Impacts of Antibiotic-Resistant Bacteria (Part 8 of 12) - Princeton

| 127

6

NewTechnologiesfor Infection

Diagnosis andControl

hree major options exist for the controlof bacterial diseases: 1) disrupt or halttransfer of bacteria from person to per-son and from the environment to people,

2) treat cases of disease with antibiotics, and3) prevent disease through vaccination. Thischapter describes diagnostic methods that guidethe selection and use of antibiotics, the use ofvaccines, methods for delivery of high concen-trations of antibiotics to areas of localized infec-tions, devices and materials designed to reducethe transfer of bacteria in the hospital, and sometreatment methods used before the antibiotic age.

The cartoon, which is adapted from one thatoriginally appeared in Science, is a humorouslook at the serious problem posed by bacteriaresistant to all available antibiotics. Some bacte-ria are expected to develop resistance to any anti-biotic introduced into medical practice.Therefore, continued improvement in infectiondiagnosis and control is necessary to optimizethe use of antibiotics and slow the spread ofresistant bacteria.

DIAGNOSTIC METHODSIn the future, science may develop a smalldevice, such as the “tricorder” used in the TVseries Star Trek, that physicians can pass over the

body of a sick person to identify the cause of adisease. Such methods are far in the future, andcurrent techniques used to identify bacteria andsusceptibility patterns are “traditional methods”that have been developed over the last century.Newer methods that involve techniques frommolecular biology and modern instrumenta-tion—not immediately at the level of Star Trek—promise to make identification and characteriza-tion faster and more certain.

❚ Traditional Methods for the Identification of BacteriaSome experts estimate that there may be a mil-lion different bacteria and that scientists haveidentified only one percent (10,000 species) ofthat total. Of those 10,000, only a fraction havebeen associated with human diseases.

When seeing a patient, a physician will askquestions, make observations, and perform teststo determine which bacteria are likely to be asso-ciated with an illness and to choose an antibiotictreatment. The physician may swab the throat inthe case of a sore throat or obtain a sample ofurine in the case of a urinary tract infection. Thecollected material on the swab or the urine can bestained with diagnostic dyes, such as the Gramstain (see chapter 2), and examined under a

T

128 Impacts of Antibiotic-Resistant Bacteria

Copyright @ 1995 Sidney Harris. Reprinted with permission,

microscope. Distinctive shapes and stainingproperties facilitate reliable preliminary rapididentification of the bacteria causing infection.

Collected samples may contain such low num-bers of bacteria as to make finding them underthe microscope difficult. The staining propertiesand shapes of the bacteria may not be unique andtherefore not identifiable. The sample may con-tain a mixture of bacteria, as is common in faecalsamples. To identify the bacteria in those cases,the physician sends a biological sample of somekind—a volume of blood or pus or other exudate,a scraping or swab from the throat or otherorifice, a sample of urine or feces—to a micro-biology laboratory.

In the laboratory, the sample is transferred toculture media specifically designed to encourage

the growth of certain pathogenic bacteria and toprevent the growth of others such as commensalbacteria that may be present in samples fromboth healthy and sick individuals. The bacteriathat are able to grow form visible colonies onagar-based media in a Petri dish or grow in brothso that the broth becomes turbid, as apple ciderdoes when yeast grow in it. In both the collectionand handling of the sample, health care personnelmust be careful to avoid contamination with thebacteria that grow literally everywhere, on thepatient’s and physician’s skin, on the surfaces offurniture and unsterilized devices in the examin-ing room, and on apparatus in the diagnostic lab-oratory.

Microbiologists can sometimes look at andsmell the colonies or liquid cultures and, based

Chapter 6 New Technologies for lnfection Diagnosis and Control 129

on their knowledge and experience, identify thebacteria in the sample. They may be able to dis-miss some bacteria from further consideration byrecognizing them as contaminants. Iterative testswith more selective media and biochemical testsmay be used for more specific identification.

Culturing and identification take time. Theshigella that might be present in a fecal sample,or the Escherichia coli that frequently cause uri-nary tract infections, grow quickly, forming col-onies in 24 hours or so, and a laboratory wouldprobably identify them in one or two days. Myco-bacterium tuberculosis grows far more slowly,and six weeks may pass before traditional meth-ods can be used to identify it.

Identifying the bacteria is often critical forchoosing the most appropriate antibiotic therapybecause some antibiotics work better against cer-tain bacteria. But identification does not provideinformation about whether the bacteria are resis-tant to the antibiotic or susceptible to it. “Suscep-tibility tests” are used to determine that.

Traditional Susceptibility TestsInformation about antibiotic-resistance/suscepti-bility is developed by testing the bacteria isolatedfrom the infection against six to 12 different anti-biotics, or more if necessary. The results fromthese tests may support the use of the antibioticthat was empirically selected by the physician,indicate that other antibiotics would work aswell, or show that the disease-causing bacteriaare resistant to the antibiotic empirically chosen.

Jorgensen (1995) describes four methods thatare currently used to determine the antibioticsusceptibility or resistance of bacteria: 1) diskdiffusion tests, 2) broth dilution tests, 3) agardilution tests, and 4) agar gradient methods.

Disk diffusion testsDisk diffusion tests measure the size of a cleararea of no bacterial growth around a sterile paperdisk containing antibiotic. The size of this area,called the “zone of inhibition,” can be measuredand reported directly, or the measurement can becompared to criteria established by the NationalCommittee for Clinical Laboratory Standards

Photo of a disk diffusion (Kirby-Bauer) susceptibility test plate.

Courtesy of James H. Jorgensen, University of Texas Health

Science Center, San Antonio, TX, 7995.

(NCCLS) to classify the bacteria as susceptible,intermediate or resistant (S, I, or R). These testsare well standardized for certain bacteria andmay be highly reproducible. However, disk testsare influenced by many laboratory variables thatcan limit accuracy unless tightly controlled.

O’Brien (1994), who initiated and runs WHO-NET, the World Health Organization-sponsoredsurveillance system for antibiotic-resistant bacte-ria, emphasizes the importance of requiring labo-ratories to report raw data about the size of thezones of inhibition (figures 6-1 and 6-2) to sur-veillance organizations. While laboratories inEurope and North America are consistent in theirmeasurement and reporting of the diameters ofzones of inhibition around antibiotic disks, theyinterpret the meaning of the measurements dif-ferently (figure 6-l). Therefore, data reported aszones of inhibition rather than as interpretationsare necessary to make any valid internationalcomparisons about the prevalence of antibiotic-resistant bacteria.

Broth dilution testsDilution tests measure the concentration of anti-biotic that is necessary to prevent the growth ofbacteria. In these tests, known amounts of bacte-ria are deposited into small test tubes containing


Figure Aa

Total number tested = 808

S = 85%

5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35


R = 7%

Figure Bb

S = 79%

L5 7 9 11 13 15 17 1921 232527 29 31 33 35

a Figure A represents a center in Europe (Center 1).b Figure B represents a center in North America (Center 5).

NOTE: European and North American centers measured similar

zones of inhibition, illustrating the reproducibility of the methods.

However, the use of different break points in the two centers wouldresult in the centers reporting different percentages of resistant

organisms. Even if the laboratory data were identical, the centerswould’ report different percentages of resistant organisms. This exam-ple demonstrates the importance of reporting raw data for making

comparisons between laboratories.

SOURCE: WHONET, 1994.


R = 160/0 I = 25% s = 59%

5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35

NOTE: This histogram illustrates the ambiguity of setting breakpointsfor classifying bacteria as susceptible, intermediate, or resistant. In

particular, the breakpoint for the division between resistant and inter-mediate appears to fall at a peak in measured zones of inhibition.

SOURCE: WHONET, 1994

1 to 2 milliliters (a teaspoonful is about 5 millili-ters) of sterile nutrient growth medium (“broth”)containing different concentrations of antibiotic(figure 6-3). The lowest concentration of antibi-otic that prevents growth of the bacteria definesthe “Minimum Inhibitory Concentration” (MIC).

While the MIC provides information about theconcentration that will inhibit the growth of abacterium, it does not say whether that concen-tration can be reached in the treated patient orwhat dose of antibiotics is needed to reach thecritical concentration. Interpretive guidelinesprovided by NCCLS publications help clinicalmicrobiologists and physicians interpret MICs asclinical categories of S (“susceptible”), I (“inter-mediate”), and R (“resistant”).

A disadvantage of this method is the largenumber of test tubes and racks and large volumesof media that are required. To test a single bacte-rial culture against six antibiotics would require42,48, or 54 tubes, depending on the lowest con-centration used. The miniaturization of this


II

2 4 8 16 32

Concentration of antibiotic A(zero time)

Concentration of antibiotic B

o 2 4 8 16 32

Concentration of antibiotic A

2

Concentration of antibiotic B(zero time)

NOTE: In both A and B, a small measured volume of bacterial culture is added to each test tube at time zero, Following incubation (the time ofIncubation depends on the growth characteristics of the bacteria), the test tubes in which bacteria were able to grow are turbid (the first three

tubes in A and all tubes in B). The absence of growth in the last three tubes that contain antibiotic A indicates that the bacteria are killed or inhib-ited by concentrations of A equal to 8 units or more, Growth in all tubes that contain antibiotic B indicates that the bacteria are resistant to all

tested concentrations of antibiotic B

SOURCE: Office of Technology Assessment, 1995.

method with microdilution trays solved thatproblem. The broth micro dilution test is cur-rently the most popular antibiotic sensitivity testin the U.S. (table 6-I; the test using test tubes iscalled the “broth macro dilution test”). The smallsize of the wells and the small volumes, about0.1 milliliter (about a drop from an eye-dropper),require that some viewing device be used todetermine which of the wells in the test plate areclear and which are turbid. There are a number ofcommercial devices that make that determina-tion, and some plot out the MICs from the tests.

To hold down costs and reduce the spaceneeded for incubation of test cultures, many lab-oratories do not use the entire series of dilutions

Photo of a broth microdilution susceptibility test tray with dispos-

able inoculator.


Science Center, Sari Antonio, TX, 7995,


Percent of laboratoriesTesting method reporting routine use

Broth macrodilution 1.8

Broth microdilution

Commercially prepared 46.2

User prepared 0.4

Agar dilution 0.2

Disk diffusion 31.8

Rapid automated 19,7

SOURCE: Jorgensen, 1995. From data collected in a 1991 Profi-

ciency Survey of 3414 laboratories conducted by the College ofAmerican Pathologists.

as diagramed on figure 6-3. Instead, based onthe NCCLS interpretive criteria, only two tothree dilutions of each antibiotic are used. One ofthe dilutions is set to match the “break-point”that defines the division between the resistantand intermediate response; another dilutionmatches the concentration that defines the break-point between the intermediate and susceptibleresponses (see figure 6-2 for examples of break-points using disk diffusion tests). When only twoor three dilutions are used for each antibiotic, thetests provide only an estimate rather than a quan-titative measurement of the MIC. The true break-point might be somewhat different from theguidelines, and this fact can cause errors in clas-sifying the bacteria as resistant or susceptible.

