Tablet Scoring: Discussion of Guidance

and Compendial Development

Historically, scoring is not a high regulatory
priority in terms of risk management

Splitting is an atypical practice not regulated by FDA

Flat pricing policies leads to more frequent
splitting

It is apparently mandated in some instances

FDA sees complaints and confusion from
patients and pharmacists
2
Introduction and Overview

FDA seeks and evaluates data, leading to Draft
Guidance

FDA also working with USP on a General
Chapter building on other compendia such as
European Pharmacopoeia (EP)
3
Introduction and Overview

This can be topic of extremes

All segments must exactly meet all criteria for new strength

Likely not practical or appropriate, especially considering manual
manipulation

It does not matter

Quality standard linked to demonstration of safety and efficacy

One standard for all brands and generics

FDA strived for middle ground, building off QbD

concepts

Bisect bar implies the tablet can be broken

A split bisected 20 mg tablet should approximately yield 2, 10 mg
segments

Desired some means to communicate to health care practitioners
which products were evaluated to aid in their splitting decisions

Thus, the functional score

concept and label was born
4
Agenda

Tablet Scoring: Discussion of Guidance and Compendial

Development

Russell Wesdyk

(FDA)

Tablet Scoring -

Background

Anthony DeStefano

(USP)

Tablet Scoring

Current USP Status

Anthony DeStefano

(USP)

Testing of Functionally Scored Tablets

Statistical Considerations

Alex Viehmann

(FDA)

Overview of the FDA Draft Guidance

Russell Wesdyk

(FDA)

Topic Wrap-Up and Questions to ACPS-CP

Russell Wesdyk

(FDA)
5
1
Food and Drug Administration
Advisory Committee for Pharmaceutical
Sciences and Clinical Pharmacology
August 9, 2012
Tablet Scoring -

Background
Anthony DeStefano, Ph.D.
Senior Vice President, Compendial

Science
United States Pharmacopeial

Convention
2

Tablet Scoring

Background
Tablets intended for oral use are the most common
dosage form in the US and many bear score marks

Patients split tablets for many reasons including to
adjust the dose, to ease swallowing and to save
money

The presence of a score mark implies to a patient
that a tablet can be split

Patients expect that the split tablet will provide the same
quality, safety and efficacy profile as a whole tablet of
equivalent dose

Currently no standards for the performance of
subdivisions of scored tablets
3
Tablet Scoring

Under Discussion for Some Time
Issues with compliance, drug acquisition costs and
patient acceptance discussed in 1999
Relationship between tablet splitting and compliance, drug
acquisition cost, and patient acceptance. Fawell, et al., Am J
Health Syst Pharm. 1999;56(24):2542-2545
Issues extensively considered in Europe by the Dutch
National Institute for Public Health and the Environment
(RIVM) and reviewed in a paper published in 2002
Breaking of scored tablets: a review. Van Santen et al., Eur J
Pharm Biopharm. 2002;53:139-145)
4
Tablet Scoring

A Long-Term Concern for USP
Journal of the APhA: March/April 2002
Title
Lack of Medication Dose Uniformity in Commonly Split Tablets
Authors
Jaja

Teng, Clara Song, Roger Williams, James Polli
Trained analyst used single edge razor blade to split
tablets from 11 products and resulting uniformity studied
5
Results from the JAPhA

Study

Uniformity Test

Protocol

Select 30 tablets from each of 11 products (4 scored, 7 unscored)

Weigh 10 tablets

Split in half and weigh each half (razor blade)

Pass: At most one of 20 halves outside the range 85-115% (but within the
range of 75-125%) of expected weight with a %RSD of 10% or less

If two outside of 85%-115% (but within 75%-125%), repeat with 20 more
tablets. Pass: All 40 additional halve must be within 85%-115%, with
%RSD for all 60 halves 10%

Fail if 3 of the first 20 halves were outside of 85%-115% or if any half tablet
was outside the range of 75%-125%

Result

1 of 4 scored tablets passed the uniformity test

2 of 7 unscored

tablets passed

No correlation with scoring, tablet shape, or tablet surface flatness

Hand splitting of 3 scored tablets soft enough to do this was worse

Conclusion: Strong suggestion that split tablets (scored or unscored)
generally fail to meet expectations for weight variation
6
A More Recent Discussion
USP Pharamcopeial

Forum, Vol

35(6); Nov-Dec
2009, pp 1598-1611
Title: Pharmacopeial Standards for the Subdivision
Characteristics of Scored Tablets
Authors: Geoff Green, Carolyn Berg, James Polli,
Dirk Barends
Top-line Observations
Presence of a score mark implies the tablet can be
subdivided into smaller doses
Extensive literature showing scored tablets can be
difficult to break and often display large variations in
mass of the subdivided parts
In a Dutch study, 39% of patients dissatisfied with
subdivision characteristics and poorly functioning score
lines were perceived as a quality defect and could lead
to reduced patient compliance with medication
7
Review of Published US Literature
The available literature suggested three areas of importance

