Application of the ROX-HR Index in Predicting High-Flow Nasal Cannula Failure in Pediatrics

The use of heated, humidified, high-flow nasal cannula (HFNC) has grown substantially in pediatric practice over the past decade.¹ Whereas historically used for infants with critical viral bronchiolitis, its use has expanded to a wide range of pediatric age groups, settings, and diagnostic conditions.^2,3 Whereas there are many proposed mechanisms to its therapeutic effectiveness, the widespread increase in use presents many challenges to providers in the pediatric ICU (PICU). These challenges include patient selection and device management strategies. Do all patients with respiratory distress benefit from HFNC therapy? Do only certain diagnostic populations or age groups? When is the optimal time to start HFNC therapy, and how do you objectively assess its efficacy? In this issue of Respiratory Care, Webb et al explore the final proposed question regarding how to objectively measure if treatment with HFNC is improving a patient’s clinical condition and how to predict which patients are likely to fail HFNC treatment.⁴

Current practice in the PICU regarding HFNC use varies widely by institution and provider.⁵ Ongoing assessment regarding the effectiveness of HFNC in a particular patient involves individual providers performing clinical assessments and using subjective metrics to determine which patients are improving and which are likely to need more invasive forms of respiratory support. In this novel study, Webb et al⁴ applied 2 objective, standardized scoring systems, the ratio of oxygen saturation (ROX) and ROX-heart rate (ROX-HR) indices, previously validated in adult patients, to predict which pediatric subjects would fail HFNC therapy. HFNC treatment failure was defined as the need for noninvasive ventilation or intubation and invasive ventilation. Their study population focused on children < 24 months old, with no restrictions on diagnosis or timing of HFNC. The study analyzed the ROX and ROX-HR indices at several time points in HFNC therapy and measured the predictive accuracy at several future time points. They concluded that the ROX-HR index was a useful tool in predicting HFNC failure in pediatric subjects < 24 months, specifically at 1 and 6 h post initiation.

The ROX-HR index represents an objective tool that is validated in adult patients on HFNC and is simple to calculate. It can be done at the bedside using readily available clinical data (heart rate, breathing frequency, S_pO2, and F_IO2). Webb et al⁴ selected the largest population of PICU HFNC users, those < 24 months, and did not limit to a specific diagnosis. This approach adds to the potential generalizability of this simple index in the PICU population. Great attention was taken to the specified index cutoff values to use. A high sensitivity is important as the primary goal of the index is to identify those most likely to fail HFNC therapy. Specificity is also important to avoid overidentification and lack of provider trust in the index. The authors selected a cutoff value of < 3 for the ROX-HR at both the 1- and 6-h time points. This provided a sensitivity of 90% at 1 h and 71% at 6 h and a specificity of 52% at 1 h and 92% at 6 h. The high sensitivity of the index at 1 h is valuable to alert clinicians to those patients at highest risk for HFNC failure. The low specificity, however, indicates that many patients who will ultimately not fail HFNC therapy will also have low ROX-HR indices at the 1-h time mark. At the 6-h mark, the ROX-HR index has lower sensitivity, meaning that a higher percentage of patients will fail HFNC therapy without obtaining a ROX-HR score < 3. The higher specificity at the 6-h mark does eliminate many “false alarms” compared to the 1-h mark. A potential application framework a clinician could employ based on this study would be to identify those at high risk of failure in the first hour of HFNC initiation using a ROX-HR index < 3. Then if the child remained on HFNC at the 6-h mark use that subsequent ROX-HR index value to help determine either the need for intubation if still < 3 or reassurance of continuing with HFNC treatment if > 3. Overall, whereas not a perfect marker of HFNC failure, the ROX-HR does have a high sensitivity to identify those patients likely to fail HFNC therapy, with attention to not cause “alarm fatigue” by overidentifying patients who will ultimately not fail, particularly at the 6-h mark.

There are several challenges to applying adult metrics and indices to pediatric populations. One of the largest lies in the high variability of pediatric vital signs by patient age relative to the more homogenous adult population. Two of the most varied are included in the ROX-HR index, breathing frequency and heart rate. It is a reasonable question to ask if this adult index will be accurate in a population with inherent differences in the input data due to these physiologic factors. The study authors attempted to address this by creating their own cutoff values using pediatric data from their study rather than making any attempt to incorporate previously defined adult values. One of the interesting findings from their results was that after age stratifying their population into < 6 months, 6–12 months, and 12–24 months the ROX-HR index was found to be a statistically significant variable in a multivariate model to predict HFNC failure only in the < 6 months age group. This age group has the highest degree of physiologic variability in both heart rate and breathing frequency. It is perhaps for this reason that the ROX-HR index can better predict HFNC failure, as the index will exhibit more changes in these patients due to the larger changes in the input variables. It could also be that the ROX-HR index was statistically significant in this age group due to the larger number of HFNC failures in those < 6 months, both in this cohort and in previous studies.⁶ It is possible that the ROX-HR index is only predictive of HFNC failure in the youngest pediatric ages as seen in this study, but it is also possible that in future studies with larger sample sizes the ROX-HR index will be found statistically significant in predicting HFNC failure in the older age groups as well.

Webb et al⁴ chose to include all patients in the specified age group who were treated with HFNC, regardless of diagnosis or other interventions. Whereas this will improve the potential generalizability of the ROX-HR index in pediatrics, it does add some potential bias to the application. One specific area surrounds the use of HFNC as a postextubation treatment modality. HFNC is frequently used as a postextubation respiratory modality in the PICU,⁷ and 32% of this study’s cohort included those using HFNC postextubation. Whereas potentially effective in preventing the need for re-intubation, it is possible that use in this setting could be fundamentally different from using HFNC for respiratory distress prior to intubation, as the potential mechanisms for HFNC failure may be different. Further studies stratifying subjects into pre- and post-intubation HFNC use may be helpful in determining the usefulness of the ROX-HR index in each setting.

The use of HFNC is increasing in the pediatric intensive care population. An accurate, easily obtained, objective index to identify patients at high risk for HFNC failure is needed to prevent both unnecessary intubation as well as late intubation. In this study, Webb et al⁴ applied the ROX and ROX-HR indices to the pediatric population and found that the ROX-HR index can serve as a useful predictor of HFNC failure in children < 24 months. This study provides a meaningful initial step in using a validated adult index to improve PICU care for this patient population. Further research is needed to validate this index with data from other institutions as well as in specific subpopulations, including specific age groups and those treated with HFNC prior to intubation and following extubation.

• Copyright © 2022 by Daedalus Enterprises