The 'levels of evidence' system, which dates back to the 1980s, ranks studies based on the strength of the corresponding study design. Level sets have been developed for studies of interventions, for ...test accuracy research, and many other forms of clinical research. For some authors, practicing evidence-based medicine has become identical to assigning levels of evidence. This is unfortunate, because more modern systems to come out of the evidence-based medicine movement, such as GRADE, distinguish between the credibility of evidence (which is not just affected by the selected design), the magnitude of the effect, and the applicability of the findings to the clinical question at hand. A new system - :fitting evidence" - has recently been developed by the Dutch Health Care Insurance Board, to evaluate evidence for reimbursement decisions. That system, with an 18-item checklist, looks at threats to the internal validity of studies as well as to the feasibility of randomized comparisons, to see whether it is "fitting" to design a randomized trial. It is unclear yet how these evaluations can and should be combined in the appraisal of the strength of the existing evidence.
To evaluate the indications, complications and survival after pancreatoduodenectomy, with special attention for the outcome after extended resection due to tumour extension and in elderly patients.
...Retrospective, partly cross-sectional study.
All 275 consecutive adult patients who underwent explorative laparotomy for a suspected resectable pancreatic head tumour or periampulary tumour in two Dutch tertiary centres between 2007 and 2010 were included. We graded the postoperative complications according to international classifications and collected data on survival.
In 218/275 patients (79%) the tumour could be resected by pancreatoduodenectomy with or without an extended resection. Malignancy was confirmed in 190/218 patients (87%); in 153/190 patients (81%) a microscopically radical (R0) resection was achieved. Fifteen percent of the patients required a re-intervention (radiological, endoscopic or surgical) because of an intra-abdominal complication. The post-operative 30-day mortality was 4.1%. Eighty-six patients (39%) were ultimately diagnosed with 'pancreatic adenocarcinoma'; they had a 1- and 2-year survival rate of 63% and 34%, respectively. In 27 patients (12%) who underwent an extended resection for oncological reasons, such as partial hepatic portal vein resection, the 30-day mortality was 0% and the survival rates were comparable to patients with a standard resection. The 81 patients (37%) aged 70 or older had a 30-day mortality and survival similar to younger patients.
More than 75% of potentially resectable tumours were resected by a pancreatoduodenectomy with or without an extended resection, with a relatively low postoperative mortality and an adequate survival benefit. After multidisciplinary assessment, both limited tumour extension and a higher age are not necessarily contraindications for a resection, as a comparable survival benefit can be obtained for these groups of patients as for other groups.
PDF