PLoS ONE plos plosone PLOS ONE 1932-6203 Public Library of Science San Francisco, CA USA 10.1371/journal.pone.0283442 PONE-D-22-04558 Research Article Biology and life sciencesBioengineeringBiotechnologyMedical devices and equipment Engineering and technologyBioengineeringBiotechnologyMedical devices and equipment Medicine and health sciencesMedical devices and equipment Medicine and health sciencesSurgical and invasive medical proceduresLaparoscopy Medicine and health sciencesSurgical and invasive medical proceduresEndoscopyEndoscopic surgery Medicine and health sciencesSurgical and invasive medical procedures Medicine and health sciencesEpidemiologyMedical risk factors Engineering and technologyMechanical engineeringRobotics Medicine and health sciencesSurgical and invasive medical proceduresEndoscopyEndoscopic surgeryEndoscopic plastic surgery Medicine and health sciencesSurgical and invasive medical proceduresPlastic surgery and reconstructive techniquesEndoscopic plastic surgery Science policyTechnology regulations Identification of predicate creep under the 510(k) process: A case study of a robotic surgical device Identification of Predicate Creep Lefkovich Charlotte Conceptualization Formal analysis Writing – original draft 1 ¤ https://orcid.org/0000-0002-0218-3248 Rothenberg Sandra Supervision Writing – review & editing 2 * Department of Public Policy, Rochester Institute of Technology, Rochester, New York, United States of America Department of Management and Public Policy, Rochester Institute of Technology, Rochester, New York, United States of America Wang Quanzeng Editor FDA: US Food and Drug Administration, UNITED STATES

The authors have declared that no competing interests exist.

Current address: Southington, Connecticut, United States of America

 * E-mail: slrbbu@rit.edu 28 3 2023 2023 18 3 e0283442 14 2 2022 9 3 2023 2023 Lefkovich, Rothenberg This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

The FDA’s 510(k) process for medical devices is based on “substantial equivalence” to devices clearedpre-1976 or legally marketed thereafter, known as predicate devices. In the last decade, several high-profile device recalls have drawn attention to this regulatory clearance process and researchers have raised questions about the validity of the 510(k) process as a broad clearance mechanism. One of the issues raised is the risk of predicate creep, a cycle of technology change through repeated clearance of devices based on predicates with slightly different technological characteristics, such as materials and power sources, or have indications for different anatomical sites. This paper proposes a new way to identify potential “predicate creep” through the use of product codes and regulatory classifications. We test this method by applying it to a case study of a Robotic Assisted Surgery (RAS) device, the Intuitive Surgical Da Vinci Si Surgical System. We find that there is evidence of predicate creep using our method, and discuss implications of this method for research and policy.

 The author(s) received no specific funding for this work. Data Availability All 510K summary files used in this research are available from the 510K database at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm Introduction

The United States Food and Drug Administration’s [FDA’s] 510(k) process for medical devices is based on “substantial equivalence” to devices cleared pre-1976 or legally marketed thereafter, known as predicate devices. In the last decade, several high-profile device recalls have drawn attention to this regulatory clearance process and researchers have raised concerns about the validity of the 510(k) process as a broad clearance mechanism. These concerns include, but are not limited to, a lack of a clear definition of “intended use,” using recalled devices as predicates, lack of publicly available data regarding predicates and determinations of substantial equivalence, equating substantial equivalence to safety, and a lack of sufficient non-clinical data to support claims of substantial equivalence, among others. As outlined in the literature review below, there is a body of research that documents these problems and the resulting safety issues that have stemmed from them.

One of the specific concerns in the 510(k) process is the risk of predicate creep, a cycle of technology change through repeated clearance of devices based on predicates with slightly different technological characteristics, such as materials and power sources, or have indications for different anatomical sites [1, 2]. Predicate creep has been identified as one of the causes of some well-known medical device failure, such as the clearance of metal on metal hip implants, surgical meshes, and the ReGen Menaflex collagen scaffold, where what might seem as small differences between a device and its predicate were enlarged over time [1, 3, 4].

While 510 (k) process is intended to allow some level of innovation and thus technical change in new devices, the challenge is to identify when these changes are not “substantially equivalent” or are changes that may lead to safety issues. One of the most direct methods for identifying predicate creep is by making technical comparisons across a predicate ancestry tree. While perhaps the most accurate, this method is time consuming and is limited by the availability of data on the technical characteristics of the products. This paper proposes using product codes and regulatory classifications as an alternate means to identify predicate creep. The methods will be tested on a Da Vinci Si Surgical System, a robotic surgical platform. In this paper we will first review the existing research on concerns with the 510(k) process and research on predicate creep. Then, we review our methods to gather and analyze predicate data. We illustrate the use of this method on the Da Vinci Si Surgical System. We conclude with a discussion of the implications for our findings for policy and future research.

 Literature review

Many of the critiques of the 510(k) process relate to the notion of substantial equivalence. First, due to the lack of a clear official definition for the key terms “indications for use” and “intended use”, the FDA has allowed permissive interpretation of these terms by applicants and inconsistent use of the terms across reviewers [5, 6]. Over time, this has resulted in the clearance of significantly altered devices, or even novel devices, as substantially equivalent to established predicates [5–7].

Second, the 510(k) process makes the implicit assumption that substantial equivalence means that a device is safe and effective, and that the predicates on which substantial equivalence determinations are based are safe. Substantial equivalence, however, only supports an assessment that the device introduces no new safety hazards and functions at least as effectively as the predicate device [8]. Moreover, if the predicate device poses risk or is ineffective, then the new device may perpetuate these flaws. It is not uncommon for a device to cite a recalled predicate [9–11], with one estimate being about 4.3% [11]. There is disagreement, however, as to whether this aspect of the process is invalid entirely or only for specific types of high risk or technologically complex devices [3, 9, 12].

Third, there are concerns regarding the use of multiple predicate devices for a single substantial equivalence determination. While most academics agree that single predicate submissions, which directly compare a new device to a single device on the market, provide significant assurance of safety and efficacy in most cases, many have raised questions about whether multiple and split predicates can provide the same level of assurance [2–4, 9].

Many of these concerns stem from an overarching issue that substantial equivalence allows a device to “piggyback” on the reasonable assurance of safety from existing predicate devices without undergoing independent testing [5, 12, 13]. These predicates can have also been cleared via substantial equivalence, creating a gap between the current device and the most recent device for which actual scientific evidence of safety was provided [14]. It is possible that, over multiple cycles of small device modifications and subsequent substantial equivalence findings, a new device may be cleared which is significantly dissimilar to the original predicate for which scientific evidence exists [7]. This process is known as predicate creep [7, 9] and may lead to a device being introduced to the market which bears no resemblance to the original device.

Within the 510(k) process, requirements for non-clinical testing are used to mitigate the risk associated with small scale predicate creep, which in most cases works effectively. However, past research on predicate networks has highlighted the lack of publicly available non-clinical scientific data in substantial equivalence determinations. Both Zuckerman et al. [9] and Liebeskind et al. [15] used the FDA 510(k) Database and/or FOIA requests to trace the predicate history of different medical devices and found that there was a dearth of publicly available scientific data to support the claim of substantial equivalence. Liebeskind et al. [15], in particular, traced the predicate ancestry of robotic surgical systems, and found that only 19 (7.3%) 510(k) clearances provided clinical data, while 73 (27.9%) did not submit any supporting data. These studies suggest that non-clinical evidence to support substantial equivalence claims may be insufficient to ensure safety when there is predicate creep.

Past research on predicate creep primarily involves creating predicate ancestry trees and using direct technical comparisons of devices and their predicates. Such studies have identified issues concerning predicate creep in surgical meshes [4] and Pathwork Tissue of Origin Test [7], DePuy ASR XL Acetabular Cup System [3], power morcellators [16], among others. Ardaugh et al. [3], for example, studied the metal-on-metal hip implant over five decades with the purpose of identifying the cause of safety flaws present in the design. They identify that a unique combination of three characteristics in the ASR XL were approved through substantial equivalence of “split predicates,” where you compare characteristics to different predicate devices. Since all three devices were deemed safe based on clinical trials and safe use of predicates in the consumer market, and the XL simply combined parts of the devices, it was placed on the market without undergoing clinical testing and the resulting product had to be recalled for particle shedding [3].

Direct technical comparisons can be challenging due to a lack of information, as discussed above, or the cumbersome size of predicate ancestry trees, particularly for newer devices. Using FOIA requests to attain more detailed information can also be onerous. This paper proposes another way to identify potentially important instances of predicate creep in a large predicate network through analysis of product classes and regulatory categorizations. We test this method on a well-known medical device, the Intuitive Surgical Da Vinci Si robotic surgical system.

 Method

The Da Vinci Si Surgical System is a robotic-based laparoscopic surgical tool initially approved by the FDA in 2000 which replaces a surgeon’s hands with robotic arms for more precise control and motion [17]. While Intuitive has subsequently brought multiple iterations of the Da Vinci to market, it remains the only full RAS platform on the market as competitors struggle to develop a viable competitor around Intuitive’s strong patent foothold [18]. The Da Vinci itself was initially approved under a 510(k) application based on a complex web of component-level substantial equivalence, most likely supplemented by additional testing. The Da Vinci Si was chosen because its function is well documented. Moreover, there is a non-negligible number of malfunctions that have occurred during the use of these types of devices [19].

Information on the devices within the Da Vinci Si Model predicate network collected through the 510(k) clearance database. We also searched other databases using the search function on the FDA website, which returns results from all FDA publications, including database information, conference presentations, regulations, and internal memos. Although this method returns significantly more results, the search function offers limited filtering options and requires manual sorting to determine whether results are relevant. Information required for determination of substantial equivalence, such as predicate devices, intended use, indications for use, and scientific evidence may be presented in the summary page. The device summary page also includes the product classification code, which identifies a more device-specific classification based on the technological characteristics and intended use of a device. This can be used to identify potential predicate devices based on the substantial equivalence parameters for a new device. From this information, we created a database of devices that included devices that were predicates of or predicated on the device; we included information on the K#, device name, clearance date, product code, and any predicate or intended use information available for all devices relevant to the database construction parameter.

Constructing predicate networks

The numbers of predicates in the Da Vinci Si predicate ancestry network made traditional tree diagrams too unwieldy, so an alternate diagram structure known as a network map was used to display predicate relationships within the clearance ancestry. After mapping the overall predicate network, we identify patterns in the use of existing mechanisms or structures which identify characteristics of the technology from a regulatory perspective. FDA product codes are particularly useful for this purpose, as they are designed to identify groups of devices with the same intended use and technological characteristics. Since possessing the same intended use is a requirement for clearance via substantial equivalence, it is likely that any device approved via 510(k) with a different product code that the predicate is of particular interest. The introduction of new product codes in the predicate ancestry tree should be indicative of the introduction of new technological characteristics or uses. While the device identification key is a set of functions or characteristics which distinguish devices in that product classification, we also looked at the regulatory description, which reflects the primary function or intended use as identified by the FDA.

 Data analysis

The network map for the Da Vinci Si is shown in Fig 1. Within this network mapping diagram each device is represented by a dot, with the dot size increasing based on the total number of substantial equivalence relationships that device is based on. Substantial equivalence relationships (also known as predicate-subject device relationships) are represented by lines drawn between the two devices involved. Each dot is labeled with the K# of the device which it corresponds to. The trace begins with the main subject device (i.e. the Da Vinci Si) on the left side of the trace, and advances toward the left, with each line originating at a subject device and terminating at its respective predicate. Therefore, the oldest devices present in the trace are located on the right side, although vertical alignment does not correlate exactly to clearance date, and the newest devices are to the left of the trace.

