The Use of an Antibiotic Order Form for Antibiotic Utilization Review: Influence on Physicians' Prescribing Patterns

Transcription

1 THE JOURNAL OF INFECTIOUS DISEASES. VOL. 150, NO.6. DECEMBER by The University of Chicago. All rights reserved /84/ $01.00 The Use of an Antibiotic Order Form for Antibiotic Utilization Review: Influence on Physicians' Prescribing Patterns Roger M. Echols and Steven F. Kowalsky From the Department of Medicine, Division of Infectious Diseases, Albany Medical College; and the Department of Clinical Science and Practice, Albany College of Pharmacy, Albany, New York An antibiotic order form was implemented for all inpatient antibiotic orders at an 800 bed hospital in April 1981 to provide an ongoing, concurrent audit of antibiotic use. The prescribing physician provided the clinical indication for the antibiotic order, and individual patient treatment courses were identified. During the 25-month study period, cephalosporins, penicillin plus ampicillin, and aminoglycosides accounted for 44070, 22%, and 17% of all treatment courses, respectively. Sixty-nine percent of firstgeneration cephalosporin treatment courses were for prophylaxis, whereas the remaining antibiotics were used for either empirical therapy or documented infection in 56%-79% of cases. After the introduction of the antibiotic order form, there was a significant decline in both the number of antibiotic treatment courses (P =.025) and the percentage of patients receiving any antibiotic (P =.007). We conclude that a specialized antibiotic order form is an effective method for antibiotic utilization review and can have a significant impact on a physician's prescribing patterns. In recent years, issues of quality assurance and cost control have often focused on the use of antibiotics. Indeed, the Joint Commission for Accreditation of Hospitals currently requires hospitals to continually review antibiotic use, not only to determine what antibiotics are being prescribed, but also to determine the appropriateness of antibiotic selection and dosing [1, 2]. Several previous reviews of antibiotic use have demonstrated frequent, inappropriate selection or dosing with antimicrobial agents [3-12]. All of these studies were either retrospective studies of randomly selected charts or short-term prevalence studies. Besides the laborintensive nature of chart review studies, audits of this type require inference by the reviewer to explain why the antibiotic was prescribed. In an extensive study of twenty general hospitals, Townsend et al. [8] determined that while patient's charts were reliable records for analyzing patterns of antibiotic use, they usually did not include a clear reason for prescribing the antibiotic. They concluded that the cost of such antibiotic utilization reviews Received for publication April 23, 1984, and in revised form June 26, Please address requests for reprints to Dr. Roger M. Echols, Division of Infectious Diseases, ME-424, Albany Medical College, Albany, New York are "prohibitiveunless thenecessaryinformationhas been recorded for easy retrieval." Morerecently, Moss et al. [14] interviewed each prescribing physician within 48 hr ofthe antibiotic order to determine why the antibiotic was given. Most antibiotic treatment courses were administered without identification or even suspicion ofa specific pathogen. Frequent empirical use of antibiotics may explain why patient's records do not explain the reason a particular antibiotic is administered. An antibiotic audit by itself cannot be expected to alter physician's prescribing patterns [15, 16]. Other measures, such as formulary restriction or the requirement ofspecific treatment approval by a pharmacist or infectious disease clinician, have been successful in altering patterns of antibiotic use [16-19J. Such restrictive policies, however, are often met with resistance on the part of practicing physicians and cannot be considered universally applicable to all hospitals. Recently, Durbin et al. [20J described the use of a prescription form used exclusively for antibiotic orders in a pediatric medical center. Although their study provided concurrent utilization data and stimulated improved use of antibiotics for surgical prophylaxis, it was limited to two wards and was discontinued after three months. Subsequently, a modified antibiotic ordering system was instituted at our 803

