Driving backwards: methodological and clinical insights from a retrospective clinical and claims data analysis

Background: As the use of claim databases has been progressively shifting from administrative needs to inferential purposes, methodological concern has arisen pertaining to the accuracy of findings obtained from administrative databases and its use in retrospective analyses. Aims and Objectives: The aim of this study is to show that adopting a looking back approach (retrospective analysis) may introduce bias due to lost information relevant to the clinical process as well as to the financial analysis. Materials and Methods: The sample includes 389 patients who underwent a primary elective or urgentcoronary artery bypass graft (CABG) surgery and had relevant cost data -these patients were included in the looking forward analysis. At discharge, of these 389 patients, 369 received a CABG DRG and were included in the looking back analysis (20 patients were discharged as non-CABG DRG and were excluded from the analysis). Statistics: Chi-square and t tests were conducted to assess differences between patients discharged as CABG and non-CABG DRGs at both baseline and 6 month follow up. Results: Patients discharged with a non-CABG DRGcompared to patients discharged with a CABG DRG, while presenting similar sociodemographic and clinical characteristics at admission, reported worse clinical outcome (greater rates of complications and death) and greater costs and length of stay (LOS) at post intervention and 6 month follow up. Conclusion: A looking back approach, excluding specific patients, may neglect information relevant both to the clinical process and to the financial analysis, leading to misleading conclusions.