10.1371/journal.pone.0283442.g001 Expanded Da Vinci Si predicate trace.

This network map included 2618 device instances (i.e. a device is cited as a predicate), with a total of 50 unique devices. The unique devices within this trace are classified under a total of 15 different product codes, with the majority of devices, including the various Da Vinci models, categorized under code NAY. Additional information about each device, including the manufacturer, clearance date, and any recalls issued, can be found in Table 1.

10.1371/journal.pone.0283442.t001 Da Vinci Si predicates.
K Number Device Name Manufacturer Clearance Date Product Code Recalls
K081137 Intuitive Surgical Da Vinci Si Surgical System: Model Is3000 INTUITIVE SURGICAL, INC. 2/18/2009 NAY 24 –Class II
K063220 Da Vinci S Surgical System-V1.1, Model Is2000 INTUITIVE SURGICAL, INC 12/1/2006 NAY 4 –Class II
K050802 Modification To Intuitive Surgical Da Vinci Surgical System And Endoscopic Instruments INTUITIVE SURGICAL, INC. 6/29/2005 NAY 0
K050369 Intuitive Surgical Da Vinci Surgical System, Model Is2000 INTUITIVE SURGICAL, INC. 4/29/2005 NAY 43 –Class II
K050404 Intuitive Surgical Da Vinci Surgical System And Endoscopic Instruments INTUITIVE SURGICAL, INC. 4/21/2005 HET 0
K043288 Modification To Intuitive Surgical Da Vinci Surgical System And Endoscopic Instruments INTUITIVE SURGICAL, INC 3/3/2005 NAY 0
K050005 Intuitive Surgical Monopolar Curved Scissors, Model 400179; Tip Cover Accessory, Model 400180 INTUITIVE SURGICAL, INC. 1/25/2005 NAY 2 –Class II
K043153 Intuitive Surgical Da Vinci Surgical System And Endoscopic Instruments, Models Is1200 & Is1000 INTUITIVE SURGICAL, INC. 12/15/2004 NAY 0
K042855 Intuitive Surgical Harmonic Curved Shears Instrument INTUITIVE SURGICAL, INC. 11/12/2004 NAY 0
K041340 Guidant Microwave Ablation System GUIDANT CORPORATION, CARDIAC SURGERY 7/28/2004 NEYOCL 0
K040237 Intuitive Surgical Da Vinci Endoscopic Instrument Control System And Endoscopic Instruments INTUITIVE SURGICAL, INC. 7/7/2004 NAY 0
K040948 Intuitive Surgical Endopass Endoscopic Delivery Instrument, Model P/N 400170 INTUITIVE SURGICAL, INC. 5/5/2004 NAY 0
K022574 Intuitive Surgical Endoscopic Instrument Control System & Endoscopic Instruments, Model Da Vinci ISI 1000/1200 INTUITIVE SURGICAL, INC. 11/12/2002 NAY 0
K021036 Intuitive Surgical Da Vinci Surgical System, Model IS1200 INTUITIVE SURGICAL, INC. 6/26/2002 NAY 18 –Class II
K013946 Flex 10 Accessory For The Afx Microwave Ablation System AFX, INC. 2/27/2002 NEYOCL 0
K013416 Intuitive Surgical Endowrist Endoscopic Instrument Family INTUITIVE SURGICAL, INC. 1/10/2002 GEI 0
K012833 Intuitive Surgical Bipolar Forceps INTUITIVE SURGICAL, INC 11/16/2001 NAY 4 –Class II
K011281 Intuitive Surgical Ultrasonic Shears INTUITIVE SURGICAL, INC. 7/24/2001 NAYLFL 0
K011002 Intuitive Surgical Da Vinci Surgical System, Model Isi 1000 INTUITIVE SURGICAL, INC. 5/30/2001 NAY 0
K003978 Afx Microwave Generator, Flex Ablation Wand, Lynx Ablation Wand, Model Series 1000, P/N 102006, P/N 102007 AFX, INC. 5/22/2001 NEYOCL 0
K002489 Intuitive Surgical Da Vinci Endoscopic Control System INTUITIVE SURGICAL, INC. 3/2/2001 NAY 0
K990144 Intuitive Surgical Endoscopic Instruments, Intuitive Surgical Endoscopic Instrument Control System INTUITIVE SURGICAL, INC. 7/11/2000 NAY 3 –Class II
K993054 Ultracision Harmonic Scalpel Coagulating Shears, Models Lcs-C5, Lcs-C1, Cs-23c, Cs-231, Cs-14c, Cs-141 ETHICON ENDO-SURGERY, INC. 12/9/1999 LFL 0
K991859 Dexide Bipolar Forceps Ii ** Device UNITED STATES SURGICAL, A DIVISION OF TYCO HEALTHC 6/23/1999 GEI 0
K980099 Ultracision Laparosonic Coagulating Shears (Lcs-5(Lcsk5 And Lcsb5)) ETHICON ENDO-SURGERY, INC. 4/9/1998 LFL 0
K974320 Cryogen Cardiac Cryosurgical System CRYOGEN, INC. 2/3/1998 OCL 0
K972662 Cryogen Cryosurgical System CRYOGEN, INC. 10/1/1997 GEH 0
K972415 Minisite*Bipolar Forceps** Device UNITED STATES SURGICAL, A DIVISION OF TYCO HEALTHC 9/19/1997 GEI 0
K965001 Intuitive Surgical Monarch Laparoscopic Manipulator INTUITIVE SURGICAL, INC. 7/31/1997 GCJ 0
K971861 Ultrasonic Hand Instruments UNITED STATES SURGICAL, A DIVISION OF TYCO HEALTHC 7/1/1997 LFL 0
K970496 Heartport Maze System: Cryoprobe Set HEARTPORT, INC. 5/9/1997 OCL 0
K964971 Cryogen Cryosurgical System CRYOGEN, INC. 3/28/1997 GEH 0
K960400 Diamond-Touch And Micro Diamond-Touch Instruments/Diamond-Line Instruments/Diamond-Port (Access Parts) SNOWDEN-PENCER 3/12/1996 FBM 0
GCJ
GEI
K953637 CMS Accuprobe 550/530 CRYOMEDICAL SCIENCES, INC. 12/4/1995 GEH 0
K953059 Kittner Dissector MEDICAL PERSPECTIVES CORP. 9/14/1995 GDY 0
K930666 Reusable Laparoscopic Instruments W/ Electrocautery SNOWDEN-PENCER 5/19/1994 GEI 0
K930667 Reusable Laparoscopic Instruments SNOWDEN-PENCER 5/16/1994 GCJ 0
K933169 Inman Endoscopic Blunt Dissector INMAN MEDICAL CORP. 4/19/1994 GCJ 0
K936308 Endex Endoscopic Positioning System ANDRONIC DEVICES, LTD. 3/31/1994 FQO 0
K931783 AESOP (Automated Endoscopic System For Optimal Positioning) COMPUTER MOTION, INC 11/22/1993 GCJ 0
K931340 Grasp Forceps/Scissors/Needle Holder/Dissector BAXTER HEALTHCARE CORP. 7/1/1993 GCS 0
K925699 Harmonic Scalpel Laparosonic Clamp Coagulator Acc. ULTRACISION, INC. 5/17/1993 GCJ 0
K914190 Auto Suture(R) Endoscopic Fan Retractor UNITED STATES SURGICAL, A DIVISION OF TYCO HEALTHC 5/6/1992 GAD 0
K912544 Bipolar Forceps EVEREST MEDICAL CORP. 6/24/1991 MAV 0
K904421 CMS Oncoprobe CRYOMEDICAL SCIENCES, INC. 4/8/1991 GEH 0
K882568 130 Cryo Unit & Assoc. Cryoprobes & Spray SPEMBLY MEDICAL LTD. 9/27/1988 GEH 0
K874367 Various Cardiac Cryoprobes Having Dia. & Cos. Diff SPEMBLY MEDICAL LTD. 1/4/1988 HQO 0
K811390 Ccs100 Cryosurgical System FRIGITRONICS OF CONNECTICUT, INC. 6/16/1981 GEH 0

A list of the codes, the device identification key, and the regulatory description for each code is shown in Table 2 for the Da Vinci Si. The device identification key is a set of functions or characteristics which distinguish devices in that product classification, while the regulatory description is the primary function or intended use as identified by the FDA.

10.1371/journal.pone.0283442.t002 Product codes in Da Vinci Si trace.
Code # Devices Device Identification Regulatory Description
FBM 1 Cannula and Trocar, Suprapubic, Non-Disposable Suprapubic urological catheter and accessories
NEY 3 System, Ablation, Microwave and Accessories Electrosurgical cutting and coagulation device and accessories
NAY 17 System, Surgical, Computer Controlled Instrument Endoscope and accessories
HET 1 Laparoscope, Gynecologic (And Accessories) Gynecologic laparoscope and accessories
GEI 5 Electrosurgical, Cutting & Coagulation & Accessories Electrosurgical cutting and coagulation device and accessories
LFL 4 Instrument, Ultrasonic Surgical N/A
OCL 5 Surgical Device, For Cutting, Coagulation, And/Or Ablation of Tissue, Including Cardiac Tissue Electrosurgical cutting and coagulation device and accessories
GEH 6 Unit, Cryosurgical, Accessories Cryosurgical unit and accessories
GCJ 6 Laparoscope, General & Plastic Surgery Endoscope and accessories
GDY 1 Gauze/Sponge, Internal, X-Ray Detectable Nonabsorbable gauze for internal use
FQO 1 Table, Operating-Room, Ac-Powered Operating tables and accessories and operating chairs and accessories
GCS 1 Endoscope, Battery-Powered and Accessories Endoscope and accessories
GAD 1 Retractor Manual surgical instrument for general use
MAV 1 Syringe, Balloon Inflation Angiographic injector and syringe
HQO 1 Unit, Cautery, Thermal, Ac-Powered Thermal cautery unit

A breakdown of the devices present in the trace color-coded by product code is pictured in Fig 2 below. The most prevalent code is NAY for computer controlled surgical systems. The next two most prevalent codes are GEH for cryosurgical units and GCJ for general laparoscopic surgery. Color-coding devices by code highlights common device functions within the trace and allows for easy identification of technical characteristics as they evolve through predicate generations.

10.1371/journal.pone.0283442.g002 Da Vinci Si trace color-coded by product classification code.

Looking at the tree breakdown by product code in Fig 2, we can see the progressive evolution of devices from more general laparoscopic surgical tools (to the right) to the more technologically complex computer-controlled system of the Da Vinci Si. The five distinct branches emerging from the intertwined trace center are each dominated by one or two distinct product codes, while the devices within the central web belong almost exclusively to the same product classification as the Da Vinci Si. This implies that the technological characteristics the FDA uses to identify devices belonging to code NAY are a combination of the characteristics present in each distinct predicate group. This illustrates how larger “jumps” in technological complexity of new devices can occur through the 510(k) process, by combining the characteristics of multiple well-understood devices into a new type of device.

Another perspective to examine the technological evolution within the ancestral trace is based on the regulatory description rather than the product code. While the product code considers both the intended use and specific characteristics of a device, the regulatory description is a broader, more general description based on the function of the device as defined by the FDA. For example, the specific device description for product code GCJ is “laparoscope, general and plastic surgery,” which specifies both a particular type of device and use, while the regulatory description “endoscope and accessories” specifies only a general classification of devices. Because of these broader descriptions, there is often overlap between the regulatory descriptions of different product codes. In this trace, devices from 15 product codes can be placed into 11 groups based on regulatory descriptions, resulting in the formation of two larger groups which contain the majority of devices within the trace.

