May 13, 2005






Hilton Washington, DC North

Gaithersburg, MD


Anesthesiology and Respiratory Therapy Devices Panel Meeting

May 13, 2005







Alan Lisbon, M.D.

Beth Israel Deaconess Medical Center 

Boston, MA


Voting Members

Charles J. Coté, M.D.

Northwestern University

Chicago, IL


Kenneth Drasner, M.D.

University of California

San Francisco General Hospital

San Francisco, CA


Avery Tung, M.D.

University of Chicago

Chicago, IL



David J. Birnbach, M.D.

University of Miami School of Medicine

Miami, FL


Andrea M. Kline, R.N., M.S., PCCNP, CPNP-AC, CCRN

Children’s Memorial Hospital

Chicago, IL


Jacqueline M. Leung, M.D., M.P.H.

University of California, San Francisco

San Francisco, CA


Robert A. Mueller, M.D., Ph.D.

University of North Carolina

Chapel Hill, NC


Industry Representative

Babatunde A. Otulana, M.D.

Aradigm Corporation

Hayward, CA



Consumer Representative

Carolyn Petersen, M.S.

Mayo Clinic, MN


Food and Drug Administration

Julian M. Goldman

Visiting Scholar, FDA Medical Devices Fellowship


Ann A. Graham, CRNA, M.P.H.

Chief, Anesthesiology and Respiratory Devices Branch


Chiu S. Lin, Ph.D.

Division Director


Neel J. Patel, M.Eng.

Executive Secretary


Sandy Weininger, Ph.D.

Electrical Engineer, Division of Electronics and Software Engineering






            Panel Chair Alan Lisbon, M.D., called the meeting to order at 8:00 a.m. to discuss and make recommendations regarding pulse oximeters. Panel Executive Neel J. Patel, M.Eng. introduced the panel members, and read the conflict of interest statement. A waiver was granted to Dr. Robert A. Mueller for his interest in a firm that could be affected by the Panel’s deliberations. The FDA also took into consideration matters regarding Drs. Cote and Leung; each have reported current or past interest in the firms at issue but in matters not related to the day’s agenda. They were allowed to participate fully in the panel’s proceedings.


FDA PRESENTATION: Challenges and Opportunities on the Critical Path to New Medical Devices

Sousan S. Altaie, Ph.D., Scientific Policy Advisor, CDRH, presented information on the Agency’s Critical Path Initiative. The FDA effort started with a white paper analyzing the hurdles of medical product development and calling for more collaboration between government research institutions and manufacturers.

Critical path research leverages basic science to help bring the products to the public faster and more safely. The FDA considers this path a “critical path” rather than a translational path because completing the path is required for the device to reach the market. In the FDA’s view, critical path research tools can have an impact on three areas: device safety; the medical utility of medical products; and the industrialization of those medical products.

Dr. Altaie said that the FDA is involved in this area because it has a uniquely broad perspective on why products fail or succeed and make it to the market. Critical path projects being developed include tools to move in vitro assays to the market; computer models of human physiology; a clear regulatory path with a consensus from the obstetric community for intrapartum fetal diagnostic devices; a pathway for the statistical validation of surrogate markers; work with medical specialty organizations to develop practice guidelines for appropriate monitoring of permanently implanted devices; and a consensus about the extent of the neurotoxcity of contact materials.


FDA PRESENTATION: Condition of Approval Studies: Recent Changes in CDRH

Thomas P. Gross, M.D., M.P.H., discussed the updated functions of the Office of Surveillance and Biometrics (OSB). The office provides support for premarket reviews, tapping a recently increased staff of epidemiologists; monitors a nation-wide passive adverse event reporting system; looks for potential device problems through their Medical Device Reporting System; conducts safety and risk issue analyses; coordinates the Center’s response to important public health issues; and interprets medical device regulations.

Condition of Approval (CoA) studies are ordered for PMA devices; this includes the periodic investigation and reporting of the safety, effectiveness, and reliability of the device for its intended use. In 2002 the FDA began looking at the CoA program by examining the 127 approved PMAs between 1998 and 2000. They focused on the 45 PMAs with clinical CoA orders. This examination uncovered a number of difficulties: the CDRH had limited procedures for tracking study results; their IT systems were found to be deficient; many lead reviewers had left which resulted in a lack of continuity; and they had limited premarket resources. Their strategy for change includes obtaining timely and useful postmarket device information; gathering information on real world use; characterizing the device’s risk-benefit profile; and adding to their ability to make sound scientific decisions.  

To accomplish this, CoA studies have moved from the premarket side of the agency to the postmarket side (to OSB). In addition, they’ve developed an automated tracking system and added epidemiologists to the PMA review team to develop a postmarket monitoring plan during the premarket review period. Dr. Gross said that these changes are expected to improve motivation for good study conduct on the parts of both the agency and industry. He added that when companies do not show due diligence in conducting CoA studies, the Agency is allowed, by virtue of Section 522 authority, to use enforcement strategies such as hefty monetary penalties. This would be done only as a last resort.


