471

Programmatic Role of the Infectious Diseases Physician in ControllingAntimicrobial Costs in the Hospital

Joseph F. John, Jr. and Neil 0. Fishman From the Division of Allergy, Immunology, and Infectious Diseases,Department of Medicine, Robert Wood Johnson Medical School,

University of Medicine and Dentistry of New Jersey, New Brunswick,New Jersey; and the Division of Infectious Diseases, University of

Pennsylvania School of Medicine, Philadelphia, Pennsylvania

Controlling antimicrobial costs has preoccupied infectious diseases physicians (IDPs). IDPs havecontrolled antimicrobial costs by the use of eight strategies: education, formulary restriction, phar-macy justification, formulary substitution, computer surveillance, laboratory item cost listing, pur-chase plans, and multidisciplinary approaches. Most strategies had input from IDPs and resultedin cost savings (up to $500,000 annually), particularly during the initiation periods. Educationalefforts were successful in reducing costs but needed continual intervention. Formulary restrictionwas the most straightforward cost-control mechanism. Restriction of "target antimicrobials" hasgiven way to "switch" therapy between expensive and less costly agents or between parenteral andoral regimens. Switch therapy is facilitated through the use of innovative order forms and on-linecomputer interaction. Computer surveillance has a capacity for interactive controls. Purchase plansmay give way to centralized pharmacy monitoring, a strategy that is attractive to managed careorganizations. Multidisciplinary antimicrobial management programs (AMPs) offer the best poten-tial for sustaining savings in antimicrobial costs. Ten recommendations lay a groundwork for IDPsto translate their expertise into leadership of AMPs.

Antimicrobic therapy saves thousands of lives and relieves much suffering, yet . . . untoward effects, harm and deathmay occur both after logical, but especially after indiscriminate, prescription. . . . The proper use of antimicrobicscan be attained by comprehension of their place, value and proper dosage in amount and time. Empiric andexperimental therapy is justified if properly controlled. The Hippocratic injunction "first do no harm" or the question"Is this drug really necessary? " are pertinent. Much needless expense, untoward effect, harm and disappointmentcan be prevented by better judgment in the use of antimicrobics for prophylaxis and therapy.

Hobart Reimann, M.D., 1961 [1]

Almost as soon as antibiotics became mainstream therapyin American medicine, warnings like the one of Reimann sur-faced in reputable journals. Now, more than 30 years later, therelentless rise in the use of antimicrobials has resulted in costsof >$7 billion annually in the United States with up to $4billion used for treatment of nosocomial infections due to anti-biotic-resistant bacteria [2]. Antimicrobials account for up to30% of hospital drug budgets [3-5]. Moreover, through thepersistent observations, studies, and publications of infectiousdiseases physicians (IDPs), it has been recognized for morethan three decades that up to 50% of antimicrobial usage inU.S. hospitals is inappropriate [1, 6, 7]. Thus, one major reason

This is one of a series of articles that focuses on the value of the infectiousdiseases specialist.

Reprints or correspondence: Dr. Joseph John, 356 MEB, 1 RWJ Place CN19,Robert Wood Johnson Medical School, New Brunswick, New Jersey 08903-0019.

Clinical Infectious Diseases 1997; 24:471-85© 1997 by The University of Chicago. All rights reserved.1058-4838/97/2403 — 0027$02 .00

for controlling antimicrobial use has been to reduce unneces-sary costs. The relative value of various strategies to reducethe hospital costs of antimicrobials is one major focus of thisreview.

Before dealing with that issue, however, it is worthwhilenoting that a second reason to control antimicrobial use is toreduce the selection of resistant microorganisms [2, 8]. Theimmense impact of antibiotic-resistant bacteria has recentlybeen summarized in a report from the now defunct Office ofTechnology Assessment that includes the problem of antibioticuse in hospitals [2]. Unfortunately, various strategies—includ-ing control of antibiotics applied in hospitals to prevent thespread of antibiotic resistance have not always resulted in in-creases in bacterial susceptibility to antimicrobials [9]. Becauseof the uncertain effectiveness of programs to contain emergingresistance, new strategies have been developed by a consensusgroup of the Society for Health Care Epidemiology of Americaand the Infectious Diseases Society of America (IDSA) [10].

In addition to rising concerns about antimicrobial resistance,there remains the impetus to reduce antimicrobial use in orderto control costs, particularly in hospitals on fixed budgets such

472

John and Fishman CID 1997;24 (March)

Table 1. Variables in studies of antimicrobial control programs.

• Antimicrobial agent(s) under consideration• Antimicrobial costs before and after implementation• Demographics of patient population• Demographics of prescriber population• Designers of program• Duration of baseline data collection• Grant support or funding source• Length of study• Method of implementation• Method of statistical analysis• Methods of tabulating costs• Number of effector personnel involved• Type of controls• Type of hospital• Ultimate effect of program• Use of rewards and punishments

as county and state hospitals or medical centers run by theU.S. Department of Veterans Affairs. Such pressures are alsoevident throughout the provinces of Canada. Furthermore, nowthat managed care organizations oversee a greater proportionof health care expenditures in the United States, there is evengreater fiscal pressure and perhaps larger financial rewardsto stockholders of for-profit health companies to limit antimi-crobial costs.

The objectives of this paper are to review the methodologyand the cost-effectiveness of antimicrobial control programs inthe United States and Canada, to determine the role of the IDPin designing and implementing these programs, and to makerecommendations that maximize the impact the IDP could haveon containing hospital antimicrobial costs.

Methods

A search of the headings of antimicrobial and antibioticcosts in the 1966-1995 MEDLINE and other databases at theUniversity of Medicine and Dentistry of New Jersey Librariesprovided references for review. Additional references were ob-tained from the personal reference libraries of the authors andfrom other articles culled from pharmacy and infectious dis-eases journals not appearing in on-line searches. Articles fromthe United States and Canada were reviewed for the characteris-tics of the control program, the specific antimicrobials underinvestigation, and the economic impact of the program afterimplementation. Specific variables considered in the review ofspecific articles are listed in table 1.

Programs were divided into one of eight types, A throughH, as shown in table 2. There was frequent overlap of strategies.For example, a hospital may have chosen to limit access togiven agents (method B, formulary restriction) but to also offerphysicians a chance to justify their choices to an IDP or apharmacy specialist (method C, pharmacy justification). It wasdifficult to separate the effect of the respective method or the

interactive effect one method had on another. In the reviewthat follows, we provide an overview of each method followedby specific details of studies that fell within a given methodol-ogy.

A. Education of Prescribers

Overview. Although education of antimicrobial prescribersis the most basic and possibly the most long-standing methodused to improve utilization of antibiotics, this technique is theleast rigorously studied. Historically, IDPs have continuouslyattempted to instruct their colleagues on the appropriate useof antimicrobial chemotherapy. Nevertheless, the significantdeficiencies in physicians' knowledge of appropriate prescrip-tion of antimicrobials [11] remain an apt target for education.Education emerges as the most palatable strategy and shouldbe the cornerstone of any antibiotic control program, but thereis little agreement in the literature as to what constitutes anoptimum program [12].

Educational programs can take two basic forms. One in-volves direct interaction with prescribers, either through discus-sion about the antibiotic order or via a more formal educationalprogram. A second employs utilization review with delayedfeedback to prescribers regarding their pattern of prescribing.

Formal educational programs designed to alter antimicrobialprescribing patterns have taken many forms, including staffconferences, lectures by outside authorities, audiovisual pack-ages, clinical pharmacy consultations, drug utilization evalua-tions, hospital-pharmacy committee newsletters, disseminationof independent sources of information, and, most recently, thedevelopment of clinical guidelines or pathways [13]. The mostuseful instructional sessions appear to be one-on-one instruc-

Table 2. Strategies to control hospital antibiotic costs.

A. Education* of prescribers1. Direct interaction2. Peer feedback from performance evaluation

a. Verbalb. Written

B. Antimicrobial formulary restrictionC. Pharmacy justification

1. Without consultation with infectious diseases physiciana. Antibiotic order formb. Stop orders

2. With consultation with infectious diseases physiciana. Direct interactionb. Simple chart entry

D. Formulary substitution or switchE. Computer surveillanceF. Clinical microbiology laboratory item costingG. Purchase plansH. Multidisciplinary approach

* Considered as education when there is human interaction over choice ofthe antimicrobial and the recommended change is not necessarily mandatory.

CID 1997; 24 (March) ID Physicians and Antimicrobial Costs 473

tion by an antibiotic utilization expert. Utilization review seemsto be less useful; the process is removed by time and spacefrom the original antimicrobial orders.

The impact of any educational program is difficult to assessbecause of the complex nature of educational variables. More-over, the diverse nature of educational efforts, the lack of stan-dardized feedback, and the complex interaction of emotional,philosophical, and microbiological (pathogenic) factors thatcontribute to physicians' decision-making processes regardingpatients with infectious diseases make formal assessment andcomparison of these interventions quite arduous.

Several attempts have been made over the years to documentthe effect of educational programs on antimicrobial prescribingpatterns, even though attempts to use education to influenceother physician behavior, such as the ordering of laboratorytests, showed little effect [14]. Physicians still lack informationabout antimicrobials [11, 15]. A recent study showed that doc-tors know less about antibiotics than they do about nonsteroidalantiinflammatory drugs and antihypertensive agents. Yet, phy-sicians seem more interested than ever in increasing theirknowledge about antimicrobials [15]. In any case, few studieshave reported the effect of education per se on altering antimi-crobial costs.

