Our testing protocols are designed to generate data suitable for submission to the US Food and Drug Administration needed for device approval. All studies are done with the approval of the UCSF Committee on Human Research (IRB).

To schedule a study, please review our Testing Costs page, and contact our clinical research coordinator.

Please be advised:

  • Our calendar fills quickly, and it is best to plan a study at least 2 months in advance if not more.

  • Our coordinator can provide you with necessary documents for your regulatory files – please request these PRIOR to your study start date, and please review the protocols in detail.

  • You must submit information on your devices to our site for IRB approval PRIOR to study initiation- we cannot conduct a study without IRB approval specific to your device. Our protocols do allow for some flexibility in terms of subject count, sample count, and number of devices tested, but we cannot add any last minute additional procedures not already in the protocol. So if there is something unique you’d like done for your study that is not already outlined in our approved protocol, please let us know well in advance so we can determine whether possible and whether additional IRB review is needed.

  • We also recommend that if you are new to clinical studies and are seeking FDA approval, that you contact the FDA to discuss your planned submission and ask questions about the process well in advance. We cannot advise on particular claims or device design aspects.



Standard Pulse Oximeter Study

We are approved by the UCSF Committee on Human Research for desaturation testing to SaO2 levels as low as 50% on healthy non-smoking volunteers; however standard testing is to a minimum of 70% SaO2 is typical.



Met-hemoglobin detection by pulse oximetry with or without simultaneous oxyhemoglobin desaturation. We are approved by the UCSF Committee on Human Research to reach a maximum level of 15% met-hemoglobin.



We can test Carboxy-hemoglobin detection by pulse oximetry with or without simultaneous oxyhemoglobin desaturation. We are approved by the UCSF Committee on Human Research to reach carboxyhemoglobin levels of 13%.


Cerebral Oximetry

We are approved by the UCSF Committee on Human Research to take samples from both the external and internal jugular veins for the purposes of testing forehead cerebral oximeters. In addition to hypoxia, we can also test with the presence of met-hemoglobin or carboxyhemoglobin.


Profound Hemodiltion

Our protocol allows us to remove up to 4 units of blood and reach a levels as low as 7g/dL in total hemoglobin. This study can be done with or without simultaneous oxyhemoglobin desaturation.


Drug-Induced Apnea Testing

Our protocol allows us to use the drugs Propofol and Remifentanil to induce an apnic event.




Transcutaneous Carbon Dioxide Sensors

Transcutaneous carbon dioxide sensors can be tested alone or in combination with pulse oximetry and oxyhemoglobin desaturations.


Motion Testing

Motion fixtures to simulate patient motion can be incorporated into the testing (provided by manufacturer.


Low Perfusion Testing

Low perfusion conditions can be created by external clamps, cuffs, etc. Arterial blood pressure monitoring can be done as well. Different body positions are also possible (e.g. Trendelenburg) during testing.



Testing is done at stable steady state levels of inspired oxygen to achieve target saturations evenly distributed between 70 and 100%. To achieve steady state levels of saturation, continuous airway gas analysis by mass spectrometer and computer is utilized. Pulse oximeter readings are compared to multiwavelength oximetry (Radiometer ABL 90). A typical test involves 12 saturation plateaus, with a total of 24 arterial blood samples for each subject. If extra blood samples are needed, our protocol allows up to a maximum of 35 samples drawn per subject.  Variations on this theme include simultaneous testing of different probe types and body locations, testing of remanufactured probes, testing of probes in selected subject populations (skin pigment, gender), etc.

This laboratory has developed methods that permit us to test multiple pulse oximeters simultaneously. We induce sudden profound and brief stable arterial oxygen desaturation in normal, paid volunteers and sample arterial blood when a stable level of hypoxia has been attained for analysis in a gold standard bench CO-oximeter. 6 to 8 subjects can be studied in a single day, with typically 20-25 arterial blood samples from each subject; 10 to 12 subjects can be studied over two days, with 20-25 arterial blood samples each. The key to this is a computer program that permits the inspired gas mixture to be adjusted by an operator who observes a breath-by-breath display of the arterial saturation computed from end tidal PO2 and PCO2, which is continuously monitored by mass spectrometry.