Dr. Lisbon noted for the record that the Panel members constituted a quorum, and he began the agenda items.



Ann A. Graham, CRNA, M.P.H., Chief, Anesthesiology and Respiratory Devices Branch, CDRH, presented information on the regulatory aspects of pulse oximeters, noting that they are Class 2 devices subject to 510(k) premarket regulations. Their intended use is the noninvasive, continuous or spot-checking monitoring of oxygen saturation, distributed by prescription. The device’s patient population is adult, pediatric, infant, and neonate. The application site can be the finger, the earlobe, the back, and forehead. They are used in a variety of settings: operating room, critical care, post-anesthesia, and the recovery room. However, their use has been moving recently into the home and other environments outside of medical or health facilities.

Ms. Graham went over the requirements of a 510(k) application. They include a description of the device, the device’s design, its comparative performance data with a legally marketed predicate device, and labeling. She noted that the accuracy specifications for the pulse oximeter depend upon the type of device sensor and the intended use population. For the transmittance wrap and clip device, the accuracy specification must be ≤3%; however, when used for neonates, the specification has a 1% degradation factor added to it, for a rate of  ≤4%. This is a result of the Agency’s agreement to accept adult data for neonatal use. Transmittance ear clip and reflectance sensors are not cleared for infants or neonates.

In the last couple of years her office has cleared about 35 510(k) applications annually—this is a substantial part of their workload. Most of these are for transmittance sensors, single use and non-sterile. The number of pulse oximeter applications into her office has dramatically increased since 1986, when they began to be used in clinical practice; there has also been a significant increase in the last 5 years.

She described the two documents the branch uses extensively, and how they need to be amended. The 1992 FDA CDRH guidance document on pulse oximeters needs updating on neonatal and labeling issues and over-the-counter (OTC) use. The 2005 ISO standards document does not offer the branch all of the guidance it currently needs, specifically on test methods and the differences between reflectance and transmission sensor types.

The Panel asked to define the term “neonatal.” She said there is guidance that defines neonates as under 30 days old with an upper weight of 10 kilos (See page 11 for her additional clarification offered later in the meeting). In response to another question, she noted that the guidelines exclude fetal use of the pulse oximeter. The FDA has two other oximeters categories of devices, she added: fetal oximeters are Class 3 devices requiring a PMA, and in-dwelling tissue oximeters. To a question about the use of predicate devices, she said that the standard has been evolving, and that manufacturers use both fairly new technology as well as older predicates for the 510(k) comparative performance data requirement.

Sandy Weininger, Ph.D., Office of Device Evaluation, Anesthesiology and Respiratory Devices Branch, CDRH, presented the regulatory history of pulse oximeters, the 1992 guidance document, and the branch’s thinking about pulse oximeter regulation given the changes in the technology since the early 1990s. He addressed the Panel’s earlier questions to Ms. Graham, stating that manufacturers will not necessarily have the same definition of neonatal as FDA, and that the FDA requires substantial equivalence of a device—that is, the manufacturer is required to demonstrate that their device is substantially equivalent to a legally marketed device.

CDRH published their 1992 oximeter guidance roughly 10 to 15 years after the devices appeared on the market. He went over what the FDA asks for in a 510(k), including identification of the configuration of any probe and monitor combinations (possibly with clinical accuracy validation), as well as any accessories—these have all been shown to affect accuracy. They also ask for engineering drawings and a description of any functional elements, alarms and alarm limits, and the defaults the manufacturer may have on these features. Manufacturers must compare their device to a predicate device, provide information on intended use, use indications, functional verification, and accuracy verification of the human subject testing. He noted that this last requirement is the most important piece of information they can get on a pulse oximeter.

The FDA also asks for a desaturation study, with a minimum of 10 healthy subjects, a range of age, skin tone, and gender, and a report on the root mean square error (Arms). They request the collection of neonatal convenience samples, while recognizing the limitations of being able to do this. To compensate for this, they degrade the adult accuracy by 1 percent. It is recommended that the manufacturer supply at least 200 data points, over a saturation range of 70 to 100 percent SaO2 (He noted that the standard does not call for this), with paired observations. They must reveal any environmental factors, including surface temperatures, electrical safety, electromagnetic compatibility, and mechanical testing of the device. Pulse rate accuracy measured using in vitro calibrators is required, set to lowest values. They also ask for sensor specifications labeling that outlines environmental specifications, sensor/monitor combinations, application time, use indications, continuous use and spot checking use specifications, and associated alarms.

For reusable probes, the manufacturers must demonstrate their ability to clean the probe and have it return to its normal and safe condition, over the lifespan claimed. The new oximeters are very software intensive, so the FDA wants to make sure that the manufacturers use the agency’s software safety guidelines to demonstrate the device’s safety and that the software development process is appropriate. The manufacturer must also present results of biocompatibility review for all materials used. Reprocessed single use devices (SUD) are now a larger part of the branch’s review activities; manufacturers must demonstrate the probes’ accuracy after the number of claimed use/cleaning cycles. They do not have to show electrical leakage or electromagnetic interference validation. 

