PROSPECTIVE VALIDATION OF THE VASOACTIVE-INOTROPIC SCORE AND
CORRELATION TO SHORT TERM OUTCOMES IN NEONATES AND INFANTS
AFTER CARDIOTHORACIC SURGERY
Jesse Davidson MD MPH1, Suhong Tong MS2, Hayley Hancock, MD1, Amanda Hauck MD1, Eduardo da Cruz MD1,
Jon Kaufman MD1
¹ The Heart Institute, Department of Pediatrics, The Children’s Hospital Colorado, USA
2 Department of Biostatistics, University of Colorado Denver, CO, USA
The original publication is available at www.springerlink.com
Jesse Davidson, M.D. M.P.H.
The Heart Institute, Department of Pediatrics, Children’s Hospital Colorado
13123 East 16th Ave, Box 100
Aurora, CO 80045, USA
Phone: +1 720-777-6820 (hours 0900-1630)
Fax: +1 720 777 7290
Purpose: Prospective validation of vasoactive-inotropic score (VIS) and inotrope score (IS) in infants after
Methods: Prospective observational study of 70 infants (≤90 days of age) undergoing cardiothoracic surgery. VIS
and IS were assessed at 24 (VIS24, IS24), 48 (VIS48, IS48), and 72 (VIS72, IS72) hours after surgery. Maximum
VIS and IS scores in the first 48 hours were also calculated (VIS48max and IS48max). The primary outcome was
length of intubation. Additional outcomes included length of intensive care (ICU) stay and hospitalization, cardiac
arrest, mortality, time to negative fluid balance, peak lactate, and change in creatinine.
Results: Based on Receiver Operating Characteristic (ROC) analysis, area under the curve (AUC) was highest for
VIS48 to identify prolonged intubation time. AUC for the primary outcome was higher for VIS than IS at all time
points assessed. On multivariate analysis VIS48 was independently associated with prolonged intubation (OR 22.3,
p=0.002), prolonged ICU stay (OR 8.1, p=0.017), and prolonged hospitalization (OR 11.3, p=0.011). VIS48max,
IS48max, and IS48 were also associated with prolonged intubation, but not prolonged ICU or hospital stay. None of
the scores were associated with time to negative fluid balance, peak lactate, or change in creatinine.
Conclusion: In neonates and infants, a higher VIS at 48 hours after cardiothoracic surgery is strongly associated
with increased length of ventilation, and prolonged ICU and total hospital stay. At all time points assessed, VIS is
more predictive of poor short term outcome than IS. VIS may be useful as an independent predictor of outcomes.
Keywords: Congenital; cardiovascular; VIS; children; outcomes; inotrope score
Cardiac surgery for the repair or palliation of congenital heart disease often results in a decrease in cardiac
output during the immediate post-operative period. In approximately 25% of infants and young children, low
cardiac output develops, and these patients are at higher risk of death in the post-operative period [1,2,4]. The
management of these patients relies on multiple strategies intended to mitigate the potential threat of low cardiac
output. As part of this management, inotropic and vasoactive agents are routinely employed after cardiac surgery in
infants to decrease the risk of low cardiac output.
In 1995, Wernovsky created an inotrope score (IS) as part of a study on post-operative hemodynamics following
the arterial switch operation . This score attempted to quantify the amount of inotropic support provided in the
post-operative period. Since this initial publication, the inotrope score has been used as a research tool to describe
the effects of various treatments on the required amount of hemodynamic support [3-6]. No study has ever shown a
correlation between the original IS and clinical outcomes of interest. Also, newer vasoactive-inotropic agents have
been introduced to pediatric cardiac intensive care, most notably milrinone and vasopressin [4, 7-9], potentially
limiting the accuracy of the inotrope score. More recently, Gaies et al published a retrospective study using an
updated vasoactive-inotropic score (VIS) . VIS incorporates the original medications from the inotrope score
and adds milrinone, vasopressin, and norepinephrine. In their population of infants undergoing cardiac surgery with
cardiopulmonary bypass (CPB), higher maximum VIS in the first 48 hours after operation was associated with
increased odds of poor short term outcomes.
