ICU admittance prediction of the seventh-day organ dysfunction clustering

Introduction

Organ dysfunction is usually evaluated by gravity scores (Logistic Organ Dysfunction Score, Multi-Organ Dysfunction Score and Sequential Organ Failure Assessment), aimed to predict the patient's hospital outcome [1, 2]. Despite their diffusion, these scores do not provide indications about the treatment's efficacy for each patient. Our aim was to utilize the previous organ dysfunction scores to perform early strategies to preserve organ function. This study tested the following hypothesis: first, that the previous organ dysfunction scores can detect clustering of organ dysfunctions; second, that each clustering could have an own sequence; and, finally, that we can develop a model to predict from the scores at ICU admittance the organ dysfunctions after the seventh day.

Methods

We retrospectively collected scores, at admission and daily for 6 days, of 85 ICU patients admitted in a 6-month period at Sant'Andrea University Hospital of Rome, Italy. The inclusion criteria were: at least two organ dysfunctions measured by one of the three scores and ICU stay ≥ 24 hours. Data were processed by cluster analysis (VARCLUS di SAS System). We used a discriminant linear model for the prediction of organ dysfunction.

Results

The results are reported in Table 1 and Figure 1. The scores we used detect three clusters of organ dysfunction characterized by a different outcome. The model we elaborate allows one to predict with great accuracy the organ dysfunction over time.

Figure 1

abstract P502

Table 1 Percentage error of model prediction for each organ

Conclusion

The present study can provide the clinician with a useful tool to enact therapeutic strategies to prevent additional organ damage.