10.5731/pdajpst.2017.007807Access the most recent version at doi: 511-52871, 2017 PDA J Pharm Sci and Tech

 Robert A. Schaut, Kyle C. Hoff, Steven E. Demartino, et al. Glass Designed To Prevent Cracked ContainersEnhancing Patient Safety through the Use of a Pharmaceutical  

on May 7, 2021Downloaded from on May 7, 2021Downloaded from

TECHNOLOGY/APPLICATION

Enhancing Patient Safety through the Use of aPharmaceutical Glass Designed To Prevent CrackedContainersROBERT A. SCHAUT*, KYLE C. HOFF, STEVEN E. DEMARTINO, WILLIAM K. DENSON, andRONALD L. VERKLEEREN

Corning Incorporated, Corning, NY 14831 ©PDA, Inc. 2017

ABSTRACT: An essential role of packaging material for the storage and delivery of drug products is to provideadequate protection against contamination and loss of sterility. This is especially important for parenteral containers,as lack of sterility or contamination can result in serious adverse events including death. Nonetheless, crackedparenteral containers are an important source of container integrity failures for injectable drugs and pose a serious riskfor patients. Despite significant investments in inspection technology, each year many injectable drugs are investi-gated and recalled for sterility risks associated with cracked borosilicate containers. Multiple studies and the manydifficulties in detection of cracked containers suggest that the magnitude of the public health risk is even larger thanthe recall rate would suggest. Here we show that the root cause of cracked parenteral containers (low internal energyfollowing annealing) is inherent to the glasses currently used for primary packaging of the majority of injectabledrugs. We also describe a strengthened aluminosilicate glass that has been designed to prevent cracks in parenteralcontainers through the use of an engineered stress profile in the glass. Laboratory tests that simulate common fillingline damage events show that the strengthened aluminosilicate glass is highly effective at preventing cracks.Significant safety benefits have been demonstrated in other industries from the use of special stress profiles in glasscomponents to mitigate failure modes that may result in harm to humans. Those examples combined with the resultsdescribed here suggest that a significant improvement in patient safety can be achieved through the use ofstrengthened aluminosilicate glass for parenteral containers.

LAY ABSTRACT: Cracks are small cuts or gaps in a container which provide a pathway for liquid, gas, or microbesthrough a glass container. When these defects are introduced to conventional glass containers holding injectablemedicines, the affected drug can pose serious risks to the patient receiving that medication. Specifically, the drugproduct may become less effective or even non-sterile, which could lead to bloodstream infections and, in some cases,death. This article presents a review of some previously documented cases of cracked glass containers that led topatient infections and deaths. Following a survey of common crack locations in glass vials, lab-based methods forreplicating these cracks are presented. These methods are then used to compare the fracture response of vials madefrom conventional borosilicate glass and strengthened aluminosilicate glass. The results show that stable cracks areessentially prevented (at least 31 times less likely to occur) in the strengthened aluminosilicate glass containers(relative to conventional borosilicate glass). This improvement in safety is similar to improvements alreadyengineered into other glass product designs by utilizing stored strain energy to mitigate certain failure modes.

Introduction

The primary purpose of the container closure systemfor a parenteral solution is to provide protection for

the dosage form in a safe and compatible manner.

Each year, several billion injectable doses are deliv-

ered successfully around the world (in a variety of

closure systems), providing lifesaving medications

and preventing various diseases. U.S. federal regula-

tions require that the closure system must not be

“reactive, additive, or absorptive so as to alter thesafety, strength, quality, or purity of the drug” and thatit must “provide adequate protection against foresee-able external factors in storage and use that can cause

* Corresponding Author: Corning Incorporated, SP-FR-05, Corning, NY 14831. Telephone: (607) 974-3199; e-mail: schautra@corning.com

doi: 10.5731/pdajpst.2017.007807

511Vol. 71, No. 6, November–December 2017

on May 7, 2021Downloaded from

deterioration or contamination of the drug product”(1). U.S. Food and Drug Administration (FDA) guid-ance lists common causes of deterioration and con-tamination such as exposure to light, loss of solvent,exposure to reactive gases (e.g., oxygen), absorptionof water vapor, and microbial contamination (2). Thepharmaceutical manufacturer must ensure that thechosen container closure system (e.g., glass vial �elastomeric stopper � aluminum crimp seal) providesthis protection for the specific drug product in a suit-able manner throughout its shelf life.

The failure of a container closure system to protect thedrug product can have serious and even deadly con-sequences. Container breakage can result in a misseddose, lacerations, or drug shortages. Inadequate seal-ing of container closure components may allow liquidor gas transport (ingress or egress), resulting in anineffective or adulterated dose (2). However, failuresfrom cracked containers, sealing problems (improperassembly, sealing-surface defects, dimensionally in-compatible components), and cored stopper materialsare especially serious because they can provide evenlarger openings for biological material transport (in-gress or egress). Injectable products require protectionfrom microbial contamination because they bypassmost of the body’s natural defenses (skin, mucousmembranes, etc.), permitting rapid and complete in-troduction of microbial contamination into a patient’scirculation. Accordingly, these breaches in containerclosure can lead to bacteremia, fungemia, sepsis, ordeath (3–5).

Glass is an ideal material for parenteral packagingbecause it uniquely combines several properties thatother materials cannot (6 –7). Glass containers aretransparent, hermetic, non-porous, easily formed intocomplex shapes, chemically durable against a widerange of solutions, and resist deformation under ap-plied loads. This unique combination of propertiesmakes glass the material of choice for packaging in-jectable products (8). Indeed approximately 98% (or23 billion) of parenteral containers were packaged inglass in 2012 (9). The majority of these parenteralcontainers are made of Type I borosilicate glass.

One drawback of conventional glass containers (cur-rent borosilicate and soda-lime silicate containers) isthat strength-reducing damage can be introduced dur-ing forming and handling. Cracks are one example ofsuch damage and are defined as a “fracture that pen-etrates completely through the glass [container] wall”

(10). They are classified as “critical” defects that are“likely to result in personal injury or potential hazardto the patient” because the nonconformity “compro-mises the integrity of the container, and risks micro-biological contamination of a sterile package” (10).Figure 1 shows a photograph of a cracked vial asso-ciated with an outbreak of bloodstream infections dueto cracked borosilicate glass (4).

There are many examples of cracks in glass containersobserved during investigations of bacteremia and fun-gemia, spanning six decades. For example:

a) In January 1968, five patients experienced septi-cemia after receiving 5% dextrose in water solu-tion following elective surgery. During the inves-tigation, a small crack was observed in the baseof a sixth glass bottle administered to a sixthpatient (11).

b) Around 1970, a patient received approximately200 mL of a glucose-saline infusion before it wasnoticed that the bottle was cracked and containedclumps of green fungus (12).

c) In another case around 1970, a separate patientreceived an infusion of glucose-saline containedin glass that was cracked and contained whiteclumps of Penicillium sp. (12).

d) In 1972, mold was introduced to an intravenoussolution through cracks in the glass from the gumadhesive on the back of its label (13).

Figure 1

Cracked borosilicate vial of contaminated humanalbumin, which upon injection resulted in bloodstream infections for patients (4).

