Validation of the vasoactive-inotropic score in predicting pediatric septic shock mortality: A retrospective cohort study.

INTRODUCTION: Mortality in pediatric septic shock remains very high. Vasoactive-inotropic score (VIS) is widely used to predict prognosis in patients with heart disease. It is a simple method that was initially used as a predictor of morbidity and mortality in postoperative patients with congenital heart diseases. Previous reports showed that high VIS score was associated with high mortality in pediatric sepsis. However, its discriminative value remains unclear. We aim to explore the discriminative value of VIS in predicting mortality in pediatric septic shock patients.
METHODS: We conducted a retrospective cohort study on medical records of septic shock patients who received care in the pediatric intensive care unit (PICU). We screened medical records of pediatric patients which were diagnosed with septic shock and admitted to the PICU and received vasoactive/inotropic score for more than 8 h. Other supporting examination results were recorded, such as organ function evaluation for calculation of Pediatric Logistic Organ Dysfunction-2 (PELOD-2) score. The outcome of patients was recorded. The receiver operating curve was constructed to calculate the area under the curve (AUC), sensitivity, and specificity of each cutoff point.
RESULTS: We obtained the optimum cutoff point of VIS > 11 with 78.87% sensitivity and 72.22% specificity. AUC positive was 0.779 (P < 0.001); predictive value and negative predictive value were 91.80% and 46.43%, respectively.
CONCLUSION: VIS > 11 has a good ability to predict mortality in children with septic shock. Copyright:

INTRODUCTION

Sepsis describes a state of organ dysfunction due to dysregulated immune response toward infections.[1] Today, uniform diagnostic criteria widely used in research to identify sepsis in the pediatric population are based on the International Pediatric Sepsis Consensus Conference (IPSCC).[2] However, the SPROUT trial found that only 42% of sepsis patients were identified as such by both the clinician and the consensus criteria.[3] According to IPSCC, sepsis is formally defined as systemic inflammatory response syndrome attributed to infection.[2] Severe sepsis is defined as sepsis with either cardiovascular organ dysfunction or acute respiratory distress syndrome or two or more other organ dysfunctions. Septic shock is particularly defined as sepsis with cardiovascular organ dysfunction.[1] Septic shock happens when sepsis causes persistent cardiovascular dysfunction despite ≥40 mL/kg fluid resuscitation.[3] Sepsis is responsible for high morbidity and mortality among children.[4] Today, septic shock mortality in developed countries is as low as 5%, whereas in developing countries, it is around 35%.[4]

Several scoring systems have been validated to predict mortality in pediatric sepsis; for instance, pediatric sequential organ failure assessment and Pediatric Logistic Organ Dysfunction-2 (PELOD-2) scores. However, the requirement for extensive laboratory examinations limit their use in terms of availability and affordability.[5] Thus, availability and affordability limit their use.

The vasoactive-inotropic score (VIS) was originally used to predict morbidity and mortality in patients with cardiac diseases.[6] It represents the need for medications to support the cardiovascular function. A study by Haque et al. demonstrates high VIS score's association with mortality in septic children with refractory shock.[7]

Calculating VIS is relatively easier than PELOD-2 score, however its use as mortality predictor in pediatric septic shock has yet to be explored.

METHODS

We conducted a retrospective cohort study on pediatric patients diagnosed with septic shock and admitted to the pediatric intensive care unit (PICU) of Cipto Mangunkusumo Hospital between December 2015 and November 2019. This study obtained ethical approval from the Faculty of Medicine Universitas Indonesia Ethical Committee with approval number KET.1198/UN2.FI/ETIK/PPM.00.02/2019 on October 2019.

