Pulse Oximeter Alarms: More Than We Bargained For
ABSTRACT & COMMENTARY
Synopsis: In this study from a pediatric ICU, a pulse oximeter's saturation alarm accounted for 43% of all alarms recorded. When it alarmed, the oximeter indicated an actual desaturation that was clinically relevant only 5% of the time.
Source: Tsien CL, et al. Crit Care Med 1997;25:614-619.
In this prospective observational study performed at Children's Hospital in Boston, a trained observer documented the sounding of monitor alarms during 298 daytime hours of continuous monitoring. For any given monitoring period, the observer concentrated on the alarms being tracked by an electronic display system at a single bedspace in the ICU and entered the observations directly into a computer data base. Using the patients' nurses for corroboration, the observer also noted whether each alarm was a true-positive and clinically relevant, a true-positive but clinically irrelevant (as when the systolic blood pressure became elevated transiently during suctioning), or a false-positive (when the alarm sounded in the absence of a change in the patient, as with manipulation of the ECG electrodes or oximeter probe).
In the 298 hours of observation, 2942 separate alarms were recorded, or an average of about 10 per hour at a given patient bedspace. Of all alarms recorded, the pat-ient's primary pulse oximeter alarmed most often; its saturation monitor accounted for 43%, and its heart rate monitor 20%. For the saturation monitor, the alarms were relevant truepositives 5% of the time, clinically irrelevant true-positives 4% of the time, and false-positives 91% of the time.
The investigators determined whether alarms occurred during patient interventions (alarms associated with procedures or other manipulation of the patient) or spontaneously in the absence of such intervention. Eighteen percent of all alarms were associated with patient interventions, 74% with no interventions, and 8% unknown. True-positive clinically significant alarms occurred more often in the absence of patient intervention (11%) than when the patient was being actively manipulated by a caregiver (2%).
COMMENT BY DAVID J. PIERSON, MD
Theoretically, the pulse oximeter represents a genuine breakthrough in ICU monitoring. Critically ill patients, who are at risk for a variety of potentially life-threatening events manifested by hypoxemia (and therefore oxyhemoglobin desaturation), can be monitored continuously and noninvasively so that such events can be detected and acted upon at the earliest possible moment. Unfortunately for both patients and caregivers, however, theoretical concept and practical reality turn out to be far apart.
In theory, we imagine desaturation to herald a serious physiologic impairment, such as produced by a pneumothorax or ventilator disconnection. While such events do occur, there is a good deal more moment-to-moment variation in arterial oxygena- tion than has been appreciated. Thorson and colleagues (Chest 1983;84:14) performed arterial blood gas analysis every 10 minutes for one hour in patients who were ill enough to be in the ICU and to have an arterial line but who were clinically stable and not being manipulated during the observation period. They found that the mean coefficient of variation for arterial PO2 was 5.1%, with a range in PO2 values during one hour of 1 to 45 mmHg (mean 16 mmHg). Hess and associates (J Clin Monit 1992;8:111) subsequently confirmed these data. Although these investigators found less variation in saturation than in PO2, most of their patients had higher baseline PO2 values than would ordinarily be maintained in critically ill patients, in which circumstances desaturations due to moment-to-moment variation would likely be much more common.
Thus, actual desaturation in the patient occurs often in the absence of the adverse events pulse oximetry is intended to detect. In addition, the oximeter is far from a perfect tool. Even when it is functioning perfectly, the 95% confidence limits for its indicated saturation are about ± 4%, so that a saturation of 92% might be indicated in the presence of a true saturation of 88%, and vice versa. Then, there are the sensors and the various connections in the system. In the study by Tsien et al, 64% of all false-positive alarms in the ICU were due to poor sensor contact, a faulty connection, and/or bad system format.
It should come as no surprise, then, that oximeter false alarms and true-but-clinically-irrelevant alarms accounted for 95% of all alarms sounded by the pulse oximeters saturation alarm. This finding is not new, and Lawless and colleagues (Crit Care Med 1994;22:981), previously reviewed in these pages, found very similar results in their pediatric ICU.
The challenge is to do something about it. As Tsien et al point out, some improvements may become available in the near future, such as using a median filter to remove transient changes in the function being monitored, performing multiple parameter detection, and making increased use of expert systems which use a specific knowledge base, rules, or belief system in determining when an alarm should sound. In the meantime, however, the clinician must contend with a cacophony of alarms-the great majority of them either false alarms or clinically irrelevant, as critically ill patients are monitored to detect potentially life-threatening adverse events.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content