The regulatory definition breakdown reveals that approximately half of the devices (24) within the trace are classified from a regulatory perspective as endoscopes and accessories. A further ~25% (13 total) are classified as electrosurgical cutting and coagulation devices, devices which use a high frequency electrical current to perform surgical operations. 6 of the devices are cryosurgical units, all classified under product code GEH, with the remaining devices representing a wide variety of functions and characteristics.

Unlike a grouping by product code, which highlights the evolution of specific technical characteristics over time as new codes are introduced to the trace, viewing the trace based on the regulatory description highlights groups of predicates based on general device functions. For example, in the product code trace the upper branch consists of four distinct product codes. However, when color-coded by regulatory description it becomes apparent that the originating predicate (branch tips) are all cryosurgical units with accessories, which serve as predicates to the electrosurgical cutting and coagulation devices that make up the middle of the branch, which in turn serve as predicates for the computer controlled surgical devices present in the web center. Fig 3 shows a direct comparison of devices in the upper Da Vinci Si predicate branch, as shown in Fig 2. We can see from this comparison that while color-coding by the product code specifically identifies groups of devices which the FDA considers similar enough to be substantially equivalent, devices with the same intended use and technological characteristics, color-coding by regulatory description makes it easier to trace the general progression of technology over time based on the general function of these devices. The new central group present in the regulatory description trace combines two product codes to create a more general device grouping.

10.1371/journal.pone.0283442.g003 Direct comparison of devices in upper Da Vinci Si predicate branch.

Fig 4 shows the devices in the entire DaVinci Si Trace color-coded by regulatory description. The teal nodes and lines represent devices approved as “endoscopic instruments and control systems” under code NAY. This style of coding based on regulatory description also draws attention to predicate devices with unusual functional characteristics that do not fit with the primary function of most devices within the trace. In some cases, this may be an indication of an unnecessary or ineffective predicate relationship, while in others it may be indicative of secondary device functions. For example, in this trace there is a device identified as a non-absorbable gauze/sponge for internal use. This device, the Medical Perspectives Kittner Dissector, is a sponge used during surgical procedures to prevent bleeding. It was identified as a predicate of the Monarch Laparoscopic Manipulator system, which included customized versions of many basic surgical instruments such as gauze. Although there is a purpose for including this device as a predicate for a custom surgical tool, this part of the system is secondary to the main function of the Da Vinci Si as a computer controlled surgical system.

10.1371/journal.pone.0283442.g004 Da Vinci Si trace with devices color-coded by regulatory description.

Sorting predicates by regulatory description appears to allow for easy identification of both primary and secondary device functions, based on the prevalence and location of a given function in the trace. Primary device functions would be those identified directly by the regulatory description of the given device, while secondary functions would be functions present in predicate devices but absent in the regulatory description of the subject device. The more prevalent a function is in the predicate history, the more likely it is to be present in the subject device in at least a secondary capacity. Additionally, the significance of a particular secondary function to the overall function of the device appears to correspond to the number of predicates with that secondary function identified as a primary function. For example, the second most prevalent regulatory description in the Da Vinci Si trace is “electrosurgical cutting and coagulation device and accessories,” which corresponds to the function of an essential component of the Da Vinci Si system. Although this is no longer listed as a primary function for the Da Vinci Si, it is an important component of the system.

 Discussion and conclusion

Given the purpose of the 510(k) process is to encourage incremental innovation in the medical device market by reducing regulatory barriers, while ensuring safety, it should be expected that there is a level of technology creep inherent in the process. If manufacturers are limited to only submitting identical devices for clearance through the 510(k) process, there can be no innovation. Even in new versions of existing devices submitted by the same manufacturer, such as the Da Vinci Models IS1000 and IS1200, there can be somewhat significant technological changes. Removing all technology creep from the process would only hinder progress within the medical devices industry. Thus, predicate creep in cases where the new device can be reasonably assured safe based on available scientific evidence, is beneficial to companies, patients, and regulators.

This research focused specifically on examining the 510(k) clearance process as it was applied to Robotic Assisted Surgical systems, an emerging technology with a high degree of technical complexity, with specific focus on the Intuitive Da Vinci Si Surgical System. The objective of this research was to examine the predicate history of these devices and assess whether significant predicate creep by tracking changes and patterns in product codes and regulatory classifications.

The major contribution of this research to the discussion surrounding the 510(k) clearance process is the use of product codes and regulatory categorizations in predicate ancestry networks as a means to identify potentially problematic predicate creep. This method allowed us to see “jumps” in technological complexity of devices new devices made by combining the characteristics of multiple well-understood devices into a new type of device, such as with the NAY product code. We were also able to see the absorption of a primary predicate device function into a secondary system function, which is an indication of increasing complexity of devices over time, where the Da Vinci Si represents a particularly large leap in complexity. While the five main predicate branch groupings each generally contain one or two primary device functions that evolve and become more complex over time, the Da Vinci Si suddenly combines all those functions together into a single device where none of the functions serve as the primary function. The listed primary function “endoscope and accessories,” is a very generalized term which only identifies the device as one used for internal imaging, even though it includes all the other secondary functions derived from the predicate devices, and is in fact used to directly perform surgeries.

Limitations

This study was limited by several factors. First, the investigations performed in this study were based solely on data available publicly through FDA databases. Predicate devices are identified within the database only when application summaries are provided, which was not required for inclusion until the mid-2000’s. This significantly limits the number of devices with traceable predicate histories. Moreover, the level of information contained within these summaries varies widely, and in many cases these summaries do not exist at all, due to the gradual evolution of requirements for 510(k) application submissions. We did consider requesting additional data through FOIA, but decided against this route given that it would add a significant amount of time to the data gathering, with an FDA estimate of 18 to 24 months [20], and there was a precedent in prior research to rely primarily on information provided in the FDA website [9]. Additionally, given we are looking backwards in time, the fact that some predicates are missing actually makes any finding of predicate creep more significant.

Another limitation is that the information available through the FDA databases is usually from general device summaries, which are written to protect intellectual property rights, including minimal details about the specific technological characteristics of each device and the evidence provided to support equivalence claims. Thus, without the information to correlate findings to actual technological characteristics and the evidence used to support their existence, it is difficult to evaluate the significance of findings. Further, there is a lack of information regarding decision making of FDA officials during the review process creates a knowledge gap, which may explain why even large-scale technological creep was allowed through this accelerated regulatory processes. Lastly, we did not determine if the product classification code was made by the manufacturer or the FDA, or if the code change was driven by an administrative need or a technical feature.

Another limiting factor which may impact the significance of conclusions is the choice to limit the scope of this research to robotic surgical devices. Conclusions drawn from this research about patterns present in the larger medical device industry may be biased by practices specific to the regulation of robotic surgical devices. In particular, the technical complexity of robotic surgical devices may lead to a higher number of predicate devices than would be present in less-complex devices. The high number of predicates per generation in this investigation was determined to be partially responsible for the high rate of predicate creep, so other types of devices may not have such significant predicate creep. Moreover, given that for many years there was little or no competition to the Da Vinci Surgical Systems, we chose a device category for which there are likely to be fewer products available to cite as predicates than for products with higher levels of market competition.

Lastly, the device investigated in this study was cleared in 2009. In 2013, the FDA completed a small sample survey to better understand the causes behind adverse events in the Da Vinci Surgical system [21]. In 2014, the FDA took steps to change the review of robotically assisted surgical devices [22]. Thus, our review does not take the evolution of regulation that has happened since 2009. While this does not limit the use of this method to investigate predicate creep, it may impact how patterns are interpreted over time. It would be interesting to compare how the predicate network changes with newer models of the Da Vinci that may have been cleared under different processes and standards.

 Implications Implications for research

This investigation identified the presence of technological creep in predicate histories and made observations about common regulatory patterns and practices. However, the choice to focus this investigation on robotic surgical systems, a device type known for its technical complexity, may not be generalizable to other medical devices. Exploring the predicate history of additional device types would allow for comparison of regulatory patterns across device types and validate the conclusions of this investigation.

It might also be interesting to compare predicate history with patent history. While the 510(k) process purports to identify predicate devices based on substantial equivalence in both functionality and technology, the patent process in the US requires proof that an idea is new and novel to secure a patent. Examination of the patent literature for devices which appear in the ancestral equivalence tree constructed from the regulatory history might allow for identification of new technological characteristics introduced in each device. Theoretically, larger technological “leaps” present in substantial equivalence trees should correspond to a stronger patent presence for a given device.

 Implications for policy

Looking at the regulatory process from an outside perspective, it is clear that predicate creep exists inherent within the substantial equivalence process. When this technology creep occurs on a small scale, introducing a new technology feature or application which slightly alters the form or function of the device while preserving the overall function and technological characteristics of the predicate, there is little potential impact to the safety of the public. In fact, purposeful inclusion of small amounts of technology creep is necessary to allow for innovation and improvement in medical device design. However, the effects of predicate creep over time have allowed for the development and clearance of entirely new devices without undergoing the stricter PMA approval process. Because this snowballing effect is directly dependent on small-scale predicate creep, it may be difficult to address the problem without negatively impacting the ability of manufacturers to bring innovative devices to market.

There have been calls for the FDA to develop a comprehensive, easily accessible database of predicate relationships. A complete map, including product codes and regulatory classifications would reveal common patterns in predicate relationships. The method used in this paper could then be used to identify critical points where a new device may be significantly different from the closest predicate in technology or intended use. At these critical points the FDA can then require additional scientific evidence of the overall device function, such as a small clinical trial, to ensure that no safety flaws have been introduced in the device due to predicate creep. This type of map would also help identify devices predicated on recalled devices or devices with known regulatory problems. This would allow for the continued use of small-scale predicate creep for technical innovation, while mitigating the introduction of untested technical characteristics and potential safety flaws over time.

Another important issue identified through this research is not the presence of predicate creep over time, but rather the sudden introduction of devices with high levels of technical complexity into the market through the 510(k) process. Starting in 2014, the FDA took steps to change the review of robotically assisted surgical devices [21]. Our review does not take these changes into account. They have also made changes that try to address the presence of extreme technology creep, so-called “leap” devices, in the regulatory process by creating more stringent guidelines for identification of predicate devices. These new guidelines reject the use of split predicates, where a device identifies one predicate for intended use and a separate predicate for technological characteristics. Interestingly, most “leap” devices identified in this research were not cleared using split predicates, but rather multiple predicates composed of “step” devices specifically designed to advance the desired use case for a particular technology.

It is possible that companies could take advantage of the substantial equivalence process by creating “step” devices, which are submitted and cleared for the specific purpose of serving as a predicate for technical characteristics, rather than as a marketable medical device. These “step” devices serve as intermediate predicates to allow more innovative devices with larger technological “leaps” into the market. Although devices with larger technological leaps are not necessarily unsafe or ineffective, for example the Da Vinci has remained on the market for over 15 years without a major recall, they do may possess more potential risk due to the fast-paced introduction of less-understood technologies into the marketplace. Although the FDA makes efforts to mitigate this risk through existing regulatory mechanisms, the lack of clearly defined substantial equivalence requirements makes it difficult to determine whether measures taken for a particular device are sufficient.

In the future, the FDA could make efforts to identify the “leap” devices and create a more targeted approval process that addresses new questions raised by these technologies. Defining clear guidelines for the amount of scientific evidence required based on the significance of new technological characteristics for device clearance would help reduce this inherent systematic risk.

For example, there may be different evidentiary requirements for a new device that is identical to an existing device but being used for a new application vs. a device utilizing a novel combination of technological characteristics from multiple previously cleared devices vs. a device that allows for a novel and significantly different use scenario.

