P~COEPIDEMIOLOGY AND DRUG UTILIZATION

Temporal decline in filling prescriptions for terfenadin .e closely in time with those for either ketoconazole or erythromycin

Temporal changes in the rates of filling terfenadine prescriptions within 2 days of those for either oral erythromycin or oral ketoconazole were described with use of paid pharmacy claims data from 1988 through 1994 in state Medicaid programs from Michigan and Ohio and in a large health maintenance organization. There were rapid and sigticant declines in the rates of filling prescriptions for either erythromycin or ketoconazole within 2 days of prescriptions for terfenadine in all three databases that coincided with 1992 publicity about the cardiovascular risk of terfenadine. These hndings suggest that the use of terfenadine with contraindicated medications has declined in response to relabeling and publicity concerning the safe use of terfenadine. Further study is necessary to estimate the absolute level of concurrent use of terfenadine with contraindicated medications. (Clin Pharmacol Ther 1997;61:93-6.)

Greg A. Burkhart, MD, Michael J. Sevka, MD, Robert Temple, MD, and Peter K. Honig, MD Rockville, Md.

Terfenadine, a nonsedating HI-receptor antag- onist, was approved in the United States in 1985 for treatment of symptoms of seasonal allergic rhinitis. Subsequent to its marketing, the U.S. Food and Drug Administration (FDA) began re- ceiving rare spontaneous reports of ventricular dysrhythmia, including torsades de pointes, that were associated with concomitant terfenadine and ketoconazole use. The first published case report of the occurrence of torsades de pointes in a patient after concomitant use of terfenadine and ketoconazole appeared in December 1990. After an FDA Pulmonary-Allergy Advisory Committee publicly discussed the potential cardiac toxicity of

From the Center for Drug Evaluation and Research, Food and Drug Administration.

The views expressed are those of the authors and not necessarily those of the FDA.

Received for publication May 8, 1996; accepted July 24, 1996. Reprint requests: Greg A. Bud&art, MD, Center for Drug Eval-

uation and Research, HFD-120, Food and Drug Administra- tion, 1451 Rockville Pike, Rockville, MD 20852.

13/l/76901

terfenadine on June 11, 1990, the FDA concluded that terfenadine labeling should be revised to warn against terfenadine use with ketoconazole or macrolide antibiotics or in patients with hepatic impairment, as well as to warn against exceeding the recommended dose. On August 6, 1990, ter- fenadine’s sponsor issued a “Dear Doctor” letter to 314,441 health professionals describing the la- beling changes that were made in the “precau- tions, ” “adverse reactions,” and “overdose” sections of the label. Shortly thereafter, in vitro and in vivo study findings confirmed that medications that in- hibit the activity of the P4503A4 metabolizing en- zyme system could block terfenadine’s metabolism, resulting in increased levels of parent terfenadine, and that plasma terfenadine levels were strongly correlated with prolongation of the QT interval. After these findings were confirmed and after con- tinued reporting of serious cardiac events occurring during terfenadine use, further labeling changes were made that elevated the warning statements to contraindications, enclosing them in a boxed warn- ing. The labeling changes were announced in a joint

93

94 Burkhart et al. CLINICAL WARMA COLOGY & THERAPEUTICS

JANUARY 1997

---C Michigan

1989 1990 1991 1992 1993 1994

Fig. 1. Proportion of filled terfenadine prescriptions that were within 2 days of a filled erythromycin prescription (adjusted for age and gender).

1995

FDA-sponsor press release followed by a “Dear Doctor” letter on July 6, 1992, that was issued to 613,328 health professionals. Also during July 1992, similar labeling changes were issued for astemizole that were also announced during a joint press re- lease followed by a “Dear Doctor” letter.

Since 1992 there has been a decline in the fre- quency of reporting cases of serious cardiovascular events associated with terfenadine use to the FDA spontaneous reporting system. Although this decline is reassuring, it is difficult to interpret because of potential for underreporting that may result because the event has been clearly labeled, perhaps reducing a health professional’s tendency to report. Thus, although there has been considerable publicity and educational effort aimed at reducing the use of con- traindicated medications with terfenadine with some evidence of success, the actual effectiveness of such

sponsor-FDA actions is not well described. With use of paid pharmacy claims from 1988 through 1994 in the state Medicaid programs from Michigan and Ohio and a large health maintenance organization (HMO), temporal changes in filling terfenadine pre- scriptions within 2 days of those for oral ketocon- azole or oral erythromycin were described.

METHODS Computerized claims files were used to identify

all paid claims for terfenadine prescriptions filled from 1988 through 1994 in Michigan and Ohio Med- icaid databases and an HMO database. With use of the date of service on the claims, the number of terfenadine fills that were within 2 days of the date of a fill for oral erythromycin or oral ketoconazole were determined. The rates of closely filling terfen- adine with either erythromycin or ketoconazole

CLINICAL f HABMACOLOGY & THERAPEUTICS VOLUME 61. NUMBER 1 Burkhart et al. 95

A

0’ _’

,’

4 +. #’ .\ I’ \

-, ,’ ‘A’ *\

+ Michigan

1988 1989 1990 1991 1992 1993 1994 1995

Year

Fig. 2. Proportion of filled terfenadine prescriptions that were within 2 days of a filled ketoconazole prescription (adjusted for age and gender).