Agar dilution tests

Agar dilution tests are similar to the broth dilu-tion tests in that they measure the MIC. In thesetests, a small volume of a bacterial suspension,usually 1–2 microliters, is transferred to each of aseries of agar plates containing known concen-trations of antibiotics. Multi-prong devices areused to transfer approximately 100 colonies atone time.

Antibiotic gradient susceptibilitytest methodsTwo commercial methods, the Etest1 (AB BIO-DISK, Solna, Sweden) and the Spiral GradientEndpoint System (Spiral Biotech Inc., Bethesda,Maryland), use antibiotic concentration gradientson agar plates. Both tests establish MICs thatcompare closely with those determined in thedisk diffusion or broth dilution tests, and both areuseful for testing anaerobic and other hard-to-grow bacteria.

The Etest has been cleared by the FDA forclinical use in the U.S. The Spiral Gradient End-

Photo of five Etest strips on the surface of an agar plate. lnter-

section of the growth ellipses with the strips indicate the MICs.



Photo of the Spiral Assay System Susceptibility Test.

Courtesy of Spiral Biotech, Inc., Bethesda, MD, 1995.

1 OTA mention of a company or product does not constitute an endorsement. Furthermore, companies and products are mentioned asexamples; competing companies and products exist.


point System has not yet been cleared by theFDA for clinical use.

These tests may have a special advantage forresistance surveillance because they have a con-t inuous concentration gradient and are able to

show subtle changes in susceptibi l i ty, and the

wide concentration gradients of these tests cover

the MIC ranges of susceptibility of a wide vari-ety of pathogens and allow both low-level andhigh-level resistance to be detected. The Etest isreportedly easy to use in most laboratory settingsand requires no complicated procedures.

Modifications of Traditional Methods ToShorten Times Necessary To Obtain ResultsThe four methods discussed require at least over-night incubation to obtain results. That time canbe shortened to four to 10 hours for certain anti-biotics and organisms by using optical devices(sometimes coupled with fluorescent indicators)more sensitive than the human eye to detectgrowth in microdilution tubes. Two commer-cially available automated systems can produceresults in four to 10 hours.

The AutoSCAN Walk/Away (Dade Micro-scan, USA, Miami, Florida) uses standardmicrodilution trays that are inoculated in thestandard way and placed in an automated incuba-tor that uses a fluorometer to detect the presenceor absence of growth at different antibiotic con-centrations. The Vitek System (bioMerieuxVitek, Hazelwood, Missouri) was developed byNASA to diagnose urinary tract infections inastronauts in space in the 1970s. It uses credit-card size reagent cards, each of which has 30 tinywells for the testing of different antibiotic con-centrations, and the assays can be completed inthree to 10 hours. While both systems providerapid results, each requires backup cultures andother tests in case of power or mechanical fail-ures.

In some cases, the automated machines canfail to detect resistance. To deal with this prob-lem, manufacturers of both of these instrumentshave developed computer software that reviewsthe results to identify those that may be false.Some of these systems can also identify unex-

Photo of bioMerieux Vitek susceptibility testing reagent card. It

was originally developed by NASA to diagnose urinary tract

infections in astronauts in space and had to be very small.



pected resistance patterns and offer suggestionsfor antibiotic treatment (Jorgensen, 1993). Com-puter analysis of the test results can also belinked to the hospital pharmacy’s computer toalert the pharmacy personnel when the wrongantibiotic therapy is being used. As discussed inchapter 4, computer networks such as this canimprove patient care and reduce costs.

Summary of the Test MethodsTable 6-1 shows the reported frequency of use ofthe various test methods in a survey of Americanlaboratories, and table 6-2 provides informationabout the relative costs of the most commonlyused methods. None of the methods differs verymuch in labor costs. Based on the costs of equip-ment and supplies, the disk diffusion method isthe least costly. O’Brien (1994) argues that it canalso be the most informative under most condi-tions because the sizes of the zones of inhibition(see photograph) provide raw data that have notbeen subject to interpretation, and zone of inhibi-tion information is more quantitative than brothdilution tests that are sometimes based on onlyone or two dilutions.

Will Faster Tests Make a Difference?A test result that shows that bacteria are resistantto the empirically chosen antibiotic will certainlycause the physician to substitute another antibi-

134 | Impacts of Antibiotic-Resistant Bacteria

TABL

E 6-

2: P

rinci

pal F

eatu

res

of P

rese

ntly

Ava

ilabl

e M

anua

l and

Aut

omat

ed A

ntim

icro

bial

Sus

cept

ibilit

y Te

stin

g Sy

stem

s

Met

hod

or in

stru

men

tTe

st fo

rmat

Mea

ns o

f gro

wth

det

ectio

n/

endp

oint

det

erm

inat

ion/

time

Equi

pmen

t req

uire

dAp

prox

imat

e co

st o

f eq

uipm

ent

Appr

oxim

ate

cost

/test

*

Bro

th m

icro

dilu

tion

(p

rep

ared

in-la

bor

ator

y)96

-wel

l pla

stic

mic

rotit

er

tray

sV

isua

l det

erm

inat

ion

of M

ICs

by

turb

idity

; 18–

24 h

ours

Tray

dis

pen

ser,

man

ual

read

er

$35,

000

$6–7

Bro

th m

icro

dilu

tion

(com

mer

cial

froz

en o

r d

ried

tray

s)

96-w

ell p

last

ic m

icro

titer

tr

ays

Vis

ual d

eter

min

atio

n or

aut

omat

ed

read

er in

terp

reta

tion

of M

ICs;

18

–24

hour

s

Cho

ice

of m

anua

l rea

der

or

aut

o-re

ader

$500

$50,

000

$7–9

Dis

k d

iffus

ion

6–12

dis

ks o

n 10

0–15

0 m

m

pet

ri p

late

sV

isua

l mea

sure

men

t of z

ones

of

inhi

biti

on; s

usce

pt.

cate

gor

ies;

16

–18

hour

s

Non

e (o

nly

stan

dar

d la

b

incu

bat

or)

Non

e$1

.50–

4

Ete

st1–

5 st

rips

on 1

00–1

50 m

m

pet

ri p

late

sV

isua

l det

erm

inat

ion

of M

ICs;

16

–18

hour

sN

one

(onl

y st

and

ard

lab

in

cub

ator

)N

one

$3–1

1

Dad

eM

icro

Sca

nW

alkA

way

96-w

ell p

last

ic m

icro

titer

tr

ays

Fluo

rog

enic

sub

stra

te a

uto-

incu

bat

or/

read

er M

ICs;

3.5

–15

hour

sA

uto-

incu

bat

or/r

ead

er $

85,0

00$1

25,0

00$1

0–12

bio

Mer

ieux

Vite

k30

–45-

wel

l pla

stic

car

ds

Kin

etic

turb

idim

etric

MIC

s or

ca

teg

orie

s; 4

–15

hour

sA

uto-

inoc

ulat

or/ i

ncub

a-to

r/re

ader

$40

,000

$152

,000

$4–6

* In

clud

es c

ost o

f con

sum

able

sup

plie

s on

ly; l

abor

req

uire

men

ts d

o no

t diff

er s

ubst

antia

lly b

etw

een

met

hod

s.

SO

UR

CE

: J.H

. Jor

gen

sen.

Mar

ch 1

995.

Cur

rent

and

futu

re p

heno

typ

ic m

etho

ds

for

det

ectio

n of

res

ista

nt b

acte

ria. R

epor

t pre

par

ed fo

r th

e U

.S. C

ong

ress

, Offi

ce o

f Tec

hnol

ogy

Ass

essm

ent.

Chapter 6 New Technologies for Infection Diagnosis and Control | 135

otic if the patient is not improving. Sometimes,however, the patient improves despite the pres-ence of apparently resistant bacteria. This canoccur because the patient’s immune system issuccessfully controlling the bacteria, or becausethe antibiotic reached a higher concentration inthe patient’s body than in the laboratory test, sothat the bacteria were killed or inhibited.

Up to 40 percent of antibiotic therapy wasinappropriate as judged by a comparison of phy-sicians’ prescriptions to an analysis of the labora-tory results for bacterial identification andantibiotic susceptibility tests (see, for example,Jorgensen and Matsen, 1987). While olderreports in the literature (see Edwards, Levin,Balogtas, et al., 1973) indicated that physicianspay little attention to microbiology test results,more recent publications (Doern, Scott, andRashad, 1982; Weinstein, Murphy, Reller, et al.,1983; Jorgensen and Matsen, 1987) indicate thatsome physicians do modify their prescriptionsupon receiving additional laboratory informa-tion. In particular, rapid susceptibility tests,which can be completed in four to 10 hours,resulted in more appropriate therapy, and Doern,Vautour, Gaudet, et al. (1994) found that rapidtests resulted in fewer additional laboratory tests,fewer invasive procedures, shortened time inintensive care, and reduced mortality.

Physicians typically receive the results of anti-biotic susceptibility tests on the morning of thesecond or third day after specimens are submit-ted to the laboratory. The faster methods producethe results more quickly, but unless the physi-cians and nursing staff are prepared to use theinformation at the earlier time, it will not be con-sidered until the next morning. In this case, tech-nological improvements can only be useful ifaccompanied by changes in habits.

Jorgensen (1995) discusses another obstacle tothe use of rapid methods. Laboratory managersoften confirm the results of the faster methodswith backup tests using the older, slower meth-ods. However, this requires performing the testsmore than once and therefore increases costs.Trade-offs must be made among the objectives

of speeding up the process, ensuring accuracy byperforming backup tests, and saving money.

❚ New Technologies for Identifying Bacteria

Antigen TestsAntigen tests use antibodies to recognize specificmolecules on or in bacterial or other cells. Forinstance, the home pregnancy test detects anti-gens that are produced only during pregnancy.

There are many versions of antigen tests todetect the presence of strep A bacteria in sorethroats, but the usefulness of these tests is limitedby low sensitivity. Traditional cultures are stillrecommended when tests are negative. Antigentests are also available to detect Clostridium difi-cile in patients with diarrhea and to determine thebacterial cause of meningitis.

Methicillin-resistant Staphylococcus aureus(MRSA) presents identification problemsbecause it and all other S. aureus grow slowly.Further, its identification is usually accomplishedby a specific test for protein antigens on its sur-face, and these tests failed to identify between 1and 25 percent of S. aureus. Kuusela, Hilden,Savolainen, et al. (1994) discovered another pro-tein on the surface of S. aureus and developed atest for it that detects both methicillin-susceptibleS. aureus and MRSA.

Tests which Directly Measure the Presence of a Bacterial or Antibiotic Resistance GeneTests which directly measure the presence of abacterial gene (discussion of tests for antibioticresistance genes follows) are fundamentally dif-ferent from the traditional tests, which measure aproperty of an organism such as its ability togrow in the presence of a certain concentration ofantibiotic. The new gene tests bring with them anew set of considerations: A bacterium mightcontain a gene for resistance, but not “express” itunder the conditions of the traditional diagnostictests, or a resistance gene may have undergone amutation that does not affect its function but thatmakes its presence undetectable, or the genes ofdead bacteria may be detected with DNA tests.