Accuracy of the splitting process

Ease of splitting scored tablets

Loss of mass of split scored tablets
Criteria

Studies conducted in US laboratories

Included reports about measuring subdivision accuracy for scored

tablets
8 studies satisfied both requirements

In 6 studies tablets were obtained commercially

In 1 study tablets were donated by the manufacturer

In 1 study professional samples were used
8
Accuracy of Subdivision
4 studies tested accuracy of subdivision using manual splitting
6 studies tested accuracy of subdivision using a splitting device
7 of 8 studies adapted USP Uniformity of Dosage Units <905> weight
variation criteria: all within 85%-115% of label claim and %RSD 6%
Results
Manually split

weight variation failed in 5 of 6 sets of tablets studied -

% of parts >115% of target for these 5 ranged between 12% and 55%
Splitter split

18 of 37 tablet sets studied showed % of parts >115% of
target ranged from 2% to 45%
Conclusions

The situation is comparable to that reported in other parts of the world

For many tablets on the US market, significant variation can occur in the mass
of subdivided tablet parts, regardless of the splitting method or person

Tablet splitter helps, but accuracy is still not acceptable for scored tablets

results vary widely depending on user and device

The presence of a score mark on a tablet does not necessarily imply the tablet
can be split into accurate partial doses.
9
Accuracy

Manually Split

US Market
Reference Panel Products Tablet
Shape
Result (% Parts
> 115% of ideal
mass
Matuschka

and
Graves 2001
Volunteers Sertraline

100 mg Capsule 0
Wilson et al.
2001
Elderly
Diabetics
Micronized gluburide

3 mg
Oval 12
McDevitt

et al.
2002
Volunteers HydroDIURIL

25mg Round 24
Teng, et al.
2002
Trained
Pharmacy
Student
HydroDIURIL

50 mg 40
Gliburide

5 mg Rectangle 15
Oretic

50 mg Round 55
10
Table 2. Accuracy of Subdivision of Scored Tablets on US Market Split by Splitter
Author
Panel and Splitting Device
Products
Tablet Shape(1)
Result (% of parts > 115%
of ideal mass)
Horn et al. 1999 Pharmacists Catapres

0.1 mg Round 12
Clonidine

0.1 mg
Round 43
EZ Dose tablet cutter Capoten

12.5 mg
Capsule 2
Sertraline

50 mg
Capsule 3
Tegretol

100 mg
Round 32
Pharmacists
Catapres

0.1 mg Round 22
Clonidine

0.1 mg
Round 42
Health Care Capoten

12.5 mg
Capsule 26
Logistics tablet cutter Amlodipine

5 mg
Octagon (modified) 17
Tenormin

25 mg
Round 18
Sertraline

50 mg
Capsule 0
Tegretol

100 mg
Round 9
Matuschka

and Graves 2001 Volunteers. LGS Health
Products pill cutter
Sertraline

100 mg Capsule 0
Rosenberg et al. 2002 (b)
Pharmacists Splitter not
specified
Buspar

5 mg Modified Rectangle 3
Captopril

6.25 mg
Capsule 13
Doxazosin

(Apotex) 0.5 mg
Capsule 10
Cardura

2 mg
Round 0
Luvox

50 mg
Oval 0
Glipizide

2.5 mg
Round 13
Hydrocholorothiazide

12.5 mg
Round 0
11
Metoprolol

(Caraco) 25 mg
Capsule 7
Metoprolol

(Mylan)
Round 0
Toprol

XL 25 mg
Oval 0
Oxybutynin

2.5 mg
Round 13
Zoloft 25 mg
Capsule 3
Zoloft Sample A 50 mg
Capsule 0
Zoloft Sample B 50 mg
Capsule 0
Trazodone

(Geneva) 25 mg
Round 14
Trazodone

(Mutual) 25 mg
Round 0
Effexor

25 mg
Pentagon 45
Coumadin 0.5 mg
Round 0
Teng

et al. 2002
Trained pharmacy student Hydrodiuril

50 mg Round 15
Glyburide

5 mg
Rectangle 15
Oretic

25 mg
Round 45
Oretic

50 mg
Round 20
Razor blade Zoloft 100 mg
Capsule 0
Polli

et al. 2003 Trained pharmacy student Coumadin 5 mgorientation 1 Round 0
Coumadin 5 mgorientation 2
Round 0
ACE-LIFE Pill Cutter
Furosemide

