[Digest Note] Draft Guideline for the Revised De Novo Classification Process

1. Introduction

On August 14, 2014, the U.S. Food and Drug Administration (FDA) issued Draft Guidelines on the direct de novo classification process, a means of accelerating the approval of new types of medical devices posing only low to moderate health risks.[1]  The FDA created de novo classification in 1997, but after the process failed to achieve its purpose of expediting approval, the FDA introduced an alternative de novo process called “direct” de novo.

A. Medical Device Classification by the FDA

The FDA, part of the Department of Health and Human Services (HHS), is responsible for protecting public health by assuring the safety and efficacy of a range of products, such as drugs, biological products, and food.[2] The FDA also regulates medical devices, and one way it attempts to efficiently review new types of devices is by classifying them based on “the level of control necessary to assure the safety and effectiveness of the device.”[3]

Class I devices pose a low to moderate risk to health and are subject only to general controls, which include regulations on adulteration, misbranding, and manufacturing practices.[4] Examples of Class I devices include dental floss, elastic bandages, and examination gloves.[5]

Class II devices pose a moderate to high risk to health, and general controls alone are insufficient to provide reasonable assurances of safety and effectiveness. Class II devices require “special controls,”[6] which include performance standards, premarket data requirements, and special labeling requirements.[7] Examples of Class II devices include condoms, medical lasers, endoscopes, and radiofrequency ablation systems.[8]

Class III devices pose the highest risk to health and are subject to the most stringent controls. Manufacturers of most Class III devices must receive premarket approval (PMA) from the FDA in order to market their devices.[9] To obtain PMA, the manufacturer must provide scientific evidence demonstrating with reasonable assurance that the device is safe and effective for its intended use.[10] As a result, the PMA process takes about a year to complete, on average.[11]

Marketing Class I or II devices is less burdensome. Manufacturers must notify the FDA of their intent to market their medical devices unless their devices are exempt from such requirement. This process is known as premarket notification (or 510(k), as Section 510(k) of the Food, Drug, and Cosmetic Act (FDCA) governs this process).[12] 510(k) applicants must demonstrate that their device is at least as safe and effective (“substantially equivalent”) to a legally marketed device (“predicate device”) not subject to PMA.[13] Manufacturers can prove substantial equivalence by showing the same intended use and the same technological characteristics as the predicate device.[14] Compared to PMA, 510(k) requires a lower burden of proof to manufacturers.[15] Most Class I and a few Class II devices are exempt from the premarket notification process, so manufacturers of these exempted devices are not required to file a premarket notification application and receive FDA clearance before marketing the devices in the market.[16] However, 510(k) exempt devices are still required to be suitable for their intended use, be adequately packaged and properly labeled, and be manufactured under a quality system.[17]

Table 1. Risk levels and the approval procedures for Class I, II, and III devices[18]

  Class I Class II Class III
Risk Level Low to moderate risk Moderate to high risk High risk
Approval Procedure Almost all devices are exempted from the premarket notification requirement (510(k))[19] Premarket Notification (510(k)) unless exempted (only a few Class II devices are exempt)[20] Premarket Approval (PMA)
Examples Dental floss, elastic bandages, patient examination glove[21] Condoms, medical lasers, endoscopes, and radiofrequency ablation systems[22] Cardiovascular permanent pacemaker electrode[23]

B. De novo Process and its Problem

The de novo classification process was introduced by the Food and Drug Administration Modernization Act of 1997 (FDAMA) and aimed to accelerate the FDA approval process for Class I and II devices. Before FDAMA, new types of devices not classified by the FDA were automatically assigned Class III.[24] As a result, they were automatically subject to PMA. But Section 513(f) of FDAMA created an alternative, accelerated route for devices of a new type by allowing manufacturers to submit a de novo request for the FDA to make a risk-based evaluation for classification of these devices into Class I or II. In order to submit a de novo petition, device manufacturers had to first show that their devices were of a new type and would have been automatically classified into Class III if they were to go through premarket notification. That is, in order to submit a de novo petition, a device must have first been found not substantially equivalent (NSE) to legally marketed devices via premarket notification.