Inclusion criteria were pediatric patients aged 1 month to 18 years, with ICD 10 administrative coding of “other sepsis” (A41) and/or “streptococcal sepsis” (A40) and/or septic shock (R65.21) and received treatment at the PICU. We further screen for patient presenting with sepsis directly admitted to the PICU from the emergency room, exclusive of patients who developed septic shock during their hospital stay and transferred to the PICU (floor admission). Exclusion criteria were incomplete information to calculate VIS and PELOD-2 score, the patient referred by other PICU, and those with hospital LOS <8 h (including those who died or transferred to other hospitals). PELOD-2 score calculated was based on the examination performed within the first 24 h of sepsis diagnosis. VIS was calculated based on the maximum dose of vasoactive and inotropic drugs used within the first 24 h since the septic shock diagnosis. The value of VIS equals to dopamine dosage (μg/kg/min) + dobutamine dosage (μg/kg/min) + 100 × epinephrine dosage (μg/kg/min) + 10 × milrinone dosage (μg/kg/min) + 10,000 × vasopressin dosage (U/kg/min) + 100 × norepinephrine dosage (μg/kg/min).[6]

The minimal sample size was calculated using that of diagnostic value cutoff point estimation, with 5% significance, absolute accuracy of 10%, and expected sensitivity of 85%. The minimal sample size calculated was 85 samples. We identified 213 medical records through electronic health record (Integra eOffice© SEVIMA, Surabaya, Indonesia). Incomplete records were excluded upon physical medical records screening. We collected the following variables for each data set: demographics data, chief complaint (related to sepsis diagnosis), history of surgery during hospital stay, history of mechanical ventilator use during hospital stay, PICU LOS, mortality outcome, maximum VIS score calculated on the 1st day of vasoactive-inotropic use, laboratory and physical examination data needed for PELOD-2 score on the 1st day of sepsis diagnosis, any diagnosis/suspected congenital heart defects, and immunosuppression (any presence/suspect of either malignancy or immunological cause of impaired immune system). We performed statistical analysis using SPSS version 17.0 (IBM Corp, Armonk, USA). Descriptive data are presented as percentage. Discriminative value of VIS was obtained using receiver operator characteristic (ROC) curve and expressed as area under the curve. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) are presented as 95% confidence interval.

RESULTS

The study team reviewed 114 medical records, 12 were excluded due to incomplete data, and 13 were screened doubly. A final total of 89 medical records of pediatric patients diagnosed with septic shock who were admitted to PICU were reviewed and analyzed.

The distribution of patient characteristics is described in Table 1. The majority of septic shock patients were <5 years of age and admitted with a respiratory chief complaint. During hospitalization, 23.6% of patients underwent surgery, 79.8% required mechanical ventilation, and 67.5% stayed in the PICU for more than 48 h. The overall patient mortality was 79.8%.

Table 1

Patient characteristics (n=89)

Variable	n (%)
Age (year-old)
1-5	65 (73.0)
6-18	24 (27.0)
Gender
Male	49 (55.1)
Female	40 (44.9)
Initial presenting diagnosis
Respiratory system	49 (55.1)
Gastroenterology and hepatology system	29 (32.6)
Neurology system	23 (25.8)
Cardiovascular system	14 (15.7)
Malignancy	13 (14.6)
Malnutrition	7 (7.9)
Endocrinology system	4 (4.5)
Nephrology system	2 (2.2)
Immunology	2 (2.2)
Trauma	0 (0.0)
Etc.	8 (9.0)
Surgery
Yes	21 (23.6)
No	68 (76.4)
Mechanical ventilation
Yes	71 (79.8)
No	18 (20.2)
PICU length of stay (h)
≤48	29 (32.5)
>48	60 (67.5)
Maximum VIS score
<25	57 (64.0)
≥25	32 (36.0)
PELOD-2 score
˃9	48 (53.9)
≤9	41 (46.1)
Outcome
Survivors	18 (20.2)
Nonsurvivors	71 (79.8)

PICU: Pediatric intensive care unit; VIS: Vasoactive-inotropic score; PELOD-2: Pediatric Logistic Organ Dysfunction-2

We performed bivariate analysis on each risk factor that may affect patient's survival. The use of mechanical ventilator and PICU LOS were significantly associated with mortality. However, on further adjustment to other variables using multiple logistic regressions, no significant association was found [Table 2]. The mortality rate was statistically different between patients with high VIS score (VIS > 25) and low VIS score, P = 0.0051. Similarly, the mortality rate was statistically different between patients with PELOD-2 score >9, PELOD-2 score 4–9, and PELOD-2 score 0–3 [Table 3].