A member Panel asked Dr. Weininger how manufacturers track can SUD reprocessing. He stated that the SUD manufacturers must have a tracking process in place that the Agency will review. The Panel also addressed the question of how well distributed the 200 samples must be. He said that they must be equally distributed across the entire range between 70 to 100 percent.

Dr. Weininger’s second presentation examined the pulse oximeter standard and highlighted the differences between the standard and the 1992 guidance document. ASTM created the F1415 standards prior to 1992; this led to the ISO 9919 standard and the European standard (EN 865). ASTM completed ISO 9919 (entitled Particular Requirements for the Basic Safety and Essential Performance of Pulse Oximeter Equipment for Medical Use) in January 2005, with a large number of clinicians and engineers contributing to it. The FDA is in the process of recognizing it.

The standard covers safety and performance, while the 510(k) process looks at substantial equivalence requiring the manufacturer to show by objective evidence how their device compares to a predicate device. The standard applies to all originally manufactured equipment, reprocessed probes, and extender cables. There are extensive requirements for labeling, documentation, and information annexes.

The standard’s accuracy specifications must be defined over an evenly distributed range of 70 to 100 percent SaO2; the Arms must be less than 4 percent. Dr. Weininger said that he hopes that the recommended desaturation profile becomes the standard profile in the next version of the standard. The standard requires that if there is a physiologic alarm there also be a low SpO2 alarm; the low SpO2 default limit in the 1992 standards was 80 percent saturation, but that has been raised to 85 percent. The device must also include a description of the signal adequacy indicator. The standards ask for information on how the device is being used, addressing patient population, the body part, and environmental information about the application of probe. The maximum application time must be disclosed, as should the rationale for that limit. The standards require a variable pitch tone on the oximeter, where the pitch lowers as the SpO2 reading lowers. They also require information on safer surface temperature standards. The IRC 60601-1 General Standard coming out this year or next year raises the surface temperature from 41C to 43 C, based on a required risk analyses. The ISO 9919 adds more constraints to this: for adults it allows 41C for four hours, 42 C for 8 hours, and 43 C for 4 hours, with neonates limited to 41 C. The standard also requires an automatic reset after 4 hours. He stated that many of the burns reported in the literature are actually pressure contusions.

Dr. Weininger discussed what items are not in the standard. The standard does not require a disclosure of the oximeter’s possibly poorer performance in neonates. There are no requirements or test methods outlines for motion artifact or low perfusion, and no single test method for accuracy and surface temperature. He believes a single test method would be helpful.

The Panel directed several questions to Dr. Weininger. Panel members were concerned about the standard for audible tones on pulse oximeters, and that there are still devices in the market where the tone does not drop as the saturation drops. Dr. Weininger said compliance with the ISO standards is voluntary, and there is no recommendation in the 1992 guidance to address this. Several members said that the FDA must consider a tone requirement in their guidance documentation. The Panel also addressed concerns about the fact that the surface temperature standards are based on adult data, with no data on neonates. Dr. Weininger said that the standards committee tried to identify the “worst case scenario” for neonates in an incubator, and that they believe that 41C is safe in that situation.

Panel members discussed how a pulse oximeter is cleared. Dr. Weininger said that the FDA can use the standards in their decision as to whether a device meets substantial equivalence, but it is not a requirement. The predicate device still exits in the market place as long as a safety hazard has not been identified. Manufacturers can get clearance by establishing substantial equivalence to a predicate device even though they do not meet the standards set by the FDA, as long as there is no obvious safety risk. The Panel asked for clarification between requirements and recommendations. Ms. Graham said that in a 510(k), the FDA makes recommendations while recognizing voluntary standards. There are regulatory requirements in a 510(k) set out in the CFR, but those are generally a broad list of requirements, including labeling and manufacturer’s discloses. With respect to pulse oximeter alarms, Ms. Graham said that there are expectations that these will be present in the device. Because these devices have now become primary clinical decision making tools, the 1992 guidance document asks for in vitro validation of the pulse rate and other data they expect the manufacturer to submit. However, this is a recommendations document that does specify requirements.

The Panel also raised the question of the origin of the 1 percent degradation for neonates. Dr. Weininger said that he believes that it is based on concerns raised in the mid-1990s about the influence of fetal hemoglobin (HbF) on the co-oximeter. In 1989 a NIH consensus panel discussed HbF influence, but they did not believe it to be clinically significant. The compromise was to allow a 1 percent degradation.

Ms. Graham clarified the definitions of neonates, as requested earlier in the meeting. On May 14, 2004 the FDA published FDA Guidance for Industry and FDA Staff: Premarket Assessment of Pediatric Medical Devices in which neonates are defined as from birth to 1 month, infants from greater than 1 month to 2 years, children from greater than 2 years to 12 years, and adolescents from greater than 12 years to 21 years. The document recommended that industry use these categories.