But by requiring manufacturers to go through the premarket notification process as a preliminary step, FDAMA slowed down the approval process. On average, the FDA took 213 days to go through a 510(k) process, determine the device was NSE, and classify the device as Class III. Then, when the manufacturer filed a de novo petition to reclassify the device, the FDA took on average another 267 days to process the petition and reclassify the device as Class I or II.[25] In total, manufacturers that wanted marketing approval of their devices via a de novo route were expected to wait about 479 days before they could market their devices—longer than the traditional premarket approval process. Therefore, manufacturers were commercially justified in foregoing a de novo petition. Indeed, only 119 de novo petitions were received between 1998 and 2009.[26]

2. New de novo Process: “Direct” de novo Process

In order to resolve the administrative inefficiency of the de novo process, the Food and Drug Administration Safety and Innovation Act (FDASIA) of 2012 provided an alternative de novo pathway that removed prerequisite premarket notification and NSE determination. If a manufacturer believes that its device is classified into Class I or Class II and that there is no similar legally marketed device, it can directly submit a de novo petition for the FDA to make a risk-based classification into Class I or II. The Draft Guidelines published in August include these major changes in the de novo process made by the FDASIA. The Guidelines outline two steps for manufacturers: pre-submission and direct de novo.

A. Pre-submission

Pre-submission is a formal written request from an applicant for feedback from the FDA.[27] Pre-submission is not a required prerequisite for the de novo review, but is helpful for submitters as it provides submitters preliminary feedback from the FDA and facilitates the subsequent de novo review. In response to pre-submission, the FDA provides feedback on necessary regulatory controls to mitigate risks and on the submitted evidence of safety and effectiveness. The FDA strongly recommends pre-submission prior to a de novo review, especially for devices that have never been reviewed under 510(k).[28]

The Draft Guidelines list recommended content for pre-submissions.

  1. Device description
  2. Proposed intended use
  3. Previous submissions
  4. Proposed class (I or II)
  5. Proposed applicability of 510(k) requirement
  6. Searches of FDA public databases that submitters used to show that there are no predicate devices
  7. Risks associated with the device
  8. Information regarding the safety and effective of the device
  9. Protocols for performance and clinical testing
  10. Proposed mitigations and controls for the risks
  11. Specific questions on issues relevant to a planned de novo[29]

The comprehensive nature of the pre-submission review demonstrates the FDA’s intention to facilitate a de novo process by tackling important issues before the formal review process.

B. Direct de novo process

The direct de novo process no longer requires a preceding 510(k) or NSE determination. When the FDA determines that the device is Class III because it is not substantially equivalent to any legally marketed device, the manufacturer of the device can file a de novo review request, or request to reclassify the device into Class I or II based on its low to moderate risk profile.

The direct de novo process does not substitute the traditional de novo process.  The FDA may still make a NSE determination and recommend a manufacturer of a device under traditional 510(k) review to file a de novo review request when the agency concludes that the device presents a low to moderate risk (Class I or II) and that regulatory controls (i.e., general and special controls) provide reasonable assurance of safety and effectiveness.[30]

The de novo review, whether or not preceded by 510(k), consists of two main stages: classification review and substantive review. Classification review determines whether there is no existing legally marketed device of the same type. For a direct de novo (i.e., de novo without 510(k)), the submitter must show that there are no legally marketed devices to which a device under review is substantially equivalent. For a traditional de novo (i.e., de novo after 510(k) and NSE determination thereafter), the submitter must show that no device of the same type has legally entered the market since the NSE determination.[31]

Substantive review analyzes 1) whether the device under review poses a low to moderate risk and 2) whether regulatory controls (general or special) provide reasonable assurance of safety and effectiveness. The Draft Guidelines cite two main criteria that the substantive review takes into account. First, the device should meet the statutory standards for Class I or Class II devices. Second, the applicant should identify the risks and benefits of the device and the methods to effectively mitigate all known risks of the device.

