Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
| 69
4
AntibioticUse in
Hospitals
t any given time, 25 to 35 percent ofhospitalized patients are receiving sys-temic antibiotics (Eickhoff, 1991) totreat active infections or to prevent
potential infections. The heavy use of antibioticsin the hospital exerts enormous selective pres-sure for the emergence and spread of antibiotic-resistant bacteria. Consequently, many of the twomillion bacterial infections acquired in the hospi-tal are antibiotic-resistant, and a few are resistantto every antibiotic currently approved for use.Some hospitals have reduced infections fromantibiotic-resistant bacteria through a combina-tion of infection control procedures that preventthe spread of the resistant organisms and throughmonitoring and control of antibiotic use.
This chapter 1) describes antibiotic use in hos-pitals and its contribution to the rise of antibiotic-resistant nosocomial infections, 2) discusses cur-rent efforts to control antibiotic-resistant infec-tions, 3) explores medical and financial factorsthat make such efforts difficult to implement inhospitals, and 4) discusses some possible solu-tions.
INFECTIONS ACQUIRED IN THE HOSPITALThe Centers for Disease Control and Prevention(CDC) estimates that 1 out of 20 patients(2 million per year) acquire infections in the hos-pital (Haley et al., 1985).1 Nosocomial infectionscost $4.5 billion a year (1992 dollars) in terms ofextra treatment and days of hospitalization,directly cause 19,000 deaths, and contribute to58,000 deaths annually (table 4-1). The 19,000deaths per year directly caused by nosocomialinfections makes them the 11th leading cause ofdeath in the U.S. population (Martone et al.,1992).
Recent data from the National NosocomialInfections Surveillance (NNIS) system show thatnosocomial infections are increasing (figure 4-1).The number of blood stream infections increased279 percent in small non-teaching hospitals, 196percent in large non-teaching hospitals, by 124percent in small teaching hospitals, and by 70percent in large teaching hospitals during the1980s. It might be discouraging that the rates ofblood stream infections have been increasing
1 Based on data from CDC’s 1976 Study on the Efficacy of Nosocomial Infection Control (SENIC). This number is still widely quoted inrecent reports (see, for example, IOM, 1992).
A
70 | Impacts of Antibiotic-Resistant Bacteria
TABL
E 4-
1: E
stim
ated
Ext
ra D
ays,
Ext
ra C
harg
es, a
nd D
eath
s At
tribu
tabl
e to
Nos
ocom
ial I
nfec
tions
Ann
ually
in U
.S. H
ospi
tals
Extra
day
sEx
tra c
harg
esDe
aths
dir
ectly
cau
sed
by in
fect
ions
Deat
hs to
whi
ch
infe
ctio
ns c
ontr
ibut
ed
Avg.
per
infe
ctio
naEs
t. U
.S. t
otal
b
Avg.
ext
ra c
harg
es p
er
infe
ctio
n in
199
2 do
llars
cEs
t. U
.S. t
otal
in
1992
dol
lars
bEs
t. U
.S. t
otal
bEs
t. U
.S. t
otal
b
Sur
gic
al w
ound
infe
ctio
n7.
33,
726,
000
$3,1
52$1
,609
,000
,000
3,25
19,
726
Pne
umon
ia5.
91,
339,
000
$5,6
83$1
,290
,000
,000
7,08
722
,983
Bac
tere
mia
7.4
762,
000
$3,5
17$3
62,0
00,0
004,
496
8,84
4
Urin
ary
trac
t inf
ectio
n1.
090
3,00
0$6
80$6
15,0
00,0
0094
76,
503
Oth
er s
ite4.
81,
946,
000
$1,6
17$6
56,0
00,0
003,
246
10,0
36
All
site
s4.
0d8,
676,
000
$2,1
00$4
,532
,000
,000
19,0
2758
,092
a Ad
apte
d fr
om R
.W. H
aley
et
al.,
Ext
ra d
ays
and
pro
long
atio
n of
sta
y at
trib
utab
le t
o no
soco
mia
l inf
ectio
ns:
A p
rosp
ectiv
e in
terh
osp
ital c
omp
aris
on. A
mer
ican
Jou
rnal
of M
edic
ine
70:5
1, 1
981,
by
poo
ling
dat
a fr
om th
e th
ree
SE
NIC
pilo
t stu
dy
hosp
itals
.b E
stim
ated
by
mul
tiply
ing
the
tota
l num
ber
of n
osoc
omia
l inf
ectio
ns e
stim
ated
in th
e S
EN
IC P
roje
ct (R
.W. H
aley
et a
l., T
he n
ation
-wid
e no
soco
mia
l inf
ectio
n ra
te: A
new
nee
d fo
r vita
l sta
tistic
s. A
mer
-ic
an J
ourn
al o
f Ep
idem
iolo
gy
121:
159,
198
5) b
y th
e av
erag
e ex
tra
day
s, a
vera
ge
extr
a ch
arg
es, o
r p
erce
ntag
e of
infe
ctio
ns c
ausi
ng o
r co
ntrib
utin
g to
dea
th, r
esp
ectiv
ely.
c 19
92 d
olla
rs e
stim
ated
from
R.W
. Hal
ey e
t al.,
Am
er. J
. Med
. 70:
51,
1981
, by
poo
ling
dat
a fr
om th
e th
ree
hosp
itals
and
ad
just
ing
for
the
annu
al r
ate
of in
flatio
n of
hos
pita
l exp
ense
s fr
om 1
976
to19
92 (
rang
e 5.
0 to
12.
3p
erce
nt)
obta
ined
from
the
Am
eric
an H
osp
ital A
ssoc
iatio
n’s
Nat
iona
l Pan
el S
urve
y. E
stim
ates
for
infla
tion
rate
s in
199
1 an
d 1
992
wer
e p
roje
cted
from
198
7 to
199
0 tr
end
.d N
atio
nwid
e es
timat
e ob
tain
ed b
y su
mm
ing
the
pro
duc
ts o
f the
site
-sp
ecifi
c es
timat
e of
the
aver
age
extr
a d
ays,
ave
rag
e ex
tra
char
ges
, or
the
per
cent
age
of in
fect
ions
cau
sing
or
cont
ribut
ing
tod
eath
, res
pec
tivel
y, fr
om th
e S
EN
IC p
ilot s
tud
ies
(R.W
. Hal
ey e
t al.,
Am
er. J
. Med
. 70:
51, 1
981)
, and
the
natio
nwid
e es
timat
e of
the
pro
por
tion
of n
osoc
omia
l inf
ectio
ns a
ffect
ing
the
site
from
the
mai
nS
EN
IC a
naly
sis
(R.W
. Hal
ey e
t al.,
Am
er. J
. Ep
idem
iol.
121:
159,
198
5).
SO
UR
CE
: W
.J. M
arto
ne,
W.R
. Ja
rvis
, D
.H.
Cul
ver,
et
al. 1
992.
Inc
iden
ce a
nd n
atur
e of
end
emic
and
ep
idem
ic n
osoc
omia
l inf
ectio
ns. I
n: J
.V.
Ben
nett
and
P.S
. B
rack
man
(ed
s.)
Hos
pita
l Inf
ectio
nsTh
ird e
diti
on. B
osto
n, M
A. L
ittle
, Bro
wn
and
Com
pan
y, p
p.5
93.
Chapter 4 Antibiotic Use in Hospitals 71
despite guidelines developed by CDC and theadoption of “universal precautions” to controlinfections. However, these increasing rates arepartially due to recent advances in medicine.Increasing rates of surgery and catheterizationprovide opportunities for bacteria to penetrateinto the body where they can cause infections. Inaddition, tissue and organ transplants, which arebecoming more frequent and successful, requireimmunosuppression so that the foreign tissue isnot rejected by the transplant recipient. Conse-quently, immunosuppressed patients are depen-dent on antibiotics to control bacterial infections.
Treatment with an antibiotic may suppressenough normal microbial flora (commensals) toleave a patient susceptible to infection by otherorganisms—especially antibiotic-resistant bacte-ria-unaffected by the antibiotic. Kollef (1 994)cites studies that show intensive care unitpatients who had received antibiotics were morelikely to develop ventilator-associated pneumo-
,80 81 82 83 84 85 86 87 88 89
SOURCE: S.N. Banerjee, T.G. Emori, D.H. Culver, et al, 1991. Ameri-
can Journal of Medicine 91 (Suppl. 3B):86S-89S
nia caused by virulent species such asPseudomonas aeruginosa or Acinetobacter, andthat patients with those infections were almosttwice as likely to die from them as patientsinfected with less virulent species.
THE RISE OF ANTIBIOTIC-RESISTANTINFECTIONS IN HOSPITALSCDC operates the NNIS system that gathers vol-untary information from approximately 200 hos-pitals, and through NNIS, CDC has documentedincreases in the number of nosocomial infectionscaused by antibiotic-resistant bacteria. Twoimportant cases are the increasing numbers ofinfections caused by methicillin-resistant Stap/ty-10CoCCUS aureus (MRSA) and vancomycin-resis-tant Enterococci (VRE). Resistant strains ofKlebsiella, Pseudomonas, Escherichia coli, andcoagulase-negative Staphylococci also causeserious problems in hospitals.
❚ Methicillin-Resistant Staphylococcusaureus (MRSA)Nosocomial Staphylococcus aureus infectionshave been a recurrent problem in hospitals formany years. This is partially due to the high rateof colonization in the population: about50 percent of the population are intermittent car-riers of Staph. aureus, and about 30 percent ofthe population are prolonged carriers of the bac-teria in their nostrils or on their skin (Waldvogel,1995). When these colonizing organisms enterinternal organs of the body through invasive sur-gery, catheterizations, or other hospital proce-dures, they can cause infection. Strains resistantto penicillin were identified soon after its intro-duction (Spink and Ferris, 1945). Currently,more than 90 percent of all Staph. aureus areresistant to penicillin (Mandell and Sande, 1990).These strains of staphylococci were most likelyresistant through the production of beta -lactamasesthat destroy penicillin and penicillin-like antibi-otics.
The synthetic penicillin, methicillin, intro-duced in 1960, is not affected by many beta-lacta-mases. However, strains of staphylococci that
72 Impacts of Antibiotic-Resistant Bacteria
contain a chromosomal gene called mec A whichencodes a modified penicillin-binding proteinhave been identified. These strains, commonlyreferred to as MRSA, are resistant to all beta-lactamantibiotics, and frequently also contain plasmid-encoded genes for resistance to other antibiotics(see chapter 2). MRSA were initially susceptibleto the fluoroquinolones introduced in the 1980s,such as ciprofloxacin, but they quickly becameresistant to these antibiotics. NNIS data docu-ment the increase in MRSA (figure 4-2). By1992, more than 40 percent of Staph. aureusinfections in large hospitals were methicillin-resistant. Some strains of MRSA are resistant toall antibiotics currently approved by the U.S.Food and Drug Administration (FDA), with theexception of vancomycin; others are susceptibleto other antibiotics as well as vancomycin (seechapter 5).
❚ Vancomycin-Resistant EnterococcusSome strains of Enterococcus are resistant to all
available antibiotics approved by FDA, and theyare, therefore, untreatable with antibiotics. NNISdata showing the increase in VRE are presentedin figure 4-3. As of 1994, almost 13 percent ofenterococci acquired in intensive care units
(ICUs) were resistant to vancomycin, and about8 percent of enterococci acquired outside ofICUs were resistant. There is currently no FDA-approved antibiotic to treat many of these infec-tions. 2
❚ Vancomycin-Resistant MRSA?A huge fear among clinicians and epidemiolo-gists is the possibility of the emergence of vanco-mycin-resistant strains of MRSA that are bothhighly virulent and untreatable. As this reportgoes to press, no confirmed vancomycin-resis-tant strain of MRSA has been reported to publichealth officials at CDC or elsewhere. However,Noble, Virani, and Cree (1992) demonstrated the
<200 beds200-499 beds>500 bedsYear MRSA exceeded 5%
1975 77 79 83 85 87 89 91
SOURCE: National Nosocomial Infections Surveillance System,
Centers for Disease Control, Atlanta, GA
1989 1990 1991
SOURCE: National Nosocomial Infections Surveillance System
Centers for Disease Control, Atlanta, GA.
transfer of a vancomycin resistance gene from anEnterococcus to Staph. aureus in the laboratory,indicating that the clinical emergence of vanco-
2 Chapter 5 describes two new drugs, quinupristin/dalfopristin and teicoplanin, currently in clinical trials that may have activity againstsome strains of VRE, These drugs are available from the manufacturers on a compassionate-use basis to patients with VRE infections (TheMedical Letter on Drugs and Therapeutics, 1994, at p. 31).