Our study sought to prospectively validate VIS and IS in a population of infants 90 days of age or less
undergoing cardiothoracic surgery. We hypothesized that higher VIS would correlate with worse short term clinical
outcomes in both CPB and non-CPB patients and that VIS between 48 and 72 hours would more accurately predict
patients at risk for poor short term outcome.
We conducted a prospective observational study of infants ≤90 days of age undergoing cardiothoracic surgery.
Patients were enrolled as part of an Institutional Review Board approved study on the post-operative kinetics of the
inflammatory marker procalcitonin (PCT). Analysis of VIS and outcomes was included as a sub-analysis of the
parent PCT trial. Seventy patients were enrolled between July 2009 and September 2010. Infants were excluded if
they were born at less than 34 weeks estimated gestational age, if informed consent could not be obtained, or if the
patient was less than 1200g at the time of surgery due to concerns for excessive blood draw volume in the PCT
portion of the study.
Demographic and pre-operative clinical information collected on all patients included gender, ethnicity,
gestational age at delivery, age and weight at the time of surgery, anatomic diagnosis, surgical procedure, Aristotle
comprehensive complexity score [11-13], use of pre-operative steroids, CPB time, aortic cross-clamp time, and deep
hypothermic circulatory arrest time. Pre-operative laboratory tests included a basic metabolic panel (BMP) as part
of the surgical clinical guidelines and PCT and liver function tests (LFTs) as part of the research protocol.
Patients entered one of three groups based on the following operative characteristics: 1) delayed sternal closure
(DSC), 2) CPB without DSC, and 3) no CPB or DSC. Group status determined the blood draw schedule for the
procalcitonin portion of the study and did not affect the VIS portion of the study other than for purposes of statistical
analysis. The decision to perform DSC was made by the attending surgeon as a precaution in patients thought to be
at high risk for low cardiac output with primary closure. Inotropic and vasoactive medications were initiated in the
operating room at the discretion of the attending surgeon and cardiac anesthesiologist. Decisions regarding ongoing
titration of vasoactive/inotropic medications were made by the CICU physician team and did not follow a pre-
established protocol. Neither the surgical team nor the CICU team was aware of the intention to study
vasoactive/inotropic support. Doses of milrinone, dopamine, epinephrine, vasopressin, and norepinephrine were
recorded hourly throughout the course of CICU admission. IS was calculated as per Wernovsky et al  and VIS
was calculated as per Gaies et al  at 24, 48, and 72 hours after admission to the CICU (figure 1): ISa =
dopamine dose (µg/kg/min) + dobutamine dose (µg/kg/min) + 100 x epinephrine dose (µg/kg/min) VISb = IS + 10 x
milrinone dose (µg/kg/min) + 10,000 x vasopressin dose (Units/kg/min) + 100 x norepinephrine dose (µg/kg/min).
In addition, maximum VIS and IS during the first 48 hours of the CICU admission were recorded.
Hemodynamic monitoring included invasive arterial pressure monitoring and central venous pressure monitoring
in all cases. Additional monitoring was instituted as necessary by the intensivist. All groups had a BMP, arterial
blood gas and lactate performed on admission and on post-operative day one. Patients in groups 1 and 2 also had a
BMP and lactate drawn at 12 hours after admission and LFTs on post-operative day one. Additional laboratory
testing was performed as directed by the intensivist.
The primary outcome determined a priori was time to first extubation in hours. Secondary outcomes included
length of CICU stay in days, length of hospital stay, time to negative fluid balance, peak lactate, and change in
creatinine. Due to the low expected mortality, a combined dichotomous poor outcome variable was utilized, defined
as any one of the following: cardiac arrest requiring chest compressions, death within 30 days or at any point prior to
discharge, renal replacement therapy, or mechanical circulatory support .