Specific studies. Most early studies revealed that educationitself may have a marginal effect in reducing antimicrobialprescribing. Once education efforts are removed, their effectis short-lived [16, 17]. Just when new parenteral and oral cepha-losporins were arriving on the market in the early 1980s, severalstudies tried to reduce the frequency of use of these agents. Inreducing the use of cephalexin via phone consultation, Selig-man [18] showed a decrease in antibiotic costs of 29% at amunicipal hospital in Brooklyn, New York.

Mary Ma, a clinical pharmacist working at the large WestLos Angeles Veterans Affairs Medical Center (VAMC), in LosAngeles, implemented an educational system as early as 1975[19]. She reviewed all incoming requests for antimicrobialsusing an order form that had to be completed by the prescriber.Use of 15 agents required the approval of an IDP. The pharma-cist reviewed all other orders for "potential problems," invok-ing "consultation with the Infectious Disease Section whenevernecessary." Though the data presented in this overview articleare scanty, the cost benefits were stated to be $67,000 for the1975-1976 academic year.

Other VAMC programs sought to broaden the educationalbasis for change, as suggested earlier [17]. One of the mosteffective programs utilized several strategies that were adaptedto local conditions and were supported by the hospital staff.Before concurrent audits were performed, specific trends weredetermined for antimicrobial usage, susceptibility data, andcosts [20]. After these trends were determined, a utilizationcontrol program was implemented in a stepwise fashion andincluded a defined hospital formulary, reporting of susceptibil-ity data, use of an antibiotic order form, and mandatory consul-tation in defined circumstances. If the first reviewer considered

antimicrobial use to be inappropriate, an IDP further reviewedthe case and made recommendations to the prescriber for alter-native therapy.

At the relatively small Wichita VAMC (Wichita, KS), afterthe fourth quarter costs in 1982 for parenteral antimicrobialagents had risen to $28,100, application of this system produceda decline to an average of $16,500 per quarter over the nextseven quarters. Educational feedback from use of an antibioticorder form, review of antimicrobial requests, and eliminationof 11 0-lactams and netilmicin provided the bulk of the savings.Further cost reduction was obtained by requiring consultationwith an IDP for use of seven additional agents: amikacin, cefo-nicid, cefoperazone, high-dose cefotaxime, cefoxitin, andpiperacillin.

Interaction with prescribing physicians has worked in a 261-bed community teaching hospital in Philadelphia. A patientcard used to record the antibiotic presently being used wasreviewed for appropriateness by an antibiotic team includingan IDP [21], who conveyed any need for change to the houseofficer. The effect of the program was measured in a short 11-week study of 164 therapeutic regimens. Overall, 49 of 78recommended changes were followed, resulting in a savingsof $9,759 that would be annualized to $291,486. Only one-sixth of the proposed changes were not followed. Conductingthe program as outlined took an estimated 5 hours per weekor about $177 per hour based on the amount of savings. Ulti-mately, it was believed that continuity, comprehensiveness,and concurrence of the team as well as house staff educationaccounted for the program's success.

Some educational approaches utilizing a format of retrospec-tive peer review of antimicrobial usage have failed to showcost-effectiveness. A prominent study using feedback from uti-lization review of antibiotic use was done in 1985 at BostonUniversity Hospital (Boston) [22]. Physicians in the upper 50percentiles of antibiotic usage were sent letters each month.During the study, the mean antibiotic cost for surgical patientswas $655 compared to a cost of $1,943 for nonsurgical patients(P < .001). The hematology and oncology departments hadthe highest costs of usage. More than 80% of the costs weregenerated by 30% of the prescribers. When expenditures werecompared quarter to quarter, costs did not decrease after feed-back was given. It is important to note that this program tar-geted only attending physicians, not house staff physicians,did not involve IDPs, and had no additional educational inter-vention.

Another team approach to reducing antimicrobic costs wasused at Monroe Community Hospital in Rochester, New York.The team consisted of a clinical pharmacist, the infection con-trol nurse, and the director of laboratory services but no IDP[23]. The physicians on an acute care ward and the team agreedto what would be considered appropriate use of antimicrobialsduring the study. In fact, the team in essence was the clinicalpharmacist who conducted interactive, educational ward roundswith the physicians on the study ward. The study involved only

474 John and Fishman CID 1997;24 (March)

126 patients over 6 months, yet savings were $29,400, for aprojected annual savings of $58,827 (58.7%) for that wardalone.

Hartford Hospital (Hartford, CT) sought to simplify the com-plexities of and thus economize on parenteral and combinationtherapy by using a concept termed "streamlining" [24]. Aninnovative group (the Department of Pharmacy Services andthe Division of Infectious Diseases) placed in the patient'schart an Antibiotic Audit Form ("not part of the medical re-cord") stating "We recommend the following changes in yourpatient: xxxx. Although these recommendations are not manda-tory, if significant differences of opinion exist, we suggestobtaining an Infectious Disease consult."

During the 7-month study, —70% of patients required rec-ommendations for streamlining, i.e., a switch from expensiveagents to less expensive and/or oral agents. At a large facilitylike Hartford Hospital, this translated to 133-340 interventionsper month (after the initial month). Almost 83% of the recom-mendations were accepted by the attending physicians. Inter-ventions decreased from 98% at the start of the study to 54%,suggesting that the educational efforts were working. In termsof cost-effectiveness, nearly $40,000 was saved in the initial7-month study period. The projected annual savings leveledoff at —$100,000. A postimplementation data analysis furtherprojected that $32.46 per patient would be saved for annualsavings of $107,637.

B. Antimicrobial Formulary Restriction

Overview. Control of the hospital formulary is the most di-rect method of influencing antimicrobial utilization and costs.Limiting the available agents on formulary portends an effec-tive means of prohibiting the use of newer, more expensiveantibiotics in place of older, equally effective agents [25]. Al-though such an approach seemed heavy-handed originally [26],most well-structured formularies today meet the therapeuticneeds of most medical staffs [27]. The economic impact ofthis change has not been well studied until the current decade.

In their 1988 Guidelines for Improving the Use of Antimicro-bial Agents in Hospitals [6], the IDSA advanced the followingrecommendations for formulary development, part of whosestated purpose was to control costs: (1) restriction of the formu-lary to the minimum number of agents required for effectivetherapy; (2) elimination of duplicate agents within a givenclass; (3) consideration of susceptibility patterns of nosocomialpathogens; (4) restriction of certain agents on the basis ofspecial indications, toxicity, or excess costs; and (5) periodicreview.

The authors advocated the establishment of a multidiscipli-nary "antimicrobial agents team" led by an IDP to optimizeuse of antiinfectives and to consider the hospital antimicrobialformulary as a dynamic, not a static, document. Investigatorsin Ireland have also advocated a similar interactive system thatthey have shown can decrease costs [28].

Specific studies. One of the first studies to test the hypothesisthat formulary restriction per se would limit costs was reportedin 1985 [29]. Cefotaxime was first available on open formularyfor 12 months. Thereafter, its use during an 18-month periodof restriction showed an actual increase, 85% of which wasdeemed appropriate by published guidelines [30, 31]. The easyavailability of cefotaxime as a nonformulary agent probablyinfluenced the outcome of this study. Thus, restriction of indi-vidual formulary items will not necessarily result in decreasedcosts for the drug unless it is unavailable through nonformularyordering, as later studies were to show [32]. Hartford Hospitalhas been able to develop steps for establishing an antibioticformulary after years of demonstrating the beneficial effect ofan IDP's input with the help of clinical pharmacists in alteringantimicrobial use [24], particularly through antibiotic stream-lining [25].

At the Dayton VAMC (Dayton, Ohio), because of escalatingcosts for antimicrobial prophylaxis, Scher and colleagues [33]started a restriction program in October 1986 by limiting useof cephalosporins to cefazolin and cefotetan. A special formwas required to order any other antibiotic, and all prophylacticantimicrobials were to be automatically discontinued 24 hoursafter the initial dose. The study ran from January to September1987. Although surgical class I and II cases increased in num-ber during the study, cephalosporin doses greatly decreased(from 947 g to 577 g), resulting in a saving of $32,163.84 or$15.36 per case in the 2,094 cases studied. The authors believedthat educational efforts to change surgeons' minds had notchanged their practices. Although surgeons predicted "direconsequences," there was actually a slight decrease in the rateof wound sepsis.

Woodward and colleagues from St. Louis [34] published amajor study in 1987 to address the effect of restrictions oncost savings and the quality of antibiotic therapy. Tobramycin,cephalosporins (except cefazolin), and vancomycin were ini-tially restricted. When compared to costs during the nonrestric-tion period, the cost savings were $2.61 per antibiotic day(P < .0046), $34,597 per month for the restricted agents and$24,620 (P < .03) per month for all antibiotics. Inappropriateuse of antibiotics before 1984 was due to use of an unnecessar-ily expensive agent 37% of the time compared to only 2% ofthe time after July 1985. There was no decrease in the qualityof outcome for infected or bacteremic patients.