40 mgorientation 1 Round 0
Furosemide

40 mgorientation 2
Round 0
Glipizide

10 mg
Round 0
Metoprolol

50 mg
Capsule 0
Zoloft 100 mg
Capsule 0
12
Peek et al. 2002
Elderly patients using cutter
A; brand not specified
Metoprolol

50 mg Capsule Tablet portions deviated 9%
from their intended ideal
mass
Elderly patients using cutter
A; brand not specified
Warfarin

5 mg Round Tablet portions deviated 9%
from their intended ideal
mass
Elderly patients using cutter
B; brand not specified
Metoprolol

50 mg Capsule Tablet portions deviated
20% from their intended
ideal mass
Elderly patients using cutter
B; brand not specified
Warfarin

5 mg Round Tablet portions deviated
26% from their intended
ideal mass
(a) All tablets single scored on one side only.
(b) Tablet mass reported for Rosenberg et al. in the "Products" column are the ideal half tablet mass.
13
Loss of Mass
4 of 8 studies reported data on loss of mass
All tablets split in halves
Loss of mass calculated by dividing total unaccounted for mass for
all tablets split by the weight of all whole tablets
Results:

Only 3 of 117 with average % loss of mass greater than 1%
Conclusion

Consistent with other studies, most tablets, on average, lost less then 1%
of the intact tablet mass upon subdivision
14
Table 3: Loss of Mass on Subdivision of Scored Tablets on US Market
Author
Panel Splitting method (a) Product Percent Loss of Mass (b)
(Range)
McDevitt

et al. 1998 Volunteers Manual Hydrochlorothiazide 25 mg 1.06 (0 to 19.4)
Matuschka

and Graves 2001 Volunteers LGS Health Products Sertraline

100 mg 0.08 (NR)
Cutter
Volunteers
Manual Sertraline

100 mg 0.06 (NR)
Polli

et al. 2003 Trained pharmacy student
ACE-LIFE tablet cutter
Coumadin 5 mgorientation 1 0.0 (NR to 0.18)
Coumadin 5 mgorientation 2 0.5 (NR to 1.4)
Furosemide

40 mgorientation 1 0.8 (NR to 1.7)
Furosemide

40 mgorientation 2 1.3 (NR to 7.3)
Glipizide10 mg 0.08 (NR to 0.95)
Metoprolol

50 mg 0.1 (NR to 0.4)
Zoloft 100 mg 0.1 (NR to 0.3)
Teng

et al. 2002 Trained individual in
laboratory conditions
Razor blade
Zoloft (sertraline) 100 mg 0.4 (NR to 1.2)
Glyburide

5 mg 2.6 (NR to 6.7)
Hydrodiuril

(hydrochlorothiazide)
50 mg
0.8 (NR to 3.0)
Oretic

(hydrocholorthiazide) 50 mg 0.8 (NR to 2.0)
Trained individual in
laboratory conditions
Manu
Glyburide

5 mg 0.4 (NR to 1.2)
Hydrodiuril

(hydrochlorothiazide)
50 mg
0.3 (NR to 0.7)
Oretic

(hydrocholorthiazide) 50 mg 0.4 (NR to 0.5)
(a) All tablets split into halves
(b) Mean loss of mass calculated by dividing the total unaccounted mass for all tablets split by the sum of theoretical weight of all whole tablets.
NR = Not reported.
15
Ease of Subdivision
Assessment of individuals

ability to subdivide
tablets regardless of accuracy or loss of mass
Studied in detail by RIVM research group but
only 2 of 8 US studies included this attribute
Results
Wilson, et. al., elderly diabetics split glyburide

tablets: 7.7 on 10 point visual analog scale
Teng

et al., 50-mg hydrochlorothiazide tablets
were hard to split
Net

very limited US data
16

Drug Content and Weight Variation
Shaynan

W. Hill, Andrew S. Varker, Kelly Karlage

and Paul B. Myrdal, Journal of Managed Care
Pharmacy, 15(3), 2000; 253-261

Examined drug content uniformity (HPLC Assay)
and weight variation for six commonly split
medications

Warfarin

Na 5 mg, scored

Simvistatin

80 mg, not scored

Metoprolol

succinate

200 mg, not scored

Metoprolol

tartrate

25 mg, scored

Citalopram

40 mg, scored

Lisinopril

40 mg, not scored

All split with tablet splitter
17

Hill, et al., Key Findings
Whole tablets all fell within USP accuracy and %RSD range

Drug Content

43 of 180 half-tablets (23.9%) out of USP drug content range. All %RSD
less than 10.5%

5 of 180 (2.78%) out of drug target range on a weight adjusted basis

Weight Variation

23 of 180 (12.8%) half-tablets fell out of USP weight variation range.
Mean weight loss -

5 of 6 <0.6%

Scored vs. Nonscored

little difference -

Drug Content: 20 of 90 out of range scored, 23 of 90 nonscored

-

wider
range

Weight Variation: 10 of 90 scored 13 of 90 nonscored

Key Conclusions:

Weight variation is a good surrogate for content uniformity

Dose is primarily determined by ability to split the tablet
18

The Situation in Europe
The European Pharmacopoeia has studied this
issue for a number of years

Presented pharmacopoeial

standards for
subdivision performance in 2002.