Outcomes were classified as upper 25% vs. lower 75% of the measures. The primary outcome was used to
determine the best predictors and their cut-offs among the various VIS and IS time points. All time points were
modeled as predictor using unconditional logistic regression by Lambert’s SAS Macro . The sensitivity,
specificity, total accuracy, weighted error ratios and AUC were then calculated and plotted. The score with the
largest AUC was chosen as the best predictor of the outcomes. Additional time points were also chosen for
modeling to allow more direct comparison to prior studies. The best cut-off was determined by maximizing
combined sensitivity, specificity, and accuracy while minimizing weighted error ratio.
For multiple logistic regression modeling, baseline characteristics were assessed exhaustively between groups
with appropriate descriptive statistics. Characteristics that differed significantly among groups were then chosen as
candidate covariates. Best subset logistic regression model technique was employed to choose the best fit models
for each of the outcomes. Group effect and presence of a functional single ventricle were kept in all multivariate
models to adjust for clinical differences.
Between July 2009 and September 2010, seventy patients were successfully enrolled. All patients completed the
procalcitonin study protocol. Of the 70 patients enrolled, 56 underwent CPB, of which 26 required DSC. One
additional patient did not undergo CPB but required DSC due to shunt malfunction and immediate reoperation for
shunt revision. No patients underwent sternal opening in the CICU. After completion of trial participation, one
patient was found to have an inaccurate date of birth in the medical record and was older than inclusion criteria
permitted. This patient was excluded from all subsequent analyses.
Demographic and baseline surgical data are shown in table 1. Overall, the DSC and non-bypass patients were
younger than the general CBP patients, and the DSC patients were significantly lower weight at the time of
operation. The DSC group had higher Aristotle scores, a higher prevalence of single ventricle physiology, and
greater use of pre-operative steroids. Median bypass and circulatory arrest times were longer in the DSC group,
although aortic cross clamp time was comparable. Median time to sternal closure in the DSC group was 2 days
Median VIS and IS for each group are also shown in table 1. Patients with DSC had significantly higher VIS
and IS at all time points assessed. The remaining CPB patients initially had a higher VIS than the non-bypass
patients, but this difference had largely disappeared by 72 hours. Post-operative outcomes are provided in table 2.
ROC analysis was performed for VIS24, VIS48, VIS72, IS24, IS72, IS48, VIS48max, and IS48max. AUC and
confidence intervals for each time point are shown in table 3. All time points performed well with an AUC of at
least 0.85. AUC was highest for VIS48 (0.93). At similar time points, the AUC was greater for VIS than IS.
Scores at 48 and 72 hours had higher AUC than maximum scores in the first 48 hours. None of these differences,
however, reached statistical significance.
Given the overall strong performance of VIS and IS at predicting prolonged intubation, we chose to perform
multiple logistic regression modeling for several different time points. VIS48 was chosen as the primary score to
model due to its high AUC and ease of clinical computation. We also modeled VIS48max to compare to prior
studies . In order to offer a comparison between VIS and IS, we added models using IS48 and IS48max. Cut-
offs for each predictor variable were chosen to maximize total accuracy and minimized weighted error ratios as
described by Lambert et al . Based on this technique we chose cut-offs for each score as listed in table 4.
The results of the multivariate logistic regression are presented in table 5. All models were controlled for
operative group and presence of single ventricle physiology. Additional covariates including age, weight, bypass
time, cross clamp time, circulatory arrest time, and Aristotle score were assessed but did not significantly improve
the fit of the model and were not included. High VIS48 was independently associated with greatly increased odds of
prolonged intubation (OR 22.3, 95% confidence interval (CI) 3.2-157.7, p value 0.002). High VIS48max, IS48, and
IS48max were also associated with increased odds of prolonged intubation [(OR 14.9, 95%CI 2.4-94.1, p=0.004),
(OR 18.1, 95%CI 2.4-138.1, p=0.005), and (OR 7.1, 95%CI 1.3-37.6, p=0.021) respectively].