Once the device passes both steps, the FDA issues a written order granting the de novo. The order also specifies the classification of the device (Class I or II) and its exemption status from premarket notification (i.e., future devices of the same type are exempt from 510(k)). Applicants can market the device immediately after they are granted a de novo.

Table 2. Timeline of the traditional de novo review and the direct de novo review[32]

Traditional de novo review

Direct de novo review

Filing for 510(k) review

Filing for direct de novo review

Pre-NSE 510(k) review

Direct de novo review

NSE determination

Filing for de novo review

(within 30 days after NSE determination)

Post-NSE de novo review

Device classification

(within 120 days after filing)

Device classification

(within 120 days after filing)

3. Examples of the Medical Devices Approved by the Direct De Novo Review


A. Dexcom Continuous Glucose Monitor Secondary Display

Dexcom is a San Diego, CA company that develops and manufactures continuous glucose monitoring systems for diabetes patients. The system consists of a sensor, transmitter, receiver, and mobile app. The sensor, which has a needle that is inserted through the skin, constantly monitors blood glucose level underneath the skin. The transmitter is directly attached to the sensor and wirelessly sends data to the receiver. The receiver (or the primary display device) receives the data from the transmitter and displays how the blood glucose level changes over time.[33] The mobile apps (or the secondary display device) fetch and display data from the receiver.[34]

Dexcom filed for premarket approval (PMA) for its G4 Platinum continuous glucose monitoring system in 2012. At that time, the continuous glucose monitoring system was composed only of the sensor, the transmitter, and the receiver and did not include the mobile app. Because the glucose monitoring system was not substantially equivalent to any legally U.S. marketed devices (i.e., predicate devices)[35], the continuous glucose monitoring system was evaluated by the FDA as a Class III medical device and received premarket approval in 2012.[36]

In 2014, Dexcom developed mobile apps that enable users to continuously monitor their blood glucose level via mobile devices and share their blood glucose level with their doctors in real time.[37] If Dexcom filed for 510(k) clearance of its mobile apps, the apps would have been classified as Class III devices because they are not substantially equivalent to any predicate devices.[38] However, the mobile apps, which passively display the data measured by and transmitted from the glucose monitoring system already approved by the FDA, would not pose a high enough risk to patients to require a premarket approval. Thus, Dexcom filed for the direct de novo classification of the mobile apps in December 15, 2014, and the FDA approved the mobile apps on January 23, 2015, less than one month after the filing.[39] In its letter to Dexcom, the FDA classified the mobile apps into class II under the generic name, “continuous glucose monitor secondary display.”[40] The category “continuous glucose monitor secondary display” is also exempt from the premarket notification requirements. This means that similar products manufactured by Dexcom or other companies can also benefit from the exemption once the FDA determines that the products are categorized as continuous glucose monitor secondary displays.

B. 23andMe Bloom Syndrome Carrier Status Test

23andMe is a Mountain View, CA, company that provides saliva-based direct-to-customer DNA test.[41] It discontinued the sale of its Saliva Collection Kit and Personal Genome Service after the FDA warned the company in 2013 that the Personal Genome Service is a medical device within the meaning of the FDCA that requires regulatory approval from the FDA in order to be marketable.[42]

In May 2014, 23andMe filed for 510(k) clearance petition for its Bloom Syndrome Carrier Status Test.[43] Bloom Syndrome is a rare autosomal recessive disorder characterized by short stature, a skin rash that develops after sun exposure, and a greatly increased risk of cancer.[44] 23andMe’s test assesses users’ carrier status for Bloom syndrome based on saliva collected from the users.[45] According to 23andMe’s press release, the FDA determined that the Bloom Syndrome Carrier Status Test did not have an applicable predicate device and converted it to a direct de novo request and classified the test as class II device in February 2015.[46] The test was categorized as a new class called “autosomal recessive carrier screening gene mutation detection system” under Section 866.5940.[47] Based on the FDA’s press release, the FDA even intends to exempt this class of genetic test from FDA premarket review.[48]

