3
Assessment of Safety Data

Concerns about the safety of the hypnotic drug Halcion (triazolam) have existed since C. van der Kroef reported a syndrome of adverse reactions to the drug in 1979 (van der Kroef, 1979)1. Many changes in the labeling guidelines have been made since then, but concerns have persisted and are described most succinctly in the 1992 Public Citizen petition to remove Halcion from the U.S. market. In its investigation of these and other concerns, the Institute of Medicine (IOM) committee reviewed and assessed the relevant data from Upjohn's New Drug Application (NDA; NDA 17-892) and other sources, including the published literature, as they pertain to whether triazolam (1) produces a unique profile or syndrome of adverse events and (2) produces adverse effects that are qualitatively similar but quantitatively different from those associated with other benzodiazepine hypnotic agents.

More specifically, the committee reviewed and evaluated (1) protocols and data from well-controlled pre- and postmarketing clinical trials; (2) data from studies performed in countries other than the United States, including data from a cohort study; (3) spontaneous reports of adverse events; and (4) data from the published literature. This chapter is organized according to these sources of data and their analyses. The committee then presents its conclusions and recommendations at the end of the chapter.

WELL-CONTROLLED PREMARKETING CLINICAL TRIALS

In assessing the data from well-controlled premarketing clinical trials, the IOM committee reviewed and performed independent statistical reanalyses of two comprehensive summaries of the safety data: Upjohn's Integrated Summary of Safety (ISS) (The Upjohn Company, 1991) and the reconstructed tabular data of adverse events analyzed by the U.S.

1  

An English text description of the Dutch experience is available (Meyboom, 1992).



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data 3 Assessment of Safety Data Concerns about the safety of the hypnotic drug Halcion (triazolam) have existed since C. van der Kroef reported a syndrome of adverse reactions to the drug in 1979 (van der Kroef, 1979)1. Many changes in the labeling guidelines have been made since then, but concerns have persisted and are described most succinctly in the 1992 Public Citizen petition to remove Halcion from the U.S. market. In its investigation of these and other concerns, the Institute of Medicine (IOM) committee reviewed and assessed the relevant data from Upjohn's New Drug Application (NDA; NDA 17-892) and other sources, including the published literature, as they pertain to whether triazolam (1) produces a unique profile or syndrome of adverse events and (2) produces adverse effects that are qualitatively similar but quantitatively different from those associated with other benzodiazepine hypnotic agents. More specifically, the committee reviewed and evaluated (1) protocols and data from well-controlled pre- and postmarketing clinical trials; (2) data from studies performed in countries other than the United States, including data from a cohort study; (3) spontaneous reports of adverse events; and (4) data from the published literature. This chapter is organized according to these sources of data and their analyses. The committee then presents its conclusions and recommendations at the end of the chapter. WELL-CONTROLLED PREMARKETING CLINICAL TRIALS In assessing the data from well-controlled premarketing clinical trials, the IOM committee reviewed and performed independent statistical reanalyses of two comprehensive summaries of the safety data: Upjohn's Integrated Summary of Safety (ISS) (The Upjohn Company, 1991) and the reconstructed tabular data of adverse events analyzed by the U.S. 1   An English text description of the Dutch experience is available (Meyboom, 1992).

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data Food and Drug Administration (FDA) (Laughren and Lee, 1992); both had been presented at a May 1992 meeting of the FDA Psychopharmacologic Drugs Advisory Committee (PDAC). The committee first analyzed the frequency of adverse events and then examined the frequency and nature of adverse events that led to the withdrawal of subjects from the trials (i.e., dropouts). Adverse Events In 1992 Upjohn created a new database of safety information by reentering the data from the case-report forms for the 116 clinical trials in the NDA. FDA oversaw this effort and verified that ''there was a highly accurate transfer of pertinent data" from the case reports to the new database (Laughren and Lee, 1992, p. 9). Upjohn also developed an ISS using results from 79 studies (from among the 116 studies from the NDA, but without the Phase I studies and including the available postmarketing clinical trials) comparing approximately 4,000 subjects treated with Halcion (0.1 to 1.0 mg), 1,300 subjects treated with flurazepam (Dalmane; 15 to 30 mg), and 2,100 subjects treated with placebo. A subset of the 116 studies was selected specifically to compare Halcion with placebo and another benzodiazepine hypnotic drug. That subset consisted of the 25 studies that (1) involved subjects with insomnia, (2) had a parallel-group design, (3) were at least 1 week in duration, and (4) involved placebo or flurazepam as the comparator drug. Upjohn used this subset of 25 studies in its comparative analysis in the ISS, and FDA used this subset in its analysis of dropouts. The IOM committee first examined Upjohn's comparative analysis of the frequency of adverse events as described in the ISS. Integrated Summary of Safety Table 3-1 indicates the frequency of adverse events involving the central nervous system (CNS) from the 79 clinical trials in the ISS. Tables 3-2 through 3-5 present tabular summaries of the data from the subset of 25 studies using the following classifications: adult and geriatric insomniac populations, low and high doses, and shorter and longer durations. In reviewing these data, the committee made the following three observations. Observation 1: Comparable Safety Profile Halcion at the currently recommended doses (0.125 and 0.25 mg) and for the shortest durations of use (1 to 7 days) has a safety profile comparable to those of both placebo and the lowest dose of flurazepam (15 mg), even for those undesirable events associated with the pharmacologic activity of benzodiazepines, namely, restlessness, nervousness, drowsiness, impaired coordination, light-headedness, and dizziness (Tables 3-2, 3-4, and 3-5). Rates of response for the most frequently occurring CNS-related adverse event (sedation or drowsiness) were 16.7, 23.5, and 11.3 percent for the three groups (Halcion, flurazepam, and placebo), respectively (see Table 3-1), Dizziness and headache were second and third most common adverse events. Other events of interest described in the tabulated summaries were impaired memory (at rates of 0.7, 0.5, and 0.2 percent, respectively) and impaired coordination (at rates of 3.4, 4.6, and 1.5 percent, respectively).

