Journal of Pediatric Critical Care

P - ISSN : 2349-6592    |    E - ISSN : 2455-7099

Original Article
Year : 2019 | Volume : 6 | Issue : 3 | Page : 21 - 24

Study of PRISM III Score as A Predictor of Mortality in PICU

Snehal Patel1, Halak Vasavada2, Purvi Patel3, Brijeshkumar Patel4, Parth Shukla4

1Professor & Head of Unit, 2Associate Professor, 3Assistant Professor, 4Resident Doctor Department of Pediatrics, N.H.L. Medical college, Shardaben Municipal General Hospital, Ahmedabad, India.

Correspondence Address:

Dr. Purvi R. Patel
95, MadhavbagTenaments, Nirnaynagar, Ahmedabad-382481
Mob. 9825661647, E-mail – evergreen.purvi@yahoo.in
Received:17-Mar-19/Accepted: 22-May-19/Published online:05-Jun-19

Source of Funding:None Conflict of Interest:None

DOI:10.21304/2019.0603.00499


ABSTRACT
Objectives:
To determine the discriminative ability of PRISM III scoring system in predicting mortality in children admitted to PICU.

Methods:
Prospective observational study in all children 1 month to 12 years of age admitted to PICU during July2017 to June2018.

Result:
Out of 250 patients admitted in the PICU, 133 were male & 117 were female. In the age group of 1 month to 12 months, 30(22.22%) expired while in 12 months to 12 years,18(11.65%) expired. Due to respiratory system involvement, mortality was 24(16.43%), due to central nervous system 8(17.39%) and septicemia 12(46.15%) . At 12 hours of admission, 214 patients had PRISM score 0 to 4, out of which 14 expired. 17 patients had score of 5 to 9, out of which 15 expired.19 patients had score >10 having 100% mortality. PRISM 24 score was taken 12 hours after the 1st score. 216 patients had PRISM score between 0 to 4, out of which 14 expired. 8 patients had score of 5 to 9 & 26 patients had score >10, out of which all expired.

Conclusion:
Since mortality rises with increase in PRISM 3 score at the time of admission, PRISM III score can be taken as an indicator of the initial severity of illness. PRISM III can help in selecting sick children for PICU admission and optimum utilization of limited PICU resources.

Introduction
In context of intensive care, a rational and objective way to defi ne and quantify severity of illness is through the development of probability models predicting mortality risks. Such predictive models or scoring systems have been developed for all age groups including pediatric.
Scoring systems are arrived at for evaluation of the patient’s mortality risk in the ICU by assigning a score to patient and predicting the outcome. However, patient’s mortality is not only affected by ICU performance but also depends on many other factors such as demographic and clinical characteristic of population, infrastructure and nonmedical factors (management and organization), case mix and admission practice. Most important factor is patients’ physiological reserve and severity of illness, irrespective of the primary illness.
Critical illness causes alteration in physiological status and biochemical parameters; hence the severity of illness may be predicted by the degree of alteration in these. PRISM III is score that uses both physiological and biochemical parameters.There are few published studies in Indian scenario to study the utility of PRISM III score in our PICUs, hence this study was undertaken.
The PRISM III score is an improved version of the PRISM score developed at the Children’s National Medical Center in Washington, DC based on data collected at 32 pediatric intensive care units using 11,165 admissions.1-3
This study is helpful to compare PRISM III score at 12 and 24 hours and to determine the discriminative ability of PRISM III scoring system in predicting the mortality and outcome in children admitted to pediatric intensive care unit (PICU).It is institution independent and can be used within limits to compare different intensive care units.3

Materials and Methods
Study Type: Prospective observational study Study population: All children 1 month to 12 years of age admitted to PICU.

Inclusion criteria:
1. All children, 1 month to 12 years of age admitted directly to PICU of a tertiary care center of a teaching hospital.
2. All children, 1 month to 12 years of age admitted to pediatric ward and shifted to PICU later.

Exclusion criteria:
1. PICU stay <4 hour (expired or transferred within 4 hour of admission).
2. Patients who left against medical advice.
3. Non-operative cardiovascular disease.
4. Chromosomal anomaly.
5. Cancer.
6. Post-operative patients.
Study period: July 2017 to June 2018.
The PRISM III score evaluation was done as per recommendation of Pollack et al.11st PRISM score (PRISM 12) was taken within 12 hours of admission to PICU and repeat PRISM score was taken 12 hours after the 1st PRISM score.Non-invasive blood pressure was recorded using blood pressure monitor and oxygen saturation measured with pulse-oximeter at admission.
Arterial blood gas analysis by radial artery prick using standard practice.Standard laboratory techniques were utilized to measure blood levels of total bilirubin, potassium, calcium and glucose. Prothrombin time and partial thromboplastin time were measured.The clinical assessment of heart rate, respiratory rate and pupillary reaction was made by a pediatric resident doctor.The Glasgow Coma score and modifi cation of Glasgow Coma score was utilized. The children were followed up during hospital stay and the outcome measures were recorded as died or survived at the end of the hospital stay.
Since the association between mortality and the PRISM score turned out to be signifi cant in the initial analysis, a logistic regression analysis was done on the discharge status (discharged/Expired), taking PRISM score as a predictor for mortality. The analysis was done using the following equations:
1. For PRISM 12
a logit r = (0.822* PRISM 12 score) – 3.4114,5 The probability of death was calculated by formula: Probability of death = er/(1 + er), where r = (0.822 * PRISM 12) – 3.411 (In our study). Here –3.411 is constant.
2. For PRISM 24
a logit r = (0.853* PRISM 24 score) – 3.4334,5 The probability of death was calculated by formula: Probability of death = er/(1 + er), where r = (0.853* PRISM 24) – 3.433 (In our study). Here –3.433 is constant.