Currently presents standards for accuracy of
subdivision but not ease of subdivision or loss or
mass
19
European Pharmacopoeia Standard for Scored Tablets -

PhEur

7.6 (2013)

Subdivision of Tablets
Tablets may bear a break-mark or break-marks and may be
subdivided in parts, either to ease the intake of the medicinal
product or to comply with the posology. In the latter case,
subdivision must be assessed and authorised

by the
competent authority. In order to ensure that the patient will
receive the intended dose, the efficacy of the break-mark(s)
must be assessed during the development of the product, in
respect of uniformity of mass of the subdivided parts. Each
authorised

dose must be tested using the following test.
20
European Pharmacopoeia Standard for Scored Tablets -

PhEur

7.6 (2013)

Subdivision of Tablets
Take 30 tablets at random, break them by hand and, from all
the parts obtained from 1

tablet, take 1

part for the test and
reject the other part(s). Weigh each of the 30 parts
individually and calculate the average mass. The tablets
comply with the test if not more than 1

individual mass is
outside the limits of 85

per cent to 115

per cent of the
average mass. The tablets fail to comply with the test if more
than 1

individual mass is outside these limits, or if 1

individual
mass is outside the limits of 75

per cent to 125

per cent of the
average mass.
21

FDA Guidance
Draft Guidance for Industry

Tablet Scoring:
Nomenclature, Labeling, and Data for Evaluation

Issued
August, 2011

Provides guidelines and criteria for assessing
characteristics of scored tables during development

Proposes the nomenclature functional score

for tablets
meeting the criteria

Consistent with European Pharmacopoeia guideline

contains drug development guidelines and acceptance
criteria

FDA Guidance provides a pathway for manufacturers to
demonstrate functionality of scoring

QbD

risk-based approach
22
FDA Guidance and USP Standard -

USPs

Role
Develop post-release testing requirements for tablets labeled
Functional Score

to show they perform as expected
throughout their shelf life
Provide a means to confirm quality of functional scoring
Specific tests
Acceptance criteria
Application of USP standard will be triggered by FDA
approved labeling
23
1
Food and Drug Administration
August 9, 2012
Advisory Committee for Pharmaceutical
Sciences and Clinical Pharmacology
Tablet Scoring

Current USP Status
Anthony DeStefano, Ph.D.
Senior Vice President, Compendial

Science
United States Pharmacopeial

Convention
2
Topics
USP and FDA
Expert Committee Deliberations
Assumptions
Current Focus
USP Process

Next Steps
3
USP Role Under Federal Law -

DRUGS

1820 USP

independent, national pharmacopeia

1906 Food & Drugs Wiley

Act

Feds can act if adulterated or misbranded

USP strength, quality & purity

1938 Federal Food, Drug, and Cosmetic Act (FD&C Act)

FDA application

safety

but no preapproval

USP identity (drug named in official compendium)

USP packaging & labeling

1962 FD&C Drug Amendments

FDA pre-market approval authority; safety & efficacy

FDA authority to require manufacturing controls:
GMPs -

assure safety + identity, strength, quality & purity

1997 FDA Modernization Act Amendments

USP Positron Emission Tomography (PET) standards
4
Overview

Role of USP-NF Standards

USP has produced uniform voluntary drug quality
standards for over 190 years

At first, recipes for compounding pharmacists.

Later, focus shifted to chemical formulations, identifications and
assays, for drug ingredients and finished drug products, to
foster conformity in manufacturing and dispensing drug
products

Roles of USP and FDA have changed over time, along
with changing medical science and public policy

Today, FDA enforces USPs

public standards (failure to satisfy
USP standards can cause article to be deemed adulterated

or
misbranded)

FDA also enforces manufacturers

private specifications in
approved drug applications (NDAs

and BLAs), and GMPs
5
May 2010 FDA-USP Quarterly Meeting

Russ Wesdyk, FDA, discussed the work of an Office of
Pharmaceutical Science Working Group on Tablet
Scoring

Increase in scored tablets through the years

Consumer expectation that a scored tablet

Was meant to be split

Split portions would have key quality attributes similar to those of
whole tablets of the same nominal dose