Julian M. Goldman, M.D., Departments of Anesthesia and Biomedical Engineering, Massachusetts General Hospital, and Visiting Scholar, FDA Medical Devices Fellowship Program, clarified the issue of variable saturation tone in pulse oximeters, raised earlier in the meeting. The ASTM and the ISO standards do not require use of a variable saturation tone, but they state that if one is present it must function such that the tone decreases if the saturation decreases, and vice versa. Some pulse oximeters are designed for spot checking and therefore have no need for physiological alarms.

Dr. Goldman’s presentation covered the clinical considerations of pulse oximeters, as related to the three questions the Panel has been asked to address. He stressed that it is important to understand the underlying assumptions of the pulse oximeter model. One important assumption is that the change in finger blood volume over time is important. As well, the Panel should keep in mind that the clinical use of the pulse oximeter will be most successful if the use conditions match the model’s assumptions, and if the calibration is performed on healthy adults under laboratory conditions—not on sick patients in the hospital. When clinical conditions diverge from laboratory conditions, inaccuracies may appear.

He covered the FDA’s questions for the Panel and the issues related to these questions. Question 1 deals with the differences between transmission and reflectance pulse oximeters. Dr. Goldman noted that both techniques are based on the same model and the same assumptions, but there are still a number of other effects that may be different between the two. He went over some of these differences, but asked if there was something that could be done from a design or educational perspective to have clinicians use the devices properly. As well, he raised the issue of whether reflectance probes should be specified for use only on validated sites, and whether it clinically acceptable to specify the use of a probe for other sites.

With respect to Question 2, he raised the issue as to whether the 1 percent degradation is clinically appropriate. While the rational is unclear and supporting data is lacking, experience in the clinical setting does seem to support the current approach. Postmarket convenience samples may be useful to support clinical assumptions.   

Dr. Goldman addressed the third question, dealing with OTC use of the device. He noted clearing the devices for OTC use would require premarket submissions and assessments of the instructions for lay persons. This raises issues about incorrect applications leading to bad data and bad outcomes, and how accurately would lay persons interpret the data. He noted that most errors produced show falsely low numbers—most consumers, he believes, would follow instructions and call a clinician for advice. The benefits to allowing OTC distribution of pulse oximeters may include empowering patients, moving healthcare into the home and creating the “ambulatory practice of the future,” and prompting additional applications.

The Panel asked Dr. Goldman about the use of pulse oximeters in the home. He said that the benefits from distribution of the devices in homes would include giving clinicians better information about their patient’s condition than simply a report by the patient of how they felt. He and Panel members suggested that data looking at the possibility of erroneously high saturation values during OTC use would be valuable.



Paul B. Batchelder, LCRP, RRT, Chief Clinical Officer, CLINIMARK, presented an evaluation of 43 articles in peer-reviewed journals looking at pulse oximetry accuracy in the clinical environment for adults and neonates. This literature review resulted in three primary points. The first is that the pulse oximeter accuracy is not dissimilar between adults and neonates in hospital setting; second, that the influence of HbF is not clinically significant, making the 1 percent degradation based only on anecdotal evidence; and, finally, that most neonatologists surveyed stated that collecting a statistically meaningful number of neonates to verify accuracy of SpO2 and SaO2 numbers over the range of 70 to 100 percent is unfeasible.  

Accuracy rate for adults was 3.26 percent (± 1.32 percent) between the laboratory and the hospital, and the neonates had a rate of 3.44 percent (± 1.35 percent)—showing essentially no difference, he said. He suggested that the 1 percent degradation factor may actually reflect the laboratory-hospital difference, and not a difference between neonates and adults. The factor may have resulted from when the Nellcor N-100 was evaluated in the 1980s.

He offered a number of suggestions for how to acquire workable data and suggested that  the adult laboratory accuracy validation combined with physical tests in the intended population and convenience samples continue to be used. As well, he suggested that the 1 percent degradation factor should be discontinued and clinical use performance should be clarified in the device labeling.

Paul D. Mannheimer, Ph.D., Principal Scientist, Nellcor/Tyco Healthcare, discussed the difference between reflectance and transmission pulse oximetry in adult and neonatal applications. Reflectance and transmission pulse oximetry technologies should be treated equivalently; there is value in manufacturers’ ability to provide products using both reflectance and transmission pulse oximetry devices; and both geometries should be treated the same when validating new products for the neonatal population.

He discussed the principles of pulse oximetry, noting that they are the same independent of emitter-detector type and position. Transmission sensors are used on parts of the body that are thin enough to allow light to transfer from the emitter to the detector, such as fingers and earlobes. Reflectance probes diffuse light parallel to the body’s surface—they do not reflect light, despite the name.  

Dr. Mannheimer noted that there has been a renewed interest in reflectance oximetry. Early reflectance devices, from the mid-1980s, could not tolerate extremely low pulses—however, he said that their performance in these situations has dramatically improved. He presented the accuracy performance data for the Nellcor N-400, a reflectance probe designed for use on fetuses.

