Medical Devices: FDA Review Time (Letter Report, 10/30/95,
GAO/PEMD-96-2).

Pursuant to a congressional request, GAO reviewed the Food and Drug
Administration's (FDA) review of medical devices, focusing on how FDA
review time has changed from fiscal year 1989 to May 18, 1995.

GAO found that: (1) FDA review times for medical device applications
remained stable from 1989 to 1991, increased sharply in 1992 and 1993,
and dropped in 1994; (2) in 1994, the median review time for 510(k)
applications was 152 days, which was higher than the median review time
during 1989 through 1991; (3) the review time trend for original
premarket approval (PMA) applications was unclear because many
applications remained open; (4) the median review time for original PMA
applications peaked at 984 days in 1992; (5) the review time trend for
supplementary PMA applications fluctuated slightly in the first 3 years,
peaked in 1992, and declined to 193 days in 1994; (6) in many instances,
FDA placed 501(k) applications on hold while waiting for additional
information, which comprised almost 20 percent of its total elapsed
review time; and (7) the mean review time for investigational device
exemptions was 30 days.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  PEMD-96-2
     TITLE:  Medical Devices: FDA Review Time
      DATE:  10/30/95
   SUBJECT:  Agency proceedings
             Medical equipment
             Food and drug law
             Safety regulation
             Product performance evaluation
             Medical research
             Consumer protection
             Product safety
             Commodity marketing

             
**************************************************************************
* This file contains an ASCII representation of the text of a GAO        *
* report.  Delineations within the text indicating chapter titles,       *
* headings, and bullets are preserved.  Major divisions and subdivisions *
* of the text, such as Chapters, Sections, and Appendixes, are           *
* identified by double and single lines.  The numbers on the right end   *
* of these lines indicate the position of each of the subsections in the *
* document outline.  These numbers do NOT correspond with the page       *
* numbers of the printed product.                                        *
*                                                                        *
* No attempt has been made to display graphic images, although figure    *
* captions are reproduced. Tables are included, but may not resemble     *
* those in the printed version.                                          *
*                                                                        *
* A printed copy of this report may be obtained from the GAO Document    *
* Distribution Facility by calling (202) 512-6000, by faxing your        *
* request to (301) 258-4066, or by writing to P.O. Box 6015,             *
* Gaithersburg, MD 20884-6015. We are unable to accept electronic orders *
* for printed documents at this time.                                    *
**************************************************************************


Cover
================================================================ COVER


Report to the Chairman, Subcommittee on Oversight and Investigations,
Committee on Commerce, House of Representatives

October 1995

MEDICAL DEVICES - FDA REVIEW TIME

GAO/PEMD-96-2

FDA Medical Device Review Time

(973425)


Abbreviations
=============================================================== ABBREV

  CT - Computed tomography
  DCLD - Division of Clinical Laboratory Devices
  DCRND - Division of Cardiovascular, Respiratory and Neurological
     Devices
  DGRD - Division of General and Restorative Devices
  DOD - Division of Ophthalmic Devices
  DRAER - Division of Reproductive, Abdominal, Ear, Nose and Throat,
     and Radiological Devices
  FDA - Food and Drug Administration
  IDE - Investigational device exemption
  PMA - Premarket approval

Letter
=============================================================== LETTER


B-261822

October 30, 1995

The Honorable Joe Barton
Chairman, Subcommittee on Oversight
 and Investigations
Committee on Commerce
House of Representatives

Dear Mr.  Chairman: 

As you know, the Food and Drug Administration (FDA) regulates the
manufacture and marketing of medical devices in this country.  Some
criticism has been expressed that FDA's review of medical devices is
excessively lengthy and can impose inordinate delays upon the
introduction of new devices into the market.  At your request, we
examined FDA's review time and how it has changed from fiscal year
1989 to May 18, 1995.  We analyzed data provided by FDA on
applications to market new devices or to begin clinical research on
unapproved devices.  We briefed your staff on the findings of our
preliminary analysis in June 1995, and we have since requested and
received comments on these findings from FDA. 


   BACKGROUND
------------------------------------------------------------ Letter :1


      TYPES OF FDA REVIEWS
---------------------------------------------------------- Letter :1.1

Medical devices can range in complexity from a simple tongue
depressor to a sophisticated CT (computed tomography) x-ray system. 
Most of the devices reach the market through FDA's premarket
notification (or 510(k)) review process.\1 Under its 510(k)
authority, FDA may determine that a device is substantially
equivalent to a device already on the market and therefore not likely
to pose a significant increase in risk to public safety.  When
evaluating 510(k) applications, FDA makes a determination regarding
whether the new device is as safe and effective as a legally marketed
predicate device.  Performance data (bench, animal, or clinical) are
required in most 510(k) applications, but clinical data are needed in
less than 10 percent of applications.\2

An alternative mode of entry into the market is through the premarket
approval (PMA) process.  PMA review is more stringent and typically
longer than 510(k) review.  For PMAs, FDA determines the safety and
effectiveness of the device based on information provided by the
applicant.  Nonclinical data are included as appropriate.\3 However
the answers to the fundamental questions of safety and effectiveness
are determined from data derived from clinical trials.\4

FDA also regulates research conducted to determine the safety and
effectiveness of unapproved devices.  FDA approval is required only
for "significant risk" devices.\5 Applicants submit applications for
such devices to obtain an investigational device exemption (IDE) from
regulatory requirements and approval to conduct clinical research. 
For an IDE, unlike PMAs and 510(k)s, it is the proposed clinical
study that is being assessed--not just the device. 


--------------------
\1 Premarket notification is commonly called 510(k) in reference to
section 510(k) of the Federal Food, Drug, and Cosmetic Act. 

\2 510(k) applications must contain a description of the device,
description of the predicate device with which it is substantially
equivalent, proposed labeling, intended use, and directions for use. 

\3 Nonclinical data may include microbiological, toxicological,
immunological, biocompatibility, shelf life, animal, engineering
(stress, wear, fatigue) data. 

