Review Of: Touger M, Birnbaum A, Wang J, et al. Performance of the RAD-57 Pulse CO-Oximeter Compared with Standard Laboratory Carboxyhemoglobin Measurement. From the Department of Emergency Medicine, Jacobi Medical Center, Albert Einstein College of Medicine, Bronx, N.Y. Ann Emerg Med. 2010;56:382—388.
The Science
This is a cross-sectional cohort study (an observational study), which tries to validate the use of Masimo’s RAD-57 Pulse CO-Oximeter by comparing carbon monoxide (CO) readings obtained by the RAD-57 and carboxyhemoglobin lab values. The study was a convenience sample of 120 patients, including adult and pediatric patients presenting to a single emergency department (ED). The investigators of the study gathered bedside RAD-57 readings and then obtained a whole blood sample analyzing CO-oximetry. The primary measurement was agreement between the RAD-57 device and the blood gas levels. The authors concluded that the level of agreement observed suggests RAD-57 measurements may not be used interchangeably with standard laboratory measurement.
Medic Marshall: I’m not entirely sure how I feel about this study. I’m afraid people will look at it as saying the RAD-57 doesn’t really work, and a valuable tool for evaluating potential CO exposures for patients and firefighters in rehab will be discarded. I don’t believe this to be the case. Like any instrument, it’s an adjunct to making better clinical decisions– not the end-all-be-all answer. If you’re evaluating a patient who happens to be a firefighter, and they’re exhibiting signs and symptoms of CO exposure but the RAD-57 device is reading in a normal range, you aren’t (or at least you shouldn’t) allow that firefighter to return to duty.
When I dug a little deeper and looked at the study’s limitations, I saw that the authors noted approximately three quarters of the subjects were of Hispanic or African descent. The RAD-57 device has already been shown to have poor performance when measuring carboxyhemoglobin in dark-pigmented individuals. A significant potential for bias (at least in my humble opinion) exists in this study. Take it for what it’s worth; I personally don’t give it a whole lot of merit.
Doc Wesley comments: To echo Marshall’s comments, I too, won’t alter my opinion of the value of the RAD-57 to detect CO. Additional issues are present with this study beyond the known inaccuracies of the device with dark skin. Examination of the data reveals that a significant number of RAD-57 readings were 0%. These results should have been excluded as a technical error because it’s unlikely to have a CO level of 0%, especially in a suburban/urban environment. To illustrate this, I took a RAD-57 and walked through my ambulance company going through administration, billing, operations and fleet. I performed 30 consecutive readings and never got a reading of 0%. Including these readings in the study creates an incorrect error rate.
Another issue to consider: What was the “gold standard” to which the RAD-57 readings were compared? In this study, a venous sample was sent for evaluation on a blood gas machine. Although this is acceptable practice to obtain CO levels, the question is how many different blood gas machines were used, and how were they calibrated? Literature exists that indicates results between machines can vary as much as +/- 5%.
Bottom line? I won’t be changing my advocacy for the use of the RAD-57 to detect CO exposure, and I hope you won’t either.
Abstract
Touger M, Birnbaum A, Wang J, et al. Performance of the RAD-57 Pulse Co-Oximeter Compared with Standard Laboratory Carboxyhemoglobin Measurement. From the Department of Emergency Medicine, Jacobi Medical Center, Albert Einstein College of Medicine, Bronx, N.Y.
Study objective: We assess agreement between carboxyhemoglobin levels measured by the Rad-57 signal extraction pulse CO-oximeter (RAD), a Food and Drug Administration—approved device for noninvasive bedside measurement, and standard laboratory arterial or venous measurement in a sample of emergency department (ED) patients with suspected carbon monoxide poisoning.
Methods: The study was a cross-sectional cohort design using a convenience sample of adult and pediatric ED patients in a Level I trauma, burn, and hyperbaric oxygen referral center. Measurement of RAD carboxyhemoglobin was performed simultaneously with blood sampling for laboratory determination of carboxyhemoglobin level. The
difference between the measures for each patient was calculated as laboratory carboxyhemoglobin minus carboxyhemoglobin from the carbon monoxide oximeter. The limits of agreement from a Bland-Altman analysis are calculated as the mean of the differences between methods +/-1.96 SDs above and below the mean.
Results: Median laboratory percentage carboxyhemoglobin level was 2.3% (interquartile range 1 to 8.5; range 0% to 38%). The mean difference between laboratory carboxyhemoglobin values and RAD values was 1.4% carboxyhemoglobin (95% confidence interval [CI] 0.2% to 2.6%). The limits of agreement of differences of measurement made with the 2 devices were -11.6% and 14.4% carboxyhemoglobin. This range exceeded the value of +/-5% carboxyhemoglobin defined a priori as clinically acceptable. RAD correctly identified 11 of 23 patients with laboratory values greater than 15% carboxyhemoglobin (sensitivity 48%; 95% CI 27% to 69%). There was one case of a laboratory carboxyhemoglobin level less than 15%, in which the RAD device gave a
result greater than 15% (specificity of RAD 96/97=99%; 95% CI 94% to 100%).
Conclusion: In the range of carboxyhemoglobin values measured in this sample, the level of agreement observed suggests RAD measurement may not be used interchangeably with standard laboratory measurement.
