Validación de un modelo pronóstico para pacientes pluripatológicos en atención primaria: Estudio PROFUND en atención primaria()

AIMS: to validate the PROFUND index in PP in Primary Health Care (PHC). DESIGN: two-year prospective multicenter study. LOCATION: three health care centers in Seville Province (Spain). SUBJECTS OF THE ASSESSMENT: PP with signed informed consent. Sample: n = 446 (p = 20%; α = 5%; β = 99%); consecutive sampling. MEASUREMENT: Dependent variable: mortality (2 years). Independent variables: socio-demography, clinic, anthropometric, laboratory, pharmacologic prescriptions, functional, cognitive and socio-familiar evaluation and the use of health resources. INFORMATION SOURCE: interview with patients and clinical charts. STATISTICAL ANALYSIS: uni and multivariate analysis according to the variables; Accuracy was assessed in the cohort by risk terciles calibration, and discrimination power, by ROC curves. Finally, accuracy of the index was compared with that of the Charlson index. RESULTS: 446 subjects were included (53.8% men); average age was 75.44 yr (Confidence interval 95% 74.58–76.31). Average of diagnostic categories was 2.37 (Confidence interval 95% 2.30–2.44). Prevalent categories were: A (64.1%), F (41.7%) and E (33.5%). Mortality within 2 years was 24.1%. Calibration in predicted/observed mortality along the three established risk strata was 16%/16.7% for PP with 0–2 points, 22%/19.5% for PP with 3–6, and 34%/36% for PP with 7 or more points (Hosmer-Lemeshow test with p = 0.119). Discrimination power of PHC PROFUND's by area under the curve was (AUC) ROC was 0.622 (Confidence interval 95% 0.556–0.689; p < 0.001), and that of Charlson index 0.510 (Confidence interval 95% 0.446 - 0.575; p > 0.005). CONCLUSIONS: The PROFUND index is a good indicative tool in the stratification of 2-year mortality risk polypathological patients in PHC.