Improvements in methods of periodontal probing: comparison of relative attachement level data selected by outlier reduction protocols from Florida disc probe measurements.

Abstract

OBJECTIVES: To compare relative attachment level data (RAL) selected by the Option-4 algorithm (O-4), Modified Option-4 algorithm (MO-4), Option-3 method (O-3) and Double Pass method (DP) from a common dataset and to determine the most efficient method in eliminating outliers. MATERIAL AND METHODS: A single clinician recorded full mouth RAL with the Florida Disc Probe on four occasions over 6 months in 16 subjects (mean age 48.1 years) with untreated moderate Chronic Adult Periodontitis (mean Probeable Crevice Depth 2.9 mm). RESULTS: 2312 sites were available for analysis. Within-visit correlation coefficients for the two selected RAL measurements were 0.98 (P < 0.001) for O-4, MO-4 and O-3 and >or= 0.92 (P < 0.001) for DP. The maximum mean differences of within-visit RAL were - 0.05 mm for O-4, - 0.03 mm for MO-4, - 0.03 mm for O-3 and - 0.02 mm for DP. The standard deviations of these differences were <or= 0.44 mm for O-4, <or= 0.47 mm for MO-4, <or= 0.45 for O-3 and <or= 0.96 mm for DP. Repeated Measures Analysis of Variance (RMAV) showed a significant difference in RAL between visits for all methods (P < 0.003). RMAV, investigating predetermined between-method comparisons, showed a significant difference in RAL between visits (P = 0.0002) and a significant interaction between the order of the selected RAL and method (P = 0.0011). Within-visit RAL agreement to within 1.0 mm was achieved at >or= 99.9% sites for O-4, >or= 99.9% sites for MO-4, >or= 99.3% sites for O-3 and >or= 85.6% sites for DP. Remeasurement (in addition to two passes) was required over the study period at 16.6% sites for O-4, 13.2% sites for MO-4 and 13.0% sites for O-3: DP, by definition, required no additional measurements. The mean site-specific variances at all visits were <or= 0.1 mm(2) for O-4, MO-4 and O-3 and <or= 0.44 mm(2) for DP. CONCLUSIONS: The Option-4 algorithm was found to be the most effective outlier reduction protocol currently available, producing the most reproducible data.