CURRENT THERAPEUTIC RESEARCH@ VOL. 61, No. 8, AUGUST 2000

Commentary

Disagreement Among Drug Compendia on Inclusion and Ratings of Drug-Drug Interactions

Thomas R. Fulda,’ Robert J. Valuck,’ Jeanne Vander Zanden,3 Sandra Parker,2 and Patricia J. Byrns4 for the US Pharmacopeia Drug Utilization Review

Advisory Panel*

‘U.S. Pharmacopeia, Rockville, Maryland, ‘Uniuersity of Colorado Health Sciences Center, Denver, Colorado, ‘JVZ Consulting, Denver, Colorado, and 4University of North Carolina,

Chapel Hill, North Carolina

ABSTRACT

Objective: The aim of this study was to determine whether there is consistency in the listing and clinical significance ratings of drug- drug interactions in 5 nationally recognized drug information sources.

Background: The Omnibus Budget Reconciliation Act of 1990, which established the Medicaid Drug Utilization Review (DUR) Pco- gram, required that the program assess data based on “explicit pre- determined standards” including drug-drug interactions and other data specified in the statute. These standards, according to the stat- ute, were to be based on 3 drug compendia and the peer-reviewed literature. This focus on explicit standards, plus a requirement that DUR screening criteria be in the public domain, has provided an opportunity to assess these criteria and the adequacy of the infor- mation on which they are based.

Methods: We compared the listing and clinical significance ratings of drug-drug interactions for a representative sample of angiotensin-converting enzyme inhibitors, beta-blockers, benzodi- azepines, calcium channel blockers, and nonsteroidal anti- inflammatory drugs presented in US Pharmacopeia Drug Informa- tion, American Hospital Formularg Service Drug Information, Hansten’s Drug Interactions Analgsis and Management, Drug In- teraction Facts (Facts and Comparisons), and the Micromedex DRUG-REAX@ system.

Results: In all the drug classes studied, individual drug-drug or drug-class interactions were rarely listed in 21 or 2 drug informa- tion sources. Furthermore, when particular drug-drug or drug-class interactions were listed in >2 of the information sources reviewed,

Members of the US Pharmacopeia Drug Utilization Review Advisory Panel are listed in the Acknowledgments. .ddress correspondence to: Thomas R. Fulda, Program Director, DUR, U.S. Pharmacopeia, 12601 Twin-

brook Parkway, Rockville, MD 20852 (e-mail: trf@usp.org). Accepted for publication April 12, 2000. Printed in the USA. Reproduction in whole or part is not permitted.

540

T.R. FULDA ET AL.

agreement on clinical significance rating generally decreased as the number of information sources in which the interaction was listed increased.

Conclusions: Results of the study suggest that there are dis- crepancies in the listing and clinical significance ratings for drug- drug interactions among the leading drug information sources, al- though the reasons for these discrepancies could not be determined. The study findings also suggest the need to develop methods for resolving discrepancies based on review of the scientific evidence to ensure that DUR is based on high-quality screening criteria.

Key words: drug utilization review, drug-drug interactions, clinical significance, evidence-based medicine. (Curr Ther Res Clin Exp. 2000;61:540-548)

INTRODUCTION

As a result of the passage of the Omnibus Budget Reconciliation Act (OBRA) of 1990, the US Congress required that a Medicaid Drug Utiliza- tion Review (DUR) Program be established to ensure that prescriptions “are appropriate, are medically necessary, and are not likely to result in adverse medical results.“l The statute further required that such pro- grams “assess data on drug use against explicit, predetermined standards including: monitoring for therapeutic appropriateness, overutilization and underutilization, appropriate use of generic products, therapeutic duplica- tion, drug-disease contraindications, drug-drug interactions, incorrect dos- age or duration of drug treatment, and clinical abuse/misuse.“’

The explicit, predetermined standards called for by the statute were to be based on 3 compendia and the peer-reviewed medical literature. The compendia identified in the statute were the American Hospital Formulary Service (AHFS) Drug Information, the US Pharmacopeia Drug Znforma- tion (USP DZ), and the American Medical Association (AMA) Drug Evalu- ations. After 1995, the AMA discontinued its publication, and OBRA was amended to include the Micromedex DRUGDEX@ system. OBRA 1990 also required that each state establish a DUR board and specified the applica- tion of standards as one of the DUR boards’ activities.’