Chapter 4 Antibiotic Use in Hospitals | 73
mycin-resistant MRSA is possible. The onlytreatment available for some strains of MRSA isvancomycin, and the emergence of vancomycin-resistant MRSA may be inevitable. It will presenta crisis in treatment.
THE USES OF ANTIBIOTICS IN HOSPITALS
❚ Prophylactic Use of AntibioticsIn large surgical hospitals, half of all antibioticsare used to prevent possible infections (prophy-laxis) (Kernodle and Kaiser, 1990). More than 30years ago, Burke (1961) showed that prophylac-tic use of antibiotics before surgery reduces post-operative infection rates. Classen et al. (1992)investigated the timing of administration of anti-biotics for prophylaxis and confirmed that antibi-otics can prevent infections when administeredtwo hours prior to surgery. They also suggestedthat antibiotics given at times other than in the 2hours before surgery (one-third of all prophylac-tic antibiotics were given earlier than 2 hoursbefore surgery or after surgery in this study of2,847 patients) are not as effective in preventinginfections (see table 4-2). Approximately
12 percent of the patients received antibioticsmore than 2 hours before surgery; and more than70 percent of the antibiotics given had half-livesranging from 0.7–1.9 hours (Wenzel, 1992), sug-gesting that these antibiotics washed out of thepatients’ system before surgery began. In thesecases it is clear that the use of antibiotics wasinappropriate and that appropriate use of antibi-otics would reduce the rate of infections and theirassociated costs because of decreases in the num-ber of days that a patient is hospitalized. More-over, appropriate use would reduce antibiotic useand help control antibiotic resistance.
Studies raise questions about the effects ofprophylactic antibiotic use other than to preventsurgical wound infections. Kollef (1994a) foundthat prophylactic use of antibiotics for selectivedigestive decontamination designed to reducenosocomial pneumonia reduced the incidence ofpneumonia, but it had no effect on mortality.Apparently this phenomenon occurred becauseantibiotic-resistant bacteria that colonized somepatients following the prophylactic treatmentwere harder to treat.
Classen et al. (1992) reported that more than50 percent of the nosocomial infections they
TABLE 4-2: Temporal Relation between the Administration of Prophylactic Antibiotics and Rates of Surgical-Wound Infection
Time of administration* No. of patients No. (%) of infections Relative risk (95% CI) Odds ratio** (95% CI)
Early 369 14 (3.8)‡ 6.7 (2.9–14.7) 4.3a (1.8–10.4)
Preoperative 1708 10 (0.59) 1.0
Perioperative 282 4 (1.4)b 2.4 (0.9–7.9) 2.1c (0.6–7.4)
Postoperative 488 16 (3.3)‡ 5.8‡ (2.6–12.3) 5.8d (2.4–13.8)
All 2847 44 (1.5) ——————— ——————
* For the administration of antibiotics, “early” denotes 2 to 24 hours before the incision, “preoperative” 0 to 2 hours before the incision, “perioper-ative” within 3 hours after the incision, and “postoperative” more than 3 hours after the incision.** As determined by logistic-regression analysis.‡ P<0.0001 as compared with preoperative group (all P values were determined by logistic-regression analysis).a P = 0.001.b P = 0.12 as compared with preoperative group.c P = 0.23.d P = 0.0001.
SOURCE: C. Classen, R.S. Evans, S.L. Pestotnik, et al. 1992. The timing of prophylactic administration of antibiotics and the risk of surgical-wound infection. New England Journal of Medicine 326(5):283.
74 | Impacts of Antibiotic-Resistant Bacteria
studied were caused by organisms resistant to theantibiotic used. In these cases the infections mayhave been caused because the resistant organ-isms were able to multiply when the susceptiblenormal bacterial flora of the patients was inhib-ited by the prophylactic antibiotics. Siegel et al.(1980) reported an especially tragic example ofprophylactic use gone awry based on examina-tion of the results of giving a single dose of peni-cillin to ward off streptococcal infections insome 9,000 newborns. Although penicillin-sensi-tive infections were reduced by the prophylactictreatment, infections with penicillin-resistantbacteria were more frequent in the babies whoreceived the antibiotic, and mortality was higherfrom the resistant infections (15 of 35) than fromthe sensitive infections (3 of 27). Overall, thedeath rate from streptococcus infections was3 times higher in the babies that received penicil-lin (1.2/1,000 vs. 0.43/1,000 live births).
❚ Antibiotic Use to Treat Active InfectionsThe remainder of antibiotic use in hospitals is fortreatment of active infections. It takes at leasttwo days to identify the bacteria causing aninfection and to determine its antibiotic suscepti-bility (see chapter 6). Therefore, the physicianoften has to make an empirical judgment aboutthe identity of the bacteria and prescribe an anti-biotic before the laboratory test results are avail-able. If a patient is very sick, the physician willoften use multiple antibiotics. If the patient isimproving when the laboratory tests arrive, thephysician might ignore the results of the tests andcontinue the patient on the empiric antibiotics. Itis difficult to determine inappropriate antibioticuse and how to improve use in such cases.
The appearance of unexpected resistant organ-isms in one patient may influence a physician toroutinely prescribe newer or broader spectrumantibiotics. A letter to the editor of the NewEngland Journal of Medicine (Lonks et al.,1995) illustrates a case where a patient sufferedbecause he was infected with an unlikely resis-tant strain. Physicians knew that no highly resis-tant strains of pneumococci had been reported inProvidence, Rhode Island; only 2.3 percent of
isolates obtained in hospitals in 1990 and 1991showed intermediate-level resistance to penicil-lin, and none was highly resistant. An otherwisehealthy 33-year-old man, who lived a little morethan 30 miles from the city, was treated in thehospital for a Streptococcus pneumoniae infec-tion. Assuming that the strain was not ceftriax-one-resistant, doctors treated the patient withdexamethasone and ceftriaxone for the first fourdays. After initial improvement, encephalitisdeveloped, and doctors switched drugs to vanco-mycin and rifampin based on antibiotic-suscepti-bility test results that showed the infecting strainswere resistant to penicillin and ceftriaxone. Thepatient’s condition eventually improved and hewas sent home. Based on this experience, theauthors concluded that “all patients with the pre-sumptive diagnosis of pneumococcal meningitisshould receive high-dose ceftriaxone (or cefo-taxime) plus vancomycin, with or withoutrifampin, until the isolate is proved to be suscep-tible to penicillin or ceftriaxone” [emphasisadded]. It may be true that following this advicewill prevent a few adverse outcomes such asthose described in the letter to the journal. How-ever, if similar reasoning is applied in manycases, the widespread use of antibiotics such asvancomycin will increase the risk for the emer-gence of antibiotic-resistant organisms.
In a study of the reasoning strategies used byphysicians in empiric antibiotic selection, Yu etal. (1991) found that unexpected organismsappeared in 3.8 percent of all blood cultures. Inthese cases, antibiotics had been prescribedwhich were not the antibiotics of choice based onlogical reasoning, but which did cover the unex-pected organisms. The authors comment that“[t]hese memorable situations may have a dis-proportionate influence in these physicians’future selection of antibiotic therapy.” They fur-ther conclude that “our disturbing and unex-pected finding is that reflex prescription ofbroad-spectrum antibiotic therapy that is so oftendecried by academicians may have a rationalbasis” and that “educational efforts that empha-size narrow, rather than broad-spectrum prescrib-ing may be inadequate to change physicianprescribing habits.”
Chapter 4 Antibiotic Use in Hospitals | 75
LEGAL ASPECTS OF ANTIBIOTIC USEMalpractice concerns might provide an addi-tional incentive to prescribe antibiotics. Accord-ing to data published by St. Paul Fire and MarineInsurance Company, a large nationwide malprac-tice insurer, a significant number of claims arerelated to infection-related illnesses and antibi-otic use (St. Paul Fire and Marine Insurance Co.,1995). It is reasonable to speculate that fear of
malpractice litigation may contribute to prescrip-tion of overly broad spectrum antibiotics or ofantibiotics in cases where the chance of a bacte-rial infection is small. Box 4-1 contains excerptsfrom a commentary in the medical journal Lan-cet discussing the medical and legal controversyover the use of prophylactic antibiotics to pre-vent neonatal bacterial sepsis caused by Group Bstreptococcus.
BOX 4-1: Group B Streptococcus: The Controversy
Group B streptococcus (GBS) is the leading cause of neonatal bacterial sepsis in the United States,infecting about 12,000 newborns annually. Some newborns infected with GBS may die or have perma-nent neurological damage from meningitis. In 1992, both the American Academy of Pediatrics (AAP) andthe American College of Obstetricians and Gynecologists (ACOG) issued protocols regarding thescreening of pregnant women to detect and treat carriers of GBS in an effort to prevent neonatal GBSsepsis.
AAP called for universal prenatal GBS screening for all pregnant women at 26–28 weeks’ gestation.Because certain population groups are more likely to carry GBS, ACOG advocated for optional screeningtargeted to certain populations where the incidence of neonatal GBS infection is inordinately high, suchas populations where sexually transmitted diseases are common.
Inasmuch as GBS is part of the normal gut flora of some women and may or may not become a patho-gen during pregnancy, both AAP and ACOG recommended intrapartum (during delivery) antibiotic treat-ment only to women with positive cultures who have additional high-risk factors such as preterm labor orpremature rupture of the membranes before 37 weeks’ gestation, fever in labor, multiple births, rupture ofmembranes for more than 18 hours at any gestational age, or a previous affected child.
The AAP and ACOG protocols leave a number of issues unresolved that expose obstetricians, familypractitioners, and nurse midwives to considerable medicolegal liability. Screening for GBS during preg-nancy does not provide certainty as to whether or not intrapartum antibiotic treatment is warranted. Astudy found that in women who were culture-positive at 28 weeks’ gestation, 30 to 50 percent were cul-ture-negative at the time of delivery; in women who were culture-negative at 28 weeks, 8 to 15 percentwere culture-positive at the time of delivery. Consequently, some women will be treated unnecessarilyand some who need treatment will be ignored.
Moreover, if only certain groups are targeted for screening in keeping with ACOG’s protocol, canexcluded groups hold health care professionals responsible if their newborn babies developed undetec-ted GBS sepsis? Further, would the withholding of treatment in a pregnant woman with a positive culturewho has no additional risk factors absolve a health care professional from medicolegal liability if that babywere affected?
The best approach to the management of GBS sepsis would be a rapid screening test during labor todetermine whether antibiotic therapy is warranted, but the poor sensitivity of such tests currently rendersthem clinically useless. Until these tests are improved, health care professionals will most likely err on theside of caution and prescribe antibiotics even in extremely low-risk cases.
SOURCE: C.V. Towers, 1995, Lancet 346:197–198.
76 | Impacts of Antibiotic-Resistant Bacteria
The following review of some malpracticesuits exemplifies the dramatic consequences thatcan occur due to undertreating with antibiotics.In Hellwig v. Potluri (Case No. WL 285712,Ohio Court of Appeals 7th Circuit, 1991), thedefendant emergency room physician was heldliable for failing to prescribe antibiotics for theplaintiff who had stepped on a rusty nail at hishome. The plaintiff developed osteomyelitiswhich forced him to “wear an appliance in hisshoe and have an altered gait for the rest of hislife.” In Toler v. United States of America aplaintiff claimed that failure of a VeteransAdministration (VA) hospital to administer anadequate course of antibiotics resulted in sepsisand death. In Griffith v. West Suburban Hospital(Case No. 86L-23904, Cook County, Illinois Cir-cuit Court, 1993), a jury returned a $3.5-millionverdict for failure to diagnose and timely treat aGroup B Strep infection. In this case, a patientshowed signs of respiratory distress shortly afterbirth, and although he was moved to an intensivecare crib, antibiotics were not administered.Seven hours later, after being transferred toanother hospital which then administered antibi-otics, the patient died.
The medical and financial consequences offailing to prescribe prophylactic antibiotics forendocarditis can be considerable. In 1993, a den-tist was held liable in Orbay v. Castellanos (CaseNo. 91-36124, Dade County Circuit Court,Miami, Florida, 1993) for failing to prescribeprophylactic antibiotics prior to tooth extraction.Soon after the tooth extraction, the plaintiff wasdiagnosed with bacterial endocarditis and under-went open heart valve replacement surgery. Thedefendant was held liable for failure to prescribeprophylactic antibiotics and failure to obtain afull medical history or medical clearance for apatient at risk of developing bacterial endocardi-tis. The jury awarded the plaintiff $1.24 million,which was reduced to $964,000 to reflect thedecision that the plaintiff was 20 percent com-paratively negligent for failure to take appropri-ate care of himself. However, a standard medicaltextbook comments:
The issue of professional liability in the pro-phylaxis of endocarditis often has led to allega-tions of negligence and malpractice suits. . . . [Itis hard] to prove that the failure of a physicianor dentist to administer antibiotics was thedirect cause of a patient acquiring endocarditis.If a strict demonstration of proximate causewere always required, it is doubtful that anyclaim based on the failure to administer prophy-laxis could succeed, but juries are sometimescapricious in deciding liability in malpracticecases. . . (Mandell et al., 1990).
