ADVERTISEMENT

Issue Navigator

Volume 13 No. 02
Earn CME
Accepted Papers





Scientific Investigations

Spontaneous Adverse Event Reports Associated with Zolpidem in the United States 2003–2012

Carmen K. Wong, BPharm1; Nathaniel S. Marshall, PhD2,3; Ronald R. Grunstein, MD2; Samuel S. Ho, BPharm1; Romano A. Fois, PhD1; David E. Hibbs, PhD1; Jane R. Hanrahan, PhD1; Bandana Saini, PhD1,2
1Faculty of Pharmacy, The University of Sydney, Sydney, Australia; 2NHMRC Centre for Integrated Research and Understanding of Sleep (CIRUS) and NeuroSleep Centre, Woolcock Institute of Medical Research, The University of Sydney and Sydney Local Health District, Sydney, Australia; 3Sydney Nursing School, The University of Sydney, Sydney, Australia

ABSTRACT

Study Objectives:

Stimulated reporting occurs when patients and healthcare professionals are influenced or “stimulated” by media publicity to report specific drug-related adverse reactions, significantly biasing pharmacovigilance analyses. Among countries where the non-benzodiazepine hypnotic drug zolpidem is marketed, the United States experienced a comparable surge of media reporting during 2006–2009 linking the above drug with the development of complex neuropsychiatric sleep-related behaviors. However, the effect of this stimulated reporting in the United States Food and Drug Administration Adverse Event Reporting System has not been explored.

Methods:

Using disproportionality analyses, reporting odds ratios for zolpidem exposure and the following adverse events; parasomnia, movement-based parasomnia, nonmovement-based parasomnia, amnesia, hallucination, and suicidality were determined and compared to all other medications in the database, followed by specific comparison to the benzodiazepine hypnotic class, year-by-year from 2003 to 2012.

Results:

Odds ratios were increased significantly during and after the period of media publicity for parasomnias, movement-based parasomnias, amnesias and hallucinations. We also observed that zolpidem adverse drug reaction (ADR) reports have higher odds for parasomnias, movement-based parasomnias, amnesias, hallucinations, and suicidality compared to all other drugs, even before the media publicity cluster.

Conclusions:

Although our results indicate that zolpidem reports have higher odds for the ADR of interest even before the media publicity cluster, negative media coverage greatly exacerbated the reporting of these adverse reactions. The effect of such reporting must be borne in mind when decisions around drugs which have been the subject of intense media publicity are made by health professionals or regulatory bodies.

Citation:

Wong CK, Marshall NS, Grunstein RR, Ho SS, Fois RA, Hibbs DE, Hanrahan JR, Saini B. Spontaneous adverse event reports associated with zolpidem in the United States 2003–2012. J Clin Sleep Med. 2017;13(2):223–234.


INTRODUCTION

Zolpidem, a hypnotic binding at the benzodiazepine receptor site, was initially marketed in the United States in 1992 and in Australia at the end of 2000, for the short-term treatment of insomnia. Zolpidem was proposed to possess an improved safety profile with minimal residual effects and lower potential for physical tolerance and dependence.1,2 Due to the favorable pharmacokinetic profile of zolpidem,3 it quickly became the third most widely prescribed hypnotic in Australia.4 However, the assumed safety benefits of zolpidem were quickly challenged following extensive news and media publicity linking the agent with a variety of potentially dangerous sleep-related behaviors.2

BRIEF SUMMARY

Current Knowledge/Study Rationale: The behavioral influence of the media on health scares and stimulated reporting have been studied for a number of medications including the contraceptive pill, vaccinations, paroxetine, and triazolam; however, the impact and consequences of media publicity have been observational and have yet to be rigorously analyzed using robust quantitative pharmacovigilance disproportionality measures.

Study Impact: To our knowledge, this is the first report that thoroughly explores the extent of stimulated reporting of adverse drug reactions associated with zolpidem exposure in the United States Food and Drug Administration Adverse Event Reporting System (FAERS) database using appropriate disproportionality analyses. Analysis of the FAERS data can provide a better profile of the level of adverse events associated with zolpidem and showcase the effect of media publicity during 2006–2009 on adverse event reporting. Drug regulatory bodies should consider this notoriety bias when making drug-related decisions based on adverse event risk.

Zolpidem has been the subject of repeated media publicity about adverse events including bizarre and potentially dangerous movement-based parasomnias such as sleepwalking,5 sleep-eating,6 and sleep-driving.7 In particular, following the implication of zolpidem in high profile cases, a prominent media publicity cluster occurred associating zolpidem with a number of complex neuropsychiatric adverse reactions. This cluster or collection of news and media coverage predominantly concentrated on the development of parasomnias, amnesia, hallucinations, and suicidality following zolpidem exposure.4 World-wide zolpidem adverse drug reaction (ADR) reports collated by the World Health Organization (WHO) (Figure S1 in the supplemental material) suggest that compared to most of the world, Australia and the United States were significantly affected by media coverage with an apparent surge in the reporting of adverse reactions.8

