Diseño de un sistema de información para detección precoz y seguimiento de pacientes con sepsis

Zusammenfassung

Introduction. In 2016, sepsis was defined as a life-threatening organic dysfunction caused by a deregulated response of the host to the infection, in parallel with the Surviving Sepsis Campaign where treatment guidelines are established and the "golden hour" is highlighted, this being the first hour in which we must act to reduce mortality. The importance of early diagnosis to start treatment has been demonstrated and given that it is a syndrome characterized by hemodynamic alterations and elevated biomarkers, we decided to carry out the design of a computer system that would help in the early detection of this syndrome based on the characteristic hemodynamic and analytical variables for its detection. Objectives 1. Main objective To evaluate an automated system for the early detection of patients with septic processes who attend a hospital emergency service. 2. Secondary objectives o Review the cut-off points of the clinical variables o Review the cut-off points of the analytical variables o Review the cut-off points of the demographic variables o Review the cut-off points for cardiovascular risk factors o Identify the analytical and hemodynamic variables with the greatest "weight". o Determine the clinical conditions of the patient that influence the diagnosis of sepsis and should be considered in an automated detection algorithm. Patients and method Observational, descriptive, cross-sectional and retrospective study. The target population is patients older than 14 years seen in the emergency service of the Virgen de La Arrixaca University Hospital in the period from January 1, 2019, to July 31, 2019, with a total sample of 19,733 patients. Results In 2,883 cases of the total number of patients treated, the alarm was activated due to possible sepsis. Of the 2,883 cases detected by the system, 1,685 (58.4%) were confirmed as sepsis by the doctors while 41.6% were not. By type, of the 1,872-sepsis detected by the system, 35.4% (n = 662) were not confirmed by doctors, 63.5% (n = 1,118) were confirmed as sepsis while 1.2% (n = 22) were confirmed as severe sepsis. Regarding severe sepsis, of the 1,011 detected by the system, 53% (n = 536) were not confirmed by doctors, 23.3% (n = 236) were confirmed by 23.6% (n = 239) were confirmed as sepsis. For temperature, PCR and procalcitonin the curves showed that new cut-off points for detection can be established. The optimal cut-off points for the detection of temperature sepsis are established at 37ºC, which determines a sensitivity of 74.3%, a specificity of 54.9% and an AUC = 0.733 (p <0.001). For CRP, the cut-off point is set at 6.1, which determines a sensitivity of 70.4%, a specificity of 69.4%, and an AUC = 0.777 (p <0.001). ROC curves for temperature and CRP are shown in Figure 12. For procalcitonin, the cut-off point is set at 0.3, which determines a sensitivity of 60.6%, a specificity of 61.2%, and an AUC = 0.674 (p <0.001). Conclusions o The computerized alert system created based on the SIRS criteria works. o The variables entered the alert system could be reduced in number. o There are variables that could be ignored for early detection. o By modifying the cut-off points for the main variables, we would increase the sensitivity and specificity of detection. o The implementation of an alert program helps early identification.