Most state Medicaid DUR programs have adopted review criteria and software supplied by commercial vendors or academic group~.~ For both prospective and retrospective DUR, it appears that the primary decision of the state Medicaid DUR board has been whether or not to adopt particular criteria offered by its chosen vendor. The role of state DUR boards in the criteria development, review, and approval process has not been studied.

The emphasis on explicit standards, together with a requirement that screening criteria for Medicaid DUR be in the public domain3 has provided an opportunity to assess these criteria and the adequacy of the information on which they are based.4*5 Media critiq ues6 charging that pharmacists fail to catch potentially harmful drug-drug interactions at the point of dispens- ing have added a sense of urgency to this endeavor.

541

CURRENT THERAPEUTIC RESEARCH@

At its 1995 meeting, the USP DUR Advisory Panel concluded that “DUR criteria should address instances where there is an increased risk for drug-induced illness that is preventable.“7 The panel further suggested that USP use “its standing advisory committees and staff or consultants to produce criteria based on rating of content by the level of scientific evi- dence . . . and quantification of risk for drug-induced illness.” The panel directed that drug-drug interactions be the first focus of this effort because “these constitute both a large number of the current conflicting criteria and the screening tools used in prospective DUR.“7 The goal was the production of a validated set of DUR criteria that might include evidence tables and bibliography. To accomplish this, USP entered into a contract with the University of Colorado Health Sciences Center in Denver.

The purpose of this paper was to compare the leading drug informa- tion sources to gauge their consistency in listing particular drug-drug in- teractions and the extent to which the different information sources that listed the same interaction agreed with regard to rating the clinical sig- nificance of that interaction.

USP chose to assess information on drug-drug interactions in 5 lead- ing drug information sources: USP DI,’ AHFS Drug Information,g Hansten’s Drug Interactions Analysis and Management, lo Drug Znterac- tion Facts,11 and the Micromedex DRUG-REAX@ system.12 These sources and the rationale for choosing them are shown in Table I.

Table I. The 5 sources of drug-drug interaction information and the rationale for their selection.

Information Source Selection Rationale

USP DI

AHFS Drug Information Hansten’s Drug Interactions

Analysis and Management Drug Interaction Facts Micromedex DRUG-REAX@

Mentioned in OBRA 1990; required in pharmacies by the laws of most states

Mentioned in OBRA 1990 Gold standard drug-interaction reference

Reference commonly found in pharmacies Widely available drug-interaction database

LISP DI = US Pharmacopeia Drug Information; OBRA = Omnibus Budget Reconciliation Act; AHFS = American Hospital Formulary Service.

The AMA Drug Evaluations, Evaluation of Drug Interactions, and the Physicians’ Desk Reference@ also were considered for inclusion in the study. The AMA Drug Evaluations was rejected because it was about to be combined with the USP DI. Evaluation of Drug Interactions was rejected because it was not listed in OBRA 1990 and because it had a lower market share than the other texts. The Physicians’ Desk Reference@ was rejected because it rates likelihood rather than severity of drug interactions.

542

T.R. FULDA ET AL.

METHODS

To document the extent to which the selected information sources agreed or disagreed with regard to the inclusion and clinical significance ratings of drug-drug interactions, 5 drug classes were chosen for analysis. These classes were selected for the following reasons: they include frequently prescribed drugs; serious consequences in terms of morbidity and mortal- ity are associated with the occurrence of drug-drug interactions involving drugs in these classes; and the USP DUR Panel expected high levels of disagreement in the listing of drug-drug interactions for these drug classes across information sources. The 5 drug classes and the drugs included in each class are as follows: angiotensin-converting enzyme (ACE) inhibitors (benazepril, captopril, enalapril, fosinopril, lisinopril, quinapril, and ramipril); benzodiazepines (alprazolam, chlordiazepoxide, clonazepam, clorazepate, diazepam, estazolam, flurazepam, halazepam, lorazepam, midazolam, nitrazepam, oxazepam, prazepam, quazepam, temazepam, and triazolam); beta-blockers (acebutolol, atenolol, betaxolol, bisoprolol, carteolol, esmolol, labetalol, metoprolol, nadolol, penbutolol, pindolol, pro- pranolol, sotalol, and timolol); calcium channel blockers (amlodipine, be- pridil, diltiazem, felodipine, isradipine, nicardipine, nifedipine, nimo- dipine, and verapamil); and nonsteroidal anti-inflammatory drugs (diclofenac, diflunisal, etodolac, fenoprofen, flurbiprofen, ibuprofen, indo- methacin, ketoprofen, ketorolac, meclofenamate, mefenamic acid, nabu- metone, naproxen, oxaprozin, phenylbutazone, piroxicam, sulindac, and tolmetin).

