Pulse Oximetry in Bronchiolitis: Have We Reached Saturation? Free

After its introduction into the market in the 1970s, pulse oximetry became a natural focus for clinicians and patients. After all, it is noninvasive and provides objective data on oxygenation that is easy to communicate to families. Its entry into the medical field coincided with impressive reductions in anesthesia-related fatalities,¹  inspiring its wide adoption in clinical care as a continuous data stream. Pulse oximeters can now be found in clinical settings—physician’s offices, emergency departments, hospital wards—as well as on the consumer market, from baby monitors to high-performance smartwatches. In an age of technology, we are inundated with data, and for the most part we love it. Our love for pulse oximetry endures, despite nontrivial limitations in accuracy: even Food and Drug Administration–approved devices have ∼4% margin of error,²  which is likely worse in patients with darker skin pigmentation.³  Continuous pulse oximetry also comes with the risks of alarm fatigue and increased workload for clinicians⁴  and patient immobility and skin injury.⁵  And although nurses report feeling overwhelmed by alarms,⁶  nurses are more likely than doctors to find reducing pulse oximetry monitoring to be unsafe.⁷  Compounded by problems with chronic nurse understaffing, bedside nurses may view physiologic monitoring as an aid to alert them to the status of their patient when they are not able to visualize them.

Fast forward from the 1970s to now, where pulse oximetry has emerged as the mainstay of bronchiolitis care, without the same regard for other markers of disease severity such as respiratory distress, heart rate, or ability to feed. This does not come without harm. With its now-ubiquitous availability, pulse oximetry use is associated with increased hospitalization rates and prolonged lengths of stay (LOS) without evidence that it averts acute clinical deterioration or improves morbidity in children hospitalized with bronchiolitis. Furthermore, although supplemental oxygen is generally viewed as harmless, growing research shows hyperoxia and oxygen toxicity can lead to worse outcomes, including mortality, in critically ill children,⁸  with anticipated results from Oxy-PICU,⁹  a multicentered randomized controlled trial of higher versus lower oxygen saturations, forthcoming. This should cause us all to reflect on our relationship with “pulse ox,” and with the idea that higher is always better when it comes to oxygen saturations.

In this issue of Hospital Pediatrics, Im et al¹⁰  capitalize on the different target oxygen saturation policies used at 6 hospitals in Canada to evaluate the association between the policy and LOS, and other outcomes such as initiation and discontinuation of oxygen, revisits to the hospital, or ICU transfers. The study was a secondary analysis of data collected from hospitals participating in a randomized controlled trial, wherein patients with bronchiolitis were randomized to intermittent or continuous pulse oximetry monitoring. Im et al included 2 hospitals (1 community, 1 children’s) with policies for target oxygen saturations of 90% at all times, and 4 hospitals (2 community, 2 children’s) with policies allowing oxygen saturation down to 88% when asleep and 90% while awake before administering supplemental oxygen. They hypothesized that the more permissive saturation goal allowing 88% when asleep would be associated with a decrease in LOS, presumably because of decreased time on oxygen. However, they found that there were no significant differences in LOS, nor their other clinical outcomes, based on the target oxygen saturation policy. They conclude that given the similar outcomes, clinicians and hospitals may prefer the lower target strategy because it may reduce sleep disruption and nursing workload associated with transient oxygen desaturation and monitor alarms (though these outcomes were not studied). Alternatively, they acknowledge hospitals may choose 90% awake and asleep for ease of implementation.

Strengths of the study include that it was performed in 6 different institutions within a similar health care system and included both community and children’s hospitals. Each institution already had existing policies and education or instruction in place surrounding the target oxygen saturations, mitigating effects related to dissemination and uptake of the policies. However, although the authors argue this was a pragmatic cohort study of real-world practice, the lack of data to evaluate adherence to the policies makes it difficult to know whether there was truly a difference in care received at bedside, and thus difficult to interpret the impact the policy has on the outcome. And although power calculations were not indicated as a substudy analysis, it is likely that there were not enough subjects to detect a difference in outcome based on such a small difference in pulse oximetry percentage; given the known margins of error in measurement, detecting that small of a difference in saturation percentage is also of questionable clinical significance.

Nevertheless, this inconclusive finding on the impact to LOS mirrors other recent studies that have failed to find a difference in LOS based on pulse oximetry monitoring status.^11–13 

It may be time to redefine the clinician’s relationship with pulse oximetry. First, we need to explore what a pulse oximetry number means to us now and how our behavior and approach to transient desaturations have evolved over time. Growing evidence has reaffirmed to us there is less value in focusing on the specific pulse oximetry number. Although 15 to 20 years ago, catching a desaturation overnight may have led to administration of supplemental oxygen and another 24 hours in the hospital, we now seemingly accept transient hypoxia as a natural part of the disease course, and prioritize other clinical markers of improvement to evaluate discharge readiness. Contributing factors may include secular trends because of our emphasis on high-value care, well-publicized efforts such as Choosing Wisely¹⁴  and the Eliminating Monitor Overuse studies,^7,15  research confirming desaturations occur commonly even when infants are not hospitalized,^16,17  understanding that clinical deterioration after initial improvement is rare,¹⁸  and a shift from tolerating 94% to 90% as acceptable saturation thresholds,^19,20  perhaps even creeping toward 88% while awake and asleep.²¹  If we are not aiming for saturations of 95% and greater, knowing the number itself (whether we catch it on continuous monitoring or not) does not seem to matter as much to us anymore, both in the moment of a desaturation, and affecting our decisions downstream.

And if knowing the number no longer means the same to us, has pulse oximetry been ousted from its position as the primary driver for length of stay in bronchiolitis hospitalizations?^22,23  If lowering target saturations or reducing continuous monitoring are not impacting length of stay, then what are the other drivers and other aspects of institutional culture that may now be more important? Our field has started to see albuterol use and viral testing creeping back into practice, and some say, “deep suctioning is the new pulse oximetry.” Whereas many are comfortable discharging a patient who experienced brief self-resolved desaturations overnight, there is greater variation in how nurses perform various levels of suctioning and how doctors incorporate it into their discharge decisions. This replacement of actions is not surprising; our drive to “do something” is so strong and a unique psychological barrier known in deimplementation work,²⁴  and it is normal to feel a sense of loss when taking away data we were previously accustomed to using in our assessment. Substitution of action or seeking alternative objective data are common either as an active deimplementation strategy or natural phenomenon to assure ourselves we are doing something. Perhaps we have successfully decoupled pulse oximetry and hypoxia from our assessment of a patient’s clinical status, and it is now time to turn our attention to the other behaviors impacting the quality and value of care for bronchiolitis. Previous single-center studies have suggested nursing-led oxygen weaning protocols,²⁵  specific suctioning practices,²⁶  and clear discharge criteria²⁷  as factors improving LOS, but evidence is sparse compared with that addressing pulse oximetry and other overutilization, and there is no published consensus.

Both continuous pulse oximetry and aiming for higher oxygen saturations undoubtedly contribute to overdiagnosis of hypoxia and overuse of resources. Continuous pulse oximetry still requires attention for deimplementation, as does the overutilization of heated high-flow nasal cannula and other low-value diagnostics and treatments. Furthermore, deimplementation research in these areas will provide us with a better understanding of strategies that can be used for reducing other low-value care. To continue improving the value of care for bronchiolitis, we need to understand what other drivers most affect LOS, identify best practices, and implement those on a larger scale.

But we may have reached our “saturation point” in pulse oximetry’s impact on LOS.