FDA Guidance under development

QbD

approach for requirements (content uniformity, other) to
support label statement (functional score)

Guidance would be a going forward

document and apply
only to those products that would be labeled as Functionally
Scored
6

2010-2015 Council of Experts
820 expert volunteers serving on 22 Expert Committees
and 67

Expert Panels

350 Expert Committee members

362 Expert Panel members*

103 FDA Liaisons
* This number does not include Expert
Committee members also serving on
Expert Panels.
7
20102015 USP Council of Experts
8
Expert Committee Deliberations
Questions
Should the standard address quality attributes for any
tablet that has been subdivided, whether scored or
not or should it mirror the FDA Guidance?
Should the chapter be:
Written as a guideline for information only (numbered
above <1000>) or
Required when called for in a USP product monograph
(numbered below <1000>)?
Should the full monograph standard be applied to the
split tablets?
If not the full standard, which procedures and criteria
should be applied?
9
Expert Committee Deliberations

Should the standard address quality attributes for any
tablet that has been subdivided, whether scored or not?

Unscored

tablets are being split

Would manufacturer be held accountable for actions of
patients and practitioners that are not addressed in the
labeling?

FDA draft Guidance provides basis for expectations of
products with approved labeling that indicates functional
scoring
10
Expert Committee Deliberations
Should the chapter be informational (numbered above
<1000>) or required if referenced in the drug product
monograph (numbered below <1000>)?
An informational chapter can be broader in scope

Address issues identified in the literature such as ease of
splitting

Provide guidelines but these would not be required

FDA Guidance and literature already provide substantial
background
Below 1000 would require limitation of scope

Specific tests (procedures and criteria)

What would trigger application (e.g., the approval of the
Functionally Scored

designation)?
11
Expert Committee Deliberations
Should the full monograph standard be applied to the split
tablets?
The going-in assumption is that the intact tablet meets all its USP
monograph requirements
From the FDA Guidance: split portions meet same testing
requirements as whole tablet of same strength
Avoid redundant testing requirements such as impurities,
identification, and assay (content uniformity for the split pieces)
Concentrate on attributes that may be affected by splitting (e.g.,
dissolution and weight variation)
12
Expert Committee Deliberations
What procedures and criteria should be applied? The draft
Guidance serves as a starting point.
USP <905> Uniformity of Dosage Forms

What is the appropriate sample size?

What aspects of uniformity are of interest?
Dissolution is specifically mentioned in the FDA draft Guidance

Approaches for immediate and modified release tablets

Immediate release by appropriate monograph test

Modified release testing by monograph test or profile
comparison with similarity factor (f2) criteria

12 split portions serve as the sample
13

Tablets labeled as functionally scored have been reviewed
by the FDA based on expectations detailed in the draft
guidance

Subdivided portions shown to meet same testing
requirements as intact tablets of the same strength

Demonstrated 90-day stability for subdivided portions

<905> Uniformity of Dosage Units

Weight variation in place of Content Uniformity (if split
portion meets 25 mg AND 25% requirement)
Assumptions
14

Current Focus
Uniformity

As an attribute of the functional scoring

Dissolution

Disintegration (when used as a surrogate for
dissolution)
15
FDA Guidance and USP Standard
USP standard will provide means to confirm quality of
functional score for drug products throughout the shelf life
of the product
Specific tests
Acceptance criteria
Application of USP standard will be triggered by FDA
approved labeling and referenced in the product
monograph
16
Documentary Standards-setting Process
17
Standards Development Process: PF

All proposed revisions required to be published in
Pharmacopeial Forum (PF)

for public review and comment

90-day comment period

Second round of notice and comment in PF can occur if changes
require substantial new compendial

requirements beyond proposal,
or if the EC or CoE

Chairperson determines reprinting is necessary
due to the significance of comments received or changes made

PF is a free, on-line only publication
18

USP Process

Next Steps
Discussion within Expert Committee

In conjunction with the appropriate FDA Liaisons, the committee will
reach conclusions on the issues outlined in the presentation and

any
others that arise during their deliberations.

Publish Draft Chapter and PF Stimuli Article in the same issue of
Pharmacopeial

Forum

Stimuli article to provide background and rationale (the why)

Draft chapter to provide procedures and criteria (the what)

90 day public comment period but will likely be posted on the
USP web site prior to official publication in PF to provide
additional time

Comments are all addressed by the committee before the standard
becomes official

Additional information gathering tools (e.g., webinars) if needed

Timing

Target publication is first half of calendar 2013.
19
Testing of Functionally Scored Tablets
Statistical Considerations
ACPS-CP
August 9, 2012
Alex Viehmann
FDA/CDER/OPS
Introduction and Objective

How to test functionally scored tablets?