Excluded from the above drug groups were drugs not available in the United States and 6 new drugs (moexipril, perindopril, carvedilol, nitren- dipine, celecoxib, and rofecoxib) that were not listed in 23 of the 5 drug information sources.

RESULTS

To determine the extent of agreement among the information sources re- garding drug-drug interactions, investigators identified all drug-drug in- teractions for each of the 64 drugs selected for study, the number of infor- mation sources listing each interaction, and the number of sources having similar clinical significance ratings. An example for 1 (captopril) of the 64 drugs for which information was obtained is shown in Table II.

Of the 66 drug-drug interactions involving captopril and another drug or drug class listed in the 5 information sources, 40 were listed in only 1 of the texts. The drug-drug interactions listed were between captopril and amlodipine, azosemide, bendroflumethiazide, benzthiazide, capsaicin, chlorothiazide, chlorthalidone, cimetidine, cyclothiazide, erythropoietin, ethanol, glycerin, hydrochlorothiazide, hydralazine, hydroflumethiazide, indapamide, insulin, interferon (alpha and beta), ketorolac, magnesium

543

CURRENT THERAPEUTIC RESEARCH@

Table II. Agreement among 5 information sources regarding drug-drug interactions involv- ing captopril.

No. of Texts No. of Texts with Listing Drug-Drug Interaction Similar Significance Rating

Interacting drug Allopurinol I$i,l;mJde

Azathioprine Bumetanide Chlorpromazine Cyclosporin Digoxin Ethacrynic acid Furosemide lndomethacin Lithium Potassium Probenecid Spironolactone Triamterene

Interacting drug class Antacids NSAlDs Antidiabetic agents Antihypertensive drugs Beta-blockers (eyedrops) Diuretics Nitrates Potassium-sparing diuretics Potassium supplements Salicvlates

NSAlDs = nonsteroidal anti-inflammatory drugs

carbonate, metolazone, naproxen, pergolide, piretanide, polythiazide, pra- zosin, quinethazone, ticrynafen, torsemide, trichlorothiazide, or xipamide. The drug-class interactions listed were between captopril and adrener- gic blocking agents, beta-blockers, bone marrow depressants, calcium channel blockers, estrogens, ganglionic blocking agents, phenothiazines, and sympathomimetics.

There was considerable variation in the number of drug-drug interac- tions listed for each drug within a drug class. In the beta-blocker class, for example, the number of interactions listed for a particular beta-blocker and all other drugs or drug classes interacting with it ranged from 54 for betaxolol to 187 for propranolol.

Overview of Agreement Among In$ormation Sources on Drug-Drug Interactions

It was not feasible to provide a detailed listing similar to the one provided for captopril for all 64 drugs reviewed. Instead, we have summa- rized the findings by drug class in Table III. The percentages of drug-drug and drug-class interactions for drugs in each of the 5 classes are given by

544

T.R. FULDA ET AL.

Table III. Summary of overall agreement among the information sources regarding the list- ing of drug-drug interactions in 5 drug classes.