512 PDA Journal of Pharmaceutical Science and Technology

on May 7, 2021Downloaded from

e) In 1975, a bottle with cracks near its base wasdiscovered because its solution was abnormallycolored and slightly turbid, yet none of the fluidhad leaked (14).

f) In 1979, a patient received 50 mL of a 5% dex-trose solution in normal saline. The infusion wasstopped when a “fungus ball” was observed as aresult of a hairline crack in the bottle. An associ-ated survey showed 24% of responding physiciansexperienced contaminated solutions within an 11year span. In 17 of 21 reported cases, the glasscontainer was visibly defective, and 11 of 12 cases(where information was available) specified “hair-line cracks” as the cause of the contamination(15).

g) Around 1990, a patient received a glucose-salineinfusion contaminated with Aerobacter cloacaeand subsequently died. Inspection of the containershowed a long crack in the vial heel, where thecrack was opened �1.9 �m. This spacing be-tween crack faces is insufficient to equalize airand vacuum pressure inside and outside the vial,but liquid can transport through capillary actionand contaminate the contents (16).

h) In 1996, patients in two distant U.S. states devel-oped similar bloodstream infections due tocracked containers from the same drug manufac-turing lot. The infections prompted a recall of theaffected lots, and the corresponding investigationnoted that on more than one occasion during themanufacturing process, cracks were introduced tothe vials when pallets of filled vials fell duringtransport by forklifts. At the end of the recall,only 6525 of 17,000 affected vials were recovered(4).

i) And in 2010, three of 11 infected neonates diedwhen their parenteral nutrition was prepared froma bottle that contained a crack (5, 17).

The occurrence rate of containers having a crack islikely underrepresented by these incidents for a varietyof reasons. These reports are limited to cases wherebacteremia or fungemia occurred, the cause was de-termined to be a crack in a glass container, and thecase was published. The symptoms associated withbacteremia, fungemia, and sepsis (fever, chills, pain,fast heart rate, mental confusion) coincide with theadverse events associated with many parenteral prod-

ucts, obscuring detection and quantification. Further-more, the recipients of certain parenteral medicationshave multiple risk factors for infection because theytypically have many medical problems and have un-dergone multiple invasive procedures, decreasing thelikelihood of investigation and identification of acracked container cause (4). These events often re-quire multiple affected patients (epidemic) to promptinvestigation, further complicating quantification ofthe frequency. The transition from multi-dose to sin-gle-dose packaging (18) should decrease the scale ofepidemics from a cracked container, but it would notprevent or reduce the occurrence of isolated cases(11). In fact, isolated cases caused by single-use con-tainers may go undetected—potentially affecting morepatients over a longer period of time. Wang et al. statethat poor tracking of broadly distributed medications(and their patients’ responses) can mask identificationof even systemically cracked populations (crack oc-currences as high as 1.5% within a lot) (4).

The pharmaceutical manufacturing industry has in-vested significantly in inspection technology, but eachyear injectable drugs continue to be recalled or inves-tigated by the manufacturer for sterility risks associ-ated with cracks. Optical inspection methods for glasscontainers rely upon special alignment of the detector,light source, and crack for detection, resulting in sig-nificant risk of false negatives. As a result, thesemethods can often fail to detect cracks effectively.Recently revised U.S. Pharmacopoeia chapters onpackage integrity testing (19) describe that classicaldye- or microbial-ingress methods are probabilisticand therefore cannot reliably ensure container closureintegrity. Online, deterministic techniques are morereliable at detecting lower frequency violations in anobjective manner. Yet within the last 5 years, therehave been at least 11 recalls due to cracked containers,affecting millions of vials (see Table I).

Considering the increasing use of parenteral medicines(20 –22), the seriousness of issues with container in-tegrity failure, the challenges with detecting cracks,and the likely underreporting, it is highly desirable toprevent these failures in an effort to improve patientsafety.

Understanding Crack Formation in ConventionalGlass Containers

Brittle materials such as glass are very strong whenplaced under compression, but they can break when

513Vol. 71, No. 6, November–December 2017

on May 7, 2021Downloaded from

TA

BL

EI

Ele

ven

Rec

alls

Issu

edfo

rIn

ject

able

Pha

rmac

euti

cals

wit

hin

the

Las

t5

Yea

rsdu

eto

aL

ack

ofSt

eril

ity

Ass

ocia

ted

wit

hC

rack

edG

lass

Con

tain

ers

Dru

gD

ate

Com

pany

Cou

ntry

Sour

ceR

ecal

lN

umbe

r

Yer

voy

10/1

8/20

14B

rist

ol-M

yers

Squ

ibb

Can

ada

http

://h

ealt

hyca

nadi

ans.

gc.c

a/re

call

-ale

rt-r

appe

l-av

is/h

c-sc

/201

4/41

861a

-eng

.php

*

Am

oxil

10/1

5/20

14G

SK

UK

http

s://

ww

w.g

ov.u

k/dr

ug-d

evic

e-al

erts

/dru

g-al

ert-

amox

il-v

ials

-for

-inj

ecti

on-5

00m

g-an

d-1g

-aug

men

tin-

intr

aven

ous-

600m

g-an

d-1-

2g-

crac

ks-i

n-vi

als-

used

-for

-pac

kagi

ng

*

Met

hotr

exat

eS

odiu

m12

/18/

2013

Tev

aC

anad

aht

tp:/

/ww

w.h

ealt

hyca

nadi

ans.

gc.c

a/re

call

-ale

rt-

rapp

el-a

vis/

hc-s

c/20

13/3

7319

a-en

g.ph

p*

Onc

aspa

r11

/15/

2013

Lin

kM

edic

alP

rodu

cts

Pty

Ltd

Aus

tral

iaht

tp:/

/app

s.tg

a.go

v.au

/PR

OD

/SA

RA

/arn

-de

tail

.asp

x?k�

RC

-201

3-R

N-0

1226

-1R

C-2

013-

RN

-012

26-1

Peg

aspa

rgas

eO

ncas

par

11/1

/201

3S

igm

a-T

auP

harm

aceu

tica

lsU

SA

http

://w

ww

.fda

.gov

/Saf

ety/

Rec

alls

/Enf

orce

men

tR

epor

ts/d

efau

lt.h

tmD

-736

-201

4

Rec

ombi

vax

HB

6/26

/201

3M

erck

US

Aht

tp:/

/ww

w.f

da.g

ov/S

afet

y/M

edW

atch

/Saf

ety

Info

rmat

ion/

Saf

etyA

lert

sfor

Hum

anM

edic

alP

rodu

cts/

ucm

3594

93.h

tm

B-0

625-

14

Cef

azol

in5/

30/2

013

San

doz

US

Aht

tp:/

/ww

w.f

da.g

ov/S

afet

y/R

ecal

ls/E

nfor

cem

ent

Rep

orts

/def

ault

.htm

D-5

97-2

013

Van

com

ycin

Hyd

roch

lori

de2/

7/20

13P

harm

aceu

tica

lP

artn

ers

ofC

anad

aC

anad

aht

tp:/

/ww

w.h

ealt

hyca

nadi

ans.

