2. METHOD

Computer-assisted searches of bibliographical data bases were conducted to identify scientific publications for the initial review. Specifically, MEDLINE and related search engines were used to identify well-designed human studies investigating the behavioral, cognitive and sedative effects of antihistamines. Search terms included: antihistamines, H1-antagonists, psychomotor performance, driving, performance impairment, and cognitive effects. Publications through the end of 1998 were included; no particular date limit was set regarding earlier publications, although it should be noted that MEDLINE typically does not include publications prior to 1966. This primary computer-assisted search was supplemented by review of the references cited in the retrieved publications as well as consideration of reports of pertinent studies known to the authors. Therefore, in addition to published journal articles, some abstracts, proceedings, and reports of conference presentations also were included. Although an extensive literature search was conducted, the results cannot be viewed as exhaustive.

The titles (or abstracts) of the identified references were evaluated for initial inclusion according to the following criteria: the article (or detailed abstract) was available in English, the study tested healthy human subjects (or allergy patients), measures included driving-related performance tasks, antihistamines were administered in an experimental setting, a placebo control treatment was included, and statistical tests compared the treatment(s) to placebo. All publications appearing to meet these initial criteria were indexed as the master reference set and copies of the articles were sought for intensive review. This master set included 386 references selected from more than 500 titles/abstracts reviewed in the initial focused search. Of the 386 references identified, 256 were excluded from the intensive review and analysis for the following reasons, as shown in Table 1 below:

The remaining 130 publications which met all inclusion criteria were then subjected to intensive review and the findings were coded and entered into the data base for summary and analysis. The complete citations for this final set of 130 publications appear as REFERENCE LIST B at the end of this report. Originally, 132 articles were deemed appropriate for the intensive review and so they are indexed in the data base (and in all appended Tables and listings) as Reference Numbers 1-132 in alphabetical order by first author. Subsequently, four of these articles were excluded from the review set (Ref# 10, 52, 72, and 118) and two additional publications (published in late 1998) were identified and added to the data base. However, to avoid major recoding and reorganization of the data base, the two added articles simply were indexed as Reference #133 (Comer et al., 1998) and #134 (Scavone et al., 1998). As such, they appear at the end of the data base and reference list, rather than in alphabetical order.

A complete listing of the individual studies, with impairment findings grouped according to the behavioral skill categories (discussed in detail below), is presented in Appendix A. In addition, for each task category, an overall summary table of “Skills performance impairment as a function of antihistamine (Drug/Dose), task category and dosing (Acute/Repeated)” was generated to present the counts of YES and NO for significant impairment. An example of such a “YES/NO Counts” table is presented in Appendix B. The tables in Appendix A and B also summarize the findings by drug generation as a class.

Details of the data base coding system can be found in an example of one of the individual Study Summary Sheets which were generated for all 130 articles (see Appendix C for an example). In brief, each article was reviewed and the information for the Citation, Method, and Results of each reported study was entered in the data base. Of note, seven publications reported more than one experiment; in these cases, the data base includes the single Citation, but separate Methods and Results sections for each of the studies which are indexed by the single Reference number plus a letter; (e.g., Reference #18 reports two separate studies: these are indexed as Reference #18A and #18B). The 130 publications reflect a total of 138 separate studies; these are included in the data base for this review.

The results from each study were coded, at the level of drug dose and task measure, for evidence of significant impairment, i.e., YES or NO. With few exceptions, “significant” means that the study reported a statistical test of the given drug's dose versus placebo at p < 0.05.

Nearly 40 different antihistamines were represented in the master data set of publications which were identified in the initial, focused literature search. In many cases, only a few studies (and sometimes only one study) examined a given drug, and many of the drugs are (were) only available in Europe. Consequently, in order to ensure an adequate sample of studies for this review, and to be relevant to the medications available to the U.S. population of drivers, we decided to focus only on the five most widely prescribed and/or studied drugs from each generation. These 10 drugs are described in detail in Table 2, as shown on the next page.

A table which lists all of the studies in this review, presented with the YES/NO codes across all 10 drugs, is presented in Appendix D. This listing provides a concise overview of the specific drugs examined in a given study. Of note, the majority of studies focused on only one of the 10 drugs. Only 12 studies involved a comparison of two different 2nd -generation drugs, and only a single study (Simons, 1996) examined a group of 2nd -generation drugs in comparison to placebo and to a 1 st -generation antihistamine as the positive control (e.g, diphenhydramine).

TABLE 2. THE TEN DRUGS SELECTED FOR REVIEW

Each drug was indexed in the data base and in all generated figures and listings by a drug code number: D1-D5 and N1-N5, respectively, reflect the five drugs in the 1st -, and 2nd -, generations. Only one study of fexofenadine's effects on driving-related behavior has been published to date (i.e., through the end of 1998 in this review). Its inclusion in this review, however, is warranted by its current status as one of the most widely prescribed antihistamines and the fact that its chemical structure is identical to terfenadine's active metabolite, except that fexofenadine is the hydrochloride salt. Terfenadine was taken off the market in early 1998 after increased reports of cardiovascular adverse effects. Nonetheless, as the parent drug to fexofenadine, the many studies of terfenadine are included in this review for their continued relevance for understanding the drug mechanism and impairment effects of the 2 nd generation drugs. In addition, astemizole recently was removed from the market due to safety concerns.

In addition to the drug coding, the results for each study were entered in the data base according to the planned analysis of the ten behavioral skill categories, as shown in Table 3 below. In addition, as noted earlier, subjective measures of sedation were analyzed if the study also had tested at least one behavioral or cognitive measure.

Table 3. NUMBER OF STUDIES AND TEST FINDINGS FOR EACH SKILL CATEGORY
AND SUBJECTIVE SEDATION: ACUTE (A) AND REPEATED (R) DOSING

Note: A = ACUTE Doses; R = REPEATED Doses ; (excluded 3 studies with only Residual effects). Many studies tested more than one skill category, measure, drug, and dose level and schedule.

It should be noted that the terms “test” or “finding” are used interchangeably in this report to describe the unit of data analysis for this comprehensive review of 138 studies. A given study, for example, may have evaluated several drugs and doses, both acute and repeated dosing schedules, and included multiple behavioral measures and subjective measures. The resultant total number of specific “tests” or “findings” from that single study, therefore, would be the product of multiplying all the levels for each factor studied. Thus, as shown in the table above, there is a total of 1,261 test findings; obviously this number is much greater than the total number of studies included in the review. Also, the number of findings for repeated dosing was rather limited (n=323) compared to the greater number for acute dosing (n=938).

The decision for classifying the many performance measures into 10 behavioral categories is admittedly somewhat arbitrary. Prior reviewers also have noted the difficulties inherent in this process of assigning a given task to a specific category (e.g., Adelsberg, 1997; Rombaut & Hindmarch, 1994), but most concur with the general areas of actual driving, simulated driving, various psychomotor skills, sensorimotor tasks, cognitive effects, and subjective measures of sedation. In order to evaluate more precisely the drug effects on the wide variety of measures, we also included sub codes in an effort to restrict the variability of findings within a given area. Specific task names and the individual response measures can be found in the detailed tables which list the impairment results by study (see Appendix A).