2 804 Echols and Kowalsky 800-bed medical center [21]. This computerenhanced system has provided concurrent information on all antibiotics prescribed for inpatients at our hospital since April 1, We believe this concurrent reviewprocess has had a significant impact on the use of antibiotics. Materials and Methods The Albany Medical Center Hospital is an 800-bed teaching hospital with a housestaff of 350 and an attending medical and surgical staffofan additional 600. On April 1, 1981, a preprinted Antibiotic Order Sheet (AOS) was instituted for all inpatient antibiotic orders. The AOS was included in each patient's chart on their admission to the hospital and was used in place of the regular physician order form. Each AOS was used only once, although more than one antibiotic could be prescribed simultaneously. The AOS required the physician to identify one of five clinical indications for the antibiotic prescription [21]: (1)surgical prophylaxis (SP); (2) empirical therapy (ET); (3) documented infection (DI); (4) nonsurgical prophylaxis (NSP); and (5) other (OTH). Space was provided for the physician to indicate the type of surgical procedure in the case of SP or the suspected site and causative pathogen(s) ofinfection when the indication was for treatment (either ET or DI). Incorporated into the AOS system was an automatic discontinuance after 48 hr if surgical prophylaxis was identified. All other indications required renewal after five days. After each order was processed by the pharmacy, the AOS was entered as a data file in a computerized collection system (Apple II Plus [48K]; Apple Table 1. Computer Inc., Cupertino, Calif). Data input included patient identification number, order date, antibiotic prescription (including dosage and interval), administration route, the service ordering the drug, and the clinical indication. All antibiotic orders, whether completed on the AOS or written on a regular physician order form, were entered. Since no indication would be provided on a regular physician order form, these were counted as "noncompliant" orders. The informational unit provided by the program was labeled an Antibiotic Treatment Course (ATC), which identified a specific patient-antibiotic combination. A patient receiving combination therapy, such as gentamicin and clindamycin, would generate two separate ATCs. A patient treated with ampicillin iv and subsequently treated with ampicillin orally would have one ATC. The program eliminated duplicate orders that included reordering the same drug for purposes of dosage change or antibiotic renewal. Since the combined indications of SP, ET, and DI accounted for >85% of all ATCs for each antibiotic group, we divided the indication data into either prophylaxis (SP plus NSP) or treatment (ET plus DI). Weclassified antibiotics into sevenseparate groups based on their patterns of clinical usage. These groups, designated sentinel antibiotics, constituted (\)90% of all ATCs and were all parenterally administered agents. The groups include the following: (1) first-generation cephalosporins; (2) second- and third-generation cephalosporins; (3) penicillin and ampicillin; (4) aminoglycosides (gentamicin, tobramycin, and amikacin); (5) antistaphylococcal drugs (oxacillin, nafcillin, and vancomycin); (6) antianaerobe drugs (clindamycin, chloramphenicol, and Change in distribution and number of sentinel antibiotics prescribed over a 25-month period. Distribution/ no. of ATCs Group no. Sentinel antibiotic Mean usage (070) Range (0J0) Trend P value st generation cephalosporins 2nd and 3rd generation cephalosporins Penicillin and ampicillin Aminoglycosides Antistaphylococcal Antianaerobe Antipseudomonal /~ ti~ ~/~ /~.../ / / /.041 NOTE. = no change; ATC = antibiotic treatment course.

3 Antibiotic Order Forms 805 metronidazole); and (7) antipseudomonal penicillins (carbenicillin, ticarcillin, mezlocillin, and piperacillin). These antibiotics- with the exception ofgroup 2 antibiotics, mezlocillin, and piperacillin - were formularly drugs throughout the study period. Cefoxitin became a formularly drug in October Monthly reports included a summary ofall ATCs listed by drug and subdivided by indication. An identical format was provided for each of 18clinical services. A separate report listing all ATCs detailed the dosage, frequency interval, indication, and clinical service. Calculation ofthe monthly prevalence ofantibiotic use was based on the number of patients receiving one or more antibiotics, including nonsentinel antibiotics, compared with hospital census data. For calculationofsentinel antibioticatcs, the number of patients receiving any antibiotic was multiplied by the ratio of total sentinel ATCsto total ATCs. Monthly reports were submitted to the Infection Control Committee for discussion. Certain antibiotics, because of their frequent misuse or high cost, were subjected to special scrutiny. Remedial measures generated by the Committee were forwarded to the hospital administration for implementation. Statistical results were obtained by a least squares regression analysis. P values ~.05 were considered statistically significant. Results Compliance. Physician compliance with the use of the AOS was during April 1981 and subsequently improved to >90%. This represented an underestimation becausesome physicians wrote orders twice: once on a regular physician order form and