The “capricious” nature of the juries mightbias physicians in favor of prescribing antibiot-ics, even when the risk of endocarditis (or otherdisease) is very minimal.
CONTROLLING THE EMERGENCE AND SPREAD OF ANTIBIOTIC RESISTANCE IN HOSPITALSPart of the difficulty in controlling antibioticresistance in hospitals is incomplete understand-ing of all the factors that contribute to the emer-gence and spread of antibiotic resistance ingeneral. Most hospital personnel would agreethat infection control is critical, but there aremany disagreements about the benefits vs. costof various infection control procedures. Few, ifany, scientists disagree that the use of antibioticsis related to the emergence and spread of antibi-otic resistance. Nevertheless, there are manycontroversies about how to implement programsto control the use of antibiotics.
❚ Infection Control in HospitalsIn 1847, Ignac Semmelweis noticed that the rateof childbed fever in new mothers was muchhigher when the babies were delivered by obste-tricians and medical students than by midwivesand midwifery students. Semmelweis surmisedthat the high rate was due to the transmission ofinfectious particles from cadavers by the obste-tricians and medical students and instituted themeasure of handwashing in a chlorine solution.This measure greatly decreased the incidence ofchildbed fever (reviewed by Sanford, 1992).
Chapter 4 Antibiotic Use in Hospitals | 77
In hospitals today, infection control proce-dures are considered absolutely essential. In1976, CDC conducted a comprehensive Study onthe Efficacy of Nosocomial Infection Control(SENIC) that measured the extent and effective-ness of infection control procedures in U.S. hos-pitals. The SENIC study included a survey of allhospitals in the United States and detailed inter-views with representative hospitals. Twentyyears later, the study remains the most compre-hensive survey of the effectiveness of infectioncontrol procedures.3 The study concluded thathospitals with intensive infection surveillanceand control programs were able to reduce the rateof nosocomial infections by 32 percent (Haley etal., 1985). Yet the study found that only about0.2 percent of U.S. hospitals had programs thateffectively controlled all four of the major typesof infections: surgical wound infection, urinarytract infection, primary bloodstream infection,and lower respiratory tract infection.
❚ Infection Control ActivitiesThe SENIC study concluded that a successfulinfection control program required leadership bya trained infection control physician, an infectioncontrol nurse for every 250 beds, organizedinfection surveillance efforts, and a system forreporting infection rates to practicing surgeons.
Handwashing and Other PrecautionsSimple infection control procedures, such ashandwashing and wearing gloves, reduce thespread of infections in hospitals, lowering theneed for antibiotics and thereby reducing selec-tive pressure for the spread of antibiotic-resistantbacteria. Health care workers have a large incen-tive to follow procedures such as universal pre-cautions4 because they were designed to protectthem from infection from organisms such as the
3 SENIC data have the serious shortcoming that they were collected before implementation of current infection control procedures suchas universal precautions, which were instituted beginning in 1985 largely because of the fear of transmission of the human immunodeficiencyvirus (HIV).
4 Universal precautions include requirements that gloves be worn when handling bodily fluids, that needles and other sharp objects bedisposed of in special containers to help prevent needle-stick accidents, and that health care workers with open or infected wounds haverestricted contact with patients or patient care equipment (Garner, 1993).
human immunodeficiency virus (HIV). How-ever, in the hospital setting health care workerswho respond to a life-threatening emergencyoften do not have time to put on gloves and fol-low proper infection control procedures. Willy etal. (1990) found that health care workers’ per-ception of their own risk and potential spread ofinfections to patients is surprisingly low. In ananonymous nationwide survey of health careworkers who might have frequent exposure toblood and other bodily fluids, only 55 percent ofthose responding reported routinely practicinguniversal precautions.
Human nature seems to prevent the full imple-mentation of one of the simplest, yet most effec-tive infection control method: handwashing.Handwashing is a proven method for reducingnosocomial infections, but the practice is notstrictly followed. Handwashing compliance ratesof less than 50 percent were observed in twostudies of intensive care units (Simmons et al.,1990; Doebbeling et al., 1992). Goldmann andLarson (1992) make the following commentsabout the lack of compliance with handwashing:
Experts in infection control coax, cajole,threaten, and plead, but still their colleaguesneglect to wash their hands.... Education andpersuasion do not generally lead to sustainedimprovement in handwashing. Physicians havebeen particularly refractory. Innovativeapproaches are needed desperately, but fewhave emerged.... There is so little confidence inhand-washing habits that hospital isolation poli-cies now assume noncompliance.... [Originalreferences not included].
Simmons et al. (1990) revealed one clue tohandwashing noncompliance: nurses who werequestioned about their handwashing practicesbelieved they were washing their hands nearly
78 | Impacts of Antibiotic-Resistant Bacteria
90 percent of the time, when actual rates werebetween 22 and 29 percent.
Research into the seemingly simple questionof which soap to use for washing hands may beuseful in helping to prevent infections. Severalstudies have shown that a 7- to 10-second hand-wash with a non-antibacterial soap increased thetransmission of bacteria due to the shedding ofbacteria-laden skin cells, but that handwashingwith antiseptic soaps reduces the rates of nosoco-mial infections (Martin, 1994). Rotter (1988)compared the efficacy of different antiseptics forwashing hands and found that antiseptics con-taining isopropanol alcohol were significantlybetter at reducing skin bacteria than liquid soap.
Applying Infection Control Procedures to Control Antibiotic-Resistant Bacteria: Some Case StudiesBox 4-2 describes the successful countrywidecontrol of MRSA in Denmark. The followingcase studies describe attempts to apply infectioncontrol procedures to control MRSA in nursinghomes and hospitals in the United States.
Case 1: Successful control in a (mostly chronic care) VA medical center (Murray-Leisure et al., 1990)The Lebanon, Pennsylvania, Medical Center isan 884-bed facility which successfully controlledan epidemic of MRSA patients during 1988–1989 within six months of instituting aggressiveinterventions. These interventions included con-fining known active MRSA carriers and MRSA-infected patients to one nursing unit, screeningpatients transferred into the facility for MRSA,using gown and glove isolation and treating bothcolonized and actively infected patients with top-ical and enteral antibiotics.
Case 2: Unsuccessful control in a VA medical center (Strausbaugh et al., 1992)The Portland, Oregon, VA Medical Center Nurs-ing Home Care Unit (NHCU) is a 120-bed facil-ity that attempted to control MRSA primarilythrough administration of the antibioticsrifampin, trimethoprim-sulfamethoxazole, and
clindamycin, used either alone or in differentcombinations, to asymptomatic carriers ofMRSA. Other measures included restrictingMRSA-infected or colonized patients to a smallcluster of rooms, glove use to prevent the spreadof any body fluids, and frequent environmentalsurface decontamination. The majority of MRSApatients in this facility remained either colonizedor became recolonized during a 30-day follow-up period after treatment. Furthermore, a mostdisturbing byproduct of the Portland VA studywas the emergence of resistance to rifampin aftertherapy.
Case 3: Coordination of infection control practices between a hospital and nursing homes to manage MRSA (Jewell, 1994)The Christ Hospital and Medical Center, OakLawn, Illinois, is an 823-bed teaching hospitalthat serves many patients who live in regionalnursing homes. Before 1991, nursing homesoften required three successive test results show-ing the patient was not carrying MRSA beforethey would accept a patient from the hospital.This led to extended stays in the hospital forpatients who were colonized with MRSA, butotherwise did not need to be in the hospital. Aquality improvement team including clinicians,hospital administrators, and nursing home repre-sentatives adopted guidelines that allowed colo-nized patients to be returned to the nursinghomes. When these new guidelines wereadopted, the hospital did not see any change inthe number of patients infected or colonized withMRSA. It did see an average decrease of over10 days in the length of stay in the hospital, areduction in the readmission rate of patients col-onized with MRSA from 8.7 to 2.7 percent in1992, and total cost savings of over $1.9 million.
These case studies illustrate the complexitiesin determining which infection control practicesare the most likely to help control antibiotic-resistant bacteria such as MRSA. In the firstcase, a combination of isolation of patients colo-nized or infected with MRSA and antibiotic ther-apy seemed to control MRSA, but in the secondcase similar procedures failed to produce posi-
Chapter 4 Antibiotic Use in Hospitals 79
tive results. Further, the second case illustrates a ing homes saved money and significantlydanger in antibiotic-therapy for decolonization: reduced the length of hospital stays. Hospitalsthe emergence of new antibiotic-resistant strains. and nursing homes need to examine cases suchAnd the third case illustrates that isolation of as these along with specific conditions in theirpatients colonized with antibiotic-resistant bacte- own facilities to determine the best practices forria can be taken too far: in this case allowing reducing the spread of antibiotic-resistantpatients colonized with MRSA to return to nurs- bacteria.
In Denmark the frequency of methicillin-resistant Staph aureus (MRSA) rose to 15 percent between
1967 and 1971, but decreased to 0.2 percent by 1984, and has remained at that low level (see figure).
Hans Jern Kolmos of the Hvidovre Hospi-
tal, University of Copenhagen, discussed the
dramatic decline in MRSA at a recent meeting
of the Association of Practitioners of Infection
Control and Epidemiology. Kolmos attributes
the decline to strict control of antibiotic use in
hospitals. He acknowledges one of the funda-
mental dilemmas in antibiotic prescribing: “In
a situation of doubt, where the clinician
stands face to face with an ill patient, fear of
overlooking an infection-or pressure from
the patient—will often outweigh the fear of
side effects in the doctor’s mind, and the
result will be prescription for safety’s sake, ”
Kolmos stresses the value of including clinical
microbiologists in the decision-making pro-
cess: “In Denmark the clinical microbiologist
is a medical doctor, who has a clinical educa-
tion in addition to his laboratory education.
Frequency of Methicillin-ResistantStaphylococcus aureus (MRSA) in Denmark
I
I
1960 1965 1970 1975 1980 1985SOURCE: V.T. Rosdahl, and AM. Knudson. 1991. The decline ofmethicillin resistance among Danish Staphylococcus aureus strains.
Infection Control and Hospital Epidemiology 12(2):83-88.
This means that he takes part not only in laboratory work, but also in the treatment of patients, either bed-
side or at conferences with the clinical staff. Formally, he is only an advisor; it is the clinician who has the
power to decide. However, the influence of the clinical microbiologist is great, partly because he is well-
known from his frequent visits to the clinical units and partly because he has the same educational back-
ground as the clinicians. ”
The low rates of MRSA in Denmark may also be due to strict compliance with infection control proce-
dures. Westh et al. (1992) note that “Isolation of a methicillin-resistant strain triggers an immediate visit to
the patient involved and the staff caring for that patient by a microbiologist and an infection control nurse.
Patients are isolated, and hygienic precautions are taken in an effort to prevent acquisition and carriage
of the resistant strain by staff members. ” They also comment that “Such precautions at institutions in
countries not yet overwhelmed by high rates of isolation of methicillin-resistant S. aureus might likewise
hinder the spread of these strains. ”
SOURCES: V.T. Rosdahl and A.M. Knudson, 1991. The decline of methicillin resistance among Danish Staphylococcus aureus
strains. Infection Control and Hospital Epidemiology 12(2):83-88; H. Westh, J.0. Jarlov, H. Kjersem, et al. 1992. The disappear-
ance of multiresistant Staphylococcus aureus in Denmark: Changes in strains of the 83A complex between 1969 and 1989. Clini-
cal Infectious Diseases 14(6) .1186-1194
80 | Impacts of Antibiotic-Resistant Bacteria
HOSPITAL ACCREDITATION AND INFECTION CONTROL REGULATIONS UNDER MEDICARECurrent hospital accreditation and Medicare reg-ulations recognize that each hospital must ana-lyze conditions in its own facility to determinethe best methods of infection control.
Loeb and O’Leary of The Joint Commissionon Accreditation of Healthcare Organizations(JCAHO) explain that
The Joint Commission historically has usedcompliance with contemporary standards as itsbasic measure of health care quality in theaccreditation process. In recent years, however,there has been growing interest in monitoringand evaluating the actual results of care. . .