 Conclusions

The primary method used to address these objectives was the development of multiple predicate ancestry trees using information available through FDA databases. Although the amount of available data is significantly limited due to missing data in the FDA database, the predicate traces developed contained enough information to determine if there was evidence of predicate creep. Through analysis of these traces, with a focus on regulatory information such as product classification codes, we were able to conclude that there is indeed both large and small scale predicate creep present within the 510(k) clearance process. It is important to note that while our research focuses on ways to identify predicate creep, it does not take the next step in identifying if this creep is, in fact, problematic. If all the preceding predicate devices are safe, and precautions are taken to mitigate small-scale predicate creep, in many cases devices exhibiting large-scale predicate creep may still be safe. Additional research is needed to assess the actual risk of technology creep across different types of medical devices and to develop mechanisms to identify when this creep may pose additional risk.

 References FDA. Review of the ReGen Menaflex®*: Departures from Processes, Procedures, and Practices Leave the Basis for a Review Decision in Question. Preliminary Report. September 2009. Accessed December 22, 2022 at https://fda.report/media/77734/Review-of-the-ReGen—Menaflex.pdf. IOM (Institute of Medicine). 2011. Medical Devices and the Public’s Health: The FDA 510(k) Clearance Process at 35 Years. Washington, DC: The National Academies Press. Ardaugh BM, Graves SE, Redberg RF. The 510 (k) ancestry of a metal-on-metal hip implant. New England Journal of Medicine. 2013 Jan 10;368(2):97–100. doi: 10.1056/NEJMp1211581 23301729 Zargar N, Carr A. The regulatory ancestral network of surgical meshes. PLoS One. 2018 Jun 19;13(6):e0197883. doi: 10.1371/journal.pone.0197883 29920525 Center for Devices and Radiological Health. (2010, August). 510(k) Working Group Preliminary Report and Recommendations. Accessed December 22, 2022, from https://www.fda.gov/files/about%20fda/published/CDRH-Preliminary-Internal-Evaluations—-Volume-I—510(k)-Working-Group-Preliminary-Report-and-Recommendations.pdf. Heneghan C, Thompson M. Rethinking medical device regulation. Journal of the Royal Society of Medicine. 2012 May;105(5):186–8. doi: 10.1258/jrsm.2012.12k030 22637765 Hines JZ, Lurie P, Yu E, Wolfe S. Left to their own devices: breakdowns in United States medical device premarket review. PLoS medicine. 2010 Jul 13;7(7):e1000280. doi: 10.1371/journal.pmed.1000280 20634922 Medtronic, Inc. v. Lohr, 518 U. S. 470 (1996) Accessed December 23, 2022 at https://supreme.justia.com/cases/federal/us/518/470/case.pdf. Zuckerman D, Brown P, Das A. Lack of publicly available scientific evidence on the safety and effectiveness of implanted medical devices. JAMA Internal Medicine. 2014 Nov 1;174(11):1781–7. doi: 10.1001/jamainternmed.2014.4193 25265047 Kadakia KT, Dhruva SS, Caraballo C, Ross JS, Krumholz HM. Use of Recalled Devices in New Device Authorizations Under the US Food and Drug Administration’s 510(k) Pathway and Risk of Subsequent Recalls. JAMA. 2023;329(2):136–143. doi: 10.1001/jama.2022.23279 36625810 Everhart AO, Sen S, Stern AD, Zhu Y, Karaca-Mandic P. Association Between Regulatory Submission Characteristics and Recalls of Medical Devices Receiving 510(k) Clearance. JAMA. 2023;329(2):144–156. doi: 10.1001/jama.2022.22974 36625811 Lennox K. Substantially unequivalent: reforming FDA regulation of medical devices. U. Ill. L. Rev.. 2014:1363. Rathi VK, Ross JS. Modernizing the FDA’s 510 (k) Pathway. The New England journal of medicine. 2019 Nov 14;381(20):1891–3. doi: 10.1056/NEJMp1908654 31644865 Fargen KM, Frei D, Fiorella D, McDougall CG, Myers PM, Hirsch JA, et al. The FDA approval process for medical devices: an inherently flawed system or a valuable pathway for innovation? Journal of neurointerventional surgery. 2013 Jul 1;5(4):269–75. doi: 10.1136/neurintsurg-2012-010400 22764203 Liebeskind AY, Chen AC, Dhruva SS, Sedrakyan A. A 510 (k) ancestry of robotic surgical systems. International journal of surgery (London, England). 2022 Feb;98:106229. doi: 10.1016/j.ijsu.2022.106229 35032697 Waetjen LE, Parvataneni R, Varon S, Saberi NS, Jacoby VL. Obstacles to studying emerging technologies. Obstetrics & Gynecology. 2015 Aug 1;126(2):391–5. doi: 10.1097/AOG.0000000000000914 26241430 Thai MT, Phan PT, Hoang TT, Wong S, Lovell NH, Do TN. Advanced intelligent systems for surgical robotics. Advanced Intelligent Systems. 2020 Aug;2(8):1900138. Wang J. Research on Development Status of Surgical Robot Based on Patent Analysis. In2021 6th International Conference on Automation, Control and Robotics Engineering (CACRE) 2021 Jul 15 (pp. 33–37). IEEE. Alemzadeh H, Raman J, Leveson N, Kalbarczyk Z, Iyer RK. Adverse events in robotic surgery: a retrospective study of 14 years of FDA data. PloS one. 2016 Apr 20;11(4):e0151470. doi: 10.1371/journal.pone.0151470 27097160 FDA, How to Make a FOIA Request Accessed February 15, 2023 at https://www.fda.gov/regulatory-information/freedom-information/how-make-foia-request. Center for Devices and Radiological Health, FDA, Small Sample Survey–Final Report. Topic: da Vinci Surgical System. 1013, November. Accessed December 23, 2022 at https://www.fda.gov/media/87485/download. FDA, Discussion Paper: Robotically-Assisted Surgical Devices. Accessed December 23, 2022 at https://www.fdanews.com/ext/resources/files/07-15/072015-robots-surgery.pdf?1436973876. 10.1371/journal.pone.0283442.r001 Decision Letter 0 Wang Quanzeng Academic Editor 2023 Quanzeng Wang This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Submission Version 0

14 Sep 2022

PONE-D-22-04558Identification of Predicate Creep under the 510K process: A Case Study of Robotic Surgical DevicesPLOS ONE

Dear Dr. Rothenberg,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please sufficiently address the comments and submit your revised manuscript in six months. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Quanzeng Wang

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Please ensure that you include a title page within your main document. We do appreciate that you have a title page document uploaded as a separate file, however, as per our author guidelines (http://journals.plos.org/plosone/s/submission-guidelines#loc-title-page) we do require this to be part of the manuscript file itself and not uploaded separately.

Could you therefore please include the title page into the beginning of your manuscript file itself, listing all authors and affiliations.

3. We noted in your submission details that a portion of your manuscript may have been presented or published elsewhere. [this paper is from a Master's Thesis, which was published by Rochester Institute of Technology.] Please clarify whether this [conference proceeding or publication] was peer-reviewed and formally published. If this work was previously peer-reviewed and published, in the cover letter please provide the reason that this work does not constitute dual publication and should be included in the current manuscript.

4. Please include a separate caption for each figure in your manuscript.

5. We note that Figure 3 in your submission contain copyrighted images. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

 a. You may seek permission from the original copyright holder of Figure 3 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

 b. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

6. Please include your tables as part of your main manuscript and remove the individual files. Please note that supplementary tables (should remain/ be uploaded) as separate "supporting information" files.

Additional Editor Comments (if provided):

(1)

You cited the following paper in your manuscript.

B. M. Ardaugh, S. E. Graves, and R. F. Redberg, "The 510 (k) ancestry of a metal-on-metal hip implant," New England Journal of Medicine, vol. 368, no. 2, pp. 97-100, 2013.

This paper analyzed FDA documents obtained from both the agency’s database and Freedom of Information Office through the Freedom of Information Act (FOIA). In this manuscript, however, you only used data from the agency’s database, which is not complete. As a result, the predicate device trees (e.g., Fig. 3) are not accurate. Some devices were cleared after more generations than described in the manuscript. You should obtain a complete predicate history of Da Vinci Surgical system through FOIA and perform further analysis. If needed, you can contact the agency’s Division of Industry and Consumer Education (DICE, https://www.fda.gov/medical-devices/device-advice-comprehensive-regulatory-assistance/contact-us-division-industry-and-consumer-education-dice) for help to perform a through predicate analysis.

(2)

You mentioned that some devices were cleared without necessary clinical studies. However, you only analyzed data until 2008. Since the new 510(k) guidance published in 2014, FDA has already been steadily enhancing the review process of Robotic Assisted Surgery (RAS) devices to address risk. Clinical data has been required to clear new surgical procedures for Da Vinci Surgical system (e.g., K182371). You should discuss recent data from 2008 to present to provide a full and complete picture of FDA regulatory landscape.

(3)

If for whatever reason you cannot perform a thorough study as suggested, you should provide justifications and sufficiently discuss limitations of this study in the paper.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: N/A

Reviewer #2: N/A

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: 

The case of RASD is an unusual situation. There is likely not other case with greater creep. When the figures are separated from the text it makes if more challenging to read. There are other cases of creep for sure but this is not typical.

Reviewer #2: 

Thank you for the opportunity to review this important study on the role of predicate creep in medical device regulation. The excessive use of predicates, and the technological and regulatory divergence among devices in the Da Vinci lineage, is striking.

Several thematic and organizational comments on the manuscript below, with comments on Figures at the end:

– The literature already contains several reports about predicate creep, including a recent study focused specifically on some of the same Da Vinci robotic surgery systems that this study focuses on (https://pubmed.ncbi.nlm.nih.gov/35032697/). The studies have some differences; the one I have cited is more focused on ancestry, and this study is more focused on predicate creep. However, I think the authors would benefit from more clearly distinguishing their work and articulating what the new contribution to the literature is.

– While predicate creep is a tricky phenomenon to study, I am not entirely sure how the methodology of this study differs from previous research on predicate creep; as previous studies have also looked at predicates, decision memos, product codes, etc. The article may benefit from a more explicit discussion of what is new in their methodology (especially since the authors state on Page 17 that they “follow a method similar to Ardaugh and Zuckerman”). The findings can still be helpful even if the methodology is similar, but the manuscript’s narrative would need to be reframed.

– The discussion of using different Da Vinci databases to perform predicate traces is a bit confusing for the reader.

– One of the strongest findings to me is in Figures 5-7, which nicely capture the evolution in product codes and regulatory descriptions over time. This is a finding that certainly has important regulatory implications given that the FDA is moving to create product-specific guidance documents for 510(k).

– The introduction and literature review sections are lengthy and contain important information; however, given the length of the manuscript, these sections may benefit from revisions for brevity to improve the readability of the manuscript.

Specific stylistic comments:

– The manuscript itself is quite lengthy and would benefit from revisions for brevity.

– Throughout the document, the authors use the term “510K Process”. Per FDA documents and other published reports in the literature, I believe the more appropriate term is “510(k) pathway” rather than “process”.

– Throughout the document, the authors alternate between using “510K” and “510(k)”. Please revise to the latter, which is how the pathway is described in FDA documents.

– There are some typographical errors in the manuscript, and additional references may be needed at different junctures. Please see the comments below for specific instances.

The below comments are recommendations for specific-line edits. The manuscript version I was provided does not include line numbers; consequently, I have tried to provide page numbers (referring to the page in the document) and sentence quotes for the authors’ ease of reference:

– Page 8: in the sentence “…..drawn attention to this regulatory approval process”; suggest changing “approval” to either “clearance” or “authorization”. Devices cleared under 510(k) are not technically “approved”.

– Page 8: in the sentence “…predicate creep, a cycle of technology change”; according to the FDA’s definition of predicate creep, the changes may not just be limited to technology, but can also refer to indications. Would suggest revising to reflect this.

