Premium
The use of likelihood ratio methodology to find predictors of treatment outcome in patients with dental injury diagnoses
Author(s) -
EMSHOFF R.,
GERHARD S.,
ENNEMOSER T.,
HÄCHEL O.,
SCHERL M.,
STROBL H.
Publication year - 2010
Publication title -
journal of oral rehabilitation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.991
H-Index - 93
eISSN - 1365-2842
pISSN - 0305-182X
DOI - 10.1111/j.1365-2842.2009.02025.x
Subject(s) - medicine , confidence interval , dental trauma , prospective cohort study , dentistry , likelihood ratios in diagnostic testing , univariate analysis , multivariate analysis , surgery
Summary The purpose of this prospective, cohort study of patients with dental injuries was to develop prediction rules to predict treatment response related to the management of dental injuries. The study comprised of 130 patients with a single permanent incisor affected by a dental displacement ( n = 100) or fracture injury ( n = 30). Laser Doppler flowmetry (LDF) measurements of pulpal blood flow (PBF) were taken 6 and 18 weeks after dental injury Treatment response (success or failure) was categorized based on findings of clinical and radiographical evaluation after 9 months. Forty‐four (34%) subjects were categorized as treatment success (absence of loss of sensitivity, periapical radiolucency and grey discolouration of crown), 43 (33%) as treatment failures (loss of sensitivity, and periapical radiolucency and/or grey discolouration of crown) and 43 (33%) as acceptable outcome (loss of sensitivity). After using univariate analysis to determine the association between potential clinical and LDF predictor variables and treatment response status, preliminary prediction rules were developed for prediction of success [positive likelihood ratio (LR), 29·0; 95% confidence interval (CI), 1·7–496·4] and failure (negative LR, 0·55; CI, 0·4–0·7). The most important variables were subluxation, root fracture, baseline PBF level and change in PBF level at 3‐month follow‐up. Outcome following the management of dental injuries may be predicted from variables collected from LDF and physical examination. Predictive modelling may provide clinicians with the opportunity to identify ‘at‐risk’ patients early and initiate specific treatment approaches.