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data TABLE 3-1 Number (percent) of Subjects Reporting CNS-Related Adverse Events in Adequate and Well-Controlled Phase II/III Studies () with a Duration of Treatment of 1 to 92 days Medical Event Halcion (n = 3,982) Flurazepam (n = 1,295) Placebo (n = 2,151) Othera (n = 1,098) Drowsiness/sedation 664 (16.7) 304 (23.5) 244 (11.3) 178 (16.2) Dizziness 350 (8.8) 68 (5.3) 132 (6.1) 80 (7.3) Headache 307 (7.7) 110 (8.5) 175 (8.1) 41 (3.7) Tiredness 183 (4.6) 77 (5.9) 171 (7.9) 122 (1.1) Nervousness 164 (4.1) 61 (4.7) 117 (5.4) 19 (1.7) Impaired coordination 136 (3.4) 60 (4.6) 33 (1.5) 3 (0.3) Depression 100 (2.5) 47 (3.6) 81 (3.8) 1 (0.1) Insomnia 73 (1.8) 28 (2.2) 98 (4.6) 9 (0.8) Confusion 53 (1.3) 14 (1.1) 50 (2.3) 1 (0.1) Excitement 49 (1.2) 1 (0.1) 52 (2.4) 0 Euphoria 38 (1.0) 25 (1.9) 26 (1.2) 2 (0.2) Memory impairment 27 (0.7) 7 (0.5) 5 (0.2) 0 Tremor 22 (0.6) 5 (0.4) 24 (1.1) 1 (0.1) Concentration difficulty 21 (0.5) 15 (1.2) 16 (0.7) 5 (0.5) Vasomotor disturbances 21 (0.5) 5 (0.4) 8 (0.4) 0 a Pentobarbital, secobarbital, chloral hydrate, methaqualone, diazepam, and oxazepam. SOURCE: The Upjohn Company (1992).

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data TABLE 3-2 CNS-Related Medical Events for Adult Insomniac Subjects, 1 to 2 Weeks of Treatment   Triazolam 0.25 mg (n = 35) vs. Placebo (n = 35)a Triazolam 0.25-0.5 mg (n = 121) vs. Flurazepam 15-30 mg (n = 52) and Placebo (n = 77)b Triazolam 0.5 mg (n = 418) vs. Flurazepam 30 mg (n = 216) Flurazepam 30 mg (n = 216)c Medical Event TZ PBO TZ FLU PBO TZ FLU PBO Drowsiness/ sedation 3 (8.6) 4 (11.4) 55 (45.5) 11 (21.2) 47 (61.0) 107 (25.6) 68 (31.5) 29 (13.7) Headache 9 (25.7) 6 (17.1) 12 (9.9) 4 (7.7) 2 (2.6) 65 (15.6) 24 (11.1) 36 (17.0) Dizziness 5 (14.3) 0 25 (20.7) 1 (1.9) 21 (27.3) 39 (9.3) 15 (6.9) 10 (4.7) Nervousness 1 (2.9) 1 (2.9) 6 (5.0) 2 (3.8) 1 (1.3) 27 (6.5) 11 (5.1) 9 (4.2) Tiredness 0 0 46 (38.0) 1 (1.9) 59 (76.6) 13 (3.1) 6 (2.8) 2 (0.9) Paresthesia 3 (8.6) 0 0 0 0 2 (0.5) 0 0 Dysesthesia 1 (2.9) 0 0 0 1 (1.3) 1(0.2) 0 0 Insomnia 1 (2.9) 0 0 0 0 2 (0.5) 3(1.4) 3 (1.4) Impaired coordination 0 0 2 (1.7) 2 (3.8) 0 22 (5.3) 7 (3.2) 6 (2.8) Memory impairment 0 0 0 0 0 2 (0.5) 1 (0.5) 0 Confusion 0 0 0 0 0 2 (0.5) 0 1 (0.5) Disorientation 0 0 0 0 0 1 (0.2) 0 0 Vasomotor disturbances 0 0 3 (2.5) 1 (1.9) 0 3 (0.7) 0 0 Derealization 0 0 1 (0.8) 0 0 0 1 (0.5) 0 Dream abnormalities 0 0 1 (0.8) 0 0 1 (0.2) 2 (0.9) 1 (0.5) Euphoria 0 0 I (0.8) 0 0 1 (0.2) 0 0 Fear 0 0 1 (0.8) 0 0 1 (0.2) 0 0 Intellectual impairment 0 0 1 (0.8) 0 0 1 (0.2) 0 0 Irritability 0 0 1 (0.8) 0 2 (2.6) 2 (0.5) 3 (1.4) 3 (1.4) Shakiness 0 0 1 (0.8) 0 0 0 0 0 Depression 0 0 0 0 0 7 (1.7) 4 (1.9) 3 (1.4)

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data   Triazolam 0.25 mg (n = 35) vs. Placebo (n = 35)a Triazolam 0.25-0.5 mg (n = 121) vs. Flurazepam 15-30 mg (n = 52) and Placebo (n = 77)b Triazolam 0.5 mg (n = 418) vs. Flurazepam 30 mg (n = 216) Flurazepam 30 mg (n = 216)c Medical Event TZ PBO TZ FLU PBO TZ FLU PBO Intoxicated/ inebrious state 0 0 0 0 0 2 (0.5) 0 0 Concentration difficulty 0 0 0 0 0 1 (0.2) 2 (0.9) 0 Muscle tone disorders 0 0 0 0 0 1 (0.2) 1 (0.5) 1 (0.5) Syncope 0 0 0 0 0 1 (0.2) 0 0 NOTE: Abbreviations: n, number of subjects in the treatment group; TZ, triazolam; FLU, flurazepam; PBO, placebo. a Includes Protocol 6401. b Includes flexible-dose Protocols 6400 and 2401. c Includes Protocols 6004, 6016, 6042, 6043, and 6044. SOURCE: The Upjohn Company (1992).