The program took 24 h/mo of an infectious diseases fellow'stime, 76 h/mo for the pharmacy, and 10 h/mo for the infectiousdiseases faculty, but costs for these services were not deductedfrom savings in antimicrobial use. This study was the firstcomprehensive, statistically based study to demonstrate thatrestricting specific agents produced cost savings with no de-crease in the quality of patient care.

Himmelberg et al. [35] published in 1991 an interestingstudy based on formulary restriction. The University of NorthCarolina hospitals had a long-standing policy for restriction ofantibiotics, requiring an infectious diseases fellow first to ap-

CID 1997; 24 (March) ID Physicians and Antimicrobial Costs 475

prove use of an antibiotic and next to notify the pharmacy ofthe approval. The restriction policies were removed from Julyto December 1988. So, the potential value of this study wasto analyze the effect of removal of restriction policies.

As for expenditures after restriction removal, total drug costsfor restricted agents increased 103%, from $154,542 to$313,905 for an added cost of $159,363. The cost of imipenemincreased the most ($44,423 to $126,332). The costs of un-restricted drugs were much smaller and did not change appreci-ably ($84,061 to $80,659), which implied that unnecessary useof restricted agents was the cause of the increase in costs, notsubstitution of drugs.

Cefoperozone and cefotaxime were little used before or afterremoval of restriction policies, so they were not retained onthe restriction list. Ceftazidime, ceftriaxone, and imipenem ac-counted for —75% of usage before and after restriction. Theappropriateness of use of ceftazidime and imipenem was stud-ied in 42 patients. There was a statistically significant differencein appropriate use in only a few of the eight categories usedfor analysis.

Since this study was published, we found no additional ana-lyses of the effect of strict formulary restriction on cost. Ineffect, on the basis of the work of Himmelberg's group andearlier studies, it is likely that many hospitals actually haveimplemented an approach of antimicrobial restriction at theformulary level because it works to reduce costs.

However, there are recent clinical trials that indicate that theuse of an expensive agent as monotherapy may be less costlythan a combination of agents needed to offset the spectrum ofthe more expensive single-agent regimen [36]. Specifically,in a comparative trial of imipenem versus clindamycin plusaminoglycoside for treatment of pneumonia, the cost of imi-penem was $507 per case versus $813 for the combination.The cost of drug preparation and delivery of the two agentsplus pharmacokinetic monitoring made the combination ther-apy more expensive. Recent studies of single daily dose therapywith aminoglycosides suggest that the costs of pharmacokineticmonitoring could be reduced or eliminated [37, 38].

C. Pharmacy Justification

Overview. This section will focus on restriction policies thatrequire approval of a designated physician or clinical pharma-cist for the use of certain agents. Several innovative methodsof justification have been utilized to control antibiotic usage.These are manifested in a variety of restriction and approvalmechanisms, such as direct communication with the pharmacy,antibiotic order forms (written justification), implementation ofcontrol categories, and automatic stop orders. These processesmay or may not involve an IDP's added approval or consulta-tion. These strategies are probably the most onerous to prescrib-ing physicians, yet they stand as some of the more effectivemethods of controlling the antimicrobial budget.

Specific studies. The landmark study of McGowan and Fin-land [39] first used justification by prescribers and approval byan IDP to limit use of most systemic antimicrobials. Formalconsultation was in general not required, and it seems thatmost approvals were given by phone. The numbers of requests,approvals, and denials were not recorded. Costs were not re-corded, but doses of individual agents were charted. For ampi-cillin, nafcillin, and cloxacillin, restrictions were removed in1968 or 1969 and usage increased. Chloramphenicol restrictionresulted in decreases in annual usage from 20,000 g to 3,000g. This study was the first to use a novel, direct interventionby IDPs: a method as simple as making a phone call, whendone regularly, had a large impact on antimicrobial usage.

Methods similar to those of McGowan and Finland wereused several years later at the Coney Island Hospital (Brooklyn,NY) [40]. An IDP considered requests by phone from housestaff or attending physicians. If there was disagreement, a con-sultation with the infectious diseases department would takeplace. This was one of the first studies to show actual reductionin costs—$131,826 in the first year and about the same amount($113,742) in the third year compared to baseline figures(1975). By year two, costs had decreased by 38%.

As in later studies, there was a concomitant educationalprogram presented by the IDP. Besides the reduction in costby the educational program, the study also reduced all typesof inappropriate use of antibiotics [31]; the study's success wassimilar to that of an earlier attempt to improve appropriatenessusing education alone [17]. Instead of using telephones, otherearly studies proved that an antibiotic prescription form couldalter the use of empirical and prophylactic antimicrobials [41].

A study by Coleman et al. from the Palo Alto VAMC (PaloAlto, CA) [42] used pharmacy justification and is widelyquoted, primarily because of their efforts to critically analyzeantibiotic expenditures associated with control programs inlong-term studies. In fact, the Stanford University IDPs initi-ated the study since they believed that the standard controlmeasures previously used had only limited success. They thusconducted a long "pre—study period" from March 1986 toAugust 1987 to gather baseline cost data. Next, four restrictedusage categories were developed that included cephalosporins,penicillins, aminoglycosides, and miscellaneous agents. Amaximum iv dose was assigned to seven agents. An InfectiousDiseases Surveillance (IDS) team established all these policiesand then enforced them for 26 months. The key differencefrom the old system was the mandate that the IDPs had toreview the plan for usage prior to pharmacy distribution andfollow through with prospective review of the cases.

Significant reductions of $7,600 per month translated into ayearly savings of $91,200. The IDS team altered therapy in35.4% of 1,769 reviews. The doses and cost of antibioticsdecreased for the diagnosis-related groups of pneumonia(P < .05), urinary tract infection, and septicemia but actuallyincreased for cellulitis. The cost per treatment day, the cost peradmission, the number of doses, and the percentage of the

476 John and Fishman CID 1997;24 (March)

reimbursement rate as set by the U.S. Health Care and Financ-ing Administration all were significantly different from thosein the pre—study period (for all, P < .05). The length of therapywas not different in the two study periods. For 11 restricteddrugs, pre- and post-restriction comparisons showed significantdecreases in cefoxitin and clindamycin costs and increases forticarcillin/clavulanate and cefotetan costs. For "special re-view" categories, there was a 42% rate of intervention by theInfectious Diseases Service. This involvement added 50-70patients per month to the IDS workload in large part man-aged by a clinical pharmacist, who eventually became part ofthe IDS team.

Educational feedback through target drug programs can alsocontrol the costs of individual agents. A good example was astudy in Indianapolis at Indiana University Hospitals in 1988aimed at reducing the costs of imipenem by implementingapproved indications for the drug's use [43]. After review ofthe patient's chart, the pharmacy would advise an alternativedrug or dose modification. If the attending physician disap-proved, an infectious disease consultation was sought. Duringthe 9-month period of implementation of the target drug pro-gram, intervention resulted in savings of $6,033.

Enforcing a restricting policy can be difficult, particularlyin hospitals that have rotating house staffs. Ben Taub GeneralHospital in Houston is a municipal hospital serving the poor.The drug budget at the hospital increased from $11.4 millionin 1990 to $16.9 million in 1992, with antimicrobials account-ing for 35% of the budget [44]. A new policy was approvedthat required an attending IDP' s approval for use of any ofeight intravenous antimicrobials. A clinical pharmacist special-ist (CSP) program was developed to enforce the program,which officially began on 1 January 1994.

After 6 months of a similarly designed pilot program at anaffiliated hospital, cost savings were $75,000. Ben Taub Gen-eral Hospital then adopted the program and realized savings inthe first 2 months of >$82,000 on restricted agents, thoughthis figure was counterbalanced by increases of the same mag-nitude in the costs of nonrestricted agents. Therefore, a nonre-stricted agent (aztreonam) was reclassified as restricted. Itsaddition to the list of restricted agents resulted in a cost avoid-ance for all antibiotics of $300,235 in 9 months. The investiga-tors believed that acceptance of the program by medical physi-cians developed more easily than with surgeons, but overallacceptance was better than the CSP had anticipated.

D. Formulary Substitution or Switch

Overview. Even though formulary restriction and justifica-tion appear to be effective means of regulating antimicrobialuse and costs, such policies place certain physicians in a polic-ing role that is viewed frequently as being punitive [45]. Inthe private practice setting, IDPs are not enthusiastic aboutantibiotic restriction policies for fear that this perception oftheir role will damage their traditional referral patterns [20].

Formulary substitution or "switch" therapy appears less dicta-torial and may avoid nonproductive disagreements but is lesslikely to be enforced as strictly as a formulary restriction.

Two types of switch therapies have been used: the substitu-tion of a member of a class of agents for a "therapeutic equiva-lent" and the change to a different class of agents, usuallywith the aim of employing a less expensive oral regimen fora costly parenteral one. The latter type of switch therapy hasbecome more popular in the past 5 to 7 years with the advent ofbroad-spectrum oral cephalosporins and quinolones. Statisticalanalysis of the cost-effectiveness of these studies is difficultsince the spectrum of coverage is usually narrowed after theswitch.

