Scoring to Predict Mortality in Neonatology

1. Introduction

Neonatal mortality scoring systems quantify the "burden of illness" to predict outcomes (mortality/morbidity). They are essential tools in modern neonatology for:

Benchmarking: Comparing performance between different NICUs (risk-adjusted mortality).
Audit & Quality Improvement: Tracking unit performance over time.
Resource Allocation: Determining nurse-patient ratios.
Prognostication: Assisting (but not dictating) clinical decision-making and parental counseling.
Research: Stratifying patients in clinical trials to ensure baseline comparability.

2. Principles of Scoring Systems

To be effective, a score must possess:

Discrimination: Ability to correctly distinguish between survivors and non-survivors (Measured by Area Under ROC Curve - AUC). An AUC > 0.8 is considered excellent.
Calibration: Agreement between predicted probability and observed outcome across different risk strata (Measured by Hosmer-Lemeshow Goodness-of-Fit test).
Simplicity: Minimal data points, routinely available, and easy to compute.
Robustness: Independent of treatment variations (avoiding "treatment bias").

3. Classification of Scoring Systems

Category	Basis	Examples
Demographic/Anatomic	Static variables (Birth weight, Gestational Age)	Birth Weight alone, NMOS
Physiologic	Derangement of homeostasis	SNAP, SNAP-II, SNAP-PE
Combined	Physiology + Demographics	CRIB, CRIB-II
Therapeutic	Intensity of intervention/workload	NTISS, TISS
Transport	Stability during transfer	TRIPS, MINT

4. Major Physiologic & Combined Scores (The "Gold Standards")

A. CRIB (Clinical Risk Index for Babies) #scoring

Designed for VLBW infants (<1500g).

CRIB I (1993):
- Timing: Data collected in first 12 hours.
- Items (6): Birth Weight, GA, Congenital Malformations, Max Base Excess, Min FiO2, Max FiO2.
- Limitations: Heavily influenced by early treatment practices (FiO2).
CRIB II (2003): Updated to remove treatment bias (FiO2 removed).
- Timing: First hour of admission (Admission Score).
- Items (5):
  1. Birth Weight
  2. Excess Base
  3. Sex
  4. Temperature at the time of admission (Hypotherimia risk)
  5. Gestational Age

🧠 Mnemonic

BEST G

Advantages: Simpler, immediate assessment, less influenced by unit protocols.

B. SNAP (Score for Neonatal Acute Physiology) #scoring

Applicable to all birth weights and gestations.

SNAP (1993): Included 34 physiologic variables collected over first 24 hours. Highly accurate but clinically cumbersome.
SNAP-PE: Added "Perinatal Extension" (Apgar score, Birth weight, SGA status).
SNAP-II (2001): Simplified using multivariate regression to 6 key variables (collected in first 12 hours):
1. pH - Lowest
2. Urine Output (<1ml/kg/hr)
3. Mean Arterial Pressure (MAP) - Lowest
4. PaO₂/FiO₂ ratio - worst
5. Seizures (multiple or repeatitive)
6. Temperature (lowest)

🧠 Mnemonic

PUMP SeT

SNAP-PE-II: SNAP-II + Birth Weight + SGA + Apgar (5 min).
Utility: Considered the "Gold Standard" for physiology-based risk adjustment in mixed NICU populations.

5. Therapeutic & Workload Scores

NTISS (Neonatal Therapeutic Intervention Scoring System):
- Measures severity based on the intensity of therapy (e.g., ventilation mode, number of lines, medications).
- Main Use: Nursing manpower planning and resource allocation.
- Flaw: "Circular logic" – sicker babies get more treatment, but aggressive treatment can inflate the score independent of actual mortality risk (Treatment Bias).

6. Transport Scores

TRIPS (Transport Risk Index of Physiologic Stability):
- Assesses stability at the referring hospital and on admission to NICU.
- Components: Temperature, BP, Respiratory status, Response to noxious stimuli.
- Use: Audit of transport team quality.

7. Comparison: CRIB-II vs. SNAP-PE-II

Feature	CRIB-II	SNAP-PE-II
Target Population	Preterm (<31 weeks ) / VLBW only (<1500 grams)	All Neonates
Complexity	Simple (5 items)	Moderate (6 physiologic + 3 perinatal)
Data Window	Admission (0-1 hr)	First 12 hours
invasiveness	Requires ABG (Base Excess)	Requires ABG, BP, Urine output
Performance	Excellent for VLBW	Excellent for Term & Preterm

8. Application in LMIC (Low-Middle Income Countries)

Challenges:
- Lack of routine ABG/BP monitoring in peripheral centers makes SNAP/CRIB difficult.
- Differences in mortality drivers (Sepsis/Asphyxia vs. Prematurity alone).
Solutions:
- NMOS (Neonatal Mortality Outcome Score): Uses simplified parameters (Age, WT, RR, Cyanosis, Capillary refill).
- Simplified SNAP-II: Using SpO2 instead of PaO2/FiO2 ratio.

9. Conclusion

No single score is perfect. CRIB-II is preferred for VLBW benchmarking due to simplicity, while SNAP-PE-II provides a comprehensive physiologic assessment for the entire NICU population. For true quality improvement, units must participate in collaborative networks (e.g., Vermont Oxford Network) using standardized, risk-adjusted mortality scores.