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data TABLE 3-3 CNS-related Medical Events for Adult Insomniac Subjects, 4 to 6 and 12 to 13 Weeks of Treatment   4-6 Weeks' Duration       12-13 Weeks' Duration   Triazolam 0.25 mg (n = 54) vs Flurazepam 30 mg (n = 27)a Triazolam 0.5 mg (n = 220) vs Flurazepam 30 mg (n = 121) and Placebo (n = 96)b Triazolam 0.5 or 0.6 mg (n = 119) vs Flurazepam 30 mg (n = 97)c Medical Event TZ FLU TZ FLU PBO TZ FLU Drowsiness/ sedation 14 (25.9) 15 (55.6) 62 (28.2) 33 (27.3) 10 (10.4) 38 (31.9) 40 (41.2) Headache 5 (9.3) 4 (14.8) 38 (17.3) 10 (8.3) 12 (12.5) 31 (26.1) 20 (20.6) Dizziness 7 (13.0) 7 (25.9) 49 (21.8) 10 (3.3) 3 (3.1) 9 (7.6) 5 (5.2) Impaired coordination 6 (11.1) 8 (29.6) 17 (7.7) 8 (6.6) 0 9 (7.6) 7 (7.2) Tiredness 4 (7.4) 1 (3.7) 4 (1.8) 5 (4.1) 2 (2.1) 2 (1.7) 7 (7.2) Insomnia 3 (5.6) 1 (3.7) 2 (0.9) 0 2 (2.1) 2 (1.7) 1 (l.0) Depression 2 (3.7) 0 3 (1.4) 3 (2.5) 0 6 (5.0) 5 (5.2) Memory impairment 0 0 2 (0.9) 0 0 11 (9.2) 1 (1.0) Confusion 1 (1.9) 0 0 0 0 3 (2.5) 0 Disorientation 0 0 2 (0.9) 0 0 1 (0.8) 1 (1.0) Paresthesia 0 0 1 (0.5) 0 0 0 1 (1.0) Dysesthesia 0 0 1 (0.5) 0 1 (1.0) 1 (0.8) 0 Vasomotor disturbances 1 (1.9) 0 2 (0.9) 0 0 3 (2.5) 2 (2.1) Derealization 0 0 1 (0.5) 0 0 0 0 Dream abnormalities 0 0 1 (0.5) 0 1 (1.0) 2 (1.7) 2 (2.1) Increased motor activity 0 0 0 0 0 1 (0.8) 0 Intellectual impairment 5 (9.3) 0 0 0 0 2 (1.7) 0

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data   4-6 Weeks' Duration       12-13 Weeks' Duration   Triazolam 0.25 mg (n = 54) vs Flurazepam 30 mg (n = 27)a Triazolam 0.5 mg (n = 220) vs Flurazepam 30 mg (n = 121) and Placebo (n = 96)b Triazolam 0.5 or 0.6 mg (n = 119) vs Flurazepam 30 mg (n = 97)c Medical Event TZ FLU TZ FLU PBO TZ FLU Irritability 0 1 (3.7) 1 (0.5) 0 0 1 (0.8) 0 Shakiness 1 (1.9) 0 0 0 1 (1.0) 1 (0.8) 0 Excitement 0 0 0 0 0 1 (0.8) 0 Mood changes 0 0 2 (0.9) 0 0 1 (0.8) 0 Apathy 1 (1.9) 0 1 (0.5) 0 0 0 0 Drug withdrawal symptoms 1 (1.9) 0 0 0 0 0 0 Personality changes 0 0 1 (0.5) 0 0 0 0 Hallucinations 0 0 0 0 0 2 (1.7) 0 Concentration difficulty 0 0 0 0 0 1 (0.8) 0 Drug abuse 0 0 0 0 0 1 (0.8) 0 Agitation 0 0 0 0 0 1 (0.8) 0 NOTE: Abbreviations: n, number of subjects in the treatment group; TZ, triazolam; FLU, flurazepam; PBO, placebo. a Includes Protocol 6401. b Includes flexible-dose Protocols 6400 and 2401. c Includes Protocols 6004, 6016, 6042, 6043, and 6044. SOURCE: The Upjohn Company (1992).

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data TABLE 3-4 CNS-Related Medical Events for Geriatric Subjects, 1 Week of Treatment   Triazolam 0.25 mg (n = 46) vs Placebo (n = 44)a Triazolam 0.125-0.25 mg (n = 18) vs Placebo (n = 19)b Medical Event TZ PBO TZ PBO Drowsiness/sedation 5 (10.9) 4 (9.1) 2 (11.1) 1 (5.3) Headache 4 (8.7) 3 (6.8) 0 1 (5.3) Dizziness 2 (4.3) 2 (4.5) 0 1 (5.3) Tiredness 1 (2.2) 0 0 1 (5.3) Nervousness 1 (2.2) 4 (9.1) 0 3 (15.8) Memory impairment 1 (2.2) 0 0 0 NOTE: Abbreviations: n, number of subjects in the treatment group; TZ, triazolam; PBO, placebo. a Includes Protocol 6401. b Includes flexible-dose Protocols 6400 and 2401. SOURCE: The Upjohn Company (1992).

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data TABLE 3-5 CNS-Related Medical Events for Geriatric Subjects, 1 to 2 and 4 Weeks of Treatment   1-2 Weeks' Duration 4 Weeks' Duration         Triazolam 0.25 mg (n = 35) vs Flurazepam 15 mg (n = 58) and Placebo (n = 48)a Triazolam 0.25 mg (n = 14) vs Flurazepam 15 mg (n = 13) and Placebo (n = 14)b Triazolam 0.25-0.5 mg (n = 40) vs Flurazepam 15-30 mg (n = 40) and Placebo (n = 41)c Medical Event TZ FLU PBO TZ FLU PBO TZ FLU PBO Drowsiness/ sedation 19 (18.1) 17 (29.33) 3~ (6.3) 4 (25.6) 3 (23.1) 2 (14.3) 24 (60.0) 31 (77.5) 19 (46.33) Dizziness 1 (1.0) 0 0 0 0 0 22 (55.0) 28 (70.0) 15 (43.9) Headache 15 (14.3) S (13.8) 4 (8.3) 1 (7.1) 1 (7.7) 0 5 (12.5) 3 (7.5) 4 (9.8) Impaired coordination 2 (1.9) 1 (1.7) 0 1 (7.1) 1 (7.1) 0 5 (12.5) 7 (17.5) 4 (9.8) Nervousness 3 (2.9) 0 2 (4.2) 1 (7.1) 0 0 12 (30.0) 8 (20.0) 6 (14.6) Memory impairment 0 0 0 0 0 0 1 (2.5) 0 0 Confusion 1 (1.0) 1 (1.7) 0 0 0 0 0 1 (2.5) 1 (2.4) Disorientation 0 0 0 0 0 0 2 (5.0) 0   Paresthesia coordination 0 0 0 0 0 0 1 (2.5) 1 (2.5)   Dysesthesia impairment 0 0 0 0 0 0 2 (5.0) 1 (2.5)   Insomnia 1 (1.0) 0 0 0 0 0 5 (12.5) 3 (7.5) 1 (2.4) Vasomotor disturbances 0 1 (1.7) 0 0 0 0 4 (10.0) 0 1 (2.4) Dream abnormalities 0 0 0 0 0 0 1 (2.5) 1 (2.5) 2 (4.9) Tremor 0 0 0 0 0 0 1 (2.5) 0 0 Fear 0 0 0 1 (7.1) 0 0 0 0 0 Intellectual impairment 0 0 0 0 0 0 1 (2.5) 0 0 Irritability 0 0 0 1 (7.1) 0 0 2 (5.0) 0 1 (2.4) Shakiness 0 0 0 0 0 0 1 (2.5) 0 0 Excitement impairment 0 0 0 0 0 0 1 (2.5) 0 1 (2.4) Mood changes 1 (1.0) 0 0 0 0 0 0 0 0

