Non-invasive blood gas interpretation
by Lawrence Martin, M.D.

Order book from or Lippincott W&W

7. Pulse oximetry instead of PaO2 and SaO2

Summary: The pulse oximeter measures SpO2, which in most situations closely correlates with SaO2 as measured by the co-oximeter. However, in several situations the pulse oximeter can be dangerously misleading, and should not be used without blood gas confirmation.


Pulse oximetry has been heralded as "arguably the most significant technological advance ever made in the monitoring of the well being and safety of patients during anaesthesia, recovery and critical care" (Severinghaus 1986) and as "the greatest advance in patient monitoring since electrocardiography" (Hanning 1995). Clearly, this is a device with which all care givers should be familiar. So much has been written about pulse oximetry that reviews of the subject appear frequently (Severinghaus 1992; JAMA 1993; Wahr 1995).

Despite its acknowledged importance and simplicity of use, the pulse oximeter is often mis-used and its measurement misunderstood. The following points bear emphasis for anyone using a pulse oximeter

  • Pulse oximetry does not differentiate carboxyhemoglobin from oxyhemoglobin (Barker 1987; Raemer 1989). Pulse oximetry emits two wavelengths of light, 660 nm and 940 nm. Oxygenated (HbO2) and deoxygenated hemoglobin (Hb) reflect these two wavelengths differently, allowing the oximeter to distinguish between them. Light transmission at 660 nm is mainly from HbO2, and at 940 nm is mainly from Hb. It turns out that COHb reflects just as much 660 nm wavelength light as HbO2, and thus COHb is read as HbO2 by the pulse oximeter. Thus, for example, a patient with a true SaO2 of 85%, plus 10% COHb, will have a pulse oximetry SpO2 reading of about 95%. For this reason pulse oximeters should never be used to assess oxygenation in anyone who might have CO poisoning.

  • Pulse oximetry does not reliably distinguish between oxygen desaturation from a low PaO2 and from excess methemoglobin (metHb). Unlike carbon monoxide, metHb does depress the SpO2 reading, but not linearly (Eisenkraft 1988; Barker 1989; Watcha 1989; Ralston 1991). MetHb decreases SpO2, but the fall in SpO2 is only by about one-half of the metHb concentration, until a reading of 85% is reached; at this point, further increases in %metHb do not lower the SpO2 any further. Thus a pulse oximetry reading of 90% could represent:

    a) a low PaO2 causing oxygen desaturation of 10% (i.e., a true SaO2 of 90%); or
    b) normal PaO2 with methemoglobin in excess of 10%; or
    c) some combination of a low PaO2 and excess metHb

  • As with CO, pulse oximeters should never be used to assess oxygenation in anyone who might have excess methemoglobin. Methylene blue is used to treat severe cases of excess methemoglobin; like many intravenous pigments, methylene blue causes a major drop in SpO2, and is another reason to avoid the pulse oximeter altogether when managing this problem (Wahr 1995).

  • Clinically acceptable precision for SpO2 is within 3% of the SaO2, but the degree of precision varies among oximeter models (Leasa 1992). Numerous studies have appeared correlating the precision or accuracy of different oximeters, and the results are variable. Knowing whether a particular model at a particular time is over- or under-estimating oxygen saturation would require measuring SaO2 (in a co-oximeter) at the time SpO2 is also measured. This is obviously not practical nor desirable. Instead, it seems prudent to assume that SpO2 is over-estimating SaO2, and to take some action whenever SpO2 falls below 93%. Such action would depend on the clinical situation, of course (e.g., close monitoring of vital signs and cardiac rhythm; adding or increasing supplemental O2; beginning another treatment; checking an arterial blood gas).

  • Pulse oximetry may give a false sense of security if the patient has adequate oxygen saturation but a declining PaO2. Because of the relatively flat portion of the O2 dissociation curve above a PaO2 of 60 mm Hg, and especially above 100 mm Hg, PaO2 could drop significantly without an appreciable change in SpO2 (Figure 6-5).

  • Pulse oximetry may give a false sense of security if the patient has adequate oxygen saturation but a rising PaCO2. A sedated or anesthetized patient receiving supplemental oxygen can maintain adequate SaO2 without adequate ventilation. In the most extreme cases this is called apneic oxygenation: diffusion of a high FIO2 into the lungs maintains oxygenation, while the PaCO2 rises to life-threatening levels (Davidson 1993; Hutton 1993, Ayas 1998). Even with extreme acidosis SaO2 may stay in the normal range, especially if PaO2 is maintained above normal (Figure 6-6). (Oxygenation by diffusion is also used in the "apnea test" for brain death. Patients without spontaneous breathing, in whom neuro-logic exam points to brain death, are oxygenated by diffusion though an endotracheal tube without any mechanical breathing. Total absence of respirations when PaCO2 reaches 60 mm Hg or higher, without confounding factors such as hypothermia or drug overdose, confirms brain death.)

  • Pulse oximetry may be unreliable if there is poor tissue perfusion, vasoconstriction or hypothermia. This problem is most often seen in patients with decreased vascular flow to the extremities. The machines only work if there is a strong pulse. With a weak pulse, the SpO2 reading may "stick" on a falsely low or high value.

  • Pulse oximetry can be mis-used by people unfamiliar with how it works and what is measured. A 1994 study showed that doctors and nurses were surprisingly ignorant about some basic oximetry principles, and made serious errors in interpretation of readings (Stoneham 1994). For example, 30% of doctors and 93% of nurses thought the oximeter measured PaO2 or oxygen content. In the same study only one doctor and one nurse (3% of each sample) knew that an oximeter requires pulsatile flow of blood under the sensor.

Of course the pulse oximeter is so simple to use that anyone can record an SpO2, whereas only trained laboratory personnel can run a blood sample through a co-oximeter. This ease of use invites mis-application. I have seen health care personnel at all levels -- physicians, nurses, therapists -- unintentionally chart a false reading because they were unfamiliar with the device and the pitfalls mentioned above. This problem is compounded when a charted value is taken as reliable by other health care workers, when in fact it may be totally erroneous.

A patient in the ER has a blood gas drawn at the same time as a pulse oximetry measurement. The arterial PaO2 is 77 mm Hg (breathing room air). Below are three values for arterial oxygen saturation, with information about how each was determined.

  • 95%, calculated from PaO2 value
  • 98%, from pulse oximeter
  • 85%, from co-oximeter

    Why do they disagree and which value is the more reliable?

This question illustrates the three ways oxygen saturation is usually obtained. By far the most reliable method is direct measurement of an arterial sample in the co-oximeter. Using four wavelengths of light, the co-oximeter makes direct measurements of oxyhemoglobin, carboxyhemoglobin and methemoglobin, and so gives the only true measurement of SaO2. Pulse oximetry, on the other hand, includes carboxyhemoglobin and a variable amount of any excess met-hemoglobin in its reading of oxygen saturation, and so will overestimate SaO2 in the presence of excess dyshemoglobins.

As for the calculated SaO2, it is only reliable if nothing else but oxygen is binding to hemoglobin, which you can't know from just the PaO2. In this example the patient actually had 10% carboxy-hemoglobin and 2% methemoglobin, so the calculated SaO2 of 95% was misleading. In summary, it is always important to know where your oxygen saturation value is coming from, and to interpret it accordingly.

Return to Introduction of "Non-invasive blood gas interpretation"
Return to Table of Contents for All You Really Need to Know to Interpret Arterial Blood Gases