出版社：Utrecht University, Maastricht University, Groningen University
摘要：Introduction : The activation of integrated care pathways is not always easy and is time-consuming, so the standardized use of stratification tools (ST) can support health care professionals in decision-making processes. Several ST –developed and validated in the United States– have been implemented in the Valencia Region (Spain) at primary care (PC) centres to identify elders with multimorbidity at risk of suffering future hospital admissions (FHA). In spite of the usefulness of these instruments, it would be recommended to apply STs which have been specifically designed on the basis of the Spanish healthcare system and the characteristics of its population. Thus, the objective of this study was to design a ST aimed at the identification of elders at risk of hospital admissions in the following 12 months in the Valencian Healthcare System. Methods : The study started with the organization of focus groups (FG) with six multidisciplinary PC professionals aimed at selecting potential variables to be included in the ST. From the FG’s results a retrospective study was carried out with the objective of identifying the most significant variables. These variables were tested in a sample of 107 elders. In the last stage, the combination of variables derived from the previous steps was tested in a sample of 1,000 patients in order to select a final combination of variables and to build a predictive algorithm through binary logistic regression analysis. Results : A set of 13 potential variables to be included in our ST were identified in the FG. These variables comprised data on socio-demographics, clinical and social aspects and registers on the previous use of health resources. As part of the retrospective studies, binary logistic regression analysis determined that the following variables were statistically significant as predictors of FHA: ‘chronic respiratory disease’ (OR= 2.32, p= 0.015), ‘chronic heart disease’ (OR= 2.10, p= 0.019), ‘palliative care’ (OR= 22.53, p= 0.04), and the ‘number of previous visits to emergency department at hospital’ (OR= 1.82, p< 0.001). The model derived a risk score from 0 to 1 for each patient, classifying patients as high risk of FHA at a risk score threshold of 0.5 or higher. The model had a sensitivity of 42% and specificity of 96% and the AUC was 0.76. Discussions : The model developed within this study has shown an optimal performance, and an acceptable ability to identify patients at high risk of FHA who really were admitted