No. of

Drug Class Drug-Drug

Interactions

ACE inhibitors 288 Benzodiazepines 1061 Beta-blockers 1267 kal;K channel blockers 485

1004

Sources Listing Drug-Drug Interactions

1 2 3 4 5

58.0% 20.1% 14.7% 3.5% 1 .O% 68.3% 20.2% 5.0% 4.6% 1.9% 60.6% 24.8% 10.5% 3.6% 0.5% 69.1% 19.0% 6.7% 2.8% 2.8% 64.4% 21 .O% 8.0% 3.5% 3.1%

ACE = angiotensin-converting enzyme; NSAlDs = nonsteroidal anti-inflammatory drugs.

the number of information sources listing the interactions. In the case of ACE inhibitors, there were 288 drug-drug or drug-class interactions in- volving 7 ACE inhibitors and some other drug. Of these interactions, 167 (58.0%) were listed in only 1 information source. Only 3 interactions (1.0%) were listed in all 5 information sources reviewed. As Table III in- dicates, for all drugs in the 5 drug classes studied, it was more likely for a drug-drug interaction to be listed in 1 or 2 information sources than in all 5 sources.

Overview of Agreement Among InJormation Sources on Clinical Significance Ratings

An additional objective of this study was to analyze the extent to which there was agreement regarding the clinical significance of any drug- drug interaction listed in 32 of the 5 information sources studied. Since ~50% of the drug-drug interactions listed for all 5 drug classes appeared in only 1 of 5 information sources, the number of drug-drug interactions considered in the analysis of clinical significance agreement is consider- ably smaller than the number of interactions in the previous analysis of the overall compendia1 listing agreement.

Table IV presents an overview of the extent of agreement on clinical significance ratings for drug-drug interactions involving the 5 drug classes listed in 2, 3, 4, or all 5 of the information sources studied. In the case of benzodiazepines, for example, 214 drug-drug or drug-class interactions were listed in 2 information sources; for 106 (49.5%) of the 214 interac- tions, there was agreement with regard to clinical significance rating. Only 17 (34.7%) of 49 interactions listed in 4 information sources agreed with regard to the clinical significance rating. Thus, for the benzodiazepines, the degree of agreement on clinical significance generally decreased as the number of information sources listing a drug-drug interaction increased. This pattern was also observed in the 4 other drug classes studied.

545

CURRENT THERAPEUTIC RESEARCH@

Table IV. Agreement among the information sources regarding clinical significance ratings of drug-drug interactions in 5 drug classes.

Agreement Among Sources

Drug Class

ACE inhibitors Benzodiazepines Beta-blockers ;a&K channel blockers

2 3 4 5

46.6% 38.0% ;;;;:f

66.7% 49.5% 28.3% 43.3% 16.5% 2.20; $2

46.7% 25.0% 7.7% 56.4% 8.8% 2.9% ;I;

ACE = angiotensin-converting enzyme; NSAlDs = nonsteroidal anti-inflammatory drugs.

DISCUSSION AND CONCLUSIONS

This study does not provide an explanation for the discrepancies observed among the leading information sources with regard to the listing of drug- drug interactions for the 5 drug classes studied. It also does not explain the variation in the clinical significance ratings when a particular drug-drug interaction is listed in 22 information sources. This research does, how- ever, raise questions as to whether the kinds of discrepancies observed in the listing of drug-drug interactions also exist for other drug standards (eg, dosage, duration of therapy, therapeutic duplication, drug-disease contra- indications) for which DUR criteria are frequently being developed.

The observed discrepancies present a significant problem for develop- ers of DUR criteria and health care providers. Failure to resolve the dis- crepancies could result (and may have already resulted) in the develop- ment and use of criteria that lack adequate scientific evidence. To ensure that screening criteria focus on significant preventable problems, the reso- lution of the discrepancies described here is of primary importance. A methodology for accomplishing this has been proposed.13 The implications of this investigation have extended beyond considerations of the effective- ness and efficiency of given DUR programs and have raised questions as to whether information commonly used by practitioners may contribute to variability in medical practice and the difficulty in practicing evidence- based medicine at the treatment level.

Acknowledgments

Financial support was provided by the USP Inc., Rockville, Maryland, through a contract with the University of Colorado Health Sciences Cen- ter, Denver, Colorado.

