When (If) to Worry? Evaluating the Accuracy of Home-Grade Pulse Oximeter Devices

Respiratory complaints, and concern for the presence and/or development of hypoxemia, compose one of the most common reasons for pediatric patients to present for health care provider evaluation, and disposition decisions for these providers are strongly influenced by S_pO2 measured by pulse oximetry.^1,2 Whereas in the acute care setting S_pO2 monitoring with medical-grade devices is standard practice, pulse oximetry monitoring in the home setting is increasingly common either for children with chronic cardiorespiratory disease or by individuals who are simply curious about their oxygen status. There is wide variability in provider decisions to prescribe home pulse oximetry.³ However, feasibility and efficacy of this practice have been described in several adult and pediatric conditions including obstructive sleep apnea,⁴ acute COVID-19,⁵ bronchopulmonary dysplasia,⁶ and congenital cardiac disease.^7,8 Whereas monitoring in these scenarios is typically performed using medical-grade devices, patients and/or their parents are increasingly able to obtain non–medical-grade (home grade) devices designed for out-of-hospital use, which can be inexpensively purchased in person or online. Whereas medical-grade pulse oximeters are approved by regulatory bodies such as the United States FDA and must be accurate within 3% of a measured oxyhemoglobin saturation, these home-grade pulse oximeters are not subject to the same approval rigor.⁹ Because of the ease with which caregivers or patients can purchase these devices, it behooves the medical community to understand their accuracy and limitations.^9,10 Whereas some studies¹¹ have been performed to assess the accuracy of home-grade devices in adults, pediatric data are scarce and limited by either small sample sizes or lack of enrolled subjects with overt hypoxemia.^12,13 Consequently, the pediatric medical community continues to lack data upon which to adequately educate their patients on appropriate use of home-grade pulse oximeters.

In this issue of Respiratory Care, Kovesi et al¹⁴ attempted to address this deficiency by comparing the accuracy of 3 different home oximeters to reference, hospital-grade oximeters in a cohort of pediatric subjects. Using a prospective, cross-sectional study, the investigators evaluated one adult and one pediatric fingertip oximeter, as well as an oximetry app built into a Samsung smartphone in 74 subjects enrolled primarily from pulmonology and cardiology out-patient clinics. Paired measurements were obtained from the same digit with each device in random order. The fingertip devices were used in subjects based on subject weight (adult in subjects ≥ 30 kg, n = 20; and pediatric in subjects < 30 kg, n = 54). The smartphone app was attempted in all subjects. Outcomes of interest included (1) the ability to obtain a reading, (2) the time required to obtain a reading, and (3) the accuracy of the reading as assessed using Bland-Altman comparison with a hospital-grade oximeter.

The investigators were able to obtain measurements in all subjects tested with the adult fingertip oximeter, while measurements were unable to be obtained in 7.7% of attempts with the pediatric fingertip oximeter and a startling 38.8% of attempts with the smartphone-based app. The smartphone failure rate was significantly lower than that for the fingertip device in both weight cohorts. This compared to a 1.2% failure rate with the hospital-grade oximeter. A modest negative correlation with age was found for time to successful reading with all devices, with no difference between home-grade and hospital-grade devices found. Most importantly, the pediatric fingertip oximeter tended to underestimate S_pO2 in the younger population (mean difference −4.5% compared to the reference device) and demonstrated significant variability with a range in discrepancy from readings > 8% to < 33% the reference oximeter. The degree of variability also appeared to be wider at S_pO2 values < 90%. The mean difference and degree of variability improved in a post hoc analysis of this device in children > 1 y of age. Conversely, the adult-based fingertip device did not differ significantly from the reference device (mean difference 0.7%) although no measurements in this cohort occurred in subjects with hypoxemic S_pO2 values. Lastly, with the smartphone-based oximeter, S_pO2 was routinely underestimated at S_pO2 values < 94% but routinely overestimated at S_pO2 values > 94%; and, as with the pediatric fingertip device, substantial variability between the two devices was noted, especially at lower S_pO2 values and in younger subjects. The authors concluded that the pediatric fingertip and smartphone oximeters evaluated were “not well suited for infants” whereas both fingertip devices performed “reasonably well” in larger children and those who were not overtly hypoxemic.

The study adds to our knowledge regarding home-grade oximeters in valuable ways. It is the largest study published to date evaluating a pediatric-specific fingertip device. The finding that this device demonstrated widely variable discrepancies with medical-grade devices, and was especially inaccurate in the infant population, is new and important considering that infants, with or without underlying pulmonary disease, are at highest risk for adverse outcomes from respiratory infections. Consequently, they may be the population that parents are particularly motivated to monitor, especially if viral symptoms develop. However, regardless of age, detection of an S_pO2 value > 90% is likely to be considered reassuring. Knowing that it may take up to 3 min, regardless of device used, to obtain a stable reading is also helpful when providers counsel parents on appropriately using these devices, should they insist upon doing so. Similarly, understanding that the smartphone-based oximeter was highly unreliable and inaccurate in smaller children, especially infants, should instruct health care providers to steer parents away from these devices, even though they are likely to already be present in many households.

However, the study has several important limitations, many of which are acknowledged by the authors. Inadvertently and unintentionally, multiple hospital-grade devices were used as the reference device. Whereas this potentially adds additional variability, this is unlikely to have accounted for a significant number of the discrepancies between hospital- and home-grade devices observed given accuracy requirements for device approval from the FDA.⁹ An unavoidable limitation is that only 3 of thousands of available home-grade devices were tested, and it is possible that the authors randomly evaluated one of the poorest performing pediatric-based devices. This limits the generalizability of the findings to other devices while emphasizing the logistical difficulties of medical science to keep pace with consumer developments.

Despite being one of the larger pediatric studies evaluating home-grade oximeters, the sample size is still quite small, especially when one considers the age/subject size differences noted. Related to this, since an important feature of home-based oximeters should be to accurately detect hypoxemia, it is significant that < 40% of the readings were obtained in subjects with hypoxemia-level S_pO2 values. Whereas it would be more logistically challenging to perform, a more accurate evaluation of the devices would have been to perform the study in hospitalized subjects with arterial lines in whom noninvasive S_pO2 values could be compared to actual measured arterial oxyhemoglobin saturation values.

Despite these limitations, the authors should be applauded for taking a first step into better understanding the accuracies and limitations of commercially available oximeters designed for home use. As has been discussed, use of these devices in the home will, undoubtably, increase in coming years. The medical community must attempt to better understand appropriate interpretation of results, whether they be based on a telephone call between patient/parent and their on-call provider or part of the history provided when a patient presents to a clinic or emergency department. Based on the results of this study, providers should be advised to inquire about what type of device was used (in particular a smartphone-based monitor if the patient is younger) and the duration of monitoring. Furthermore, as should always be the case, counseling parents on their child’s medical status should never rely on a single data point, and clarifying questions designed to put the S_pO2 value into the context of the patient’s other symptoms remains a necessity. It is our hope, as also stated by the authors, that this study stimulates others to perform similar, larger-scale evaluations of home-grade devices, especially in both younger and hypoxemic pediatric subjects. As the market becomes increasingly saturated with these devices, knowing when, or even if, we can reliably detect the presence of desaturation bears understanding.