Table 1. Studies assessing and/or comparing US and EU systems for approval and post-market surveillance of medical devices.

Article (Year)	System Studied	Methodology	Results	General Conclusions
Studies of pre-market evaluation: empirical basis for device approval
Dhruva et al. [16] (2009)	US	Descriptive analysis of pre-market studies of cardiovascular devices performed for PMA process, 2000–2007	33 of 123 studies (27%) were randomized and 14% were blinded. 111 of 213 primary end points (52%) were compared with controls (31% retrospective). Of the primary end points, 88% were surrogate markers of effectiveness rather than “final” outcomes (e.g, mortality).	(1) Studies supporting cardiovascular PMAs lack some of the hallmarks of well-designed clinical trials, they may not accurately represent women, and the results may not be reported fairly.(2) All four studies focused on original PMA applications in one organ system, challenging generalizability.(3) Certain classic features of clinical trials, such as randomization and blinding, are not always possible in device trials.(4) Comparative studies not performed in EU due to lack of data availability.
Dhruva et al. [16] (2009)	US	[Same]	Gender not reported in 34 of 123 studies (28%); 41% of studies included analysis or comment on gender-specific outcomes.
Chen et al. [17] (2011)	US	[Same]	Only 20 of 123 (16%) studies included “training patients” (the first individuals in whom a device is tested). All training patients were excluded from effectiveness analyses, and 95% were excluded from safety analyses, suggesting that the data submitted to FDA may be biased in favor of the device’s effectiveness and safety.
Kramer et al. [20] (2009)	US	[Same]	Pivotal trial end points were of “high quality” for 82% of effectiveness end points and 60% of safety end points. Subject accounting was “high quality” in 77% of studies.
Studies of pre-market evaluation: timing of approval
Gollaher and Goodall [21] (2011)	US and EU	Descriptive analysis of length of pre-market evaluation and comparison of US versus EU time differences for convenience sample of devices	In the US, after 2007, mean approval times increased for devices approved via both PMA (by 75%, to 27 mo) and 510(k) (by 43%, to 4.5 mo) processes. PMA pathway devices approved on average 3 y earlier in EU, compared with 4 mo earlier for 510(k) pathway devices.	(1) EU process appears faster than FDA for high-risk devices (based on a small sample size), but only slightly faster for medium-risk devices.(2) Most FDA-approved high-risk devices are supported by advisory panels; delays in high-risk device approval are multifactorial and include multiple application amendments.
Jugo [23] (2007)	US	Descriptive analysis of 42 PMA applications submitted to FDA (2002–2007) and records of advisory committee meetings	31 of 38 applications (82%) were approved (four still pending), 29 having received positive recommendation from the advisory committee. PMA applications took an average of 1.4 y to complete, with delays due in part to numerous amendments from applicants (on average 13/application).
Jugo [24] (2008)	US	[Same]	Among the seven rejected applications in the sample, common reasons for rejection included design flaws and statistical concerns.
Jugo [23] (2008)	US and EU	Same, with additional comparison to EU regulatory history of same devices	Of the 42 devices, 23 confirmed to receive US and EU approval. Among these devices, EU approval was received on average 3.5 y earlier than US.
Studies of pre-market evaluation: classification in the US
GAO [25] (2009)	US	Descriptive analysis of device submissions to FDA (2003–2007)	13,199 submissions for Class I and II devicesvia the 510(k) process (90% cleared); 217 original and 784 supplemental PMA submissions for Class III devices (78% and 85% cleared, respectively). 342 submissions for Class III devices went through the 510(k) process; 228 (67%) cleared.	Many high-risk devices are still cleared through the 510(k) process; no analysis into why these mischaracterizations occurred.