Members of the US Pharmacopeia Drug Utilization Review Advisory Panel are: Terrence F. Blaschke, MD, Chair, Stanford University, Palo Alto, California; David M. Angaran, RPh, Powell, Ohio; Edward P. Arm- strong, PharmD, University of Arizona, Tucson, Arizona; James L. Black-

546

T.R. FULDA ET AL.

burn, PharmD, University of Saskatchewan (emeritus), Saskatoon, Sas- katchewan, Canada; Catherine E. Burley, MD, Georgia Drug Utilization Review Board, Fayetteville, Georgia; Patricia J. Byrns, MD, University of North Carolina, Chapel Hill, North Carolina; Elizabeth A. Chrischilles, PhD, University of Iowa, Iowa City, Iowa; Theodore M. Collins, RPh, Uni- versity of Wisconsin-Madison, Madison, Wisconsin; Robert P. Craig, PharmD, PCS Health Systems Inc., Scottsdale, Arizona; W. Gary Erwin, PharmD, Omnicare Inc., King of Prussia, Pennsylvania; Stan N. Finkel- stein, MD, Massachusetts Institute of Technology, Cambridge, Massachu- setts; Catherine A. Harrington, PharmD, PhD, The Lewin Group, Fairfax, Virginia; Mark L. Horn, MD, Pfizer Inc., New York, New York; Judith K. Jones, MD, PhD, The Degge Group, Ltd., Arlington, Virginia; Michael L. Kelly, RPh, Mississippi Pharmacists Association, Jackson, Mississippi; Duane M. Kirking, PharmD, PhD, University of Michigan, Ann Arbor, Michigan; Ann M. Koeniguer, RPh, Research Medical Center, Parkeville, Missouri; David Lee, MD, Management Sciences for Health, Arlington, Virginia; Gary M. Levine, RPh, Medicine Shoppe International, St. Louis, Missouri; Gladys Peachy, RN, Med, MHSc, Dundas, Ontario, Canada; Eleanor M. Perfetto, PhD, QualityMetrics Inc., Stevensville, Maryland; T. Donald Rucker, PhD, University of Illinois (emeritus), Adamstown, Mary- land; Daniel W. Saylak, DO, Texas Drug Utilization Review Board, Bryan, Texas; Fredrica E. Smith, MD, New Mexico Drug Utilization Review Board, Los Alamos, New Mexico; Brian L. Strom, MD, MPH, University of Pennsylvania, Philadelphia, Pennsylvania; and Ilene H. Zuckerman, PharmD, University of Maryland, Baltimore, Maryland.

References:

1. Pub L No. 101508, 104 Stat 1388 151-155.

2. Lipowski EE, Collins T. Medicaid DUR Programs: 1993. Washington, DC: American Pharmaceutical Association; 1995.

3. US Senate Special Committee on Aging. Implementation of the Drug Use Review Pro- visions of the Medicaid Prudent Pharmaceutical Provisions of the Omnibus Budget Rec- onciliation Act of 1990. US Special Committee on Aging, l-6, 1991.

4. Byrns PJ, Bondy J, Valuck R, et al. Opening the black box: A case study of conflicting review criteria. Clin Pharm Ther. In press.

5. Lipton HL, Bird JA. Drug utilization review in ambulatory settings: State of the science and directions for outcomes research. Med Care. 1993;31:1069-1082.

6. Danger at the drug store. US News World Report. August 26, 1996.

7. Minutes dated January 11, 1997 of the USP DUR Advisory Panel Meeting of December 8-9, 1996. Unpublished.

8. USP Drug Information Volume I: Drug Information for the Health Care Professional. 16th ed. Rockville, Md: US Pharmacopeial Convention; 1996.

547

CURRENT THERAPEUTIC RESEARCH@

9. AHFS Drug Zr$ormatiolz. Bethesda, Md: American Society of Hospital Pharmacists; 1996.

10. Hansten PD, Horn JR, eds. Drug Interactions Analysis and Management. Vancouver,

WA: Applied Therapeutics; October 1996.

11. Drug Interaction Facts. St. Louis, MO: Facts and Comparisons; October 1996.

12. DRUG-REAXB system [database on CD-ROM]. Englewood, Colo: Micromedex; November 1996.

13. Valuck RJ, Byrns PJ, Fulda TR, et al. Methodology for assessing drug-drug interaction

evidence in the peer-reviewed medical literature. Curr Ther Res Clin Exp. 2000;61:553- 568.

548