Clinical Profile and Validity of PRISM III among Children Admitted in A PICU Over One Year
- 1. Military Hospital Dehradun, India
ABSTRACT
Introduction: Ability to quantify the severity of the patient and predict the probability of death based on the clinical state of admitted child can help decision making selection of treatments, ethical issues and economic strategies.
Aim and objectives: To analyze clinical and demographic profile of children admitted in PICU and predictive value of PRISM score.
Material and methods: It was a prospective, observational, Clinical study. Minimal sample size for the sample to be representative was 240. Consecutive type of non-probability sampling was used during the study. Receiver operator characteristic curve analysis was done for evaluating the efficacy of PRISM Score to predict mortality.
Results: During the study period we had total 315 admissions with 112 being direct admissions while rests 202 were transfer in from ward. Out of these 315 patients 262 were either transferred out of PICU or discharged to home while 49 children died thus mortality rate was 15%. At cut-off of 10.5, PRISM score at 12 hours and at cut-off of 8.5, PRISM score at 24 hours had good sensitivity and specificity. An inverse correlation was observed between PRISM score at 12 hours and 24 hours with hospital stay i.e. higher PRISM score was correlating with decreased hospital stay.
Conclusion: There was male predominance and younger rage groups among PICU admissions. PRISM score at presentation may be used as a tool in predicting mortality in critically ill children. Hemato-onclogy and postoperative status were the commonest indications for PICU admissions in our set up.
CITATION
Ashish Kumar S, Garg P (2020) Clinical Profile and Validity of PRISM III among Children Admitted in A PICU Over One Year. Ann Pediatr Child Health 8(2): 1173.
KEYWORDS
- PICU
- PRISM score
- Children
- Mortality
- Critically ill children
INTRODUCTION
Pediatric intensive care units (PICU) are meant to provide sophisticated care for children and adolescents, but increased quantum of therapy can also increase the suffering, financial burden and in the end, may just prolong the death process [1]. Because of this, it is important to be able to quantify the severity of the patient and predict the probability of death based on the clinical state of admitted child [2]. This should guide decision making in important issues like selection of treatments, ethical issues and economic strategies [3]. Furthermore, comparing the mortality to the severity of illness, they can be also used to classify patients and may be used to compare clinical studies and technological resources [4].
The Pediatric Risk of Mortality (PRISM) score is one of the main indicators used in the PICU. It was obtained and validated from the Physiologic Stability Index (PSI) [5]. It uses 14 parameters (physiological and laboratory data) and for each one was used the highest severity value recorded in the first 24 hours. The risk of death is calculated by a logistic regression equation using the value of the PRISM, patient age and need of surgery on admission to the PICU [6]. The PRISM has been developed and validated in tertiary PICUs in mostly developed countries [7,8], with very few reports from India [9-11]. Due to this fact, this study has been planned to evaluate the PRISM score and to correlate it with mortality in patients admitted to the PICU of a northern Indian tertiary referral hospital. We collected data in order to analyze clinical and demographic profile of children admitted in PICU and to see predictive value of PRISM score in our setting for predicting mortality and length of hospital stay.
MATERIAL AND METHODS
This study was done at PICU of a teaching hospital and tertiary care referral centre of north India. Study Population included all the children who were admitted in PICU between ages 1 month to 18 years of age. It was a prospective, observational, Clinical study conducted from December 2017 to November 2018 with consecutive admitted patients to PICU who were eligible for recruitment in the study. Exclusion Criteria being patients not completing 24 hours in PICU either as they were transferred out or expired. Children whose parents were unwilling to give consent were also no included in the study. After taking informed consent, clinical details and blood samples, relevant for PRISM III (Table 1), calculations were collected at 12 and 24 hours. Minimal sample size for the sample to be representative was 240 calculated based on Danielle’s formula with prevalence of PICU admission in our hospital taken as 13% from data of last 3 years. Consecutive type of non-probability sampling was used during the study. Receiver operator characteristic curve analysis was done for evaluating the efficacy of PRISM Score to predict mortality.