gc.c

a/re

call

-ale

rt-

rapp

el-a

vis/

hc-s

c/20

13/2

3809

r-en

g.ph

p*

Cya

noco

bala

min

9/27

/201

2F

rese

nius

Kab

iU

SA

http

://w

ww

.fda

.gov

/Saf

ety/

Rec

alls

/Enf

orce

men

tR

epor

ts/d

efau

lt.h

tmD

-029

7-20

15

Cya

noco

bala

min

Inje

ctio

n4/

2/20

12A

mer

ican

Reg

ent

US

Aht

tp:/

/ww

w.f

da.g

ov/S

afet

y/R

ecal

ls/

ucm

2985

45.h

tm*

Mid

azol

am/H

epar

in/K

etor

olac

Tro

met

ham

ine/

Ond

anse

tron

/D

iaze

pam

7/8/

2011

Hos

pira

US

Aht

tp:/

/ww

w.f

da.g

ov/S

afet

y/R

ecal

ls/E

nfor

cem

ent

Rep

orts

/ucm

2828

59.h

tm*

514 PDA Journal of Pharmaceutical Science and Technology

on May 7, 2021Downloaded from

placed in tension. The strength of glass under tensionis limited in practice by flaws, which are cuts or gapsin the atomic bonds at the glass surface. Appliedtension concentrates at the tip of flaws (i.e., checks,scratches, and even cracks), causing flaw extension(propagation) if the concentrated tension exceeds thestrength of individual atomic bonds. Figure 2 illus-trates the stress concentration at the flaw tip when thesample is experiencing uniform tension. Flaw exten-sion can stop if the applied tension decreases overtime, or if the flaw grows into a region of low tension.If the applied tension is sufficiently uniform and out-lasts the flaw propagation, the flaw will extend (boththrough the wall thickness and away from the origin),and the container will break. Cracks occur wheneverthe stress applied allows the flaw to extend through thewall thickness, but the propagation stops before thecontainer breaks. Cracks occur (and are stable) inconventional glass containers because they containonly minimal residual stresses, driven low by thermalannealing during forming.

The initial surface flaws in glass containers are intro-duced primarily from contact with other surfaces dur-ing forming, transport, filling, and handling. Thesetypes of contact are pervasive throughout the conven-tional glass container manufacture process (regardless

of manufacturer), and therefore all glass containersexhibit surface flaws (with significant variation inseverity). When these surface flaws subsequently ex-perience applied tension, they may grow into cracks orbreaks depending upon the duration of the appliedtension and the alignment of the original flaw with theapplied tension. As a result, cracks form in certainregions of the vial more frequently than other regionsdue to the severity and frequency of the original flawsand their alignment with the applied tension.

A survey of failed vials was conducted to determinethe most common locations of cracks in borosilicatecontainers. Type I borosilicate vials returned to phar-maceutical companies due to the presence of crackswere shared with Corning Incorporated for this study.The returned containers included a wide range ofdimensions (3 to 40 mL) and both molded and tubularforming methods. Modern fractographic techniques(16) were applied to determine the origin and catego-rize each crack location. The location categories areillustrated in Figure 3. In total, 81 cracked vials wereinspected, and the results showed that more than 90%of cracks occur in the heel, footprint, and body re-gions. The remaining regions (bottom, shoulder, neck,flange, seal surface) experienced cracks much lessfrequently (�10%).

Figure 2

Cross-sectional schematic of a glass article underuniform applied tension, showing low level stress(grey shading) throughout the part and high stressconcentration (dark grey shading) near the flaw tip.

Figure 3

Vial schematic labeled with common locations. Thehighlighted regions (Body, Heel, & Footprint) showwhere more than 90% of cracks are introduced.

515Vol. 71, No. 6, November–December 2017

on May 7, 2021Downloaded from

A separate fractographic survey inspected more than200 borosilicate vials that were returned to variouspharmaceutical companies as part of customer com-plaints (not necessarily cracked). The inspection cat-egorized sources of breakage, cracks, and damageintroduced by the pharmaceutical filling lines and in-field handling. This fractographic survey and rootcause investigations revealed three common modes ofcontainer damage leading to breakage and crack for-mation: bump checks, bottom lensing, and neck cracks.

Bump checks, sometimes called bruises, are typicallycrescent-shaped surface marks caused by a mechanicalbump or glass-to-glass contact. They are classifiedwithin PDA Technical Report 43 as either Minor orMajor A defects based on their size because while aminor nonconformity may “not impact product qualityor [filling] process capability,” a major defect maylead “to serious impairments or malfunction thatmakes the packaging unusable” (10). Checks by them-selves are not classified as Critical because they do notyet present a sterility risk, but larger bump checkswould be considered Major A because they can lead tocracks by subsequent applied tension. This 200 partsurvey showed that approximately 80% of the cus-tomer complaint containers exhibited bump checks inthe heel. Given their heel location and high frequency,bump checks are a leading root cause of many crackedcontainers.

Lensing is defined as a “glass container bottom that iscompletely separated from its body” and is classifiedas a Critical defect because it “compromises the in-tegrity of the container and risks microbiological con-tamination of a sterile product” (10). The mechanicalprocess that creates lensing also creates bottom lenscracks, where the footprint and heel regions contain alarge crack but the bottom does not completely sepa-rate. These cracks can be difficult to detect duringautomated or visual inspection due to the narrow crackopening widths and the three dimensional curvature ofthe vial heel and footprint. Bottom lens cracks, whilethey do not occur as frequently as bump checks, arehigh risk to patient safety because the difficult detec-tion and large crack paths provide opportunity for vastmicrobial transport.

Neck cracks were observed less frequently in bothsurveys (�10% of cracked container field returns) butmay occur during capping operations on commercialfilling lines and may be very difficult to detect. Allcracks are classified as Critical defects (10), but neck

cracks are especially concerning because they are gen-erally more difficult to detect than cracks in the bodydue to the three dimensional curvature in this regionand the potential masking by the aluminum cap. Whilethere are many mechanisms for introducing neckcracks, several involve misalignment of cappingequipment. In one case, the authors observed a fixedcrimp seal rail was configured to seal a differentgeometry than the vial being processed (i.e., config-ured for smaller neck diameter). Neck cracks wereintroduced where the crimping equipment contactedthe neck of the vial.

Once they are introduced, cracks in conventional con-tainers will remain stable until another force is ap-plied. For example, if additional tension were appliedafter a crack were present, the tension could cause thecrack to continue growing larger or could cause thecontainer to break. However, without an applied ten-sion (mechanical load, thermal gradient, pressure gra-dient, etc.), the cracks will persist indefinitely becauseother drivers to propagate cracks—such as residualstresses (10)— have been relieved or homogenized bythermal annealing during the forming process. If anapplied tension of sufficient magnitude and durationwere present, cracks would propagate to failure(breakage) rather than present a container integrityrisk.

Conventional glass containers are therefore suscepti-ble to forming cracks because (1) they contain initialsurface flaws from forming, handling, and transportdamage; (2) they experience ample tension duringfilling and handling that can cause these flaws to growinto cracks; and (3) they have insufficient tension tocause the crack to propagate and lead to breakage.Most cracks are introduced in the body, heel, andfootprint regions of a vial as a result of bump checks,while less frequent high-risk cracks (because detectionis difficult), such as neck cracks and bottom lensing,are introduced in the neck and footprint.

Three Methods for Replicating Stable Cracks inParenteral Containers

Fractographic evidence observed during inspection offield returns revealed a variety of ways that cracks areintroduced to different regions of borosilicate glassvials, and three common failure modes are explored inmore detail here. To demonstrate understanding of theconditions causing failure, controlled lab experimentsare conducted to replicate the failure, including its

516 PDA Journal of Pharmaceutical Science and Technology

on May 7, 2021Downloaded from

fractographic features. Inclusion of fractographic fea-tures in the replication process assures that the stressespresent at the moment of fracture within the glasscontainer are also being reproduced. Well-controlledmethods can then be applied to evaluate differences intest condition or container design. Here, we demon-strate the ability of these methods to replicate thefractographic features of the three failure modes andlater apply the methods to evaluate crack behavior invials with an engineered stress profile.

