Background: The vasoactive-inotrope-score (VIS) has been used to objectively quantify the degree of cardiovascular support required in patients undergoing congenital heart surgery or in pediatric sepsis. While most studies measured VIS at predetermined time points, this method may not accurately reflect the totality of support needed or the varying requirements for vasoactive support during an ICU stay, as doses can change substantially over a short period of time. Our goal was to use a continuously calculated, cumulative VIS in all patients admitted to our PICU who received vasoactive infusions to determine if this can more accurately predict a complicated PICU course. 

Methods: Retrospective data was collected on patients admitted to Akron Children’s PICU between 2019-2021 that received vasoactive infusions during admission. Cumulative VIS was calculated using an algorithm which totaled VIS each time a patient was started on or had a dose change in inotropic medication using automated real-time infusion pump verification in the electronic health record. A composite poor outcome was defined a priori as either mortality, cardiac arrest in the ICU, or ECMO support. Statistical analysis was conducted using Wilcoxon Rank-Sum Test, Chi-Square Test of Independence, and logistic regression with ROC curve analyses. Stats shown as median and IQR. 

Results: Of the 2173 patients studied, 57 patients had the defined poor outcome. Variables studied included length of stay (LOS), age, severity of illness scores, cumulative VIS, and the total and average VIS at 6, 12, 18, 24, 36, and 48 hours. 171 patients were post-op CV surgery, of which 9 had a poor outcome. Age of good outcome was older (61 mo[13, 162.5] vs 50[3,152], p=0.38). LOS for good outcome was shorter (28 days [18,56] vs 98[24,294] p<0.001). The cumulative VIS of patients with good outcome was significantly smaller (0.0[0,0] vs 13.4[2.9,40.5], p<0.001). Odds ratio for a good outcome in non-CV surgery patients was 2.26[1.09,4.69, p=0.024). 

Using univariate logistic regression, ROC curves demonstrated significant p values (<0.001) for total cumulative VIS and cumulative VIS at 6, 12, 18, 24, 36, and 48 hours. Total cumulative VIS has an AUC= 0.90 which is similar to the AUCs for PIM2/3, PRISM3, PELOD (0.92, 0.91, 0.96 respectively). Measurements of cumulative VIS at 6 and 12 hours had lower AUC than all other time points, yet both >=0.87. 

Conclusion: Approximately 2.3% of our population experienced poor outcome of which they had higher LOS, severity scores, and higher total cumulative VIS. Total VIS at predetermined cut points showed similar significance to the total cumulative VIS but with lower AUC. Cumulative VIS performed similar to PIM2, PIM3, and PRISM3 in predicting poor outcome but was outperformed by PELOD. Total cumulative VIS remains a strong predictor for poor outcome. Using real-time infusion pump data, the VIS is captured in real-time, which if integrated into a real-time alert system, could give clinicians more data for patient management, especially in post-op cardiac patients.