Reliability of linear alveolar bone loss measurements of mandibular posterior teeth from digitized bitewing radiographs

• Published in: Journal of clinical periodontology 1998-11, Vol.25 (11), p.850-856
• Main Authors: Hildebolt, C. F., Pilgram, T. K., Yokoyama-Crothers, N., Fletcher, G., Helbig, J. L., Bartlett, T. Q., Gravier, M., Vannier, M. W., Shrout, M. K.
• Format: Article
• Language: English
• Subjects: alveolar bone; Alveolar Bone Loss - diagnostic imaging; Analysis of Variance; Confidence Intervals; Dentistry; Humans; Mandible - diagnostic imaging; Observer Variation; radiographic bone height; Radiography, Bitewing - methods; Radiography, Bitewing - statistics & numerical data; Radiography, Dental, Digital - methods; Radiography, Dental, Digital - statistics & numerical data; reliability; Reproducibility of Results; Tooth - diagnostic imaging
• ISSN: 0303-6979; 1600-051X
• DOI: 10.1111/j.1600-051X.1998.tb02381.x

. Observer reliability in performing linear measurements between the cementoenamel junction and alveolar crest was determined for mandibular posterior teeth from digitized clinical bitewing radiographs acquired during recall examinations. 6 measurements (corresponding to traditional probing measurements were made per tooth by 3 observers. Mesial and distal measurements made to the most coronal aspects of the alveolar crest were the most reliable and least biased, As was anticipated, intra‐observer reliability was better than inter‐observer reliability although the 3 observers of our study were able to detect a significant mean change (0.1 mm. p < 0.0001) in alveolar bone height over a 1‐year period for 10 patients. For our most reliable and unbiased measurements (mesial measurements to the alveolar crest), a change of 0.54 mm (90th percentile) would be required to indicate change at a site from one time to the next. Based on the reliability of our digital radiographic measurements, with the α error rate set at 0.05 and β at 0.20. a difference in alveolar bone height of 0.3 mm could be detected with a patient sample size of between 13 (best case) and 54 (worst case).