\4 For PMAs, information on the device and its components, the
manufacturing process, labeling that includes its intended use and
directions for use as well as clinical and nonclinical studies are
included in the submission. 

\5 A "significant risk" device is one that is intended as an implant,
used in supporting or sustaining human life, of substantial
importance in diagnosing, curing, mitigating or treating disease, or
otherwise prevents impairment of human health, and presents a
potential for serious risks to the health, safety, or welfare of a
subject (21 C.F.R.  812.3(m)).  For a nonsignificant risk device only
Institutional Review Board approval is required. 


      MODIFICATION OF CLEARED OR
      APPROVED APPLICATIONS FOR
      DEVICES
---------------------------------------------------------- Letter :1.2

Modifications of medical devices, including any expansion of their
labeled uses, are also subject to FDA regulation.  Applications to
modify a device that entered the market through a PMA are generally
linked to the original PMA application and are called PMA
supplements.  In contrast, modifications to a 510(k) device are
submitted as new 510(k) applications.  References may be made to
previous 510(k) applications. 


      MEASURING THE LENGTH OF FDA
      REVIEWS
---------------------------------------------------------- Letter :1.3

FDA uses several measures of duration to report the amount of time
spent reviewing applications.  In this letter, we use only three of
those measures.  The first is simply the time that elapses between
FDA's receipt of an application and its final decision on it (total
elapsed time).  The second measure is the time that FDA has the
application under its review process (FDA time).  This includes both
the time the application is under active review and the time it is in
the FDA review queue.  The amount of time FDA's review process has
been suspended, waiting for additional information from the
applicant, is our third measure (non-FDA time). 

Our measures of review time are not intended to be used to assess the
agency's compliance with time limits for review established under the
Federal Food, Drug, and Cosmetic Act (the act).\6 The time limits for
PMA, 510(k), and IDE applications are 180, 90, and 30 days,
respectively.  FDA regulations allow for both the suspension and
resetting of the FDA review clock under certain circumstances.\7

How review time is calculated differs for 510(k)s and PMAs.  If a PMA
application is incomplete, depending on the extent of the
deficiencies, FDA may place the application on hold and request
further information.  When the application is placed on hold, the FDA
review clock is stopped until the agency receives the additional
information.  With minor deficiencies, the FDA review clock resumes
running upon receipt of the information.  With major deficiencies,
FDA resets the FDA clock to zero upon receipt of the information.  In
this situation, all previously accrued FDA time is disregarded.  (The
resetting of the FDA clock can also be triggered by the applicant's
submission of unsolicited supplementary information.) The amount of
time that accrues while the agency is waiting for the additional
information constitutes non-FDA time.  For 510(k)s, the FDA clock is
reset upon receipt of a response to either major or minor
deficiencies. 

For this report, we define FDA time as the total amount of time that
the application is under FDA's review process.  That is, our measure
of FDA time does not include the time that elapses during any
suspension, but does include time that elapsed before the resetting
of the FDA clock.  The total amount of time that accrues while the
agency is waiting for additional information constitutes non-FDA
time.  (The sum of FDA and non-FDA time is our first measure of
duration--total elapsed time.)


--------------------
\6 FDA, as indicated by its own reports, has sometimes failed to meet
these time limits.  For example, in 1994, only 45 percent of its
510(k) reviews were completed within 90 days. 

\7 See 21 C.F.R.  814.37 and 814.40 for PMAs and 21 C.F.R.  807.87(k)
for 510(k)s.  The review of IDE applications is not subject to the
resetting of the FDA review clock; investigations for which IDE
applications are submitted may begin within 30 days of application
receipt if FDA fails to act (see 21 C.F.R.  812.30). 


      CLASSES AND TIERS OF MEDICAL
      DEVICES
---------------------------------------------------------- Letter :1.4

The act establishes three classes of medical devices, each with an
increasing level of regulation to ensure safety and effectiveness. 
The least regulated, class I devices, are subject to compliance with
general controls.  Approximately 40 percent of the different types of
medical devices fall into class I.  At the other extreme is premarket
approval for class III devices, which constitute about 12 percent of
the different types of medical devices.  Of the remainder, a little
over 40 percent are class II devices, and about 3 percent are as yet
unclassified.\8

In May 1994, FDA implemented a three-tier system to manage its review
workload.  Classified medical devices are assigned to one of three
tiers according to an assessment of the risk posed by the device and
its complexity.  Tier 3 devices are considered the riskiest and
require intensive review of the science (including clinical data) and
labeling.  Review of the least risky devices, tier 1, entails a
"focused labeling review" of the intended use.  In addition to the
three tiers is a group of class I devices that pose little or no risk
and were exempted from the premarket notification (510(k))
requirements of the act.\9

Under the class and tier systems, approximately 20 percent of the
different types of medical devices are exempted from premarket
notification.\10 A little over half of all the different types of
medical devices are classified as tier 2 devices.  Tiers 1 and 3
constitute 14 and 12 percent of the different types of medical
devices, respectively.\11


--------------------
\8 General controls for class I devices include registering device
manufacturing facilities, providing FDA with regularly updated lists
of marketed devices, complying with good manufacturing practices (as
established by FDA), and maintaining records and filing reports of
device-related injuries and malfunctions.  The Safe Medical Devices
Act of 1990 revised the requirements for class II devices, subjecting
them to both general and special controls.  Special controls include
performance standards, postmarket surveillance, patient registries,
and other controls as deemed necessary.  Class III devices require
clinical data to demonstrate safety and effectiveness. 

\9 These exempted devices remain subject to other requirements of the
act.  (See footnote 8.)

\10 Our 20-percent figure was determined by obtaining a frequency
distribution by tiers of the information FDA provided. 

