A predictive model for osteoporotic fractures

• Published in: Revista electrónica "Dr. Zoilo E. Marinello Vidaurreta" Vol. 44; no. 2
• Main Authors: Zaily Fuentes-Díaz, Orlando Rodríguez-Salazar, Elizabeth Vidor-Guerra, Luis Manuel Amador-Aguilar
• Format: Journal Article
• Language: Spanish
• Published: Universidad de Ciencias Médicas de Las Tunas 01-03-2019
• Subjects: adult; forecasting; fractures, bone; osteoporosis
• ISSN: 1029-3027

Background: osteoporosis is a silent disease that causes a progressive deterioration and fragility of the bone tissue. The people who suffer from this condition may have a probable fracture.Objective: to design a predictive model for fractures in adult patients, defined from a study of surgical patients operated on at the department of orthopedics of the “Manuel Ascunce Domenech” General Teaching Hospital, from 2014 through 2018.Methods: a predictive model for fractures in adult patients was created. To this aim a quasi-experimental was carried out with surgical patients diagnosed with a fractured and operated on as either elective or emergency surgeries, at the aforementioned hospital department and during the period declared in the objective.Results: the predictive model for fractures based on the logistic regression reached good values regarding verisimilitude and the coefficient of determination, R squared. In the operative characteristic curve of the receptor (COR) associated with the model, for a threshold of 0,220, 94,4 % of true positives were obtained and 6,5 % of false positives, fracture probability was declared over the threshold. With the predictive contingency of the model for fractures 79 patients with bone fragility were confirmed as true positives, prognosticating fracture risk. Among the patients with a good bone quality, 14 did not have a prognosis of risk, but 7 did have it. The application of the model gave good solutions that supported treatment.Conclusions: the predictive model for fractures in adult patients based on the logistic regression was reliable and was validated in a local population.