Looking at the vulnerabilities of transmission and reflectance pulse oximetry, he said that there is no practical difference between the two types. Both probes can be improperly applied—often resulting in shunting—or applied to an inappropriate location.  All of these vulnerabilities are present in adults, children, and neonates. Reflectance probes offer additional value in that they can be uses on parts of the body that are not opaque—such as the back or the forehead. A recent abstract (Bebout DE, Mannheimer PD. Anesthesiology 2000; 96:A558) showed little vasoactivity when a reflective probe was used on the forehead and the patient was sifted from 74 F to 58 F. In another study (MacLeod DB, et al. Anesthesia 2005; 60 (1):65-71), researchers briefly induced hypothermia in a healthy adult to evaluate the delay in the blood reaching monitoring sites.

He suggested that neonatal 510(k) submissions for pulse oximetry should be tested as for any other pulse oximetries, independent of sensor geometry. They should be validated in controlled laboratory studies on healthy adults; form, fit, and function tests should be done to demonstrate physical safety and appropriateness on neonates; and that convenience samples should be taken on neonates around the 90 percent saturation rate to confirm system accuracy. He recommended simultaneously testing predicate device and the new device, to make sure the level of performance is comparable to what is being used clinically.

The Panel asked him why it would not be possible to do studies with multiple monitor sites on the neonate. Dr. Mannheimer said that it is not unreasonable, as long as the patient is not made hypoxic.

Philip Isaacson, Founder and Director of Nonin Medical, Inc., presented information on pulse oximeter applications for sports. He noted that they are showing up in many places, including among patients who receive prescriptions for them. Nonin Medical makes no medical claims about their product, the SportStat Fingertip Pulse Oximeter, because it is being sold for consumer use rather than for medical use (i.e. prescription or over-the-counter (OTC) use).

Their consumer use product is used by mountain climbers to avoid acute mountain sickness, high altitude pulmonary edema, and high altitude cerebral edema. Hikers, skiers, bikers, and those who work out in gyms have also used the device The product has been used in a variety of environments, including outdoors in extreme cold and heat and at high altitudes.

Brodie Pedersen, Senior Quality Design Engineer, Nonin Medical, Inc., spoke on the consumer and aviation applications of pulse oximeters. Pilots have used these devices to detect hypoxia while flying without oxygen. He noted that hypoxia can occur when a pilot is flying as low as 5,000 feet. Personal tolerance for high altitude can vary daily, and he said that Nonin’s device helps consumers understand how their body is responding to various conditions; consumers might not recognize when they are in need of oxygen because their mental acuity will be dulled when their oxygen levels drop. Using their device can help pilots constantly monitor when they should adjust oxygen flow. He briefly noted that when the device is sold over-the-counter (OTC), safety and effectiveness are regulated; when they are sold in the consumer market, safety is not regulated.  

The Panel asked Mr. Pederson and Mr. Isaacson whether any of their products have been tested and compared to the medical device under clinical conditions. Mr. Pedersen said the consumer devices are identical to the medical device. The Panel expressed concern that consumers may not have the training to use the devices appropriately. The Panel also asked whether the manufacturer has tested the device at 30,000 feet, the level at which they claim operational efficacy. Mr. Pedersen said that they have been tested in simulators to show that the electronics in the device are not affected by that altitude, but the tests did not involve humans.



Dale Gerstmann, M.D., Director of Neonatal Research, Utah Valley Regional Medical Center, discussed his study of the functional performance of neonatal pulse oximeters, specifically addressing bias, precision, and resolution. The patient sample was taken between 2002 and 2005, from a 40-bed, level-three NICU with about 500 admissions annually. The NICU does not take care of neonates with complex congenital heart disease, complex pediatric surgical problems, and neither do they conduct extracorporeal membrane oxygenation (ECMO). The study included 114 subjects. All data were collected under two IRB-approved protocols (SpO2:SaO2 pairs with informed consent, and SpO2 recording with anonymous waived consent). There were 1,248 pairs of data.

Bias is the difference between the pulse oximetry and the arterial saturation. He presented Bland-Altman plots displaying bias, precision, and Arms for six different devices. A bias measurement at 0.0 is ideal, but only three devices come close to this level. He noted that the average bias by saturation is not uniform across the range for which clinicians typically see saturations in sick neonates on ventilators and whose oxygen is being controlled; as the saturation drops, the bias increases. More clinically acceptable would be to have a device that shows flat performance. Blood gases are drawn less frequently on ventilated neonates with no complications; therefore, convenience data is becoming harder and harder to obtain. Dr. Gerstmann presented a table extracted from the same data showing that all of the Arms values are under what is currently acceptable (<4); this shows that the 1 percent HbF adder is unnecessary.