For example, samples from a patient who isbeing successfully treated with anti-TB drugsoften test positive in DNA tests, but negative inculture-based tests that rely on growing theorganism. These are problems that must be con-sidered in designing new genetic tests and usingthem in clinical practice.

One huge advantage of tests that measure thepresence of a bacterial gene is that they arequick; many tests take only a few hours or less.Another advantage is that they generally havemuch higher sensitivity than the antigen andother enzymatic tests described above, althoughin some cases the sensitivity is not as high as thatof traditional culture methods. The speed com-bined with the sensitivity is very useful. How-ever, some tests require culturing, i.e., thegrowing of bacteria from the clinical samples,before the genetic test can be performed. Theculturing requirement adds time to the process.To date there exists no test so rapid that it willconfirm bacterial identification and susceptibilitybefore a patient leaves a physician’s office. Thedevelopment of faster and more susceptiblegenetically based tests for bacteria started in theearly 1980s, but most are still not available forroutine use. Nevertheless, some of these tests,such as those that are able to diagnose tuberculo-sis in a few hours instead of a few weeks, repre-sent a significant technological advancement thathas improved clinical practice.

DNA probe assaysSingle-stranded fragments of DNA or RNA thatare complementary to a target DNA or RNAsequence will form a double-stranded molecule,known as a “stable hybrid,” under certain reac-tion conditions. Diagnostic fragments, or probes,which will bind to target DNAs or RNAs, arelabeled with enzymes or dyes so that the bindingof the probe to the target can be detected.

At the present time, several commercial DNAprobe tests in clinical use are manufactured byGen-Probe, Inc. (San Diego, California), Gene-Trak (Framingham, Massachusetts), Ortho Diag-nostic Systems (Raritan, New Jersey), andothers. Some of these tests are designed to con-

firm the identification of cultivated colonies,such as tests for M. tuberculosis, M. avium (animportant pathogen in patients with AIDS), andNeisseria gonorrheae (the agent of gonorrhea).Other tests can be used for the direct detection ofbacteria in clinical samples, such as Neisseria,Chlamydia trachomatis (an agent of urethritis,cervicitis, and pelvic inflammatory disease) andS. pyogenes (a cause of suppurative tonsillitis or“strep throat”). These are organisms that for themost part are difficult or slow to cultivate andidentify in the laboratory. The tests requireapproximately two to four hours to complete andcost the patient approximately $20–40 per test.

One important disadvantage of probe-basedmethods to date has been their low sensitivitycompared to culture-based methods. Probeassays for M. pneumoniae (an agent of atypicalor “walking” pneumonia) and Legionella pneu-mophilia (the agent of Legionnaire’s Disease)are no longer much used, primarily because ofthis problem. New technologies in development,which provide the ability to amplify probe orprobe-linked signals after binding to the target,may help increase the sensitivity of these tests.

One promising probe-based test that doeshave adequate sensitivity is a rapid direct DNAprobe test from Gen-Probe that can identifyGroup A Streptococcus directly from throatswabs. In comparative studies, test results agreedclosely with those from older and slower tests(see, for example, Rippin, et al., 1994; Heiter andBourbeau, 1993), unlike the quick antigen streptests described above. However, Heiter andBourbeau conclude that because this test requiresseveral instruments not routinely found in doc-tors’ offices and because it still requires twohours, the test will not be useful for point-of-caretesting in doctors’ offices or emergency roomclinics.

Target amplification methodsOne of the most promising approaches forincreasing the sensitivity of probe-based DNAtests is to amplify the target DNA sequencethrough such methods as polymerase chain reac-tion (PCR), which can rapidly generate millions


of copies of bacterial or resistance gene DNA orRNA sequences. PCR requires identifying andsynthesizing short sequences complementary tothe target gene that act as “primers” for the syn-thesis of the DNA. It is relatively easy to synthe-size millions of these short sequences, but itwould be difficult to synthesize larger pieces ofDNA. Starting with as little as one strand ofDNA from the sample, PCR uses enzymes toelongate the “primers” into full copies of theDNA. PCR can generate millions of copies of aparticular DNA in hours.

Species-specific PCR detection assays havebeen developed for at least 50 different bacterialpathogens, and specific sequences are availablefrom a much larger number of species, for whichPCR primers can be designed. However, only afew standardized kits for performing these testson specific bacterial species are commerciallyavailable in the U.S. Among those kits that eitherhave been cleared, or are nearing clearance, bythe Food and Drug Administration are PCRassays for C. trachomatis, N. gonorrheae, and M.tuberculosis. Even without a commercially avail-able standardized kit, the tests can still be set upindividually by service labs. However, there areseveral disadvantages to performing these testswithout using a standard kit. First, most of theseassays do not perform as well in detecting bacte-ria in clinical samples as they do in purified cul-tures; suboptimal sample preparation proceduresand reaction conditions are probably to blame.Second, unless physical, chemical, or enzymaticprecautions are in place, PCR and other targetamplification methods are easily jeopardized bycontaminating nucleic acid, either from prioramplification reactions or from positive clinicalsamples. Third, there is dramatic interlaboratoryvariability in the test results for the same groupof clinical samples. Many of these problems maybe solved by the availability of standardizedcommercial kits.

After the nucleic acid is isolated and amplifiedby a technique such as PCR, the nucleic acid canbe sequenced to identify the organism. Auto-mated sequencers marketed by Applied Biosys-tems can determine 48 independent DNA

sequences of 400–500 nucleotides in length inapproximately 8 hours, and speed and sequencelength capabilities are continually beingimproved. Automated sequencing systemsrequire an initial investment of approximately$55,000 (Molecular Dynamics) to $125,000(Applied Biosystems, including sequence analy-sis software). It is estimated that identification ofa single bacterial isolate with an automated pro-cedure will cost approximately $75.

Another way to identify the organism is tobind the nucleic acid to probes that recognizespecific sequences. Currently, sequences pre-pared from specific reference strains of bacteriaare used. New strategies are expected to use ran-dom sequences of nucleic acid bound in orderlyarrays on micro-scale photolithographic siliconchips, and the nucleic acid can be identified bydetermining which probes bind to it. Because ofthe microscopic scale of these tests, the boundnucleic acid must be detected with a laser confo-cal microscope. This approach has already beenshown to be useful for the detection of singlebase pair mutations in the human immunodefi-ciency virus. This technology offers significantpotential for rapid sequence determination ofspecific gene targets and for the detection ofspecific identifying signature sequences or anti-biotic-resistance-associated sequences. First-generation chip-based sequencing systems maybe available for research by 1996.

Using rapid DNA tests to diagnose tuberculosisDiagnosing tuberculosis is difficult because ithas many different clinical manifestations. More-over, many physicians were not trained to recog-nize tuberculosis because its prevalence wasdecreasing until about 10 years ago. The recentresurgence of this disease is a huge problem,both in the United States and around the world,and rapid diagnosis is critical so that patients canbe treated before they pass this highly infectiousdisease to others. Quick determination of the sus-ceptibility of the infecting organism is alsobecoming increasingly important because many


drugs are inactive against some of the multi-resistant strains of tuberculosis.

The tuberculin skin test is often used as thefirst diagnostic indication that a person has beeninfected with tuberculosis. A positive tuberculinskin test does not mean that the person has activedisease, only that the person has been exposed totuberculosis. Haas and Des Prez (1995) reviewstudies of the interpretation of positive tuberculinskin tests in nursing homes which show that3.8 percent of men who were tuberculin-positiveon admission to nursing homes subsequentlydeveloped active disease, and that 11.6 percentof men who became positive while in the nursinghome later developed the disease. The percent-age developing active disease could be reducedto 0.2–0.3 percent with the prophylactic use ofthe antibiotic INH. However, the level of hepatictoxicity from INH was 3–4 percent, and therewere other side effects. Deciding when to pre-scribe antibiotics for a patient with a positiveskin test but no other symptoms is very compli-cated because the toxicity of the drug must beweighed against the probability that the patientwill develop tuberculosis. The same consider-ations apply to new diagnostic tests based on thedetection of the DNA of M. tuberculosis.

Isolating the mycobacteria causing tuberculo-sis requires from three to eight weeks, and sus-ceptibility testing by agar dilution methodsrequires another three to six weeks. Highly vari-able results have been observed between two dif-ferent clinical laboratories using culture methods(Hewlett, Horn, and Alfalla, 1995). The identifi-cation and susceptibility testing of drug-resistantTB can be significantly hastened by using theBACTEC radiometric method, but the timerequired is still 20 days or more. Recent data onthe Etest for susceptibility testing of mycobacte-ria suggest that MIC values can be obtained infive to 10 days, a significant improvement overcurrent methods (Wanger and Mills, 1995).

With PCR and probe-based DNA tests, physi-cians will have the ability to identify mycobacte-ria in the sputum of patients within a few hoursto a few days. All tests that are currently clearedby the FDA require some culturing of the clinical

sample, but newer tests in development willallow identification of mycobacteria directlyfrom clinical samples. These tests are used inmany other countries, including much of Europe.Some laboratories are promoting clinical use ofPCR tests in the U.S. Macher and Goosby (1995)document a difficulty in interpretation of PCRtests in the absence of other clinical signs oftuberculosis. On the basis of two (out of three)positive PCR tests, the patient received antituber-culosis chemotherapy and was placed in isola-tion. Later, six cultures turned out to be negativefor tuberculosis, and the patient was taken offdrugs for active tuberculosis and placed on INHalone for preventive therapy. This case studyindicates that the DNA probe tests might be toosensitive: they might detect non-viable mycobac-teria from a previous exposure. This result iscomparable to a positive tuberculin skin test,which, as discussed above, indicates past expo-sure to mycobacteria but does not necessarilysignify active tuberculosis.

❚ New Technologies for Detecting Antibiotic ResistanceThe increasing prevalence of antibiotic-resistantbacteria is leading manufacturers to develop testsspecifically to identify resistant strains. In gen-eral, these tests are designed to produce results ina few hours. Discrepancies may arise betweenthe results of old and new methods. The oldermethods directly measure whether an organismexpresses resistance and can grow in the pres-ence of an antibiotic. Some of the newer methodsindicate whether an organism has a gene encod-ing for resistance. However, the organism maynot “express” this resistance even if it has thegene. In some cases, it is unknown whether thepresence of the gene or the expression of thegene under laboratory conditions is the moreimportant predictor of clinical outcome.

Enzymatic TestsEnzymatic tests can directly measure the pres-ence of an enzyme that confers antibiotic resis-tance, such as β-lactamases that inactivate


penicillins and other β-lactam antibiotics and theenzyme that inactivates chloramphenicol. Thedetection of the β-lactamases requires only a fewminutes (Stratton and Cooksey, 1990), but it islimited to only a few bacterial species. More-over, it does not detect penicillin resistancecaused by other mechanisms, such as the produc-tion of modified penicillin binding proteins. Thetest for the chloramphenicol inactivating enzymerequires one to two hours and can be used todetect the most common form of chlorampheni-col resistance, but it has decreasing utilitybecause of the declining use of this antibiotic.

