Objective: To estimate the costs and decrease in mortality that could be achieved at Neonatal Intensive Care Units (NICUs) if the intake of at least 90% of extremely low birth weight (ELBW) infants was at least 98% human milk.
Study design: economic study by means of Markov Chain Monte Carlo (MCMC) simulation modelling.
Setting: hospital-based study.
Study population: simulated population (n = 24 149) sourced from 2012 United States vital statistics data. Extremely low birth weight infants born at 23 to 32 week’s gestation with weight between 400 and 1000 g, followed up from birth to 36 weeks’ postmenstrual age. Infants that died in the first 72 hours of life were excluded.
Risk factor assessment: exclusive human milk feeding during the period under study. The simulated cohort with an optimized feeding (OF) pattern consisted of 24 149 ELBW infants of who more than 90% received at least 98% of human milk (HM), compared to a simulated cohort of ELBW infants with suboptimal feeding (SOF) patterns: some were exclusively fed preterm formula (PF) and the rest a mixed diet (MD). The model did not consider infants fed with donor human milk.
Outcome assessment: the primary outcomes were the incidence of necrotising enterocolitis (NEC) Bell stage II or higher, and death. The actual feeding pattern of each ELBW infant during the period under study was estimated using the weekly percentages reported in a previous retrospective single-centre study (n = 285)1 and the multicentre Glutamine Trial (GT)2 dataset (n = 1433). The probability of developing NEC with the corresponding 95% confidence intervals (95 CIs) was estimated based on a subset of 848 infants through logistic regression modelling, obtaining adjusted odds ratios (aORs). Mortality due to NEC was estimated using rates from the Vermont Oxford Network (VON). The secondary outcome was reduction in costs based on Medicaid and Medicare data (increasing hospital hosts by 15% and adding the fees of neonatologists). The direct and indirect marginal costs were estimated adjusting for inflation and expressed in 2014 US dollars (US$).
Main results: the GT analysis found that 9% of infants had an OF pattern and the rest a SOF pattern (MD in 67.5% and PF in 22.4%). The incidence of NEC was 1.3% for OF, 11.1% for MD and 8.2% for PF (P < .002). The adjusted regression models showed an increased risk of NEC in infants fed a MD (aOR, 8.7; 95 CI, 8.7 to 65.2) or PF (aOR: 12.1; 95 CI, 1.5 to 94.2) compared to infants with an OF pattern. The study found no significant differences between infants with a MD and infants fed PF (aOR, 1.39; 95 CI, 0.83 to 2.33).
In the MCMC simulation, there was an annual excess of 928 cases of NEC (95 CI, 830 to 1036) and 121 deaths (95 CI, 108 to 134; 51% higher). The annual excess cost associated with SOF was 27.1 million US$ (95 CI, 24 to 30.4) in direct costs and 563 655 US$ (95 CI, 476 191 to 599 069) in indirect costs. The annual cost attributable to premature deaths due to NEC was 1.5 thousand million US$ (95 CI, 1.3 to 1.6).
Conclusion: in ELBW infants, exclusive human milk feeding is associated with a lower risk of NEC and death due to NEC, and lower medical costs.
Conflicts of interest: none noted.
Funding source: none noted.