Weibull survival modeling of laminate veneers using aggregated data from published clinical cohorts

TUZLALI M., BAKİ N.

Journal of Prosthetic Dentistry, 2026 (SCI-Expanded, Scopus)

Özet

Statement of problem: Clinical laminate veneer (LV) cohorts often report time-point outcomes, limiting inferences on longer-term survival and subgroup performance beyond the observed follow-up. Purpose: The purpose of this study was to model LV survival using a parametric Weibull approach applied to aggregated clinical cohort endpoints and to provide pooled and subgroup survival projections with bootstrap uncertainty. Material and methods: Study arms reporting mean follow-up, number of LVs, and failures were pooled. A 2-parameter Weibull survival function, S(t)=exp(-(t/η)β), was fitted by maximum likelihood using a binomial likelihood for failures at each arm's reported mean follow-up. Pointwise 95% confidence intervals (CIs) were obtained by study-arm bootstrap (1200 replicates). Survival projections were reported at 5, 10, 15, and 20 years overall and by restoration material and fabrication method. Subgroup projections beyond twice the subgroup’s maximum observed mean follow-up were not reported, and values beyond the subgroup maximum were flagged as parametric extrapolations (†). Results: The dataset comprised 65 arms (12 617 LVs; 542 failures; observed arm-level survival 95.7% [1−d/n]). The pooled Weibull fit yielded β=.76, with model-based survival of 95.8% (95% CI, 93.1–97.3) at 5 years, 92.9% (95% CI, 86.4–96.2) at 10 years, and 88.3% (95% CI, 70.0–95.3) at 20 years. Long-horizon subgroup estimates showed widening uncertainty when follow-up and failures were sparse. Conclusions: Aggregated-endpoint Weibull modeling supported favorable LV survival through 10 years; projections to 20 years remained model-based and increasingly uncertain at later horizons. Long-term subgroup inferences—particularly for digital workflows and zirconia—should be limited to the observed horizon.