Tests Based on Indicator Dyes or Light-Producing EnzymesSome tests add indicator dyes to a bacterial cul-ture and then detect the presence of living organ-isms by a color change in the indicator dye. Anexample is the Crystal MRSA Rapid ID test fromBeckton Dickinson. This test, which can detectMRSA in four hours in cultured bacteria, uses anindicator dye that can be observed under an ultra-violet light source in the absence of oxygen. Inthis test, three samples of bacteria are incubatedwith the indicator dye. In addition, one of thesamples is incubated with oxacillin (a semi-syn-thetic penicillin similar to methicillin) and one ofthe samples is incubated with vancomycin. If thebacteria survive, they will use the oxygen in thesamples and the dye changes color. If the samplecontains MRSA, the organism will survive in thepresence of oxacillin but not in the presence ofvancomycin. If the organism is susceptible tooxacillin, it will not survive either antibiotic. Thetest, which costs about five dollars, does notrequire expensive instrumentation. Kohner, Kol-bert, Geha, et al. (1994) found that this system isan effective rapid screening method for MRSAbut has poor performance for coagulase-negativeStaphylococci, which often present a greaterdiagnostic dilemma.

A more complicated test for multiresistanttuberculosis is currently in very early develop-ment (Jacobs, Barletta, Udani, et al., 1993). Inthis test, the gene for the light-producing enzymefrom fireflies was cloned into a virus that infectsM. tuberculosis. The virus is added to a sampleof sputum from the patient. The virus will infectany mycobacteria that are present. If the virusinfects living mycobacteria, the viral DNA isactivated, and the firefly enzyme will cause theculture to give off light. When antibiotics areadded to the test, only resistant mycobacteria willsupport viral growth; susceptible ones will not,and susceptible cultures will not light up. Thussusceptibility can be determined. Research iscurrently underway to determine if this test canmeasure as few as 100 live M. tuberculosis bac-teria, and would therefore work directly onpatient samples in a few hours (Jacobs, NIHGrant R01AI27235). However, this sensitivitymay be difficult or impossible to achieve becauseof background signals in the sample.2

DNA-Based Methods for Testing Antibiotic ResistanceCurrent susceptibility patterns suggest thatrifampin resistance in M. tuberculosis can beused as a predictive marker of multidrug resis-tance. In general, surveillance indicates thatresistance to rifampin correlates well with resis-tance to three or more antituberculosis drugs.Furthermore, virtually all of the highly resistantmycobacterial strains (resistant to greater thanfive drugs) are rifampin-resistant. However, thismay change in the future if M. tuberculosisundergoes further genetic mutation.

PCR tests are in development to detectrifampin resistance in M. tuberculosis caused bythe rpoB gene. The use of the signaturesequences in the rpoB gene assumes that thereare not significant numbers of rifampin-resistantM. tuberculosis strains in the community withother, uncharacterized rpoB mutations in thegene.

2 All samples “glow in the dark”—some background signals are detected. This test will only achieve high sensitivity if the signal from thefirefly enzyme is significantly larger than the background signal.


MRSAs are currently identified by using thetraditional tests discussed in the first part of thischapter. The performance of these tests may bevariable. Factors such as the inoculum size, incu-bation time and temperature, pH of the medium,salt concentration of the medium, and prior expo-sure to β-lactam antibiotics all influence theexpression of resistance. To complicate mattersfurther, only some bacteria in a culture mayexpress methicillin resistance, even if all havethe gene. Taking into account these factors, theNational Committee for Clinical LaboratoryStandards (NCCLS) has recommended guide-lines to optimize the detection of resistance.However, occasional organisms have been iso-lated that are difficult to characterize by thesemethods. The results produced by various meth-ods of disk diffusion and those from agar dilutionmethods are often not consistent. In addition, it isdifficult to separate organisms that are highlyresistant due to overproduction of β-lactamasefrom organisms that have the mecA gene encod-ing for an altered penicillin binding protein.Organisms resistant due to the mecA gene oftenrequire vancomycin therapy, while organismsresistant due to overproduction of β-lactamasemight actually respond better to treatment withβ-lactam antibiotic/β-lactamase inhibitor combi-nations than vancomycin.

PCR and DNA-probe techniques have nowbeen developed to identify the mecA gene. Ingeneral, the studies to date show a high degree ofcorrelation between traditional and DNA-basedtests and allow accurate classification of highlyresistant and borderline resistant strains.

Guidelines for interpretation of the mecAdetection result will need to be formallyaddressed as more laboratories begin to use thisand other genetic methods. Proposals have beenmade to regard mecA-positive organisms (bothcoagulase-negative staphylococci and S. aureus)as intrinsically resistant to all antibiotics exceptvancomycin and to report immediately all mecA-positive results, which can be available wellbefore results from traditional methods. Thereare situations where the mecA-positive organismdoes not express resistance clinically and may

respond to β-lactam therapy. It is important todocument these cases carefully to avoid unneces-sary use or overuse of vancomycin. Neverthe-less, all mecA-positive organisms may be highlyresistant if the organism expresses the mecAgene. This may lead to treatment failures if β-lac-tam antibiotics are chosen.

Surveillance and DNA-based diagnosticsSurveillance of genetic mutations in bacteria willbe essential in the use of new DNA diagnostics,which measure the presence of specific geneticsequences. Mutations might alter thesesequences, or new genes conferring resistancemay spread. For example, widespread surveil-lance efforts are necessary to insure that signa-ture sequences represent the majority ofmutations in the rpoB gene that confer rifampinresistance in M. tuberculosis.

❚ Regulation of Diagnostic TestsThe Clinical Laboratory Improvement Amend-ments (CLIA ’88) were passed by Congress toregulate the quality of diagnostic testing. Regula-tions under CLIA, which became effective inSeptember of 1992, require all clinical laborato-ries that perform certain diagnostic tests to regis-ter with the federal government and performquality control tests and document quality assur-ance. However, certain tests are “waived” underCLIA; this means that the test can be performedin any physician’s office, whether or not theoffice is registered under CLIA. Other tests, gen-erally the more complex ones, can be performedonly in offices that comply with the CLIA regu-lations for laboratories.

The CLIA regulations may be a disincentiveto performing tests. Complying with themincreases the cost of testing and may delayresults. For example, physicians may choose notto register their offices under CLIA and willtherefore be compelled to send out numeroustests that they formerly performed. This mayresult in the performance of fewer diagnostictests, which could contribute to the overuse ofantibiotics. A physician might decide that it is


easier and more cost-effective to prescribe anti-biotics for all sore throats rather than performthroat cultures. However, this negative potentialconsequence of the CLIA regulations must beweighed against whatever positive effects theyhave had on the quality and consistency of test-ing that is done in the clinical laboratories thatmeet CLIA standards.

❚ Getting New Tests to MarketThe worst outcome for a sensitivity test is toindicate that bacteria are susceptible to an antibi-otic when the antibiotic has no effect against thatstrain. Such an error, which can result in apatient’s death, is called a “very major error” intesting. The second worst outcome is to reportthat bacteria are resistant to an antibiotic that isin fact effective against them. That error, whichcould result in treatment with a more toxic, moreexpensive antibiotic than necessary, is termed a“major error” (Jorgensen 1995).

It is impossible to design and perform teststhat are completely error free. The manufactur-ers, the FDA, health care providers, and the pub-lic have to decide what levels of errors areacceptable. Often, new tests are compared with a“gold standard”—an older test that has beenproved to be reliable. However, the “gold stan-dard” is also not completely error free. Thereforeit is sometimes difficult to interpret differencesbetween a new test and a “gold standard.” Forexample, culturing M. tuberculosis is consideredthe “gold standard” for the diagnosis of tubercu-losis. However, Abe, et al. (1993) found thatsome patient samples were positive for M. tuber-culosis by DNA-based techniques but negativewhen cultured; these patients had clinical signsof tuberculosis, including characteristic radio-graphs, clinical manifestations of the diseaseand/or clinical response to antituberculosis che-motherapy.

Two FDA centers are involved in approvingtest methods for antibiotic susceptibility. TheFDA Center for Drug Evaluation and Researchcertifies that disks are available for each antibi-otic on the market in the United States, and it

assures the potency of the disks and that criteriafor interpretation of the disk assays are availablewhen an antibiotic goes on the market. The FDACenter for Devices and Radiological Health hasresponsibility for determining the safety andeffectiveness of other devices and materials,including computer software, for susceptibilitytesting.

A new diagnostic device can be reviewed byFDA under two different procedures. A device ormethod that employs principles similar to thoseused by products already on the market and thatrequires an incubation period of 16 hours ormore is reviewed under the “510(k) clearance”process. The performance of the new device ormethod is compared to the performance of theproduct already marketed to determine whetherthe two are “substantially equivalent.” If theyare, the new device or method is cleared for mar-keting without undergoing the more extensiveprocedures, known as “pre-market approval.”The 510(k) process is also used when a manufac-turer wants to add a new antibiotic to the batteryof antibiotics already included in a test kit.

New diagnostic tests that are not “substan-tially equivalent” to any product on the marketmust submit an application for “pre-marketapproval” (PMA) to the FDA. Because theapproval process under the PMA review is sub-stantially more difficult, manufacturers have adisincentive to develop novel products.

Any device that requires less than 16 hours’incubation is required to undergo the pre-marketapproval process, which takes longer and is sub-stantially more difficult to complete than the510(k) clearance process. Jorgensen (1995)claims that there is no clear justification for the16-hour incubation period serving as the cutoffbetween a 510(k) review and a PMA reviewbecause there is no indication that more rapiddevices are inherently less accurate than others.The difference in the time required to obtain a510(k) clearance, as opposed to a pre-marketapproval, is a matter of contention. According toJorgensen (1995), the requirements for a 510(k)clearance have grown since 1990, and they arenow approaching those required for a PMA. On


the other hand, FDA (1995) asserts that therehave been marked improvements in the process-ing of 510(k) applications.

VACCINESPerhaps the ultimate weapon against antibiotic-resistant bacteria is the development of vaccinesand pre-emptive immunization. In concept, vac-cines are simple. When a person receives aninoculation of a preparation of killed or attenu-ated (“weakened”) disease-causing bacteria orvirus, a component of such an agent, or a relatedorganism that does not cause disease, the inocu-lated person’s immune system will respond andproduce antibodies to antigens on the injectedmaterials. The immune system has a “memory.”As a result, if the person is subsequently infectedby the organism for which the vaccine was pre-pared, he or she will produce antibodies that caninactivate the agent and remove it from the body.“Natural immunity” is produced in a similarway; once a person has had a disease, theimmune system recognizes the organism thatcaused it and eliminates it from the body.

In practice, preparation of the specific mate-rial for the inoculation—the antigen—can be dif-ficult. Preparing it so that the production ofantibody is stimulated without objectionable tox-icity, either at the time of inoculation or later,may not be simple.

The success of Haemophilus influenzaetype B (Hib) vaccines, which were introduced in1988, demonstrates that antibacterial vaccinescan be quite successful. Countering that greatsuccess is the more than 75 years’ experiencewith an antituberculosis vaccine.