\11 Medical devices have both a class and a tier designation
associated with them.  Although tiers were not implemented until
1994, for this report, we have applied the tier classification
retrospectively to our data to examine review time. 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :2

Review times and trends for medical device applications varied widely
over the period beginning October 1, 1988, through May 18, 1995.  For
510(k) applications submitted in a given fiscal year, the review time
remained stable over the 3 years from 1989 to 1991, then rose sharply
in 1992 and 1993 before dropping in 1994.  For 1994, the median was
152 days.  The mean time to a decision was higher, at 166 days, and
this mean will continue to grow as the remaining open cases (13
percent) are completed.\12

The review time trend for original PMAs was less clear, in part
because a large proportion of the applications have yet to be
completed.  Open cases ranged from 4 percent of 1989 to 81 percent of
1994 applications.  More than 40 percent of the 1992 and 1993
applications were still open.  The median for 1994 was undetermined
as less than 50 percent of the applications were completed.  For
1993, the median review time was 804 days. 

The review time for PMA supplements, however, fluctuated slightly in
the first 3 years, before peaking in 1992 and declining thereafter. 
The median for 1994 was 193 days as opposed to a mean of 162 days. 
Again, the mean will increase when the remaining open cases (21
percent) are closed. 

Not all the time that elapsed between an application's submission and
its final determination was spent under FDA's review process.  In
many instances, FDA had to wait for additional information.  This
non-FDA time comprised about one-fifth of total elapsed review time
for 510(k)s.  It constituted about one-fourth of total elapsed review
time for original PMAs and one-third for PMA supplements. 


--------------------
\12 We report our findings here in terms of two measures:  median
review time (that is, how long the case representing the midpoint in
review time took to complete review), and mean review time (the
average time to complete review).  The median includes all cases (so
long as at least one-half of cases submitted in a given year were
completed).  By necessity, the mean includes only those cases that
have been completed.  Both measures are reported by year of
submission, not year of decision.  For greater detail on the two
measures and the implications of their use, see pp.  12-14. 


   PRINCIPAL FINDINGS
------------------------------------------------------------ Letter :3


      PREMARKET NOTIFICATIONS
      (510(K)S)
---------------------------------------------------------- Letter :3.1

From 1989 through 1991, the median time between the submission of a
510(k) application and FDA's decision (total elapsed time) was
relatively stable at about 80 to 90 days.  The next 2 years showed a
sharp increase that peaked at 230 days in 1993.  Although the median
review time showed a decline in 1994 (152 days), it remained higher
than that of the initial 3 years.  (See figure 1.)

   Figure 1:  Median Review Time
   for 510(k)s by Fiscal Year\ a

   (See figure in printed
   edition.)

\a The median includes open cases. 

Similarly, the mean also indicated a peak in review time in 1993 and
a subsequent decline.  The mean review time increased from 124 days
in 1989 to 269 days in 1993.  In 1994, the mean dropped to 166 days;
however, this mean will increase as the 13 percent of the
applications that remained open are closed.  (See table II.1.)

Of all the applications submitted to FDA to market new devices during
the period under review, a little over 90 percent were for 510(k)s. 
Between 1989 and 1994, the number of 510(k) applications remained
relatively stable, ranging from a high of 7,023 in 1989 to a low of
5,774 in 1991.  In 1994, 6,446 applications were submitted. 

Of the 40,950 510(k) applications submitted during the period under
review, approximately 73 percent were determined to be substantially
equivalent.  (That is, the device is equivalent to a predicate device
already on the market and thus is cleared for marketing.) Only 2
percent were found to be nonequivalent, and 6 percent remained open. 
Other decisions--including applications for which a 510(k) was not
required and those that were withdrawn by the applicant--account for
the rest.  (See appendix I for details on other FDA decision
categories.)

For applications determined to be substantially equivalent, non-FDA
time--the amount of time FDA placed the application on hold while
waiting for additional information--comprised almost 20 percent of
the total elapsed time.  (See table II.7.) Figure 2 displays FDA and
non-FDA time to determine equivalency for 510(k) applications. 

   Figure 2:  Mean Time to
   Determine Equivalency for
   510(k)s by FDA and Non-FDA
   Time\a

   (See figure in printed
   edition.)

\a Open cases are not included.  The means shown will increase when
open cases are completed. 


      PREMARKET APPROVALS (PMAS)
---------------------------------------------------------- Letter :3.2

The trends in review time differed for original PMAs and PMA
supplements.  There was no clear trend in review times for original
PMA applications using either medians or means since a large
proportion of the applications had yet to be completed.  The median
time between the submission of an application and FDA's decision
(total elapsed time) fluctuated from a low of 414 days in 1989 to a
high of 984 days in 1992.  Less than 50 percent of the applications
submitted in 1994 were completed; thus, the median review time was
undetermined.  (See figure 3.)

   Figure 3:  Median Review Time
   for Original PMAs by Fiscal
   Year\ a

   (See figure in printed
   edition.)

\a The median includes open cases.  Data for 1994 are omitted because
less than 50 percent of the applications submitted had been
completed. 

Except for 1989, the means were lower than the medians because of the
large number of open cases.  The percent of applications that
remained open increased from 4 percent in 1989 to 81 percent in 1994. 
The means, then, represent the time to a decision for applications
that were less time-consuming.  When the open cases are completed,
lengthy review times will cause an increase in the means.  (See table
III.1.)

For PMA supplements, the median time ranged from 126 days to 173 days
in the first 3 years, then jumped to 288 days in 1992.  In 1993 and
1994, the median declined to 242 and 193 days, respectively.  (See
figure 4.)

   Figure 4:  Median Review Time
   for PMA Supplements by Fiscal
   Year\ a

   (See figure in printed
   edition.)

\a The median includes open cases. 

This trend was reflected in the mean review time that peaked at 336
days in 1992.  Although the mean dropped to 162 days in 1994, this is
expected to increase because 21 percent of the applications had not
been completed at the time of our study.  (See table III.7.)

Applications for original PMAs made up less than 1 percent of all
applications submitted to FDA to market new devices in the period we
reviewed.  PMA supplements comprised about 8 percent of the
applications. 

The number of applications submitted for PMA review declined each
year.  In 1989, applications for original PMAs numbered 84.  By 1994,
they were down to 43.  Similarly, PMA supplements decreased from 804
in 1989 to 372 in 1994.  (See tables III.1 and III.7.)