Is the Uniformity of Dosage Units test
appropriate for assessing functionally scored
tablets ability to split?

Create a statistically valid sampling plan that
characterizes functionally scored tablets ability
to split
2
Committee Considerations

USP <905> Uniformity of Dosage Units

Two sided tolerance interval

Controls for a proportion to be within specified limits

Two one-sided tolerance interval approach

Controls for a proportion of lot to be above a lower limit

Controls for a proportion of lot to be below an upper limit

Attribute sampling plan

Go / no-go decision on each unit

Acceptable Quality Level (AQL) and an Unacceptable
Quality Level (UQL).
3
USP <905>
Table 1. Uniformity of Dosage Units Test Procedure
All measurements of dosage units and criteria values are in percentage label claim
(%LC).
At each stage calculate the sample average, X , and the sample standard deviation s.
Stage Number tested Pass stage if:
S
1
10 |M -X| + 2.4s 15.0, where M is defined below.
S
2
20

i) |M -X| + 2.0s 15.0 using all 30 results (S
1
+ S
2
)

ii) No dosage unit is outside the maximum allowed range of 0.75*M
to 1.25*M.

M is defined as follows:
If T is less than or equal to 101.5%LC, and
(i) If X is less than 98.5%LC, then M =98.5%LC.
(ii) If X is between 98.5 and 101.5%LC, then M =X .
(iii) If X is greater than 101.5%LC, then M = 101.5%LC.
If T is greater than 101.5%LC, and
(i) If X is less than 98.5%LC, then M =98.5%LC.
(ii) If X is between 98.5 and T, then M =X .
(iii) If X is greater than T, then M = T.
T is the Target content per dosage unit at the time of manufacture, expressed as percentage label
claim. Unless otherwise specified in the individual monograph, T is 100.0%LC.

4
USP <905>

Apply current procedure to tablet segments

Example: Bi-sect 15 tablets; analyst has no less than 30 tablet
segments

The current procedure allows for a +/-

1.5% indifference zone.

The procedure is based on a two-sided tolerance interval approach.

An interval that contains p percent of the population measurements

N=10 / k
2

=2.4

N=30 / k
2

=2.0

k
2

is a tolerance interval factor that is affected by sample size,
desired confidence, and coverage (k
2

is specific to a two
sided tolerance interval)

K
2
is determined so that the interval covers at least a
proportion p of the population with a confidence c

Based upon the criteria of the test (k
2

value), the metrics provided are

87% confident that 91% of the population lies between 83.5-116.5%; this
is due to the 1.5% indifference zone

The second aspect of the procedure is no tablet will be outside ~ 73.9-

126.9%
5
USP <905> Acceptance Limits
Acceptance Limits
0
1
2
3
4
5
6
7
8
1
1
7
1
1
5
1
1
3
1
1
1
1
0
9
1
0
7
1
0
5
1
0
3
1
0
2
9
9
.
5
9
8
.
5
9
7
9
5
9
3
9
1
8
9
8
7
8
5
8
3
.
5
Mean
M
a
x

S
D
n=30
n=10
6
Two One-Sided Tolerance Interval

Ensures that p percent of population measurements will not fall
below a lower limit

Lower limit is pre-determined (lower specification)
X
L
= -k
1
*s (lower tolerance limit)

Ensures that p percent of population measurements will not fall
above an upper limit

Upper limit is pre-determined (upper specification)
X
U
= +k
1
*s (upper tolerance limit)
k
1
is a tolerance interval factor that is affected by sample
size, desired confidence, and coverage (k
1
is specific to a
one sided tolerance interval)
k
1
is determined so that the interval covers at least a
proportion p of the population with a confidence c
x
x
7
Two One-Sided Tolerance Interval

Compare upper specification to upper tolerance limit (X
U

)

Compare lower specification to lower tolerance limit (X
L
)

If lower tolerance limit > lower specification and upper
tolerance limit < upper specification

Y% confidence that at least P% of population lies
below upper specification

Y% confidence that at least P% of population lies
above lower specification
8
Two One-Sided Tolerance Interval

Example Stage 1

Take no more than 15 tablets and split accordingly; this will leave analyst with no
less than 30 units.

Take a random sample of 10 units.

Calculate Mean and Standard Deviation

Calculate Lower Tolerance Limit

Xbar

k
1

*s

K
1

=3.4 (95% confident that 97.5% of population lies above Lower
Tolerance Limit based on sample size of 10)

Lower Tolerance Limit

75%

Calculate Upper Tolerance Limit

Xbar + k
1

*s

K
1

=3.4 (95% confident that 97.5% of population lies below Upper
Tolerance Limit based on sample size of 10)

Upper Tolerance Limit

125%

If original 10 fail, move to Stage 2

If lot complies at Stage 1; analyst is 95% confident that at least 97.5% of lot lies
above 75%LC and 95% confident that at least 97.5% of lot lies below 125% LC.