Table 2

Association between patient characteristics and mortality

Variable	Nonsurvivors (n=71) n (%)	Survivors (n=18) n (%)	Crude RR (95% CI)	P	Adjusted RR	P
Age (year-old)
1-5	53 (81.5)	12 (18.5)	1.09 (0.84-1.41)	0.556a
6-18	18 (75.0)	6 (25.0)
Gender
Male	37 (75.5)	12 (24.5)	0.89 (0.72-1.09)	0.268
Female	34 (85.0)	6 (15.0)
Surgery
Yes	18 (85.7)	3 (14.3)	1.10 (0.89-1.36)	0.546
No	53 (77.9)	15 (22.1)
Immunosuppression (malignancy, immunological)
Yes	14 (93.3)	1 (6.7)	1.21 (1.01-1.46)	0.288
No	57 (77.0)	17 (23.0)
Congenital heart defects
Yes	10 (71.4)	4 (28.6)	0.88 (0.62-1.24)	0.470
No	61 (81.3)	14 (18.7)
Mechanical ventilator
Yes	61 (85.9)	10 (14.1)	1.55 (1.01-2.36)	0.008a,*	3.24 (0.99-10.58)	0.052
No	10 (55.6)	8 (44.4)
PICU length of stay (h)
≤48	28 (96.6)	1 (3.4)	1.35 (1.13-1.60)	0.006*	8.06 (0.98-66.28)	0.052
>48	43 (71.7)	17 (28.3)

Chi-square test., aFisher’s exact test, *P<0.05. RR: Risk ratio, PICU: Pediatric intensive care unit, CI: Confidence interval

Table 3

Relationship between vasoactive-inotropic score, PELOD-2 score, and mortality in shock septic patients

	Survivors (n=18) n (%)	Nonsurvivors (n=71) n (%)	RR (95% CI)	P
VIS score
<25	16 (27.8)	41 (72.2)	1.3034 (1.08-1.57)	0.0051*
≥25	2 (6.7)	30 (93.3)
PELOD-2 score
>9	4 (8.3)	44 (91.7)	4.583 (0.79-26.51)	0.08*
4-9	10 (27.8)	26 (72.2)	3.611 (0.62-21.09)	0.15
0-3	4 (80.0)	1 (20.0)	Reference

Fisher’s exact test. *P<0.05. VIS: Vasoactive inotropic score, PELOD-2: Pediatric Logistic Organ Dysfunction-2, RR: Risk ratio, CI: Confidence interval

We obtained cutoff value for VIS and PELOD-2 score to predict mortality. VIS score of 11 was shown to be the optimum cutoff point with 78.87% (67.56%–87.97%) sensitivity, 72.22% (46.52–90.30) specificity, 91.80% (84.04–95.97) PPV, and 46.43% (33.71–59.63) NPV in predicting septic shock mortality in our study. The AUC of VIS score was 0.757 and was shown to be able to discriminate nonsurvivors among our pediatric septic shock patients (P = 0.001) [Table 4 and Figure 1].

Table 4

Discriminative value of PELOD-2 score, PELOD-2 score without lactate, and vasoactive-inotropic score in predicting mortality outcome in pediatric shock sepsis patients

	PELOD-2	PELOD-2 without lactate	VIS
Cutoff point	>10	>9	>11
Sensitivity, % (95% CI)	61.97 (49.67-7324)	67.61 (55.45-78.24)	78.87 (67.56-87.67)
Specificity, % (95% CI)	83.33 (58.58-96.42)	66.67 (40.99-86.66)	72.22 (46.52-90.31)
PPV, % (95% CI)	93.62 (83.71-97.67)	88.89 (80.32-94.0)	91.80 (84.04-95.97)
NPV, % (95% CI)	35.71 (27.90-44.37)	34.29 (24.62-45.46)	46.43 (33.71-59.63)
AUC	0.757	0.753	0.779
P	0.001*	0.001*	<0.001*