– Page 9: In the sentence starting “In the last decade…”, would suggest referencing the 2011 IOM report on 510(k), which references “creep” on page 89 and 230. https://nap.nationalacademies.org/download/13150 – Page 9: In the introduction, I suggest the authors explicitly reference the FDA’s 2009 report on the ReGen MenaFlex device, which (to my knowledge) is the first time the agency explicitly acknowledged the phenomenon of predicate creep and provided a formal definition: https://int.nyt.com/data/int-shared/nytdocs/docs/104/104.pdf.

– Page 9: In the sentence “…there is a lack of data to support concerns surrounding the 510(k)”; I am not sure this is true. There are many case studies in the literature (see Kadakia 2021 in JAMA IM, Freeman 2014 in Annals of Health Law, Ardaugh 2013 in NEJM) and some systematic studies as well (Pai 2021 in PLOS ONE, Zuckerman 2014 in JAMA IM). I believe the introduction might benefit from a more robust discussion of the existing literature, and how the methodology/intervention of this study contributes to and/or differs from existing approaches and knowledge.

– Page 10: In the sentence “…these methods will be tested on the Da Vinci Surgical System”; a recent study published in the International Journal of Surgery took this same approach (https://pubmed.ncbi.nlm.nih.gov/35032697/). I would suggest the authors contextualize their findings and methods given this recent research.

– Page 10: Rather than quoting the FDA definition, the authors could consider paraphrasing for brevity.

– Page 11: The sentence “…manufacturers are able to modify the wording of the stated intended use to make changes in device function appear minimal” may benefit from a citation given it is a claim of manufacturer intent.

– Page 11: In the sentence “In his concurring opinion…”; I believe this should be “her”, as Justice O’Connor is a female.

– Page 12: In the sentence “…deemed safe based on market performance”; would clarify for the reader what “market performance” actually means.

– Page 13: In the sentence “…maintained by the FDA, with mixed results”; I am not sure what makes the previous studies “mixed”. Please clarify.

– Page 13: In the sentence “…it is hard to do without the use of mandatory clinical trials, which would defeat the purpose of the 510(k) process entirely”. I am not sure this is true. The authors could consider the role of registries and post-market surveillance platforms in advancing evidence generation, and clearer premarket standards in mitigating predicate creep.

– Page 14: In the sentence “…documents obtained through the FDA 510(f) database”; there appears to be a typo, the “(f)” should be a “(k)”

– Page 14: The authors write “Neither article, however, clearly specified the exact methodology used to trace predicates”. I am not sure this is true, as the preceding sentence indicates that Zuckerman used the 510(k) database, and the original manuscript cited here states in their methods section that “we analyzed the type of scientific evidence the company provided to the FDA and the public to support the claim of substantial equivalence to a device already on the market or to establish safety or effectiveness” – which appears to be a similar method to what the authors are proposing? The main difference appears to be that the previous study did not look at differences in product code. (Additionally, the authors state on Page 17 that they “follow a method similar to Ardaugh and Zuckerman” – so it’s not clear what the flaw/differences are).

– Page 15: The sentence “…more precise control and motion” is a claim of performance that should include a citation. Similarly, the argument in the following sentence about “…Intuitive’s strong patent foothold” also requires a citation.

– Page 15: The paragraph “The Da Vinci is an interesting case study” may benefit from being moved to the methods. It should also include a discussion of a recent paper published on its 510(k) ancestry in the International Journal of Surgery.

– Page 16: In the sentence “For this paper, most data….”; please define what the “data” are.

– Page 16: In the sentence, “…were collected through 510(k) approval database”; a “the” appears to be missing, and “approval” should be “clearance”.

– Page 16: In the sentence “When either of these databases did not provide the needed information, we also conducted targeted search…”; the “either” should be “neither”, and an “a” is missing before “targeted”.

–Page 17: Portions of the figure description could be moved from the methods in to a caption for the figure to improve brevity.

–Page 18: In the sentence “….increased number of predicates appear make traditional tree diagrams unwieldy”; a “to” appears to be missing”.

– Page 18: The majority of the paragraph under the header, “Measuring Predicate Creep” could be deleted as this set-up has already been completed in the Introduction/Literature Review (which is the more appropriate place for it).

– Page 21: The sentence “Since the successful performance of each predicate device in the market is part of the body of evidence to support the safety claims of the new device, a smaller number of unique devices with market performance data effectively reduces the level of assurance of safety for the subject device” – does not make sense. Per the IOM report and the authors’ own arguments in the introduction, the 510(k) pathway does not support justifications of safety and effectiveness. Additionally, per the authors’ arguments, shouldn’t the quality of the substantial equivalence determination rather than the number of unique devices be the real determinant of safety?

– Page 21: In the section on “Direct Comparison of Technological Characteristics”; it is not clear what makes “the degree of technological differences ‘striking’” or (per page 22) what constitutes a “large change in technology within a single predicate relationship”. (Some of this is of course intuitive, but it is worth spelling out for the reader – because by the FDA’s guidance, the subject and predicate are not required to be identical, so some differences are expected”). For example, the authors could consider interweaving a discussion of the clinical/non-clinical evidence here.

– Page 24: What is the difference between “device instances” and “unique devices”?

– Page 24: “Grouping by regulatory mechanisms”; devices can have their product codes reclassified over time; was that the case for any of the devices in the sample?

– Page 29: The sentence “…methods of predicate analysis; The” I think the semicolon is supposed to be a period.

– Page 30: In the sentence “the 510(k) process was officially implemented via the 1990 Safe Medical Devices Act” – technically the pathway was established as part of the 1976 device amendments. Additionally, this paragraph could be potentially condensed, as the use of these different forms is a moot point since digital copies anyway are not available until ~2000.

– Page 32: The first paragraph under “Implications for Research” should be moved to the limitations section (as it is essentially a question of generalizability)

– Page 34: The sentence “Although devices with larger…” is grammatically off (the clause in the middle about Da Vinci creates an awkward break”).

– Page 35: The sentence “The FDA has recently begun taking steps…” could use a citation. Also, I am not sure that “split predicates” are really the problem for “leap devices” as identified in the authors’ research.

Comments on Figures

– Figure 1: While this figure is well-intentioned, I don’t think the message is quite clear to the reader. Perhaps some simple labeling like “Device 1, Device 2”, etc. could be helpful. Also perhaps worth noting somewhere here the lack of testing.

– Figure 2: Would recommend reformatting this figure, given the unevenness of the red branch line. Could either draw a more symmetric shape, or could consider just highlighting the lines in colors rather than creating a box.

– Figures 5, 6, 7: These figures are really nicely done, and nicely capture the regulatory effect of predicate creep. Kudos to the authors.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

 10.1371/journal.pone.0283442.r002 Author response to Decision Letter 0 Submission Version 1

23 Dec 2022

Thank you for giving us the opportunity to revise this paper. Based on the reviewer comments we have significantly shortened and altered the framing of this paper. We now focus on what reviewer 2 thought was the primary contribution of the paper, identification of potential “predicate creep” through the use of product codes and regulatory classifications. Below you will find a detailed response to yours and the reviewers’ comments. If you refer to the attached response documents, our responses are in italics.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

We have reformatted the title page and paper according to these guidelines.

2. Please ensure that you include a title page within your main document. We do appreciate that you have a title page document uploaded as a separate file, however, as per our author guidelines (http://journals.plos.org/plosone/s/submission-guidelines#loc-title-page) we do require this to be part of the manuscript file itself and not uploaded separately.

Could you therefore please include the title page into the beginning of your manuscript file itself, listing all authors and affiliations.

We have included the title page.

3. We noted in your submission details that a portion of your manuscript may have been presented or published elsewhere. [this paper is from a Master's Thesis, which was published by Rochester Institute of Technology.] Please clarify whether this [conference proceeding or publication] was peer-reviewed and formally published. If this work was previously peer-reviewed and published, in the cover letter please provide the reason that this work does not constitute dual publication and should be included in the current manuscript.

We have included this explanation in the cover letter.

4. Please include a separate caption for each figure in your manuscript.

We have made a short caption for each figure in the manuscript.

5. We note that Figure 3 in your submission contain copyrighted images. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

a. You may seek permission from the original copyright holder of Figure 3 to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

b. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

This figure has been removed from the paper.

6. Please include your tables as part of your main manuscript and remove the individual files. Please note that supplementary tables (should remain/ be uploaded) as separate "supporting information" files.

Additional Editor Comments (if provided):

(1)

You cited the following paper in your manuscript.

B. M. Ardaugh, S. E. Graves, and R. F. Redberg, "The 510 (k) ancestry of a metal-on-metal hip implant," New England Journal of Medicine, vol. 368, no. 2, pp. 97-100, 2013.

This paper analyzed FDA documents obtained from both the agency’s database and Freedom of Information Office through the Freedom of Information Act (FOIA). In this manuscript, however, you only used data from the agency’s database, which is not complete. As a result, the predicate device trees (e.g., Fig. 3) are not accurate. Some devices were cleared after more generations than described in the manuscript. You should obtain a complete predicate history of Da Vinci Surgical system through FOIA and perform further analysis. If needed, you can contact the agency’s Division of Industry and Consumer Education (DICE, https://www.fda.gov/medical-devices/device-advice-comprehensive-regulatory-assistance/contact-us-division-industry-and-consumer-education-dice) for help to perform a through predicate analysis.

We did consider requesting additional data through FOIA. We decided against this route, however, since earlier experience with the FOIA process resulted in response times of over 6 months. Moreover, once the first level of predicates are requested, we would then need to see what predicates are used for any new devices in the tree, which would then take another 6 months. We did try sending some requests directly to the companies, but these requests were uniformly unanswered. Given we are looking backwards in time, the fact that some predicates are missing actually makes any finding of predicate creep more significant. Additional predicate data would either show no change in the amount of predicate creep or would show increased predicate creep. We now include the following in our paper:

“This study was limited by several factors. First, the investigations performed in this study were based solely on data available publicly through FDA databases. Predicate devices are identified within the database only when application summaries are provided, which was not required for inclusion until the mid-2000’s. This significantly limits the number of devices with traceable predicate histories. Moreover, the level of information contained within these summaries varies widely, and in many cases these summaries do not exist at all, due to the gradual evolution of requirements for 510(k) application submissions. We did consider requesting additional data through FOIA, decided against this route given that it would add a significant amount of time to the data gathering and there was a precedent in prior research to rely primarily on information provided in the FDA website [11]. Additionally, given we are looking backwards in time, the fact that some predicates are missing actually makes any finding of predicate creep more significant.”

(2)

You mentioned that some devices were cleared without necessary clinical studies. However, you only analyzed data until 2008. Since the new 510(k) guidance published in 2014, FDA has already been steadily enhancing the review process of Robotic Assisted Surgery (RAS) devices to address risk. Clinical data has been required to clear new surgical procedures for Da Vinci Surgical system (e.g., K182371). You should discuss recent data from 2008 to present to provide a full and complete picture of FDA regulatory landscape.

Since we are focus on one particular device and looking back, we did not feel the need to include more recent data. Given the refocusing of the paper on the method identifying predicate creep reviewer 2 thought was the primary contribution of the paper, identification of potential “predicate creep” through the use of product codes and regulatory classifications, providing more recent data does not add to the goal of the analysis. You are correct that we make statements about the FDA process without including more recent changes in process. We have addressed this in the limitations.

We now state:

“Lastly, the device investigated in this study was cleared in 2009. In 2013, the FDA completed a small sample survey to better understand the causes behind adverse events in the da Vinci Surgical system [20]. In 2014, the FDA took steps to change the review of robotically assisted surgical devices [21]. Thus, our review does not take the evolution of regulation that has happened since 2009. While this does not limit the use of this method to investigate predicate creep, it may impact how patterns are interpreted over time. It would be interesting to compare how the predicate network changes with newer models of the da Vinci that may have been cleared under different processes and standards.”

