Premium
Frequency distributions of periodontal attachment loss
Author(s) -
Socransky S. S.,
Haffajee A. D.
Publication year - 1986
Publication title -
journal of clinical periodontology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.456
H-Index - 151
eISSN - 1600-051X
pISSN - 0303-6979
DOI - 10.1111/j.1600-051x.1986.tb00855.x
Subject(s) - frequency distribution , clinical attachment loss , statistics , mathematics , periodontal disease , medicine , dentistry
3 patterns of attachment loss were described based on curve fitting the frequency distributions of baseline attachment level measurements of 61 subjects with destructive periodontal diseases. The frequency distributions of attachment level measurements of 30 group I individuals were satisfactorily fit by tachment level measurements of 30 group I individuals were saticfactorily fit by the sum of 2 normal curves, while those of 14 group II individuals were fit by the sum of 3 normal curves. Frequency distributions of 17 group III subjects were fit by a single normal curve. A computer algorithm was used to simulate attachment level loss frequency distributions by varying the “burst size” at active lesions and by imposing “immunity” for varying periods of time on sites which had been attacked. Frequency distributions of group I subjects could be simulated using burst sizes between 0.3 and 3.4 mm. These subjects could be divided into 2 groups: 14 subjects with localized minimal attachment loss required an average burst size of < 1 mm, while 16 subjects with more destruction at diseased sites required an average burst size of > 1 mm to simulate their frequency distributions. Frequency distributions of 16 of 17 group III subjects could be simulated using an average burst size of < 1 mm. However, the frequency distributions could also be simulated using larger burst sizes with the imposition of “immunity” for substantial periods of time at the attacked sites. The frequency distributions of group II subjects were more difficult to simulate. A better simulation could be obtained using a large average burst size (> 1 mm) than a small burst size. However, the most satisfactory fit was obtained by the superimposition of a large burst pattern of attack on a distribution created using small bursts. The optimal average sizes of the small and large bursts varied from subject to subject as did the mean attachment level at which the burst size was changed. The simulations suggest that the different frequency distributions result from different modes of progression of attachment loss with burst size and possibly local “immunity” playing major rôles.