cancer: a population based study inNorth-East Netherlands. Eur J Surg Oncol 2010; 36:449–455.
12 Schouten LJ, Jager JJ, van den Brandt PA. Quality of cancerregistry data: a comparison of data provided by clinicianswith those of registration personnel. Br J Cancer 1993; 68:974–977.
13 World Health Organization (WHO). InternationalClassification of Diseases for Oncology (ICD-O-3) (3rd edn).WHO: Geneva, 2000.
14 Tesser P, Van Praag C, Van Dugteren F, Herweijer L, Vander Wouden H. Rapportage minderheden 1995. Sociaal enCultureel Planbureau: Rijswijk, 1995.
15 Zeger SL, Liang KY. Longitudinal data analysis usinggeneralized linear models. Biometrika 1986; 73: 13–22.
16 Verhoef C, van de Weyer R, Schaapveld M, Bastiaannet E,Plukker JT. Better survival in patients with esophageal cancerafter surgical treatment in university hospitals: a plea forperformance by surgical oncologists. Ann Surg Oncol 2007;14: 1678–1687.
17 Dassen AE, Lemmens VE, van de Poll-Franse LV,Creemers GJ, Brenninkmeijer SJ, Lips DJ et al. Trends inincidence, treatment and survival of gastric adenocarcinomabetween 1990 and 2007: a population-based study in theNetherlands. Eur J Cancer 2010; 46: 1101–1110.
18 Isobe Y, Nashimoto A, Akazawa K, Oda I, Hayashi K,Miyashiro I et al. Gastric cancer treatment in Japan: 2008annual report of the JGCA nationwide registry. GastricCancer 2011; 14: 301–316.
19 Strong VE, Song KY, Park CH, Jacks LM, Gonen M,Shah M et al. Comparison of gastric cancer survival followingR0 resection in the United States and Korea using aninternationally validated nomogram. Ann Surg 2010; 251:640–646.
20 Statistics Netherlands. Figures on Health and Welfare.http://www.cbs.nl [accessed 13 April 2012].
21 Dutch Healthcare Inspectorate. Grote intensive care-afdelingenwerken continu aan kwaliteit. Ministry of Health Welfare andSport: Utrecht, 2011.
22 Birkmeyer JD, Sun Y, Goldfaden A, Birkmeyer NJ,Stukel TA. Volume and process of care in high-risk cancersurgery. Cancer 2006; 106: 2476–2481.
2012 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of SurgeryPublished by John Wiley & Sons Ltd
J. L. Dikken, M. W. J. M. Wouters, V. E. P. Lemmens, H. Putter, L. G. M. van der Geest, M. Verheij et al.
Supporting information
Additional supporting information may be found in the online version of this article:
Fig. S1 Hospital type–outcome analyses for the years 2005–2009: a oesophagectomy, 3-month mortality; bgastrectomy, 3-month mortality; c oesophagectomy, 3-year survival; and d gastrectomy, 3-year survival (Worddocument)
Fig. S2 Hospital type–outcome analyses with hospital of diagnosis instead of hospital of surgery, 1989–2009: aoesophagectomy, 3-month mortality; b gastrectomy, 3-month mortality; c oesophagectomy, 3-year survival; and dgastrectomy, 3-year survival (Word document)
Fig. S3 Hospital type–outcome analyses with university hospitals as reference category, 1989–2009: aoesophagectomy, 3-month mortality; b gastrectomy, 3-month mortality; c oesophagectomy, 3-year survival; and dgastrectomy, 3-year survival (Word document)
Please note: John Wiley & Sons Ltd is not responsible for the functionality of any supporting materials suppliedby the authors. Any queries (other than missing material) should be directed to the corresponding author for thearticle.
2012 British Journal of Surgery Society Ltd www.bjs.co.uk British Journal of SurgeryPublished by John Wiley & Sons Ltd
