Sedation with “non-sedating” antihistamines: four prescription-event monitoring studies in general practiceCommentary: Reporting of adverse events is worth the effort
BMJ 2000; 320 doi: https://doi.org/10.1136/bmj.320.7243.1184 (Published 29 April 2000) Cite this as: BMJ 2000;320:1184
Sedation with “non-sedating” antihistamines: four prescription-event monitoring studies in general practice
- Ronald D Mann, professor emeritus (DrMann{at}manorcottage.fsbusiness.co.uk)a,
- Gillian L Pearce, data processing managerb,
- Nicholas Dunn, senior research fellowb,
- Saad Shakir, director bb
- a Southampton University, Southampton
- b Drug Safety Research Unit, Bursledon Hall, Southampton SO31 1AA
- West Midlands Centre for Adverse Drug Reaction Reporting, City Hospital, Birmingham B18 7QH
- Correspondence to: R D Mann, Manor Cottage, Midhurst, West Sussex GU29 9RL
- Accepted 7 February 2000
Abstract
Objectives: To investigate the frequency with which sedation was reported in post-marketing surveillance studies of four second generation antihistamines: loratadine, cetirizine, fexofenadine, and acrivastine.
Design: Prescription-event monitoring studies.
Setting: Prescriptions were obtained for each cohort in the immediate post-marketing period.
Subjects: Event data were obtained for a total of 43 363 patients.
Main outcome measure: Reporting of sedation or drowsiness.
Results: The odds ratios (adjusted for age and sex) for the incidence of sedation were 0.63 (95% confidence interval 0.36 to 1.11; P=0.1) for fexofenadine; 2.79 (1.69 to 4.58; P<0.0001) for acrivastine, and 3.53 (2.07 to 5.42; P<0.0001) for cetirizine compared with loratadine. No increased risk of accident or injury was evident with any of the four drugs.
Conclusions: Although the risk of sedation was low with all four drugs, fexofenadine and loratadine may be more appropriate for people working in safety critical jobs.
Introduction
Antihistamines are often used to treat the symptoms of allergies such as seasonal and perennial allergic rhinitis and urticaria. The first generation antihistamines have been associated with side effects, particularly sedation.1 Second generation antihistamines are therefore favoured over the first generation drugs, not because of greatly improved efficacy but because they have fewer side effects, especially sedation.2–4
Although the second generation antihistamines are known to all have similar efficacy,3 the extent of their sedative effects is not well established. To further examine the sedative effects of four commonly prescribed antihistamines—loratadine, cetirizine, fexofenadine, and acrivastine—we analysed the results of four non-interventional observational cohort studies of these drugs performed by the Drug Safety Research Unit. These studies correlated prescriptions issued in general practice with events reported by the patients to their general practitioners after the drug was dispensed. By monitoring these events in a substantial population of allergy sufferers, without the restrictions imposed by clinical trials methodology, it was possible to measure differences in side effects between these drugs.
Methods
The methods of prescription-event monitoring have been previously described in detail.5 In brief, the general practitioner writes a prescription which the patient takes to the pharmacist. The pharmacist sends all these prescriptions to the Prescription Pricing Authority, which under conditions of full confidentiality, provides electronic copies of the exposure data to the Drug Safety Research Unit. After three, six, or 12 months, “green form” questionnaires are sent to the general practitioners who wrote the original prescriptions. These questionnaires seek to determine any event experienced by patients while they were taking the drug and for a period afterwards. General practitioners are also asked to indicate whether the event was considered to be related to the drug, although they are not required to make this connection. Additionally, the prescribers are asked to indicate whether the drug has been stopped and, if so, the reason for this. All reported pregnancies are followed up to determine the outcome and the cause of all deaths are established. Both the exposure (prescription) and the outcome (event) data are computerised for analysis.
Statistical analysis
The number of events observed during the treatment period in each individual patient is recorded and the incidence density for each event is calculated using the equation:
The incidence density is the measure of the number of reports of each event per thousand patient-months of exposure to the drug. We calculated incidence densities for various time intervals: the first month of exposure (ID1), during months 2-6 (ID2), and during all months of treatment (IDA). The difference between the incidence density in the first month and that in the second to sixth months (ID1−ID2) and the 99% confidence interval for this difference were also calculated. Incidence densities were calculated for all of the events reported, to give an indication of which events were reported significantly more frequently in the first month of exposure. We calculated non-adjusted and age and sex adjusted odds ratios for drowsiness or sedation for fexofenadine, cetirizine, and acrivastine using loratadine as baseline.