A bump check contains unique fractographic featuressuch as frictive transfer of material at the contact site,a crescent-shaped initial crack, and in more severecases the fracture extends in opposite directions awayfrom the origin along a curved path having a “wing-shaped” appearance. Figure 4 is a photograph of acrack in a field-returned Type I borosilicate vial thatoriginates at a bump check and extends through theglass wall. This example is representative of the typ-ical features of these defects, but not all bump checkdefects contain all of these features.

Figure 5 shows a schematic of the experimental setupused to replicate this defect: a 3 mL vial placed on a33° sine plate fixture in a 5 kN load frame. Thephotograph in Figure 5 shows a 16 mm borosilicateball positioned to contact the vial sidewall approxi-mately 2.5 mm from the footprint of the vial. A load isapplied to the vial heel by vertical displacement of theball (indicated by the arrow) to a predetermined peakload. During loading, a crack forms in the tested vialthat persists after the load is removed. Cracks areconsistently produced in the vials tested using thismethod.

Figure 4

Photograph of a borosilicate vial returned from the field with a bump check crack that extends through theglass wall that was formed during shipping to the health care facility.

Figure 5

Mechanical testing equipment designed to replicatethe features of a bump check crack by loading avial heel with a sacrificial borosilicate glass ball.

517Vol. 71, No. 6, November–December 2017

on May 7, 2021Downloaded from

Figure 6 compares optical microscope images of arepresentative field return bump check defect to arepresentative defect created by this experimentalsetup. Both checks contain evidence of frictive mate-rial transfer near the conical center of the crack as wellas wing-shaped fractures extending from the origin.Differences in features between samples will exist dueto differences in loading rate, peak load, and otherunique surface characteristics of each specimen. De-spite these differences, the figures show replication ofthe key bump check fractographic features in orienta-tion and scale, indicating that the experimental setup isable to simulate the conditions that lead to bump checkfailures.

Lensing cracks were replicated via a two-step process:surface damage introduction followed by a dynamicimpact (light strike). As illustrated in Figure 7, dam-age was first introduced to the footprint of 3 mL vialsby contact with 90 grit silicon carbide–fixed abrasivegrinding paper at a controlled normal load of 10 N.Second, the vial bottom center is lightly struck with alow-pressure pneumatic actuator putting the footprintin tension and propagating the initial damage to alensing crack.

Figure 8 compares the fractographic features of a lenscrack from a field return with those generated by thismethod. The field return sample (Figure 8a) shows thecrack propagating for about half of the circumference,before stopping in the heel or lower sidewall. Thesecracks are particularly difficult to detect due to the

narrow crack opening widths (preventing light reflec-tion from its features) and the three dimensional cur-vature of the heel and footprint region. The micro-scope images (Figure 8b) show features of the lenscracks produced with this method. The photos (at

Figure 6

Optical microscopy comparison between a field return bump check crack and the laboratory replication,showing key fractographic features: frictive material transfer, crescent-shaped initial crack, and fracturepropagating away from the origin creating “wing-shaped” features. The replication of these features demon-strates that the mechanical testing equipment is able to generate bump check cracks in borosilicate vials.

Figure 7

Schematic of two-step process of creating lenscracks. At left, the borosilicate glass vial is pressedinto silicon carbide fixed abrasive paper (texturedsurface contacting vial footprint) with a 10N load,creating initial surface damage in the footprintregion. At right, the center bottom is then lightlystruck with a low-pressure pneumatic actuator put-ting the footprint in tension and propagating theinitial damage to a lensing crack.

518 PDA Journal of Pharmaceutical Science and Technology

on May 7, 2021Downloaded from

higher magnification than 8a) show the part of thecrack system in the heel where the fracture terminatesbecause the stress was removed.

Neck cracks can be introduced by interactions withcapping equipment. This defect was previously ob-

served in borosilicate vials capped by several meth-ods, including rotating the vial against a static sealingrail system and by free-spinning displacement-drivencrimping wheels. Because the introduction of thesecrack systems is difficult to control, the containerresponse for defects in this region was evaluated usinga rotary disc (low-speed diamond blade). Figure 9illustrates how the rotary disc was used to damage theneck of a vial in a similar location to that of a misad-justed automatic crimping wheel. This method moreconsistently introduced severe damage in the vialneck, and manual inspection provided observations ofdifferences in container response. The response of thevial to this type of insult can be categorized as eithercracked (if the disc induces damage completelythrough the neck and the vial remains intact), or bro-ken (if the vial flange separates from the neck beforethe disc induces damage completely through the neck).

Dye Ingress Leak Testing

Dye ingress leak testing was employed to confirm thatperceived cracks (identified by human inspection)generated in the replication experiments completelypenetrated the glass container and presented patientsafety risk. Following the recommendations in USP

Figure 8

a) Field-returned borosilicate vials that experi-enced a lensing crack that propagated up the side-wall from additional applied stress but did notbreak the vial. b) Optical microscopy image of twovials from controlled lab experiments that repli-cated these lensing cracks.

Figure 9

a) At left, seal crimping equipment configured to shape aluminum caps on vials. When operating, the crimpingwheel (at lower left of image) is displaced a fixed distance toward the neck OD (at lower right), bending thealuminum cap under the flange. The vial being processed has larger neck OD, causing the crimping wheel todamage the glass surface in the vial neck. b) At right, rotary disc blade cutting into vial neck mimicking thelocation of damage from seal crimping equipment contact. Cutting through the vial neck with the rotary discwithout the vial breaking demonstrates the stable nature of cracks in conventional glass vials. However, if thevial breaks before the disc can completely penetrate through the neck, that vial is not at risk to cause patientharm through the injection of contaminated drug product.

519Vol. 71, No. 6, November–December 2017

on May 7, 2021Downloaded from

�1207.2�, a procedure was developed in which vialswere filled to a nominal volume with high-purity wa-ter, stoppered, and sealed (23). The sealed vials weresubmerged in a 0.1% methylene blue dye solution, andvacuum of �85 kPa was applied for 60 min. Duringthis time, the pressure differential created by the vac-uum could draw gas or fluid out of the vial throughcracks. The submerged vial system was then vented toambient pressure for 60 min, allowing the dye solutionto be drawn into vials that had previously leaked gasor liquid during the vacuum step. Considering the lowpressure used in this method, it is unlikely that signif-icant flaw extension occurs from the applied pressure.Vials were removed from the solution, rinsed thor-oughly, and manually inspected to determine if dyeingress occurred.

As recommended by USP �1207.2�, reference solu-tions and negative controls were implemented to en-sure consistency of results. Reference solutions werecreated by 11 serial dilutions of the 0.1% methyleneblue solution between 0.05% and 0.000049% for semi-quantitative inspection. The test specimens were com-pared to these references and a process blank against awhite background as shown in Figure 10. Negativecontrols were created using as-received (non-dam-aged) vials that were filled, stoppered, crimp-sealedunder identical conditions, and vacuum-tested along-side each sample set to ensure any identified leakingvials were not a result of the crimping procedure. Anyleaking of these negative controls during testing wouldhave indicated poor sealing of the glass flange, rubberstopper, and aluminum cap closure system. No failuresof the negative controls were observed in any of thetest sets.

Preventing Formation of Stable Cracks

Cracks in glass parenteral containers present a seriousrisk to patient safety (ineffective dose, lack of sterility,contamination leading to sepsis or even death) and tothe drug supply chain stability (recalls and resultingshortages) and should therefore be prevented. It is wellknown in other glass applications that strengtheningtechniques can inhibit flaw introduction and growthand also control the fracture behavior when glass doesbreak (number of fragments, their shapes, etc.). Forexample, thermally tempered safety glass used in au-tomobile and architectural applications is engineeredto have a region of compressive stress at the surfacethat hinders damage introduction and flaw extension.In addition, when safety glass does break (at higherload than non-tempered glass) the fracture propagatesto the extent of the tempered region and producessmall, harmless, approximately cubic-shaped frag-ments to prevent injury from dagger-like fragmenta-tion typical of annealed glasses (24).

