More details on Hybrid:

Development:

Initially, a function was created which took the value of the mean of the NICU and Benaron-Benitz formulae for gas flow rates less than 200 mL/min, and took the value of only Benaron-Benitz otherwise. The function displayed a discontinuity at exactly 200 mL/min as NICU predicts a non-zero FiO2 for this flow rate. The function was smoothed by fitting a polynomial of second degree, the coefficients of which were derived using polynomial regression. A first degree (linear) polynomial did not approximate the function well at all, a third degree polynomial showed no significant deviation from the second degree polynomial and a fourth degree polynomial was too close of a fit to the discontinuity and hence displayed behaviour that would not be expected physically.

The quadratic fit did show unphysical behaviour in that it sometimes predicted a maximum FiO2 after which FiO2 decreases with increasing flow rate. This is in contrast to the predictions made by other models, which are all either monotonically increasing or monotonically increasing up to a point at which they become constant. A consequence of this decreasing portion was that a reverse calculator for the hybrid model would return two different gas flow rate values (before and after the maximum FiO2). To avoid ambiguity, the final hybrid calculator was made single-valued by taking the form of the quadratic fit up to the maximum FiO2 reached by Benaron-Benitz and taking that value for all greater flow rates. This maximum was chosen as it is closest to the hypopharyngeal measurments extracted from Gonzalez et al. (2019) and comes from a validated model.

If selecting flow rate for the preferred output, the formula is inverted to solve for flow rate.

Conditions:

The quadratic used is only fit using flow rate values up to 2000 mL/min (what is considered the low-flow limit), hence that is maximum value recommended for use. As Finer (NICU) is used to produce this quadratic it is best if a respiratory rate is known.

As made evident by the disclaimer on the calculator, this model has only been validated against one (small) data set (Gonzalez et al. (2019)), and has not been peer-reviewed, hence medical practitioners should always exercise caution if choosing to use the value given by this model. The peer-reviewed and validated Finer and Benaron-Benitz values are always displayed alongside the hybrid value for comparison.

Note: The Hybrid calculator uses the package regression-js to fit the polynomial under the MIT License.

This work is licensed under CC BY-NC-SA 4.0