Objectives: Intraobserver reliability of post hoc urodynamic interpretation is greater than interobserver reliability, attributable to interpreter bias. If post hoc interpretation is qualitatively similar to interpretation at the time of study by a urodynamicist in attendance ("live" interpretation), a similar intraobserver bias should be observed. We therefore evaluated the intra- and interobserver reliability of post hoc versus "live" interpretation.
Methods: Fifty-five consecutive urodynamic studies administered by each of two urodynamicists and interpreted at the time of study were de-identified and later re-interpreted by the same two physicians. History and exam findings, cystometric tracing, uroflow summary and radiographs were available for interpretation. Intraobserver (post hoc vs. live), interobserver (post hoc vs. live) and interobserver (post hoc vs. post hoc) reliability was assessed by correlation coefficients for quantitative data, and Cohen's kappa statistic for categorical data.
Results: Post hoc versus live interpretations demonstrated no superiority of intraobserver reliability. For categorical observations, intraobserver reliability was best for urodynamic observations, but worse for clinical diagnosis (kappa = 0.37) and primary treatment recommendation (kappa = 0.26). Interobserver reliabilities were similar to intraobserver. Post hoc/post hoc interobserver correlations for quantitative data were good to very good, r(2) = 0.611-0.914.
Conclusions: In contrast to previously reported post hoc/post hoc comparisons, post hoc/live comparison demonstrated no superiority of intraobserver reliability. Urodynamic observations and impressions are more reliable than are clinical diagnosis and therapeutic recommendation based upon urodynamic evaluation. Unrecorded and heuristic observations at the time of study as well as interpreter judgment may thus be determinants of diagnostic interpretation of urodynamic evaluation.
(c) 2009 Wiley-Liss, Inc.