Not all strengthening processes are able to impartspecial fracture behavior. Processes producing lowerlevels of strengthening (less stored strain energy) mayexhibit strength improvements over non-strengthenedglass but with the same dagger-like fragmentationpattern as annealed glass (24). Special fracture behav-ior is only enabled when the strengthening processproduces sufficient strain energy to influence the flawpropagation (25). Lawn and Marshall describe that forsamples with sufficient internal tension, a crack willtend to propagate catastrophically once it penetratesthe protective surface layer (26). This means that flawtips that remain within the compressive surface layerwill remain benign, and flaws that penetrate into the

Figure 10

Positive control population for semi-quantitative comparison of dye ingress tests. The vial set shows (from leftto right) a process blank (no methylene blue) and serial dilutions of methylene blue dye from 0.000049% to0.1%.

520 PDA Journal of Pharmaceutical Science and Technology

on May 7, 2021Downloaded from

tensile region (at the center of the wall thickness) canpropagate under the stored strain energy. A similarstrengthening method can be applied to glass contain-ers for pharmaceutical applications to prevent forma-tion of stable cracks.

Ion-exchange strengthening (or chemical tempering)is another method that can impart this threshold strainenergy. It is better suited to strengthen thin walls,glass compositions with low thermal expansion coef-ficient, and complex container geometries compared tothermal tempering (27). The process uses chemicalgradients and interdiffusion to substitute larger ionsfrom an external salt bath for smaller ions in the glassnetwork. This substitution creates a compressive stresson the surface of the container and a balancing centraltension over the thickness of the container wall, asillustrated in Figure 11. Because the stress profile is

the result of interdiffusion, exposure time and temper-ature can be controlled and monitored during the man-ufacture of the glass containers to maintain the strainenergy (or central tension) above the minimum tensionthreshold value for crack prevention.

An aluminosilicate glass, previously shown to be suit-able for pharmaceutical use (28), was strengthened bythe ion-exchange process as part of this study. A seriesof glass vials was prepared with increasing levels ofcentral tension (or stored strain energy) by varying theion-exchange process (time and temperature) to ex-plore changes in the material’s fracture response. Thecontainers were then subjected to controlled damageto observe the fracture response. Figure 12 illustratesthe decrease in frequency of cracked vials produced asa function of increasing central tension (as surrogatefor stored strain energy). The graph shows a clear“threshold” response, above which damage suitable toproduce cracks in unstressed vials causes obviousbreakage patterns (stable cracks are not maintained).

Demonstrating Prevention of Stable Cracks

To demonstrate the prevention of stable cracks withthe strengthened aluminosilicate vial, several methodswere employed: (i) replication of common crack fail-ure modes using lab tests, (ii) statistical analysis andinterpretation of the lab results, (iii) a line simulationwith a misadjusted capper to assess performance usingactual pharmaceutical processing equipment, and

Figure 11

Illustration of the engineered stress profile result-ing from ion-exchange of a thin-walled glass article,where the abscissa is the wall thickness (t, radialdirection) and stress is the ordinate. High compres-sive stress (CS) is installed at the surfaces, and itdecreases to the depth of the compressive stresslayer (DOL). The compressive strain energy in-duced by this ion-exchange process is balanced bytensile strain energy, measurable as central tension(CT).

Figure 12

Percent of containers exhibiting stable cracks fromdamage introduction, as a function of increasingcentral tension. The response shows a clear“threshold” response, above which damage (severeenough to cause cracks) causes obvious breakinginto fewer pieces with clean edges, facilitatingdetection.

521Vol. 71, No. 6, November–December 2017

on May 7, 2021Downloaded from

(iv) evaluation of automatic visual inspection systemin use on pharmaceutical filling lines to assess preven-tion of cracks relative to the current state-of-the artprocessing equipment used for this purpose.

Lab Replication of Common Failure Modes

Differences in vial cracking response were illustratedby applying the crack replication methods to typicalborosilicate and ion-exchange strengthened alumino-silicates vials. Glass vials of equivalent dimensions (3mL nominal fill volume) were formed from tubes oftwo glass compositions (51-expansion borosilicate,and strengthened aluminosilicate). After forming andannealing, the aluminosilicate glass vials were ion-exchanged to establish the internal stored strain energyabove the minimum tension threshold. The strength-ened aluminosilicate glass vials received a low coef-ficient of friction surface treatment, which resists glassdamage and reduces particle formation (29). The boro-silicate vials underwent typical annealing and bulkpackaging. Both populations of vials were then testedvia the three high-risk crack replication methods de-scribed previously followed by dye ingress testing toconfirm leaking cracks.

The bump check crack replication method was re-peated on 100 typical borosilicate vials and 100strengthened aluminosilicate glass vials, then all vialswere subjected to the dye ingress leak testing protocoldescribed previously. The borosilicate vial populationexhibited perceived cracks (by human visual inspec-tion) with 100% of the population tested. Dye ingresstesting confirmed that 20 vials leaked (20%), as sum-marized in Table II. The strengthened aluminosilicate

glass vials showed no perceived cracks and no leakingupon dye ingress testing.

The bump check crack introduction method was thenincreased in a stepwise manner to test if cracks couldbe introduced in the strengthened aluminosilicate glassvials. Despite increasing the load by 35%, the testsshowed no observation of cracks; instead the vialsbegan to break— consistent with the bimodal responseexpected.

The absence of cracks in the aluminosilicate vialsunder the same conditions that consistently crackedthe borosilicate vials demonstrates a significant differ-ence in crack introduction behavior. The load neededto observe cracks will be just below that needed tobreak the container. The absence of cracks and onsetof breaking at higher loads indicates that flaws wereintroduced that exceeded the aluminosilicate compres-sive layer depth. The binary response of the alumino-silicate vials (intact or broken) is expected because thestored strain energy was above the threshold shown inFigure 12.

The lensing crack replication method was performedon 50 typical borosilicate and 50 strengthened alumi-nosilicate glass vials. After going through this damagereplication process, manual inspection confirmed thatall conventional borosilicate vials contained perceivedcracks. Aluminosilicate vials that had been throughthe same crack replication method showed no signs ofdamage and zero perceived cracks by manual inspec-tion. Dye ingress testing confirmed that a large frac-tion (70%) of the cracked borosilicate vials leaked andnone of the aluminosilicate vials exhibited leaks.

In parallel, the crack replication conditions (load ap-plied during initial damage introduction and peak loadduring dynamic impact) were increased to observe achange in the aluminosilicate fracture behavior. Afterincreasing the dynamic impact peak load by 60%, thealuminosilicate vials began to exhibit breaks. For theengineered stress profile imparted to these alumino-silicate vials, cracks were not created for any combi-nation of initial damage introduction or dynamic im-pact conditions. When damage was introduced thatpenetrates deeper than the compressive layer, thestored strain energy serves to extend the flaw, propa-gating to cause breakage and make the damage obvi-ous.