THIS IS AN EXAMPLE ONLY

SPECIFICATION / % CONFIDENCE /
PROPORTION CAN BE ALTERED
9
Two One-Sided Tolerance Interval

Example Stage 2

Analyze remaining 20 units (30 total; Stage 1 + Stage 2)

Calculate Mean and Standard Deviation of all 30 units

Calculate Lower Tolerance Limit

Xbar

k
1

*s

K
1

=2.6 (95% confident that 97.5% of population lies above Lower
Tolerance Limit based on sample size of 30)

Lower Tolerance Limit

75%

Calculate Upper Tolerance Limit

Xbar + k
1

*s

K
1

=2.6 (95% confident that 97.5% of population lies below Upper
Tolerance Limit based on sample size of 30)

Upper Tolerance Limit

125%

If lot complies at Stage 1; analyst is 95% confident that at least 97.5% of lot
lies above 75%LC and 95% confident that at least 97.5% of lot lies below
125% LC.

THIS IS AN EXAMPLE ONLY

SPECIFICATION / % CONFIDENCE /
PROPORTION CAN BE ALTERED
10
Two One-Sided Tolerance Interval

Acceptance Limits
Max SD for TOSTI (n=10)
0.00
1.00
2.00
3.00
4.00
5.00
6.00
7.00
8.00
1
2
5
1
2
3
1
2
1
1
1
9
1
1
7
1
1
5
1
1
3
1
1
1
1
0
9
1
0
7
1
0
5
1
0
3
1
0
1 9
9
9
7
9
5
9
3
9
1
8
9
8
7
8
5
8
3
8
1
7
9
7
7
7
5
mean of 10
M
a
x
S
D
Max SD
Max SD for TOSTI (n=30)
0.00
2.00
4.00
6.00
8.00
10.00
12.00
1
2
5
1
2
3
1
2
1
1
1
9
1
1
7
1
1
5
1
1
3
1
1
1
1
0
9
1
0
7
1
0
5
1
0
3
1
0
1
9
9
9
7
9
5
9
3
9
1
8
9
8
7
8
5
8
3
8
1
7
9
7
7
7
5
mean of 30
M
a
x
S
D
Max SD
11
Tolerance Interval
Concerns

Parametric

Assume the data follow a normal distribution

Analyst bi-sects tablet and one segment is
crushed in to powder

Each granule of powder in crushed segment must be
considered a segment in the random sample

Probability of passing becomes nearly impossible
12
Attribute Sampling Plan

Why?

Easy to implement

Counting test

Non parametric

Does not assume tablet segments follow a
specific distribution

Crushing one tablet segment will not
guarantee failure
13
Attribute Sampling Plan

Development Questions?

What is the desired attribute?

What is the sample size?

What is the acceptance/rejection criteria?

What is the Acceptable Quality Level (AQL)?

Percent Defective that is the base line requirement for the quality of
the producers product. There is a high probability (e.g. 95%) of
accepting a lot that has a defect level less than or equal to the AQL.

Type I error (

risk): probability of rejecting a lot that has a defect
level equal to the AQL.

What is the Unacceptable Quality Level (UQL)?

High defect level that would be unacceptable. There is a low
probability (e.g. 10%) of accepting a lot with a defect level as high


as the UQL.

Type II error (

risk): The probability of accepting a lot with a defect
level equal to the UQL.
14
Attribute Sampling Plan

Attribute

Functionally-scored tablet breaks into the desired number
of segments, each segment containing +/-

25% LC

Accounts for all segments of the tablet

Sample size

Random sample of 30 tablets (legacy number)

Acceptance criteria

A range of acceptance numbers (Ac) were investigated

0,1,2,3,4,5,6
15
Operating Characteristic Curves

Plots the probability of accepting the lot (Y-axis) versus
the lot percent defectives (X-axis)
25 20 15 10 5 0
1.0
0.8
0.6
0.4
0.2
0.0
Lot Percent Defect ive
P
r
o
b
a
b
i
l
i
t
y

o
f

A
c
c
e
p
t
a
n
c
e

0.95 ( AQL)
0.1 ( UQL)
n sample size
c acceptance number
30 0
30 1
30 2
30 3
30 4
30 5
30 6
n c
Operati ng Characteri sti c ( OC) Curve
16
AQL / UQL
Ac AQL (95% Pa)
UQL (10% Pa; 90%
confident defect level
does not exceed)
0 0.18% 7.47%
1 1.2% 12.28%
2 2.88% 16.7%
3 4.75% 21%
4 6.78% 24.7%
5 9.2% 28.8%
6 11.22% 32.5%
17
Example

Analysis by Weight
1.