Of the 401 applications submitted for original PMAs, 33 percent were
approved, 26 were withdrawn, and nearly a third remained open.  The
remainder (about 9 percent) fell into a miscellaneous category.  (See
appendix I.) A much higher percentage of the 3,640 PMA supplements
(78 percent) were approved in this same period, and fewer PMA
supplements were withdrawn (12 percent).  About 9 percent of the
applications remained open, and 2 percent fell into the miscellaneous
category. 

For PMA reviews that resulted in approval, non-FDA time constituted
approximately one-fourth of the total elapsed time for original PMAs
and about one-third for PMA supplements.  The mean FDA time for
original PMAs ranged from 155 days in 1994 to 591 days in 1992. 
Non-FDA times for those years were 34 days in 1994 and 165 days in
1992.  For PMA supplements, FDA review times were lower, ranging from
a low of 105 days (1990) to a high of 202 days (1992).  Non-FDA time
for those years were 59 days (1990) and 98 days (1992), respectively. 
(See table III.13.) Figures 5 and 6 display the proportion of FDA and
non-FDA time for the subset of PMAs that were approved. 

   Figure 5:  Mean Time to an
   Approval for Original PMAs by
   FDA and Non-FDA Time\a

   (See figure in printed
   edition.)

\a Data for 1994 are omitted because less than 50 percent of the
applications submitted had been completed.  Open cases are not
included.  The mean times shown will increase when open cases are
closed. 

   Figure 6:  Mean Time to an
   Approval for PMA Supplements by
   FDA and Non-FDA Time\a

   (See figure in printed
   edition.)

\a Open cases are not included.  The mean times shown will increase
when open cases are closed. 


      INVESTIGATIONAL DEVICE
      EXEMPTIONS (IDES)
---------------------------------------------------------- Letter :3.3

For IDEs, the mean review time between submission and FDA action was
30 days, and it has not changed substantially over time.  Unlike
510(k)s and PMAs, IDEs are "deemed approved" if FDA does not act
within 30 days.  Of the 1,478 original IDE submissions from fiscal
year 1989 to 1995, 33 percent were initially approved (488) and 62
percent were denied or withdrawn (909).  The number of IDE
submissions each year ranged from a high of 264 in 1990 to a low of
171 in 1994.  (See table IV.1.)


   OBJECTIVES, SCOPE, AND
   METHODOLOGY
------------------------------------------------------------ Letter :4

Our objective was to address the following general question:  How has
the time that 510(k), PMA, and IDE applications spend under FDA
review changed between fiscal year 1989 and the present?  To answer
that question, we also looked at a subset of applications that were
approved, distinguishing the portion of time spent in FDA's review
process (FDA time) from that spent waiting for additional information
(non-FDA time).  For applications that were approved, we present the
average number of amendments that were subsequently added to the
initial application as well as the average number of times FDA
requested additional information from the applicant.  (Both of these
activities affect FDA's review time.)

We used both the median and mean to characterize review time.  We use
the median for two reasons.  First, a large proportion of the
applications have yet to be completed.  Since the median is the
midpoint when all review times are arranged in consecutive order, its
value can be determined even when some applications requiring lengthy
review remain open.  In contrast, the mean can only be determined
from completed applications.  (In this case, applications that have
been completed by May 18, 1995.) In addition, the mean will increase
as applications with lengthy reviews are completed. 

To illustrate, for applications submitted in 1993, the mean time to a
decision was 269 days for 510(k) applications that have been closed. 
However, 3 percent of the applications have yet to be decided.  If
these lengthy reviews were arbitrarily closed at May 18, 1995 (the
cutoff date for our data collection), the mean would increase to 285
days.  In contrast, the median review time (230 days) would remain
the same regardless of when these open applications were completed. 

The second reason for using the median is that the distributions of
review time for 510(k), original PMA, and PMA supplement applications
are not symmetric, that is, having about the same number of
applications requiring short reviews as lengthy reviews.  The median
is less sensitive to extreme values than the mean.  As a result, the
review time of a single application requiring an extremely lengthy
review would have considerably more effect on the mean than the
median.  Figure 7 shows the distribution for 510(k)s submitted in
1993, the most recent year in which at least 95 percent of all 510(k)
applications had been completed.  The distribution is skewed with a
mean review time of 269 days and a median review time of 222 days for
all completed applications.\13

   Figure 7:  Time to a Decision
   for 510(k)s Submitted in Fiscal
   Year 1993\a

   (See figure in printed
   edition.)

\a The number of cases in this frequency distribution is 6,101.  (The
one application with over 1,500 days was dropped from this figure.)
As of May 18, 1995, 3 percent of applications submitted had not been
completed and were not included.  When these applications are closed,
the mean will increase. 

To provide additional information, we report on the mean review times
as well as the median.  The discrepancy between the two measures
gives some indication of the distribution of review time.  When the
mean is larger than the median, as in the case of the 510(k)s above,
it indicates that a group of applications required lengthy
reviews.\14 Another reason we report the means is that, until
recently, FDA reported review time in terms of means. 

In appendix I, we provide the categories we used to designate the
different FDA decisions and how our categories correspond to those
used by FDA.  Detailed responses to our study objective are found in
tabular form in appendixes II, III, and IV for 510(k)s, PMAs, and
IDEs, respectively. 

We report our findings according to the fiscal year in which the
applications were submitted to FDA.  By contrast, FDA commonly
reports review time according to the fiscal year in which the review
was completed.\15 Although both approaches measure review time, their
resultant statistics can vary substantially.  For example, several
complex applications involving lengthy 2-year reviews submitted in
1989 would increase the average review time for fiscal year 1989 in
our statistics and for fiscal year 1991 in FDA's statistics. 
Consequently, the trend for review time based on date-of-submission
cohorts can differ from the trend based on date-of-decision cohorts. 
(See appendix V for a comparison of mean review time based on the two
methods.)

