Sepsis-3 task force emphasized that sepsis was the primary cause of death from infection, especially if not recognized and treated early. Thus, its identification requires urgent attention [5]. According to the degree of priority assigned during triage, continued monitoring can rapidly identify sepsis [16]. Furthermore, attempts to identify patients with sepsis at the triage had continued [17]. Hayden et al. evaluated the efficacy of a sepsis work-up and treatment (SWAT) protocol for rapid identification of sepsis during triage [18]. Although the qSOFA score was recommended by the Sepsis-3 task, its usefulness has remained debatable. Previously, several studies compared the accuracy of different scoring systems, such as qSOFA, SIRS, and SOFA. Raith et al. reported that an increase in SOFA score of 2 or more points indicated greater prognostic accuracy for in-hospital mortality than SIRS or qSOFA [19]. On the other hand, Park et al. found that qSOFA is more effective than SIRS in predicting the occurrence of organ failure in patients with suspected infection [20].
In the investigation of pneumonia, CRB and CRB-65 scoring systems are easy to use, especially in those cases where laboratory result of blood, urea, and nitrogen is unavailable [21, 22]. These systems were proven highly effective in predicting the prognosis and were used widely for several years [23]. In previous studies for pneumonia, CRB and CRB-65 were similar and did not provide additional predictive performance compared with qSOFA [9, 24]. Both CRB and qSOFA had three identical vital signs as criteria: respiratory rate, mental status, and blood pressure. Although CRB and qSOFA used the same vital signs as mentioned above, the thresholds for respiratory rate and blood pressure were stricter for CRB than for qSOFA (CRB: respiratory rate > 30, systolic blood pressure < 90 or diastolic blood pressure ≤ 60; qSOFA: respiratory rate ≥ 22, systolic blood pressure ≤ 100). Therefore, CRB was expected to be more effective in predicting outcomes than qSOFA. CRB-65 used an added parameter of age ≥ 65 years, so it was expected to provide additional predictive performance. However, in this study, the AUROCs of CRB, CRB-65, and qSOFA revealed similar effectiveness in prediction of outcomes. In addition, the differences were not statistically significant, but compared with SIRS, CRB and CRB-65 were significantly more effective in predicting ICU admission. AUROC value of SIRS for predicting ICU admission was < 0.6, which indicated its poor effectiveness. On the other hand, AUROC values of CRB and CRB-65 were between 0.7 and 0.8, which is described as “adequate”. For predicting in-hospital mortality, only CRB-65 provided better predictive performance than SIRS. AUROC values of CRB-65 were between 0.7 and 0.8, which was quite accurate. In the comparison between CRB and CRB-65, despite the addition of age as a parameter, CRB-65 was statistically superior only in predicting in-hospital mortality (p = 0.02). In previous studies that compared the effectiveness of prediction by qSOFA and SIRS in UTI patients, qSOFA had a higher predictive accuracy for in-hospital mortality and ICU admissions than SIRS [25, 26]. Likewise, in our study, the ability of qSOFA to identify the requirement of ICU admission and in-hospital mortality in patients with UTI was better than that of SIRS.
There are several limitations to this study. First, this was a single-center, retrospective study. Thus, selection bias may exist because of the limited sample size available from a single institute. Therefore, caution should be used in generalizing our results, and further studies are required with multi-center, prospective designs for generalization. Second, patients with UTI, especially the elderly, tended to have co-morbidities. Thus, multiple organ dysfunction syndrome may have affected the prognosis. Third, being a large tertiary academic hospital, our institution receives patients transferred from smaller hospitals and primary healthcare institutions who are already in a poor condition. Thus, their mortality is generally higher than normal, which may result in inaccurate study results.