❚ Hib Vaccines, a Success StoryBefore the introduction of vaccines against it,Hib (Haemophilus influenzae type B) was theleading cause of invasive bacterial disease inchildren under five years of age, and it causedabout 20,000 cases of meningitis and another3,000 to 5,000 cases of invasive Hib diseaseannually. The mortality rate was 3 to 5 percent;moreover, up to 20 percent of the survivors of

meningitis suffered hearing loss or mental retar-dation, and resistance to ampicillin was increas-ing (Adams, Deaver, Cochi, et al. 1993).

In 1993, five years after the introduction ofHib vaccines, a number of researchers publishedreports about the incidence of Hib diseases inchildren up to five years old. Those vaccinatedwith Hib vaccine generally had disease rates 80to 90 percent below the rates seen in unvacci-nated children (Wenger 1994). The rates of Hibmeningitis began to fall in 1989, after the intro-duction of the vaccine, and they continued to fallthrough 1991 (the last year for which data wereavailable). In contrast, the rates of meningitisfrom Neisseria meningitidis and Streptococcuspneumonia remained unchanged, ruling out ageneral decline in meningitis as the explanationfor the Hib results. An unexpected result of theHib vaccination program was a reduction in thenumber of children who carry Hib in their upperairways. That, in turn, reduced the number ofchildren who could infect others, and the rates ofHib disease have fallen in both vaccinated andunvaccinated children.

A polysaccharide (a polymer of sugar mole-cules that is unique to the Hib bacteria) vaccinelicensed in 1985 had no effect on the occurrenceof invasive Hib disease in Los Angeles County(see figure 6-4) and, in fact, it was of little valuein disease prevention. Three years later, a conju-gate vaccine, prepared by chemically joining thepolysaccharide to a protein that was known tostimulate antibody production, was licensed.This vaccine was very successful. Even when useof this vaccine was restricted to children olderthan 18 months (from 1988 through 1990), therewas a drop in the Hib invasive disease rate inyounger children. Vaccination of the older chil-dren had reduced infections of the younger ones,due to reduced transmission of the bacteria.Licensing of a vaccine for 2-month-old childrenin 1990 led to great reductions in the disease inLos Angeles County by 1992.


Conjugate vaccinelicensed for 18 month-olds

Polysaccharide vaccinelicensed for

24 month-olds

I

I1 1

Conjugated vaccinelicensed for

-oIds

1983 1984 1985 1986 1987 1988 1989 1990 1991 1992

SOURCE: Division of Microbiology and Infectious Diseases, NationalInstitute of Allergy and Infectious Diseases. 1994. Annual Report,p. 1

❚ BCG Vaccine, 75 Years’ ExperienceAlbert Calmette and Camille Guerin at the Pas-teur Institute in Paris first produced BCG as avaccine for the prevention of TB. BCG is madefrom preparations of a live, attenuated strain ofM. bovis, which is closely related to M. tubercu-losis, and over 70 percent of children worldwidenow receive the vaccine. It is compulsory in 64countries and recommended in 118 others (OTA,1993). People who have received this vaccinetypically show a positive response to tuberculinskin tests. This is considered a great disadvan-tage in the U. S., where tuberculin skin tests areused to screen for exposure to tuberculosis.

Colditz, Brewer, Berkey, et al. (1994) reportedthe most thorough review of the efficacy of BCGvaccine. Their meta-analysis of the worldwideliterature led to the conclusion that BCG reducedthe risk of TB by about 50 percent, but the suc-cess rate vanes from batch to batch of the vac-cine, which is prepared in different laboratoriesunder different conditions around the world. The50 percent effectiveness conclusion was chal-

lenged by a number of scientists (Benin, 1994;Wheeler, Rodrigues, and Diwan, 1994; Com-stock, 1994), but the authors replied that “. . .meta-analysis shows that the preponderance ofevidence reveals that BCG vaccine is effective inpreventing TB” (Coldwitz, Brewer, Berkey, etal., 1994a).

The United States has never required the vac-cine because of questions about its efficacy andits usefulness in a population with a low inci-dence of TB. In 1979 the Centers for DiseaseControl and Prevention (CDC) recommended thevaccine for health care workers in contact withTB patients, but CDC’s 1988 policy statementreversed that recommendation because of thelack of evidence for increased TB among healthcare workers (OTA, 1993). CDC recommendBCG for members of high risk groups who havelimited access to health care. However, the CDCbelieves that the rate of tuberculosis is so low inthe general population of the U.S. that the advan-tages conferred by vaccination are outweighedby the disadvantage of being unable to screen thepopulation by using the tuberculin skin test.

❚ Vaccine ResearchSuccessful vaccines are available for use againstviral diseases such as measles, mumps, andrubella, and against bacterial diseases such asdiphtheria, tetanus, and pertussis (whoopingcough). Currently, researchers are pursuing newvaccines against bacterial pathogens, such as theStreptococcus species, Staph. aureus a n dHelicobacter pylori, which are common prob-lems, in part because of high rates of antibioticresistance.

Active Systemic ImmunizationActive immunization is the process of adminis-tering specific microbial antigens that stimulatethe host’s immune system to produce protectiveantibodies. Active immunization can besystemic-the traditional method—or mucosal(discussed below). Systemic immunity is accom-plished by injection, the result being long-livedproduction of circulating immunoglobulin G


(IgG) antibodies. For bacterial vaccines, polysac-charides from the outside capsule of the bacteriaare generally employed, but as was seen with theHib vaccines (Wenger, 1994), capsular polysac-charides alone do not always stimulate sufficientantibody production. To raise sufficient levels ofantibodies, the polysaccharides may have to beconjugated with “carrier proteins,” potent immu-nogens that provoke an immune system responseto the entire complex (i.e., to polysaccharideantigen and protein carrier). The combination ofpolysaccharide and protein is called a “conjugatevaccine.” Finding the proper carrier is one of themore difficult aspects of vaccine development,but four different proteins—all of bacterial ori-gin—work well in Hib vaccines.

Streptococcus pneumoniae vaccinesVaccine development is further complicatedbecause different strains of the same bacteriahave different polysaccharide antigens. Forinstance, Streptococcus pneumoniae has 84 dis-tinct capsular polysaccharides. A vaccine thatcontains 23 different polysaccharides providesprotection against 90 percent of invasive pneu-mococcal strains (Siber, 1994). That vaccine is75 percent effective when administered to immu-nocompetent adults, and its wider use might pre-vent half of the 80,000 annual pneumococcalpneumonia deaths among older people (MedicalWorld News, 1993). It is not, however, reliablein children under two years of age. For the vacci-nation of children, several companies are devel-oping conjugated vaccines against thepolysaccharides of the seven strains of pneumo-coccus that most commonly infect children, andthese are currently undergoing human trials. Inaddition, researchers are investigating the possi-bility of using a polysaccharide that is commonto all pneumococcus conjugated to one or moreof several proteins that are common to all pneu-mococcus as vaccines, but there is no definitiveevidence for their usefulness (Siber 1994).

Vaccines against otitis mediaS. pneumonia is one of several bacteria that causeotitis media. That illness is so notorious that phy-

sicians who surveyed parents about their willing-ness to have their children immunized against thedisease titled their report “The Surprisingly HighAcceptability of Low-Efficacy Vaccines for Oti-tis Media...” (Wischnack, Jacobson, Poland, etal., 1995). Although no such vaccine is now inuse, the interviewers asked about five hypotheti-cal vaccines that had different efficiencies in dis-ease prevention, and side effects that rangedfrom the temporary discomfort of a “shot” in allchildren to a few days of flu-like symptoms in upto half of vaccinated children. About half of the601 interviewed parents would accept any vac-cine if it would prevent three or more infectionsin the next six months. Parents were less accept-ing of vaccines with lower efficiencies and worseside effects. The authors of the study concludethat parents, even of children who have not hadotitis media, are willing to accept some discom-fort in their children to obtain protection againstthe disease and that parents’ willingness isgreater than the medical establishment’s orFDA’s.

A biotech firm, MicroCarb (Bethesda, Mary-land), has licensed one vaccine against Haemo-philus influenza, another cause of otitis media, toPasteur Merieux Serums et Vaccines S.A. If thisvaccine proves successful, it will be of valueagainst both antibiotic susceptible and resistantH. influenza, which are increasingly common.

Staphylococcus aureus vaccinesVaccines against Staph. aureus, which is oftenantibiotic resistant, would be helpful to patientsat high risk for infection with this organism,including renal dialysis patients, or patientsreceiving prosthetic devices like hips or vasculargrafts, which act as sites for infection (Univax,1994). Researchers are pursuing vaccines madeof capsular polysaccharide types 5 and 8, whichwould encompass 90 percent of Staphylococcussystemic infections. Recent research has shownthat high levels of biologically active antibodiesagainst Staph. aureus types 5 and 8 can be stimu-lated in human subjects when the antigens arecombined with a protein from P. aeruginosa as


the carrier (Fatton, et al., 1990; Fatton, et al.,1993).

Active Mucosal ImmunizationThe second approach to active immunization isto stimulate the immune defenses of the mucosallinings of the gastrointestinal, respiratory andurogenital tracts, the nasal passages, and theinner ear. These mucosal linings produce immu-noglobulin A (IgA). IgA diminishes microbialvirulence by preventing microbial adherence tohost cells. It also coats the surface of the antigen,making an antigen/IgA complex that stimulateswhite blood cells to recognize, engulf, anddestroy any pathogen expressing that antigen.Mucosal lymphocytes also trigger production ofcirculating IgG antibodies. Current targets formucosal immunity include Helicobacter pylori,Clostridium difficile, Shigella flexneri, Campylo-bacter strains, and certain strains of Escherichiacoli.

Mucosal vaccines are immunogenic only ifthey reach specific immune response tissuesbeyond the stomach, which requires their surviv-ing passage through stomach acid and enzymes.Some researchers are testing synthetic polymersto protect their vaccines. Another strategy is touse liposomes, lipid-containing vesicles madefrom the same natural materials that composemammalian cell membranes.

Helicobacter pylori vaccinesThe discovery by Marchetti, Arico, Burroni, etal. (1995) that bacteria isolated from humanswith ulcers could infect mice and cause diseaseprocesses that mimic those seen in humans hasspurred progress toward a vaccine against H.pylori, the causative organism. Those researchersfound that mice were protected from infectionafter administration of disrupted H. pylori bacte-ria. This finding, characterized as “of extremepractical importance” (Tompkins and Falkow1995) may lead to a vaccine to protect the50 percent of the world’s population that are cur-rently infected by H. pylori. While H. pylori ismost often associated with gastritis and ulcers in

the United States, elsewhere in the world it isalso a common cause of stomach cancers.

Campylobacter vaccinesCampylobacter strains have recently emerged asone of the common causes of diarrhea and maycause 2.5 million cases annually in the UnitedStates. Treatment is increasingly complicated byantibiotic resistance. In 1994, the U.S. Navysigned a Cooperative Research and DevelopmentAgreement (CRADA) with MicroCarb Inc. forclinical trials of a vaccine against Campylo-bacter.

