Related Experiment Video
Updated: Sep 23, 2026
Assessment and Evaluation of the High Risk Neonate: The NICU Network Neurobehavioral Scale
Published on: August 25, 2014
Daily cost prediction model in neonatal intensive care
John A F Zupancic1, Douglas K Richardson, Bernie J O'Brien
1Department of Neonatology, Beth Israel Deaconess Medical Center, and Harvard Medical School, Boston, MA 02458, USA. zupancic@post.harvard.edu
Insights
Identifying key resource items can predict daily ancillary costs for extremely low birth weight infants, reducing the burden of economic evaluations in neonatal trials.
Area of Science:
- Neonatal intensive care
- Health economics
- Clinical trial methodology
Background:
- Economic evaluations alongside neonatal randomized controlled trials are challenged by the high cost of detailed patient resource data collection.
- Accurate cost prediction is crucial for resource allocation and trial feasibility.
Purpose of the Study:
- To identify key resource items that accurately predict daily ancillary costs for extremely low birth weight infants.
- To reduce the data collection burden in neonatal clinical trials.
Main Methods:
- Utilized stepwise linear regression on data from 385 infants in tertiary neonatal intensive care units.
- Developed models correlating total daily ancillary costs with 89 non-personnel resource items across five admission periods.
- Validated models using a randomly split half of the patient day data.
Main Results:
- Regression models explained 76% to 94% of daily ancillary cost variability across different admission periods.
- Key predictors included surfactant, chest radiographs, red blood cell transfusions, and cranial ultrasounds.
- Models demonstrated excellent predictive power on validation data.
Conclusions:
- Daily non-personnel costs for extremely low birth weight infants are influenced by a limited set of key resource variables.
- Predicting total ancillary costs with minimal data collection can facilitate economic evaluations in neonatal trials.
Objectives:
One barrier to economic evaluation alongside neonatal randomized controlled trials is the expense of collecting detailed patient resource information. To reduce this data collection burden, we identified the key resource items that predict daily ancillary costs for extremely low birth weight infants.
Methods:
Participants were 385 infants enrolled in the Trial of Indomethacin Prophylaxis for Preterms in nine tertiary level neonatal intensive care units in Canada. Information on eighty-nine nonpersonnel resource items was abstracted from the hospital chart from admission to tertiary hospital discharge. Unit costs were derived from a provincially standardized cost accounting system. Using stepwise linear regression, models correlating total daily ancillary costs with key resource items were constructed for each of five periods of admission. Models were derived in a randomly split half of the total sample of patient days and validated against the remainder.
Results:
The 385 infants contributed resource information from 23,354 admission days. The regression model for weeks one to twelve included the covariates surfactant, chest radiograph, red blood cell transfusion, cranial ultrasound, abdominal radiograph, parenteral amino acid infusion, surgery, platelet transfusion, and echocardiogram and explained 91% of the variability in daily nonpersonnel costs (P<.0001). Models for other admission periods similarly included between four and eight covariates, were highly significant (P<.0001) and explained between 76% and 94% of daily ancillary cost variability. The regression equations showed excellent predictive power when applied to the second half of the patient data set.
Conclusions:
Daily nonpersonnel costs for extremely low birth weight infants are driven by a limited number of key resource variables. The ability to predict total ancillary costs with minimal data collection will facilitate inclusion of economic evaluations in neonatal trials.