An evaluation of the container response to severe neckcracks was performed on a smaller population of 20

TABLE IIResults of Dye Ingress Testing for Borosilicateand Strengthened Aluminosilicate Vials afterExperiencing Bump Check Damage Replication

Glass Type

Dye IngressTesting

QuantityTested

PercentLeaking

Borosilicate 100 20%

Strengthened Aluminosilicate 100 0%

No leaking cracks were observed for the aluminosili-cate vials strengthened with an engineered stress pro-file, while 20% of the borosilicate vials experiencedleaking cracks under the same conditions.

522 PDA Journal of Pharmaceutical Science and Technology

on May 7, 2021Downloaded from

vials per glass type. The method was designed todemonstrate the inherent nature of crack stability inconventional glass vials and illustrate the release ofstored strain energy via fracture with the strengthenedaluminosilicate vials. Figure 13 compares representa-tive micrographs of a conventional vial to a strength-ened aluminosilicate vial after damage from a rotarydisc, showing different fractographic responses. Forthe conventional borosilicate vial, the rotary disc com-pletely penetrates through the vial neck with no frac-ture observed, only a roughened cut surface from thedisc. This means that the rotary disc carved a largepathway through the neck thickness (an exaggeratedcrack) of the conventional borosilicate vial without theflange separating from the neck. For the strengthenedaluminosilicate vial, the image shows that the rotarydisc penetrates less than 75% through the thickness ofthe neck before the stored strain energy (central ten-sion) caused the glass to fracture and the flange toseparate from the neck.

These results illustrate that strengthened aluminosili-cate glass vials show a marked differentiation in frac-ture behavior, exhibiting no cracks in any of the testsand no detection issues during automated inspection.

Statistical Analysis of Lab Replication Results

A statistical analysis was performed to quantitativelycompare the probability of failure of these two vial

types (borosilicate or aluminosilicate) for the twoprobabilistic damage replication methods (bump checkand lensing crack). The binomial distribution wasused, in which case the median probability of thefailure mode occurring was calculated, that is, theprobability for which the cumulative binomial proba-bility is 0.5. Additionally, the 80% two-sided and 95%two-sided confidence bounds around these median val-ues were determined by calculating the failure rate forwhich the cumulative binomial probability is 0.1 and0.9 (for the 80% level) and 0.025 and 0.975 (for the95% level).

Because the tests of aluminosilicate vials did not pro-duce any failures, the calculated probability valuesonly represent an upper bound. For example, in thelensing crack replication method, there were 50 alu-minosilicate samples tested with no failures. As such,there is 97.5% confidence that the true probability ofoccurrence is less than 0.071, and a 90% confidence itis less than 0.045.

The tabulated results for producing leaking crackedvials are shown in Table III. The results show thatdifferences between the borosilicate and aluminosili-cate failure probabilities are significant (evident by theseparation in the confidence bounds). Differences inmedian values are greatest for bump check and lensingcrack replication methods, where borosilicates are at

Figure 13

a) At left, an image of a conventional borosilicate vial neck showing rotary disc damage going through the neckthickness without flange separating from the neck. b) At right, an image of a strengthened aluminosilicate vialneck after a rotary disc damage demonstrating that the disc penetrated less than 75% of the way through thethickness before central tension from the engineered stress profile (ESP) initiated the glass fracture andseparated the flange from the neck. This response prevents the vial from being at risk of containing acontaminated drug product that could be administered to a patient.

523Vol. 71, No. 6, November–December 2017

on May 7, 2021Downloaded from

least 30 to 50 times more likely to exhibit leakingcracks.

Because the rotary disc cut demonstrated a targetedresponse with the binary result of cracked or broken,the predicted results of a statistical analysis is onlysubject to the number of specimens tested and moreinsight can be gained from the fracture response thanprobabilities. As described previously, the fracto-graphic response of the conventional borosilicate vialis to allow the disc to cut completely through the neckthickness; in contrast, the strengthened aluminosilicatevial will tolerate the disc cutting only up to 75% of theneck thickness before the vial breaks. This demonstra-tion of the response due to stored elastic strain energyis physics-based and not probability-based.

It was previously noted that 90% of cracks observed infailed vials occurred in regions of the vial evaluated bythese crack replication methods (�80% bump checks,�10% lensing cracks). Assuming that these replica-tion methods are representative of actual stresses in-curred on filling lines and in the field, we can apply therelative frequency of occurrence to estimate a cumu-lative probability of crack occurrence in each vialtype. The total median failure probabilities for leakintroduction in each glass, weighted by these percent-ages, are for borosilicate 0.200 and for aluminosilicateless than 0.0064. This implies that for �90% ofcracked containers, conventional glass is more than 31times more likely to fail due to leaks than alumino-silicate, essentially preventing cracks in the strength-ened aluminosilicate containers. Given the physics-based understanding of the fracture response, this

scale factor (31 times) is smaller than what largerpopulations are expected to show due to the smallsample sizes used in this study.

Line Simulation with Misadjusted Capper Equipment

A mixed population of 400 typical borosilicate and200 strengthened aluminosilicate glass vials were pro-cessed through a misadjusted capper (most vials ex-hibiting some damage), resulting in 13 perceivedcracks and 1 breakage in the typical borosilicate vialpopulation and 0 perceived cracks and 8 breakages inthe strengthened aluminosilicate glass vial population.Dye ingress testing confirmed that two of the 13perceived cracks in typical borosilicate vials wereleaking and zero of the strengthened aluminosilicateglass vials were leaking, as summarized in Table IV.

These results show that the damage introduced by themisaligned capping equipment was severe enough tocrack and break borosilicate vials and cause breakageof several aluminosilicate containers. The experimentfurther illustrates the binary response (intact or bro-ken, not cracked) for the aluminosilicate containers.Further, the engineered stress profile makes damageintroduced by filling lines more evident so that oper-ators can immediately address setup issues and pre-vent creation of systemic cracks.

Evaluation of Automated Visual Inspection System(Current State-of-the-Art)

A leading automated visual inspection was used toevaluate the efficiency of current state-of-the-art sys-

TABLE IIIStatistical Analysis of the Probability of Producing a Leaking Crack in Aluminosilicate and ConventionalBorosilicate Vials for Bump Check and Lensing Crack Replication Methods, Based upon Values in TablesII and V

Method Glass

95% Two-Sided

80% Two-Sided

Median0.0250 0.1000 0.5000 0.9000 0.9750

Bump Check Borosilicate 0.1349 0.1577 0.2060 0.2607 0.2918

Aluminosilicate �0.0003 �0.0011 �0.0069 �0.0228 �0.0362

Lensing Crack Borosilicate 0.5751 0.6228 0.7086 0.7854 0.8214

Aluminosilicate �0.0005 �0.0021 �0.0138 �0.0450 �0.0711

Both 80% and 95% two-sided confidence intervals are calculated, though the intervals reported for the aluminosilicatevials represent upper bounds due to the absence of failures.

524 PDA Journal of Pharmaceutical Science and Technology

on May 7, 2021Downloaded from

tems to reliably detect cracks relative to the strength-ened aluminosilicate vial, which was designed to pre-vent cracks.

The lensing crack replication method was performedon 50 conventional borosilicate and 50 strengthenedaluminosilicate glass vials. After experiencing thisdamage replication process, manual inspection con-firmed that all conventional borosilicate vials con-tained perceived cracks. Aluminosilicate vials that hadbeen through the same crack replication methodshowed no signs of damage and zero perceived cracksby manual inspection.