Passive immunizationPassive immunization involves administeringantibodies directed against specific pathogens toindividuals who have not developed such anti-bodies on their own but who are at risk for infec-tion. This may be due to lack of prior exposure tothe pathogen, or due to immunosuppression,which renders the individual’s immune systemunable to produce antibodies. The antibodies arepurified from the blood of healthy donors whoseantibody levels are raised by active immuniza-tion. The most common example of passiveimmunization is the administration of HepatitisB virus-specific gamma-globulin to travelers.Researchers are currently focusing efforts todevelop antibodies against Staph. aureus and P.aeruginosa, both of which are often antibioticresistant, as well as against other bacteria.

Passive immunization does not always work.Low birthweight babies are at high risk of noso-comial infections because of long hospitaliza-tions and immature immune systems. Injectionsof pooled human antibodies (“immune globu-lin”) into very low birthweight babies did notreduce the incidence of nosocomial infectionscompared to the incidence in very low birth-weight babies who did not receive the immuneglobulin (Fanaroff, Korones, Wright, et al.,1994). This failure does not invalidate the idea ofpassive immunization, even in low birthweightbabies, but it underlines the importance of trialsof the efficacy of interventions before they areintroduced widely into practice.


Vaccine SummaryVaccines are not high-profit items. UNICEF esti-mates that the entire global vaccine market isabout $3 billion, which can be compared to the$3.5 billion market for a single ulcer drug. Whilevaccine development against bacteria that havehigh frequencies of antibiotic-resistant strains,such as Staph. aureus or S. pneumonia, wouldreduce infections by those bacteria, few vaccineswill be developed for bacteria solely because ofthe problems raised by antibiotic resistance.Instead, the general problems raised by the bacte-ria may lead to development of a vaccine thatwill protect against both antibiotic susceptibleand resistant strains.

STIMULATING THE IMMUNE SYSTEMGranulocyte colony-stimulating factor (G-CSF)is a growth factor that stimulates the proliferationof neutrophil cells, important components of theimmune system. Crawford, Ozer, Stoller, et al.(1991) have shown that the administration of G-CSF to cancer patients on chemotherapy led to a51 percent reduction in culture-confirmed infec-tions, a 47 percent reduction in the mean numberof days of antibiotic use, and a 45 percent reduc-tion in the mean number of days of hospitaliza-tion. G-CSF in the form of filgrastim (Amgen,Thousand Oaks, California) has been approvedby the FDA and is clinically available.

TARGETED DELIVERY OF ANTIBIOTICSSome sites of infection or potential infection arelocalized, such as wounds or the area around ajoint replacement. Delivery of antibiotics directlyto those sites may stop the growth of susceptiblebacteria, and if the concentration can be raisedhigh enough, it may even stop the growth ofmany resistant bacteria. Direct delivery of antibi-otics in this way has the additional advantage ofproducing only very low levels of circulatingantibiotics, thus reducing pressure for the selec-tion of resistant bacteria elsewhere in the body.

❚ MicroencapsulationEntry into the body, whether surgical or trau-matic, opens pathways for infection. Surgicalpatients who develop wound infections spent, onaverage, 14.3 days longer in the hospital thanuninfected matched controls (Maderazo, Judson,and Pasternak, 1988), at an increased cost of$36,000 to $45,000 per patient (Cohen, 1994;Daly, Eliopoulos, Reiszner, et al., 1988).Twenty-four percent of United States servicemenwho sustained open fracture wounds in Panamaduring Operation “Just Cause” developed woundinfections (Jacob, Erpelding, and Murphy, 1992),and 48 percent of wounded United States sol-diers in the Persian Gulf conflict who sustainedopen fractures developed postoperative infec-tions (Travis and Cosio, 1993). Gustilo, Men-doza, and Williams (1984) report similarinfection rates in civilians with severe open frac-tures of the tibia. Many of these infections occurin patients who receive very large doses of sys-temic antibiotics.

Researchers at the Walter Reed Army Instituteof Research (WRAIR) have developed a novelbiodegradable local antibiotic delivery systemthat promises to decrease infections in wounds.They encapsulate an antibiotic in a copolymer ofpoly (DL-lactide-coglycolide) to produce micro-spheres 50 to 250 micrometers (µm) in diameter.Dusted into wounds after surgery, these micro-spheres provide an initial burst of the antibioticwithin the first few hours and prolonged drugrelease over a period of up to 21 days. After 2 to3 months, the microspheres completely degrade.As of March 1995, the WRAIR researchers hadconstructed microspheres containing ampicillin,cefazolin, cefamandole, and tobramycin.

Cefazolin-containing microspheres were usedto treat wounds in rats that had been intentionallyinfected with cefazolin-resistant MRSA, andthey were as effective as free cefazolin powder ineliminating MRSA. Systemic administration ofcefazolin, on the other hand, had no effect on theMRSA infections. In a similar experimentinvolving ampicillin-resistant MRSA, micro-spheres containing ampicillin were more effec-


tive than free ampicillin powder, and systemicampicillin had no effect.

The United States Army, which developedthis technique (Setterstrom, Tice, and Myers,1994; Jacob, Setterstrom, Bach, et al., 1991;Jacob, Cierny, Fallon, et al., 1993), has a patentpending on it. Further development will requireprivate funding to take the research from the pre-clinical stage to trials in humans.

❚ Antibiotic-Impregnated CementBone infections and infections of joint prosthesesare hard to treat with systemic antibiotics, par-tially because limited blood flow to the skeletaltissues does not allow high concentrations of thedrug to reach the area of infection. An antibiotic-impregnated polymer, poly (methyl-methacry-late) (PMMA), has been used to cement bonefractures and prostheses in place, and has shownclinical success, but its usefulness is limited bythe toxicity of the material and shrinkage whichleaves marginal mechanical support for theremaining bone. Yu, et al. (1992) describedhydroxyapatite (HAP) cement, which has thesame chemical composition as bone mineral.This material can be molded to fill the space leftby the absence of bone, and Yu, et al. demon-strated that antibiotics impregnated in this mate-rial are slowly released. They concluded that thismaterial is very promising for preventing infec-tions in bone fractures and in joining prostheses,and they propose future in vivo experiments.

❚ Biological Substances to Facilitate the Entry of Antibiotics into BacteriaOne mechanism of resistance involves bacterialcell walls in excluding antibiotics from the bacte-rial cell. Research is underway on biological sub-stances that allow antibiotics to penetrate intosuch bacteria. For example, because iron is insol-uble but necessary for bacterial metabolism, bac-teria synthesize and excrete compounds that canbind iron ions, called “siderophores.” Thesecompounds scavenge iron outside the cell, andthe cell then transports the iron-siderophore com-pound back inside the cell. Inside the cell, the

iron-siderophore complex is metabolized by thebacteria, releasing iron for bacterial use. Sidero-phores may be modified to carry antibiotics intothe bacteria. These may be especially useful inthe treatment of Gram-negative bacterial infec-tions. Although the outer cell membrane chan-nels (“porins”) of Gram-negative bacteria are toosmall to accommodate many antibiotics, sidero-phores enter the cell via a non-porin route, andresearchers reason that antibiotics attached tosiderophores might be “dragged” inside.

Over 200 siderophore molecular structures areknown. Often, only portions of the siderophoresare required to penetrate the cell. One goal ofcurrent research is to optimize synthetic sidero-phores in order to make their transport intobacterial cells more efficient. Synthetic sidero-phores, when conjugated with beta-lactam antibi-otics or erythromycin, can carry the antibioticacross bacterial cell membranes with high effi-ciency. These antibiotics kill bacteria whendelivered inside the cell in this manner (Miller,1989; McKee, Sharma, and Miller, 1991).Siderophores are also being explored for theirpotential to transport vancomycins. Althoughsiderophore/antibiotic conjugates have thus farbeen used only as antibacterials, researchers arecurrently attempting to apply the same methodol-ogy to antifungal/siderophore conjugates.

REDUCING INFECTIONS BY MODIFYING DEVICESSeveral hundred thousand cases of hospitalacquired infection per year are related to the useof medical devices such as catheters, endotra-cheal tubes and mechanical ventilators (IOM,1992). These devices provide extra opportunitiesfor bacteria to enter the body. Experience withdialysis, the filtering of the blood of patients withkidney disease, indicates that changing thedesign and materials of medical devices can min-imize infections.

❚ Infections in Dialysis PatientsIn 1991, there were approximately 120,000patients on maintenance dialysis (Favero, Alter,


and Bland, 1992) with 45,000 new patientsadded per year. Infections are the cause of deathin 15 to 30 percent of dialysis patients.

The technique called hemodialysis is used totreat approximately 85 percent of dialysispatients. Simulating the function normally per-formed by the kidney, it filters the patient’sblood through a membrane which separates outunwanted components and adds needed compo-nents. Cupraphane membranes, the most com-monly used filtration membrane in hemodialysis,are made from cotton fibers dissolved in anammonia solution of cupric oxide. Recently,membranes made of synthetic polymers such aspolysulfone (PS), polymethylmethacrylate(PMMA) and polyacrylonitirile (PAN) havebeen developed. A recent review of the proper-ties of hemodialysis membrane (Hakim, 1993)describes how the interaction of blood with cot-ton fiber membranes such as Cupraphane pro-duces a decrease in the immune functions in theblood, leaving the patient more susceptible toinfection. The membranes made of syntheticpolymers do not seem to decrease the immunefunctions in the blood. Retrospective studiesshowed that replacing a Cupraphane membranewith a polysulfone membrane eliminated50 percent of the infections.

Another 15 percent of patients are on perito-neal dialysis. In this technique, fluid is pumpedinto the patient’s abdomen, allowing exchange ofblood components through the peritoneal liningof the abdomen. A recent review (Diaz-Buxo,1993) shows that the incidence of peritonitis(peritoneal infection) was twice as high whenolder CAPD (continuous ambulatory peritonealdialysis) machines were used than when newdialysis machines of different design, such asCCPD (continuous cyclic peritoneal dialysis)machines and Y-set connections for CAPD, wereused. This may be because the order of flow isreversed in CCPD and Y-set CAPD compared toother forms of CAPD, so that the connections(and contaminating bacteria) are washed outbefore fluid is pumped into the body. Diaz-Buxocomments that CAPD machines are more com-mon than CCPD machines, partially because of

the lower cost of the machine itself. When thetotal costs of the two systems were calculated,including the cost of the machine and the cost ofhospitalization for peritoneal infections, the totalcosts were the same (King, et al., 1992).

Analyzing the costs of dialysis for kidneypatients is especially interesting because dialysispatients have been covered by Medicare since1973 regardless of their age. Medicare pays a setamount per patient for dialysis and pays sepa-rately for any hospitalization necessitated bycomplications. Under this system, physicianshave a financial incentive to use the least expen-sive equipment. However, it would be beneficialto the patients, and probably cheaper for Medi-care, to use the more expensive equipment andprevent infections that may require hospitaliza-tion. Outpatient costs, primarily dialysis,accounted for 33 percent of total costs comparedwith 44 percent of total costs attributable to hos-pitalizations (Smits, 1995). (The remainder of thecosts were for physician services, skilled nursingcare, and home health care.) This demonstratesthat investing in new technologies that preventinfections and hospitalizations can be cost-effec-tive. These investments would also reduce antibi-otic resistance by preventing infections and thusreducing the use of antibiotics.

