Development of a risk index for postoperative mortality or ICU admission in preterm and early term neonates undergoing surgery
Embargo Date
2022-06-04
OA Version
Citation
Abstract
OBJECTIVE: Despite recent scientific and technological advancements preterm birth remains one of the leading causes of neonatal mortality worldwide. With annual increases in the prevalence of preterm birth over the last decade further research is needed to discover the optimum management style for this complex population. The objectives of this study are to identify factors that may increase the risk of postoperative mortality or intensive care in preterm and early-term neonates undergoing surgery, and to create a predictive risk score to help clinicians and families better understand the probability of this postoperative outcome.
METHODS: A retrospective analysis was performed on 653 neonates born <39 weeks gestation undergoing surgery within the first 30 days of life at Boston Children’s Hospital between January 2013 and December 2018. Multivariable logistic regression analysis was used to determine independent risk factors for postoperative intensive care or mortality. Significant independent risk factors were multiplexed into a predictive risk algorithm with scores ranging from 0 to 13 points. Probabilities of intensive care or mortality were obtained with 95% confidence intervals for each score, and model validation via 1,000 bootstrap resamples was performed to assess model discrimination and calibration.
RESULTS: Among the total of 653 premature neonates undergoing surgery, 614 (94%) were deceased or required postoperative intensive care. Fifty-eight (9%) were deceased, 556 (85%) required postoperative intensive care and survived, and 39 (6%) survived without intensive care. Two hundred ninety-two neonates (45%) were female, and 356 neonates (55%) had an American Society for Anesthesiologists – Physical Status (ASA-PS) rating of IV/V. The following were determined independent predictors of postoperative mortality or intensive care: body weight < 2.5 kg (odds ratio(OR) = 0.16; 95% CI: 0.06, 0.41; P<0.001), ASA-PS IV/V (OR=14.65; 95% CI: 1.28, 168.2; P=0.031), cardiovascular condition (OR=4.44; 95% CI: 1.27, 15.55; P=0.020), neurological condition (OR=7.84; 95% CI: 1.89, 32.49; P=0.005), pulmonary condition (OR=16.11; 95% CI: 1.51, 32.06; P=0.002), and renal/urological condition (OR=6.96; 95% CI: 1.51, 32.06; P=0.013). The probability of postoperative mortality or intensive care ranged from 31.2% (95% CI: 18.6%, 47.4%) for a risk score of 0 points to 99.9% (95% CI: 99.9%, 99.9%) for a risk score of at least 9 points. The score demonstrated extremely good discriminatory performance (c-index=0.969; 95% CI: 0.950, 0.986).
DISCUSSION: Our risk score can estimate postoperative survival without intensive care for preterm (PT) and early-term (ET) neonates. Providers could use this index to guide discussions with families about possible postoperative outcomes. Our model can also be applied within healthcare facilities for a variety of purposes. This score can assist in departmental standardization for the perioperative management of a PT or ET neonate to create more regular treatment patterns. If applied correctly it could increase operating room throughput and allow for more efficient healthcare resource utilization. Finally, though many severity-of-illness and mortality risk scores exist in clinical practice presently, the proposed algorithm can be applied to a broader population with less temporal restriction.
CONCLUSION: The risk algorithm is a reliable and reproducible tool for the prediction of postoperative mortality or the likelihood of postoperative intensive care for PT and ET neonates meeting the inclusion criteria.