Higher VIS48 was also independently associated with increased odds for prolonged ICU stay (OR 8.1, 95%CI
1.4-45.4, p value=0.017) and hospitalization (OR 11.3, 95%CI 1.7-73.7, p value 0.011). High VIS48max, IS48, and
IS48max predicted mildly increased odds of prolonged ICU stay and hospitalization, but none of these associations
reached statistical significance. Overall, we did not find any association between VIS or IS and time to negative
fluid balance, change in creatinine, or peak lactate.
Only 9 patients met criteria for poor outcome. These small numbers limited the extent of statistical analysis
possible, particularly with VIS48 and IS48 where two of the patients died before 48 hours and thus did not receive a
48 hour score. VIS48max and IS48max both showed a moderate association with poor outcome with a trend
towards statistical significance when adjusted for group [(OR 6.0, 95%CI 0.8-43.4, p value=0.076), (OR 4.8, 95%CI
0.7-31.1, p value=0.099) respectively].
To our knowledge, this study is the first to prospectively validate the use of VIS and IS in infants. Our findings
confirm that in this population, VIS, assessed in the first 72 hours after surgery, correlates better with outcomes than
does IS. The exact timing of scoring, however, remains unclear. The sentinel study using VIS by Gaies et al
focused primarily on the intensity of medical cardiovascular support (maximum VIS in the first 48 hours after ICU
admission) . In our design, we opted to assess both markers of peak intensity (VIS48max) as well as markers of
prolonged high intensity therapy (VIS48). We found a striking association between VIS48 and several important
short term outcomes including length of intubation, length of intensive care unit stay, and length of hospital stay.
Overall in our study VIS48 outperformed VIS48max, suggesting that duration of intensive cardiovascular support
may be more important than maximal intensity of therapy as a predictor of these short term outcomes. We
hypothesize that VIS48 may place less emphasis on patients with transient poor function or vasoplegia whose
outcomes are likely to be good, while potentially identifying patients with persistent issues such as myocardial
injury, poor surgical repair, or capillary leak. VIS48 has the added clinical benefit of ease of calculation, requiring a
simple bedside check without review of prior records. Our patient population does differ slightly from the initial VIS
publication, focusing on younger infants (0-3 months instead of 0-6 months) and including a small sample of non-
bypass patients. This may account for some of the difference in our findings. The odds ratios for maximum VIS,
however, are quite similar in our population as compared to that of Gaies et al, suggesting that maximum VIS
behaved in a very similar manner as a predictor in both populations .
The strong association between VIS and length of intubation makes intuitive sense. Positive pressure
ventilation decreases the energy expended by the patient for breathing. In a critically ill post-operative infant who
already requires high levels of inotropic and vasoactive support, clinicians are less likely to extubate and thereby
transfer the work of breathing exclusively to the patient. So while we agree with the general idea that high VIS is
largely a surrogate marker for poor outcomes and VIS should not be targeted as a primary intervention to improve
outcomes, it is likely that therapies capable of improving post-operative VIS would directly improve intubation
times as well.
