Serum procalcitonin shows moderate accuracy for sepsis diagnosis in critically ill ICU patients

INTRODUCTION: Procalcitonin (PCT) has been a blue-eyed-boy in diagnosing sepsis in previous years. The aim of this systematic review and meta-analysis was to assess the accuracy of PCT for the diagnosis of sepsis, according to the recent Sepsis-3 criteria, in adult patients admitted to the intensive care unit (ICU). METHODS: We searched several electronic databases, including PubMed, Science Direct, Wiley, Cochrane, and Google Scholar from the inception of the Sepsis-3 diagnostic criteria (January 1, 2016) until May 31, 2025, for randomized controlled trials, cohort, and case-control studies that assessed the diagnostic accuracy of PCT for sepsis using the Sepsis-3 criteria among critically ill adult patients with suspected infection. We performed a random effect diagnostic meta-analysis, evaluated the risk of bias of individual studies using the QUADAS tool, and assessed certainty of evidence using GRADE methodology. RESULTS: We included 10 studies comprising 1098 patients. Of these, 635 patients were diagnosed with sepsis based on Sepsis-3 criteria, including 89 patients with septic shock. The pooled sensitivity of PCT for diagnosing sepsis was 0.72 (95 % CI [confidence interval], 0.68-0.75, low certainty) and the pooled specificity was 0.65 (95 % CI, 0.61-0.69, low certainty). The pooled positive likelihood ratio was 2.45 (95 % CI [confidence interval], 1.62-3.68), The negative likelihood ratio was 0.38 (95 % CI, 0.28-0.53), and the pooled diagnostic odds ratio was 7.08 (95 % CI, 3.69-13.58). The area under the summary receiver operating characteristic curve of PCT was 0.79 (95 % CI 0.73-0.86). CONCLUSION: Based on pooled analysis, PCT has a moderate sensitivity and specificity for diagnosis of sepsis in ICU patients with suspected infection. These results suggest clinicians should be cautious about using PCT to facilitate the diagnosis of sepsis in critically ill adults with suspected infection. Given the ongoing uncertainty, further high-quality data in this population is needed.