(3)

If for whatever reason you cannot perform a thorough study as suggested, you should provide justifications and sufficiently discuss limitations of this study in the paper.

Please see the additions to the limitation as listed above.

Comments to the Author

Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1:

The case of RASD is an unusual situation. There is likely not other case with greater creep. When the figures are separated from the text it makes if more challenging to read. There are other cases of creep for sure but this is not typical.

Thank you for your comment. We agree that this level of creep may not be seem across other medical devices. It is, however, a good case to review ways to measure predicate creep. We discuss the issue of generalizability as a limitation of the study. We apologize for the separation of the figures and also find this more difficult to read. Unfortunately, that was the format that was requested by the journal.

Reviewer #2:

Thank you for the opportunity to review this important study on the role of predicate creep in medical device regulation. The excessive use of predicates, and the technological and regulatory divergence among devices in the Da Vinci lineage, is striking.

Thank you in advance for these helpful comments. You are clearly very familiar with this literature and the FDA processes. Your comments have led us to new, important, references. Perhaps more important, it has led to focus the paper on the part of the analysis you found most interesting, which was the second half the paper. By focusing on this part of the paper, the paper is hopefully more concise and makes a clearer contribution.

Several thematic and organizational comments on the manuscript below, with comments on Figures at the end:

– The literature already contains several reports about predicate creep, including a recent study focused specifically on some of the same Da Vinci robotic surgery systems that this study focuses on (https://pubmed.ncbi.nlm.nih.gov/35032697/). The studies have some differences; the one I have cited is more focused on ancestry, and this study is more focused on predicate creep. However, I think the authors would benefit from more clearly distinguishing their work and articulating what the new contribution to the literature is.

Thank you for pointing out this article. We have integrated it into our paper. In addition, as summarized in response to this and your next comment. We agree that we don’t sufficiently differentiate our paper. We have modified the end of the introduction in a way that both highlights the contributions of past research, and differentiates the contribution of our own paper. Please see the changes, placed after your next comment, to see the new framing.

– While predicate creep is a tricky phenomenon to study, I am not entirely sure how the methodology of this study differs from previous research on predicate creep; as previous studies have also looked at predicates, decision memos, product codes, etc. The article may benefit from a more explicit discussion of what is new in their methodology (especially since the authors state on Page 17 that they “follow a method similar to Ardaugh and Zuckerman”). The findings can still be helpful even if the methodology is similar, but the manuscript’s narrative would need to be reframed.

We completely agree with this assessment that the technical comparison method is essentially the same as the one just by Ardaugh et. al (2013). This has led us to realize that our most important contribution is essentially our second method, which focuses on regulatory classification. The benefit of this approach is that, once the data is entered, can be done fairly quickly and be used to identify areas in a predicate network that one can focus on for direct technical comparisons. We have changed the paper narrative to this effect and have added the following:

“Within the 510(k) process, requirements for non-clinical testing are used to mitigate the risk associated with small scale predicate creep, which in most cases works effectively. However, past research on predicate networks has highlighted the lack of publicly available non-clinical scientific data in substantial equivalence determinations. Both Zuckerman et al. [11] and Liebeskind et al. [15] used the FDA 510(k) Database and/or FOIA requests to trace the predicate history of different medical devices and found that there was a dearth of publicly available scientific data to support the claim of substantial equivalence. Liebeskind et al. [15], in particular, traced the predicate ancestry of robotic surgical systems, and found that (92.7%) 510(k) clearances did not submit clinical data, including 73 (27.9%) that did not submit any supporting data. These studies suggest that non-clinical evidence to support substantial equivalence claims may be insufficient to ensure safety when there is predicate creep.

Past research on predicate creep primarily involves creating predicate ancestry trees and using direct technical comparisons of devices and their predicates. Such studies have identified concerning predicate creep in surgical meshes [6] and Pathwork Tissue of Origin Test [9], DePuy ASR XL Acetabular Cup System [5], power morcellators [16], among others. Ardaugh et al. [5], for example, studied the metal-on-metal hip implant over five decades with the purpose of identifying the cause of safety flaws present in the design. They identify that a unique combination of three characteristics in the ASR XL were approved though substantial equivalence of “split predicates,” where you compare characteristic to different predicate devices. Since all three devices were deemed safe based on clinical trials and safe use of predicates in the consumer market, and the XL simply combined parts of the devices, it was placed on the market without undergoing clinical testing and resulting product had to be recalled for particle shedding [5].

Direct technical comparisons can be challenging due to a lack of information, as discussed above, or the cumbersome size of predicate ancestry trees, particularly for newer devices. Using FOIA requests to attain more detailed information can also be onerous. This paper proposes another way to identify potentially important instances of predicate creep in a large predicate network through analysis of product classes and regulatory categorizations. We test this method on a well-known medical device, the Intuitive Surgical da Vinci Si robotic surgical system.”

– The discussion of using different Da Vinci databases to perform predicate traces is a bit confusing for the reader.

– One of the strongest findings to me is in Figures 5-7, which nicely capture the evolution in product codes and regulatory descriptions over time. This is a finding that certainly has important regulatory implications given that the FDA is moving to create product-specific guidance documents for 510(k).

We agree that this should be more of the focus of the paper, and is the main finding. This has been made clearer in the introduction and the conclusion.

– The introduction and literature review sections are lengthy and contain important information; however, given the length of the manuscript, these sections may benefit from revisions for brevity to improve the readability of the manuscript.

Specific stylistic comments:

– The manuscript itself is quite lengthy and would benefit from revisions for brevity.

We have significantly shortened the paper by reducing the literature review and taking out the material on the first Da Vinci technical comparison.

– Throughout the document, the authors use the term “510K Process”. Per FDA documents and other published reports in the literature, I believe the more appropriate term is “510(k) pathway” rather than “process”.

We have completed this change.

– Throughout the document, the authors alternate between using “510K” and “510(k)”. Please revise to the latter, which is how the pathway is described in FDA documents.

We have completed this change.

– There are some typographical errors in the manuscript, and additional references may be needed at different junctures. Please see the comments below for specific instances.

The below comments are recommendations for specific-line edits. The manuscript version I was provided does not include line numbers; consequently, I have tried to provide page numbers (referring to the page in the document) and sentence quotes for the authors’ ease of reference:

– Page 8: in the sentence “…..drawn attention to this regulatory approval process”; suggest changing “approval” to either “clearance” or “authorization”. Devices cleared under 510(k) are not technically “approved”.

Thank you for catching this. Since the FDA website refers to this is clearance – we have changed this to the term “clearance” throughout the paper

– Page 8: in the sentence “…predicate creep, a cycle of technology change”; according to the FDA’s definition of predicate creep, the changes may not just be limited to technology, but can also refer to indications. Would suggest revising to reflect this.

Thank you – using the reference provided below on the ReGen MenaFlex, we have changed the sentence to the following:

“One of the specific concerns in the 510(k) process is the risk of predicate creep, a cycle of technology change through repeated clearance of devices based on predicates with slightly different technological characteristics, such as materials and power sources, or have indications for different anatomical sites [3-4].”

– Page 9: In the sentence starting “In the last decade…”, would suggest referencing the 2011 IOM report on 510(k), which references “creep” on page 89 and 230. https://nap.nationalacademies.org/download/13150 –

Thank you for this reference. There was quite a bit of valuable material in it and we now reference it.

Page 9: In the introduction, I suggest the authors explicitly reference the FDA’s 2009 report on the ReGen MenaFlex device, which (to my knowledge) is the first time the agency explicitly acknowledged the phenomenon of predicate creep and provided a formal definition: https://int.nyt.com/data/int-shared/nytdocs/docs/104/104.pdf.

Thank you for this suggestion. We have reviewed this report and now reference it in the paper.

– Page 9: In the sentence “…there is a lack of data to support concerns surrounding the 510(k)”; I am not sure this is true. There are many case studies in the literature (see Kadakia 2021 in JAMA IM, Freeman 2014 in Annals of Health Law, Ardaugh 2013 in NEJM) and some systematic studies as well (Pai 2021 in PLOS ONE, Zuckerman 2014 in JAMA IM). I believe the introduction might benefit from a more robust discussion of the existing literature, and how the methodology/intervention of this study contributes to and/or differs from existing approaches and knowledge.

We completely agree. In fact, we contradict this statement in the literature review. We now have changed this part of the introduction to read as follows and document some of the specific problems in the literature review section of the paper:

“In the last decade, several high-profile device recalls have drawn attention to this regulatory clearance process and researchers have raised concerns about the validity of the 510(k) process as a broad clearance mechanism. These concerns include, but are not limited to, a lack of a clear definition of “intended use,” using recalled devices as predicates, lack of publicly available data regarding predicates and determinations of substantial equivelence, equating substantial equivelence to safety, and a lack of sufficient non-clinical data to support claims of substantial equivelence, among others. As outlined in the literature review below, there is a body of research that documents these problems and the resulting safety issues that have stemmed from them.”

– Page 10: In the sentence “…these methods will be tested on the Da Vinci Surgical System”; a recent study published in the International Journal of Surgery took this same approach (https://pubmed.ncbi.nlm.nih.gov/35032697/). I would suggest the authors contextualize their findings and methods given this recent research.

Again, thank you for identifying this article. We now highlight this article and its findings, but also state how what they do is different than what we will be doing. We have added the following:

“Within the 510(k) process, requirements for non-clinical testing are used to mitigate the risk associated with small scale predicate creep, which in most cases works effectively. However, past research on predicate networks has highlighted the lack of publicly available non-clinical scientific data in substantial equivalence determinations. Both Zuckerman et al. [11] and Liebeskind et al. [15] used the FDA 510(k) Database and/or FOIA requests to trace the predicate history of different medical devices and found that there was a dearth of publicly available scientific data to support the claim of substantial equivalence. Liebeskind et al. [15], in particular, traced the predicate ancestry of robotic surgical systems, and found that (92.7%) 510(k) clearances did not submit clinical data, including 73 (27.9%) that did not submit any supporting data. These studies suggest that non-clinical evidence to support substantial equivalence claims may be insufficient to ensure safety when there is predicate creep.

Past research on predicate creep primarily involves creating predicate ancestry trees and using direct technical comparisons of devices and their predicates. Such studies have identified concerning predicate creep in surgical meshes [6] and Pathwork Tissue of Origin Test [9], DePuy ASR XL Acetabular Cup System [5], power morcellators [16], among others. Ardaugh et al. [5], for example, studied the metal-on-metal hip implant over five decades with the purpose of identifying the cause of safety flaws present in the design. They identify that a unique combination of three characteristics in the ASR XL were approved though substantial equivalence of “split predicates,” where you compare characteristic to different predicate devices. Since all three devices were deemed safe based on clinical trials and safe use of predicates in the consumer market, and the XL simply combined parts of the devices, it was placed on the market without undergoing clinical testing and resulting product had to be recalled for particle shedding [5].

Direct technical comparisons can be challenging due to a lack of information, as discussed above, or the cumbersome size of predicate ancestry trees, particularly for newer devices. Using FOIA requests to attain more detailed information can also be onerous. This paper proposes another way to identify potentially important instances of predicate creep in a large predicate network through analysis of product classes and regulatory categorizations. We test this method on a well-known medical device, the Intuitive Surgical da Vinci Si robotic surgical system.”

– Page 10: Rather than quoting the FDA definition, the authors could consider paraphrasing for brevity.

We have removed this long quote, as we feel it was not needed.

– Page 11: The sentence “…manufacturers are able to modify the wording of the stated intended use to make changes in device function appear minimal” may benefit from a citation given it is a claim of manufacturer intent.