JCAHO has recently developed a system forperformance measurement called the IndicatorMeasurement System (IMSystem). Beginning in1996, the system will include several measure-ments related to antibiotic use and infection con-trol: timing of administration of prophylacticantibiotics, surveillance and prevention of surgi-cal site infection, surveillance and prevention ofventilator-associated pneumonia, and surveil-lance and prevention of primary blood streaminfections. JCAHO has recognized “. . . thealready tremendous information burdens on mostorganizations” and therefore has designed
“. . . the IMSystem to be parsimonious, thatis, to collect only those data elements that areneeded and to use all the elements that are col-lected. Whenever possible, the IMSystem usesdata elements likely to be already collected byhealth care organizations” (IMSystem GeneralInformation, JCAHO).
Participation in this system, which is volun-tary, has great potential to help hospitals identifyspecific problems in infection control.
Medicare regulations state that as a conditionof participation in Medicare, hospitals must havea quality assurance program in which “nosoco-mial infections and medication therapy must beevaluated” (42 CFR 482.21a2). Further, “theremust be an active program for the prevention,control, and investigation of infectious and com-
municable diseases” (42 CFR 482.42). This pro-gram includes the designation of an infectioncontrol officer who “must develop a system foridentifying, reporting, investigating, and control-ling infections and communicable diseases ofpatients and personnel” (42 CFR 482.42a1) and“must maintain a log of incidents related toinfections and communicable diseases” (42 CFR482.42a2).
In the past, regulations for accreditation andMedicare participation were more specificallyworded, and specifically acknowledged the prob-lems of antibiotic resistance: for example, hospi-tals had to have “measures which control theindiscriminate use of preventive antibiotics in theabsence of infection, and the use of antibiotics inthe presence of infection is based on necessarycultures and sensitivity tests” (42 CFR405.1022c6 as of Oct. 1, 1983). However, basedon past experiences such as those described inthis chapter, specific regulations such as thesemay not be applicable to every facility.
❚ Surveillance of Antibiotic-Resistant BacteriaThere is no national system for reporting thepresence and pattern of antibiotic-resistant bacte-ria, leaving physicians and scientists in the darkabout the prevalence of those organisms in dif-ferent geographical areas. Although many in-hospital, small-scale surveillance systems,designed to track the spread of disease-causingorganisms, including antibiotic-resistant bacte-ria, provide information to physicians aboutwhich antibiotics remain effective, there is nostandard format for the collection and dissemina-tion of data. Antibiotic prescriptions and micro-biology test results are often recorded onseparate slips of paper, making correlation of thetwo sets of data almost impossible. However, theincreasing use of computer technology and theInternet provides increased opportunities forstandardized record keeping in hospitals andeasy database collection and access.
At the state level, the New Jersey StateDepartment of Health started collecting dataabout antibiotic-resistant bacteria in 1991. The
Chapter 4 Antibiotic Use in Hospitals | 81
system includes the 95 acute-care hospitalslicensed by the State of New Jersey and uses datathat are already routinely collected in hospitallaboratories. All hospitals make monthly reportsto the State Department of Health, which, in turn,disseminates its compilation of information toanyone on request. This system’s tracking ofvancomycin-resistant Enterococcus (VRE)spurred collaborative efforts involving privateand public sector and academic organizations toevaluate risk factors for the disease, treatmentoptions, effectiveness of infection-control proce-dures, and the in-vitro susceptibility of VRE toantimicrobial agents during the planning of clini-cal trials (MMWR, 1995). The system is inex-pensive to operate and simple to maintain.
SCOPE, Surveillance and Control of Patho-gens of Epidemiological Importance, is anational effort established by the University ofIowa and Lederle Laboratories (now Wyeth-Ayerst Lederle Laboratories) in 1995. The pro-gram expects to collect reports of all nosocomialbloodstream infections in 48 hospitals nation-wide as well as samples of the organisms isolatedfrom the infected patients. The reports will pro-vide information about the spread of antibiotic-resistant bacteria in the participating hospitals.The bacterial samples will be banked at the Uni-versity of Iowa, and the accuracy of bacterialidentification and antibiotic resistance determi-nations will be verified for representative sam-ples. For a fee, the University will test newantibiotics from any company against bacteria inits collection. The first hospital entered the pro-gram on April 1, 1995, and 40 had entered byJune 30.
There are also other industry-funded surveil-lance systems. A number of academic and com-mercial laboratories conduct surveillance undercontract to pharmaceutical companies, but theyare not necessarily designed to obtain informa-tion most useful for public health purposes.Instead, and understandably, they collect infor-mation about the efficacy of producers’ products.
The National Nosocomial Infection Survey(NNIS), which is run by CDC, is the singlenationwide surveillance system that produces
information about antibiotic-resistant bacteria.While it is limited to reports on nosocomialinfections, it is the source for most of the data inthis OTA report about MRSA, VRE, and otherdrug-resistant bacterial infections.
CDC is in the early stages of establishingnationwide surveillance of drug-resistant Strep-tococcus pneumoniae (DRSP), which will coverinfections whether or not they occur in a hospital.The system requires that participating laborato-ries test all S. pneumoniae isolated from bloodand cerebrospinal fluid for antibiotic susceptibil-ity by using standard testing methods, and thatall test results be reported to the state healthdepartments. The CDC initiated this system in 20laboratories in New Jersey in April 1995, and iffunds are available, the organization expects thatmost of the nearly 2,000 hospital and commerciallaboratories that now have computerized recordkeeping will be in the system by 1998. As labo-ratories add computer capabilities, CDC willencourage them to enlist in the system, and itexpects that all of the nearly 5,000 laboratories inthe country will participate. If the DRSP systemworks, CDC envisions expanding it to includeother antibiotic-resistant bacteria. As an earlystep in setting up the DRSP system, and atCDC’s request, the Council of State and Territo-rial Epidemiologists has recommended DRSP forinclusion on the list of notifiable diseases, andfour states now report it.
WHONET, a surveillance project of theWorld Health Organization, was established andoperated by two scientists, and it functions on ashoestring budget. The system collects informa-tion about resistance patterns in bacteria fromabout 100 hospitals all over the world, makes thedata available to researchers, and provides muchof the available information about the interna-tional flow of antibiotic-resistant bacteria.
One of WHONET’s great strengths is that ithas demonstrated that laboratories around theworld can produce data that can be interpretedand incorporated into a system that providesresults that are comparable from country to coun-try. To do this, the network collects laboratorydata, not interpretations of the data. While rulesfor interpreting susceptibility test results differ
82 | Impacts of Antibiotic-Resistant Bacteria
among various countries, WHONET can makeinternational comparisons based on the raw data.
Participating institutions also gain fromWHONET. The network provides laboratorieswith a computer program, which can be taught inabout six hours, and, where necessary, a com-puter. The software of WHONET, set up to iden-tify unusual patterns of resistance, allows theinfection control practitioner at the hospital totrace the spread of individual strains of bacteriaand use that information to modify infection con-trol procedures.
WHONET is inexpensive, it requires littlesupervision, and it obtains raw data, the data ofmost value to researchers (see chapter 6). It hasbeen successful in obtaining information fromdeveloping countries as well as developed ones,
and it provides an example of the feasibility ofcollecting and reporting antibiotic-resistanceinformation for little money.
❚ Controlling the Use of AntibioticsMuch evidence links the use of antibiotics to theemergence and spread of antibiotic resistance.Table 4-3 summarizes some studies which dem-onstrate relationships between increased use ofantibiotics and prevalence of resistance in hospi-tal organisms. There are also many exampleswhere the prevalence of resistance in hospitalorganisms decreased when the use of antibioticswas decreased (table 4-4). McGowan (1994)recently asked the question: “Do intensive hospi-tal antibiotic control programs prevent the spreadof antibiotic resistance?” and concluded that
TABLE 4-3: Some Studies Demonstrating a Temporal Relationship Between Increased Usage of Antimicrobial Agents and Increased Prevalence of Resistant Hospital Organisms
Year Reference Setting for use of antimicrobials Organism(s) Antimicrobial(s) used
1953 1 General use Staphylococcus aureus Erythromycin
S. aureus Penicillin
S. aureus Chlortetracycline
1956 2 Burn ward S. aureus Chloramphenicol
S. aureus Chlortetracycline
1967 3 Surgical prophylaxis S. aureus Neomycin cream
1971 4 Burn ward Pseudomonas aeruginosa Gentamicin
1978 5 Surgical prophylaxis P. aeruginosa Gentamicin
Serratia Gentamicin
1979 6 Postoperative use Serratia Gentamicin
1. M.H. Lepper, B. Moulton, H.F. Dowling, et al. 1953. Epidemiology of erythromycin-resistant staphylococci in a hospital population—effect onthe therapeutic activity of erythromycin. In: H. Welch and F. Martí-Ibáñez (eds.) Antibiotics annual 1953–1954. New York, NY. Medical Encyclope-dia, pp. 308–313.2. C.D. Gibson, Jr., and W.C. Thompson, Jr. 1956. The response of burning wound staphylococci to alternating programs of antibiotic therapy. In:H. Welch and F. Martí-Ibáñez (eds.) Antibiotics annual 1955–1956. New York, NY. Medical Encyclopedia, pp. 32–34.3. P.M. Rountree, M.A. Beard, J. Loewenthal, et al. 1967. Staphylococcal sepsis in a new surgical ward. British Medical Journal 1:132–137.4. J.A. Shulman, P.M. Terry, and C.E. Hough. 1971. Colonization with gentamicin-resistant Pseudomonas aeruginosa, pyocine type 5, in a burnunit. Journal of Infectious Disease 124(suppl):S18–23.5. N.J. Roberts, Jr., and R.G. Douglas, Jr. 1978. Gentamicin use and Pseudomonas and Serratia resistance: effect of a surgical prophylaxis reg-imen. Antimicrobial Agents and Chemotherapy 13:214–220.6. V.L. Yu, C.A. Oakes, K.J. Axnick, et al. 1979. Patient factors contributing to the emergence of gentamicin-resistant Serratia marcescens. Amer-ican Journal of Medicine 66:468–472.
SOURCE: J.E. McGowan, Jr. 1983. Antimicrobial resistance in hospital organisms and its relation to antibiotic use. Reviews of Infectious Dis-eases 5(6):1033–1048.