RESULTS
During the study period we had total 347 admissions out of which 32 were excluded as 9 had died within 12 hours of admission and 23 were shifted out of PICU within 24 hours .Out of 315 admissions with 112 being direct admissions while rest 202 were transfer in from ward .None of the parents denied consent. Outcome of these 315 patients was that 262 were either transferred out of PICU or discharged to home while 49 children died thus mortality rate was 15%. Demographic distribution is shown in Table 2. We classified these admissions on basis of primary system involved as shown in table 3. Outcome was also analyzed based on primary diagnosis and revealed that highest mortality was associated with CVS, infection and hematooncology. Postoperative cases had low mortality although they were the most common indication for PICU admission. Receiver operator characteristic curve analysis was done for evaluating the efficacy of PRISM Score to predict mortality. Area under ROC curve was 0.875 (CI: 0.813-0.937), and 0.905 (CI: 0.844-0.967) for PRISM score at 12 and 24 hours respectively. At cut-off of 10.5, PRISM score at 12 hours had sensitivity and specificity of 71.8% and 85.5% while PPV and NPV was 49.1% and 94% respectively. At cut-off of 8.5, PRISM score at 24 hours had sensitivity and specificity of 86.1% and 83.8% while PPV and NPV was 49.2% and 97.1% respectively (Figure 1). An inverse correlation was observed between PRISM score at 12 hours and 24 hours with hospital stay i.e. higher PRISM score was correlating with decreased hospital stay. The contradictory findings can be attributed to higher mortality (i.e. lower hospital stay) among cases with high PRISM score (Table 4).
RESULTS
During the study period we had total 347 admissions out of which 32 were excluded as 9 had died within 12 hours of admission and 23 were shifted out of PICU within 24 hours .Out of 315 admissions with 112 being direct admissions while rest 202 were transfer in from ward .None of the parents denied consent. Outcome of these 315 patients was that 262 were either transferred out of PICU or discharged to home while 49 children died thus mortality rate was 15%. Demographic distribution is shown in Table 2. We classified these admissions on basis of primary system involved as shown in table 3. Outcome was also analyzed based on primary diagnosis and revealed that highest mortality was associated with CVS, infection and hemato-oncology. Postoperative cases had low mortality although they were the most common indication for PICU admission. Receiver operator characteristic curve analysis was done for evaluating the efficacy of PRISM Score to predict mortality. Area under ROC curve was 0.875 (CI: 0.813-0.937), and 0.905 (CI: 0.844-0.967) for PRISM score at 12 and 24 hours respectively. At cut-off of 10.5, PRISM score at 12 hours had sensitivity and specificity of 71.8% and 85.5% while PPV and NPV was 49.1% and 94% respectively. At cut-off of 8.5, PRISM score at 24 hours had sensitivity and specificity of 86.1% and 83.8% while PPV and NPV was 49.2% and 97.1% respectively (Figure 1). An inverse correlation was observed between PRISM score at 12 hours and 24 hours with hospital stay i.e. higher PRISM score was correlating with decreased hospital stay. The contradictory findings can be attributed to higher mortality (i.e. lower hospital stay) among cases with high PRISM score (Table 4).
Table 1: PRISM SCORE performa.
| Parameter | Within 12 hours | Within 24 hours | |
| 01 | Systolic BP | ||
| 02 | Diastolic BP | ||
| 03 | Heart rate | ||
| 04 | Respiratory rate | ||
| 05 | PaO2/FiO2 | ||
| 06 | PaCO2 | ||
| 07 | Glasgow coma score | ||
| 08 | Pupillary reactions | ||
| 09 | PT/PTT | ||
| 10 | Total bilirubin | ||
| 11 | S. Potassium | ||
| 12 | S. Calcium | ||
| 13 | Glucose | ||
| 14 | Bicarbonate in mEq/L |
Table 2: Demographic characteristics among children admitted in PICU during study period.
| Sex wise distribution | |
| Male | 192 (60%) |
| Female | 123(40%) |
| Age wise distribution | |
| Below 1 year | 120 (38%) |
| 1year- 5 years | 121 (38%) |
| 5-10 years | 49 (15%) |
| Above 10 years | 25 (07%) |
Table 3: Admissions and mortality data based on primary system involvement.