Resolution is an instrumentation term that relates to what a clinician would see on the device’s display. The precision or the resolution of the monitor is ±1. However, after plotting the 95 percent confidence intervals, a wide discrepancy appears between what is shown on the monitor and what the data really means, he said—the accuracy reading could actually be ±4 or 5. A clinically acceptable device would ideally show readings that are closer to where precision and the bias is acceptable, and be closer to what the resolution on the monitor shows.

The Panel asked Dr. Gerstmann whether the data had been published. He said that the smaller data set was published in the Journal of Perinatology two years ago, and that he is preparing the larger data set for publication. The Panel congratulated Dr. Gerstmann for obtaining data the manufacturers have claimed is unavailable. They also noted that it is disturbing to see positive bias measurements at low saturations, and that it seems to open the door for the appearance of false negatives.

The Panel expressed concern about the device’s reliability under adverse conditions, and whether the manufacturer had any devices returned. They asked the Nonin representatives if they recommended that climbers purchase spare devices to protect against device failure in extreme and life-threatening situations. Nonin offers a three-year warranty on the device, they said, and that they very rarely fail. There is no way to recalibrate the device in the field. When they fail, they tend to fail to shut off when expected. Panel members also expressed concern that the Nonin device’s accuracy dramatically falls off below 70 percent saturation. A Nonin representative stated that they have actually tested the device down to 50 percent, but with limited data. Panel members also expressed concern that the manufacturers have not tested the device at high altitude on humans, and that there is little training provided with the device.

The Panel asked Dr. Goldman why the pulse oximeter error rate tends to read a low saturation, and how a recreational consumer user might respond to this. He said that historically, the older pulse designs read low in the presence of low perfusion and motion conditions, such as shivering. Today, there is a better understanding of why that happens, so these errors have generally been corrected. Typically, high readings (which are very uncommon) are due to equipment problems, such as a damaged cable. He added that the presentation previously that showed positive bias in the lower saturation ranges raises the question of how heavily laboratory testing should be relied on to indicate real world performance. He urged increase testing on the population of intended use, otherwise there will be surprises.



Dr. Lisbon read the first question into the record. Panel members generally agreed that there seem to be no differences between transmission and reflectance oximeters; however, there are differences in their clinical applications. Panel members suggested that pulse oximeter device should include clear, population-specific instructions on probe locations and positions that will achieve laboratory level accuracy. For example, some areas on a neonate’s head may be more efficient than others. These usage instructions should be supported by data focused on the clinical applications on specific body parts.

Panel members noted that clinicians are using these devices to drive decision making, so they must understand more about signal quality, device limitations, warnings, and how the devices perform using different types of sensors. A possible answer to the need for additional information may be to ask manufacturers to include instructional inserts with each probe, similar to drug information inserts. Some Panel members noted that in the pediatric world, reflectance technology is rarely used, and clinicians would need more guidance on where to place these types of devices on a child’s body.

A Panel member brought up the issue of an excess of oxygen in pre-term neonates and how to protect against that. If these devices are underreading saturation, that could have significant implications. Many Panel members stated that validation studies in the neonate population are feasible and should be done. In fact, some suggested designing studies that randomize where on the surface of the neonate’s body the pulse oximeter is placed to understand which will provide the most accurate reading—this would involve multiple oximeters placed on a single patient’s body at the same time.

Panel members suggested premarket studies for oximeters not currently on the market, and postmarket studies for those already in use. This last effort would help everyone understand how the algorithms in the current devices are different from the predicate devices. Panel members acknowledged, however, that postmarket studies would involve some logistical concerns. One member suggested that postmarket studies could be funded by manufacturer user fees, similar to drug studies, or pulse oximeter industry contributions to a fund.


Dr. Lisbon read the second question into the record. Panel members generally agreed that the 1 percent degradation to compensate for the accuracy rates between neonates and adults is unnecessary. Members discussed how saturation measurements below 70 percent can be validated; there are some clinical situations where this data could be collected and interpreted, such as cardiac surgery. Some members expressed concern about Dr. Gerstmann’s presentation that shows large biases in pediatric saturation measurements at the lower end. This underscores the need for postmarket research, publication, as well as work by the manufacturer to see if there are some systems that have lower biases at lower saturations for neonates.

Panel members discussed the FDA’s practice of considering pulse oximeter systems as single units when the end users can substitute different probes. Some suggested that there should be testing for each probe with each possible unit, especially given the difference in device algorithms. The FDA said that this issue has come up in their discussions but with no resolution. This also could be managed with labeling, noted on member, because getting each sensor approved for each unit may be very burdensome. One Panel member suggested that the 70 percent validation limitation on oximeters for neonate use should be dropped to 60 percent, and require additional validation testing for these levels, to accommodate the increasing numbers of very sick children many pediatricians are seeing.  

Panel members said that it is important to understand the biases of each range in which these devices are used, because the biases are different at low ranges versus high ranges. As well, because the biases are different among different boxes and probes, the average physician needs better guidance on which probe and which box fulfills his requirements. Physicians also need help interpreting the numbers cyanotic children may produce. For example, 25 percent saturation isn’t a meaningful number; it only tells the physician that the child is in very serious trouble. The labels should reflect the fact that the numbers have variable interpretations from patient to patient. There are populations that should be studied to understand better what these numbers mean, despite the cost in time and money.

