Original study - ZZI 02/2016

The influence of various implant types on peri-implant bone loss – a retrospective radiological evaluation

Although prospective, controlled, randomized clinical studies produce the best proof of which of these factors influence crestal bone loss most, in dental practices success can only be evaluated with radiographs taken during recall appointments. The objective of the present study was to evaluate the influence of 3 manufacturers’ implants on peri-implant bone loss.

Material and methods

Study population and test method

This retrospective study collected data from patients provided with implants between March 29, 1995 and December 31, 2012. The data were collected from the patients’ records at Dr. Knöfler’s dental practice, Leipzig, Germany.

The patients were informed that their data were to be used for statistical evaluation. No data were collected other than those usually gathered during a check-up. The evaluation included all patients where at least one postoperative and a further check-up radiograph were available. All radiographs of these patients were digitized.

To ensure that they cooperated, all patients included in this study took part in a professional dental hygiene program.

Patients with bone diseases, non-regulated diabetes mellitus or severe, aggressive periodontitis were excluded. Further exclusion criteria were bisphosphonate treatment and smoking more than 20 cigarettes per day.

All implants with peri-implant diseases during the observation period, either with or without augmentation, and all other implants which would have failed for other reasons were excluded from the study, regardless of the cause of implant failure.

The implant types1 used in this study are shown in table 1.

Augmentation procedure

The various augmentation procedures employed during the study period included internal sinus lifting, bone compaction, bone splitting, alveolar ridge spreading, bone onlay grafting and even external sinus lifting.

Two or more radiographs of each patient were then measured as follows. On each digitized image a line (A) was drawn perpendicular to the implant axis (at a tangent to the implant tip). Then the following values were recorded: the distance from the apical implant tip to the implant platform (implant length = Li) and the lengths of the 2 lines parallel to it – one running from line A to the bony ridge and to the mesial of the implant (Lm) and one running from line A to the bony ridge and to the distal of the implant (Ld) (fig. 1).

For each radiograph the mean depth of the bony cavity (Dorg) was calculated with the following formula: Dorg(0, 1, ...n)= (Li(0, 1, 2 ...n) – [(Ld(0, 1, 2...n)+ Lm(0, 1, 2 ...n) )/2]) x Li org)/Li(0, 1, 2...n) represented the known implant length and Li the measured implant length. Use of this formula compensates for distortion of the radiograph images in comparison to reality. The mean depth of the bony cavity calculated from radiographs taken immediately after implant placement was designated Dorg0 and the depths calculated from images of the same patient taken later were designated Dorg1, Dorg2 etc.

Normally, bone loss data was available for several implants per patient. In addition, information on bone loss per implant at different times was available for many patients. Accordingly, the decision was taken to observe these data points separately, i.e. as if each element of information on bone loss taken at every point in time, stemmed from a different patient (“independent examination“). For this type of evaluation a dot was entered on the time diagram for every measurement and a scatter plot created. The progression was indicated by a curve surrounded by a cloud of dots.

As several implants were placed epicrestally, some supracrestally and a few subcrestally, an attempt was made to determine whether the depth of placement [(a) epicrestal, (b) supracrestal or (c) subcrestal)] led to differences in the progression of bone loss (fig. 2). For every implant, the bone loss values taken later were compared with the previous value and these values used for drawing another scatter plot.

Quantitative evaluation


The data were evaluated using the SAS 9.2 program (SAS Institute, Cary, North Carolina, U.S.A.). Mean values, standard deviations, median, minimum and maximum were used for variables with uniform distribution. With discretely distributed variables absolute and relative frequencies were specified.

In order to describe bone loss as a function of time (quantile regression, PROC quantile), non-parametric regression procedures were employed. Regression analysis of the implants which failed was not carried out as all implant failures were not included.

This enabled the time-dependent behaviour of the parameters tested to be shown more clearly in comparison to a scatter plot where the regression functions for a series of quantiles (5 %, 10 %, 25 %, 50 %, 75 %, 90 %, 95 %) of the variables undergoing examination were determined and shown as lines. With quantile regression, hypotheses on the basis correlation function were established. A polynomial of the third degree was used for the decade logarithm, with which the time-axis was compacted (compression of long time periods), to achieve uniform distribution along the time-axis.

PAGE: 1 | 2 | 3 | 4 | 5