| Primary System | Total admissions( % of total admissions) | Total mortality | Mortality (% among admissions in that group) |
| Hemato-oncology | 71 (22%) | 17 | 23% |
| Surgical | 71 (22%) | 3 | 4% |
| Respiratory | 23 (7%) | 5 | 21% |
| CNS | 49 (15%) | 4 | 8% |
| CVS | 23 (7%) | 7 | 30% |
| Nephrology | 30 (9%) | 2 | 6% |
| GIT | 18 (5%) | 5 | 27% |
| Metabolic | 11 (3%) | - | |
| Infection | 12 (3%) | 3 | 25% |
| Other | 3 (1%) | - |
DISCUSSION
Analysis of demographic distribution shows that in PICU more male children got admitted during study period than female children and this difference was statistically significant. This may reflect male susceptibility to various serious diseases but it could also be because of more likelihood of a male child to be brought to PICU as compared to female child due to still existing preference to a male child among Indian families. This finding is similar to PICU data generated from North India like recently published study by Makhija et al. [12]. Age wise distribution emphasizes vulnerability in early age with maximum admissions in first year of life (38%), and in age group between 1-5 years (38%). This finding highlights the importance of infant mortality rate and under five mortality rates as overall indices for health care system of a country. In our experience most common PICU admission indications were related to hemato-oncology and surgical and neurological involved were followed by this and that. This is very different from other studies where infection and respiratory distress are the commonest indications [13]. This reflects the difference in our patient population and it being a referral center where admissions are based on referral and direct admissions are catered at another hospital. This gives a future direction that PICUs specifically catering to hemato-oncolgy and post surgical can be developed in future like pediatric cardiac ICUs which are already functional in our hospital as well as at many institutes in India.
Overall mortality among PICU admission was 15% which is comparable to other PICUs as reported by Singhal D et al. [14], Madaan et al. [15], and Hassan et al. [16], with mortality rates of 18%, 12.5% and 17% respectively. Probability of death based on primary system involved revealed that children with cardiovascular system involvement were most likely to die in PICU. The reason of this is that only children who came to general PICU were the ones where surgery could not be offered either because of poor general condition or secondary to inoperable cardiac condition. Although admission numbers of children requiring surgical management and those having hematooncolgy diseases were similar there was marked difference in mortality with former group having 4% mortality while later had significantly higher rate of 23%. This reflects impact of basic disease and treatment related complications like febrile neutropenia are more common with hematoncological disorders.
In present study, mean PRISM score at 12 and 24 hours was significantly higher among expired. PRISM 12 score of 10.5 and PRISM 24 score of 8.5 were reasonable cut off points in our study to predict chances of survival and mortality. Our results are also comparable to other studies showing PRISM score as a as a sensitive predictor of mortality [13-17]. El?Nawawy A et al. [18], in their study observed that non-survivors had a significantly higher mean score compared with survivors (36 vs. 17). Bellad et al., reported an overall mortality of 16.7% with 89.2% accuracy at cut-off score of 15 [19]. Costa A et al. [20], observed median pediatric risk of mortality score as significantly lower in patients who survived (p8 as a significant predictor of mortality (chi-square value of 29.615 and a OR of 9.28 (9 times more risk of dying compared to patients with a prism score >8). Dey et al. [21], in their study observed PRISM as a sensitive predictor of outcome at a cut-off point of 13.5.
An inverse but non-significant correlation was observed between PRISM score with PICU and hospital stay. These contradictory findings can be attributed to higher mortality among cases with high PRISM score. At cut-off of 15, hospital stay decreased significantly i.e. higher mortality (i.e. lower hospital stay) was associated with PRISM score > 15. Madaan et al. [15], did not observe any difference in the mean duration of ICU stay among the survivor and non-survivors. Bellad et al. [18], reported that the mean ICU stay among non-survivor was shorter when compared to those survivors. As the inverse of the correlation between PRISM and mean ICU stay is considered to be an indicator of the quality of PICU services [22], a difference in the quality of care available at various centers may explain the observed variability in the ICU stay being reported in different studies. Brindha et al. [23], attributed the median length of PICU stay with the type of primary disease condition affecting the patients and not to the PRISM score of the patients.
Table 4: correlation of Prism score with length of hospital stay and PICU stay.
| Pearson co-relation | ||
| PRISM 24 | r- value | p- value |
| PICU | -0.04 | 0.585 |
| Hospital | -0.13 | 0.043 |
| PRISM 12 | ||
| PICU | -0.06 | 0.399 |
| Hospital | -0.17 | 0.011 |
CONCLUSION
The present study was done in a tertiary care referral hospital to evaluate the demography, clinical profile and outcome of patients admitted to PICU by using PRISM score. There was male predominance among PICU admissions and younger rage groups (infants and below five were disproportionately represented in the PICU population. PRISM score at presentation may be used as a tool in predicting mortality in critically ill children. In our setup hemato-oncolgy and postoperative status were the commonest indications for PICU admissions giving us a roadmap for future development of PICUs specifically catering to this population.