Specific studies. Some switch strategies are agent specific.Achusim [46] applied a cefoxitin-to-cefotetan switch, pre-dicting that the change would save $100,000. Indeed, in a 1-year study, savings were $106,059 on the cephalosporin aloneand $259,281 in total cost savings for the switch. Concomitantdecreases in costs occurred for tobramycin (43%), cefamandole(34%), cefoperazone (31%), cefoxitin (93%), ampicillin (18%),mezlocillin (40%), piperacillin (36%), and ticarcillin (32%),whereas increases in costs were seen with ceftriaxone (33%),aztreonam (32%), cefotetan (313%), oxacillin (48%), clinda-mycin (31%), and metronidazole (52%). IDP input may havetaken place at the pharmacy and therapeutics committee level,but there was no active IDP input during the study.

. Physicians at times have chosen to perform a clinical trialin order to determine the wisdom of formulary substitution.Edwards et al. [47] sought to substantiate the apparent superior-ity of cefamandole over cefazolin in perioperative prophylaxis.From August 1990 to May 1992, cefazolin or cefamandole wasgiven to 893 patients at the time of anesthesia induction, duringsurgery, and 24 hours postoperatively. There was no differencein infection rates between the two groups. There was a 2.8times greater charge for cefamandole than cefazolin that wouldtranslate into a $95,000 annual savings if cefazolin were usedas the sole prophylaxis. Cefazolin remained the agent of choicefor perioperative vascular prophylaxis.

Cooperative studies have also been used to determine thevalue of switch therapy. Grasela et al. [48] formed a DrugSurveillance Network of 78 pharmacists from 54 institutionsand enrolled 766 patients with infections serious enough torequire at least a 7-day course of parenteral antimicrobial ther-apy. The pharmacists encouraged a switch from a parenteralregimen to oral ciprofloxacin. Of 766 patients discharged, 496continued ciprofloxacin therapy. The savings in Red Bookwholesale costs from not using the iv therapy for 16,732 unuseddoses were $187,146.50 or $274.81 per patient.

Ehrenkranz et al. [49] used an established group of hospitalscalled the South Florida Infection Control Consortium to studythe effect of switching from parenteral antimicrobial therapyto comparable oral regimens as treatment for pulmonary infec-tions. Nurse "emissaries" at three Dade County hospitals,closely supervised by an IDP, picked a total of 82 patients, of

CID 1997;24 (March) ID Physicians and Antimicrobial Costs 477

which 53 were cases and 29 were controls. An interesting thirdgroup termed noncompliant because they would not enterthe study also were analyzed for outcome. There were 32parenteral regimens used, with oral substitutes proscribed bywritten guidelines. The control group had an average cost of$1,801, and the noncompliant group had a cost of $2,505, whilethe group of study patients averaged $917. The savings of $884per patient in comparison with the control group were duemostly to a shortened mean length of stay of 2.4 days. Patientsof compliant and noncompliant physicians had no differencein outcome.

The approach of Ehrenkranz et al. has been supported insubsequent studies. In three different trials, Ramirez [50] stud-ied the effect of switch therapy in 150 patients with community-acquired pneumonia in 1994. Patient outcome for switch ther-apy was as favorable as that for conventional therapy. In 1994the total hospital savings were projected to be $119,947, includ-ing, interestingly, cost avoidance of $3,707 for "line sepsis."

An intensive switch program for clindamyin and metronida-zole was devised by pharmacists at Vancouver General Hospi-tal (Vancouver, British Columbia, Canada) [51]. After a base-line period of study in 1988, each of the next four study yearsshowed a higher percentage in the use of oral clindamycin(44%) and of oral metronidazole (79%). Cumulative costs from1988 through 1991 were $31,920 for metronidazole and$53,880 for clindamycin.

The investigators' program involved the timely use of stick-ers applied to the front of patients' medical charts indicatingthe cost of the agents involved in treatment. Their success withclindamycin and metronidazole encouraged them to expand theprogram to fluconazole, cefuroxime, and ciprofloxacin.

Okpara et al. [52] at a general hospital in Houston negotiateda collaborative iv to oral switch among several departments.Parenteral therapy was usually switched to oral ciprofloxacinor oral amoxicillin/clavulanate. In the first 2 months of theprogram, physicians accepted intervention in 84.6% of cases,resulting in a savings of $12,935.

Another Canadian group composed of pharmacists and IDPstargeted clindamycin [53]. A baseline study revealed that36.4% of clindamycin use was determined to be inappropriate.During the subsequent intervention period, 76 cases neededintervention, resulting in a cost avoidance of $9,722 or $15,798annualized. Empirical use was the most common reason forthe need for intervention.

A secondary effect was a dramatic decline in clindamycin usagecosts of $41,000 and only a $3,400 increase in the use of metroni-dazole. The authors emphasize that these so-called Targeted DrugMonitoring (TDM) programs will succeed only with the strongsupport of hospital administration and medical staff. TDM alsoshows the impact that a single antimicrobial agent can have ona hospital budget. Before TDM of clindamycin, this agent aloneaccounted for $160,000 of the 1987-1989 budget.

Altering dosing strategies for other agents can be a basisfor other cost benefits. At Hartford Hospital, Quintiliani and

colleagues [54] have shown in a crossover study that a doseof 4 g of piperacillin every 8 hours has preferable pharmacody-namics to a dose of 5 g of mezlocillin given every 8 hours.The theoretical cost savings would be about $17,000 if thisregimen of piperacillin were used instead of mezlocillin.

E. Computer Surveillance

Overview. The computer may be the ultimate method forantibiotic surveillance and education. Many hospitals now usean order entry system but lack an antimicrobial cost-contain-ment software program. The unique aspect of computerizedentry of agents like antimicrobials is the opportunity for instan-taneous feedback, education, and a seamless alteration in thepreference for certain agent(s) [55]. Interaction with the com-puter departs so radically from earlier written and even oralfeedback that it becomes a so-called automated antibiotic con-sultant [56, 57].

At the LDS Hospital in Salt Lake City, the computer displayfor a patient-specific antibiotic option lists susceptibility resultsfor the past 5 years and the past 6 months as well as a cost perantibiotic for a 24-hour course of treatment [55]. The automatedantibiotic consultant was able to select appropriate therapy 94%of the time compared to the laboratory susceptibility tests.Moreover, the average cost per day ($41.08) for the antibioticsuggested by the computer was significantly less (P < .01)than the cost of the antibiotics actually prescribed ($51.93).

Specific studies. One computerized ordering program hasbeen applied at the University of Utah for monitoring the dura-tion of prophylaxis in surgical patients [58]. After a baselineanalysis using the HELP (Health Evaluation through LogicalProcessing) Hospital Information System, a study was per-formed to determine the effect of stop orders triggered bycomputer analysis of duration of prophylaxis.

According to the baseline year of study, 40% of surgical pa-tients received prophylaxis for a longer duration than necessary.In the intervention year, 35% of patients undergoing surgery stillreceived prophylaxis for too long a period; however, the durationof therapy was shorter (19 doses versus 13 doses, respectively).In year 2 the computer identified 1,061 patients with excessprophylaxis for a potential savings of $44,562. The investigatorssuggest that we are just beginning to realize the capabilities ofa system like HELP to alter antibiotic prescribing.

Antibiotic utilization review can be performed efficientlywhen a computerized ordering system forms the basis of thedata base. This approach has been described for an integratedprogram at the three major teaching hospitals of the Universityof Colorado Medical School [59]. Prescribers used a computer-assisted antibiotic order-entry program to select each inpatientantibiotic and to designate the purpose for use. After the entry,the computer informs the physician if a restricted antimicrobialhas been ordered. There are also programmed dosing selectionschosen by an IDP and a pharmacist. Combining the orderingdata base with in vitro susceptibility information from the clini-

478 John and Fishman CID 1997;24 (March)

cal microbiology laboratories allows for rapid feedback to pre-scribers. Susceptibilities are reported only for the least expen-sive antibiotics. Cost analysis of such a system should berelatively easy on the basis of preliminary reports [57].

F. Clinical Microbiology Laboratory Antibiotic Cost Listing

Several papers suggest the potential economy of antimicro-bial cost listing through susceptibility reports from the clinicalmicrobiology laboratory [20, 59]. One such program at theDenver VAMC has a computerized feedback system that places"on line" the costs of the least expensive, most clinicallyrelevant agents [59]. The cost-effectiveness of making theselistings available to prescribers has not been published.

G. Purchase Plans

Although many medical centers participate in pharmaceuti-cal purchase plans, to our knowledge there are few papersthat describe the cost-effectiveness of purchase plans and theireffect on the antimicrobial budgets of participating institutions.One pharmacy management company, Owen Healthcare inHouston, provides a wide variety of pharmacy services to —27C1hospitals in 41 states [60]. The intended purpose of these ser-vices is to "affect physician prescribing of antimicrobialagents" by use of several methods. Using their analysis of120 hospitals, for example, Owen executives concluded thattreatment with cefotaxime instead of ceftriaxone would save$15.27 per course and should be the workhorse third-generationcephalosporin in their hospitals. Similar switch strategy to cefo-taxime from ceftriaxone in a 300-bed hospital can save up to$30,190 per year [61].

Owen Healthcare also can track expensive items at individualhospitals and report to the hospital the cost of antibiotics perpatient day (for example, a monthly range of $8.22 to $11.41).Managed care companies, especially those that eventually con-trol or own more than one hospital, will undoubtedly makeuse of pharmacy management companies that can provide ornegotiate drug purchase plans with pharmaceutical companies.