Chapter 4 Antibiotic Use in Hospitals | 83
TABLE 4-4: Some Studies Demonstrating a Temporal Relationship Between Decreased Usage of Antimicrobial Agents and Decreased Prevalence of Resistant Organisms
Year Reference Setting for use of antimicrobials Organism(s) Antimicrobial(s) used
1953 1 General use Staphylococcus aureus Chloramphenicol
1954 2 General use S. aureus Erythromycin
1956 3 Burn ward S. aureus Chlortetracycline
S. aureus Chloramphenicol
1960 4 General use S. aureus Penicillin
1960 S. aureus Tetracycline
1966 5 Pediatric ward S. aureus Erythromycin
1967 6 Surgical prophylaxis S. aureus Neomycin cream
1970 7 General use Escherichia coli Streptomycin
Klebsiella, Enterobacter Streptomycin
1970 8 Neurosurgical unit Klebsiella “All”
1970 9 General use S. aureus Erythromycin
S. aureus Novobiocin
1971 10 Burn ward Pseudomonas aeruginosa Gentamicin
1972 11 Burn ward “Enterobacteriaceae” Carbenicillin
Pseudomonas aeruginosa Carbenicillin
1973 12 Nursery “Enterobacteria” Carbenicillin
1974 13 Urology ward “Gram-negative bacilli” 5 agents
1975 14 Nursery E. coli Kanamycin
1978 15 Surgical prophylaxis Pseudomonas aeruginosa Gentamicin
16 Serratia Gentamicin
1. W.M.M. Kirby, and J.J. Ahern. 1953. Changing pattern of resistance of staphylococci to antibiotics. Antibiotics and Chemotherapy 3:831–835.2. M.H. Lepper, B. Moulton, H.F. Dowling, et al. 1953. Epidemiology of erythromycin-resistant staphylococci in a hospital population—effect onthe therapeutic activity of erythromycin. In: H. Welch and F. Martí-Ibáñez (eds.) Antibiotics annual 1953–1954. New York, NY. Medical Encyclope-dia, pp. 308–313.3. C.D. Gibson, Jr., and W.C. Thompson, Jr. 1956. The response of burning wound staphylococci to alternating programs of antibiotic therapy. In:H. Welch and F. Martí-Ibáñez (eds.) Antibiotics annual 1955–1956 New York, NY. Medical Encyclopedia, pp. 32–34.4. M. Barber, A.A.C. Dutton, M.A. Beard, et al. 1960. Reversal of antibiotic resistance in hospital staphylococcal infection. British Medical Journal1:11–17.5. A.W. Bauer, D.M. Perry, and W.M.M. Kirby. 1960. Drug usage and antibiotic susceptibility of staphylococci. Journal of the American MedicalAssociation 173:475–480.6. J.O. Forfar, A.J. Keay, A.F. Maccabe, et al. 1966. Liberal use of antibiotics and its effect in neonatal staphylococcal infection, with particularreference to erythromycin. Lancet 2:295–300.7. P.M. Rountree, M.A. Beard, J. Loewenthal, et al. 1967. Staphylococcal sepsis in a new surgical ward. British Medical Journal 1:132–137.8. R.J. Bulger, E. Larson, and J.C. Sherris. 1970. Decreased incidence of resistance to antimicrobial agents among Escherichia coli and Kleb-siella-Enterobacter: observations in a university hospital over a 10-year period. Annals of Internal Medicine 72:65–71.9. D.J.E. Price, and J.D. Sleigh. 1970. Control of infection due to Klebsiella aerogenes in a neurosurgical unit by withdrawal of all antibiotics. Lan-cet 2:1213–1215.10. M. Ridley, D. Barrie, R. Lynn, et al. 1970. Antibiotic-resistant Staphylococcus aureus and hospital antibiotic policies. Lancet 1:230–233.11. J.A. Shulman, P.M. Terry, and C.E. Hough. 1971. Colonization with gentamicin-resistant Pseudomonas aeruginosa, pyocine type 5, in a burnunit. Journal of Infectious Disease 124(suppl):S18–23.12. E.J.L. Lowbury, J.R. Babb, and E. Roe. 1972. Clearance from a hospital of gram-negative bacilli that transfer carbenicillin-resistance toPseudomonas aeruginosa. Lancet 2:941–945.13. J.A. Franco, D.V. Eitzman, and H. Baer. 1973. Antibiotic usage and microbial resistance in an intensive care nursery. American Journal of Dis-eases of Children 126:318–321.14. H. Søgaard, C. Zimmermann-Nielsen, and K. Siboni. 1974. Antibiotic-resistant gram-negative bacilli in a urological ward for male patients dur-ing a nine-year period: relationship to antibiotic consumption. Journal of Infectious Disease 130:646–650.15. J.B. Howard, and G.H. McCracken, Jr. 1975. Reappraisal of kanamycin usage in neonates. Journal of Pediatrics 86:949–956.16. D.L. Palmer. Epidemiology of antibiotic resistance. 1980. Journal of Medicine 11:255–262.
SOURCE: J.E. McGowan, Jr. 1983. Antimicrobial resistance in hospital organisms and its relation to antibiotic use. Reviews of Infectious Dis-eases 5(6):1033–1048.
84 | Impacts of Antibiotic-Resistant Bacteria
. . . in a few institutions there has been anincrease in susceptibility to antimicrobials fol-lowing intensive control or monitoring . . . in afew hospitals, intensive antibiotic control forselected drug-organisms pairs was associatedwith a high prevalence of susceptibility, and theproportion susceptible fell abruptly when con-trol or monitoring was relaxed or removed.
This latter finding indicates that the decreasein resistance may not be stable: reintroduction ofthe antibiotic can cause the resistance to immedi-ately return.
There are also counterexamples where antibi-otic control programs do not increase susceptibil-ity. In one example, resistance patterns inEnterobacter cloacae but not Pseudomonasaeruginosa were related to ceftazidime use in 18different hospitals in different geographical loca-tions (Ballow and Schentag, 1992). Silber et al.found that “facilities with restriction programswere as likely as those without to have had a caseof VRE bacteremia.” In Denmark the use ofmethicillin increased substantially in the 1970swhile the prevalence of MRSA decreased sub-stantially. The decrease in MRSA was correlatedwith a decrease in the use of tetracycline andstreptomycin (Rosendal et al., 1977). This mightbe explained by the use of tetracycline and strep-tomycin selecting for bacteria with multi-resis-tant plasmids (see chapter 2) also containinggenes for resistance to methicillin. Takentogether, these examples indicate that it is notsimple to determine the specific relationshipbetween antibiotic use and antibiotic resistance.
CDC recently began a systematic study of therelationship between antibiotic use and antibioticresistance. In the initial phase of the I-CARE(Intensive Care Antimicrobial Resistance Epide-miology) project, eight pilot hospitals monitoredthe use of antibiotics and the numbers of antibi-
otic-resistant bacteria. The results for MRSA(shown in figure 4-4) indicate that some hospi-tals use large amounts of methicillin and havehigh frequencies of resistant organisms(hospital B), while others use very little methicil-lin, but still have high frequencies of resistantorganisms (hospital E).
One possible explanation for this is suggestedby the Klebsiella results in figure 4-5: hospital Emay be receiving many patients from anotherhospital (or nursing home) that uses a lot ofmethicillin. Hospital H is interesting in that it hasone of the lowest rates of MRSA and the highestuse of methicillin of any of the eight pilot hospi-tals. This result might be related to a recent resultfrom a French 15-year study (Loulergue et al.,1994) that showed the prevalence of MRSA wasunrelated to cloxacillin (a semisynthetic penicil-lin derivative closely related to methicillin) useon some wards of a hospital where none of thestaff was a carrier of MRSA. This study indi-cated that carriage of MRSA by hospital staff isone risk factor for patients becoming infectedwith MRSA. The data from I-CARE correlate theemergence and spread of antibiotic resistancewith different causes in different hospitals.Moreover, the pilot study demonstrates how use-ful a system such as I-CARE can be in compar-ing an individual hospital to national trends andusing that comparison to design antibiotic useand infection control procedures specifically tai-lored to the problems in the individual hospi-tal.Antibiotics are widely used by physicians incommunity practice as well as by physicians inthe hospitals. In one study (table 4-5), about halfof the cardiac surgery patients colonized withcefazolin-resistant strains of bacteria were colo-nized upon admission to the hospital.5 There-fore, some antibiotic-resistant strains arise in thecommunity, indicating that antibiotic use must be
5 Cefazolin is commonly administered to cardiac patients as prophylaxis to prevent infections during the surgery. The risk of developinga Staph. aureus infection after cardiac surgery has been estimated as 15-44 percent (Mandell, Bennet, Dolin, page 2747). Colonization of thepatient or attending staff with cefazolin-resistant strains would be a significant risk factor for surgical infections when cefazolin is used forprophylaxis.
Chapter 4 Antibiotic Use in Hospitals | 85
controlled by community physicians as well asby hospital physicians in order for hospital-basedprograms to be fully effective. (For more infor-mation about antibiotic-resistant bacteria andantibiotic use in the community, see chapter 3.
❚ Improving Antibiotic Use
Antibiograms To guide physicians in the use of antibiotics,many hospitals provide “antibiograms” thatdescribe the susceptibility of commonly encoun-tered bacteria to various antibiotics. As shown intable 4-6, the vast majority of causes of bacterialinfections in both inpatients and outpatientsremain sensitive to the modern antibiotics. Onthe other hand, many Staph. aureus, coagulase-negative Staphylococci, and S. pneumoniae areresistant to many commonly used antibiotics, andsome Enterococcus are resistant to all antibiotics.
FormulariesThe use of all drugs in hospitals is increas-
ingly controlled by hospital formularies, whichwere set up to control the costs of drugs. The for-mularies may have the added benefit of helpingto control the use of antibiotics and the antibioticresistance problem. In Denver, Colorado, areahospitals (North, 1993), a formulary is combinedwith a computerized antibiotic order form. Thissystem restricts some antibiotics to approvedindications, and use of others requires approvalby specialists in infectious disease. This systemhas saved the hospitals money, and allowed themto easily change the formulary when susceptibil-ity testing indicated a problem of increased resis-tance to a specific antibiotic ) .
Physician EducationPhysician education is crucial to avoid mistakesmade by inadequate knowledge of antibiotic
FIGURE 4-4a: Percent of Staphylococcus Aureus Resistant to Methicillin
SOURCE: National Nosocomial Infections Surveillance System, Centers for Disease Control, Atlanta, GA.
FIGURE 4-4b: Grams of Methicillin Used per 1,000 Patient Days
SOURCE: National Nosocomial Infections Surveillance System, Centers for Disease Control, Atlanta, GA.
FIGURE 4-4c: Percent of Staphylococcus aureus Resistant to Methicillin/Methicillin Use
SOURCE: National Nosocomial Infections Surveillance System, Centers for Disease Control, Atlanta, GA.
Chapter 4 Antibiotic Use in Hospitals 85
I
0 10 20 30 40 50
5 , I
0 20 40 60 80 100
5 , I
0 20 40 60 80 100
SOURCE: National Nosocomial Infections Surveillance System, Centers for Disease Control, Atlanta, GA.
I
86 Impacts of Antibiotic-Resistant Bacteria
I
801
6 0
501
40+
30
20
10
86 87 88 89 90 91 92 93
SOURCE: National Nosocomial Infections Surveillance System,Centers for Disease Control, Atlanta, GA.
1
I / I
I
SOURCE: National Nosocomial Infections Surveillance System,Centers for Disease Control, Atlanta, GA.
controlled by community physicians as well asby hospital physicians in order for hospital-basedprograms to be fully effective. (For more infor-mation about antibiotic-resistant bacteria andantibiotic use in the community, see chapter 3.
90-
80-
70-
86 87 88 89 90 91 92 93
SOURCE: National Nosocomial Infections Surveillance System,Centers for Disease Control, Atlanta, GA.
❚ Improving Antibiotic Use
AntibiogramsTo guide physicians in the use of antibiotics,many hospitals provide “antibiograms” thatdescribe the susceptibility of commonly encoun-tered bacteria to various antibiotics. As shown intable 4-6, the vast majority of causes of bacterialinfections in both inpatients and outpatientsremain sensitive to the modern antibiotics. Onthe other hand, many Staph. aureus, coagulase-negative Staphylococci, and S. pneurnoniae areresistant to many commonly used antibiotics, andsome Enterococcus are resistant to all antibiotics.
FormulariesThe use of all drugs in hospitals is increas-
ingly controlled by hospital formularies, whichwere set up to control the costs of drugs. The for-mularies may have the added benefit of helpingto control the use of antibiotics and the antibioticresistance problem. In Denver, Colorado, areahospitals (North, 1993), a formulary is combinedwith a computerized antibiotic order form. Thissystem restricts some antibiotics to approvedindications, and use of others requires approval
Chapter 4 Antibiotic Use in Hospitals | 87
tion about susceptibilities of different organisms.Physicians must learn to check other reliable up-to-date sources of information about antibioticssuch as The Medical Letter On Drugs and Thera-peutics (New Rochelle, NY: The Medical Letter,Inc.) and to consult with infectious diseaseexperts who are aware of susceptibility patternsin the specific hospitals.
Computerized Systems for Antibiotic MonitoringThe LDS Hospital in Salt Lake City, Utah, hasdeveloped a computerized antibiotic monitoringsystem, which is part of a larger computerizedpatient record system that automatically collectssurveillance data and generates antibiograms(see table 4-6) (Evans and Pestotnik, 1994).When the microbiology laboratory results areentered into the computer, the computer checksthe susceptibilities of the organisms against theantibiotic prescribed for the patient and generatesan alert when an antibiotic is inappropriate. Inone year, the system generated an alert for32 percent of the patients. However, many physi-cians did not change the antibiotic based on thealert, often because the patient was clinically
improving even though the susceptibility resultsindicated that the antibiotic was inappropriate.6
The system also notifies physicians of the opti-mum time for administration of prophylacticantibiotics. Use of the system saved $42 perpatient in the first year of use, with a projectedreduction in the costs of prophylactic antibioticsof over $89,000 per year in a single hospital(Evans et al., 1990).
Another part of the antibiotic monitoring sys-tem at the LDS hospital is a computerized antibi-otic consultant (Evans et al., 1994). This systemuses surveillance data together with informationabout the site of the infection and patient aller-gies to determine the best choice of empiric anti-biotic therapy. The computer consultant wasbetter at choosing antibiotics than the physiciansin the hospital. The computer chose antibiotics towhich the infecting bacteria were susceptible94 percent of the time; the physicians chose cor-rectly 77 percent of the time.