The relationship between high VIS48 and prolonged ICU and hospital stay is less obvious. Our very young
patient group experiences a wide range of ICU and hospital days. In many cases, this variation is due to factors not
immediately associated with VIS such as poor feeding, vocal cord paralysis/paresis, phrenic nerve injury, and
chylothorax. However, even in the presence of these other factors, VIS48 showed a strong, independent association
with length of ICU and hospital stay. Likely high VIS is simply a marker for poor physiology in the immediate
post-operative period. This poor physiology may in turn lead to prolonged therapies, more frequent complications,
and borderline cardiac and pulmonary function that impair convalescence, particularly feeding. Conversely, in the
face of conflicting data in the literature concerning the risk/benefit of specific medications for cardiovascular
support [4, 7-9, 15], our study results do not exclude a true causal relationship between high levels of
vasoactive/inotropic support and the need for extended intensive and general hospital care. Our population had too
few events in the poor outcome variable to undergo rigorous statistical testing. Trends in our population, however,
agree with the findings of the prior study, associating high VIS scores with increased odds of cardiac arrest and
death . Contrary to the retrospective study, we did not find an association between high VIS and time to
negative fluid balance. This difference may be explained by variable clinical strategies regarding the use of
vasoactive medications to stabilize hemodynamics and augment urine output. We also found no association
between VIS and biochemical markers such as peak lactate or change in creatinine. In most of our patients, peak
lactate was found immediately after arrival in the ICU and for the most part, represents physiology during the
operation. Changes in creatinine were small and would require a much larger cohort to detect statistical differences.
We attempted to design our study to minimize issues associated with observational studies. By utilizing
prospectively collected data, we hoped to reduce the potential biases inherent in retrospective studies, particularly
recording bias in the medical record. We also felt that it was important to control for the effects of DSC on our
outcome measures. Not only does DSC account for a very different pre-operative risk profile, we hypothesized that
it also represents the culmination of multiple intra-operative issues that might not be captured by specific variables
such as bypass or cross clamp time. In addition, following the operation, DSC has direct effects on length of
intubation due to the inability to extubate a patient with an open sternum. Failure to account for DSC during
multivariate analysis risks introduction of a significant confounding variable in our study. Lastly, the inclusion of
non-bypass patients in the analysis increases the applicability of VIS to the general cardiac intensive care
Our study has a number of limitations. It reflects a single institutional experience in a well defined patient
population. The sample size is relatively small with significant variability in the outcome measures. Because
clinical management was not under protocol, patient progression may have been affected by variations in attending
physicians’ practices. Finally, VIS is subject to the same inherent weakness of many scoring system: it attributes
arbitrary power to the different factors included in the equation without any assessment of the relative importance of
the individual components.
Additional research is needed to better refine the potential uses of VIS. Based on the findings of this study and
those of Gaies et al , VIS should replace IS as the best measure of cardiovascular support available for research
involving this patient population. From a clinical perspective, a high VIS at 48 hours should trigger physician
awareness that the infant in question continues to be at risk for poor outcome. Use of this scoring system for
different age groups and in different disease entities may be warranted but requires further research. Future studies
should be multi-centered and adequately powered to detect small differences short term outcomes, particularly
cardiac arrest and mortality. They should also address some important unanswered questions such as the relative
importance of different vasoactive/inotropic medications and should include a long term follow-up plan to assess
neurologic outcome and long term morbidity and mortality.
Vasoactive-inotropic score at 48 hours after cardiac surgery is a simple clinical tool that can provide valuable
information regarding likely length of intubation, intensive care unit stay, and hospital stay. VIS at 48 hours
performs better than maximum VIS in the first 48 hours after surgery in predicting poor short term outcomes.
Within the first 72 hours after surgery, VIS is a stronger predictor of poor short term outcome than inotrope score.
Given these findings, we believe that VIS, particularly at 48 hours, should replace the previous inotrope score as the
best available measure of cardiovascular support after cardiac surgery in infants.
We would like to thank Dr. James Jaggers for his assistance in preparing this manuscript. We would also like to
recognize the nurses and staff of the CICU and CTRC at the Children’s Hospital of Colorado for their valuable
contributions to our project. This study was supported by grant MO1-RR00069, General Clinical Research Centers
Program, NCRR, NIH.
1. Parr GV, Blackstone EH, Kirklin JW (1975) Cardiac performance and mortality early after intracardiac
surgery in infants and young children. Circulation 51, 867-874.