Donald Melnikoff, Smiths Medical PM, addressed the Panel. He said that there are inserts accompanying disposable reflectance probes but they are not packaged with each probe—they are placed in the box of probes. The inserts are already very cumbersome because of their size, so they typically do not get attached to each individual probe. These inserts do indicate that there is no validation for use below 70 percent.

Panel members discussed why the device displays a number, such as 40 percent, that is uninterpretable. One member suggested that a solution may be to provide no numbers below 70 percent or some other lower level. Other Panel members noted that the trend is actually more important in such situations, so a clinician would still want numbers below 70 percent to track a trend. The instructions may need to be simplified and streamlined so that health care workers can quickly extract meaningful information from them. However, other members warned that the data is truncated below 70 percent, so there is no continuous scale to track. This may cause additional misinterpretation of the numbers.   

Before opening Panel discussion on the third question, Dr. Lisbon asked Ms. Graham to describe the differences between OTC products and consumer products (nonmedical). She said that OTC devices would have a medical claim associated with them. OTC use of a pulse oximeter, for example, would identify intended use and a patient population.  A nonmedical device, such as the SportStat, is not medical device because no medical claims are made.  An OTC product cleared through FDA requires the same level of evidence to make a determination of substantial equivalence is the same as a prescription product. OTC requires additional information, she added, such as appropriate labeling outlining intended use. The FDA would also want to include language that indicates when the use of an OTC pulse oximeter may not appropriate, but they need the Panel’s suggestion on how to word this. Additionally, the FDA has more control over the manufacturing processes for an OTC device than for a nonmedical device. A Panel member asked what kind of claims the manufacturer can make on a label. Ms. Graham stated that pulse oximeters are currently labeled to show intended use, whether for monitoring continuously or on a spot-check basis; any other claims would need additional validation. Mark Melkerson, Acting Associate Director, Office of Device Evaluation, noted that a treadmill marketing for general exercise use would be marketed as a nonmedical consumer product, while the same product intended to restore muscle function would be marketed as a medical device.

Dr. Lisbon read the third question into the record. There was some disagreement among Panel members as to the value a home pulse oximeter would have on consumers’ health and health decision making. Some members who did not feel comfortable with the consumer use of pulse oximeters in the home said that if a person is sick enough to need the device in their home, this use should be under their doctor’s direction. They expressed concern that consumers would not know what to do if the device showed a low saturation level, and that this would give them a false sense of security. While this device may encourage patients to take more responsibility for their health, it may also discourage them from following up with their health care provider. There is no guarantee that patients will receive the education they need to manage the decisions brought about by the use of a pulse oximeter.

Panel members who felt that pulse oximeters should be sold OTC said that parents with a sick child would not buy an oximeter for home use without consulting their pediatrician. These Panel members suggested that making the devices fall under OTC guidelines would actually improve consumer guidance on their use and limit the claims manufacturers can make. The introduction of these devices in the home could empower consumers and patients, and encourage them to take actions to maintain their health.

Some Panel members suggested that the devices should include some kind of calibration device to indicate when device is working properly. Another Panel member suggested that to reduce the burden on consumers to interpret the oximeter’s results correctly, home-use models could simply feature indicator lights. Other Panel members suggested that the FDA should select a patient population and conduct validation studies in that population before allowing the device to be marketed OTC.

Some Panel members proposed a compromise that would involve having the device undergo trials in a single community; this may address some members’ concerns that there are no studies showing how consumers would handle the device. Another suggestion was that these devices would work especially well where there are public defibrillators; this was agreed to by most Panel members. Many members noted that numerous medical devices are OTC, such as glucometers and blood pressure monitors, and they are being used by consumers with very good results.



Dr. Lisbon opened the floor to members of the audience who want to make a statement. Mr. Melnikoff asserted that customers are buying the devices as commercial products, but industry would like to market them as OTC products while providing a higher level of education.

A Panel member asked the FDA if they only need to identify accuracy data for OTC use, or whether efficacy data also necessary. Ms. Graham stated that the end point would be in the comparison to a predicate device.

Dr. Gerstmann said that the way pulse oximeters are being used today has made it one that controls disease, rather than a monitoring device. He said that he continues to be concerned that the national studies are using devices that are not as accurate as they should be.

Dr. Batchelder said that no validation is required in the neonatal population currently, as there is usable information available. Reflecting what a colleague, a pulmonary cardiologist, has said, Dr. Batchelder asserted that physicians would like to have pulse oximeters sold OTC because many do not prescribe them. Consumers are buying them on the Internet; these should have same accuracy as medical devices, along with better information and explanations of the accuracy. He thinks they should survey pulmonologists in the United States to solicit feedback on how the devices are being used among consumers at home.