Table I. Rates of filling prescriptions for terfenadine within 2 days of those for erythromycin or ketoconazole (Michigan and Ohio Medicaid and an HMO) from 1988 to 1994

Michigan Medicaid Ohio Medicaid HMO Teflenadine fills E * Kf Terfenadine fills E + Kf Terfenadine fills E* Kf

Total Age 64

1-11 12-30 31-50 51-70 271

Sex Male Female

549,376 5 7.2 403,852 3.3 9.7 535,737 1.9 3.4

3.0% 3.8 3.6 1.7% 2.4 0 3.0% 1.3 0.6 31.2% 6.6 6.6 24.2% 4.5 8.8 27.7% 2.3 4.1 35.7% 5.4 9.4 37.4% 3.5 14.5 53.2% 1.8 3.5 19.3% 3.6 6.6 21.8% 2.6 8 15.3% 1.5 2.3 10.8% 2.2 3.4 14.9% 2.1 2.7 0.7% 0.9 0

22.7% 4.3 11.4 23.9% 2.7 16.4 36.2% 1.9 3.7 77.3% 5.2 5.9 76.1% 3.5 7.6 63.8% 1.9 2.8

*Oral erythromycin, number of terfenadine prescriptions within 2 days of oral erythromycin per 100 terfenadine filled prescriptions. iOral ketoconazole, number of terfenadine prescriptions within 2 days of oral ketoconazole per 10,000 terfenadine filled prescriptions

were defined as the number of terfenadine fills within 2 days of that drug divided by the number of terfenadine fills. Separate rates were computed for oral erythromycin and oral ketoconazole. Variation in the rates were described by calendar year, age, and gender with unconditional logistic regression.

Yearly and biannual rates were adjusted for any age and gender changes over time separately in each database.

RESULTS Table I shows the extent of terfenadine exposure

96 Bud&art et al.

by age and gender in each database. Overall, there were about 1.5 million terfenadine prescriptions filled from 1988 through 1994, with the age group from 31 to 50 years accounting for one-third to one-half of all use in all three databases. Likewise, twofold to threefold more prescriptions were filled in females than males. Table I also shows the overall rates of filling terfenadine within 2 days of erythro- mycin and ketoconazole and the variation in rates by age and gender. Without considering temporal changes, from about 2 to 5 per 100 prescriptions for terfenadine were filled within 2 days of a prescrip- tion for erythromycin, whereas from about 3 to 10 per 10,000 terfenadine prescriptions were filled within 2 days of a prescription for oral ketoconazole. Males and patients from 12 to 50 years of age had consistently higher rates of filling terfenadine within 2 days of ketoconazole in each database.

Although the HMO database had consistently lower rates of filling either erythromycin or oral ketoconazole within 2 days of terfenadine, the vari- ation in the rates by age, gender, and year was consistent across the three databases for each drug. Figs. 1 and 2 graph these rates by year, respectively, for erythromycin and ketoconazole separately for each database from 1988 through 1994 adjusting for any temporal changes in age and gender. There were sharp declines in filling terfenadine close to either oral erythromycin or oral ketoconazole. Al- though the largest decline for both terfenadine and erythromycin was in the second half of 1992 in all three databases, there is suggestion that the declines for both began in 1991 in the HMO and Ohio Medicaid. After the decline in 1992, the rates have remained relatively constant. In the second half of 1994, from 3 to 9 per 1000 terfe- nadine prescriptions were filled within 2 days of erythromycin, whereas from 0 to 5 per 10,000 terfenadine prescriptions were filled within 2 days of oral ketoconazole.

DISCUSSION These data show significant declines in the rates

of filling terfenadine prescriptions within 2 days of ketoconazole or erythromycin. Although the rate of filling terfenadine with contraindicated medications is probably an imperfect estimate of actual concur- rent use, as discussed below, the magnitude of the decrease and the likely correlation with concurrent use make it clear that use of contraindicated medi- cations with terfenadine has declined markedly

CLINICAL PHARMA COLOGY &THERAPEUTICS JANUARY 1997

since 1992. This decline correlates temporally with the general publicity and changes in terfenadine labeling regarding its safe use.

Although claims data are useful for evaluation of temporal trends in concurrent use of terfenadine with contraindicated medications, the rate of filling terfenadine prescriptions within 2 days of those for contraindicated medications may be an imperfect estimate of actual concurrent use for at least two reasons. First, both physicians and pharmacists may advise patients not to start one of the filled prescrip- tions until completing the other, meaning that coprescribing as identified in claims data would overestimate concurrent use. Second, claims data do not account for medications that are stored at home and used episodically by patients, so that concurrent use would be underestimated when claims data were used to identify coprescribing. Thus a limitation of this study is the inability to estimate the absolute rate of concurrent use of terfenadine with contrain- dicated medications. Even if only 3 per 10,000 ter- fenadine prescriptions were filled and used concur- rently with ketoconazole, a sizable number of patients would be exposed to potentially dangerous levels of plasma terfenadine, given extensive terfe- nadine use. Further study should therefore focus on prospectively querying patients after prescription fills for the drugs of interest, measuring parent ter- fenadine levels, and focusing on cardiovascular out- comes that could result from concurrent use. Nev- ertheless, it is clear that concomitant use of ketoconazole and erythromycin with terfenadine has fallen by 80% or more.

References Monahan BP, Ferguson CL, Killeavy ES, Lloyd BK, Troy J, Cantilena LR. Torsades de pointes occurring in association with terfenadine use. JAMA 1990;264: 2788-90. Honig PK, Woosley RL, Zamani K, Conner DP, Can- tilena LR. Changes in the pharmacokinetics and elec- trocardiographic pharmacodynamics of terfenadine with concomitant administrations of erythromycin. Clin Pharmacol Ther 199252:231-S. Honig PK, Wortham DC, Sarmani K, Conner DP, Mulin JC, Cantilena LR. Terfenadine-ketoconazole in- teraction: pharmacokinetic and electrocardiographic consequences. JAMA 1993;269:1513-8. Woosley RL, Chen Y, Freiman JP, Gillis RA. Mecha- nism of the cardiotoxic actions of terfenadine. JAMA 1993;269:1.532-6.