All vials were inspected by an automated camerasystem to evaluate the capability of the inspectionequipment to detect cracks in the footprint and heelareas. The commercially available camera system wasspecifically designed to identify cracks in this regionof the vial with the camera orientation illustrated inFigure 14a. Table V quantifies the results and showsthat the automated camera system captured 74% of thecontainers with known cracks and failed to detect 26%of the cracked vials processed through the system. Dyeingress testing was performed and identified that 70%of all of the lensing replication borosilicate vials wereleaking, including 35% of the cracks that had beenaccepted by the visual inspection (Figure 14b). Thisshows that even the best automated visual inspectionsystem, performing 100% inspection, is unable to re-liably detect cracks that present sterility risks.

In contrast, none of the strengthened aluminosilicateglass vials showed evidence of damage or perceivedcracks, despite having been subjected to the samedamage process. Automated inspection did not rejectany of the aluminosilicate vials, and dye ingress test-

ing demonstrated that none of that population exhib-ited leaking cracks.

The reduction in crack occurrence demonstrates thatstable cracks are essentially prevented when glassvials are strengthened to levels above the minimumtension threshold. While automated inspection candetect some cracks that are introduced during thefilling line process, it does not protect against crackoccurrence after the filling process from prior or newdamage events. The stored strain energy that preventscracks in the aluminosilicate containers persiststhrough handling, transport, and end use at the patient.

TABLE IVInspection and Dye Ingress Testing Results for Vials Processed with Misaligned Capping Machine,Replicating Neck Crack Damage Introduction

Glass TypeVial Quantity Processed

by Capping Machine

HumanInspection

Dye IngressTesting

PercentCracked

PercentLeaking

Borosilicate 400 3.25% 0.5%

Strengthened Aluminosilicate 200 0% 0%

Human inspection identified 13 (3.25%) borosilicate vials that contained neck cracks, while no (0%) strengthenedaluminosilicate vials were identified as cracked. Leak testing of these vials showed that two (0.5%) of borosilicatevials exhibited leaking cracks and none (0%) of the strengthened aluminosilicate vials leaked.

Figure 14

a) At left, an illustration shows the orientation of astate-of-the-art automated inspection camera de-signed to reject vials with cracks in the lensingcrack region, with an example photo at bottom. b)At right, a photo of vials that were accepted as goodby the automated inspection system shows that theyclearly failed the dye ingress testing as indicated bythe presence of blue dye.

525Vol. 71, No. 6, November–December 2017

on May 7, 2021Downloaded from

Engineered Safety Benefits of Glass in OtherIndustries

Engineering safety into glass product design is not anovel concept. As mentioned earlier, automotive andarchitectural glasses are designed to reduce harm inthe event of glass breakage (24). Automotive glassuses lamination and thermal tempering to increase thesafety of passengers in the event of a collision. Thelamination prevents glass fragment ejection and pre-vents passenger ejection during a collision (30). Au-tomotive side and rear windows are made of thermallytempered glass, which has high stored strain energy inthe glass. This stored strain energy aids in egress ofpassengers from an overturned vehicle by causing thewindow to fracture into many small pieces after beingstruck with an emergency hammer.

Architectural glass is also thermally tempered. If anoverstressing situation occurs, the thermal temperingalso causes the window to fragment or dice into smallsquared-off pieces of low mass that minimize harm inoverhead applications. Both laminated and temperedglass for automotive and architectural use are sub-jected to strict testing standards to ensure manufac-tured product performs as expected during use. Theirreliability has been demonstrated, and now federalregulations require their use in many applications,such as shower doors. This approach of engineeringsafety into product design and manufacturing throughstandards setting can also be applied to pharmaceuticalglass packaging to prevent harm from cracked glasscontainers.

Conclusions

Glass is the preferred primary packaging material forparenteral drug products, but the risk of cracks inconventional glass containers cannot be mitigatedthrough inspection alone. The low-energy state ofconventional glass vials allows crack systems to per-sist indefinitely, presenting opportunities for sterilityloss and patient harm if the compromised dose isadministered. Introduction of stored strain energy wasshown to provide sufficient driving force to preventstable cracks in vials made from an aluminosilicateglass.

Three crack failure modes were identified in variousfield return surveys and replicated in controlled labo-ratory damage introduction methods. Differences incracking behavior were observed when these methodswere applied to both conventional borosilicate andstrengthened aluminosilicate glass vials. Specifically,the conventional glass vials exhibited cracks with highfrequency and were not detected reliably by state-of-the-art automated inspection equipment. Strengthenedaluminosilicate glass vials did not exhibit cracks underany of the testing methods. In one instance (neckcracks from line simulation with misadjusted capperequipment), the strengthened aluminosilicate glass vi-als broke in obvious ways to signal a setup issue, whilethe borosilicate glass vials sustained a high frequencyof stable cracks that were difficult to detect with visualinspection.

These crack failure modes observed in conventionalglass containers can be prevented through stored strainenergy as imparted through an ion-exchange process.

TABLE VAutomated Inspection and Dye Ingress Testing Results for Vials That Had Experienced Lensing CrackReplication Damage

Glass TypeAutomated

Inspection Results

Dye Ingress Testing

Quantity Tested Percent Leaking

Borosilicate 26% Accepted 17 35%

74% Rejected 33 88%

Strengthened Aluminosilicate 100% Accepted 50 0%

0% Rejected 0 0%

The automated inspection rejected 74% of the damaged borosilicate vials, 88% of which leaked in dye ingress testing.Of the 26% damaged borosilicate vials that were accepted, 35% of them leaked in dye ingress testing. While thestrengthened aluminosilicate population experienced the same damage replication event, 100% of the vials wereaccepted by the automated inspection equipment and none of the vials leaked.

526 PDA Journal of Pharmaceutical Science and Technology

on May 7, 2021Downloaded from

By utilizing glass with sufficiently high stored strainenergy, containers for pharmaceutical packaging canfollow the automotive and architectural industries todesign for improved patient safety.

Acknowledgments

The authors are grateful for the assistance of Jamie T.Westbrook, Eric Allington, Douglas M. Noni, StevenA. Tietje, and Stephen Robinson in conducting themechanical testing and leak detection. We appreciatethe assistance of pharmaceutical collaborators whoconducted various machinability studies, providedpharmaceutical context, and allowed us to inspectcountless damaged samples.

Conflict of Interest Statement

All authors were employed by Corning Incorporatedduring the execution of this research.

References

1. U.S. National Archives and Records Administra-tion. Title 21. Drug Product Containers and Clo-sures. Section 211.94 In Code of Federal Regu-lations; 2016.

2. U.S. Department of Health and Human Services,Food and Drug Administration, Center for DrugEvaluation and Research, Center for BiologicsEvaluation and Research. Container Closure Sys-tems for Packaging Human Drugs and Biologics;1999.

3. Maki, D. G.; Goldmnan, D. A.; Rhame, F. S.Infection Control in Intravenous Therapy. Annalsof Internal Medicine 1973, 79 (6), 867– 887.

4. Wang, S. A.; Tokars, J. I.; Bianchine, P. J.; Car-son, L. A.; Arduino, M. J.; Smith, A. L.; Hansen,N. C.; Fitzgerald, E. A.; Epstein, J. S.; Jarvis,W. R. Enterobacter cloacae Bloodstream Infec-tions Traced to Contaminated Human Albumin.Clinical Infectious Diseases 2000, 30 (1), 35– 40.

5. Bhakdi, S.; Krämer, I.; Siegel, E.; Jansen, B.;Exner, M. Use of Quantitative MicrobiologicalAnalyses to Trace Origin of Contamination ofParenteral Nutrition Solutions. Medical Microbi-ology and Immunology 2012, 201 (2), 231–237.