The two methods provide different information and are useful for
different purposes.  Using the date-of-decision cohort is useful when
examining productivity and the management of resources.  This method
takes into consideration the actual number of applications reviewed
in a given year including all backlogs from previous years. 
Alternatively, using the date-of-submission cohort is useful when
examining the impact of a change in FDA review policy, which quite
often only affects those applications submitted after its
implementation.\16 To minimize the effect of different policies on
review time within a cohort, we used the date-of-submission method. 

We conducted our work in accordance with generally accepted
government auditing standards between May and June 1995. 


--------------------
\13 See appendix III for the distribution of review time for original
PMAs and PMA supplements. 

\14 For original PMAs, the mean is smaller than the median.  The
smaller mean results from the large number of open cases. 
Applications requiring lengthy reviews have yet to be completed.  As
these reviews are completed, the mean will increase. 

\15 Using date-of-decision cohorts obviates the problem of open
cases.  Both means and medians can be easily determined. 

\16 FDA has indicated that it plans to include statistics on review
time based on the year of submission in its reports. 


   AGENCY COMMENTS
------------------------------------------------------------ Letter :5

Officials from FDA reviewed a draft of this report and provided
written comments, which are reproduced in appendix VI.  Their
technical comments, which have been incorporated into the text where
appropriate, have not been reprinted in the appendix. 

FDA believed that the report misrepresented the current state of the
program as the draft did not acknowledge recent changes in the review
process.  FDA officials suggested a number of explanations for the
apparent trends in the data we reported (see appendix VI).  Although
recent initiatives to improve the review process provide a context in
which to explain the data, they were outside the scope of our work. 
We were not able to verify the effect these changes have actually had
on review time.  To the extent that these changes did affect review
time, they are reflected in the review times as presented and are
likely to be reflected in future review times. 

The agency also believed that the draft did not reflect the recent
improvements in review time.  We provided additional measures of
review time in order to present the review times for the more recent
years.  We have also included more information on the difference
between the date-of-submission and date-of-decision cohorts, and we
have expanded our methodological discussion in response to points FDA
made on the clarity of our presentation.  (Additional responses to
the agency comments are included in appendix VI.)

As agreed with your office, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 30 days
after its date of issue.  We will then send copies to other
interested congressional committees, the Secretary of the Department
of Health and Human Services, and the Commissioner of Food and Drugs. 
Copies will also be made available to others upon request. 

If you or your staff have any questions about this report, please
call me at (202) 512-3092.  The major contributors to this report are
listed in appendix VII. 

Sincerely yours,

Kwai-Cheung Chan
Director of Program Evaluation
 in Physical Systems Areas


FDA DECISION CODES AND GAO'S
CATEGORIES
=========================================================== Appendix I

FDA uses different categories to specify the type of decision
rendered for 510(k)s, PMAs, and IDEs.  For our analysis, we collapsed
the multiple decision codes into several categories.  The
correspondence between our categories and FDA's are in table I.1. 



                               Table I.1
                
                 Correspondence Between GAO Categories
                         and FDA Decision Codes

Type of
application         GAO category  FDA decision code
------------------  ------------  ------------------------------------
Premarket           Equivalent    Equivalent
notification

                    Nonequivalen  Nonequivalent
                    t

                    Other         Additional information requested;
                                  applicant cannot respond within 30
                                  days

                                  Forwarded to drugs/biologics

                                  Deleted/duplicate

                                  Deleted

                                  Drug (CDER) review required

                                  Exempted by regulation

                                  General purpose article

                                  Closeout letter issued

                                  Not actively regulated

                                  Not a device

                                  Not a finished product

                                  Not a required submission

                                  Preamendment exempt

                                  Refuse to accept

                                  Reconditioner/remanufacturer

                                  Transitional device

                                  Withdrawn by applicant

Premarket approval  Approved      Approved

                    Denied        Denied

                    Withdrawn     Withdrawn

                    Other         Abandoned

                                  Converted

                                  Reclassified

                                  Other

Investigational     Approved      Approved
device exemption

                                  Approved with conditions

                                  Deemed approved and request
                                  information

                    Denied        Disapproved

                                  Refuse to accept

                    Withdrawn     Deemed approved/immediate withdraw

                                  Immediate withdrawal by FDA

                                  Withdrawn by sponsor

                    Other         Acknowledge incoming

                                  Study exempt from part 812

                                  Inadequate incoming

                                  Incomplete

                                  Product jurisdiction pending

                                  Product jurisdiction transferred

                                  No response necessary

                                  Nonsignificant risk device study

                                  Other

                                  Request for progress report

                                  Investigation terminated/
                                  inadequate/no final report

                                  Telephone response

                                  Voluntary termination requested
----------------------------------------------------------------------

PREMARKET NOTIFICATION (510(K))
TABLES II.1 - II.12
========================================================== Appendix II

Tables Page

Table II.1: Average Days to Decision on 510(k)s
and Number of Open Cases by Fiscal Year 27

Table II.2: Summary of Average Days to Decision on 510(k)s
and Number of Open Cases 28

Table II.3: Average Days to Decision and Number of Open
Cases for 510(k)s by Class of Medical Device 29

Table II.4: Average Days to Decision and Number of Open Cases
for 510(k)s by Tier of Medical Device 30

Table II.5: Average Days to Decision and Number of Open Cases
for 510(k)s by Medical Specialty of Device 31

Table II.6: Average Days to Decision and Number of Open Cases
for 510(k)s by Reviewing Division 32

Table II.7: Average Days to Equivalency for 510(k)s
by Fiscal Year 33

Table II.8: Average Days to Equivalency for 510(k)s by Class,
Tier, Medical Specialty, and Division 34

Table II.9: Average Days to Equivalency for 510(k)s by Class
of Medical Device 35

Table II.10: Average Time to Equivalency for 510(k)s by Tier
of Medical Device 36

Table II.11: Average Time to Equivalency for 510(k)s by
Medical Specialty of Device 37

Table II.12: Average Time to Equivalency for 510(k)s by
Reviewing Division 38


REVIEW TIME FOR PREMARKET
NOTIFICATION

The following tables present the data for premarket notifications, or
510(k)s, for fiscal years 1989 through May 18, 1995.  The first set
of tables (tables II.1 through II.6) presents the time to a
decision--from the date the application is submitted to the date a
decision is rendered. 

