Patient-Centered Outcomes: A Way Forward When Evidence Is Lacking

High value, patient-centered care must consider the impact interventions have on patients and their families. The American Academy of Pediatrics Clinical Practice Guideline on the inpatient management of bronchiolitis focuses on “supportive care” such as oxygen therapy and hydration, without strong recommendation to the mechanism of administration (eg, high-flow nasal cannula [HFNC], low-flow nasal cannula [LFNC], nasogastric [NG], intravenous [IV]).¹  This focus on supportive care is because of lack of improvement in health outcomes in hospitalized children with bronchiolitis with interventions such as albuterol,²  nebulized hypertonic saline,^3,4  steroids,⁵  antibiotics,⁶  racemic epinephrine,⁷  nasal phenylephrine,⁸  and, most recently, HFNC.⁹  The guidelines even suggested, that “clinicians may choose not to” use continuous pulse oximetry for patients not requiring oxygen as new data emerge that we have also become overreliant on monitoring tools such as continuous pulse oximetry, which leads to unnecessary admission,¹⁰  prolongs hospital length of stay, and contributes to alarm fatigue within the hospital.^1,11  In contrast to more robustly studied interventions, nasal suctioning was left completely out of the American Academy of Pediatrics guidelines, despite its frequent use, because of a lack of high-quality data to support or disprove its use. Because bronchiolitis remains the leading cause of hospitalization of patients <2 years old at a direct cost of >$700 million annually,¹²  the challenge of how to treat this common condition and our inherently human “need to do something” is juxtaposed with what seems like an inadequate toolbox of low-quality, poorly studied interventions.

Especially in cases where limited or poor evidence exists (eg, nasal suctioning, pulse oximetry), patient reported outcomes (PROs) have the potential to provide a fuller picture of an interventions’ effect on patients and their families. Historically, a PRO is an outcome reported directly by the patient without the interpretation of health care personnel, though for patients too young or unable to communicate, family-reported outcomes serve a similar purpose.¹³  Examples include perceived harm or benefit of interventions, or quality-of-life issues such as the impact of an intervention on the disruption of breastfeeding or caregiver comfort.¹⁴  Though there is limited evidence to support patient- or family-reported outcomes in bronchiolitis specifically, routine measurement of PROs has been shown to improve patient–clinician communication.¹⁵  Furthermore, in a systematic review of PRO measurement in oncology, measuring PROs can increase monitoring of treatment responses, increase the detection of previously unrecognized problems, have a positive effect on patient satisfaction, and even lead to a modest to strong improvement in health outcomes.¹⁶ 

In this edition of Hospital Pediatrics, Lane and Foster¹⁷  examined the impact of levels of respiratory support and NG tube use in hospitalized children with bronchiolitis, across 13 domains using a previously validated questionnaire.¹⁸  Three out of 13 questions highlighted statistically significant differences in mean impact score across varying levels of support. These areas included caregiver holding, caregiver bonding, and breastfeeding disruption. Mean impact scores decreased between no support and low support (defined as LFNC), but steadily increased in moderate support (HFNC without NG tube), high support (HFNC and NG tube), and positive pressure (with or without NG tube) across the 3 domains. Furthermore, their use of open-ended questions for qualitative responses identified tubes and lines as the most often-cited barriers to holding, bonding, and breastfeeding. There were no patients who required LFNC and NG tube support, and the impact of IV fluids was not explored in this study. Though maybe counterintuitive, the authors posit that the decrease in mean impact score between no support and low support might be because of a placebo effect provided by some degree of intervention, but that at increasing levels of support, the interventions become more of a barrier to holding, bonding, and breastfeeding. This is supported by the identification of oxygen and NG tube supplementation as specific therapies they felt helped their child recover more quickly, in another open-ended question.

This study provides an important first step in evaluating the impact of respiratory support and NG tube supplementation on hospitalized infants with bronchiolitis. However, there are some important limitations. First, the groups that made up their levels of support were not determined a priori, but rather on the basis of balanced patient distribution across the groups. Therefore, as an example, their low support group was made up of patients with a max support of LFNC or NG tube use, whereas it may have been more helpful to attribute responses to a particular intervention by breaking these down further (eg, LFNC, LFNC with NG, LFNC without NG). Second, there was no attempt to consider the effect admission to the PICU, which is required at their institution for all patients who received higher than floor-standard HFNC flow rates and positive pressure ventilation, might have had on survey results and qualitative responses. This makes it hard to distinguish the effects of the respiratory support modality from PICU level practices such as more intense monitoring, more frequent vital sign gathering, and presence of nothing by mouth status, which are likely very disruptive to things like sleeping, bonding, and breastfeeding. Lastly, the timing of the survey (1 to 28 days from discharge) has the potential to affect some domains more than others, in addition to introducing a variable amount of recall bias. For example, a financial hardship might not be present immediately upon discharge, but may become apparent later when bills from the hospital start to arrive.

Despite its limitations, this study highlights the important notion that, even when we limit our care to supportive care interventions, we have the potential to create impact. Though it’s common for supporters of unproven therapies or more aggressive monitoring parameters to cite improved comfort,¹⁹  safety, or parental preference,²⁰  the degree to which these interventions might provide relief from symptoms and improved safety, compared with the consequences of intervening, must be considered to provide high-value, patient-centered care. One example of this phenomenon is the use of HFNC. Though HFNC is posited to reduce work of breathing in infants with bronchiolitis, and thus it could be argued that using it may improve comfort, it is unclear if the potential to improve comfort from work of breathing outweighs the comforting effects of a caregiver’s ability to hold, bond, and breastfeed. Few would argue utopia is an achievable goal within a hospital setting, but what if we created a management plan that prioritized patients and their caregivers? In otherwise stable, hospitalized patients with bronchiolitis, imagine the removal of cardiac monitors and continuous pulse oximetry to empower caregivers to hold, comfort, and breastfeed. Imagine the clustering of vitals, the administration of medications (eg, antipyretics), and nasal suctioning being timed just before an infant being ready to feed. What if we offered parents the choice between IV or NG hydration, when either are appropriate? Empowering caregivers to comfort their child would not only give them ownership of care, but could also serve as the impetus to deimplement care for which we don’t have clear evidence. This overly idealistic scenario could exist in parallel with robust clinical trials that answer clinical questions such as the utility of nasal suctioning, which oxygen delivery device (HFNC versus LFNC) or hydration mechanism (IV versus NG) is more appropriate for which patient, which patients warrant continuous pulse oximetry monitoring and cardiac monitoring and which patients don’t, and what oxygen parameters avert harm while not overtreating a number. Though it is clear we need more high-quality evidence from appropriately powered randomized controlled trials in pediatrics, the use of qualitative data and patient- or family-reported outcomes has the potential to fill knowledge gaps and guide treatment, especially where evidence is poor. It’s time to consider how our supportive care interventions affect patients and if they truly palliate patients, or if we’re palliating ourselves.