Setting up a comprehensive patient data sys-tem requires significant financial investment byhospitals. However, the hospitals will realizecost savings just from improvement in the use ofantibiotics. Forty to fifty percent of hospital
6 Many patients recover from bacterial illnesses on their own without the help of an antibiotic.
TABLE 4-5: Characteristics of Cardiac Surgery Patients Colonizedwith Cefazolin-Resistant Gram-Negative Bacilli
Location at first positive culture (% patients)
Species
Number of patients colonized
(n = 87) At admission48–72 hr into
CSICU> 72 hr into
CSICU
Percent of colonization due
to horizontal transmission
Percent developing
clinicalinfection
Enterobacter species
58 50 34 16 16 21
Citrobacter species
37 49 22 29 ? 3
Pseudomonas aeruginosa
33 55 12 33 9 27
Serratia marcesens
7 43 57 0 29 29
KEY: CSICU = cardiac surgery intensive care unit; ? = unknown (no typing system used).
SOURCE: Adapted from D.M. Flynn, R.A. Weinstein, and S.A. Kabins. 1988. Infections with gram-negative bacilli in a cardiac surgery intensivecare unit: The relative role of Enterobacter. Journal of Hospital Infections 11:367.
88 | Impacts of Antibiotic-Resistant Bacteria
TABL
E 4-
6: A
ntib
iotic
Sus
cept
ibilit
y of
Com
mon
Org
anis
ms
(Num
bers
Indi
cate
Per
cent
Sus
cept
ible
) Jan
uary
–Dec
embe
r 199
4 (P
age
1 of
2)
IN P
ATIE
NTS
Stap
h.
aure
us(1
080)
Coag
ulas
e-ne
g. S
taph
. (6
47)
Stre
p.
pneu
mon
iae
(75)
Ente
roco
ccus
sp
p.(6
12)
E. coli
(808
)
K.
spec
ies
(396
)
C.
dive
rsus
(3
5)
C.fr
eund
ii (4
2)
Ente
roba
cter
spec
ies
(309
)
Serr
atia
sp
ecie
s (5
5)
Prot
eus
mir
abili
s (1
00)
P.
aeru
gino
sa
(520
)
A.
anitr
atus
(6
0)
X.
mal
toph
ilia
(42)
H.
influ
enza
e (1
28)
B.
frag
ilis
grou
p (3
0)
Pen
icill
in5
686
3
Am
pic
illin
8854
00
00
091
75
Am
pic
illin
-su
lbac
tam
***
***
5859
9150
1513
9754
093
Oxa
cilli
n70
26
Tica
r-cl
avul
ante
***
***
*77
7596
6261
8810
067
5852
100
Cep
halo
thin
****
***
*73
7197
31
094
Cef
azol
in**
***
*88
7897
73
097
Cef
otet
an**
***
*99
8897
6855
9299
70
Cef
urox
ime
***
***
99
Cef
otax
ime
***
***
9399
9597
7766
9499
100
76
Cef
tazi
dim
e84
6923
Imip
enem
***
***
100
100
960
Chl
oram
phe
nico
l99
Ery
thro
myc
in91
Trim
eth-
sulfa
8339
7780
7710
058
8893
9570
9088
Cip
roflo
xaci
n70
5421
9998
100
100
9699
100
8261
14
Nitr
ofur
anto
in87
9661
4515
Gen
tam
icin
9791
9593
100
9884
6650
Am
ikac
in**
**10
098
9984
9135
Clin
dam
ycin
7248
92
Met
roni
daz
ole
100
Van
com
ycin
100
100
99
* P
iper
acill
in-t
azob
acta
m w
as a
dd
ed to
the
form
ular
y la
te in
the
year
. Its
act
ivity
is e
qui
vale
nt to
or
som
ewha
t sup
erio
r to
tica
rcill
in-c
lavu
lana
te.
**Te
sted
for
the
oral
cep
halo
spor
ins,
cep
hale
xin
and
cep
hrad
ine.
***
Oxa
cilli
n-re
sist
ant s
tap
hylo
cocc
i are
als
o re
sist
ant t
o b
eta-
lact
amas
e in
hib
itor
com
bin
atio
ns, c
epha
losp
orin
s, a
nd im
ipen
em; o
xaci
llin-
susc
eptib
le s
tap
hylo
cocc
i are
sus
cep
tible
to th
ose
agen
ts.
****
Am
ikac
in r
epor
ted
onl
y on
Gen
tam
icin
res
ista
nt is
olat
es.
Chapter 4 Antibiotic Use in Hospitals | 89
TABL
E 4-
6: A
ntib
iotic
Sus
cept
ibilit
y of
Com
mon
Org
anis
ms
(Num
bers
Indi
cate
Per
cent
Sus
cept
ible
) Jan
uary
–Dec
embe
r 199
4 (P
age
2 of
2)
OUT
PATI
ENTS
Stap
h.
aure
us(9
66)
Coag
ulas
e-ne
g. S
taph
. (3
20)
Stre
p.
pneu
mon
iae
(103
)
Ente
roco
ccus
sp
p.(6
32)
Esch
eric
hia
col
i(2
558)
K.
spec
ies
(405
)
C.
dive
rsus
(7
6)
C.
freu
ndii
(48)
E.
spec
ies
(136
)
Serr
atia
sp
ecie
s (4
6)
Prot
eus
mira
bilis
(2
44)
P.
aeru
gino
sa (2
34)
Salm
onel
la
spec
ies
(25)
Shig
ella
sp
ecie
s(5
1)
H.
influ
enza
e (1
21)
Pen
icill
in4
1383
Am
pic
illin
9453
20
62
089
9233
72
Am
pic
illin
-su
lbac
tam
***
***
7275
9581
3712
97
Oxa
cilli
n89
52
Tica
r-cl
avul
ante
***
***
*90
9010
090
9293
100
85
Cep
halo
thin
****
***
*65
8696
33
094
Cef
azol
in**
***
*90
9197
134
093
Cef
otet
an**
***
*99
9710
091
8390
100
Cef
urox
ime
***
***
100
Cef
otax
ime
***
***
9299
9810
094
9294
100
100
100
Cef
tazi
dim
e95
Chl
oram
phe
nico
l92
8710
0
Ery
thro
myc
in84
Trim
eth-
sulfa
9254
8081
8310
061
9488
9310
070
94
Cip
roflo
xaci
n88
7224
9997
9792
9796
9987
100
100
Nitr
ofur
anto
in91
9962
9190
510
0
Gen
tam
icin
9996
100
9798
9595
93
Am
ikac
in**
**93
100
93
Clin
dam
ycin
9176
Van
com
ycin
100
100
100
* P
iper
acill
in-t
azob
acta
m w
as a
dd
ed to
the
form
ular
y la
te in
the
year
. Its
act
ivity
is e
qui
vale
nt to
or
som
ewha
t sup
erio
r to
tica
rcill
in-c
lavu
lana
te.
**
Test
ed fo
r th
e or
al c
epha
losp
orin
s, c
epha
lexi
n an
d c
ephr
adin
e.**
*O
xaci
llin-
resi
stan
t sta
phy
loco
cci a
re a
lso
resi
stan
t to
bet
a-la
ctam
ase
inhi
bito
r co
mb
inat
ions
, cep
halo
spor
ins,
and
imip
enem
; oxa
cilli
n-su
scep
tible
sta
phy
loco
cci a
re s
usce
ptib
le to
thos
e ag
ents
.**
**A
mik
acin
rep
orte
d o
nly
on G
enta
mic
in r
esis
tant
isol
ates
.K
EY
:S
tap
h. =
Sta
phy
loco
cci
P. a
erug
inos
a =
Pse
udom
onas
aer
ugin
osa
Str
ep. =
Str
ephc
occu
s A
. ani
trat
us =
Aci
neto
bac
ter
anitr
atus
K. s
pec
ies
= K
leb
siel
la s
pec
ies
X. =
Xan
thom
onas
mal
top
hilia
C. d
iver
sus
= C
itrob
acte
r d
iver
sus
H. i
nflu
enza
e =
Hae
mop
hilu
s in
fluen
zae
C. f
reun
dii
= C
itrob
acte
r fr
eund
ii B
. tra
gill
is g
roup
= B
acte
roid
es tr
agill
is g
roup
E. s
pec
ies
= E
nter
obac
ter
spec
ies
SO
UR
CE
: Uni
vers
ity H
ealth
Sys
tem
, San
Ant
onio
, TX
, 199
4. W
ith p
erm
issi
on o
f J.H
. Jor
gen
sen.
90 | Impacts of Antibiotic-Resistant Bacteria
BOX 4-3: “Food-Borne” Outbreak of Expensive Antibiotic Usein Community Teaching Hospital
To the Editor—Drug utilization review assures cost-effective use of medications in hospitals. Wepresent an example of drug utilization review that began with the identification of an “index case” of acostly therapeutic decision. Subsequent investigation lead to the identification of a prescribing outbreakas well as its probable source.
Report of a Case—A 32-year-old man had been on a camping trip and noted an insect bite at the topmargin of his sock. The next day he noted redness and swelling at the site of the bite. The third day hewas febrile and the redness began to spread. On the fourth day, red streaks extended 15 cm above thesite of injury. He felt ill and came to the emergency department. His examination demonstrated a temper-ature of 39.4°C, sickly appearance, and a tender cellulitis of his lower leg. Blood pressure was normaland he did not have a truncal rash. Therapy with a new, expensive, broad-spectrum antibiotic was initi-ated. When asked about his antibiotic choice, the admitting intern noted at morning report that he hadplanned on giving penicillin or nafcillin, but had been overruled by the supervising resident who insistedon a “more modern choice for a severely ill patient.”
Comment—Following discussion of this case, we evaluated the use of the new antibiotic in our hospi-tal. We found that use had transiently increased following its addition to our formulary in February 1994,then abruptly increased in June and July. After conducting interviews with our house officers, it wasrevealed that an extravagant dinner party had been held for incoming and current house staff the thirdweek of June. The sponsor of this dinner was the manufacturer of the antibiotic. The increase in use ofthis agent bore a striking temporal association with this dinner. Furthermore, the prescribing resident hadattended the dinner and directed the admitting intern to use the drug instead of nafcillin.
The prescribed antibiotic exhibits a broad spectrum of activity, including β-lactamase-producingstrains of staphylococci, Haemophilus influenzae, anaerobes, and facultative gram-negative rods. Theagent would be expected to be effective in most settings where nafcillin might be used. Although thisagent is not contraindicated in treating uncomplicated cellulitis, it is much more expensive ($183.20 perday) than other effective drugs such as nafcillin ($84 per day). In this single case, the daily excess cost oftherapy would approximate $100. The relationship between pharmaceutical marketing maneuvers andprescribing is controversial. Previous ecological studies have found an association between educational“enticements” and hospital formulary additions and prescribing trends. However, we are not aware of adetailed case description where a more expensive therapeutic choice was made when less expensivetherapeutic alternatives were indicated. We do not know if the resident’s attendance at the dinner causedhis therapeutic choice. However, the striking epidemiological association between resident attendance atthis drug company-sponsored event and the subsequent changes in hospital-wide prescribing practicesshould prompt training programs to be wary of such outside sources of medical education.
SOURCE: Quoted from R.I. Shore and W.L. Greene, letter to the editor, Journal of the American Medical Association273(24):1908. Copyright 1995, American Medical Association.
Chapter 4 Antibiotic Use in Hospitals | 91
FIGURE 4-6: An Antibiotic Advertisement from a Medical Journal
SOURCE: A major pharmaceutical company.
92 | Impacts of Antibiotic-Resistant Bacteria
pharmacy budgets are for antibiotics, and one-fourth of that in some hospitals is for vancomy-cin alone (Modern Healthcare, 1994). Eliminat-ing unnecessary use of antibiotics will decreasetotal pharmacy expenditures. Treating infectionswith appropriate antibiotics and administeringprophylactic antibiotics with appropriate timingwill also increase the quality of patient care anddecrease the number of days spent in the hospi-tal. (OTA’s report Bringing Health Care Online:The Role of Information Technologies, Septem-ber 1995, discusses costs and benefits of comput-erized patient record systems.)
Practice GuidelinesPractice guidelines, or practice protocols, aremedical guidelines that “encompass a broadrange of strategies designed to assist practitio-ners in the clinical decision-making process”(Shanz, 1993). More specifically, they are “stan-dardized specifications for care developed by aformal process that incorporates the best scien-tific evidence of effectiveness with expert opin-ion” (Leape, 1990). These guidelines are set byexperts from specific areas of the medical profes-sion to advise about recommended standards ofcare. For example, the goal of practice guidelinesestablished by the Agency for Health Care Policyand Research, a federal agency empowered toestablish practice guidelines, is to encouragephysicians and other health care providers tochange their practice behavior, thus improvingpatient care, patient outcomes, and quality of life(AHCPR, 1994).