Ethical considerations
Prescription-event monitoring is a form of pharmacovigilance, an exercise which has its legal basis in European Union directives 65/65 and 75/319 and in regulation 2309/93. The method of study (records only research) also complies with the guidelines on the practice of ethics committees in medical research involving human subjects issued by the Royal College of Physicians of London in August 1996.
Results
The data collection periods for the four drugs were May to August 1989 for cetirizine and loratadine, May 1989 to September 1990 for acrivastine, and March to August 1997 for fexofenadine. The response rates (number of green forms returned/number of green form sent) were 50.7% for loratadine, 50.9% for fexofenadine, 56.5% for acrivastine, and 57.4% for cetirizine.
Table 1 gives the age and sex distribution of patients treated with each antihistamine. The demographics of each cohort were roughly similar. A higher proportion of women than men were prescribed antihistamines, and younger people were more likely to receive the drugs than elderly people.
Number (percentage) of patients treated with antihistamines according to age and sex
Figure 1 shows the most frequently reported events for loratadine in the first month of treatment and corresponding values for the other antihistamines. The differences between the antihistamines in the incidence density of events classified as “drowsiness or sedation” are further investigated in table 2. The unadjusted and age and sex adjusted odds ratios show that loratadine and fexofenadine are associated with a lower incidence of sedation than acrivastine and cetirizine (table 3). Since sedation may result in an increased risk of other events such as accident and injury, we analysed the incidence density of these events in the first month of treatment (fig 2). There was no increased risk of accident or injury with any of the four drugs.
Eight most commonly reported events for loratadine in first month of treatment and corresponding incidence densities for acrivastine, cetirizine, and fexofenadine
Incidence densities and number of reports of sedation with four antihistamines
Odds ratios for the risk of drowsiness and sedation associated with antihistamines
Incidence density of events related to sedation in the first month of treatment for four antihistamines
Discussion
It has been recognised for over 30 years that drug safety depends not only on preclinical studies but also on post-marketing surveillance.6 Post-marketing prescription-event monitoring studies observe large cohorts and aim to provide data on around 10 000 patients for each drug. The data consist of the real life experiences of patients, who are often taking concomitant drugs for other conditions. To date, 65 prescription-event monitoring studies have been undertaken with a mean cohort size of 11 055 patients.
Prescription-event monitoring has various strengths and weaknesses.5 The method is non-interventional and does not interfere with general practitioners' decisions about the most suitable treatment for a patient. It therefore avoids the selection bias inherent in clinical trials. It is carried out on a national scale and is representative of the whole population using the drug. As all events are monitored, the technique can pick up trends in events that might not be considered to be related to the drug by doctors seeing individual patients. Additional information, such as use during pregnancy, can be monitored, and data on the use of certain drugs during the first trimester of pregnancy have been published.7 A disadvantage of the method is that it relies on general practitioners returning completed green forms. Thus there may be a bias caused by the lack of data from non-responders. However, there is no reason to suppose that any such bias would be different for the drugs compared in this study. Even though the data collection period for fexofenadine was later than for the other drugs, we are not aware of any publicity that might have affected the reporting of sedation. Also, there is unlikely to be any hidden confounding of these results, since all the drugs are prescribed for well defined, similar, indications. Adjustment for age and sex did not greatly alter the odds ratios.
Sedative effects of antihistamines
The number of reports of sedation with all four antihistamines was low. However, the adjusted odds ratios suggest that cetirizine was 3.5 times more likely and acrivastine 2.8 times more likely to result in reports of sedation than loratadine; there was no significant difference between loratadine and fexofenadine. Sedation might result in an increased risk of accident and injury, but we found no such difference between the antihistamines.
The second generation antihistamines are difficult to separate in terms of efficacy. Previous investigations have shown the potential cardiotoxic effects of astemizole, ebastine, and terfenadine which can, in serious cases, result in torsade de pointes.3 4 Fexofenadine, a metabolite of terfenadine, has also been connected with cardiotoxicity,8 although other reports suggest that this is not the case.9 However, no cardiotoxic events of relevance were noted for any of the antihistamines in this study.