This is a good observation. Rereading the cited papers, the issue of manufacturers intent was referred to papers that focused on the process in the NICE Evaluation Pathway (EP) Programme in the United Kingdom. Thus, we have changed the sentence to be as follows, which more accurately reflects the sentiment in the studies focused on the 510(k) process:

“Many of the critiques of the 510(k) process relate to the notion of substantial equivelence. First, due to the lack of a clear official definition for the key terms “indications for use” and “intended use”, the FDA has allowed permissive interpretation of these terms by applicants and inconsistent use of the terms across reviewers [7-8]. Over time, this has resulted in the clearance of significantly altered devices, or even novel devices, as substantially equivalent to established predicates [7-9].”

– Page 11: In the sentence “In his concurring opinion…”; I believe this should be “her”, as Justice O’Connor is a female.

Great catch. We actually took out this sentence to make the review more concise.

– Page 12: In the sentence “…deemed safe based on market performance”; would clarify for the reader what “market performance” actually means.

We have changed the sentence to read as follows:

“Since all three devices were deemed safe based on clinical trials and safe use of predicates in the consumer market, and the XL simply combined parts of the devices, it was placed on the market without undergoing clinical testing and resulting product had to be recalled for particle shedding [5]. “

– Page 13: In the sentence “…maintained by the FDA, with mixed results”; I am not sure what makes the previous studies “mixed”. Please clarify.

We have removed this sentence.

– Page 13: In the sentence “…it is hard to do without the use of mandatory clinical trials, which would defeat the purpose of the 510(k) process entirely”. I am not sure this is true. The authors could consider the role of registries and post-market surveillance platforms in advancing evidence generation, and clearer premarket standards in mitigating predicate creep.

We ended up removing this sentence, but you are correct. As the editor has indicated, there has been attention by the FDA to both of these areas for robotic surgical devices. Documents describing these changes (or intended changes) are now referenced in the limitations sections.

– Page 14: In the sentence “…documents obtained through the FDA 510(f) database”; there appears to be a typo, the “(f)” should be a “(k)”

This has been fixed.

– Page 14: The authors write “Neither article, however, clearly specified the exact methodology used to trace predicates”. I am not sure this is true, as the preceding sentence indicates that Zuckerman used the 510(k) database, and the original manuscript cited here states in their methods section that “we analyzed the type of scientific evidence the company provided to the FDA and the public to support the claim of substantial equivalence to a device already on the market or to establish safety or effectiveness” – which appears to be a similar method to what the authors are proposing? The main difference appears to be that the previous study did not look at differences in product code. (Additionally, the authors state on Page 17 that they “follow a method similar to Ardaugh and Zuckerman” – so it’s not clear what the flaw/differences are).

We have removed this statement, and more clearly differentiate our method from Zuckerman and others as focusing on product codes and regulatory classifications.

– Page 15: The sentence “…more precise control and motion” is a claim of performance that should include a citation. Similarly, the argument in the following sentence about “…Intuitive’s strong patent foothold” also requires a citation.

We have added recent references to support these two statements.

“The Da Vinci Surgical System is a robotic-based laparoscopic surgical tool initially approved by the FDA in 2000 which replaces a surgeon’s hands with robotic arms for more precise control and motion [17]. While Intuitive has subsequently brought multiple iterations of the Da Vinci to market, it remains the only full RAS platform on the market as competitors struggle to develop a viable competitor around Intuitive’s strong patent foothold [18].”

– Page 15: The paragraph “The Da Vinci is an interesting case study” may benefit from being moved to the methods. It should also include a discussion of a recent paper published on its 510(k) ancestry in the International Journal of Surgery.

We have made both of these changes. In the methods we state:

“The Da Vinci Surgical System is a robotic-based laparoscopic surgical tool initially approved by the FDA in 2000 which replaces a surgeon’s hands with robotic arms for more precise control and motion [17]. While Intuitive has subsequently brought multiple iterations of the Da Vinci to market, it remains the only full RAS platform on the market as competitors struggle to develop a viable competitor around Intuitive’s strong patent foothold [18]. The Da Vinci itself was initially approved under a 510(k) application based on a complex web of component-level substantial equivalence, most likely supplemented by additional testing. The Da Vinci was chosen because its function is well documented. Moreover, there is a non-negligible number of malfunctions that have occurred during the use of these types of devices [19].”

– Page 16: In the sentence “For this paper, most data….”; please define what the “data” are.

We have changed this sentence to “For this paper information on the devices within the Da Vinci Si Model predicate network”

– Page 16: In the sentence, “…were collected through 510(k) approval database”; a “the” appears to be missing, and “approval” should be “clearance”.

We have changed this sentence to read: through the 510(k) clearance database

– Page 16: In the sentence “When either of these databases did not provide the needed information, we also conducted targeted search…”; the “either” should be “neither”, and an “a” is missing before “targeted”.

In shortening the paper, we deleted this particular sentence. It now reads “Information on the devices within the Da Vinci Si Model predicate network collected through the 510(k) clearance database. We also searched other data bases using the search function on the FDA web site, which returns results from all FDA publications, including database information, conference presentations, regulations, and internal memos.”

–Page 17: Portions of the figure description could be moved from the methods in to a caption for the figure to improve brevity.

–Page 18: In the sentence “….increased number of predicates appear make traditional tree diagrams unwieldy”; a “to” appears to be missing”.

We have changed the sentence: “The numbers of predicates in the Da Vinci Si predicate ancestry network made traditional tree diagrams too unwieldy, so an alternate diagram structure known as a network map was used to display predicate relationships within the clearance ancestry.”

– Page 18: The majority of the paragraph under the header, “Measuring Predicate Creep” could be deleted as this set-up has already been completed in the Introduction/Literature Review (which is the more appropriate place for it).

This method description has been condensed (and some of it deleted given the paper refocusing).

– Page 21: The sentence “Since the successful performance of each predicate device in the market is part of the body of evidence to support the safety claims of the new device, a smaller number of unique devices with market performance data effectively reduces the level of assurance of safety for the subject device” – does not make sense. Per the IOM report and the authors’ own arguments in the introduction, the 510(k) pathway does not support justifications of safety and effectiveness. Additionally, per the authors’ arguments, shouldn’t the quality of the substantial equivalence determination rather than the number of unique devices be the real determinant of safety?

In shortening the paper, this sentence was removed. However, it does raise an interesting question that could be explored empirically.

– Page 21: In the section on “Direct Comparison of Technological Characteristics”; it is not clear what makes “the degree of technological differences ‘striking’” or (per page 22) what constitutes a “large change in technology within a single predicate relationship”. (Some of this is of course intuitive, but it is worth spelling out for the reader – because by the FDA’s guidance, the subject and predicate are not required to be identical, so some differences are expected”). For example, the authors could consider interweaving a discussion of the clinical/non-clinical evidence here.

We have removed this section.

– Page 24: What is the difference between “device instances” and “unique devices”?

This sentence has been reworded for clarity “This network map included 2618 device instances (i.e. a device is cited as a predicate), with a total of 50 unique devices.”

– Page 24: “Grouping by regulatory mechanisms”; devices can have their product codes reclassified over time; was that the case for any of the devices in the sample?

Yes, there were some products for which product codes were changed. These are listed in the table as having multiple product codes. For example, Guidant Microwave Ablation System was first coded as NAY and then given a subsequent product code OCL. For this particular device, which was cleared by the FDA in 2002, a letter notifying the manufacturer of this change was sent in 2008. Not all devices with multiple codes, however, had similar documentation as to when the code was switched.

– Page 29: The sentence “…methods of predicate analysis; The” I think the semicolon is supposed to be a period.

This sentence was changed.

– Page 30: In the sentence “the 510(k) process was officially implemented via the 1990 Safe Medical Devices Act” – technically the pathway was established as part of the 1976 device amendments. Additionally, this paragraph could be potentially condensed, as the use of these different forms is a moot point since digital copies anyway are not available until ~2000.

We now simply state: Due to changes in regulation over time, the level of information available in the 510(k) database varied widely based on the date of approval.

– Page 32: The first paragraph under “Implications for Research” should be moved to the limitations section (as it is essentially a question of generalizability)

You are correct – it has been moved and merged into the limitation section

– Page 34: The sentence “Although devices with larger…” is grammatically off (the clause in the middle about Da Vinci creates an awkward break”).

We have changed this to read: “Devices with larger technological leaps are not necessarily unsafe or ineffective. For example, the Da Vinci has remained on the market for over 15 years without a major recall. Devices with large technological leaps, however, may possess more potential risk due to the fast-paced introduction of less-understood technologies into the marketplace.”

– Page 35: The sentence “The FDA has recently begun taking steps…” could use a citation. Also, I am not sure that “split predicates” are really the problem for “leap devices” as identified in the authors’ research.

You are absolutely correct that this paragraph was not well cited and was internally inconsistent. We ended up deleting this paragraph.

Comments on Figures

– Figure 1: While this figure is well-intentioned, I don’t think the message is quite clear to the reader. Perhaps some simple labeling like “Device 1, Device 2”, etc. could be helpful. Also perhaps worth noting somewhere here the lack of testing.

We have deleted this figure with the new structure of the paper.

– Figure 2: Would recommend reformatting this figure, given the unevenness of the red branch line. Could either draw a more symmetric shape, or could consider just highlighting the lines in colors rather than creating a box.

We have deleted this figure with the new structure of the paper.

– Figures 5, 6, 7: These figures are really nicely done, and nicely capture the regulatory effect of predicate creep. Kudos to the authors.

Thank you. We intend to be the focus of the revised paper.

Submitted filename: repsonse to reviewers.docx

 10.1371/journal.pone.0283442.r003 Decision Letter 1 Wang Quanzeng Academic Editor 2023 Quanzeng Wang This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Submission Version 1

23 Jan 2023

PONE-D-22-04558R1Identification of predicate creep under the 510(k) process: A case study of a robotic surgical devicePLOS ONE

Dear Dr. Rothenberg,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Feb 20. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Quanzeng Wang

Academic Editor

PLOS ONE

Journal Requirements:

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript. If you need to cite a retracted article, indicate the article’s retracted status in the References list and also include a citation and full reference for the retraction notice.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: N/A

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: Thank you very much for the opportunity to review a revised version of the manuscript by Lefkovich and Rothenberg. The authors should be applauded for their detailed and thorough revisions of the manuscript. I have included below some further substantive and stylistic comments for their consideration. Please note that page numbers below correspond to the 103-page document I was sent for my review, which includes both the authors’ responses, the clean version of the manuscript, and the tracked changes version of the manuscript.

Substantive:

-The updated focus on product codes is nicely done, and a good addition to the literature. The revised Figure 2 captures this quite well in particular. Kudos.

-In the discussion, I think the authors should be clear that the limitations of their work are not simply to a specific type of device (Robotic Surgery devices), but also to a device category which is highly consolidated (limited competition – which means there are fewer available predicates that likely can be cited) and is by definition composed of multiple different devices (a similar issue for devices like orthopedic implants, but perhaps less of an issue for devices such as respiratory machines or endovascular catheters)

-Introduction (page 18): the paragraph starting with “FDA has responded to these concerns…” could be deleted from the introduction. Some of this content is already in the discussion, and that is a more appropriate place for it. This change would also help shorten the introduction.

-Literature Review (page 23): Please revise this sentence: “The uniqueness of the device as an emerging technology makes it representative of the many challenges…”. It does not follow that a “unique” example can also be “representative”.

