Platelet count decline - An early prognostic marker in critically ill patients with prolonged ICU stays

Background: Thrombocytopenia is common in ICU patients. The objective of this study was to evaluate possible links between declining platelet counts early in the ICU stay and survival. Methods: All patients who were admitted to the ICU for at least 5 days and had no thrombocytopenia at the time of admission were included in the study. A multivariable logistic regression model, with hospital mortality as the outcome variable, was built. Results: We included 1,077 patients in the study. At ICU admission, the median platelet count was not significantly different in survivors (256 X 10(9) cells/L; interquartile range [IQR], 206 to 330 X 109 cells/L) and nonsurvivors (262 X 109 cells/L; 211 to 351 X 10(9) cells/L). Median simplified acute physiology scores 11 (SAPS 11) at ICU admission was worse in nonsurvivors than in survivors (50 [IQR, 37 to 63] vs 37 [IQR, 27 to 48], respectively; p < 0.0001), as was the mean (+/- SD) sequential organ failure assessment (SOFA) score on day 3 (6.3 +/- 3.24 vs 4 +/- 2.8, respectively; p < 0.0001). Absolute platelet counts were lowest on day 4, but differed significantly between survivors and nonsurvivors only on day 7. Conversely, any percentage decline in platelet counts from 10 to 60% on day 4 was significantly associated with mortality. By multivariable analysis, a 30% decline in platelet count independently predicted death (odds ratio, 1.54; 95% confidence interval, 1.12 to 2.14; p = 0.008), in addition to increasing or stable SOFA scores from ICU admission to day 4, older age, male gender, ICU admission for coma, worse SAPS 11 score at ICU admission, transfer from another ward, and comorbidity. Conclusion: In patients who spend > 5 days in the ICU and have normal platelet counts at ICU admission, a decline in platelet counts provides prognostic information. This parameter deserves to be included in new scoring systems.