Practice guidelines on infection control or theprudent use of antibiotics might be helpful incontrolling antibiotic resistance. For example,practice guidelines might specify that older anti-biotics such as amoxicillin be tried for commu-nity-acquired infections before newer, broaderspectrum antibiotics are used. Under managedcare, insurers may adopt guidelines such as thesebecause they will save money as older antibioticsare generally much less expensive than newerantibiotics.
Practice guidelines may also be of use in med-ical malpractice litigation. A major difficulty in
medical malpractice cases is establishing theappropriate standard of care before “layperson”decision-makers on juries. Practice guidelineshave the potential to reduce such difficulties. Byestablishing an unbiased standard of care, prac-tice guidelines should “significantly reduce themost vexing problem in malpractice litigation:the battle of the experts” (West, 1994). In theory,a physician could rely on the practice guidelineas the appropriate standard of care without hav-ing to worry whether a judge or jury, in a medi-cal malpractice case, would consider the careadministered appropriate. The only remainingissues to be determined in medical negligence lit-igation would be whether the practice guideline“is relevant to the case at hand, and whether it isappropriate to use the [guideline] to establish thestandard of care” (West, 1994).
On the other hand, practice guidelines whichsuggest any benefit from the use of antibioticsmay be used as evidence against the physician inthe case of a bad outcome. For example, a guide-line on the treatment of otitis media with effusionpublished by the Agency for Health Care Policyand Research concludes:
Meta-analysis for Guideline developmentshowed a 14 percent increase in the probabilitythat otitis media with effusion would resolvewhen antibiotic therapy was given versus notreatment. . . . When this small improvement inresolution of otitis media with effusion isweighed against the side effects and cost ofantibiotic therapy, antibiotic therapy may not bepreferable to observation in management of oti-tis media with effusion in the otherwise healthyyoung child. . . . To assist in making choices formanagement of otitis media with effusion,health care providers need to inform parentsfully as to the side effects and costs of antibiotictherapy, as well as the benefits and harms ofother options for care (AHCPR, 1994).
A physician who elects not to prescribe anantibiotic, foregoing the 14-percent increasedprobability that the condition “would resolve,”might be held legally liable for any negative out-come. Such potential liability might encouragephysicians to prescribe antibiotics even when
Chapter 4 Antibiotic Use in Hospitals | 93
they may not be necessary. Further, the aboveguidelines do not instruct physicians to considerthe spread of antibiotic resistance in the decisionto prescribe antibiotics. If practice guidelines aregoing to have an effect on promoting prudentantibiotic use, they have to acknowledge that thebenefit to a few patients from routine use ofnewer and broader spectrum antibiotics may beoutweighed by the public health benefitsexpected from reducing the prevalence of antibi-otic-resistant bacteria.
One concern of practice guidelines relevant toantibiotic use is that national standards of con-duct do not adequately reflect the localizedaspect of antibiotic-resistant bacteria outbreaks.The National Health Lawyers Associationaddressed this concern in its 1995 ColloquiumReport on Legal Issues Related to Clinical Prac-tice Guidelines, which conceded that “[s]omelocal adaptation of national guidelines is proba-bly inevitable and may be useful, because evenwell-developed guidelines may have gaps andmay not foresee significant local objectives orconstraints” (National Health Lawyers Associa-tion Colloquy, 1995). One solution may be theuse of an online computer system that allowshealth care practitioners in a particular geo-graphic area to consult with each other and localexperts concerning appropriate local adaptationsto practice guidelines (Meyers, 1995). Such asystem would also allow health care practitionersto disseminate the specifics of their cases, as wellas establish a record of compliance with the prac-tice guidelines in the event of future litigation(Meyers, 1995).
COSTS OF CONTROLLING THE EMERGENCE AND SPREAD OF ANTIBIOTIC-RESISTANT BACTERIAHospitals cannot charge costs of infection con-trol procedures and the monitoring of antibiotic-resistant bacteria directly to insurance compa-nies. As a result, although these proceduresimprove the quality of patient care, hospitals’efforts to minimize costs may retard spending onthem. Haley et al. (1987) commented that hospi-
tals might not be placing enough emphasis oninfection control because “the direction and mag-nitude of the financial incentive to prevent noso-comial infections are not clear to many hospitaladministrators.” They analyzed the financialincentives for hospitals to prevent nosocomialinfections under the prospective payment systemand concluded that
Assuming an average nosocomial infectionrate of 5.7 percent, one would expect. . . a hos-pital with 10,000 admissions annually to haveapproximately 570 nosocomial infections peryear in the absence of an effective infectioncontrol program. If the average 1985 marginalcost of providing extra care for a nosocomialinfection were approximately $1800, the totalcost of treating these infections would amountto approximately $1 million per year, not count-ing physicians’ fees or medicolegal losses. . . .From the nationwide SENIC project evaluation,we know that at least 32 percent of the infec-tions can be prevented, thus indicating that aneffective infection control program could pro-duce a gross financial savings of approximately$305,000 per year. . . nearly five times the costsof the program.
A computerized antibiotic monitoring system,such as that of the LDS Hospital, reduces costsboth by controlling the use of antibiotics andreducing the length of hospital stays, but the LDSsystem has been in development for 20 years, itis based on obsolete computer technology, and itis not exportable. Developing a system on cur-rent computer technology will take a significantinvestment in research and development. Givenall the costs involved in control and monitoring,it would be useful to calculate the total cost tohospitals of antibiotic resistance to judgewhether infection control procedures and moni-toring of antibiotic-resistant bacteria will have afinancial payoff.
Many different factors can be considered in acalculation of the cost of antibiotic-resistant bac-teria: the direct cost of time in the hospital, thecosts of extra physician visits when antibioticsare ineffective, the extra hospitalizations due tocommunity-acquired resistant infections, and thecosts of newer antibiotics to replace antibiotics
94 | Impacts of Antibiotic-Resistant Bacteria
such as penicillin to which organisms havebecome resistant. To those must be added theindirect costs to patients from lost days of work,increased illness, and, at worst, death. It is diffi-cult to estimate the costs of all of these factors.
Phelps (1989) made such an estimate and con-cluded that antibiotic-resistant bacteria cost thenation between $0.1 billion and $30 billion annu-ally. Use of different values for the value of a lifeaccounted for almost all of the 300-fold range inthe estimate. The National Foundation for Infec-tious Disease (1990) estimated that the costs ofnosocomial infections caused by antibiotic-resis-tant bacteria could be as high as $4 billion annu-ally, and CDC has estimated the costs of allnosocomial infections at $4.5 billion per year, anestimate that includes costs from both antibiotic-resistant and susceptible infections.
Here, OTA estimates the effects of antibiotic-resistant bacteria on the costs of some hospital-izations. The national costs of five classes ofnosocomial infections—surgical wound infec-tions, pneumonia, bacteremias, urinary tractinfections, and others—are taken from the resultsof the SENIC project (see table 4-1). Those costsare shown on the first data line in table 4-7 (forinstance, the cost of all surgical wound infectionsis $1.6 billion annually). The calculation of thecosts of each of the infections caused by each ofsix different antibiotic-resistant bacteria is illus-trated by the example of MRSA-associated sur-gical wound infections. Staph. aureus isassociated with 19 percent of all surgical woundinfections, and 15 percent of all Staph. aureus isMRSA. Therefore, the hospital cost of MRSA-associated surgical wound infections is$50 million [$1.6 billion × 0.19 × 0.15 =$50 million]. Repeating this process for the fivekinds of infections and the six different antibi-otic-resistant bacteria produces an annual total of$661 million (1992) for hospital costs.
Using the estimate of Holmberg, Solomon andBlake (1987) that antibiotic resistance doublesthe cost of nosocomial infections, the minimumextra cost of antibiotic-resistant bacteria in hos-pitals is $661 million annually (1992 dollars) andthe minimum total cost of antibiotic-resistant
bacteria in hospitals is $1.3 billion annually(1992 dollars). The actual hospital costs arebound to be much higher as this calculation con-siders only six species of bacteria, and in somecases considers strains of bacteria that are resis-tant to only one antibiotic and not other strains ofthe same bacteria that are resistant to other anti-biotics. Further, the trends in antibiotic resistanceindicate that the number of antibiotic-resistantinfections is likely to be increasing rapidly.Finally, the OTA estimate considers only onefactor among many that increase the costs ofantibiotic-resistant bacteria; it ignores costs ofother infections, costs of days of work lost, andpost-hospital care, and other major costs. Forthese reasons, the OTA estimate of $1.3 billionmust be considered a minimum estimate .
CONCLUSIONSTwenty-five to 35 percent of all hospitalizedpatients receive antibiotics, which producesenormous pressure for the selection of antibiotic-resistant bacteria. The result of that pressure isincreasing frequencies of antibiotic-resistant bac-teria in hospitals: Some strains of vancomycin-resistant Enterococcus are now resistant to allFDA-approved antibiotics, and some strains ofStaphylococcus aureus, a common cause ofnosocomial infections, are resistant to all antibi-otics except vancomycin. Many experts fear theemergence and spread of Staph. aureus strainsresistant to all antibiotics, including vancomycin,which would pose a major health care crisis.
Two avenues are open to reduce the spread ofantibiotic-resistant bacteria. One is infection con-trol to reduce the rate of hospital infections, andthe other is the reduction in the use of antibioticsto reduce selection pressures. While infectioncontrol programs have worked well in someinstitutions, similar programs have produced nopositive results elsewhere. The mixed resultsindicate that more research into what makes sys-tems work and why is needed to guide infectioncontrol efforts. Formularies, lists of drugs thatare available for use in a hospital, were estab-lished to control drug costs, but they can be tied
Chapter 4 Antibiotic Use in Hospitals | 95
to information about antibiotic susceptibility pro-duced by the hospital microbiology laboratory toinform physicians’ prescription decisions. Posi-tive results have been reported in the few placesthis has been tried, but more evaluation will benecessary before it is widely adopted.
Surveillance systems are designed to collectand disseminate information to physicians andothers about the presence and prevalence of anti-biotic-resistant bacteria. They are common inhospitals, but far less common between andamong hospitals and across larger geographicalunits. New Jersey has the only statewide system
TABLE 4-7: Costs of Stays in Hospital Associated with Antibiotic-Resistant BacteriaSurgical wound
infection Pneumonia BacteremiaUrinary tract
infection Other Total
Total cost of nosocomial infectionsa 1.6 1.3 0.36 0.61 0.66 4.5
Staph. aureus 19% 20% 16% 2% 17%
Methicillin resistant 15% 15% 15% 15% 15%
Cost of MRSA b 50 40 10 1.8 20 122
Enterococcus 12% 2% 9% 16% 5%
Vancomycin resistant 7.9% 7.9% 7.9% 7.9% 7.9%
Cost of VRE b 20 2 2.6 10 2.4 37
Pseudomonas 8% 16% 3% 11% 6%
Imipenem resistant 7.8% 16.9% 10.3% 6.9% 12.5%
Cost of impenem-resistant pseudomonas b
10 40 1 4.6 5 61
Coagulase-negative Staphylococcus (CoNS)
14% 2% 31% 4% 14%
Methicillin resistant 50% 50% 50% 50% 50%
Cost of methicillin-resistant CoNSb
112 13 56 12 46 239
E. Coli 8% 4% 5% 25% 4%
Ampicillin resistant 35% 35% 35% 35% 35%
Cost of ampicillin-resistant E. Coli b
45 18 6 5 9 83
Enterobacter 7% 11% 4% 5% 4%
Resistant 37% 37% 37% 37% 37%
Cost of resistant enterobacter b 41 52 5 11 9.7 119
TOTAL COSTb 661
a In billions of 1992 dollars.b In billions of 1992 dollars.NOTE: The costs were estimated by multiplying the total cost of nosocomial infections from a specific category (e.g., urinary tract infections) bythe fraction of infections in that category caused by a specific organism (e.g., E. coli) and the fraction of the organism resistant to one specificantibiotic (e.g., ampicillin). The data from the fraction of infections caused by specific organisms and organisms resistant to a specific antibioticwere taken from the CDC/NNIS system. This calculation represents a minimum estimate of the costs of antibiotic resistant bacteria: it onlyaccounts for charges in a hospital for nosocomial acquired infections due to six different antibiotic resistant species.
SOURCE: Office of Technology Assessment, 1995, based on data from the Centers for Disease Control, National Nosocomial Infections Surveil-lance (CDC/NNIS) System, Atlanta, GA.
96 | Impacts of Antibiotic-Resistant Bacteria
in the country, and CDC is only now establishinga nationwide system for one kind of antibiotic-resistant bacterium. In addition, a number of pri-vately supported surveillance systems collectdata for pharmaceutical companies, but, under-standably, those systems collect information fortheir clients rather than for general public healthinformation. On the international level, WHO-NET collects data from over 100 institutionsaround the world. Chapter 1 discusses some fea-tures that could be built into a national surveil-lance system directed at antibiotic-resistantbacteria and offers an option for its implementa-tion.
