No abstract

The low response rate was mainly due to nondelivery of questionnaires at T1 during the regular postoperative visit by the operating physician. Though nonresponse occurs under conditions of routine care, meaningful information was gained, which should be used for quality improvement activities. Because the preoperative level is a major determinant of the postoperative health outcomes, the prospective pre-post measurement should be preferred if institutional comparisons are intended.

BackgroundMeasuring change is fundamental to evaluations, health services research and quality management. To date, the Gold-Standard is the prospective assessment of pre- to postoperative change. However, this is not always possible (e.g. in emergencies). Instead a retrospective approach to the measurement of change is one alternative of potential validity. In this study, the Gold-Standard 'conventional' method was compared with two variations of the retrospective approach: a perceived-change design (model A) and a design that featured observed follow-up minus baseline recall (model B).MethodsIn a prospective longitudinal observational study of 185 hernia patients and 130 laparoscopic cholecystectomy patients (T0: 7-8 days pre-operative; T1: 14 days post-operative and T2: 6 months post-operative) changes in symptoms (Hernia: 9 Items, Cholecystectomy: 8 Items) were assessed at the three time points by patients and the conventional method was compared to the two alternatives. Comparisons were made regarding the percentage of missing values per questionnaire item, correlation between conventional and retrospective measurements, and the degree to which retrospective measures either over- or underestimated changes and time-dependent effects.ResultsSingle item missing values in model A were more frequent than in model B (e.g. Hernia repair at T1: model A: 23.5%, model B: 7.9%. In all items and at both postoperative points of measurement, correlation of change between the conventional method and model B was higher than between the conventional method and model A. For both models A and B, correlation with the change calculated with the conventional method was higher at T1 than at T2. Compared to the conventional model both models A and B also overestimated symptom-change (i.e. improvement) with similar frequency, but the overestimation was higher in model A than in model B. In both models, overestimation was lower at T1 than at T2 and lower after hernia repair than after cholecystectomy.ConclusionsThe retrospective method of measuring change was associated with a larger improvement in symptoms than was the conventional method. Retrospective assessment of change results in a more optimistic evaluation of improvement by patients than does the conventional method (at least for hernia repair and laparoscopic cholecystectomy).

The low response rate was mainly due to non-delivery of questionnaires at T1 during the regular post-operative visit by the operating physician. Though non-response occurs under conditions of routine care, meaningful information was gained which should be used for quality improvement activities. Because the pre-operative level is a major determinant of the post-operative health outcomes, the prospective pre-post measurement should be standard, in case institutional comparisons are intended.

No abstract