We first present a summary table on the time to a decision by fiscal
year (table II.1).  The grand total for the number of applications
includes open cases--that is, applications for which there had not
been any decision made as of May 18, 1995.  As the distribution for
time to a decision is not symmetric (see figure 1 in the letter), we
present the means and percentiles to characterize the distribution. 
(The means and percentiles do not include open cases.)

The second table is a summary of the time to a decision by class,
tier, medical specialty of the device, and reviewing division (table
II.2).  The next four tables (II.3 through II.6) provide the details
for these summary tables.  The totals in these tables include only
applications for which a decision has been rendered. 

The class, tier, and medical specialty of some of the devices have
yet to be determined and are designated with N/A.  Medical
specialties other than general hospital or general and plastic
surgery include anesthesiology; cardiovascular; clinical chemistry;
dental; ear, nose, and throat; gastroenterology/urology; hematology;
immunology; microbiology; neurology; obstetrics/gynecology;
ophthalmic; orthopedic; pathology; physical medicine; radiology; and
clinical toxicology. 

The five reviewing divisions in FDA's Center for Devices and
Radiological Health are Division of Clinical Laboratory Devices
(DCLD); Division of Cardiovascular, Respiratory and Neurological
Devices (DCRND); Division of General and Restorative Devices (DGRD);
Division of Ophthalmic Devices (DOD); and Division of Reproductive,
Abdominal, Ear, Nose and Throat, and Radiological Devices (DRAER). 

The second set of tables (tables II.7 through II.12) presents the
mean time to determine equivalency.  We provide the means for total
FDA time, non-FDA time, and total elapsed time.  FDA time is the
total amount of time the application was under FDA review including
queue time--the time to equivalency without resetting the FDA review
clock.  The total elapsed time, the duration between the submission
of the application and FDA's decision, equals the sum of the FDA and
non-FDA time. 

We deleted cases that had missing values or apparent data entry
errors for the values relevant to calculating FDA and non-FDA time. 
Therefore, the total number of applications determined to be
equivalent in this group of tables differs from that in the first
set.  Again, we have two summary tables, followed by four tables
providing time to determine equivalency by class, tier, medical
specialty, and reviewing division (tables II.7 through II.12). 




   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)


PREMARKET APPROVAL TABLES III.1 -
III.18
========================================================= Appendix III

Tables Page

Table III.1: Average Days to Decision on Original PMAs by Fiscal
Year and Number of Open Cases 47

Table III.2: Summary of Average Days to Decision on PMAs and
Number of Open Cases 48

Table III.3: Average Days to Decision on Original PMAs by Class
of Medical Device and Number of Open Cases 49

Table III.4 Average Days to Decision on Original PMAs by Tier
of Medical Device and Number of Open Cases 50

Table III.5: Average Days to Decision on Original PMAs by
Medical Specialty of Device and Number of Open Cases 51

Table III.6: Average Days to Decision on Original PMAs by
Reviewing Division and Number of Open Cases 52

Table III.7: Average Days to Decision on PMA Supplements by
Fiscal Year and Number of Open Cases 53

Table III.8: Summary of Average Days to Decision on PMA
Supplements and Number of Open Cases 54

Table III.9: Average Days to Decision on PMA Supplements by
Class of Medical Device and Number of Open Cases 55

Table III.10: Average Days to Decision on PMA Supplements by
Tier of Medical Device and Number of Open Cases 56

Table III.11: Average Days to Decision on PMA Supplements by
Medical Specialty of Device and Number of Open Cases 57

Table III.12: Average Days to Decision on PMA Supplements by
Reviewing Division and Number of Open Cases 58

Table III.13: Average Days to Approval for PMAs by Fiscal Year 59

Table III.14: Average Days to Approval for PMAs by Class, Tier,
Medical Specialty, and Reviewing Division 60

Table III.15: Average Days to Approval for PMAs by Fiscal Year
and Class of Medical Device 61

Table III.16: Average Days to Approval for PMAs by Fiscal Year
and Tier of Medical Device 62

Table III.17: Average Days to Approval for PMAs by Fiscal Year
and Medical Specialty of Device 63

Table III.18: Average Days to Approval for PMAs by Fiscal Year
and Reviewing Division 64


REVIEW TIME FOR PREMARKET APPROVAL

In reviewing a PMA application, FDA conducts an initial review to
determine whether the application contains sufficient information to
make a determination on its safety and effectiveness.  A filing
decision is made--filed, filed with deficiencies specified, or not
filed--based on the adequacy of the information submitted.  The
manufacturer is notified of the status of the application at this
time, especially since deficiencies need to be addressed. 

As part of the substantive review, a small proportion of PMA
applications are also reviewed by an advisory panel.\1 These panels
include clinical scientists in specific medical specialties and
representatives from both industry and consumer groups.  The advisory
panels review the applications and provide recommendations to the
agency to either approve, deny, or conditionally approve them.  FDA
then makes a final determination on the application. 

To examine in greater detail those cases where the intermediate
milestones were applicable, we calculated the average duration
between the various dates--submission, filing, panel decision, and
final decision.  The number of applications differs for each of the
milestones as not all have filing or panel dates.  (See figure
III.1.)

   Figure III.1:  Average Days to
   Decision on Premarket Approvals
   by Milestone by Year, October
   1, 1988 - May 18, 1995

   (See figure in printed
   edition.)

 

The following tables present information on review time for PMA
applications for fiscal years 1989 through 1995.  Original PMA
applications are distinguished from PMA supplements.  Some
observations were deleted from our data because of apparent data
entry errors.  The first set of tables (tables III.1 through III.6)
presents the time to a decision for original PMAs--from the date the
application is submitted to the date a decision is rendered.  The
second set of tables (tables III.7 through III.12) provides similar
information, in the same format, for PMA supplements. 