H. Multidisciplinary Control Programs

Overview. Evident from our review thus far is the fact thatmany programs have a central methodology but in fact employseveral different strategies to achieve control of antimicrobialuse. These programs are usually termed multidisciplinary sincethey involve input from various hospital services includinginfectious diseases, clinical pharmacy, infection control, nurs-ing, and particularly administration. Almost 20 years ago,Counts [62] suggested a collaborative "multidisciplinary anti-microbial utilization committee" much along the lines of thosethat were developed. Although several of the programs de-scribed earlier were applied in a multidisciplinary fashion [20,34, 42], we will describe those comprehensive antimicrobial

management programs (AMPs; used synonomously with anti-microbial management teams) that included extensive analysesof cost-effectiveness.

Specific studies. At the University of Minnesota, expendi-tures for antiinfectives were sharply increasing in the 1980s,from a total of $961,518 for parenteral antibiotics and antifun-gals in 1981 to $2,907,071 for the same agents in 1988 [63].In 1984, 1985, and 1986, the pharmacy and therapeutics com-mittee offered three alternatives to prescribers that were rein-forced by written and verbal follow-up. The changes involvedsubstituting the combination of gentamicin and nafcillin forthe combination of vancomycin and tobramycin; using metroni-dazole instead of clindamycin and cefoxitin; and extending theinterval between doses to 8 hours whenever "cefazolin every6 hours" was ordered. No data are given as to the acceptanceof the program, but expenditures in 1985 through 1988 werebelieved to be reduced by $400,000 per year, assuming anunrestricted rate of escalation of drug use.

Another multidisciplinary program utilized comprehensivehospitalwide antimicrobial management at two hospitals (Mil-lard Fillmore Hospitals) in Buffalo over a 10-year period, cul-minating in publication of a highly detailed program of control[26]. At these hospitals, costs had been rising —15% annually.A formulary established in 1984 had little effect on costs from1984 through 1986. An antimicrobial order sheet failed to havean impact between 1986 and 1988 [64]. The new programinvolved the input of "clinical pharmacists working in conjunc-tion with infectious disease personnel."

There were two major components of the system. The firstwas based on consultation by the IDPs for empirical therapy,focusing on cases of "unreasonable empiricism." The secondwas a streamlining program [24] that was actively implementedby a full-time Pharm.D. antimicrobial specialist in each of thetwo hospitals. Streamlining involved simplifying the antimicro-bial regimen, switching to oral antimicrobials (usually ci-profloxacin) early in the course of therapy, or actually shorten-ing therapy, for example, from 10 to 5 or 7 days for nosocomialpneumonia. Culture and susceptibility data were reviewed onday 3 with the aid of on-line computer interaction between theclinical microbiology laboratory and the pharmacy. The authorstried to determine if the program could pay for itself.

In the first 266 patients reviewed, dosage adjustment wasthe most frequent change (40%), followed by discontinuationof antibiotics (18%), switch to oral agents (17%), change inregimen (14%), or diversion to a clinical trial protocol (11%).Actual cost savings in June—December 1989 were $47,340 thatannualize to $81,155. Antimicrobial expenditures during thattime fell to 20.2% of a total budget of $3.6 million.

A Canadian multidisciplinary hospital-based program hasbeen developed at Henderson General Hospital in Hamilton,Ontario [5]. Three major strategies were used: (1) automatictherapeutic interchanges whereby nine automatic substitutionscould replace an order for one of eight specified antimicrobials;(2) Antimicrobial Restriction Policy, in which indications were

CID 1997;24 (March) ID Physicians and Antimicrobial Costs 479

developed by the Infectious Diseases Division and wereapproved by the pharmacy, and then for only 3 days; and (3)conversion from parenteral to oral antimicrobials in which asticker was applied to the patient's chart advising of the change.

Before the institution of the program (1991), the antimicro-bial budget of $1,539,553 was 41.6% of the total drug budgetof $3,697,017. This percentage steadily declined over the next4 years until it reached 28% or $918,301 for antimicrobials.Strategy 2, using restriction, resulted in the largest cost reduc-tions followed by those resulting from use of strategy 1. Theprogram of substituting oral antimicrobials was nonproductiveat first and resulted in a change in only 27% of cases. After apharmacist was assigned specifically to monitor the program,75% of the recommendations were followed, resulting in anannual savings of $50,000. The program overall produced sav-ings of —$46,000 per month or a reduction in the mean dailycost per dose from $11.88 to $10.16 (Canadian dollars).

Table 3 summarizes cost savings or cost avoidance in thestudies we have reviewed. Of those studies listed, 79% includedIDPs in some fashion in the control program. Annualized costsavings with or without inclusion of cost avoidance estimatesranged from $7,440 for a single drug target to $483,032 for acomprehensive program.

Quality assurance or improvement has not routinely beenmeasured in studies of antimicrobial costs. Others have pro-posed that traditional quality assurance approaches simply havenot worked and that the policy of identifying outliers was,in retrospect, punitive [45]. Jewell et al. with their proposedcontinuous quality improvement (CQI) team actually offer acommercial service that applies a system of CQI to existingcontrol programs [20, 26]. This CQI model includes an infec-tion control nurse and an IDP as part of an eight-person team.The concept has already reduced costs (up to $12,000 perquarter) by lowering rates of cardiac surgery infection,but results have not been reported for antimicrobial costcontrol [45].

Discussion

In this era of return to primary care, subspecialists need toprovide evidence that their nonpatient care activities remaincost-effective. The goal of this review was to document therole of the IDP in programs to control nosocomial antimicrobialcosts. Our review clearly provides data to support the conclu-sion that IDPs have been necessary, even crucial, to the successof cost-effective antimicrobial control programs, but their inputalone is not sufficient. We believe that there is a need for newrecommendations to insure the continued participation of oursubspecialty in controlling antibiotic costs.

Descriptions of many types of strategies and programs tocontrol antimicrobial use have been published [1, 70, 71]. Sinceinvestigators typically desire to publish useful and positivedata, nearly all of the published pharmacoeconomic analysesof antimicrobial control programs in our review describe a

beneficial effect, all of which suggests that programs lackingsuch an effect are underrepresented. Recently, the field of phar-macoeconomics has burgeoned. The number of papers in thefield heretofore concentrated in medical journals hasbeenexpanding rapidly, and they now appear in pharmacy and healtheconomic journals; their quality has been improving as well[72]. As a result, our review attempted to integrate data pub-lished in all these fields.

In the large representative sample of articles, we found thatan IDP was involved —85% of the time in either planning orexecuting an antimicrobial control program. The IDP fre-quently relied on the cooperation of pharmacy staff for success.Most programs at their inception tended to report large savings;some approached a half-million dollars, not so surprising sinceearlier studies had shown that --.50% of antimicrobial usagewas inappropriate. Savings generally plateaued after the firstyear or two [6]; nonetheless, input by IDPs remained crucialfor adjusting to changes in nosocomial flora, antimicrobial sus-ceptibility patterns, and the addition of new antimicrobialagents to the formulary.

Direct educational efforts, usually by the IDP, or programsbased on utilization review either failed or needed to be rein-forced constantly in order to maintain any impact [27]. Eventu-ally, administrators realized that comprehensive programs hadthe greatest chance of success and that the IDP figured promi-nently into almost all successful programs [20].

Comprehensive programs, those that used four or more strat-egies to control costs, usually featured multidisciplinary teams.The members of teams learned that by working together theycould implement multiple strategies and obtain the best eco-nomic success [26, 68]. For example, an AMP at HartfordHospital emphasizes formulary streamlining reinforced by for-mulary restriction and review, antibiotic order forms, and edu-cational efforts [25]. Within the team framework [73], IDPshave the broadest expertise to direct multidisciplinary teamsthat use a multifaceted approach to control prescription of anti-biotics [74]. The structure of these teams often centers arounda quality improvement hub and is designed to address a broadrange of nosocomial problems besides antimicrobial use andcost, such as outcome analysis and hospital epidemiology.

A study benchmarking the activity in streamlining found thatmany hospitals still have no activity in this area and could benefitgreatly from the cost savings reported with the use of AMPs [75].Infectious diseases training programs should educate prospectiveIDPs to perform the tasks of a team leader, including participationin the growing field of outcomes research [76].

How should hospitals tackle the problem of antimicrobialcontrol? First, it is necessary for a hospital administration torealize the multiple benefits of new strategies such as streamlin-ing to curb antibiotic costs [75]. Next, a hospital should desig-nate leaders to choose and to implement strategies that willwork in their institution.