again on the AOS. Table 2. Prescribing patterns. The relative frequency with which the sentinel antibiotics were prescribed is shown in table 1. First-generation cephalosporins were the most frequently ordered antibiotics. There was no significant change in their relative use from April 1981 through April Penicillin plus ampicillin and the aminoglycosides were the second and third most commonly prescribed antibiotics. Use of drugs from group 3 lessened progressively (P =.013)in relation to use ofdrugs from other groups. Group 2 included only cefoxitin and cefamandole in April The third-generation cephalosporins were introduced from July 1981 to April Although a significant trend towards increased use of group 2 agents occurred (P =.003), they accounted for only 5% of sentinel ATCs. The remaining groups - representing aminoglycosides, antistaphylococcal, antianaerobe, and antipseudomonal antibiotics - were used consistently throughout this study period. Clinical indications for the antibiotics prescribed are listed in table 2. Sixty-nine percent (range, 610,10 76%) of first-generation cephalosporin use was for prophylaxis, most often for surgical procedures. Although other groups were also used for prophylaxis, there was a significant decrease in the use of group 2 antibiotics for prophylaxis (P =.001) and a concomitant increase in the prophylactic use of group 1drugs (P =.001). Among group 7 antibiotics, there was a significant decrease in prophylactic use (P =.05) accompanied by an increase in use for treatment (P =.001).Groups 4, 5, and 6 did not undergo any change in their clinical indications. Prescribing frequency. The most significant change detected was that fewer antibiotic treatment courses were prescribed in our hospital (figure 1). Change in clinical indications for sentinel antibiotics over a 25-month period. Prophylaxis/treatment Mean Group no. Sentinel antibiotic usage (0/0) Range (%) Trend 1 1st generation cephalosporins 69/ /9-25 t/ 2 2nd and 3rd generation 13/ / It cephalosporins 3 Penicillin and ampicillin 18/ /56-75 t/ 4 Aminoglycosides 18/ / Antistaphylococcal 26/ / Antianaerobe 8/ / Antipseudomonal 6/ / It NOTE. infection; = no change. P value.001/ / / /.001 Prophylaxis = surgical prophylaxis plus nonsurgical prophylaxis; Treatment = empirical treatment plus documented

4 806 Echols and Kowalsky 0-0 % Pt. Admissions _ Number of ATe's Figure 1. Change in number of ATCs and percentage of admitted patients receiving antibiotics over a 25 month period. ATC = antibiotic treatment course. A M J J A SON olj i In April 1981, a total of 1,313 sentinel ATCs were administered. The number decreased dramatically in May and June with a rise in July and August, which coincided with the arrival of new housestaff members. Thereafter, the number ofatcs continued to decline to rv900 per month. This decrease in ATCs of rv30ojo was statistically significant (P =.02). The percentageof patients receiving sentinel antibiotics also declined significantly (P =.007) from 47% in April 1981 to rv30% over the last 18 months (figure 1). There was a concomitant increase in the proportion of patients receiving more than one antibiotic (P =.001). Thus, during the survey there resulted a relative increase in multiple drug therapy, with fewer ATCs and fewer patients receiving antibiotics. Discussion We relied on the prescribing physician rather than on the chart reviewer to document the clinical indication for the antibiotic prescription. To substantiate the accuracy of the method, we performed three separate chart review audits (30-40 charts per audit) of individual antibiotics. The reviews demonstrated rv95ojo concordance between the indication provided by the prescribing physician on the AOS and the clinical indications determined by the reviewers. Much of our data regarding the distribution of antibiotics is similar to previous studies [9, 12, 26], although comparative data from the United Kingdom differs in several regards [13]. We found firstgeneration cephalosporins prescribed most commonly and used most frequently for prophylaxis in surgical procedures. The greatest decrease in ATCs was among the first-generation cephalosporins. We did not observea decrease in the relative use ofaminoglycosides after the introduction of the secondand third-generation cephalosporins. Although the total number of aminoglycoside ATCs decreased significantly, aminoglycoside use relative to the use of other sentinel groups remained stable. Gentamicin continues to be the most frequently prescribed aminoglycoside and accounts for approximately twothirds of all aminoglycoside ATCs. The overall decrease in ATCs was not the result of an increase in mono-drug therapy. Rather than single, broad-spectrum antibiotics replacing combination therapy, multiple antibiotic use has continued while the number of patients treated has declined. The decreased use