Studies of device recalls: US
Zuckerman et al. [26] (2011)	US	Descriptive analysis of FDA recalls (2005–2009) and approval pathways of devices that were recalled	Of 113 recalls, 80 (71%) were cleared via 510(k) process while eight (7%) had been exempted from review. 13 (16%) 510(k)-cleared devices had been designated as Class III high-risk devices, potentially more appropriate for PMA pathway.	(1) Most recalls affect devices cleared via non-PMA pathways, though majority of cleared devices are not recalled.(2) Recalls of specific devices can affect very few or millions of units/patients, making the clinical impact of recalls overall more difficult to adjudicate.(3) Some recalls may not have been avoided with more thorough pre-market review.(4) Recalls poorly measure system performance, given complexities in recognizing post-market events and translating these into FDA action
Hall [28] (2010)	US	Descriptive analysis of FDA recalls and comparison to 510(k) clearances during same time period (2005–2009)	31 of 118 recalls (28%) involved automated external defibrillators and infusion pumps. 95 (81%) were approved through the 510(k) process. During this period, 99.6% of 19,873 510(k)-approved devices did not lead to safety-related recalls.
Maisel [29] (2010)	US	Descriptive analysis of FDA 510(k) submissions and recalls using time-to-recall as outcome measure	For approximately 3,000 devices cleared annually, recall rate of 1.3%–1.5%/year in first 4 y, 1.0%/year thereafter.
Villarraga et al. [30] (2007)	US	Descriptive analysis of FDA recalls (2004–2006) and reasons for recall	70 (average 23.3/y) recalls involving 184 devices were adjudicated as mechanical (37%) or electrical (19%) issues. Among recalls, 57% were caused by medium-risk devices and 20% by high-risk devices.
Battelle Memorial Institute [31] (2010)	US	Descriptive analysis of FDA recalls and comparison to 510(k) and PMA approvals during that time (2005–2010)	Recalls involved 0.16% of devices approved via 510(k) pathway and 0.85% of devices approved via PMA pathway. About half of recalls were due to pre-market design deficiencies, 29% to manufacturing problems, and 6% to labeling problems.
GAO [32] (2011)	US	Descriptive analysis of FDA recalls (2005–2009) and interviews with FDA staff	Of 3,510 device recalls, 140 (4%) were high-risk recalls involving substantial danger for patients. Interviews revealed insufficient FDA internal review of recall episodes.
Studies of device recalls: EU and US/EU comparisons
Heneghan et al. [33] (2011)	EU	Descriptive analysis of UK recalls (2006–2010) and requests to manufacturers for data on recalled devices	2,124 Field Safety Notifications and 447 Medical Device Alerts were found, covering 197 withdrawn or recalled devices. Only 2% of manufacturers provided data for evaluating the pre-market data and safety assessments.	(1) Details of pre-market evaluation and specific safety problems are rarely accessible in EU for recalled devices.(2) Timing of high-risk recalls similar in US and EU. Rates of recalls similar on an absolute scale, but not a relative scale, given greater number of device approvals in EU
Davis et al. [34] (2011)	EU and US	Descriptive analysis of devices recalled in EU (2005–2009), results matched to US recalls, and comparisons made to existing US data	The authors estimated about 105 high-risk recalls in the EU during this period, similar on an absolute scale to the US. 45% were due to pre-market problems, comparable to rates in the US. Authors matched 126 moderate- or high-risk recalls to recalls in the US. Recall notifications posted about 3 wk earlier in the EU than in the US.
Surveys
Makower et al. [35] (2010)	US and EU	Survey of industry leaders	EU process characterized by high levels of predictability (85%, versus 22% for FDA), reasonability (91% versus 25%), and transparency (85% versus 27%). Overall, 75% of respondents viewed EU experience as excellent or very good, versus 16% for FDA experience.	(1) EU viewed as a simpler, less rigorous process, but result may be related to response bias.(2) FDA data requirements and process inefficiencies considered by industry to impede innovation and limit patient access to novel devices.(3) Surveys poorly done, limited by small sample sizes
PricewaterhouseCoopers [1] (2011)	US and EU	Structured interviews with industry leaders	FDA perceived as having more rigorous and lengthier device approval process than EU.
PricewaterhouseCoopers [36] (2011)	US and EU	Survey of industry leaders	40% of respondents believed FDA rejected applications because of lack of resources. 63% reported FDA changed its opinion during the review process.