As an example, one of us (N.O.F.) has devised a multidisci-plinary program much like several of those described in our

Focus of program

Imipenem target drug programAntibiotic use following therapeutic

interchange programAntibiotic streamliningCost-effectiveness of continuous

parenteral antibiotic therapyEffect of formulary restriction on

cefotaxime usageOrder forms and prescription patternsCefamandole vs. cefazolin prophylaxis:

cost-effectivenessIntervention to discontinue parenteral

antibiotic therapy in patients withpulmonary infections

Computer surveillance of totalantimicrobial use

Antibiotic use patterns and expendituresover 7 years in a university hospital

Intravenous to oral step-down programfor clindamycin at VancouverGeneral Hospital

Targeted Drug Monitoring forclindamycin cost avoidance

Oral ciprofloxacin as follow-up to ivciprofloxacin in 54 institutions

Home iv .antibiotic therapy in HMOcompared to affiliated hospital costs

All-inclusive antibiotic reviewEffect of removing formulary restriction

policy on use of antimicrobial drugsat university hospital

An early comprehensive controlprogram performed in three phases

Team approach to reduce antibiotic costImipenem vs. clindamycin/

aminoglycoside for LRITargeted multidisciplinary switch

therapy from ceftriaxone andcefotetan to cefotaxime andcefmetazole, respectively

Prototype study of required antibioticorder form

Modify residents' selection ofantimicrobial agents

Challenge of increasing antibiotic costs $67,000using an antibiotic order form

Enforcing a policy for restricting $300,325 $400,433antibiotic use

Controlling costs of antibiotic resistance NAat VAMC

Intravenous to oral conversion for $12,934/mo $77,604select antimicrobials in 580-bedcommunity hospital

Nonvalue of retrospective peer 0comparison feedback in containingantibiotic costs

Graduated comprehensive utilization $11,200/quarter $44,800program to control costs at VAMC

Savings of studySavings

annualized

$6,033/9 mo $7,440$259,281

$107,637$7,600 $91,200

NG

NG$95,000

$884/patient $50,000

$10.85/patient/24 h

$406,133

$85,800

$37,600

$187,146

$116/d

$61,660/11 w $291,486$159,463

$483,032

$58,000$306/patient

$52,873

NA

$16.29/order

480 John and Fishman CID 1997;24 (March)

Table 3. Economic outcome of different types of antibiotic control programs.

First author[reference]

Yearprogram

beganIDPinput Strategy*

Abel [43] 1991 No DAchusim [46] 1992 No D

Briceland [24] 1988 Yes AColeman [42] 1991 Yes C2

DiVito [29] 1985 Yes Cl

Echols [64] 1984 Yes ClaEdwards [47] 1993 Yes H

Ehrenkranz [49] 1992 Yes D

Evans [55] 1994 Yes E

Fletcher [63] 1990 Yes H

Frighetto [51] 1992 No D

Gin [53] 1994 Yes D

Grasela [48] 1991 No D

Grizzard [65] 1991 Yes D

Heineman [21] 1986 Yes AHimmelberg [35] 1991 Yes B

Hirschman [66] 1988 Yes H

Karki [23] 1990 No AKreter [36] 1992 No D

Lee [67] 1995 Yes H

McGowan [39] 1974 Yes C2a

MacCosbe [68] 1985 Yes Al

Ma [19] 1984 Yes Cl

Maswoswe [44] 1995 Yes C2

North [59] 1993 Yes E, F

Okpara [52] 1995 Yes D

Parrino [22] 1989 Yes A2

Pelletier [20] 1985 Yes H

CID 1997;24 (March) ID Physicians and Antimicrobial Costs 481

Table 3. (Continued)

First author[reference]

Yearprogram

beganIDPinput Strategy* Focus of program Savings of study

Savingsannualized

Quintiliani [54] 1993 Yes Cost-effective dosing method forpiperacillin

$16,435

Rawlings [61] 1995 No Automatic switch from ceftriaxone tocefotaxime in 1993

$11.75/patient/d $47,997

Recco [40] 1979 Yes C2a Early study of control in municipalhospital using IDP interaction

$122,784

Salama [5] 1996 Yes Integrative program in Canadianteaching hospital

$27,590/mo $331,080

Schentag [26] 1993 Yes Change in antibiotic usage resultingfrom interaction with CP, IDP, andmicrobiology laboratory

$200,000

Scher [33] 1990 Yes Reducing costs of prophylaxis bylimiting number of cephalosporins

$32,163

Seligman [18] 1981 Yes A Value of phone call to IDP to limitcephalexin use

$303,939

Soumerai [69] 1993 No C 1 a Encourage use of gentamicin instead oftobramycin

$2,340/8 w $15,210

Woodward [34] 1987 Yes B, Cl Cost savings from formulary restrictionand physician monitoring in medicalschool—affiliated hospital

$24,620/mo $295,440

NOTE. CP = clinical pharmacist; HMO = health maintenance organization; IDP = infectious diseases physician; LRI = lower respiratory tract infection;NA = not available; NG = not given; VAMC = Veterans Affairs Medical Center.

* See table 2 for type of method used in the antimicrobial control program.

review [20, 26, 45]. Strategies already adopted include a re-stricted formulary, published treatment guidelines, an approvalprocess for restricted agents, an antibiotic streamlining pro-gram, and formulary designations based on hospital susceptibil-ity patterns and costs. Implementation involves competitivebidding between therapeutic equivalents and dosing restrictionsbased on pharmacodynamic properties [77]. Published guide-lines serve as the basis for all approvals or denials of restrictedagents. Utilization review and outcome data are complied andpresented at quarterly hospital quality assurance meetings andto the hospital administration. Although it is difficult to deter-mine the relative effect of individual elements, in over 3 yearsthe program has proven to be very cost-effective, saving anaverage of $385,000 annually.

How are antibiotic control teams to be financed? Bryan [74]has urged hospitals to employ the expertise of IDPs "for theexplicit purpose of formulating and implementing antimicrobialtherapy guidelines." Our review demonstrated that cost savingscan often offset the budget of an AMP. Smaller hospitals with-out full-time IDPs may have to depend on smaller teams toconduct their programs and may have to pay IDP consultantson a time basis for review of antimicrobial control activities.

How successful has this strategy been? Only one institutionhas reported creation of new salary lines for IDPs to manageantimicrobial use [78]. From personal communications, weknow of several other lines for IDPs that have been created.Despite the paucity of published reports of success in obtaining

salary lines, we still strongly advise those IDPs who aspire toconduct antimicrobial control programs for their hospitals toproceed along specified tracks. To do this, they must first assessthe programs on hand for cost-effectiveness and quality andthen meet with administrators and justify the economic valueof their proposed services [79]. The more inappropriate usagethat exists in a center, the more initial impact an antimicrobialcontrol program will have. After initial successes, IDPs mustdemonstrate their ability to sustain cost-avoidance expenses[66].

As prospective AMP team leaders, IDPs must understandwhat we would term the "antibiotic profile" of their hospitals:the amount of antimicrobials used, their costs, the distributionof those costs among specialties, the percentage spent on pro-phylactic agents, the use of buying cooperatives, the power ofthe pharmacy and therapeutics committee in formulary devel-opment, and the willingness of prescribers to learn and changetheir habits. IDPs need to prepare themselves for their ex-panding and changing roles as facilitators of antimicrobial con-trol teams in order to justify a charge for these services.

As more hospitals fall under the ownership or control ofmanaged care organizations (MCOs) [80], it will be importantto understand the goals of these companies. The focus of MCOswill not be limited to the inpatient setting inasmuch as capitatedcare will pursue savings on all pharmaceuticals, particularlyantimicrobials for outpatients [65]. It may be reasonable to askthe MCOs to supply the funds to create a multidisciplinary

482 John and Fishman CID 1997;24 (March)

Figure 1. Participation in an anti-microbial management programwith the infectious diseases physi-cian as the hub and director. Note thebidirectional interaction (4-) impliedbetween the member/functions andthe director that are shaded in gray.P & T = pharmacy and therapeutics.

AMP. MCOs may wish to develop their own teams that serveseveral hospitals. MCOs will be in a better position than smallerhospitals to provide the initial capital outlay and may also owna large centralized data processing center that can analyze thedata in a timely fashion. Formularies already published bylarge pharmaceutical claims companies, some insuring up to59,000,000 lives, should be increasingly attractive to MCOs[81]. These mammoth consortiums may be approached by IDPsto develop incentive programs based on expenditures for bothoutpatient and inpatient drugs and devices used to treat infec-tions [79].

The cost savings priorities of MCOs, while increasing thepotential value of the IDP, may either meet unexpected barriersor become springboards for improving public health [82]. Bar-riers may come in the form of regulations by the state, localdemographics, or cultural beliefs [83]. Older MCOs, like thoserun by the U.S. Department of Veterans Affairs and KaiserPermanente have, in general, allowed their facilities to formu-late their own methodology locally to control antibiotic costs[20]. Similarly, the Canadian National Health Service has en-couraged individual hospitals to develop and implement controlpractices [5]. Whether the barriers to the growth of managed

care will reduce cost savings in newer for-profit MCOs remainsspeculative [83].

MCOs may have a new role in prevention that could includemanagement of antimicrobial use [84]. In this changing healthcare scenario, IDPs should become aware of MCO initiativesand position their hospital to benefit from resources availablethrough MCOs that would possibly control but certainly wouldinfluence antimicrobial use. Perhaps some MCOs may alsoprovide research funding to explore these issues.

Computers are an emerging new tool in the control of anti-microbials use. It remains to be shown if computers will consis-tently have an impact on antimicrobial costs similar to thoseof AMPs. Early reports are encouraging [55, 56, 58, 85]. Thegroup at LDS Hospital has recently published the results ofa 7-year antibiotic management program [86]. The programdeveloped a set of computer-assisted guidelines based on localclinician consensus to manage all antibiotic agents used in theinstitution. Overall antibiotic costs and adjusted costs per pa-tient were reduced drastically from a high of $122.66 in 1988to $51.90 in 1994.