2. Wernovsky G, Wypij D, Jonas RA, Mayer JE, Jr., Hanley FL, Hickey PR, Walsh AZ, Chang AC,
Castaneda AR, Newburger JW (1995) Postoperative course and hemodynamic profile after the arterial
switch operation in neonates and infants. A comparison of low-flow cardiopulmonary bypass and
circulatory arrest. Circulation 92, 2226-2235.
3. Froese NR, Sett SS, Mock T, Krahn GE (2009) Does troponin-I measurement predict low cardiac output
syndrome following cardiac surgery in children? Crit Care Resusc 11, 116-121.
4. Hoffman TM, Wernovsky G, Atz AM, Kulik TJ, Nelson DP, Chang AC, Bailey JM, Akbary A, Kocsis JF,
Kaczmarek R, Spray TL, Wessel DL (2003) Efficacy and safety of milrinone in preventing low cardiac
output syndrome in infants and children after corrective surgery for congenital heart disease. Circulation
5. Millar KJ, Thiagarajan RR, Laussen PC (2007) Glucocorticoid therapy for hypotension in the cardiac
intensive care unit. Pediatr Cardiol 28, 176-182.
6. Moffett BS, Mott AR, Nelson DP, Goldstein SL, Jefferies JL (2008) Renal effects of fenoldopam in
critically ill pediatric patients: A retrospective review. Pediatr Crit Care Med 9, 403-406.
7. Burton GL, Kaufman J, Goot BH, da Cruz EM (2011) The use of Arginine Vasopressin in neonates
following the Norwood procedure. Cardiol Young, 1-9.
8. Mastropietro CW, Clark JA, Delius RE, Walters HL, 3rd, Sarnaik AP (2008) Arginine vasopressin to
manage hypoxemic infants after stage I palliation of single ventricle lesions. Pediatr Crit Care Med 9, 506-
9. Mastropietro CW, Rossi NF, Clark JA, Chen H, Walters H, 3rd, Delius R, Lieh-Lai M, Sarnaik AP (2010)
Relative deficiency of arginine vasopressin in children after cardiopulmonary bypass. Crit Care Med 38,
10. Gaies MG, Gurney JG, Yen AH, Napoli ML, Gajarski RJ, Ohye RG, Charpie JR, Hirsch JC (2010)
Vasoactive-inotropic score as a predictor of morbidity and mortality in infants after cardiopulmonary
bypass. Pediatr Crit Care Med 11, 234-238.
11. Jacobs ML, Jacobs JP, Jenkins KJ, Gauvreau K, Clarke DR, Lacour-Gayet F (2008) Stratification of
complexity: the Risk Adjustment for Congenital Heart Surgery-1 method and the Aristotle Complexity
Score--past, present, and future. Cardiol Young 18 Suppl 2, 163-168.
12. Lacour-Gayet F, Clarke D, Jacobs J, Gaynor W, Hamilton L, Jacobs M, Maruszewski B, Pozzi M, Spray T,
Tchervenkov C, Mavroudis C (2004) The Aristotle score for congenital heart surgery. Semin Thorac
Cardiovasc Surg Pediatr Card Surg Annu 7, 185-191.
13. Li J, Zhang G, Holtby H, Cai S, Walsh M, Caldarone CA, Van Arsdell GS (2008) Significant correlation of
comprehensive Aristotle score with total cardiac output during the early postoperative period after the
Norwood procedure. J Thorac Cardiovasc Surg 136, 123-128.
14. Lambert J and Lipkovich I, A macro for getting more out of your ROC curve. SAS Global Forum 2008.
2008, SAS: San Antonio, Texas.
15. Li J, Zhang G, Holtby H, Humpl T, Caldarone CA, Van Arsdell GS, Redington AN (2006) Adverse effects
of dopamine on systemic hemodynamic status and oxygen transport in neonates after the Norwood
procedure. J Am Coll Cardiol 48, 1859-1864.