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data   1-2 Weeks' Duration 4 Weeks' Duration         Triazolam 0.25 mg (n = 35) vs Flurazepam 15 mg (n = 58) and Placebo (n = 48)a Triazolam 0.25 mg (n = 14) vs Flurazepam 15 mg (n = 13) and Placebo (n = 14)b Triazolam 0.25-0.5 mg (n = 40) vs Flurazepam 15-30 mg (n = 40) and Placebo (n = 41)c Medical Event TZ FLU PBO TZ FLU PBO TZ FLU PBO Depression 3 (2.9) 0 0 0 0 0 1 (2.5) 3 (7.5) 2 (4.9) Concentration difficulty 0 0 0 0 0 0 1 (2.5) 2.(5.0) 0 Drug withdrawal symptoms 1 (1.0) 0 0 0 0 0 0 0 0 Drug habituation 0 0 0 1 (7.1) 0 0 0 0 0 Agitation 0 0 1 (2.1) 0 0 0 1 (2.5) 0 0 Sexual dysfunction 0 0 0 0 0 0 1 (2.5) 1 (2.5) 1 (2.4) Apathy 0 0 0 0 0 0 2 (5.0) 1 (2.5) 1 (2.4) NOTE: Abbreviations: n, number of subjects in the treatment group; TZ, triazolam; FLU, flurazepam; PBO, placebo. a Includes Protocol 6401. b Includes flexible-dose Protocols 6400 and 2401. c Includes Protocols 6004, 6016, 6042, 6043, and 6044. SOURCE: The Upjohn Company (1992).

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data Observation 2: Comparable Rates of CNS-Related Events In studying the data in Tables 3-2 through 3-5 the committee concluded that for studies comparing Halcion, flurazepam, and placebo in non-geriatric adults, and geriatric subjects in studies of comparable length, the rates of CNS-related adverse events for the two drugs and placebo do not differ remarkably. There were, however, a very few instances in which the rate of CNS-related adverse events was at least threefold greater for Halcion relative to that for placebo. In comparing Halcion with flurazepam, the committee found a few comparisons in which the rate of adverse events for flurazepam is lower by twofold and a few instances in which the rate is lower by as much as threefold. For example, a significant difference in memory impairment was noted for the 0.5-mg Halcion dose. As seen in the last two columns of Table 3-3, the occurrence of memory impairment in 9.2 percent of subjects treated with Halcion (0.5 or 0.6 mg) stands out in comparison with a rate of memory impairment of only 1 percent among subjects treated with flurazepam (30 mg). It is important to note that the data in these two columns are from studies in which relatively high doses (higher than the current Halcion labeling recommends as an initial or starting dose) of the two drags (0.5 or 0.6 mg of Halcion or 30 mg of flurazepam) were used, and in which there was a long duration of treatment (12 to 13 weeks). These differences were not observed at 4 to 6 Weeks of treatment. Observation 3: Increased Sensitivity in Geriatric Subjects Consistent differences can be seen for geriatric subjects, suggesting an increased risk of adverse CNS-related events with the highest Halcion dose and the longer duration of treatment compared with the risk with the lowest dose and shorter durations (Tables 3-4 and 3-5), and the same is true for flurazepam. Comparing the results for adults (younger and middle-aged adults) in Tables 3-2 and 3-3, there is little evidence of increases in the rate of any of the more than two dozen types of CNS-related adverse events listed for those receiving Halcion, and the same is true for flurazepam. However, the rates (in Tables 3-2 to 3-5) come from different studies and it is difficult to evaluate the significance of the comparisons in terms of possible dose and duration effects. No statistical analyses were presented as an aid in differentiating among the many possible comparisons to focus on those possibly occurring on a more than random basis. Summary Based on a visual inspection of the tabular data, the committee concludes that the currently recommended doses of Halcion (0.125 and 0.25 mg) and the shortest durations of use (1 to 7 days) have a safety profile comparable to those of both placebo and the lowest dose of flurazepam (15 mg), including comparable rates of CNS-related adverse events. The data also suggest that geriatric subjects may be at an increased risk of adverse CNS-related events with the highest Halcion dose and the longest duration of treatment compared with the risk with the lowest dose and shorter durations. IOM Analysis of Upjohn's Integrated Summary of Safety To provide a broader view of the dose-response relation, particularly for the lower doses that are relevant to current labeling (i.e., 0.25 mg for non-geriatric subjects and 0.125 mg for