6. Schaut, R. A.; Weeks, W. P. Historical Review ofGlasses Used for Parenteral Packaging. PDAJ. Pharm. Sci. Technol. 2017, 71 (4), 279 –296.

7. Akers, M. J. Parenteral Preparations. In Reming-ton: Essentials of Pharmaceutics; Felton, L., Ed.;Pharmaceutical Press: London, U.K., 2013; pp495–532.

8. Eakins, M. N. Parenteral Products: PharmacopeialControl of Containers, Storage and Distribution.Pharmacopeial Forum 2011, 3, 4.

9. Kuehn, S. E. Clash of the Titans: Super-Rivals Glassand Plastic Square Off for Patient Safety. Pharma-ceutical Manufacturing 2014 (October), 3–9.

10. PDA Technical Report 43. Identification andClassification of Nonconformities in Molded andTubular Glass Containers for PharmaceuticalManufacturing: Covering Ampoules, Bottles, Car-tridges, Syringes and Vials. Parenteral Drug As-sociation: Bethesda, MD, 2013.

11. Sack, R. A. Epidemic of Gram-Negative Organ-ism Septicemia Subsequent to Elective Operation.American Journal of Obstetrics and Gynecology1970, 107 (3), 394 –399.

12. Robertson, M. H. Fungi in Fluids—A Hazard ofIntravenous Therapy. Journal of Medical Micro-biology 1970, 3 (1), 99 –102.

13. Maddocks, A. C.; Raine, G. Contaminated DripFluids. British Medical Journal 1972, 2 (5807), 231.

14. Elin, R. J.; Lundberg, W. B.; Schmidt, P. J. Eval-uation of Bacterial Contamination in Blood Pro-cessing. Transfusion 1975, 15 (3), 260 –265.

15. Daisy, J. A.; Abrutyn, E. A.; Mac, G. R. Inadver-tent Administration of Intravenous Fluids Con-taminated With Fungus. Annals of Internal Med-icine 1979, 91 (4), 563–565.

16. Frechette, V. D. Failure Analysis of Brittle Mate-rials; American Ceramic Society, Inc.: Wester-ville, OH, 1990; p 116.

17. Bloomfield, S.; Exner, M.; Flemming, H.-C.;Goroncy-Bermes, P.; Hartemann, P.; Heeg, P.;Ilschner, C.; Krämer, I.; Merkens, W.; Oltmanns,

527Vol. 71, No. 6, November–December 2017

on May 7, 2021Downloaded from

P.; Rotter, M.; Rutala, W. A.; Sonntag, H.-G.;Trautmann, M. Lesser-Known or Hidden Reser-voirs of Infection and Implications for AdequatePrevention Strategies: Where To Look and WhatTo Look for. GMS Hygiene and Infection Control2015, 10, 1–14.

18. Swift, R. Glass Containers for Parenteral Prod-ucts. In Pharmaceutical Dosage Forms—Paren-teral Medications, 3rd ed.; Nema, S., Ludwig,J. D., Eds.; CRC Press: Boca Raton, FL, 2012;Vol. 1, pp 287–304.

19. USP Committee of Revision. �1207� PackageIntegrity Evaluation—Sterile Products. In TheUnited States Pharmacopeia, Thirty-ninth Revi-sion; US Pharmacopeia, 2016.

20. Maki, D. G.; Martin, W. T. Nationwide Epidemicof Septicemia Caused by Contaminated InfusionProducts. IV. Growth of Microbial Pathogens inFluids for Intravenous Infusion. Journal of Infec-tious Diseases 1975, 131 (3), 267–272.

21. Microbiological Hazards of Infusion Therapy,Phillips, I., Meers, P.D., D’Arcy, P.F., Eds.;Springer: Amsterdam, 1976; p 186.

22. Pharmaceutical Technology Editors. Injectablesto Show Double-Digit Growth Through 2020PharmTech.com [Online], 2015. http://www.pharmtech.com/injectables-show-double-digit-growth-through-2020.

23. USP Committee of Revision. �1207.2� PackageIntegrity Leak Test Technologies. In The UnitedStates Pharmacopeia, Thirty-ninth Revision; USPharmacopeia, 2016.

24. Gardon, R. Thermal Tempering of Glass. In Elas-ticity and Strength in Glasses: Glass: Science andTechnology, Uhlmann, D., Ed.; Elsevier Science:Amsterdam, 2012; pp 145–213.

25. Tandon, R.; Glass, S. J. Controlling the Fragmen-tation Behavior of Stressed Glass. In FractureMechanics of Ceramics: Active Materials, Nano-scale Materials, Composites, Glass and Funda-mentals, Bradt, R. C., Munz, D., Sakai, M., White,K. W., Eds.; Springer: Boston, MA, 2005; pp77–91.

26. Lawn, B.; Marshall, D. Contact Fracture Resis-tance of Physically and Chemically TemperedGlass Plates—A Theoretical Model. Physics andChemistry of Glasses 1977, 18 (1), 7–18.

27. Gy, R. Ion Exchange for Glass Strengthening.Materials Science and Engineering B 2008, 149(2), 159 –165.

28. Schaut, R. A.; Peanasky, J. S.; DeMartino, S. E.;Schiefelbein, S. L. A New Glass Option for Par-enteral Packaging. PDA J. Pharm. Sci. Technol.2014, 68 (5), 527–534.

29. Timmons, C.; Liu, C. Y.; Merkle, S. ParticulateGeneration Mechanisms during Bulk Filling andMitigation via New Glass Vial. PDA J. Pharm.Sci. Technol. 2017, 71 (5), 379 –392.

30. Clark, C. C.; Yudenfriend, H.; Redner, A. S. Lac-eration and Ejection Dangers of AutomotiveGlass, and the Weak Standards Involved. TheStrain Fracture Test. Annual Proceedings/Associ-ation for the Advancement of Automotive Medi-cine 2000, 44, p 117–132.

528 PDA Journal of Pharmaceutical Science and Technology

on May 7, 2021Downloaded from

Authorized User or for the use by or distribution to other Authorized Users·Make a reasonable number of photocopies of a printed article for the individual use of an·Print individual articles from the PDA Journal for the individual use of an Authorized User ·Assemble and distribute links that point to the PDA Journal·Download a single article for the individual use of an Authorized User·Search and view the content of the PDA Journal  permitted to do the following:Technology (the PDA Journal) is a PDA Member in good standing. Authorized Users are An Authorized User of the electronic PDA Journal of Pharmaceutical Science and 

copyright information or notice contained in the PDA Journal·Delete or remove in any form or format, including on a printed article or photocopy, anytext or graphics·Make any edits or derivative works with respect to any portion of the PDA Journal including any·Alter, modify, repackage or adapt any portion of the PDA Journaldistribution of materials in any form, or any substantially similar commercial purpose·Use or copy the PDA Journal for document delivery, fee-for-service use, or bulk reproduction orJournal or its content·Sell, re-sell, rent, lease, license, sublicense, assign or otherwise transfer the use of the PDAof the PDA Journal ·Use robots or intelligent agents to access, search and/or systematically download any portion·Create a searchable archive of any portion of the PDA JournalJournal·Transmit electronically, via e-mail or any other file transfer protocols, any portion of the PDAor in any form of online publications·Post articles from the PDA Journal on Web sites, either available on the Internet or an Intranet,than an Authorized User· Display or otherwise make any information from the PDA Journal available to anyone otherPDA Journal·Except as mentioned above, allow anyone other than an Authorized User to use or access the  Authorized Users are not permitted to do the following:

on May 7, 2021Downloaded from