We first present a summary table on the time to a decision by fiscal
year (tables III.1 and III.7).  Again, the grand total for the number
of applications includes the number of open cases--that is,
applications for which there had not been any decision made as of May
18, 1995.  As with 510(k)s, the distributions of time to a decision
for original PMAs and PMA supplements are not symmetric.  Thus we
report means and percentiles to characterize these distributions. 
(These means and percentiles do not include open cases.)

Figure III.2 presents the distribution for original PMAs submitted in
1989, the most recent year for which at least 95 percent of the
applications had been completed.  Figure III.3 presents the
distribution for PMA supplements submitted in 1991, the most recent
year with at least a 95-percent completion date. 

   Figure III.2:  Time to a
   Decision for Original PMAs
   Submitted in Fiscal Year 1989\a

   (See figure in printed
   edition.)

\a The number of cases in this frequency distribution is 81.  As of
May 18, 1995, about 4 percent of the applications submitted had not
been completed and were not included. 

   Figure III.3:  Time to a
   Decision for PMA Supplements
   Submitted in Fiscal Year 1991\a

   (See figure in printed
   edition.)

\a The number of cases in this frequency distribution is 577.  As of
May 18, 1995, 3 percent of the applications submitted had not been
completed and were not included. 


The second table is a summary of the time to a decision by class,
tier, relevant medical specialty of the device, and reviewing
division (tables III.2 and III.8).  The two summary tables are
followed by four tables (tables III.3 through III.6 and III.9 through
III.12) presenting the details by class, tier, medical specialty, and
reviewing division.  The totals in these tables include only
applications for which a decision has been rendered. 

The class, tier, and medical specialty of some of the devices have
yet to be determined and are designated with N/A.  Medical
specialities other than cardiovascular or ophthalmic include
anesthesiology; clinical chemistry; dental; ear, nose, and throat;
gastroenterology/urology; general and plastic surgery; general
hospital; hematology; immunology; microbiology; neurology;
obstetrics/gynecology; orthopedic; pathology; physical medicine;
radiology; and clinical toxicology. 

The third set of tables provides information on the time to an
approval, for both original PMAs and PMA supplements (tables III.13
through III.18).  Four different measures of duration are
provided--total FDA time, non-FDA time, total elapsed time, and FDA
review time.  Total FDA time is the amount of time the application is
under FDA's review process.  Non-FDA time is the time the FDA clock
is suspended waiting for additional information from the applicant. 
The total elapsed time, the duration from the date the application is
submitted to the date of FDA's decision, equals the sum of total FDA
and non-FDA time.  FDA review time is FDA time for the last
cycle--excluding any time accrued before the latest resetting of the
FDA clock. 

Again, we first provide a summary table for time to an approval by
fiscal year (table III.13).  In this table, we also provide the
number of amendments or the number of times additional information
was added to the initial submission.  Not all amendments were for
information requested by FDA as can be seen from the number of
requests for information. 

Table III.13 is followed by a summary by class, tier, medical
specialty, and reviewing division (table III.14).  Tables III.15
though III.18 provide the details for these two summary tables. 




   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)


--------------------
\1 Of the 401 original PMAs, 87 (22 percent) were reviewed by panels. 
Of the 3,640 PMA supplements, only 9 (0.2 percent) received panel
review. 


INVESTIGATIONAL DEVICE EXEMPTION
TABLES IV.1 - IV.6
========================================================== Appendix IV

Tables Page

Table IV.1: Average Days to Decision on Investigational Device
Exemptions by Fiscal Year 67

Table IV.2: Summary of Average Days to Decision on
Investigational Device Exemptions 68

Table IV.3: Average Days to Decision on Investigational Device
Exemptions by Class of Medical Device 69

Table IV.4: Average Days to Decision on Investigational Device
Exemptions by Tier of Medical Device 70

Table IV.5: Average Days to Decision on Investigational Device
Exemptions by Medical Specialty of Device 71

Table IV.6: Average Days to Decision on Investigational Device
Exemptions by Reviewing Division 72


REVIEW TIME FOR INVESTIGATIONAL
DEVICE EXEMPTIONS

The following tables present the average days to a decision for
investigational device exemptions.  The first table presents the
averages for the years from October 1, 1988, through May 18, 1995. 
This is followed by summaries by class, tier, medical specialty, and
then reviewing division.  The next four tables (tables IV.3 through
IV.6) provide the details for these summary tables. 




   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)



   (See figure in printed
   edition.)


COMPARISION IN TABLE V OF
ALTERNATIVE METHODS FOR
DETERMINING AVERAGE DAYS TO
DECISION BY FISCAL YEAR
=========================================================== Appendix V

ALTERNATIVE CALCULATION OF REVIEW
TIME BY YEAR OF DECISION

We reported our findings according to the fiscal year in which the
applications were submitted to FDA (date-of-submission cohort).  By
contrast, FDA commonly reports review time according to the fiscal
year in which the review was completed (date-of-decision cohort). 
This led to discrepancies between our results and those reported by
FDA.  The following table illustrates the differences in calculating
total elapsed time by the year that the application was submitted and
the year that a decision was rendered.  Comparisons are provided for
510(k)s, PMA supplements, original PMAs, and IDEs. 

Our dataset did not include applications submitted before October 1,
1988.  Consequently, the results presented in the following table
understated the number of cases, as well as the elapsed time, when
calculated by the year of decision.  That is, an application
submitted in fiscal year 1988 and completed in 1989 would not have
been in our dataset. 



   (See figure in printed
   edition.)




(See figure in printed edition.)Appendix VI
COMMENTS FROM THE FOOD AND DRUG
ADMINISTRATION
=========================================================== Appendix V



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)

comments, which have been incorporated into the text where
appropriate but have not been reprinted here. 



(See figure in printed edition.)



(See figure in printed edition.)