*P<0.05. PELOD-2: Pediatric Logistic Organ Dysfunction-2, VIS: Vasoactive inotropic score, AUC: Area under curve, NPV: Negative predictive value, PPV: Positive predictive value, CI: Confidence interval

Figure 1

Receiver operating curve of vasoactive inotropic score,Pediatric Logistic Organ Dysfunction-2 score without lactate in predicting mortality in pediatric septic shock

Using ROC, PELOD-2 without lactate score was able to discriminate nonsurvivors with AUC of 0.753 (P < 0.001). Using PELOD-2 without lactate, we obtain optimal cutoff point of >9 with 67.61 (55.45–78-24) sensitivity and 66.67 (40.99–86.66) specificity. Similarly, PELOD-2 score (with lactate) was able to discriminate nonsurvivors with AUC of 0.779 (P = 0.001). Using PELOD-2 cutoff point of >10, sensitivity and specificity were 78.87 (67.56–87.67) and 72.22 (46.52–90.31), respectively [Table 4].

DISCUSSION

This study demonstrates that VIS can be used to predict mortality in our pediatric septic shock patients. Secondarily, we also re-validate the use of PELOD-2 score in predicting clinical outcome in pediatric shock sepsis. In addition, we also found that omitting the lactate component in PELOD-2 did not compromise its discriminative value, however different cutoff value should be used.

Our data demonstrate high prevalence of pediatric sepsis cases that fell into septic shock of 17.6% with mortality rate reaching 78.6%. Our mortality rate is considerably higher compared to previously published data. A meta-analysis by Tan et al. comprised of 94 studies with a cumulative of 7561 severe sepsis and septic shock patients demonstrate that mortality rate is significantly higher in developing countries compared to developed countries.[4] Our even higher mortality rate may be due to the single-center nature of our study which took place in a tertiary hospital. Cipto Mangunkusumo is a teaching hospital and a national referral center. Hence, most pediatric septic shock cases included in this study have underlying disease.

In this study, none of the risk factors identified independently affect the mortality outcome [Table 2]. This is contrary to the findings of Shime et al. that identified hematological diseases, malignancy, and shock as factors that affect mortality among severe sepsis subjects.[8] We think that this may be due to the minimal sample size in this study as the sample size calculated was for determination of cutoff for VIS score and not to find difference in proportion between identified risk factors.

In this study, the value of VIS score used to predict mortality was the maximum dosage of vasopressor and inotropes used in the first 24 h of septic shock diagnosis. This was based on previous study by Musick et al. on 20752 patients admitted to the PICU. Their study found that the initial 24h VIS was better than the following 48h VIS in predicting mortality in critically ill PICU patients (AUC 0.736 vs 0.788 respectively).[9]

We obtained the optimal cutoff value of VIS in predicting mortality in pediatric septic shock of > 11 with AUC 0.779 [Table 4]. This supports previous study by Musick et al. and Haque et al. that demonstrate the usefulness of VIS in predicting mortality in critically ill patients.[79] In their study, Haque et al. used different cutoff or VIS of 20 to analyze the increased risk of mortality among their pediatric septic shock patients. However, the nature of their study was to find a correlation between high VIS score, instead of finding the optimum cutoff point for VIS in predicting mortality in pediatric septic shock patients. Using our data ROC, VIS > 20 has an AUC of 0.734, which is lower than VIS > 11. Furthermore, using VIS > 20, the sensitivity calculated was 57.75%, hence certain cases in which mortality risk should be anticipated might be underestimated (data not shown).