❚ Infections from Sutures and CathetersImprovements in the materials used for othermedical devices such as sutures and catheterscould also greatly reduce the rate of infection. Inparticular, sutures made of synthetic materialssuch as dacron and nylon have lower infectionrates compared to natural sutures such as cotton,silk and catgut, and monofilament sutures havelower infection rates compared to polyfilamentsutures.

Studies of the colonization of medical devicesby coagulase-negative staphylococci (Chris-tensen, Baldassarri, and Simpson, 1994) providessome insight into why some suture materials areassociated with infections more than others. Theprocess of colonization of non-biological sur-faces by coagulase-negative staphylococci is


shown in the photograph. The first step in coloni-zation is binding and/or trapping a “unique site”on the surface, such as a microscopic crack ordepression in the surface of the material. Syn-thetic materials such as nylon and plastics aregenerally much smoother than natural materialssuch as cotton and silk and therefore have fewerunique sites. Similarly, monofilament aresmoother than polyfilaments. Therefore, it is notsurprising that the natural materials and polyfila-ments are more often associated with infectionsthan the synthetic materials and monofilaments.Knowledge about the colonization and infectionprocess for non-biological materials will helpguide new designs of medical devices that mayminimize infections and reduce the need for anti-biotics.

Maki (1 994) reviewed innovative designs thathelp prevent infections in intravascular cathetersused for infusion therapy. Some catheters have anew design that creates mechanical barriersagainst infection at the entrance of the catheter to

the skin. Other designs create a closed system;for example, they replace the stopcocks used toobtain blood specimens from arterial lines with adiaphragm. Such closed systems reduce the rateof infection.

Another strategy for preventing infections isto coat the materials used in medical deviceswith antibiotics or other antibacterial agents.Like the microencapsulated antibiotics and anti-biotic-impregnated cement, these coated cathe-ters may have the advantage of delivering highconcentrations of antibiotics to the site of poten-tial infection with much lower systemic antibi-otic concentrations. In one system, the cathetersare coated or impregnated with silver ions, whichare bactericidal but non-toxic to humans. (Manu-facturers include Arrow International and C.R.Bard Urological Division; Maki, et al., 1991;Stamm, 1991). In another system, catheters arecoated with materials bearing positively chargedchemicals, to which negatively charged antibiot-

Series of four microphotograph demonstrating the colonization of a non-biological surface.

Courtesy of Meryl E. Olson, University of Calgary, Calgary, Canada

4


ics are bound (Cook Bio-Guard AB coated cathe-ters, Cook Critical Care). A trial with thesecatheters coated with cefazolin showed a seven-fold decrease in the infection rate (Kamal,Pfaller, Rempe, et al., 1991). Further, a reductionin the infection rate was seen even if the cathe-ters were changed only once every seven days(compared to once every four days for standardcatheters; Kamal, Divishek, Adams, et al., 1994).The longer life of the coated catheter compen-sates for its higher cost (about $4.50 more percatheter).

OLD THERAPIESIn the pre-antibiotic era, scientists and physicianstried different methods to treat bacterial infec-tions. Two of those methods, “phage therapy”and “serum therapy,” are now mentioned as pos-sible treatments in a post-antibiotic era.

❚ Phage TherapyWhile most people may not recognize the term“phage therapy,” many people read about it inArrowsmith. The hero of that novel tried to treatbacterial infections by the use of viruses thatwould specifically attack the bacteria, and in reallife, many physicians tried the same method inthe early part of this century. Because virusesthat infect bacteria are called “bacteriophages”(literally, eaters of bacteria) or “phages” forshort, the treatment is called “phage therapy.”Phage therapy has remained outside the main-stream of medicine because of doubts about itsefficacy and the success of antibiotics.

Phages recognize specific binding sites on thebacteria. Therefore, phages that infect E. coligenerally do not infect other bacteria, and, infact, sometimes will only recognize a singlestrain of bacteria. This specificity offers thepromise of being able to prepare phages to attackparticular bacteria.

Levin and Bull (1995) and Levin, DeRouin,Moore, et al. (1995) review the literature aboutphage therapy. They focus on some recent exper-iments with systems that involve mice infectedwith E. coli and argue that phage therapy is

worth renewed investigation. While they do notthink that it will replace antibiotics, they believethat it may have some future use in treating anti-biotic-resistant bacteria. They also argue that thetime to develop alternatives to antibiotic therapyis now, when antibiotics remain effective againstmost diseases.

❚ Serum TherapyTextbooks of medicine and of microbiology pub-lished before 1940 are filled with instructions forserum therapy. In some respects similar to pas-sive immunization, serum therapy involves tak-ing blood serum from horses or rabbits that havesurvived an intentional bacterial infection andinjecting it into a patient suffering from an infec-tion by the same organism.

Serum is still used in the treatment of somediseases that involve bacterial toxins; in particu-lar, tetanus and botulism are treated with horseserum. Serum for the treatment of botulism iskept at several major airports around the country,ready for shipment to hospitals that diagnose therare disease. (According to the CDC [1979],there were about 10 outbreaks of botulism,involving about 2.5 people per outbreak, eachyear in the period 1899 through 1977.) For otherinfections, serum therapy was replaced as antibi-otics became available. A patient’s possible ana-phylactic response to chemical substances in theanimal serum is the chief danger.

Serum therapy may have application in treat-ing Escherichia coli O157:H7, which becamefamous as the cause of more than 500 cases ofdisease and perhaps four deaths in people whoate under-cooked fast-food hamburgers in thePacific Northwest in early 1993. The usual treat-ment for the disease does not include antibiotics(Salyers and Whitt, 1994). Antibiotics have notbeen shown to shorten the course of the diseaseor to reduce the occurrence of kidney complica-tion. Further, antibiotic treatment may cause thebacteria to increase the production of the bacte-rial toxin that causes the disease. The cause ofdisease in E. coli O157:H7 infections is a toxinthat resembles the Shigella toxin that causes dys-


entery. That toxin has been isolated and purified.Antibodies generated against the toxin havepotential in treating E. coli O157:H7-caused dis-eases, but the market for such a drug is small,and no trials are in progress.

SUMMARYThis chapter reviews some new technologies thatwill help health care providers use antibioticsmore effectively. Diagnostic technologies helpthe clinician identify the specific bacteria caus-ing the infection and its susceptibility to antibiot-ics. This information is critical for choosing themost appropriate antibiotic. New technologies,such as DNA identification of antibiotic resis-tance genes, have the potential to provide thisinformation more quickly than is possible withtraditional diagnostic tests, which require grow-ing the bacteria in cultures. These new diagnostictechnologies have already proven useful in diag-nosing tuberculosis. Many companies are rapidlydeveloping additional tests for TB and other bac-teria. There are unresolved issues with respect tothe accuracy, sensitivity, and reproducibility ofthese tests. These issues may not be resolveduntil the tests have received FDA review and arewidely used in clinical settings. This chapter dis-cusses some of these issues.

Preventing infections is another way to slowthe increase of antibiotic-resistant bacteriabecause prevention will reduce the total use ofantibiotics. Methods of preventing infectioninclude vaccines and changes in the design andcomposition of medical devices to prevent thegrowth of bacteria. The recent introduction of avaccine against Hemophilus influenza B resultedin a dramatic reduction in the incidence of child-hood diseases caused by this bacteria. A numberof other vaccines are under development, includ-ing those for Staphylococcus aureus, as well asbetter vaccines for Streptococcus pneumoniae.Indwelling devices and sutures are often ports ofentry for bacteria into the body, and improveddevices and materials have been shown to reduceinfection rates. Further research and applicationcould produce further reductions.

REFERENCESAbe, C., K. Hirano, M. Wada, et al. 1993. Detec-

tion of Mycobacterium tuberculosis in clini-cal specimens by polymerase chain reactionand Gen-Probe Amplified MycobacteriumTuberculosis Direct Test. Journal of ClinicalMicrobiology 31:3270–3274.

Adams, W.G., K.A. Deaver, S.L. Cochi, et al.1993. Decline of childhood Haemophilusinfluenzae type b (Hib) disease in the Hibvaccine era. Journal of the American Medi-cal Association 269:221–226.

American Medical News. March 27, 1995. Chi-cago, IL. American Medical Association.

Bellin, E. 1994. Letter to the Editor. Journal ofthe American Medical Association 272: 765.

Christensen, G.D., L. Baldassarri, and W.A.Simpson. 1994. Colonization of medicaldevices by coagulase-negative staphylo-cocci. In: A.L. Bisno and F.A. Waldvogel(eds.) Infections Associated with IndwellingMedical Devices. Washington, DC. ASMPress.

Cohen, M.L. 1994. Antibiotic resistance: Prog-nosis for public health. Trends in Microbiol-ogy 2:422–425.

Coldwitz, G.A., T.F. Brewer, C.S. Berkey, et al.1994. Efficacy of BCG vaccine in the pre-vention of tuberculosis: Meta-analysis of thepublished literature. Journal of the AmericanMedical Association 271:698–702.

Coldwitz, G.A., T.F. Brewer, C.S. Berkey, et al.1994a. Letter to the Editor. Journal of theAmerican Medical Association 272:766.

Comstock, G. 1994. Letter to the Editor. Journalof the American Medical Association272:766.

Crawford, J., H. Ozer, R. Stoller, et al. 1991.Reduction by granulocyte colony-stimulat-ing factor of fever and neutropenia inducedby chemotherapy in patients with small-celllung cancer. New England Journal of Medi-cine 325:164–170.

Daly, J.S., G.M. Eliopoulos, E. Reiszner, et al.1988. Activity and mechanism of action ofDuP105 and DuP721, new oxazolidononecompounds. Journal Antimicrobial Chemo-therapy 21:721–730.


Diaz-Buxo, J.A. 1993. Differences betweenCAPD and CCPD: Implications for therapy.Seminars in Dialysis 6:312–315.

Doern, G.V., D.R. Scott, and A.L. Rashad. 1982.Clinical impact of rapid antimicrobial sus-ceptibility testing of blood culture isolates.Antimicrobial Agents and Chemotherapy21:1023–1024.

Doern, G.V., R. Vautour, M. Gaudet, et al. 1994.Clinical impact of rapid in-vitro susceptibil-ity testing and bacterial identification. Jour-nal of Clinical Microbiology 32:1757–1762.

Edwards, L.D., S. Levin, R. Balogtas, et al. 1973.Ordering patterns and utilization of bacterio-logic identification. Archives of InternalMedicine 132:678–682.

Fanaroff, A.A., S.B. Korones, L.L. Wright, et al.1994. A controlled trial of intravenousimmune globulin to reduce nosocomialinfections in very-low-birth-weight infants.New England Journal of Medicine330:1107–1113.

Fatton, A., R. Schneerson, S.C. Szu, et al. 1990.Synthesis and immunologic properties inmice of vaccine composed of Staphylococ-cus aureus type 5 and type 8 capsularpolysaccharides conjugated to Pseudomonasaeruginosa exotoxin A. Infection and Immu-nity 58:2367–2374.