-When tracking product code divergence, did the authors also confirm that the change in product code was a decision by the manufacture, and not because of a reclassification order by FDA? As the latter would mean that it wasn’t so much an instance of “predicate creep” as it was updated regulation

-Data Analysis (page 29): “This illustrates how larger jumps in the technological complexity of devices new devices can occur through the 510(k) process” – great summary statement. Note that there seems to be a typo, with “devices” repeated twice. Otherwise, this well captures the contributions of this study.

-Discussion (page 32): The sentence “…to reduce barriers to bringing new medical devices” is not quite correct (that is perhaps a summation of the “Least Burdensome Principle”). The purpose of 510(k) is to facilitate incremental innovation in medical device development; not just reduce barriers. Please correct.

-Discussion (page 32): The sentence “..predicate creep where the new device can be guaranteed safe…”. “Guarantee” may not be appropriate here. The IOM report on 510(k) is clear that 510(k) clearance does not demonstrate safety. Would consider revising.

-Discussion (page 33): “We were also able to see the absorption of a primary predicate device function into a secondary system function” – a really important interesting insight. Nicely done.

-Limitations (page 35): In the sentence “scope of this research to robotic surgical devices”; it also may be useful to point about the fact about consolidation/monopolization of this space, which makes it harder to generalize to other product codes where there may be more device makers (and hence, more predicates)

-Implications for Research (Page 36): the point about patents is really interesting. Of note, there’s no good resource for this for device (e.g., there’s no equivalent of the FDA Orange Book for generic medicines as there is for devices).

-Implications for Policy (Page 38): The discussion about Step and Leap devices is interesting; I think the authors could provide more weight and clarity to this section by inverting the order of the current organization. First, highlight what FDA has already done, and then offer recommendations; bounces around a little too much right now.

Stylistic

-The figures appear a bit blurry in my version of the PDF. Assuming Editors will work with the authors to ensure the figures in the final version of the paper are of a higher resolution

-At several points in the paper, the authors use the word “approved” (for example, page 21 of the Literature “These predicates can have also been approved…”). 510(k) does not provide “approvals”; the appropriate term is either “clearance” or “authorization”. Please revise all references throughout

-Throughout the article, the authors alternate from “Da Vinci” and “Da Vinci Si”. Please use just one consistent name.

-In the Implications for Future Research section, consider revising instances of “should” to “could” when discussing recommendations for FDA

-Abstract (page 17): last sentence has a grammatical issue (“…discuss implications of this method finding for research and policy”). Perhaps the authors meant “methodological”?

-Literature Review (page 20): in the sentence “Substantially equivalence, however, proves only…”); I’d advise the authors to find an alternative word for “proves”. SE determinations are less of a tried-and-true fact, and more of a reflection of a regulatory assessment

-Literature Review (page 20): For the sentence “It is not uncommon for a device to cite a recalled predicate”, consider citing the new study by Everhart and colleagues in JAMA 2023 which estimates the prevalence of this phenomenon to be ~5% of all 510(k) devices

-Literature Review (page 21): Seems like there’s a typo in this sentence (“…systems, and found that (a cycle.7%) 510(k) clearances did not..”). Please fix.

-Literature Review (page 23): The authors use “database” as one word and “data bases” as two words in on this page. Please stick to just one word.

-Literature Review (page 22): When discussing FOIA requests, could reference FDA’s website for this process, which provides timelines for how long this takes. Could also reference studies from the literature showing the limitations of what information FDA is willing to redact

-Implications for Policy (page 38): “…”step” devices, which are submit and approved…”. “Submit” should be “Submitted”.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

 10.1371/journal.pone.0283442.r004 Author response to Decision Letter 1 Submission Version 2

20 Feb 2023

Thank you for giving us the opportunity to revise this paper. We are particularly grateful for the attention to detail of the reviewer, and willingness to look through a 103 page document! Below you will find a detailed response to yours and the reviewers’ comments. We have also included this as a separate document, in which our responses are in italics. This may be easier to follow.

Substantive:

-The updated focus on product codes is nicely done, and a good addition to the literature. The revised Figure 2 captures this quite well in particular. Kudos.

Thank you again for your suggestion to change the focus of the paper. We agree that it improved the paper significantly.

-In the discussion, I think the authors should be clear that the limitations of their work are not simply to a specific type of device (Robotic Surgery devices), but also to a device category which is highly consolidated (limited competition – which means there are fewer available predicates that likely can be cited) and is by definition composed of multiple different devices (a similar issue for devices like orthopedic implants, but perhaps less of an issue for devices such as respiratory machines or endovascular catheters)

You are correct that this is a limit. We have added the following statement.

Moreover, given that for many years there was little or no competition to the Da Vinci Surgical Systems, we chose a device category for which there are likely to be fewer products available to cite as predicates than for products with higher levels of market competition.

-Introduction (page 18): the paragraph starting with “FDA has responded to these concerns…” could be deleted from the introduction. Some of this content is already in the discussion, and that is a more appropriate place for it. This change would also help shorten the introduction.

We agree and have taken out this paragraph

-Literature Review (page 23): Please revise this sentence: “The uniqueness of the device as an emerging technology makes it representative of the many challenges…”. It does not follow that a “unique” example can also be “representative”.

You are absolutely right, and the sentence was also a bit repetitive. We removed this sentence.

-When tracking product code divergence, did the authors also confirm that the change in product code was a decision by the manufacture, and not because of a reclassification order by FDA? As the latter would mean that it wasn’t so much an instance of “predicate creep” as it was updated regulation

We did not determine if the product classification code was made by the manufacturer or the FDA, or if it was driven by an administrative need or a technical feature. We now mention this in the limitations section. “Lastly, we did not determine if the product classification code was made by the manufacturer or the FDA, or if it was driven by an administrative reason or a technical feature.”

-Data Analysis (page 29): “This illustrates how larger jumps in the technological complexity of devices new devices can occur through the 510(k) process” – great summary statement. Note that there seems to be a typo, with “devices” repeated twice. Otherwise, this well captures the contributions of this study.

Thank you for catching this. It has been corrected.

-Discussion (page 32): The sentence “…to reduce barriers to bringing new medical devices” is not quite correct (that is perhaps a summation of the “Least Burdensome Principle”). The purpose of 510(k) is to facilitate incremental innovation in medical device development; not just reduce barriers. Please correct.

We have rewritten the sentence as “Given the purpose of the 510(k) process is to encourage incremental innovation in the medical device market by reducing regulatory barriers…..”

-Discussion (page 32): The sentence “..pre…”. “Guarantee” may not be appropriate here. The IOM report on 510(k) is clear that 510(k) clearance does not demonstrate safety. Would consider revising.

Thus, predicate creep in cases where the new device can be reasonably assured safe based on available scientific evidence, is beneficial to companies, patients, and regulators.

-Discussion (page 33): “We were also able to see the absorption of a primary predicate device function into a secondary system function” – a really important interesting insight. Nicely done.\\

Thank you

-Limitations (page 35): In the sentence “scope of this research to robotic surgical devices”; it also may be useful to point about the fact about consolidation/monopolization of this space, which makes it harder to generalize to other product codes where there may be more device makers (and hence, more predicates)

We have added this sentence:

Moreover, given that for many years there was little or no competition to the Da Vinci Surgical Systems, we chose a device category for which there are likely to be fewer products available to cite as predicates than for products with higher levels of market competition.

-Implications for Research (Page 36): the point about patents is really interesting. Of note, there’s no good resource for this for device (e.g., there’s no equivalent of the FDA Orange Book for generic medicines as there is for devices).

That is true. We have noticed this when comparing research approaches with our research group’s work on pharmaceuticals

-Implications for Policy (Page 38): The discussion about Step and Leap devices is interesting; I think the authors could provide more weight and clarity to this section by inverting the order of the current organization. First, highlight what FDA has already done, and then offer recommendations; bounces around a little too much right now.

Thank you for this suggestion. We have changed the order and it does make more sense now.

Stylistic

-The figures appear a bit blurry in my version of the PDF. Assuming Editors will work with the authors to ensure the figures in the final version of the paper are of a higher resolution

We agree. We will have to talk to the journal about this. The figures in our document are clear - but they change when uploaded into the system.

-At several points in the paper, the authors use the word “approved” (for example, page 21 of the Literature “These predicates can have also been approved…”). 510(k) does not provide “approvals”; the appropriate term is either “clearance” or “authorization”. Please revise all references throughout

Very good point. You are correct. We have changed all “approved” to “cleared”

-Throughout the article, the authors alternate from “Da Vinci” and “Da Vinci Si”. Please use just one consistent name.

Thank you for this suggestion. Except for cases where we are talking about the Da Vinci systems as a whole or another model, we have changed all mentions to Da Vinci Si

-In the Implications for Future Research section, consider revising instances of “should” to “could” when discussing recommendations for FDA

We have made this change

-Abstract (page 17): last sentence has a grammatical issue (“…discuss implications of this method finding for research and policy”). Perhaps the authors meant “methodological”?

We have changed this to read “We find that there is evidence of predicate creep using our method, and discuss implications of this method for research and policy.”

-Literature Review (page 20): in the sentence “Substantially equivalence, however, proves only…”); I’d advise the authors to find an alternative word for “proves”. SE determinations are less of a tried-and-true fact, and more of a reflection of a regulatory assessment

You are absolutely correct. We have changed this to read “Substantial equivalence, however, only supports an assessment that the device introduces no new safety hazards and functions at least as effectively as the predicate device.”

-Literature Review (page 20): For the sentence “It is not uncommon for a device to cite a recalled predicate”, consider citing the new study by Everhart and colleagues in JAMA 2023 which estimates the prevalence of this phenomenon to be ~5% of all 510(k) devices

Thank you for pointing out this article. I added this and another article in that same issue as references.

-Literature Review (page 21): Seems like there’s a typo in this sentence (“…systems, and found that (a cycle.7%) 510(k) clearances did not..”). Please fix.

This has been fixed

-Literature Review (page 23): The authors use “database” as one word and “data bases” as two words in on this page. Please stick to just one word.

We have made this correction.

-Literature Review (page 22): When discussing FOIA requests, could reference FDA’s website for this process, which provides timelines for how long this takes. Could also reference studies from the literature showing the limitations of what information FDA is willing to redact

We have added a reference to the FDA time estimate to our discussion of FOIA, which is now in the conclusion. Since the issue of redacted data was not discussed in the paper, we did not add references regarding this issue. It now reads “We did consider requesting additional data through FOIA, but decided against this route given that it would add a significant amount of time to the data gathering, with an FDA estimate of 18 to 24 months [24], and there was a precedent in prior research to rely primarily on information provided in the FDA website [11].

-Implications for Policy (page 38): “…”step” devices, which are submit and approved…”. “Submit” should be “Submitted”.

This has been corrected

Thank you again for your detailed comments. It has significantly improved the quality of this paper.

Submitted filename: Response to Reviewer 2nd revision.docx

 10.1371/journal.pone.0283442.r005 Decision Letter 2 Wang Quanzeng Academic Editor 2023 Quanzeng Wang This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Submission Version 2

9 Mar 2023

Identification of predicate creep under the 510(k) process: A case study of a robotic surgical device

PONE-D-22-04558R2

Dear Dr. Sandra,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Quanzeng Wang

Academic Editor

PLOS ONE

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: N/A

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: Thank you very much for the detailed and responsive edits; I appreciate the engagement with the reviewer & editorial feedback; and believe the manuscript will be an important contribution to the literature. No further comments from me. Congratulations to the authors.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

**********

 10.1371/journal.pone.0283442.r006 Acceptance letter Wang Quanzeng Academic Editor 2023 Quanzeng Wang This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

17 Mar 2023

PONE-D-22-04558R2

Identification of predicate creep under the 510(k) process: A case study of a robotic surgical device

Dear Dr. Rothenberg:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Quanzeng Wang

Academic Editor

PLOS ONE