Our findings suggest that in situations in which even very infrequent reports of sedation are undesirable (for example, when prescribing for flight crew) loratadine or fexofenadine are preferable to acrivastine or cetirizine.
What is already known on this topic
Second generation “non-sedating” antihistamines are usually considered to be equivalent in efficacy but their sedating properties are less clear
Prescription-event monitoring is a well established method of recording events experienced after routine prescription of drugs
What this study adds
Loratadine and fexofenadine resulted in a significantly lower incidence of sedation than cetirizine and acrivastine
No cardiotoxic events of relevance were noted for any of the four antihistamines studied
Acknowledgments
We thank all the general practitioners who completed and returned their green forms. Without their cooperation this investigation would not have been possible.
Contributors: RDM first noticed the relevant differences in reporting rates, did the initial analyses, wrote the first draft of the paper, and is the guarantor. GP was responsible for the data processing assisted by Gillian Pearce. ND undertook the final and independent statistical analysis. SS was responsible for the discussions balancing the statistical and clinical perspectives of the study
Footnotes
-
Funding The Drug Safety Research Unit is a registered charity, number 327206. It receives unconditional grants from a number of pharmaceutical companies. These companies have no say in the conduct of the studies and have no statistical or editorial control over analysis or reporting of results.
-
Competing interests The unit has received funds from Hoechst Marion Roussel and Schering Plough but there has been no external funding of the comparison reported in this paper.
Commentary: Reporting of adverse events is worth the effort
- R E Ferner, director.
- a Southampton University, Southampton
- b Drug Safety Research Unit, Bursledon Hall, Southampton SO31 1AA
- West Midlands Centre for Adverse Drug Reaction Reporting, City Hospital, Birmingham B18 7QH
The paper by Mann et al relies on adverse event monitoring. Adverse events are undesirable things that happen to patients and include adverse reactions caused by drugs. Some schemes are limited to adverse drug reactions. The United Kingdom Committee on Safety of Medicines, for example, collects and analyses about 20 000 yellow cards each year. They have proved valuable in detecting adverse reactions,1 and the system covers all drugs throughout their use and in all patients. It depends, however, on a reporter suspecting a reaction and having the confidence and time to commit the suspicion to paper.
Schemes that monitor events rather than reactions remove the need for individual practitioners to assess whether a relation might be causal. The Southampton “green form” scheme is a method of post-marketing surveillance in general practice.2 It aims to discover adverse events occurring in patients prescribed a selected new drug during a particular period of observation. The scheme is notified by the Prescription Pricing Authority of every prescription of the drug, and after a time sends a form to the prescriber, asking for notification of any adverse events reported by the patient to them. About half the cards sent out are returned. The collected data for a specific drug and event are used to calculate incidence density (the ratio of the number of reports of the event during treatment to the number of patient-months of exposure to the drug).2 The relative risk of a given event across a group of drugs is estimated by comparing incidence densities.
The observational studies are, of course, not randomised and so may be biased.3 For example, one preparation may be particularly used for a specific condition or age group, or there may be differential failure to detect, record, or report an adverse event. Estimates should be viewed with circumspection even if bias seems unlikely—several incidence ratios for many different events can be examined, and some will inevitably differ from others because of random variation. The associated significance will be misleading unless correction has been made for multiple comparisons. Results also have to be set in clinical context. We should certainly be reassured by the low overall incidence of sedation with selective histamine H1antagonists shown by Mann et al: fewer than one patient in 140 complained of drowsiness with any of these drugs. There are some differences between them, which may be relevant to people in safety critical jobs.
As the authors point out, the investigation would not have been possible without the cooperation of the reporters, who should be encouraged by seeing that their efforts are worth while. In due course, computerised systems such as the general practice research database may allow post-marketing surveillance without tears.4 5 Until then, we will all benefit from the information that diligent reporters send on yellow cards and green forms.
Footnotes
-
Competing interests The West Midlands Centre for Adverse Drug Reaction Reporting receives financial support from the Medicines Control Agency.