Purpose: To examine factors that influence the rate of bucco-lingual bone-loss occurs around endosseous immediate implants.

Methods: Mesial and distal vertical bone heights, bucco-lingual bone width, and plaque index were measured and calculated at implant placement (0 months), 4 months, and 12 months. Implant site bone density was given a categorical value of 0 to 3 from tooth replacement number, corresponding to posterior maxilla to anterior mandible. Data was standardized into standard normal variables. Terminal events were defined as occurring when >50% of bucco-lingual bone loss occurred (from time of implant placement). Statistically, the time data was stochastically perturbed, and all data was embedded into a proportional hazard's model with a two parameter Weibull baseline in the Bayesian paradigm in the software WINBUGS. A Markov Chain Monte Carlo (MCMC) algorithm procedure was used for parameter inference. The chain was run for 150,000 iterations, and the first 50,000 were discarded as burn-in. Predictive distributions were calculated for a variety of experimental conditions to generate predictive probability distributions of bucco-lingual bone loss of >50% occurring.

Results: Clinical covariates influence baseline bucco-lingual bone-loss rate in the rank order (comparing magnitude of expectation of coefficient): distal bone-loss rate (2.645), initial bucco-lingual bone width (1.108), oral hygiene (0.9596), mesial bone-loss rate (0.8308), and bone density (0.0569).

Conclusion: Mesial and distal bone-loss rates are strongly predictive of bucco-lingual bone loss rate. This is likely due to shared inflammatory etiology between covariates (dependency). Initial bucco-lingual bone width and plaque index are also strongly predictive. Bone density did not have a significant predictive effect.