Loading...
Publication
Modelos multivariáveis para a avaliação do risco de perda precoce de dentes em pacientes periodontais
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.4 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Introdução: Os modelos multivariáveis de predição de eventos são ferramentas com
elevada aplicação clínica. Recentemente, foram desenvolvidos modelos de previsão de
perda de dentes em pacientes com periodontite. Devido à inexistência deste tipo de
modelos clínicos para a população portuguesa, este estudo visou o desenvolvimento de
um modelo de previsão de perda precoce de dentes após diagnóstico periodontal, numa
população da Clínica Dentária Egas Moniz (CDEM).
Materiais e métodos: Foram considerados 460 pacientes do Departamento de Periodontologia da CDEM, relativos ao período de Maio de 2015 a Maio de 2018. Após
aplicação dos critérios de inclusão, 445 pacientes (9 377 dentes) foram incluídos no
estudo. Os modelos utilizaram parâmetros do nível paciente e do nível dente. O modelo
preditivo, ou seja, a capacidade de prever com exactidão a perda precoce do dente antes do tratamento periodontal não cirúrgico usando parâmetros de base, foi investigado a partir de uma abordagem multivariada.
Resultados: O modelo multivariável desenvolvido é representado pela expressão: Log
[Prob. (Ext.)/(1-Prob. (Ext.))] = -7.850 + 0.589 × TD(Incisivo) + 0.661 × PIC, com
valores de odds ratio (OR) associados de 1.80 (IC 95%: 1.04-3.12) e 1.94 (IC 95%: 1.78-
2.10), para TD (Incisivo) e perda de inserção clínica (PIC), respectivamente. O modelo
reduzido final explica 25.3% da variabilidade total e classifica corretamente 98.9% dos
casos.
Discussão: O modelo foi criado para um prognóstico dentário individual, devendo ter-se
em consideração que o destino de um dente é influenciado pelo plano final do tratamento global. Posteriormente, é mandatório realizar um estudo de validação deste modelo numa população subsequente da CDEM e/ou numa outra população representativa.
Conclusão: O modelo multivariável poderá ter relevância clínica na tomada de decisão
da manutenção ou extração dentária após diagnóstico de periodontite. Este é o primeiro
modelo preditivo periodontal desenvolvido para uma população portuguesa tendo apresentado uma elevada adequação à população estudada.
Introduction: Multivariate prediction models are powerful tools with high clinical application. Recently, predictive models for tooth loss have been developed in patients with periodontitis. Due to the inexistence of this type of models for the Portuguese population, this study aimed to develop a predictive model of early tooth loss after periodontal diagnosis in a population of the Egas Moniz Dental Clinic (EMDC). Materials and Methods: A total of 460 patients from the Periodontology Department of EMDC, from May 2015 to May 2018, were considered. After application of the inclusion criteria, 445 patients (9 377 teeth) were included in the study. The models used patientlevel and tooth-level parameters. The predictive model, i.e., the ability to accurately predict early tooth loss prior to non-surgical periodontal treatment using baseline parameters, was investigated through a multilevel approach. Results: The developed multivariate model is represented by the expression: Log [Prob. (Ext.) / (1-Prob. (Ext.))] = -7.850 + 0.589 × TD (Incisive) + 0.661 × CAL, with associated Odds Ratio (OR) of 1.80 (CI 95%: 1.04-3.12) and 1.94 (CI 95%: 1.78-2.10), for Type of tooth, TD(Incisive) and Clinical Attachment Loss (CAL), respectively. The final reduced model explains 25.3% of the total variability and correctly classifies 98.9% of the cases. Discussion: The model was created for an individual dental prognosis, so it must be considered that the destination of a tooth is often influenced by the final plan of the global treatment. Posteriorly, it is mandatory to carry out a validation study of this model in a subsequent population of the Egas Moniz Dental Clinic and/or in other representative population. Conclusion: The multivariate model may have clinical relevance in the decision making of dental maintenance or extraction after diagnosis of periodontitis. This is the first predictive periodontal model developed for a Portuguese population and presented a high adequacy for the studied population.
Introduction: Multivariate prediction models are powerful tools with high clinical application. Recently, predictive models for tooth loss have been developed in patients with periodontitis. Due to the inexistence of this type of models for the Portuguese population, this study aimed to develop a predictive model of early tooth loss after periodontal diagnosis in a population of the Egas Moniz Dental Clinic (EMDC). Materials and Methods: A total of 460 patients from the Periodontology Department of EMDC, from May 2015 to May 2018, were considered. After application of the inclusion criteria, 445 patients (9 377 teeth) were included in the study. The models used patientlevel and tooth-level parameters. The predictive model, i.e., the ability to accurately predict early tooth loss prior to non-surgical periodontal treatment using baseline parameters, was investigated through a multilevel approach. Results: The developed multivariate model is represented by the expression: Log [Prob. (Ext.) / (1-Prob. (Ext.))] = -7.850 + 0.589 × TD (Incisive) + 0.661 × CAL, with associated Odds Ratio (OR) of 1.80 (CI 95%: 1.04-3.12) and 1.94 (CI 95%: 1.78-2.10), for Type of tooth, TD(Incisive) and Clinical Attachment Loss (CAL), respectively. The final reduced model explains 25.3% of the total variability and correctly classifies 98.9% of the cases. Discussion: The model was created for an individual dental prognosis, so it must be considered that the destination of a tooth is often influenced by the final plan of the global treatment. Posteriorly, it is mandatory to carry out a validation study of this model in a subsequent population of the Egas Moniz Dental Clinic and/or in other representative population. Conclusion: The multivariate model may have clinical relevance in the decision making of dental maintenance or extraction after diagnosis of periodontitis. This is the first predictive periodontal model developed for a Portuguese population and presented a high adequacy for the studied population.
Description
Dissertação para obtenção do grau de Mestre no Instituto Universitário Egas Moniz
Keywords
Doença periodontal Periodontite Inteligência artificial Modelo preditivo de risco de perda precoce de dentes