Since the reclassification order letter from the FDA to 23andMe is not disclosed, it is unclear why the FDA determined that genetic tests for autosomal recessive carrier screening generally pose a low to medium risk to users. However, from the definition of the classification, which set the intended users as adults of reproductive age,[49] and the FDA’s press release,[50] the FDA was aware of the risk of erroneous results when the FDA determined to classify the 23andMe’s genetic test as class II. One possible reason for the FDA’s decision may be that unlike genetic tests for breast cancer susceptibility, genetic tests for autosomal recessive carriers determine whether users are at risk of passing the autosomal recessive gene on to their children. Thus, while genetic tests for breast cancer directly affect decisions regarding medical interventions on users, genetic tests for autosomal recessive carrier screening do not directly impact the users’ safety and health.

4. Effects of the Direct De Novo Review Process on the Average Duration of the De Novo Pathway

 The FDA introduced direct de novo after it was found that the de novo process took even longer than premarket approval to complete.[51] Because the direct de novo process aims to substitute both NSE determination (510(k) review) and risk-based evaluation of devices (de novo review) in the traditional de novo process, the direct de novo is effectuating its purpose only if direct de novo review time is materially shorter than the sum of 510(k) review time and de novo review time during the traditional de novo process.

Statistical analyses were conducted on the 510(k) and de novo databases published by the FDA until May 14, 2015. Data for traditional 510(k) reviews, post-NSE de novo reviews, and direct de novo processes were extracted from online databases published by FDA.[52] However, information about pre-NSE 510(k) processes in the traditional de novo reviews could not be obtained from the FDA databases. Thus, data on the pre-NSE 510(k) processes were extracted from a privately-run online database of 510(k) applications.[53] The sample size of traditional 510(k) reviews, traditional de novo reviews, and direct de novo reviews was 52,492, 83, and 29, respectively. Nineteen traditional de novo reviews were excluded when calculating the average duration because the FDA databases had incomplete information about the reviews’ filing dates or decision dates.

Table 3. Average duration of 510(k) reviews


Traditional 510(k)

Traditional de novo

Direct de novo

Average Duration (days)




Standard Deviation (days)




Sample Size




Table 3 compares among average duration of three different 510(k) review processes. The differences between the average durations of the traditional 510(k) review and the traditional de novo process are statistically significant (P-value =  << 0.01). This shows that the average duration of 510(k) review increased to a statistically significant degree after the introduction of the traditional de novo process in 1997. In addition, the differences between the average durations of the traditional de novo review and the direct de novo process are calculated to be statistically significant (P-value =  << 0.01). In fact, the direct de novo processes are on average twice as fast as the traditional de novo processes. This supports that the direct de novo process significantly reduced the average duration of 510(k) reviews for novel types of medical devices.

Table 4. Average duration of traditional de novo reviews


Pre-NSE 510(k)

Post-NSE de novo


Average duration (days)




Standard Deviation (days)




Sample Size




Table 4 breaks down the traditional de novo review into pre-NSE 510(k) and post-NSE de novo, provides a deeper insight into the reason why the direct de novo process is much faster than the traditional de novo. Tables 1 and 2 both show that the average duration of the direct de novo process (263.3 days) is not much different from that of the post-NSE de novo process (266.5 days) (p-value = 0.47). In other words, the FDA is not processing direct de novo reviews faster than post-NSE reviews under the traditional de novo system. Instead, FDA reduces the duration of direct de novo reviews by removing the NSE determination altogether, which on average takes more than 200 days.

Because the pre-NSE 510(k) reviews determine nothing more than the fact that the device under review are unlike any legally marketed devices, the NSE determination does not make a substantive assessment of safety and effectiveness of the devices. Therefore, by removing the 510(k) procedure, the direct de novo review achieved both promptness and effectiveness.

