By: Lydia Furman, MD, Assistant Editor
Discharging preterm infants is an arduous duty. Both residents and supervising neonatologists are familiar with the many pitfalls that hold up the show. Coordination of care, services and appointments, and needed equipment, are massive tasks. And there is often enough angst about readiness- i.e. “will this baby ‘fly’ and “will the parents be able to meet the baby’s needs” – to put off the discharge date several days for non-medical reasons. All that is in addition to the strong desire of parents to finally “escape” home with their baby. Any prolongation of the hospital stay is very expensive.
Dr. Temple et al. (doi: 10.1542/peds.2015-0456) have written a highly pragmatic article that gives providers a new “crystal ball” algorithm with which to plan discharges. Using daily progress note information, their work teaches us how to predict discharge within a 2-10 day period, giving providers and staff the information and a level of certitude with which to plan. They emphasize that their study is not about predicting length of stay at admission, but it’s about using “real time” data to predict future discharge during the hospital stay.
What parameters are most useful? It’s an interesting exercise to try to predict or guess ahead what information will be most useful. Will it be lab values, growth parameters, feeding information, cessation of “A’s and B’s” (apneas and bradycardias), vital signs, original birthweight or gestational age, number of medications, or some golden combination of these?
The authors evaluated a total of 4,693 patients and 103,206 patient-days, and examined four subpopulations, including premature infants, babies with cardiac disease, babies with gastrointestinal surgery, and those with neurosurgical conditions. They used progress notes to identify qualitative and quantitative parameters, and two types of “derived” or calculated data. The retrospective data they used is clinical and intuitive, and highly available, and will likely appeal to neonatologists and trainees. Ultimately with the use of just two features (no spoiler here- please read to find out!), days to discharge of 4 days can be predicted with surprising accuracy for three of the four subpopulations (neurosurgical patients were a challenge for the algorithm). This excellent work needs prospective confirmation, but the results are highly encouraging.
Clearly the most important thing is getting babies and parents home in a way that is comfortable and safe, but there is a huge carrot at the end for making this transition as timely as possible. A brief peek at some crude financial data suggests that the work of Temple and colleagues has the potential to create enormous societal savings. The average cost of a preterm birth in the US is $32,325, and for infants born at less than 28 weeks, the average cost of the hospital stay was $280,811 (March of Dimes Peristats, https://www.marchofdimes.org/peristats). Single day charges for the NICU range around $3,000, not including costs related to specific surgical procedures or imaging (http://www.managedcaremag.com/archives/1001/1001.preterm.html).
Thus any comprehensive incremental decreases in length of NICU hospital stay could have a profound impact on total health care dollars. Kudos to the authors for their forward thinking work, since ultimately safe healthcare change must be driven and led by knowledgeable physicians, rather than by administrators or insurance companies alone.