When Using Pulse Oximetry Which Co-Existing Factors May Affect the Accuracy of the Readings: A Mechanistic Guide to Confounders, Limits, and Bedside Verification

When using pulse oximetry, co-existing factors that may affect accuracy include poor peripheral perfusion, motion and low signal amplitude, venous congestion, dyshemoglobinemias such as carboxyhemoglobin and methemoglobin, dark skin pigmentation, nail cosmetics and artificial nails, ambient light and electromagnetic interference, and improper sensor placement or device conditions. Each factor distorts the reading through a distinct mechanism, and several can produce dangerous errors in opposite directions, so a normal-looking SpO2 is never proof of adequate oxygenation on its own.

Pulse oximetry is one of the most consequential numbers in clinical care, and also one of the most misunderstood. The device does not measure oxygen content, oxygen delivery, or ventilation. It estimates the percentage of hemoglobin binding sites occupied by oxygen, using the different light-absorption profiles of oxygenated and deoxygenated hemoglobin across two wavelengths. Nearly every factor that degrades accuracy does so by corrupting either the light signal, the assumption that only two hemoglobin species matter, or the algorithm's assumption of a stable arterial pulse.

This explainer goes one level deeper than the typical overview. It walks through the physical mechanism, quantifies error where reliable numbers exist, separates myths from real confounders, and ends with practical verification steps. Throughout, the governing rule is simple: treat the oximeter as a screening instrument whose output must be reconciled with the patient's physiology, not as a laboratory measurement.

Mechanism: how two wavelengths become a percentage

A conventional transmission oximeter shines red light (about 660 nm) and infrared light (about 940 nm) through a pulsatile vascular bed, usually a fingertip. Oxygenated hemoglobin absorbs more infrared and less red; deoxygenated hemoglobin does the reverse. The device measures the pulsatile (AC) and steady (DC) components at each wavelength, computes a ratio-of-ratios, and maps that ratio to an SpO2 value using an empirical calibration curve.

That curve is the first hidden assumption. Calibration is derived from controlled human desaturation studies, typically in healthy adult volunteers breathing adjusted gas mixtures. The resulting algorithm therefore reflects the physiology of that population, usually over a range of roughly 70 to 100 percent. Below about 70 percent, most consumer and many clinical devices are extrapolating beyond validated data. This is why the FDA has stated that oximeter accuracy below 70 percent is not established for cleared devices.

The second assumption is that the arterial pulse is the dominant pulsatile signal. The AC component is treated as arterial blood entering the tissue with each heartbeat, while the DC component is treated as static tissue, bone, and venous blood. Motion, venous pulsation, and low perfusion break this assumption. When the presumed arterial pulse is actually noise or venous blood, the computed ratio becomes physiologically meaningless.

An isolette warmer, a cold exam room, a tight blood pressure cuff still inflated on the same arm, and a patient shivering after surgery all feed the same error pathway. The optical math is unchanged; the input signal is corrupted.

Quantified context: how large the errors can be

The magnitude of these errors matters more than their existence. For most well-perfused patients with normal hemoglobin and light skin, modern oximeters sit within roughly 2 to 3 percentage points of arterial blood gas (ABG) measurement under stable conditions. That figure represents a population average, not a guarantee for any individual reading.

Perfusion is the dominant amplifier of error. In shock, hypothermia, or peripheral vasoconstriction, the pulsatile signal can fall below the device's detection floor, and readings become erratic or fail entirely rather than simply drifting. Studies of critically ill and low-perfusion patients show wide individual scatter, with some readings diverging by more than 5 points and others failing to acquire a signal. The mechanism is straightforward: a smaller AC signal relative to noise reduces the signal-to-noise ratio of the ratio-of-ratios computation.

Dyshemoglobinemias produce errors that are directional and often large. Carboxyhemoglobin absorbs red light similarly to oxyhemoglobin, so standard two-wavelength devices tend to read falsely high in carbon monoxide poisoning, sometimes in the normal range despite severe tissue hypoxia. Methemoglobin absorbs both wavelengths comparably, pulling the ratio toward roughly 85 percent regardless of true arterial saturation; a patient with methemoglobinemia may show a stable reading near that value while being profoundly hypoxemic.

Venous congestion inflates the venous contribution to the pulsatile signal. When venous pressure is elevated, venous blood can pulsate, and the oximeter partly averages arterial and venous saturation. This is the counterintuitive detail most thin explainers miss: a peripheral SpO2 can be pulled downward by local venous pooling without any change in arterial oxygenation.

Edge cases: skin, nails, light, motion, and device state

Skin pigmentation is a real and now well-documented confounder. Melanin absorbs light across the visible and near-infrared spectrum, reducing signal amplitude and altering the calibration relationship. The FDA has issued safety communications and convened advisory discussions on this issue, and large clinical datasets have shown that occult hypoxemia, defined as a normal-range SpO2 with a low ABG saturation, occurs more often in patients with darker skin. The error is not a fixed offset; it is variable and more pronounced at lower saturations and lower perfusion.

Nail cosmetics and artificial nails are a practical, easily overlooked factor. Dark polish, gel coatings, and acrylics can attenuate transmitted light or introduce optical scatter. Studies of nail polish show measurable bias for some colors, particularly dark and blue tones, while others show minimal effect. The pragmatic response is to rotate the sensor or remove the coating when the reading matters.