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data and alprazolam (~20 times), oxazepam (~70 times), and flurazepam and temazepam (>100 times) (see Table B-2). Generally, the affinity ratios fit the relative amounts administered clinically, although the dose of triazolam used in comparison studies is generally relatively greater than that suggested by the affinity ratios based on binding to cortical tissue. For example, consistent with the binding data, 0.5 mg of triazolam appears to be similar to 60 mg of temazepam in terms of sedation (Rush et al., 1993). At these doses, triazolam produces relatively less cognitive impairment than temazepam. However, this pattern is not always found when the ratio of the triazolam dose/temazepam dose is less favorable for triazolam. One complicating factor is that there are several benzodiazepine receptor subtypes with different receptor affinities and affinities for different anatomical locations. The importance of this effect Was illustrated in a study by Sanger and Benavides (1993) (see Table B-3 ). Although the order of affinity of benzodiazepines in the rat cortex and cerebellum is relatively consistent, most benzodiazepines have markedly less affinity for spinal cord benzodiazepine receptors because a different subgroup of receptors is expressed there. In particular, triazolam appears to bind to benzodiazepine receptor subtypes with relatively equal potency. However, zolpidem shows greater potency for subtypes bearing the alpha-1 and gamma-2 receptor subunits (Faure-Halley et al., 1993; Graham et al., 1996; Ramsey-Williams and Carter, 1996). Because triazolam has binding affinity in the brain and spinal cord comparable to those of most benzodiazepines, it is relatively more potent in the spinal cord than most benzodiazepines. The significance of spinal cord benzodiazepine receptors relative to that of brain benzodiazepine receptors is not known. Another important issue in considering the affinity of a drug for its receptors is the issue of active metabolites. Triazolam has two active metabolites, alpha-hydroxy-triazolam and 4-hydroxy-triazolam, both of which have significant affinities for benzodiazepine receptors (Sethy and Harris, 1982; Richelson et al., 1991). Both metabolites are extensively converted to the glucuronide in plasma and are present as the unbound form (the form that can enter the brain) in negligible amounts in plasma (Eberts et al., 1981; Mauri et al., 1993). Although there is no evidence that these metabolites contribute to the acute behavioral effects of triazolam, studies have not ruled out the accumulation of metabolites with long-term administration. In vivo binding data for humans obtained from position emission spectroscopy or single photon emission computerized tomography (SPECT) studies would be helpful for advancing the discussion of differences in drug affinity between triazolam and other drugs in different regions of the brain. However, these studies have not yet been conducted. Lipophilicity The time course of the availability of benzodiazepines to brain benzodiazepine receptors is highly dependent on their lipophilicity, which allows them to cross the blood-brain barrier (Arendt et al., 1987; Miller et al., 1988) (see Table B-4). In this regard, triazolam is moderately liphophilic relative to other benzodiazepines. Midazolam and diazepam appear to be more lipophilic. The lipophilicity of triazolam is comparable to that of lorazepam and alprazolam. Agonist Status Triazolam is a full agonist of benzodiazepine receptors, as are most of the other benzodiazepines studied.

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data Levels in Plasma Two issues are of particular importance with regard to the levels of benzodiazepines in plasma: time to peak concentration and plasma clearance. Peak plasma triazolam levels are reached quite rapidly, but not distinctively so among benzodiazepines. Zolpidem may reach peak levels the quickest, with some studies suggesting that peak levels are reached in. approximately 0.5 hour (Monti et al., 1994). Peak plasma triazolam and flurazepam levels are reached in approximately 1.0 hour (Greenblatt et al., 1989). However, flurazepam is much less potent than its metabolite desalkylflurazepam, which achieves peak levels later. Plasma temazepam levels peak in 1.5 hours, achieving peak levels more slowly than the other drugs mentioned. The rapidity with which peak levels are achieved in plasma is important for the behavioral effects of a drug. It is well known, for example, that more rapid onset is associated with the increased euphoric effects of drugs of abuse, as exemplified by the reduced abuse potential of methadone compared with that of heroin, Similarly, the more rapidly that peak levels are achieved in blood, the more 'rapidly sleep may be induced. Triazolam has a plasma half-life of 2 to 3 hours. This is comparable to that of zolpidem (Greenblatt et al., 1989; Monti et al., 1994). Together, they have the shortest half-lives of the routinely administered hypnotic agents. A short half-life reduces the risk of carryover sedation and cognitive impairment, whereas it increases the risk of adverse events due to withdrawal. It should be noted that the short plasma half-life of triazolam might allow for more time without significant receptor occupancy between doses. Although this has yet to be demonstrated by in vivo receptor methods, if true this quality might be associated with reduced physical dependence with long-term drug administration. Several factors alter the plasma half-life of triazolam. Pharmacokinetic interactions with other medications is a major issue for many medications. Triazolam is metabolized by P-450 CYP 3A4, an hepatic enzyme. Several drugs and foods, including ketoconazole, diltiazem, serotoninergic antidepressants, and grapefruit juice, inhibit this enzyme and produce dose-related increases in the plasma half-life of triazolam (Hukkinen et al., 1995; yon Moltke et al., 1996; Kosuge et al., 1997). One study failed to show the interaction with fluoxetine (Wright et al., 1992). Other drugs, such as rifampin, induce P-450 CYP 3A4 and substantially reduce blood triazolam levels (Villikka et al., 1997). Drug interactions are of significant clinical importance for triazolam, and doses should be adjusted accordingly. Among FDA-approved medications, however, this issue is not unique to Halcion. Other issues influence Plasma clearance. Plasma triazolam levels are doubled in elderly individuals, and therefore recommended doses are reduced for this group (Greenblatt et al., 1991). Cirrhosis does not appear to influence triazolam levels in blood (Robin et al., 1993). Pharmacodynamic Interactions Pharmacodynamic interactions are also important for benzodiazepines and have been shown to affect the response to triazolam as well. Metabolites of progesterone stimulate brain gamma-aminobutyric acid type A receptors. Consistent with this finding, progesterone appears to potentiate the effects of triazolam in postmenopausal women (McAuley et al., 1995). However, it is not yet clear that variability in progesterone and neurosteroid levels during the

