Background: Retrospective assessment of preoperative status is common in nonprospective study designs. The aim of this study was to test the hypothesis that prospective and retrospective evaluations of the preoperative patient condition in hallux surgery gives equal results and therefore can be used interchangeably.
Methods: One hundred and fifty-nine consecutive feet with hallux surgery were analyzed prospectively. Parallel to routine evaluation of the preoperative status, the AOFAS Hallux Scale was recorded prospectively. Two years after surgery, the medical records were re-evaluated for retrospective reconstruction of the AOFAS score. Simultaneously all patients were asked to assess their preoperative status retrospectively.
Results: Using the medical charts for retrospective assessments, the preoperative status was estimated too low compared to prospective evaluation. This effect was even more pronounced with the patients' own retrospective assessment of their preoperative status. Linear regression coefficient for prospective and retrospective data showed moderate correlation with r = 0.59 for the AOFAS Score using the medical charts and poor correlation of r = 0.24 using the patients' own retrospective assessments. Spearman's rank correlation index was p = 0.57 and p = 0.23, respectively. The coefficient of repeatability according to Bland and Altman was 25.7 and 48.7 points, respectively, for the AOFAS score. The difference between the mean values of prospective and retrospective assessment was 5.6 (13.0) points. This means that a retrospectively evaluated AOFAS Score may be 31.3 points below or 20.1 points above prospective assessments (61.6 points below or 35.7 points above compared to the patients' own retrospective self-assessments). Agreement between individual items using Kappa statistics showed poor results except for metatarsophalangeal joint motion.
Conclusions: Prospectively and retrospectively evaluated AOFAS scores cannot be used interchangeably for clinical outcome evaluations. Retrospective scoring gives worse results even when evaluated using conscientiously recorded medical charts and therefore leads to overestimation of the effect of surgery. These data support prospective study designs to ensure the best outcome analysis for clinical evaluation of hallux surgery.