Results
In present study, out of 250 patients, 133 were male & 117 were female. Out of 133 male patients, 106(79.7%) survived and 27(20.3%) expired. Out of 117 female patients,96(82.05%) survived and 21(17.95%) expired.
In the age group of 1 month to 12 months, there were 135 patients,out of which 105(77.87%) survived and 30(22.22%) expired. In the age group of 12 month to 12 years, there were 115 patients, out of which 97(88.35%) survived and 18(11.65%) expired. There was higher number of expiries in infant age group compared to child age group, but the difference is not signifi cant(p=0.18).Distribution according to system involvment is presented in Table 1

Table 1:
Distribution according to Primary System involved


PRISM III Score
PRISM 12 score is taken within 12 hours of admission. Out of 250 patients, there were 214 patients with PRISM score between 0 to 4, out of which 200 Survived and 14 expired. 17 patients had score of 5 to 9, out of which 2 Survived and 15 expired. 19 patients had score >10, and mortality in these cases was 100%. In ROC curve, area under curve is 0.927(92.7%)(Figure 1).

Table 2:
Total PRISM score within 12 hours of admission(PRISM 12):


PRISM 24 score was taken 12 hours after the 1stscore. Out of 250 patients, there were 216 patients with PRISM score between 0 to 4, out of which 14 expired. (Table 3) In ROC curve area under curve is 0.918(91.8%) (Figure 2).
Table 3:
Total PRISM score 12 hour after the 1st score (PRISM 24)




Figure 1 :
PRISM 12: In ROC curve, area under curve is 0.927(92.7%).



Figure 2 :
PRISM 24: In ROC curve, area under curve is 0.918(91.8%).

The results on goodness of the prediction model as seen by the Hosmer-Lemeshow goodness of fi t chisquare are presented in Tables 4 and 5. There was no signifi cant difference between the observed and the expected deaths (p =0.96 and 0.59 respectively).

Table 4:
Goodness of predictive model for PRISM 12 (Observed VS Expected Mortality)



Table 5:
Goodness of predictive model for PRISM 24 (Observed VS Expected Mortality)



Thus, PRISM score can reliably predict mortality in PICU patients and is an important marker for early prediction of mortality.

Discussion
Taking the probability of death at different levels, the plot of false positives vs true positives i.e., ROC curve analysis (Figure 1 & 2) shows that 92.7% and 91.8% of the subjects could be predicted correctly by using PRISM 12 and PRISM 24 scores respectively. Thus, the prediction of PICU mortality based on PRISM Scores was valid and highly reliable. Also, the ROC curves at 12 and 24 hours show that there is no signifi cant difference between area under the curve for 12 hours and 24 hours, which suggests that the 12 hours score is as reliable as the 24 hours score in predicting mortality. As compared to other study by Goyal et al., ROC curve analysis shows area under curve was 89.5%.6 In the present study ROC curve analysis shows area under curve was 92.7% and 91.8% at PRISM 12 and PRISM 24 scores respectively.
There was no signifi cant difference between the observed and the expected deaths (p =0.96 and 0.59 respectively). Thus, PRISM score can reliably predict mortality in PICU patients and is an important marker for early prediction of mortality.
Thus, PRISM III score can prove to be a good tool to evaluate the performance in PICUs. Thus, PRISM III score in PICUs should be considered at both the individual patient level and at the aggregate level.

Conclusion
The prediction of PICU mortality based on PRISM III Scores is valid and highly reliable in Indian circumstances. Both PRISM 12 and PRISM 24 can predict mortality equally. Thus, PRISM scoring only once within 24 hours may be enough for prediction of mortality. Apart from this, many factors like quality of PICU care and various patient factors can also affect outcome and we need large multicenter studies and meta-analysis to further validate the results of this study.

References
1. Pollack MM, Patel KM, Ruttiman UE. PRISM III: An updated pediatric risk of mortality Score. Crit Care Med 1996; 24: 743-52.
2. Pollack MM, Ruttiman UE, Getson PR. The Pediatric Risk of mortality (PRISM) SCORE. Crit Care Med 1988; 16: 1110-6.
3. Balakrishnan G, Aitchison T, Hallworth D, Morton NS. Prospective evaluation of Pediatric Risk of Mortality (PRISM) score. Arch Dis Child 1992; 67: 196-200.
4. Walker SH, Duncan DB. Estimation of the probability of an event as a function of several independent variables. Biometrika 1967; 54: 167-78.
5. Cox DR. The regression analysis of binary sequences (with discussion). J Roy Stat Soc B. 1958; 20: 215-42.
6. Goyal S, Dhyani A. Epidemiological Profi le, Outcome Analysis and PRISM III Score of Patients Admitted in PICU at Bal Chikitsalaya, M.B Hospital, R.N.T. Medical College, Udaipur. Int J Med Paediatr Onc.2015;1(1):1-5.