There is an increasing use of health administrative data for research and in health policy and planning. We examined rates of oesophagectomy on oesophageal cancer patients and whether the inclusion of patients with gastro-oesophageal cancer affected oesophagectomy rates and post-resection outcomes in NSW. Unsurprisingly, the addition of gastro-oesophageal cancer patients to our cohort increased the oesophagectomy rates by 3.5%. Clearly if hospital funding is based partly on patient admissions and the volume of procedures undertaken, care must be taken to ensure that all potential ‘qualifying’ patients are included in any performance assessment. Our reported rate of oesophagectomy in patients is comparable to previously reported European and UK studies
[30, 31] However, the UK rate has been decreasing
 and may be due to the highly specialised pre- and peri-operative assessment and management required for conducting an oesophagectomy
[39, 40], and the support for centralization of oesophagectomies being performed in dedicated multidisciplinary centres
Due to increased precision, the inclusion of gastro-oesophageal cancer resulted in an increase in the number of predictor variables associated with rates of resection and post-resection outcomes that were not seen with oesophageal cancer alone (with the exception of 30-day complication rates). For our overall cohort, factors associated with receiving an oesophagectomy included patient characteristics (age, gender, birthplace) cancer characteristics (histology, disease spread) and hospital related factors (location and funding sector); similar to previously published work
. 30-day mortality in this study was low, in-line with previous studies and international benchmarks
[9, 38, 43]. We found that patients undergoing surgical resection in public hospitals were more likely to have a length of stay >28 days. This may reflect the epidemiology of oesophageal and gastro-oesophageal cancer and that more complex cases are treated in public hospitals (teaching hospitals linked to academic centres in Australia). Thus, the addition of gastro-oesophageal cancer did impact on factors associated with receiving an oesophagectomy and on some post-surgery outcomes.
One-year cancer survival was worse in those who underwent surgery in private hospitals that were not co-located with public hospitals, potentially indicating the benefits of surgery being conducted in higher volume or teaching hospitals
[7, 9, 18, 21, 30, 41, 44, 45].
Our current study demonstrated a reduced 30-day complication rate in gastro-oesophageal cancer compared with oesophageal cancer, hence statistical associations tended towards the null in the combined cohort. This may have been due to the slightly higher rate of transhiatal oesophagectomies undertaken for adenocarcinoma, compared with that in oesophageal cancer and hence lower peri-operative morbidity
, however the only factors which were significantly associated with higher complication rate in patients with oesophageal cancer were older age and resection being conducted in some area health services.
Although we attempted to account for comorbidity with the use of the CCI, the proportion of patients with comorbidity preceding cancer diagnosis was low and as aforementioned, due to how CCI is calculated, comorbidity was most likely under-ascertained. The under-reporting of comorbidities in administrative data may be due to incomplete data transposition from medical records in individual hospitals to administrative databases
[47–49]. Further, non-surgical treatments such as chemotherapy and radiotherapy are funded by the PBS and MBS respectively and this data is not linked routinely to hospital morbidity data thus it was not available for this population linkage study. Further, the only measure of disease severity available was degree-of-spread at diagnosis; there were no indicators of performance status available.