The following are GAO's comments on the August 2, 1995, letter from
FDA. 

GAO COMMENTS

1.  The purpose of our review was to provide to FDA's congressional
oversight committee descriptive statistics on review time for medical
device submissions between 1989 and May 1995.  It was not to perform
an audit of whether FDA was in compliance with statutory review time,
nor to examine how changes in FDA management practices may have
resulted in shortening (or lengthening) review times.  FDA officials
suggested that a number of process changes and other factors may have
contributed to the trends we reported--for example, the increased
complexity of the typical submission that resulted from the agency's
exemption from review of certain low-risk devices.  We are not able
to verify the effect changes have actually had on review time, and it
may be that it is still too early for their impact to be definitively
assessed. 

2.  In discussing our methodology in the draft report, we noted the
differences between FDA's typical method of reporting review time
according to the year in which action on applications is finalized,
as opposed to our method of assigning applications to the year in
which they were submitted.  We also included an appendix that
compares the results of the two different approaches.  (See appendix
V.) We agree with FDA that it is important for the reader to
understand these differences and have further expanded our discussion
of methodology to emphasize this point.  (See p.  14.)

3.  We agree with FDA that our report "deals only with calculations
of averages and percentiles"--that is, with means, medians (or 50th
percentile), as well as the 5th and 95th percentiles.  However, FDA's
suggested additions do not extend beyond such descriptive statistics. 

We also agree that mean review times in the presence of numerous open
cases may not be meaningful.  For this reason, we have included open
cases in our tables that report review time, but we have excluded
them from the calculation of means.  FDA suggests that we include
open cases in our calculation of medians.  We have adopted this
suggestion and presented our discussion of trends in terms of the
median review time for all cases.  It should be noted, however, that
including open cases increases our estimate of review time.  (For
example, including open cases raises the calculation of 510(k) median
review time from the 126 days we reported for 1994 to 152 days.)

Figure VI.1 depicts the relationship among the three measures of
elapsed time for 510(k) submissions:  the mean of closed cases, the
median of closed cases, and the median of all cases.  The two
measures of closed cases reveal roughly parallel trends, with median
review time averaging some 45 days fewer than mean review time.  The
two estimates of median review time are nearly identical from 1989
through 1990 since there are very few cases from that period that
remain open.  The divergence between the two medians increases as the
number of open cases increases in recent years until 1995, when the
median, including open cases, is larger than the mean of closed
cases. 

   Figure VI.1:  Total Elapsed
   Time for 510(k) Reviews,
   1989-95:  Mean, Median for
   Closed Cases, and Median for
   All Cases\a

   (See figure in printed
   edition.)

\a Fiscal year 1995 includes actions only through May 18, 1995. 


4.  While we are unable to reproduce the calculations performed by
FDA, we agree in general with the trends indicated by FDA. 
Specifically,

  Our calculations, as presented in our draft report tables II.7 and
     following, showed a decrease from 1993 to 1994 in FDA review
     time for finding a 510(k) submission substantially equivalent. 
     By our calculation, this declined from a mean of 173 days in
     1993 to 100 days in 1994. 

  The proportion of 510(k) applications reaching initial
     determination within 90 days of submission increased from 15.8
     percent in 1993 to 32 percent in 1994 and 57.9 percent between
     October 1, 1994, and May 18, 1995. 

Clearly, since 1993, more 510(k) cases have been determined within 90
days, and the backlog of undetermined cases has been reduced. 
Because a review of the nature and complexity of the cases still open
was beyond the scope of this study, we cannot predict with certainty
whether, when these cases are ultimately determined, average review
time for 1995 cases will be shorter than for cases submitted in 1993. 

5.  FDA time was reported in our draft report tables II.7 through
II.12, and findings contrasting the differences between FDA time and
non-FDA time were also included.  Additional language addressing this
distinction has been included in the text of the report. 

6.  FDA's contends that 1989 was an atypical year for 510(k)
submissions and therefore a poor benchmark.  However, we do not
believe that starting our reporting in 1989 introduced any
significant bias into our report of the 510(k) workload.  Indeed, our
draft report concluded that the number of 510(k) submissions had
"remained relatively stable" over the 1989-94 period.  If we had
extrapolated the data from the first 7-1/2 months of 1995 to a full
year, we would have concluded that the current fiscal year would have
a substantially lower number of 510(k) submissions (16 percent to 31
percent) than any of the previous 6 years. 

7.  The tier classification was created by FDA to manage its review
workload; however, it was not our intention to evaluate or in any way
assess the use of tiers for such purposes.  The tier classification
was based on "the potential risk and complexity of the device."
Accordingly, both class and tier provide a rough indication of a
device's complexity. 

8.  We agree that our draft report aggregated original PMA
submissions and PMA supplements in summarizing its findings.  We have
now disaggregated PMA statistics throughout. 

9.  We interpret the figures presented by FDA to represent the mean
number of days elapsed between receipt (or filing) of a PMA
submission and a given month for cases that have not been decided. 
We agree with FDA that the average review time for open original PMAs
does not appear to have increased substantially since the beginning
of calendar 1994 and that the average review time has decreased for
PMA supplements since late 1994.  Decreasing these averages is the
product of either an increasing number of new cases entering the
system or of closing out older cases in the backlog or both.  Since
the number of PMAs (originals and supplements) submitted in recent
years has declined, the evidence suggests that the drop in average
time for pending PMA supplements resulted from eliminating lengthy
backlogged cases. 

10.  As noted earlier, assessing the impact of specific management
initiatives is beyond the scope of this report.  However, we do agree
with FDA that the approval rate for initial IDE submissions doubled
between 1994 and 1995; by our calculations, it increased from 25
percent to 54 percent.  We have not independently examined the total
time to approval for all IDEs. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================= Appendix VII

PROGRAM EVAULATION AND METHODOLOGY
DIVISION

Robert E.  White, Assistant Director
Bertha Dong, Project Manager
Venkareddy Chennareddy, Referencer
Elizabeth Scullin, Communications Analyst

*** End of document. ***