The use of VIS in septic shock patients has a good correlation with mortality as demonstrated in our study [Table 3] and Haque et al.[7] The calculation of VIS takes into account the dosage of vasoactive and inotropic agents used to support cardiovascular performance in the event of shock. The use of VIS was originally applied to predict the outcome based on the hemodynamic supports in patients post cardiac surgery.[610] Since then, VIS has been widely used in cardiogenic shock.[11] We identified two prior studies in pediatric septic shock which used VIS. Mcintosh et al. evaluated VIS in pediatric sepsis and concluded that VIS was independently associated with clinical outcome such as PICU LOS, length of ventilator use, cardiac arrest incidence, extracorporeal membrane oxygenation use, and mortality.[12] A study by Haque et al. also demonstrated that high VIS score (>21) was associated with mortality.[7] This significance of VIS in pediatric sepsis may be due to the fact that hemodynamic characteristics in pediatric septic shock are similar to those of cardiogenic shock.[13]

The use of the PELOD-2 score to predict mortality in pediatric patients has been well documented. A study by Zhong et al. demonstrated that the PELOD-2 score assessed on the first 24 h of admission has a good association to mortality outcome. Moreover, their study settings have similar characteristics to ours in which both were conducted in developing countries and in pediatric patients receiving care at the PICU. They also concluded that in developing countries, the PELOD-2 score can be used to standardize the diagnosis of sepsis using the new sepsis definition which includes the involvement of organ dysfunction.[14]

In addition, we also validated the use of PELOD-2 score without lactate component that yielded the same predictive value as PELOD-2 in predicting mortality in pediatric septic shock patients. The incorporation of lactate in the PELOD-2 score may be due to it being an indirect marker of tissue hypoperfusion.[15] However, previous studies in children demonstrated inconsistent results in determining the threshold of lactate value in pediatric sepsis to predict mortality.[1617] Similar to our study, Yuniar et al. demonstrated that lactate >2.5 mmol/L was not associated with mortality. Their study demonstrated that only 37% of pediatric shock patients had increased lactate.[18] A previous study by Scott et al. demonstrated normalization of lactate within 2–4 h of presentation, with decreased risk of persistent organ dysfunction.[19] In our study, PELOD-2 score without lactate shows similar predictive value in pediatric sepsis outcome compared to PELOD-2 score with lactate. This may be explained by the lactate value used for PELOD-2 score calculation that was examined within the first 24h, while lactate levels actually fluctuates within a narrower time frame.

Despite significant results, our study has its limitations. First, the retrospective nature of this study may overlook inconsistency in medical data recorded by different treating physicians. Furthermore, there is also a potential of recall bias as the administrative diagnosis was retrospectively assigned, which may have some discrepancy with the actual clinical diagnosis. Second, the small sample size and single-center nature of the study may limit the generalizability of the result. Potential selection bias may occur as we conducted a study in a national referral hospital. This may result in overrepresentation of septic shock in the more severe spectrum. Third, we only analyzed VIS score within the first 24 h and not serial VIS assessment. Furthermore, although with comparable discriminative value as PELOD-2 score, VIS cannot be used to substitute PELOD-2 score in detecting organ dysfunction used in the diagnosis of pediatric sepsis. However, its use to predict mortality is recommended as it is simpler and cost less than PELOD-2 examination. To our knowledge, this is the first study which identifies cutoff of VIS score in predicting mortality in pediatric sepsis patient. It is hoped that it can be used in appropriate clinical setting to help determine the management and prognosis of pediatric sepsis patients.

CONCLUSION

In conclusion, VIS can be used as predictor for mortality in pediatric septic shock with good discriminative ability. The use of VIS is comparable with the PELOD-2 score in predicting septic shock mortality, and it should be considered in appropriate setting. A future prospective, multicentered validation study needs to be conducted to extend the generalizability of its use in predicting mortality in pediatric septic shock.

Research quality and ethics statement

This study was approved by the Institutional Review Board / Ethics Committee (Approval # 1212092-1; Approval date Oct 28, 2019). The authors followed the applicable EQUATOR Network (http://www.equator-network.org/) guidelines during the conduct of this research project.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.