One estimate of the total costs associated withantibiotic-resistant bacteria had a range of$100 million to $30 billion annually, with mostof the 300-fold range in cost coming from vary-ing estimates of the value of a human life, andanother estimate said that the costs could be up to$4 billion annually. OTA estimates the minimalextra hospital costs associated with five kinds ofnosocomial infections caused by antibiotic-resis-tant bacteria to be $1.3 billion per year. The totalcosts would certainly be certainly higher whenhospital costs of other antibiotic-resistant bacte-rial infections and non-hospital costs are consid-ered.
REFERENCESBallow, C.H., and J.J. Schentag. 1992. Trends in
antibiotic utilization and bacterial resistance.Report of the National Nosocomial Resis-tance Surveillance Group. DiagnosticMicrobiology and Infectious Disease15(Suppl. 2):37S–42S.
Burke, J.F. 1961. The effective period of preven-tive antibiotic action in experimental inci-sions and dermal lesions. Surgery 50:161–168.
Classen, D.C., R.S. Evans, S.L. Pestotnik, et al.1992. The timing of prophylactic adminis-tration of antibiotics and the risk of surgical-wound infection. New England Journal ofMedicine 326(5): 281–286.
Doebbeling, B.N., G.L. Stanley, C.T. Sheetz, etal. 1992. Comparative efficacy of alternativehand-washing agents in reducing nosoco-mial infections in intensive care units. NewEngland Journal of Medicine 327(2): 88–93.
Eickoff, T.C. 1992. Antibiotics and nosocomialinfections. In: J.V. Bennett, and P.S. Brach-man (eds.). Hospital Infections Third Edi-tion. Boston, MA. Little, Brown andCompany.
Evans, R.S., and S.L. Pestotnik. 1993. Applica-tions of medical informatics in antibiotictherapy. In: J.A. Poupard, L.R. Walsh, andB. Kelger (eds.) Antimicrobial SusceptibilityTesting: Critical Issues for the 90s. NewYork, NY. Plenum Press, pp. 87–96.
Evans, R.S., S.L. Pestotnik, J.P. Burke, et al.1990. Reducing the duration of prophylacticantibiotic use through computer monitoringof surgical patients. DICP, The Annals ofPharmacotherapy 24:351–354.
Evans, R.S., D.C. Classen, S.L. Pestotnik, et al.1994. Improving empiric antibiotic selectionusing computer decision support. Archivesof Internal Medicine 154(8):878–884.
Garner, J.S. 1993. The CDC hospital infectioncontrol practices advisory committee. Amer-ican Journal of Infection Control 21(3):160–162.
Goldmann, D. and E. Larson. 1992. Hand-wash-ing and nosocomial infections [editorialcomment]. New England Journal of Medi-cine 327(2): 120–122.
Haley, R.W. 1991. Measuring the costs of noso-comial infections: Methods for estimatingeconomic burden on the hospital. AmericanJournal of Medicine 91 (Suppl. 3B):32S–38S.
Haley, R.W. 1992. Cost-benefit analysis ofinfection control programs. In: J.V. Bennettand P.S. Brachman (eds.) Hospital Infec-tions Third Edition, Boston, MA. Little,Brown and Company, pp. 507–532.
Haley, R.W., J.W. White, D.H. Culver, et al.1987. The financial incentive for hospitals toprevent nosocomial infections under the pro-spective payment system. An empirical
Chapter 4 Antibiotic Use in Hospitals | 97
determination from a nationally representa-tive sample. Journal of the American Medi-cal Association 257(12):1611–1614.
Haley, R.W., D.H. Culver, J.W. White, et al.1985. The nationwide nosocomial infectionrate: a new need for vital statistics. AmericanJournal of Epidemiology 121(2):159–167.
Holmberg, S.D., S.L. Solomon, and P.A. Blake.1987. Health and economic impacts of anti-microbial resistance. Reviews of InfectiousDiseases 9(6):1065–1078.
Institute of Medicine. 1992. Microbial Threats toHealth in the United States. WashingtonD.C., National Academy Press.
Joint Commission on Accreditation of Health-care Organizations. Feb. 1995. IMSystemGeneral Information. Oak Terrace, IL, p. 15.
Jewell, M. 1994. Cost-containment using anoutcome-based best practice model for themanagement of MRSA. Journal of Chemo-therapy VI(Suppl. 2):35–39.
Kernodle, D.S., and A.B. Kaiser. 1990. In: Man-dell, G.L., J.E. Bennett, and R. Dolin (eds.)Mandell, Douglas and Bennett’s Principlesand Practice of Infectious Diseases. NewYork, NY. Churchill Livingstone, Inc.
Kollef, M.H. 1994. Antibiotic use and antibioticresistance in the intensive care unit: Are wecuring or creating disease? Heart and Lung23:363–367.
Kollef, M.H. 1994. The role of selective diges-tive tract decontamination on mortality andrespiratory tract infections: A meta-analy-sis. Chest 105:1101–1108.
Leape, L.L. 1990. Practice guidelines and stan-dards: an overview. 16 Quality Review Bul-letin, pp. 42–49.
Loeb, J.M., D.S. O’Leary. 1995. A call for col-laboration in performance measurement.(From the Joint Commission on Accredita-tion of Healthcare Organizations). Journalof the American Medical Association273(18):1405.
Lonks, J.R., Durkin, M.R., Meyerhoff, A.N., etal. 1995. Meningitis due to ceftriaxone-resis-tant Streptococcus pneumoniae. NewEngland Journal of Medicine 332:893–894.
Loulergue, J., A. Audurier, J.M. DeLarbre, et al.1994. Changes in microbial ecology and useof cloxacillin. Journal of Hospital Infection27(4):275–283.
Mandell, G.L., J.E. Bennett, and R. Dolin. 1990.Mandell, Douglas and Bennett’s Principlesand Practice of Infectious Diseases. NewYork, NY. Churchill Livingstone, Inc.
Mandell, G.L., and M.A. Sande. 1990. Antimi-crobial agents: penicillins, cephalosporins,and other beta-lactam antibiotics. In: A.GGilman, T.W. Rall, A.S. Nies, et al. (eds.)Goodman and Gilman’s The Pharmacologi-cal Basis of Therapeutics, Eighth Edition.New York, NY. Pergamon Press, pp. 1065–1097.
Martin, M.A. 1994. Methicillin-resistant Staphy-lococcus aureus: the persistent resistantnosocomial pathogen. Current Clinical Top-ics in Infectious Diseases 14:170–91.
Martone, W.J., W.R. Jarvis, D.H. Culver, et al.1992. Incidence and nature of endemic andepidemic nosocomial infections. In: J.V.Bennett, and P.S. Brachman (eds.) HospitalInfections Third Edition. Boston, MA. Little,Brown and Company, pp. 577–596.
McGowan, Jr., J.F. 1994. Do intensive hospitalantibiotic control programs prevent thespread of antibiotic resistance? InfectionControl and Hospital Epidemiology15(7):478–483.
Meyers, K., Chief of Infectious Diseases andHospital Epidemiologist, Baptist MedicalCenter. Feb. 21, 1995. Personal communica-tion.
Murray-Leisure, K.A., Geib, S., Graceley, D. etal. 1990. Control of epidemic methicillin-resistant Staphylococcus aureus. InfectionControl and Hospital Epidemiology11(7):343–350.
National Foundation for Infectious Disease.1990. Press release at Interscience Confer-ence on Antimicrobial Agents and Chemo-therapy. Reported in D. North. 1993.
National Health Lawyers Association Colloquy.1995. Legal Issues Related to Clinical Prac-tice Guidelines. vol. 47.
98 | Impacts of Antibiotic-Resistant Bacteria
Noble, W.C., Z. Virani, and R.G. Cree. 1992.Co-transfer of vancomycin and other resis-tance genes from Enterococcus faecalisNCTC 12201 to Staphylococcus aureus.FEMS-Microbiol-Lett. 72(2): 195–198.
North, D. 1993. Controlling the Costs of Antibi-otic Resistance. Clinical Therapeutics15(Suppl. A):3–11.
Phelps, C.E. 1989. Bug/drug resistance. Some-times less is more. Medical Care 27:194–203.
Rosdahl, V.T., and A.M. Knudson. 1991. Thedecline of methicillin resistance among Dan-ish Staphylococcus aureus strains. InfectionControl and Hospital Epidemiology12(2):83–88.
Rosendal, K., O. Jessen, M.W. Bentzon, et al.1977. Antibiotic policy and spread of Sta-phylococcus Aureus strains in Danish hospi-tals, 1969–1974. Acta Pathologica etMicrobiologica. Scandinavica (Sect. B)85:143–152.
Rotter, M. 1988. Are models useful for testinghand antiseptics? Journal of Hospital Infec-tions 11 (Suppl. A):236–243.
Sanford, J.P. 1992. Forward. In: J.V. Bennett,and P.S. Brachman (eds.) Hospital Infec-tions Third Edition. Boston, MA. Little,Brown and Company.
Scott, L. June 13, 1994. Restraint on use of anti-biotics urged. Modern Healthcare, p. 47.
Shanz, S.J. Fall 1993. The emerging status ofpractice parameters. 7 Medical Staff Counsel31.
Siegel, J.D., G.H. McCracken Jr., N. Threlkeld,et al. 1980. Single-dose penicillin prophy-laxis against neonatal group B streptococcalinfections. A controlled trial in 18,738 new-born infants. New England Journal of Medi-cine 303:769–775.
Silber, J.L., S.M. Paul, G. Crane, et al. 1994.Influence of hospital antibiotic policy andusage on the incidence of vancomycin-resis-tant enterococcal (VRE) bacteremia.Abstract. Infection Control and HospitalEpidemiology 15(4):32.
Simmons, B., J. Bryant, K. Neiman, et al. 1990.The Role of Handwashing in Prevention ofEndemic Intensive Care Unit Infections.Infection Control and Hospital Epidemiol-ogy 11(11):589–594.
Spink, W.W., and V. Ferris. 1945. Quantitativeaction of penicillin inhibitor from penicillin-resistant strains of staphylococci. Science102:221.
St. Paul Fire & Marine Insurance Co. 1995. 1994Annual report to policyholders, pp. 4–5.
Strausbaugh, L.J., C. Jacobson, D.L. Sewell, etal. 1992. Antimicrobial therapy for methicil-lin-resistant Staphylococcus aureus coloni-zation in residents and staff of a veteransaffairs nursing home care unit. InfectionControl and Hospital Epidemiology13(3):151–159.
The Medical Letter on Drugs and Therapeutics.1994. Handbook of Antimicrobial Therapy.New Rochelle, NY. The Medical Letter, Inc.
U.S. Congress, Office of Technology Assess-ment. September 1995. Bringing HealthCare Online: The Role of Information Tech-nologies, Washington, DC. U.S. Govern-ment Printing Office.
U.S. Department of Public Health, Public HealthService, Agency for Health Care Policy andResearch. July 1994. Clinical PracticeGuideline Number 12: Otitis Media withEffusion in Young Children. AHCPR Pub.No. 94-0622. Rockville, MD.
U.S. Department of Public Health, Public HealthService, Centers for Disease Control. 1995.Morbidity and Mortality Weekly Report44:27.
Waldvogel, F.A. 1995. Staphylococcus aureus(including toxic shock syndrome). In: G.L.Mandell, J.E. Bennett, and R. Dolin (eds.)Mandel, Douglas and Bennett’s Principlesand Practice of Infectious Diseases Fourthedition. New York, NY. Churchill Living-stone, Inc., p. 1757.
Wenzel, R.P. 1992. Preoperative antibiotic pro-phylaxis. New England Journal of Medicine326(5):337–339.
Chapter 4 Antibiotic Use in Hospitals | 99
Westh, H., J.O. Jarlov, H. Kjersem, et al. 1992.The disappearance of multiresistant Staphy-lococcus aureus in Denmark: changes instrains of the 83A complex between 1969and 1989. Clinical Infectious Diseases14(6):1186–1194.
Willy, M.E., G.L. Dhillon, N.L. Loewen, et al.1990. Adverse exposures and universal pre-cautions practices among a group of highly
exposed health professionals. Infection Con-trol and Hospital Epidemiology 11(7):351–356.
Yu, V.L., G.P. Stoehr, R.C. Starling, et al.. 1991.Empiric antibiotic selection by physicians:evaluation of reasoning strategies. Ameri-can Journal of the Medical Sciences301:165–172.