Typically, saturation is determined once with air breathing and then reduced suddenly to one of 6 levels, range as requested, e.g. 94%, 90%, 85%, 80%, 75% and 70%, for about 30-60 seconds at each level. An arterial blood sample is obtained from an indwelling catheter at the end of each hypoxic plateau. The operator changes the inspired oxygen concentration at the end of each blood sampling and the sudden change this produces in the oximeter recording is used as a time marker for subsequent analysis. A “run” takes 5-7 minutes, during which several plateaus are obtained, e.g. 90%, 80%, and 70%; the run is terminated by a breath of 100% O2 followed by room air. Two runs together compose the six levels of saturation previously mentioned. Saturation of each arterial blood sample is determined by direct oximetry in a Radiometer ABL 90 multi-wavelength oximeter.



Normally, each tested oximeter should provide an electrical analog output signal that we record with LabVIEW in the computer for subsequent plotting and statistical analysis. The timing of the arterial sample on the pulse oximeter recording is obtained from the record of that oximeter using the abrupt fall to a lower plateau or rise produced by re-oxygenation. The mean oximeter output between 6 and 12 seconds before the sudden change of SpO2 is recorded and is read by cursor on a color terminal, and the value transferred to a file for statistical analysis in Excel. The output files include data collected at 2 Hz for each oximeter, and are available on CD or other media, for analysis.

A study normally consists of one normoxic and 19-24 hypoxic comparisons for each subject. In many modern pulse oximeters no analog output is provided. If the manufacturer chooses to record the data in a portable computer at a rate of at least 1 Hz, that data can also be submitted to us for data analysis without extra charge. If we are not asked to analyze data, the charge will be as shown for the first 4 instruments, with additional oximeters counted as non-recorded instruments in the following table. If a manufacturer prefers to collect and analyze the data, the continuous digital signal of each oximeter should be read, for comparison with the blood sample, 9 seconds before the record shows a sudden fall or rise in oxygen saturation, not at the time of blood sampling. This procedure accounts for the delays of finger circulation and uses the estimated delay from the lung to the sample site. There is no useful correlation between the actual time of blood sampling and the oximeter recording because of the extreme variability of tissue blood flow lag. Reduction of the data requires several days. Manufacturer’s representatives may be present for these tests, and may mount the probes. An extra charge is made if no representative is present, requiring us to mount the probes. In all cases, the blood analysis data are provided, including the SaO2, MetHb, COHb and Hgb concentration.



The data analysis report will consist of the following:

  • Graphic plots of the saturation values of each pulse oximeter at each blood sample time plotted against SaO2, the hemoximeter (blood) value.

  • Regression equations for the overall response of each instrument.

  • Plot of the bias values of all samples and all instruments against SaO2.

  • Tables of the mean error or bias, its standard deviation, standard error, 95% confidence interval, maximum and minimum and RMS, all computed both overall and by several sub-ranges of desaturation.

We can record data if each oximeter control unit provides a single-ended analog output signal with BNC terminal mounted on the box. We provide cables to connect this to our computer. The oximeter probe leads should be at least 8 ft long. If no analog outputs are supplied, the manufacturer will be responsible for data recording. Manufacturers must provide their own pulse oximeters and probes. Each individual manufacturer is charged an amount determined by the number of subjects, the number of samples per subject, and the number of oximeters (as shown in the cost table on the next page) plus the facility fee and UCSF overheads. Sites may include the fingers, ears, forehead and bridge of the nose. Data are not published without specific permission of the manufacturer, and in general, no plans now exist for publication of future data.
Please contact our study coordinator if you would like to view a sample of a study report.