Fatton, A., J. Shiloach, D. Bryla, et al. Compara-tive immunogenicity of conjugates com-posed of the Staphylococcus aureus type 8capsular polysaccharide bound to carrierproteins by adipic acid dihydrazide or N-succinimidyl-3(2-pyridyldithio) propionate.Infection and Immunity 60:584–589.

Favero, M.S., M.J. Alter, and L.A. Bland. 1992.Dialysis-associated infections and their con-trol. In: J.V. Bennett, and P.S. Brachman(eds.) Hospital Infections Third Edition.Boston, MA. Little, Brown and Company.

Gustilo, R.B., R.M. Mendoza, and D.N. Will-iams. 1984. Problems in the management oftype III (severe) open fractures: A new clas-sification of type III open fractures. Journalof Trauma 24:742–746.

Haas, D.W., and R.M. Des Prez. 1995. Mycobac-terium tuberculosis. In: G.L. Mandell, J.E.

Bennett, and R. Dolin (eds.) Mandel, Dou-glas and Bennett’s Principles and Practiceof Infectious Diseases Fourth edition. NewYork, NY. Churchill Livingstone, Inc.

Hakin, R.M. 1993. Clinical implications ofhemodialysis membrane biocompatibility.Kidney International 44:484–494.

Hewlette, D. Jr., D.L. Horn, and C. Alfalla. 1995.Drug-resistant tuberculosis: Inconsistentresults of pyrazinamide susceptibility test-ing. Journal of the American Medical Asso-ciation 273:916–917.

Heiter, B.J., and B.P. Bourbeau. 1993. Compari-son of the Gen-Probe Group A Streptococ-cus Direct Test with culture and rapidstreptococcal antigen detection assay fordiagnosis of streptococcal pharyngitis. Jour-nal of Clinical Microbiology 31(8):2070–2073.

Institute of Medicine. 1992. Emerging InfectiousDiseases. Washington, DC. National Acad-emy Press.

Jacob, E., G. Cierny, M.T. Fallon, et al. 1993.Evaluation of biodegradable cefazolinsodium microspheres for the prevention ofinfections in rabbits with experimental openttibial fractures stabilized with internal fixa-tion. Journal of Orthopedic Research11:404–411.

Jacob, E., J.M. Erpelding, and K.P. Murphy.1992. A retrospective analysis of open frac-tures sustained by U.S. military personnelduring Operation Just Cause. Military Medi-cine 157:552–556.

Jacob, E., J.A. Setterstrom, D.E. Bach, et al.1991. Evaluation of biodegradable ampicil-lin anhydrate microcapsules for local treat-ment of experimental staphylococcalosteromyelitis. Clinical Orthopedics267:237–244.

Jacobs, W.R., Jr., R.G. Barletta, R. Udani, et al.1993. Rapid assessment of drug susceptibili-ties of Mycobacterium tuberculosis bymeans of luciferase reporter phages. Science260:819–822)

Jorgensen, J.H. 1993. Recent developments inautomated and rapid susceptibility testingmethods. In: R.C. Spencer, E.P. Wright, and


S.W.B. Newson (eds.) Rapid Methods andAutomation in Microbiology and Immunol-ogy. Andover, England. Intercept Ltd.

Jorgensen, J.H. 1995. Current and Future Pheno-typic Methods for Detection of ResistantBacteria. Contract report to the Office ofTechnology Assessment. Photocopied type-script.

Jorgensen, J.H., and J.M. Matsen. 1987. Physi-cian acceptance and application of rapidmicrobiology instrument test results. In: J.H.Jorgensen (ed.) Automation in ClinicalMicrobiology. Boca Raton, FL. CRC Press,Inc.

Kamal, G., D. Divishek, J. Adams, et al. 1994.Reduced intravascular catheter infection byroutine use of antibiotic bonded catheters(abstract). Critical Care Medicine 22:A115.

Kamal, G.D., M.A. Pfaller, L.E. Rempe, et al.1991. Reduced intravascular catheter infec-tion by antibiotic bonding: A prospective,randomized, controlled trial. Journal of theAmerican Medical Association 265:2364–2368.

King, L.K., J.C. Kingswood, P. Sharpstone.1992. Comparison of the efficacy, cost, andcomplication rate of APD and CAPD aslong-term outpatient treatments for renalfailure. In: R. Khanna, K.D. Nolph, B.F.Prowant, et al. (eds.) Advances in PeritonealDialysis. Vol. 8. Toronto, Canada. pp. 123–126.

Kohner, P., C. Kolbert, D. Geha, et al. 1994.Evaluation of the BBL Crystal MRSA IDSystem for the detection of methicillin resis-tance in Staphylococci. Abstract presented atthe 93rd General Meeting, American Societyfor Microbiology, Las Vegas, Nevada.

Kuusela, P., P. Hilden, K. Savolainen, et al.1994. Rapid detection of methicillin-resis-tant Staphylococcus aureus strains not iden-tified by slide agglutination tests. Journal ofClinical Microbiology 32:143–147.

Levin, B.R. and J.J. Bull. 1995. Phage therapyrevisited: The population biology of a bacte-rial infection and its treatment with bacte-riophage and antibiotics. AmericanNaturalist. Submitted.

Levin, B.R., T. DeRouin, N. Moore, and J.J.Bull. 1995. Phage therapy revisited II: Anexperimental study of the population dynam-ics of a bacterial infection and its controlwith antibiotics and bacteriophage. Microbi-ology. Submitted.

Macher, A. and E. Goosby. 1995. PCR and themisdiagnosis of active tuberculosis. Letter.New England Journal of Medicine332(2):128–129.

Maderazo, E.G., S. Judson, and H. Pasternak.1988. Late infections of total joint prosthe-sis. Clinical Orthopedics 229:131–142.

Maki, D.G. 1994. Infections caused by intravas-cular devices used for infusion therapy:pathogenesis, prevention, and management.In: A.L. Bisno and F.A. Waldvogel (eds.)Infections Associated with Indwelling Medi-cal Devices. Washington, DC. ASM Press.

Marchetti, M., B. Arico, D. Burroni, et al. 1995.Development of a mouse model of Helico-bacter pylori infection that mimics humandisease. Science 267:1655–1658.

McKee, J.A., S.K. Sharma, and M.J. Miller.1991. Iron transport mediated drug deliverysystems: Synthesis and antibacterial activityof spermidine- and lysine-based sidero-phore-β lactam conjugates. BioconjugateChemistry 2:281–291.

Miller, M.J. 1989. Synthesis and therapeuticpotential of hydroxamic acid based sidero-phores and analogues. Chemical Reviews89:1563–1579.

Medical World News. 1995. CDC says many eld-erly skipping pneumovax. Medical WorldNews 34(6):11.

O’Brien, T. 1994. Remarks at U.S. Food andDrug Administration, Part 15 Hearing: Sur-veillance Systems for Antibacterial Resis-tance. Rockville, MD, November 9, 1994.

Rippin, K., C. Carlyn, J. Goldfard, et al. 1994.Rapid direct DNA probe testing of throatswabs for group A Streptocccus (GrAS)from pediatric patients. Abstract D1.Abstracts of the 34th Interscience Confer-ence on Antimicrobial Agents and Chemo-therapy. Washington, DC. American Societyfor Microbiology.


Salyers, A.A. and D.D. Whitt. 1994. BacterialPathogenesis: A Molecular Approach.Washington, DC. ASM Press.

Setterstrom, J.A., T.R. Tice, and W.E. Myers.1984. Development of encapsulated antibi-otics for topical administration to wound. In:J.M. Anderson, and S.W. Kim (eds.) 1984.Recent Advances in Drug Delivery Systems.New York, NY. Plenum Press, pp. 185–198.

Siber, G.R. 1994. Pneumococcal disease: pros-pects for a new generation of vaccines. Sci-ence 265:1385–1387.

Smits, H. 1995. Testimony presented before theSubcommittee on Health, Committee onWays and Means, United States House ofRepresentatives, on the Medicare End StageRenal Disease (Kidney Failure) Program,April 3.

Stamm, W.E. 1991. Catheter-associated urinarytract infections: epidemiology, pathogenesis,and prevention. American Journal of Medi-cine 91(3B):65S–71S.

Stratton, C.W., and R.C. Cooksey. 1990. Suscep-tibility tests: special tests. In: A. Balows,W.J. Hausler, K.L. Herrmann, et al. Manualof Clinical Microbiology, 5th Edition. Wash-ington, DC. American Society of Microbiol-ogy.

Tompkins, L.S., and S. Falkow. 1995. The newpath to preventing ulcers. Science267:1621–1622.

Travis, M.T., and M.Q. Cosio. 1993. A retro-spective review of orthopedic patientsreturning from Operation Desert Shield andDesert Storm to an Army medical center.Military Medicine 158:348–351.

Trenholme, G.M., R.L. Kaplan, P.H. Karahusis,et al. 1989. Clinical impact of rapid identifi-cation and susceptibility testing of bacterialblood culture isolates. Journal of ClinicalMicrobiology 27:1342–1345.

Univax Biologics, Inc. Rockville, MD. 1994.Personal communication to D. Frankle. Dataon file.

U.S. Congress, Office of Technology Assess-ment. 1993. The Continuing Challenge ofTuberculosis. U.S. Government PrintingOffice: Washington, DC.

U.S. Department of Health and Human Services,Food and Drug Administration. 1995. Com-ments on second draft of this OTA report.Memorandum from Associate Commis-sioner for Legislative Affairs, U.S. Food andDrug Administration to Office of Technol-ogy Assessment. June 16.

U.S. Department of Health, Education, and Wel-fare, Center for Disease Control. 1979. Botu-lism in the United States, 1899–1977.Handbook for Epidemiologists, Clinicians,and Laboratory Workers. Washington, DC.

Wanger A. and Mills K. 1995. E-Test for Sus-ceptibility Testing of Mycobacterium tuber-culosis to first line antituberculosis agents.Poster U-67. American Society for Microbi-ology 95th General Meeting. WashingtonDC.

Weinstein, M.P., J.R. Murphy, L.B. Reller, et al.1983. The clinical significance of positiveblood cultures: A comprehensive analysis of500 episodes of bacteremia and fungemia inadults. II. Clinical observations, with specialreferences to factors influencing prognosis.Reviews of Infectious Disease 5:54–70.

Wenger, J.D. 1994. Mini review: Impact of Hae-mophilus influenzae type B vaccines on theepidemiology of bacterial meningitis. Infec-tious Agents and Disease 2:324–332.

Wheeler, J.G., L.C. Rodrigues, and V. Diwan.1994. Letter to the Editor. Journal of theAmerican Medical Association 272:765–766.

Wischnack, L.L., R.M. Jacobson, G.A. Poland,et al. 1995. The surprisingly high acceptabil-ity of low-efficacy vaccines for otitis media:A survey of parents using hypothetical sce-narios. Pediatrics 95:350–354.

Yu et al. 1992. Self-setting hydroxyapatitecement: a novel skeletal drug-delivery sys-tem for antibiotics. Journal PharmaceuticalSciences 81:529–531.

Documents

Impacts of Antibiotic-Resistant Bacteria (Part 8 of 12) - Princeton