It is likely that computers will become an increasingly im-portant part of AMPs, leaving for future studies the issues of

CID 1997;24 (March)

ID Physicians and Antimicrobial Costs 483

Table 4. Steps for infectious diseases physicians to apply for estab-lishment of an antimicrobial management program.

1. Develop knowledge about the local antimicrobial budget andpatterns of usage.

2. Collect the necessary data to publish a quarterly antimicrobialsusceptibility report including the prices of all formularyantimicrobials.

3. Gather baseline usage and cost data and attempt to comparethem to those of equivalent institutions [60].

4. Outline a structure for participation in the AMP team andestimate the annual costs for funding, including salary for theIDP.

5. Meet with administrators of hospitals and managed carecompanies to discuss implementation [79].

6. Focus on the most frequently used and most costly agents onthe formulary.

7. Publish a manual, Guidelines for Antimicrobial Use, and provideappropriate educational or multimedia formats for educating theprescribing staff.

8. Ensure a fail-safe mechanism to resolve disagreements with theprescribing staff

9. Develop innovative educational methods including computerizedinterplay to explain the use of new antimicrobials, antivirals,and antiinfectious biologics.

10. Reevaluate the AMP annually with quality assurance personneland review the cost-effectiveness with the chief of staff, thepresident of the hospital, and the Vice President for ClinicalAffairs.

NOTE. AMP = antimicrobial management program; IDP = infectiousdiseases physician.

the time required by human members of the team to formulateand implement policy. The developments in computerized in-formational data bases and Internet interactions make it manda-tory that IDPs understand the capabilities of hospital computersto help implement antimicrobial control programs. One WorldWide Web site managed by the University of Wisconsin Medi-cal School at Froedfert Hospital (Milwaukee) provides an anti-microbial usage guide (http://www.intmed.mcw.edu/Antibiot-icGuide.html) . AMPs may also choose to develop their ownhome pages or link to available university and hospital Websites [87].

The IDP interested in providing reimbursable leadership,in the control of nosocomial antimicrobial costs should un-derstand that such a function can be usurped by other healthcare providers, including pharmacists and quality care ad-ministrators. The recommendations in table 4 are a series ofsteps for IDPs to translate their expertise in antimicrobialusage into leadership of AMPs. In our view, IDPs are bestequipped to address cost-containment issues and need tomobilize colleagues in pharmacy, nursing, and quality man-agement to present a comprehensive and cohesive programto the hospital administrators that will appeal to the majorMCOs in their region [2, 88] (figure 1). Specific algorithmstermed critical pathways or clinical practice guidelines havebeen published that may serve as a checklist and road map

for contemporary implementation of antimicrobial controlprograms [32, 89].

References

1. Reimann HA. The misuse of antibiotics. Med Clin North Am 1961;45:849 54.

2. Office of Technology Assessment USC. Impacts of antibiotic-resistantbacteria. Washington, D.C.: Government Printing Office, 1995.

3. Avorn J, Soumerai SB, Taylor W, Wessels MR, Janousek J, Weiner M.Reduction of incorrect antibiotic dosing through a structured educationalorder form. Arch Intern Med 1988;148:1720-4.

4. Baum C, Kennedy DK, Knapp DE, et al. Drug utilization in the UnitedStates, 1985. DHHS. Rockville, MD: Food and Drug Administration/National Center for Drugs and Biologics, 1986.

5. Salama S, Roststein C, Mandell L. A multi-disciplinary hospital-basedantimicrobial use program: impact on hospital pharmacy expendituresand drug use. Canadian Journal of Infectious Diseases 1996; 7:104-9.

6. Man JJ, Moffet HL, Kunin CM. Guidelines for improving the use ofantimicrobial agents in hospitals: a statement by the Infectious DiseasesSociety of America. J Infect Dis 1987;157:869-76.

7. Kislak JW, Eickhoff TC, Finland M. Hospital-acquired infections andantibiotic usage in the Boston City Hospital-January, 1964. N Engl JMed 1964; 271:834-5.

8. Cohen ML. Epidemiology of drug resistance: implications for a post-antimicrobial era. Science 1992;257:1050-5.

9. McGowan JE Jr. Do intensive hospital antibiotic control programs preventthe spread of antibiotic resistance? Infect Control Hosp Epidemiol 1994;15:478-83.

10. Shlaes DM, Gerding D, John JF, et al. Guidelines for the prevention ofantimicrobial resistance in hospitals. Infect Control Hosp Epidemiol1997 (in press).

11. Neu HC, Howrey SP. Testing the physician's knowledge of antibiotic use:self-assessment and learning via videotape. N Engl J Med 1975;293:1291-5.

12. Bryan CS. Strategies to improve antibiotic use. Infect Dis Clin North Am1989;3:723-34.

13. Kunin CM, Johansen KS, Worning AM, Daschner FD. Report of a sympo-sium on use and abuse of antibiotics worldwide. Rev Infect Dis 1990;12:12-9.

14. Schroeder SA, Myers LP, McPhee SJ, et al. The failure of physicianeducation as a cost containment strategy: report of a prospective con-trolled trial at a university hospital. JAMA 1984; 252:225-30.

15. Barclay LP, Hation RH, Doering PL, Shands JW. Physicians' perceptionsand knowledge of drug costs: results of a survey. Formulary 1995; 30:268-79.

16. Gilbert DN, Eubanks NM, Jackson JM. The effects of monitoring the useof gentamicin in a community hospital. J Med Educ 1978; 53:129-34.

17. Jones SR, Pannell J, Barks J, et al. The effect of an educational programupon hospital antibiotic use. Am J Med Sci 1977;273:79-85.

18. Seligman SJ. Reduction in antibiotic costs by restricting use of an oralcephalosporin. Am J Med 1981; 71:941-4.

19. Ma MY. Meeting the challenge of rising antibiotic costs. VA Practitioner1984:63-5.

20. Pelletier LL Jr. Hospital usage of parenteral antimicrobial agents: a gra-dated utilization review and cost containment program. Infect Control1985; 6:226-30.

21. Heineman HS, Watt VS. All-inclusive concurrent antibiotic usage review:a way to reduce misuse without formal controls. Infect Control 1986;7:168-71.

22. Parrino TA. The nonvalue of retrospective peer comparison feedback incontaining hospital antibiotic costs. Am J Med 1989; 86:442-8.

23. Karki SD, Holden JM, Mariano E. A team approach to reduce antibioticcosts. DICP 1990;24:202-5.

484 John and Fishman CID 1997; 24 (March)

24. Briceland LL, Nightingale CH, Quintiliani R, Cooper BW, Smith KS. 49. Ehrenkranz NJ, Nerenberg DE, Shultz JM, Slater KC. Intervention toAntibiotic streamlining from combination therapy to monotherapy uti- discontinue parenteral antimicrobial therapy in patients hospitalizedlizing an interdisciplinary approach. Arch Intern Med 1988;148: with pulmonary infections: effect on shortening patient stay. Infect Con-2019-22. trol Hosp Epidemiol 1992;13:21-32.

25. Crowe HM, Quintiliani R. Antibiotic formulary selection. Med Clin North 50. Ramirez JA. Switch therapy in community-acquired pneumonia. DiagnAm 1995;79:463-76. Microbiol Infect Dis 1995; 22:219-23.

26. Schentag JJ, Ballow CH, Fritz AL, et al. Changes in antimicrobial agent 51. Frighetto L, Nickoloff D, Martinusen SM, Mamdani FS, Jewesson PJ.usage resulting from interactions among clinical pharmacy, the infec- Intravenous-to-oral stepdown program: four years of experience in atious disease division, and the microbiology laboratory. Diagn Microbiol

large teaching hospital. Ann Pharmacother 1992;26:1447-51.

Infect Dis 1993;16:255-64. 52. Okpara AU, Maswoswe JJ, Stewart K. Criteria-based antimicrobial IV to27. Henderson P, Crowe HM. Antibiotic streamlining-scope, format, and

oral conversion program. Formulary 1995; 30:343-8.

benefits. Conn Med 1989;53:423-4. 53. Gin AS, Lipinski LA, Honcharik N. Impact of a target drug monitoring28. Feely J, Chan R, Cocoman L, Mulpeter K, O'Connor P. Hospital formular- program on the usage of clindamycin. Can J Hosp Pharm 1994; 47:

ies: need for continuous intervention. BMJ 1990; 300:28-30. 53-8.29. DeVito JM, John JF. Effect of formulary restriction of cefotaxime usage. 54. Quintiliani R, Nightingale CH, Sullivan MC. Use of pharmacodynamic

Arch Intern Med 1985; 145:1053-6. concepts in developing a cost-effective dosing method for piperacillin.30. Jogerst GJ, Dippe SE. Antibiotic use among medical specialties in a com- Clin Ther 1993; 15(suppl A):44-9.

munity hospital. JAMA 1981;245:842-6. 55. Evans RS, Classen DC, Pestotnik SL, Lundsgaarde HP, Burke JP. Improv-31. Kunin CM, Tupasi T, Craig WA. Use of antibiotics: a brief exposition of

ing empiric antibiotic selection using computer decision support. Arch

the problem and some tentative solutions. Ann Intern Med 1973; 79: Intern Med 1994;154:878-84.