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data menstrual cycle are associated with altered sensitivity to triazolam (Rukstalis and de Wit, 1995). Similarly, ethanol comparably potentiates the behavioral effects of both triazolam and zopiclone in humans (Kuitunen, 1994). Caffeine reduces many of the cognition-impairing effects of triazolam in a dose-dependent manner (Rush et al., 1994). Unique Effects of Triazolobenzodiazepines on Locus Coeruleus Neurons The Public Citizen petition highlights potential unique effects of triazolobenzodiazepines upon locus coeruleus activity. The suggestion arises from three principal sources: (1) the initial impression that alprazolam had unique efficacy against panic disorder; (2) data indicating that benzodiazepines inhibited locus coeruleus neuron firing; and (3) clinical data suggesting that benzodiazepines, particularly alprazolam, reduced yohimbine-induced panic in subjects with panic disorder. Given recent findings, these arguments are not as compelling as they may have been in 1992. First, all benzodiazepine anxiolytic compounds (diazepam, lorazepam, clonazepam, and alprazolam) have been shown to be effective in the treatment of panic disorder when adjusting for differences in the potency of each compound. Thus, alprazolam is not unique in this regard (Krystal et al., 1996). Second, the initial reports of benzodiazepine inhibition of locus coeruleus neuron activation indicated rather modest effects that emerged at rather high benzodiazepine doses (Grant et al., 1980; Beck and Fibiger, 1995). It is not clear that triazolobenzodiazepines are uniquely potent in this regard (Laurent et al., 1983; Nakane et al., 1994). In addition, there are questions about whether the effects of benzodiazepines on noradrenergie neuron activity are direct or indirect (Simson and Weiss, 1989). Third, studies by Krystal and colleagues (1996) suggest that triazolobenzodiazepines do not potently block the methoxyhydroxyphenylglycol (MHPG) increases produced by yohimbine. This finding suggests that the benzodiazepine effect is upstream from the locus coeruleus and that the primary effects of benzodiazepine withdrawal would not be via locus coeruleus activation. Summary Triazolam is a potent, broad-spectrum benzodiazepine agonist with a relatively short plasma half-life. Basic animal and clinical studies do not suggest a profile for this medication that is inconsistent with the information presented to FDA at the time of the FDA review in 1992. Its pharmacokinetic and pharmacodynamic profiles are associated with benefits and risks that are particular to the intended use of this medication. The studies reviewed here, however, suggest that most differences between triazolam (Halcion) and other benzodiazepines can be eliminated by manipulating the parameters of drug administration. In humans, when this type of parity is achieved, the effects of triazolam may be indistinguishable from those of other benzodiazepines (Oliveto et al., 1994).

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data Consideration of Amnestic Effects of Halcion The information discussed in this section is summarized in Table B-5. Several levels are considered: expected dose-related anterograde memory impairment, and unexpected amnestic events evident only in subjective reports. Concern that Halcion might uniquely produce the latter type of amnesia was raised (Krystal et al., 1995). Performance of Memory Tasks After Single and Multiple Doses All benzodiazepines have dose-related amnestic effects. These have been demonstrated by impairment in performance of a variety of the memory tasks listed in Table B-5. Several studies indicated that information that could not be recalled after a delay could not be recalled at testing 24 hours later (Greenblatt et al., 1989; Milgrom et al., 1994). Similarly, a nighttime dose of 0.5 mg of Halcion, but not 30 mg of temazepam, had residual amnestic effects in the morning (Bixler et al., 1991). Interpretation of the comparisons of amnestic effects across drugs is difficult because of differences between routinely used doses and doses that actually produce comparable sedative or amnestic effects. This was illustrated by Rush et al. (1993), who suggested that the sedative effects of 0.5 mg of Halcion per 70-kg person are equivalent to those of 60 mg of temazepam per 70-kg person, but that the amnestic effects of Halcion are less than those of temazepam at these doses. However, the sedative and amnestic effects of 0.5 mg of Halcion per 70-kg person are greater than those of 30 mg of temazepam per 70-kg person. The bottom line of these studies appears to be that Halcion produces dose-related impairment in performance of memory tasks. This impairment appears to be roughly comparable to that found after the administration of other benzodiazepines, when adjusting for their relative receptor affinities. When comparing benzodiazepines on the basis of typically prescribed doses, Halcion appears to have greater amnestic effects. This is presumably because typically prescribed doses overestimate the dose of Halcion for comparison purposes. Repeated dosing may differentially affect the amnestic properties of short- and long-acting benzodiazepines. Roehrs et al. (1983) found that single doses of Halcion at 0.5 mg had greater anterograde amnestic effects than flurazepam at 30 mg. However, after 6 days of dosing, the memory impairments associated with flurazepam became progressively worse and equivalent to those of Halcion. The increasing amnestic effects of flurazepam may have been consistent with the accumulation of this drug in the blood with chronic administration. This change did not occur with Halcion, consistent with the absence of drag accumulation with chronic administration. As a result, the report of Roehrs et al. (1983) raises the possibility that long-term treatment with Halcion or other short-acting benzodiazepines might spare subjects the progressive memory impairment that might be associated with long-term treatment with a long-acting medication. Among the publications reviewed, that of Roehrs et al. (1983) was the. only study that directly compared the time course of amnestic effects from short- and long-acting benzodiazepines. This finding awaits replication.

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data Spontaneous Reports of Memory Impairment Of the studies that directly evaluated subjective reports of memory disturbances, the report by Bixler et al. (1991) stands out for consideration. In that study, subjects received 0.5 mg of Halcion, 30 mg of temazepam, or placebo for a total of 5 nights. It is important to note that this 30-mg dose of temazepam is approximately half of the equivalent of the 0.5-mg dose for Halcion. Drug administration was divided into two sessions (initially, either 3 or 4 days of drug treatment, then 2 days of placebo, and then 1 to 2 days of active drug). Five of six subjects receiving Halcion reported daytime episodes of amnesia or subjective memory impairment, no subjects receiving temazepam reported any events, and one subject receiving placebo reported an amnestic event. One other study documented spontaneous reports of memory impairment during chronic treatment with Halcion (Fleming et al., 1990). In that study, 4 of 24 of subjects treated with 7.5 mg of zopiclone and 3 of 24 patients treated with 0.25 mg of Halcion reported memory impairment. These reports are the basis of the concern raised by these investigators in their petition to FDA. The magnitude and frequency of amnestic effects in the study of Bixler et al. (1991) are of concern. However, the high frequency of these severe effects is not evident in the other studies reviewed. Furthermore, the comparison of the low, nontherapeutically equivalent dose of 30 mg of temazepam with 0.5 mg of Halcion may have biased the study of Bixler et al. (1991) against Halcion. Halcion and State-Dependent Learning One report suggests that Halcion facilitates the recall of dissociative experiences in a state-dependent learning model (Weingartner et al., 1995a). State-dependent learning is a term applied to the tendency for the retrieval of learned information to be impaired when the behavioral state while learning (e.g., while on a benzodiazepine) is different from the behavioral state when retrieval occurs (e.g., while off of a benzodiazepine). State-dependent learning effects are generally small. However, they may contribute to the apparent anterograde memory impairment. For example, information learned while traveling after ingesting Halcion might be more difficult to recall when one has not taken Halcion. This hypothesis has yet to be tested. Summary Halcion produces dose-related amnestic effects. Particularly when higher doses are taken at night, these effects may persist into the morning. Depending on one's selection of comparator doses of other drugs, Halcion is either more amnestic than, or as amnestic as other benzodiazepines with comparable levels of benzodiazepine receptor occupancy. The controlled clinical trials do not resolve the frequency of clinically significant amnestic episodes among patients treated with Halcion, particularly at 0.5-mg dose or higher.