Take a random sample of 30 units
2.

Accurately weigh each tablet and record its weight
3.

For each tablet, determine the expected weight of the split
portions by dividing the whole-tablet weight by the designed
number of split portions indicated on the labeling.
4.

Manually break each tablet into the designed number of split
portions and weigh each split portion.
5.

For each tablet, determine the percent of the expected weight
represented by each of the split portion

An acceptable tablet breaks into the designed number of
segments, with each split portion having not less than 75% and
not more than 125% of the expected weight of the split tablet
portion
Acceptance Criteria: Not less than 28 of the 30 tablets are acceptable.
18
Statistical Metrics
18 16 14 12 10 8 6 4 2 0
1. 0
0. 8
0. 6
0. 4
0. 2
0. 0
Lo t Pe r c e nt De f e c t iv e
P
r
o
b
a
b
i
l
i
t
y

o
f

A
c
c
e
p
t
a
n
c
e

0. 95 ( A Q L)
0. 10 ( UQ L)
O pe r a ti ng C ha r a c te r i s ti c ( O C ) C ur v e
S a m ple S iz e = 30, A cce pta nce Num be r = 2

AQL = 2.88% (95% probability a lot with 2.88% defects will be accepted)

UQL = 16.7% (10% probability a lot with 16.7% defects will be accepted;
90% confident the lot contains no more than 16.7% defects)

Defect: A tablet that does not break in to the desired number of
segments, with each segment containing +/-

25% LC.
19
Conclusions

Parametric and Nonparametric methodologies' are
being investigated

Nonparametric approach (attribute sampling plan) was
identified as a viable option

No distribution assumption

Easy implementation

Attribute sampling plan is a valid statistical approach

Provides level of assurance on un-tested units ability to
split
20
Overview of the FDA Draft Guidance
ACPS-CP
August 9, 2012
Russell Wesdyk
Scientific Coordinator
FDA/CDER/OPS/IO
Agenda

Guidance as drafted

Overview of comments received

Areas of potential evolution

Rationale
2
Draft Guidance

Three principles

Consistent approach to CMC evaluation of scored products

Consistency of nomenclature (scored, bisected, etc)

Provide information through labeling or other means to health care
providers

Generally meet the specified requirements of the sub-divided
segment

Label product as functional score

Provides health care practitioners with information relevant to
splitting decisions

Focused on development and validation data; not end product
release requirements
3
Draft Guidance -

Guidelines

Not scored if

Below MTD

Not safe to handle

Release mechanism compromised by splitting

Stability on segments in pharmacy dispensing containers
at controlled room temperature (CRT) for 90 days

Risk assessment for justifying testing criteria

Testing using patient (manually) and mechanically split
products
4
Draft Guidance -

Criteria

UDU USP <905>

Loss on Mass (LOM); less than 3%

Friability

Dissolution

Staged for IR, MR ( matrix), MR (coated,
compressed components)
5
Comments on Draft Guidance

Approximately 20 comments to the docket

Broadly supportive of need for guidance

Brand, generic, provider

Comments both supportive and opposed to allowing
non-functional scores

Two opposed, one in favor

Multiple concerned re industry segment gaming

FDA comfortable tools exists to address this
6
Specific Comments on Draft Guidance

Approximately 11 comments on specific guidance
aspects

Stability (reduced requirements or clarifying questions)

Friability (elimination)

Burden of patient segment splitting guideline

LOM (elimination)

Dissolution (reduced requirements)
7
FDA Tablet Scoring Working Group

Membership from ONDQA, OGD, OC, OTR, and
OPS/IO

Functional representation includes chemists,
field investigators, industrial pharmacists,
practicing (community) pharmacists, and
medical officers

Considered the comments received to the
docket
8
Draft Guidance

Stability & Friability

Clarification of stability requirements recommended

Use of pharmacy container with no seal or desiccant

Recommended maintaining 90-day segment stability
guideline

Short term prescription dispensed with minimal downstream
handling

Reports of 3 months supply being dispensed

Chronic drugs / mail order

Further shipment or extensive handling

Recommended maintaining friability guideline also
based on potential for downstream handling/shipment
9
Draft Guidance

Other Specific Comments

Recommended modifying guideline to not use indicated
patient segment, but maintain guideline to test non-

mechanically split segments (as well as those
mechanically split)

Patient segment data could be requested when justified by risk
assessment

Recommended maintaining LOM guideline

May be subsumed in UDU, USP <905>

Potentially a deciding data set if non-functional scores allowed

Recommended dissolution guidelines remain unchanged
10
THANK YOU
11