555-60. 56. Evans RS, Pestotnik SL, Classen DC, Burke JP. Development of an auto-32. Dunagan WG, Medoff G. Formulary control of antimicrobial usage. What mated antibiotic consultant. MD Comput 1993;10:17-22.

price freedom? Diagn Microbiol Infect Dis 1993; 16:265-74. 57. Kephart PA, Fueiner M, Hibbard T, Carver E, Center VAM. Computer-33. Scher KS, Bernstein JM, Arenstein GL, Sorensen C. Reducing the cost of

aided antibiotic selection. P&T 1992;17:451, 452, 455-7.

surgical prophylaxis. Am Surg 1990; 56:32-5. 58. Evans RS, Pestotnik SL, Burke JP, Gardner RM, Larsen RA, Classen DC.34. Woodward RS, Medoff G, Smith MD, Gray JL III. Antibiotic cost savings Reducing the duration of prophylactic antibiotic use through computer

from formulary restrictions and physician monitoring in a medical- monitoring of surgical patients. DICP 1990;24:351-4.school-affiliated hospital. Am J Med 1987; 83:817-23. 59. North D. Controlling the costs of antibiotic resistance. Clin Ther 1993;

35. Himmelberg CJ, Pleasants RA, Weber DJ, et al. Use of antimicrobial

15(suppl A):3 -11.drugs in adults before and after removal of a restriction policy. Am J 60. Roark MK, Reed WE Jr. Econotherapeutics. Diagn Microbiol Infect DisHosp Pharm 1991;48:1220-7. 1995;22:209-17.

36. Kreter B. Cost-analysis of imipenem-cilastatin monotherapy compared .61. Rawlings RD. Therapeutic exchange of cefotaxime for ceftriaxone: evalua-with clindamycin plus aminoglycoside combination therapy for treat- tion, implementation, and subsequent cost savings at a 300-bed commu-ment of serious lower respiratory, intra-abdominal, gynecologic, and

nity hospital. Diagn Microbiol Infect Dis 1995;22:235-7.

urinary tract infections. Clin Ther 1992;14:110-21. 62. Counts GW. Review and control of antimicrobial usage in hospitalized37. Gilbert DN. Once-daily aminoglycoside therapy. Antimicrob Agents patients: a recommended collaborative approach. JAMA 1977; 238:

Chemoth 1991;35:399-405. 2170-2.38. Preston S, Briceland LL. Single daily dosing of aminoglycosides. Pharma- 63. Fletcher CV, Metzler D, Borchardt-Phelps P, Rodman JH. Patterns of

cotherapy 1995; 15:297-316. antibiotic use and expenditures during 7 years at a university hospital.39. McGowan JE Jr, Finland M. Usage of antibiotics in a general hospital: Pharmacotherapy 1990; 10:199 -204.

effect of requiring justification. J Infect Dis 1974;130:165-8. 64. Echols RM, Kowalsky SF. The use of an antibiotic order form for antibiotic40. Recco RA, Gladstone JL, Friedman SA, Gerken EH. Antibiotic control in utilization review: influence on physicians' prescribing patterns. J Infect

a municipal hospital. JAMA 1979;241:2283-6. Dis 1984; 150:803-7.41. Durbin WA Jr, Lapidas B, Goldmann DA. Improved antibiotic usage 65. Grizzard MB, Harris G, Karns H. Use of outpatient parenteral antibiotic

following introduction of a novel prescription system. JAMA 1981; 246: therapy in a health maintenance organization. Rev Infect Dis 1991;1796-1800. 13(suppl 2):S174-9.

42. Coleman RW, Rodondi LC, Kaubisch S, Granzella NB, O'Hanley PD. 66. Hirschman SZ, Meyers BR, Bradbury K, Mehl B, Gendleman S, KimelblattCost-effectiveness of prospective and continuous parenteral antibiotic B. Use of antimicrobial agents in a university teaching hospital: evolu-control: experience at the Palo Alto Veterans Affairs Medical Center tion of a comprehensive control program. Arch Intern Med 1988; 148:from 1987 to 1989. Am J Med 1991;90:439-44. 2001 -7.

43. Abel SR, Guba EA. Evaluation of an imipenemicilastatin target drug pro- 67. Lee J, Carlson JA, Chamberlain MA. A team approach to hospital formu-gram. DICP 1991;25:348-50. lary replacement. Diagn Microbiol Infect Dis 1995;22:239-42.

44. Maswoswe JJ, Okpara AU. Enforcing a policy for restricting antimicrobial 68. MacCosbe PE, Gartenberg G. Modifying empiric antibiotic prescribing:drug use. Am J Health System Pharm 1995; 52:1433-5. experience with one strategy in a medical residency program. Hosp

45. Jewell MA. Quantitative methods for quality improvement. Pharmacother- Formul 1985; 20:986-92.apy 1995; 15:27S-32S. 69. Soumerai SB, Avorn J, Taylor WC, Wessels M, Maher D, Hawley SL.

46. Achusim LE. Antibiotic use following implementation of a therapeutic Improving choice of prescribed antibiotics through concurrent remindersinterchange program. P&T 1992:775-98. in an educational order form. Med Care 1993; 31:552-8.

47. Edwards WH Jr, Kaiser AB, Tapper S, et al. Cefamandole versus cefazolin 70. Jawetz E. Current concepts in therapy. N Engl J Med 1958;258:747-8.in vascular surgical wound infection prophylaxis: cost-effectiveness and 71. Lepper LH. The use and misuse of antibiotics and other chemotherapeuticrisk factors. J Vasc Surg 1993;18:470-6. agents. Med Clin North Amer 1961; 45:1663-76.

48. Grasela TH Jr, Paladino JA, Schentag JJ, et al. Clinical and economic 72. Bradley CA, Iskedjian M, Lanctot KL, et al. Quality assessment of eco-impact of oral ciprofloxacin as follow-up to parenteral antibiotics. DICP nomic evaluations in selected pharmacy, medical, and health economics1991; 25:857-62. journals. Ann Pharmacother 1995; 29:681-9.

CID 1997;24 (March) ID Physicians and Antimicrobial Costs 485

73. Sherer JL. Tapping into teams. Hosp Health Netw 1995; 5:32-6.74. Bryan CS. The role of the infectious diseases physician in setting guide-

lines for antimicrobial use. Bull NY Acad Med 1987; 63:627-37.75. Schentag JJ, Paladino JA, Birmingham MC, Zimmer G, Carr JR, Hanson

SC. Use of benchmarking techniques to justify the evolution of antibioticmanagement programs in healthcare systems. J Pharm Technol 1995;11:203-10.

76. Bootman JL. Pharmacoeconomics and outcomes research. Am J HealthSystem Pharm 1995; 52(suppl 3): S16 -9.

77. Nightingale CH, Quintiliani R, Nicolau DP. Intelligent dosing of antimi-crobial. In: Remington JS, Swartz MN, eds. Curr Clin Top Infect Dis1994;14:252-65, Boston: Blackwell Scientific.

78. Lee CK, Glenn DJ. Cefotaxime and ceftriaxone use evaluation in pediat-rics: considerations of cost effectiveness. Diagn Microbiol Infect Dis1995;22:231-3.

79. Stern Z, Altholz J. The hospital director and the quality circle. Clin PerformQual Health Care 1995;3:41-4.

80. Columbia/HCA buys Tulane Hospital and Clinic. Journal of InvestigativeMedicine 1995; 43:101-4.

81. Browne RA, Mattson CN. PCS clinical formulary and prescribing guide-lines 1995-1996. Scottsdale, AZ, 1995.

82. Swartz MN. Hospital-acquired infections: diseases with increasingly lim-ited therapies. Proc Natl Acad Sci USA 1994; 91:2420-7.

83. Ermann D, Richmond J. Managed care arrangements: barriers to costsavings potential. Med Care Rev 1994; 51:2:125-48.

84. Centers for Disease Control and Prevention. Prevention and managed care:opportunities for managed care organizations, purchasers of health care,and public health agencies. MMWR Morb Mortal Wkly Rep 1995;44(RR-14):1 -12.

85. Evans RS, Larsen RA, Burke JP. Computer surveillance of hospital-ac-quired infections and antibiotic use. JAMA 1986; 256:1007-1.

86. Pestotnik SL, Classen DC, Evans RS, Burke JP. Implementing antibioticpractice guidelines through computer-assisted decision support: clinicaland financial outcomes. Ann Intern Med 1996;124:884-90.

87. Kramer JM, Cath A. Medical resources and the Internet. Making theconnection. Arch Intern Med 1996;156:833-42.

88. Schulman KA, Glick HA, Rubin H, Eisenberg JM. Cost-effectivenessof HA-1A monoclonal antibody for gram-negative sepsis: economicassessment of a new therapeutic agent. JAMA 1991;266:3466-71.

89. Ellrodt AG, Conner L, Riedinger M, Weingarten S. Measuring and improv-ing physician compliance with clinical practice guidelines: a controlledinterventional trial. Ann Intern Med 1995;122:277-82.