ofantibiotics takes on added significance when compared with other studies that have shown an increasing prevalence of antibiotic use despite restrictive policies [5, 12,23]. Only Latorraca et al. [22] have observed a decrease in antibiotic use after introducing a continuous, but retrospective, chart review audit. The AOS resulted in substantial savings by the hospital pharmacy. Since the introduction of the AOS, the proportionofthe hospital drug budget for parenteral antibiotics has decreased from to 19%. Multiple factors such as competitive bidding and generic equivalents contributed to cost savings; however, most hospitals have experienced a significant increase in antibiotic expenditures due to the introduction of costly new agents [24]. An antibiotic order form is not the only way to influence or control a physician's prescribing pat-

5 Antibiotic Order Forms 807 tern. Requiring verbal or written justification for prescriptions of selected antibiotics is effective [1, 7, 17, 19,24] in limiting their use; however, this would be logistically impossible for all antibiotic prescriptions. Because our AOS was used for all antibiotics, no single agent was targeted for surveillance, and the prescribing physician did not have to obtain prior approval for the use of specific antibiotics. Physician resistance to our surveillance method was minimal, as evidenced by the excellent compliance with the AOS. Despite the large turnover in housestaff, the continued use of the AOS has not required substantial educational efforts. References 1. Counts GW. Review and control of antimicrobial usage in hospitalized patients: a recommended collaborative approach. JAMA 1980;238: Finland M. The antibiotic audit-its value in the individual patient. In: Lorian V, ed. Significance of medical microbiology in the care of patients. 2nd ed. Baltimore: Williams & Wilkins, 1982; Scheckler WE, Bennett JV. Antibiotic usage in seven community hospitals. JAMA 1970;213: Roberts AW, Visconti JA. The rational and irrational use of systemic antimicrobial drugs. Am J Hosp Pharm 1972;29: McGowan JE Jr, Finland M. Infection and antibiotic usage at Boston City Hospital: changes in prevalence during the decade J Infect Dis 1974;129: Castle M, Wilfert CM, Cate TR, Osterhout S. Antibiotic use at Duke University Medical Center. JAMA 1977;237: Craig WA, Uman SJ, Shaw WR. Ramgopal V, Eagom LL, Leopold ET. Hospital use of antimicrobial drugs: survey of 19 hospitals and results of antimicrobial control programs. Ann Intern Med 1978;89: Townsend TR, Shapiro M, Rosner B, Kass EH. Use of antimicrobial drugs in general hospitals. I. Description of population and definition of methods. J Infect Dis 1979;139: Shapiro M, Townsend TR, Rosner B, Kass EH. Use of antimicrobial drugs in general hospitals. II. Analysis of patterns of use. J Infect Dis 1979;139: Shapiro M, Townsend TR, Rosner B, Kass EH. Use of antimicrobial drugs in general hospitals. III. Patterns of prophylaxis. N Engl J Med 1979;301: Townsend TR, Shapiro M, Rosner B, Kass EH. Use of antimicrobial drugs in general hospitals: IV. Infants and children. Pediatrics 1979;64: Stevens GP, Jacobson JA, Burke JP. Changing patterns of hospital infections and antibiotic use. Prevalence surveys in a community hospital. Arch Intern Med 1981;141: Leigh DA. Antimicrobial usage in forty-three hospitals in England. J Antimicrob Chemother 1982;9: Moss E, McNicol MW, McSwiggan DA, Miller DL. Survey of antibiotic prescribing in a district general hospital. I. Pattern of use. Lancet 1981;2: Kunin CM. Antibiotic accountability. [editorial] N Engl J Med 1979;301: Kunin CM, Thpasi T, Craig WA. Use of antibiotics. A brief exposition of the problem and some tentative solutions. Ann Intern Med 1973;79: McGowan JE Jr, Finland M. Usage of antibiotics in a general hospital: effect of requiring justification. J Infect Dis 1974;130: Seligman SJ. Reduction in antibiotic costs by restricting use of an oral cephalosporin. Am J Med 1981;71: Brooks GF, Barriere SL. Clinical use of the new beta-lactam antimicrobial drugs. Ann Intern Med 1983;98: Durbin WA Jr, Lapidas B, Goldmann DA. Improved antibiotic usage following introduction of a novel prescription system. JAMA 1981;246: Kowalsky SF, Echols RM, Peck F Jr. Preprinted order sheet to enhance antibiotic prescribing and surveillance. Am J Hosp Pharm 1982;39: Latorraca R, Martins R. Surveillance of antibiotic use in a community hospital. JAMA 1979;242: Simmons HE, Strolley Po. This is medical progress? Trends and consequences of antibiotic use in the United States. JAMA 1974:227: Recco RA, Gladston JL, Friedman SA, Gerken EH. Antibiotic control in a municipal hospital. JAMA 1979; 241: McGowan JE Jr. Antimicrobial resistance in hospital organisms and its relation to antibiotic use. Rev Infect Dis 1983;5: Kennedy DL, Forbes MB, Baum C, Jones JK. Antibioticuse in U.S. hospitals in Am J Hosp Pharm 1983; 40: