Frailty Approach for the Analysis of Clustered Failure Time Observations in Dental Research

Frailty Approach for the Analysis of Clustered Failure Time Observations in Dental Research
January 2005
By S.K. Chuang, T. Cai, C.W. Douglass, L.J. Wei, and T.B. Dodson
Journal of Dental Research
You can view the abstract online. A subscription is required to view the full text or it can be purchased online.
Abstract
Because dental implant failure patterns tend to cluster within subjects, we hypothesized that the risk of implant failure varies among subjects. To address this hypothesis in the setting of clustered, correlated observations, we considered a retrospective cohort study where we identified a cohort having at least one implant placed. The cohort was composed of 677 patients who had 2349 implants placed. To test the hypothesis, we applied an innovative analytic method, i.e., the Cox proportional hazards model with frailty, to account for correlation within subjects and the heterogeneity of risk, i.e., frailty, among subjects for implant failure. Consistent with our hypothesis, risk for implant failure among subjects varied to a statistically significantly degree (p = 0.041). In addition, the risk for implant failure is significantly associated with several factors, including tobacco use, implant length, immediate implant placement, staging, well size, and proximity of adjacent implants or teeth.