In this secondary analysis of a multicenter prospective study on patients admitted to IMUs, we assessed the potential role of a number of easily measurable variables in predicting the short-term (30 days) mortality comparing their prognostic role to PSI and CURB-65 scores. In particular, we focused our analysis on sodium disorders which are quite frequent during hospital stay [13, 14].
Despite the improvement of in-hospital care of patients with CAP, mortality is high and still remain a major concern for these patients. Thus, there is a growing need to validate new predictors of death and identify potentially “at-risk” patients in order to anticipate a potential clinical worsening [15]. PSI and CURB-65 scores are effective in predicting the short-term mortality or transfer to the ICU in patients presenting in Emergency Department with pneumonia but their prognostic role in other settings is less established [16, 17]. In our study, in patients admitted for CAP, hypernatremia (defined by a value equal to or greater than 145 mEq/L), assessed at the time of admission, is associated with a 3.8-fold increased risk of 30-day death. Interestingly, hypernatremia, together with other 5 easily assessable variables (namely cancer, age > 75, CKD, multilobar infiltrates at chest imaging and dementia), is able to define the short-term prognosis of our patients better than PSI and CURB-65 scores (AUC 0.71 vs 0.63 and 0.64, respectively). For instance, in our study, mortality in the intermediate-risk PSI group was lower than mortality in the low-risk PSI group, highlighting the poor performance of this score in patients admitted to IMUs. On the other hand, hypernatremia and hyponatremia do not appear to be associated with an increased likelihood of being transferred to the ICU.
Sodium disorders, both hypernatremia and hyponatremia, have been previously correlated with adverse outcomes (mortality and prolonged hospital stay) in a prospective cohort of 90.000 unselected hospitalized Chinese patients [9]. Sodium levels are included in the APACHE II (Acute Physiology and Chronic Health Evaluation) score which is currently used to identify critically ill patients at risk of in-hospital mortality [18]. Abnormal sodium levels, both hypo and hypernatremia at admission have been associated with higher mortality among unselected patients in cardiac intensive care units [19].
Results in patients with CAP are less compelling. Dysnatriemia has not been considered as a potential predictor in CURB-65 derivation and validation cohorts [5]. Conversely, hyponatremia only was associated with a higher risk of mortality and transfer to ICU unit in the derivation and validation cohort of the PSI score [6]. Older age and higher prevalence of comorbidities in the ICECAP study may explain different results that we found in our analysis. In the CAPNETZ study on a large population of patients with CAP both hyponatremia and hypernatremia were independently predictors of 28-day mortality [20]. However, due to the relatively young age of the included population, the prevalence of hypernatremia was extremely low (1.4%).
Hypernatremia and hyperosmolarity are common in patients with disorders of impaired thirst sensation or with reduced access to water [21], such as the old, multimorbid and fragile patients included in the ICECAP population. Several possible mechanisms can lead to hypernatremia in CAP, such as plasma volume depletion during acute infection due to increased insensible perspiration determined by fever, tachypnea, diarrhea [13] or reduced water intake due to the patient's severely compromised condition [14].
Of note, dysnatremias did not appear to be associated with transfer to the ICU. Although the lack of association may apparently appear contradictory, this may be due to other constraints in the transfer of ICU including very elderly age, severe cognitive and functional impairment or the presence of concomitant diseases, such as advanced cancer or irreversible late stage cardiovascular diseases.
Our study has some limitations, the main one being that, sodium levels were measured locally instead of being assessed by a single central laboratory. However, the large number of enrolled patients and the involvement of multiple centers in the study, make the results of this study reliable and generalizable. Furthermore, the available sodium levels were only those at the time of admission and any alteration in plasmatic sodium arising during hospitalization, is not available. Thus, any inference on the prognostic role of sodium levels measured during hospitalization should have been avoided. Last, results of our study should be considered only hypothesis generating and our score should be validated in other populations of patients admitted to IMUs for CAP before it can be implemented in clinical practice.
In conclusion, using data of a large multicenter prospective study (ICECAP study), we were able to develop a simple score useful to predict the short term prognosis of patients admitted to IMU for a CAP. Although the results of this score appear promising (with a better accuracy in comparison to PSI a CURB-65 scores), other prospective studies are needed to confirm and validate these preliminary findings.