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data Review of Data Regarding Possible Anxiogenic or Insomniac Effects Associated with Halcion Administration or Withdrawal The information discussed in this section is summarized in Table B-6. This section reviews reports that suggest that the use of Halcion as a hypnotic agent is associated with (1) an increase in daytime anxiety or (2) withdrawal-related anxiety or sleep disturbance. Halcion Effects on Daytime Anxiety Studies provide conflicting data regarding the possibility that Halcion is acutely anxiolytic (Pinnock et al., 1985; Stopperich et al., 1993). Studies also provide conflicting data on the effects of repeated administration. Some studies suggest that Halcion use as a hypnotic agent reduces daytime anxiety (Mauri et al., 1993; Saletu et al., 1994). Others (Bliwise et al., 1988; Scharf, 1993; Monti et al., 1994) suggest that use of other benzodiazepines or related agents, but not Halcion, as hypnotic agents reduces daytime anxiety. Of most concern are studies that report increased daytime anxiety. Kales et al. (1986) noted a significantly higher rate of ''excitatory events" including nervousness, anxiety, and hyperarousal among six subjects administered 0.5 mg of Halcion compared with the rate of such events among subjects given 15 mg of quazepam (both drugs were given on a short-term basis). Adam and Oswald (1988) reported a 52 percent increase in daytime anxiety on a visual analog scale for subjects treated with 0.5 mg of Halcion but not for those treated with placebo or lormetazepam. These anxiogenic effects are difficult to interpret because the effects of treatment, but not the treatment-time interaction, are significant. The authors do not present the raw data, compare baseline anxiety values between their groups, or use baseline anxiety values as a covariate. As a result, it is possible that their finding reflects a baseline difference between groups. Limited information about baseline differences between their groups also makes it difficult to evaluate the finding that 7 of 40 subjects receiving Halcion, but not those receiving placebo or lormetazepam, had panic attacks during the study. Two other studies reported infrequent anxiety-related adverse events among subjects receiving Halcion, but there is no indication of a greater frequency of adverse events for Halcion than for comparator benzodiazepines (Roger et al., 1993; Monti et al., 1994). Withdrawal-Related Anxiety or Insomnia Following Short- and Long-Term Halcion Use Rebound Anxiety Pinnock et al. (1985) did not find evidence of increased anxiety 6 hours postoperatively in subjects treated with Halcion preoperatively. However, several studies report increases in anxiety following the termination of Halcion administration with short- and long-term treatments (Lee and Lader, 1988). Some studies suggest that withdrawal-related anxiety is not greater for Halcion than for zopiclone (Fleming et al., 1990).

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data Rebound Insomnia Several studies describe increased rates of insomnia following discontinuation of Halcion treatment, even after only a few nights of treatment (Kales et al., 1986; Mamelak et al., 1990; Mauri et al., 1993). In general, the duration of rebound insomnia is limited to the first three nights of discontinuation (Adam and Oswald, 1988; Mouret et al., 1990; Elie et al., 1990; Monti et al., 1994). Several of these studies suggest that the magnitude of impairment with Halcion is greater than that with comparator benzodiazepines (Monti et al., 1994). The level of sleep impairment following termination of Halcion treatment in insomniac subjects does not generally exceed the initial level of sleep impairment (McCluskey et al., 1991). Furthermore, the degree of rebound insomnia following the discontinuation of Halcion treatment. appears to be more closely associated with the magnitude of clinical benefit than the duration of drug exposure in insomniac subjects (Merlotti et al., 1991). These data suggest that a component of the rebound insomnia is a return to the pretreatment level of insomnia. However, other data support the rebound insomnia model. For example, tapering the cessation of Halcion treatment reduces the cessation-related decline in sleep quality (Roehrs et al., 1992). An issue facing the comparison of short- and long-half-life benzodiazepines is the possibility that long-acting benzodiazepines may not produce comparable levels of acute rebound anxiety. The short-half-life agents produce a relatively short period of sleep disruption. In contrast, there may be a less severe, but more protracted, disruption of sleep associated with withdrawal from long-acting benzodiazepines (Kales et al, 1982; Gillin et al., 1989; Mitler et al., 1984). Summary The published literature has documented both anxiolytic and anxiogenic effects in relatively small populations of individuals administered Halcion, Halcion does not appear to be as effective as longer-acting benzodiazepines for reducing daytime anxiety, and it may be associated with substantial increases in anxiety. Clinically significant anxiety increases appear to be relatively infrequent. However, the frequency of these reactions cannot be adequately assessed from the data published in the literature. Similarly, there appears to be increased risk of sleep impairment after the discontinuation of Halcion administration. The frequency of severe or protracted impairment is rare, but it is also impossible to determine this frequency from the data available in the literature. Overall, the data published in the literature do not contradict FDA or IOM analyses discussed elsewere in this report. Ataxia, Disinhibition, and Psychotogenic, Confusional, or Dissociative Effects of Halcion On the basis of data published in the literature, there do not appear to be compelling data singling out Halcion use as a risk factor for falls (see Table B-7). There are insufficient data to base a reconsideration of FDA approval of Halcion on the basis of published data on behavioral dyscontrol following Halcion administration (see Table B-8). The data from published studies consisting of case reports of the emergence of paranoia, hallucinations, or