Motion artifact corrupts the pulse detection itself. Shivering, seizure activity, tremor, and agitated movement can generate pulsatile signals unrelated to arterial flow. Devices use pulse-quality indicators, perfusion indices, and signal-quality metrics to flag this, but a flagged number is still frequently read as if valid. The professional caveat: never act on a reading that the device itself marks as low quality, and never dismiss a low-quality flag because the number looks plausible.

Ambient light adds another layer. Bright surgical lights, phototherapy units, and direct sunlight can leak into the detector and saturate it. Conversely, electromagnetic interference from electrosurgical units and certain monitoring environments can inject noise. Correct sensor shielding and positioning reduce but do not eliminate these effects.

Sensor placement and device condition round out the list. A sensor on an extremity with an inflated blood pressure cuff, an arterial line, or a recent line insertion will read poorly. Reusable sensors with worn LEDs, cracked housings, or damaged cables degrade accuracy. Right-to-left intracardiac shunts and significant anemia add physiologic confounders: severe anemia can yield a reassuring SpO2 while oxygen content and delivery are inadequate, because saturation and content are different quantities.

Misconceptions: what oximetry does not tell you

The most dangerous misconception is that SpO2 equals oxygenation of the patient. It equals the saturation of available hemoglobin, assuming normal hemoglobin species and adequate perfusion. A patient with severe anemia can have a normal SpO2 and inadequate oxygen delivery. A patient with carbon monoxide poisoning can have a normal SpO2 and profound tissue hypoxia.

A second misconception is that a single reading is reliable. Trend matters more than any individual value. A falling trend in a well-perfused patient with a good signal is meaningful; an isolated reading taken through cold fingers, dark polish, or motion is not.

A third is that consumer-grade devices and clinical devices perform equivalently. Over-the-counter oximeters are cleared under different regulatory pathways and are not intended to diagnose or manage disease. Their accuracy claims and validation populations differ from hospital-grade devices.

A fourth is the belief that a low reading always means hypoxemia. Low peripheral perfusion, venous congestion, nail cosmetics, and motion can produce falsely low readings. The correct response to a low reading is to assess the patient and the signal, not to immediately act on the number.

Timeline table: what develops, and when it distorts the reading

Comparison table: alternatives when oximetry is unreliable

Practical actions: verifying a reading that matters

Bulleted red-flag warnings:

A reading that is stable, plausible, and low in a patient with cyanosis or altered mental status warrants co-oximetry, not repeated oximetry. A near-normal SpO2 in a patient with suspected carbon monoxide exposure should not reassure you; standard oximetry cannot exclude it. Any reading flagged as low quality by the device should be treated as invalid, not as a number to round. A sudden change in SpO2 without a corresponding clinical change should prompt sensor and site inspection first. A normal SpO2 in a patient with signs of poor oxygen delivery does not exclude anemia or shock.

Concrete steps for verification:

Inspect the sensor site. Confirm correct placement, remove nail polish or artificial nails if feasible, and check for edema, cold skin, or an inflated cuff on the same arm. Read the pulse-quality and perfusion indicators. If the device provides a signal-quality index or plethysmographic waveform, use it. A flat or irregular waveform invalidates the number. Warm the extremity and reposition the sensor. Recheck after a brief interval; do not average readings across changing conditions. Compare with the patient's clinical status. Work of breathing, mentation, skin color, and capillary refill should be consistent with the reading. Confirm with an alternative method when the reading will drive treatment. Use ABG with co-oximetry when dyshemoglobinemia, severe anemia, or low perfusion is suspected. Document the conditions of measurement. A reading without its quality context is of limited value later.

Thresholds deserve explicit reconciliation. Many clinicians use an SpO2 of 94 percent or below as a trigger for further assessment in adults, and 90 percent or below as a threshold for urgent action, but these thresholds are institutional and population-specific rather than universal. Device labeling, patient age, pregnancy, and chronic lung disease alter the appropriate target. The single number should never override the clinical picture.

Limitations: what the evidence cannot settle

The evidence base on oximetry accuracy has genuine gaps. Most calibration data derives from healthy volunteers with light skin, and the divergence between device performance in that population and in critically ill, dark-skinned, or poorly perfused patients remains incompletely characterized. The FDA has highlighted these concerns and continues to evaluate validation standards, but no single correction factor exists to adjust a reading for pigmentation or perfusion.

Dyshemoglobinemia detection requires co-oximetry; conventional oximetry cannot substitute. Carbon monoxide poisoning and methemoglobinemia remain time-critical diagnoses that oximetry can miss entirely. Anemia and impaired oxygen delivery are similarly invisible to a saturation reading.

The final rule for practice: use pulse oximetry as a rapid screening tool, interpret the number in the context of signal quality, perfusion, hemoglobin species, and patient physiology, and escalate to a direct measurement when the decision depends on accuracy.

This article is general educational information about pulse oximetry principles and is not a diagnosis or treatment plan. It does not replace evaluation by a qualified clinician, and no reading described here should be used to make decisions about an individual patient's care. If you or someone else is in immediate medical danger, contact local emergency services.