5. Impacts of the Direct De Novo Review on Genetic Tests and Medical Software

Promptness of the direct de novo processes will have a positive impact on genetic tests and medical software. The risk of genetic tests and mobile apps can be high enough to warrant premarket approval if the impacts of misdiagnosing users are serious. Genetic tests for cancer susceptibility genes, for example, would require premarket approval by the FDA. However, by limiting the scope of the usage of genetic tests and mobile apps, the FDA can control the risk of misdiagnosis to users, as seen in the cases of Dexcom mobile apps and 23andMe’s carrier screening test. Then, the benefits to the public from the faster market introduction of genetic tests and medical software via direct de novo reviews will outweigh the risks posed by them.

Disease diagnostic tests that are based on DNA sequences of patients had been automatically classified as Class III because there was no appropriate product classification code for such genetic tests. Before the introduction of the direct de novo review process, companies that developed genetic tests faced a business risk that FDA might not approve the genetic tests after the companies wait at least a year (via premarket approval, traditional 510(k), or traditional de novo) without garnering revenues from the genetic tests. Because the viability of such companies almost completely depends on the FDA market approval, the long waiting period and the uncertain nature of the de novo process seriously eroded the viability of genetic test companies. However, as direct de novo processes are shown to be an effective and prompt process, it is expected that more companies that develop various genetic tests will use the direct de novo pathway to obtain FDA market approval. In addition, more capital is expected to flow into the disease diagnostics industry, since one of the major risks of the industry (complicated and long FDA approval procedure) is significantly reduced by the introduction of the direct de novo process.

The direct de novo process will also spur the development of medical software for existing medical devices. As seen in the example of Dexcom, mobile apps that passively display data transmitted from medical devices already approved by the FDA do not pose greater risks to users than the existing medical devices from which the mobile apps are receiving data. By allowing medical software reach market earlier via the direct de novo pathway, the FDA will provide the public with benefits of medical software. Mobile apps combined with medical devices are expected to improve the quality of healthcare and communications between patients and healthcare providers.

Some benefits of the direct de novo review to the development of medical software are demonstrated by the FDA’s approval of Dexcom’s Cloud Data Management Software in 2014.[54] The Cloud Data Management Software was intended to assist people with diabetes and their healthcare professionals in the review, analysis, and evaluation of historical glucose measurement data to support effective diabetes management.[55] After the direct de novo pathway that took less than four months, Dexcom’s software was categorized as class I under the name “continuous glucose monitor data management system.”[56] The FDA explicitly limited the use of the software as an aid in diabetes management to minimize any unexpected risk of the software to users, as the FDA did when it classified the Dexcom’s mobile apps as class II.[57] The direct de novo pathway will facilitate the FDA review process of other medical software as it did for Dexcom’s data management software and mobile apps. As long as the FDA is well aware of the risks of medical software under review and limits the scope of the usage, the direct de novo pathway will serve its original purpose of streamlining the classification procedure of novel low-to-moderate risk devices without compromising the public health.

6. Conclusion

The direct de novo process cut down the duration of the FDA de novo review process by about 50%, achieving its original purpose of accelerating the FDA review process for new types of devices that pose only low to moderate risks. By eliminating the preceding 510(k) determination, which adds little to the understanding of effectiveness and safety of devices, the direct de novo process achieves both speed and efficacy. Also, as shown in the examples of 23andMe and Dexcom, the disease diagnostics industry and the health IT field companies that develop genetic tests will likely benefit from the direct de novo process. Companies in the diagnostics and health IT fields can use the direct de novo process to reduce the duration for the FDA approval evaluation and thus reduce business uncertainty related to such process. The official Guideline for the de novo process, which will be published in near future after FDA gathers comments on the Draft Guideline, would further clarify the de novo process, enable applicants to prepare better applications, and facilitate the overall de novo process.