Question 1:  Pulse oximeter sensors may be implemented in either a transmittance or reflectance configuration. In both configurations, light is scattered by blood, which has time dependent characteristics, and bone or other tissue structures which are not time dependent. Transmittance sensors are configured in a manner where the emitter outputs light which travels through tissue (e.g. finger, toe, and ear) and is received on the opposite side by the detector. Reflectance sensors are configured with the emitter and detector in the same plane. Emitted light must reach the detector by reflection off a surface which typically results in smaller signal strengths in comparison to transmittance sensors. Please discuss the clinical differences between transmittance and reflectance sensors. In your discussion, please specifically comment on:

a.  Any difference in performance between the two sensors.

b.  Whether the differences in performance would lead you to recommend different pre-market evaluation methods and standards, and, if so, what those would be.

c.  Whether differences in performance would exclude certain indications for use for one type compared to the other and, if so, what those would be.



a.  The Panel agreed that there are no fundamental technological differences between the two sensor types; however, each type of pulse oximeter should be validated in its own right to understand the variabilities that are present. The ways they are used in clinical settings—location of sensors and patient population—affect the resulting measurements.

b.  The Panel said that there are differences in performance requiring premarket validations based on their clinical use.

c.  The Panel agreed that there are no differences in performance that would exclude indications for use for the two sensor types. However, the Panel suggested a caveat when using these devices in children with cyanotic heart disease.


Question 2:  The agency currently recommends that pulse oximeter sensors are clinically validated to a stated accuracy (e.g. ± 2% from 70-100 %) on healthy adults under ideal laboratory conditions. Then, for transmittance sensors intended for use on neonates, a 1% degradation factor is added to the stated accuracy specification (e.g. ± 3%) to compensate for uncertainty due to the inability to perform suitable calibration studies and the apparent effect of fetal hemoglobin (FHb) on saturation measurements (there is some evidence that co-oximeters, the reference device used to analyze blood, is inaccurate at high concentrations of FHb).

a.  Please discuss whether you believe the agency should continue to follow this recommendation for validation of transmittance sensors intended for use in neonates and if this same recommendation should also be followed for validation of reflectance sensors intended for use on neonates. If the panel does not feel that this recommendation is appropriate, please provide the agency with suggestions and recommendations as to how the validation of accuracy for neonatal use should be revised.

b.  Please discuss whether you believe that the labeling for sensors, especially those for neonates, should contain information on how the saturation accuracy specification was developed and validated.


a.  Panel members agreed that the 1 percent degradation factor for neonates is unnecessary, and that neonates and adults should be evaluated under the same standards. A neonate device should be evaluated in its own right and not use adult data. There was also a request for a breakdown of the biases and increased precision over the ranges: 70 to 80 percent, 80 to 90 percent, and 90 to 100 percent. Manufacturers should provide increased disclosure as to the origin of the validation data, how it was obtained, and its limitations.


b.  Package inserts should detail how a specific device functions with a specific processor.


Question 3:  To date, the agency has not cleared any pulse oximeters for medical uses as over-the-counter (OTC) devices. Please comment on the risks and benefits of OTC pulse oximeters. Please specifically discuss:

a.  Under what circumstances, and for what general or specific clinical conditions or indications (asthma, COPD monitoring), if any, should the agency allow the clearance of an OTC pulse oximeter.

b.  Whether you believe adequate labeling can be written to ensure the safe1 and effective2 use of an OTC pulse oximeter including directions for use, contraindications, warnings, and precautions. If so, please comment on what specific elements or statements should be in such labeling.

c.  Whether the significant chance for misuse of OTC pulse oximeters exists if cleared or approved by the agency for OTC use and what the potential risks would be.



Dr. Lisbon asked Panel members how they would vote on whether to recommend pulse oximeter devices for OTC release, if they had to vote now based on what they currently know and given appropriate labeling: 4 responded yes, 4 responded no, and 2 were undecided.

The Panel was divided on the OTC issue because most felt that they did not have sufficient information and study results to make a final judgment on any part of this question. Some members suggested polling pulmonologists to gauge their sense of how patients are handling the home use of pulse oximeters. Others suggested that OTC use of the device might be introduced into a limited study population or place the device with publicly accessible defibrillators.    

Panel members also urged the FDA to consider two safety features in the future: a change in tone installed in the device to correspond to a reduction in oxygen saturation level for continuous readings; and a push for getting rid of the pulse oximeter clips because too many accidents have happened when children are fitted with adult-sized clip-ons.



Dr. Lisbon thanked the speakers and the Panel members. He adjourned the meeting at 3:10 p.m.


I certify that I attended this meeting of the Respiratory Therapy Devices Panel Meeting on May 13, 2005, and that these minutes accurately reflect what transpired.



Neel J. Patel, M.Eng.

Executive Secretary



I approve the minutes of this meeting

as recorded in this summary.



Alan Lisbon, M.D.





Summary prepared by

Susan C. Sanderson

5932 Lee Highway

Arlington, VA 22207

(703) 237-3696