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data confusion indicate that these conditions are infrequent consequences of Halcion administration (Wehli et al., 1985; Kales et al., 1986; Adam and Oswald, 1988) (see Tables B-9 and B-10). Consideration of Other Potential Adverse Effects The information discussed in this section is summarized in Table B-10. This section reviews several issues. raised in the published literature. Early Termination of Use Two reports indicate higher rates of early terminations of use after Halcion use than after use of comparator drugs (Fleming et al., 1990; Roger et al., 1993). Another report failed to find evidence of excessive terminations after Halcion use relative to that after zolpidem use (Hajak et al., 1994). These data do not permit the development of conclusions regarding an increased incidence of termination of use associated with Halcion use compared with that of other benzodiazepines. Adverse Effects Defined Generally Two studies (Wehli et al., 1985; Fleming et al., 1990) reported increased rates of adverse events in general associated with Halcion use relative to the rates associated with the use of comparator drugs. However, the difference between Halcion and the comparator drugs in the study of Wehli et al. (1985) was limited to mild side effects. Other studies failed to find differences between Halcion and comparator drugs (Roger et al., 1993; Hajak et al., 1994; Jacobsen et al., 1994) or did not note any significant or unexpected adverse effects (Thorpy et al., 1992; Mauri et al., 1993). Other Adverse Effects A recent report suggests an advantage of short- versus long-half-life benzodiazepines regarding driving safety among elderly subjects (Hemmelgarn et al., 1997). That study reported a 50 percent increase in the number of injurious motor vehicle crashes among elderly drivers during the first 7 days of use of long-half-life benzodiazepines compared with the numbers among elderly drivers Using short-half-Fife benzodiazepines or a placebo. The risk remained increased after continuous long-half-life benzodiazepine use for up to 1 year.

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data Summaries and Meta-Analyses In the time since the introduction of Halcion, a growing number of studies have reviewed the safety and efficacy of Halcion. These reviews have supported both its relative safety and efficacy (Greenblatt et al., 1984; Jonas et al., 1992; Klett, 1992; Rothschild, 1992; Mendelson and Jain, 1995; Lobo and Greene, 1997) or concluded that serious questions regarding the safety and efficacy of Halcion remain (Kales et al., 1996). These reports echo two concerns about the published literature: (1) essentially all studies evaluate only a single dose of Halcion or its comparator drug, which makes it difficult to equate Halcion doses with the doses of other medications on the basis of equal potency, and (2) most trials with a single dose of Halcion have used doses that are relatively larger (in terms of anticipated receptor occupancy) than that of the drug with which it is compared. The net. result is that many trials appear to be biased in favor of associating adverse effects with Halcion (Lobo and Greene, 1997). Overall, the reviews do not convincingly support the existence of an unexpected clinical profile for Halcion. Closing Comments The bulk of the published reports related to Halcion's safety were not designed to evaluate the relative frequency of rare, but serious, side effects. A small number of these studies suggest that Halcion use is associated with frequent and serious side effects. The frequency or severity of these side effects is not replicated broadly. in the published studies. However, the presence of these serious side effects warrants the review of other data that might provide a better understanding of these effects, which are of concern. The published studies, however, do not provide convincing evidence that there is associated with Halcion a unique or serious health risk relative to those associated with other benzodiazepines or benzodiazepine-like hypnotic agents. There is the possibility that, relative to other benzodiazepines, individuals receiving Halcion tend to remain on higher doses for longer than the recommended duration (Martinez-Cano et al., 1996). Use patterns may interact with the pharmacologic properties to give rise to increased rates of adverse effects. Even if this is true, however, this conclusion cannot be evaluated on the basis of the published literature. Kales and colleagues (1996) raised the concern that the available data from both controlled trials and SRS might be inadequate to evaluate the effects of Halcion on autobiographical memory (amnestic events). To the committee's knowledge, this phenomenon has not received direct attention in a published study and could be evaluated in a controlled trial. However, the subsequent modification to the labeling of Halcion reflects an integration of reports of serious, but infrequent, adverse effects of this drag into clinical practice. CONCLUSIONS AND RECOMMENDATIONS It is important to note that the conclusions and recommendations are based on a review of the available public information. Various types of data were reviewed and evaluated: (1)

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data randomized, controlled (dose and duration) clinical trials, (2) spontaneous reports of adverse events, (3) survey data, and (4) the published literature. The committee did not review original, raw data or case reports but, rather, the data that were summarized in the New Drug Application and other sources. The committee's analyses were based on these summary data. Clinical Trials and Surveillance The committee is confident in the quality and adequacy of the data from clinical trials (pre- and postmarketing). supporting the safety of using Halcion within the current labeling guidelines. The committee recognizes, however, that the lack of significant adverse events reported from clinical trials appear to conflict with the numbers and types of adverse events (e.g., anterograde amnesia, confusion) that have appeared in the SRS of FDA and in some case reports in the literature. Many factors contribute to this apparent conflict (which is not uncommon among drugs), including the nature and design of clinical trials and external events that can affect the reporting of adverse events. It is important to note that the statistical power to detect rare events is necessarily limited in controlled clinical trials because such trials include a small number of subjects compared with the number of patients using the drug in the postmarketing period, and subjects admitted to the trials must conform to carefully defined inclusion and exclusion criteria, narrowing the likely range of adverse events; rare events are unlikely to be detected in sample populations of a few hundred subjects. In addition, the treatment regimens in these trials are designed to avoid untoward or adverse events that might be expected to occur with higher doses or with dose dependent or duration-dependent use. With respect to surveillance and reports of adverse events, the committee notes that apparent inconsistencies in the data from clinical trials and spontaneous reports are likely to occur for the reasons stated above, and concludes the following: The popularity and consequent widespread use of Halcion produced large at-risk populations from which spontaneous reports of adverse events emerged. Many people take Halcion (and other hypnotic drugs) for more than a year and at dosages above those recommended in the labeling. In general, the types and frequencies of reported adverse events are subject to many external influences, including media attention, marketing, litigation, differential reporting rates, ability to connect drug use to a health event, and other factors, all of which affect the accuracy of interpreting the results. Recommendation. 5: Improve Surveillance, Analysis, and Integration of Findings. The committee recommends that FDA develop improved methods for integrating the findings of clinical trials and postmarketing surveillance, and for resolving discrepancies in the interpretation of data from spontaneous reports, clinical case reports, and controlled clinical trials. This would include the reestablishment of a biostatistics and epidemiology advisory committee (in addition to having biostatistics and epidemiology expertise on the other advisory committees) that

OCR for page 49
Halcion: An Independent Assessment of Safety and Efficacy Data would be charged with the rapid and thorough assessment of the potential health risks suggested by reports of adverse events, identification and resolution of conflicts that may arise in the review of clinical trial and surveillance data, and the provision of expert advice on the maintenance and operation of effective postmarketing surveillance systems.