C-reactive Protein Concentration as a Risk Predictor of Mortality in Intensive Care Unit

A Multicenter, Prospective, Observational Study

Rong Qu; Linhui Hu; Yun Ling; Yating Hou; Heng Fang; Huidan Zhang; Silin Liang; Zhimei He; Miaoxian Fang; Jiaxin Li; Xu Li; Chunbo Chen

Disclosures

BMC Anesthesiol. 2020;20(292) 

In This Article

Discussion

In our study of critically ill patients, at ICU admission, the serum concentrations of PCT and CRP and WBC count were significantly lower in survivors than in nonsurvivors, which is similar to the findings of previous studies.[12,19] As a continuous predictor, none of the inflammatory markers at ICU admission, including PCT concentrations, had sufficient discriminative ability for predicting ICU mortality. However, as a dichotomized variable, CRP > 62.8 mg/L at ICU admission was associated with an increased odds of ICU mortality regardless of whether the patient was septic or not. This association remained significant even after adjusting for the APACHE II score, WBC count, and PCT concentration. The findings have clinical implications that some clinically available inflammatory markers may be useful for helping clinicians assess the prognosis of patients in the ICU, which is worth further confirmation.

In clinical practice, PCT concentration is a well-established marker in septic patients.[20–23] However, this real-world study with a sizable sample did not provide evidence for the usefulness of PCT as a predictor of ICU mortality, and this negative association was not due to any nonlinear relationship between PCT and mortality. Although the PCT concentrations of nonsurvivors were significantly higher than those of survivors, the difference was not significant after adjusting for the APACHE II score, WBC count, and CRP concentration, and its AUC-ROC curve was not satisfactory. This result is not surprising since patients with different pathogens might have differently increased PCT values.[24] PCT elevation also occurs in nonsepsis states, such as postoperative conditions, cardiogenic shock or resuscitated cardiac arrest. In addition, PCT levels may be low in patients with viral infections, localized infections or early infections.[25] This negative finding may be due to the complicated reasons for ICU admission. On the other hand, even though the PCT concentration is increased in the serum of patients with bacterial infection, the half-life of PCT is short, and the maximum daily decrease under effective anti-infective therapy is 50%.[26] Therefore, with infection control, the PCT concentration decreases gradually, and the risk of death decreases accordingly. Otherwise, with the progression of infection, the PCT concentration continues to increase, and the risk of death also increases. This may be another reason why the discriminatory performance of PCT was poor in our population. This finding reminds us that more attention should be paid to the trend of PCT serum levels and not only one PCT value.[20,23,27] Of course, the time frame between two consecutive samples needs further investigation.

CRP has been used for many years.[28–30] However, as it is difficult to differentiate sepsis from other nonsepsis causes of SIR, its specificity has been challenged.[31,32] In the study, the CRP concentrations of nonsurvivors were significantly higher than those of survivors. However, its AUC-ROC curve was not satisfactory. As a dichotomized variable, CRP > 62.8 mg/L at ICU admission was associated with an increased odds of ICU mortality regardless of whether the patient was septic or not. This may be related to the lack of specificity of CRP in both sepsis and nonsepsis patients.[31] The findings suggested that even inflammation without infection may still be associated with ICU mortality. The negative predictive value of CRP showed that patients with a low CRP (< 62.8 mg/L) at ICU admission had a low risk of ICU mortality. Therefore, in clinical practice, CRP may be useful for helping clinicians assess the prognosis of ICU patients.

In our study, APACHE II scores other than WBC count were independently associated with the outcome of nonsurvival. As the APACHE II score is a physiologically based system containing 12 physiological parameters, it is a useful prediction tool for hospital consequences, including mortality in critically ill patients.[33] Moreover, the APACHE II scoring system includes WBC count, and thus, pathophysiological changes after systemic insult could be illustrated comprehensively and systematically by the APACHE II scoring system. Therefore, this scoring system is thought to be superior to WBC count for the prediction of adverse outcomes.[34]

Furthermore, we determined that the variable of CRP > 62.8 mg/L did not increase the prediction performance of the APACHE II score using the AUC-ROC comparison by the DeLong method. However, the AUC has recently been criticized for its insensitivity in model comparisons in which the baseline model has performed well.[35] Thus, 2 other measures have been proposed to capture the improvement in discrimination for nested models: integrated discrimination improvement and continuous net reclassification improvement. In the present study, we found that the dichotomized variable of CRP markedly improved risk reclassification over the APACHE II score, displaying a category-free NRI of 0.556 and an IDI of 0.013. On the basis of these results, we concluded that the APACHE II score plus CRP is more promising in terms of improving predictive value than the APACHE II score alone.

There are advantages in this study. To the best of our knowledge, this is the largest study exploring the prognostic value of PCT at ICU admission for patients with and without sepsis. Moreover, it is a multicenter prospective study that allows the retrieval of real-world data with potentially better data quality than retrospective designs, and our results may be generalizable to other centers that have different cases. However, there are some limitations in the study. First, many studies[36,37] revealed that distinct groups of pathogens and different foci of infection determined different PCT serum concentration. In this study, the specific information about infections were not analyzed. It was due to the incomplete data regarding the pathogens and foci of infection. In addition to pathogens, there are many other contributing factors like the source of ICU admission, which showed difference in the constitution of the two groups. Second, CRP at ICU admission was associated with ICU mortality, and this association did not seem to be different between septic patients and nonseptic patients, but the sensitivity and specificity were not perfect. Third, many confounders may still exist and potentially determine bias. For example, some preexisting clinical conditions (i.e., hypertension, diabetes mellitus, coronary disease, diabetes mellitus) presented high proportions in the non-survival than in the survival group. Thus, further studies, such as those combining multiple biomarkers, are essential for improving the prediction performance.

PCT concentration and WBC count at ICU admission were inadequate in their predictive ability of ICU mortality. As a dichotomized variable, CRP > 62.8 mg/L at ICU admission was associated with an increased odds of ICU mortality regardless of whether the patient was septic or not. The negative predictive value of CRP showed that patients with a low CRP (< 62.8 mg/L) at ICU admission had a low risk of ICU mortality. Further studies, such as those combining multiple biomarkers, are essential for improving the prediction performance.

processing....