Although media coverage remains an important source of information on health and pharmaceuticals, there is growing realization that the media can potentially influence the behaviors and perceptions of consumers and health care professionals.9 In particular, there is increasing evidence that intense negative coverage of a medication including scare stories can dramatically affect drug use, and bias ADR reporting in spontaneous reporting systems (SRS).10 While SRS are an invaluable tool in post-marketing pharmacovigilance, there are a number of recognized limitations, including under-reporting,11 the Weber effect,12 and notoriety bias.11 The Weber effect describes an epidemiologic phenomenon whereby the number of ADR reports for a newly approved drug peaks at the end of the second year of marketing and then declines rapidly despite increases in prescribing rate.13 Stimulated reporting or notoriety bias occurs when patients and healthcare professionals are influenced or “stimulated” by media publicity including regulatory alerts to report specific drug-related adverse reactions through a “bandwagon effect.”4,14 With an increasing number of countries accepting direct ADR reports from consumers, it is inevitable that extensive media coverage can potentially bias and distort drug signals.4,11 Several drugs have been the subject of stimulated reporting including the contraceptive pill,15 vaccinations,16 paroxetine,17 and triazolam.18 In a recent study that explored the effect of zolpidem associated stimulated reporting using the Australian Therapeutic Goods Administration (TGA) Database of Adverse Event Notifications (DAEN), we observed an increased risk in the development of bizarre sleep-related behaviors (parasomnia, amnesia, hallucination, and suicidality) with zolpidem exposure even before the media publicity cluster.4 Importantly, the study also demonstrated that negative media coverage markedly increased the reporting odds ratios for parasomnia and amnesia indicating the effect of intense media reporting of adverse events related to zolpidem.4

In early 2006, the United States experienced substantial media interest surrounding zolpidem induced complex sleep-related behaviors in conjunction with a class-action lawsuit and the high-profile vehicle collision of Congressman Patrick Kennedy.4 These incidents prompted the United States Food and Drug Administration (FDA) to issue warnings and safety alerts in 2007 regarding the development of these neuropsychiatric reactions post-zolpidem use. The death of Australian actor Heath Ledger, which was falsely reported to be associated with zolpidem, further intensified the controversy surrounding the hypnotic agent in both the United States and Australia.19,20

Similar to the Australian TGA DAEN, the FAERS database contains voluntary reports submitted directly from health-care professionals, consumers, and manufacturers on adverse events and medication errors. As such, the FAERS database is also highly susceptible to stimulated reporting, but currently, the effect of this stimulated reporting on signal generation in the FAERS has not been explored. With reports submitted from the United States and other countries, FAERS is a much larger database than DAEN, containing well over two million ADR reports compared to 70,000 ADR reports in our previous analysis.4 Analysis of the FAERS data can provide a better profile of the level of adverse events associated with zolpidem and showcase the effect of media publicity on adverse event reporting. In addition, compared to Australia, the use of zolpidem in the United States is far more widespread,21 with cases of zolpidem sleep-related neuropsychiatric reactions subjected to relatively lower media exposure, both in terms of the amount of and the tone of the publicity. Given the lower force of media presentations about zolpidem in the United States, the FAERS database would thus be more sensitive to drug-event association (DEA) signals. Hence the aim of this study was to quantify the association between zolpidem exposure and the reported number of adverse events year-by-year before (2003–2005), during (2006–2009), and after media publicity (2010–2012) using the FAERS database.

METHODS

Data Source

The FAERS database is publicly available and quarterly data files can be downloaded from the FDA website (http://www.fda.gov/). Each file contains raw data extracted from the FAERS database, available in either ASCII or SGML formats. These files include information on patient demographics and administrative information including the country of origin of the report, drug and reaction information, patient outcome information, and information on the source of the reports.22 Reported events are stored in a relational database, and quarterly data files were combined for the time period between January 1, 2003, and August 31, 2012. An initial analysis for the time period between January 1, 2000, to August 31, 2012, was undertaken following anecdotal reports of automatism following zolpidem intake occurring before 2003. Year-by-year un-adjusted odds ratio for the ADRs of interest are presented in the supplemental material (Figure S2). Adverse drug reactions were coded by medical officers using the preferred term (PT) terminology of the Medical Dictionary for Regulatory Activities (MedDRA). The structure of the FAERS database has previously been described in detail.23

Data Preparation

To address the issue of duplicate reports and the lack of standardization of drug names, we performed a manual and detailed reorganization of drug name variants in the database.23 The open source software, OpenRefine (previously Google Refine; version 2.5) was utilized to organize and standardize medication brand names and common misspellings to reflect their relevant active ingredients.23 All active ingredients were defined according to the World Health Organization (WHO) Anatomical Therapeutic Chemical (ATC) classification. Drugs that could not be standardized according to this classification or were not recognized as a medicinal product were left verbatim.23

Duplicates and follow-up reports, defined by the FDA as those with the same case number, were identified and merged. Records were excluded when details were absent for patient age, sex, drug exposure, or adverse reaction.

Case/Non-Case Method for Signal Detection

Case/non-case methods were used to generate signals for DEAs using disproportionality analyses of SRS data. For the purposes of analyses, cases were operationalized as reports that included any preferred term (using MedDRA version 12.1) within each of the following categories of ADR: amnesia, hallucination, parasomnia, and suicidality. Parasomnia preferred terms were further reclassified by the authors (CW, RF, and DH) into two distinct categories of movement-related parasomnia and nonmovement-related parasomnia (Table 1). As movement-based parasomnias encompassed dramatic events of sleepwalking, sleep-driving, and sleep-eating which were the focus of the media cluster, differentiation of parasomnia preferred terms into the two categories allowed us to determine whether differences between these groups were evident.

Adverse drug reaction categories of interest and preferred terms defining cases.

jcsm.13.2.223.t01.jpg

table icon
Table 1

Adverse drug reaction categories of interest and preferred terms defining cases.

(more ...)

Using the case/non-case methodology, the reporting odds ratio (ROR) can be calculated with its 95% confidence interval (95% CI) as a measure of disproportionality. The statistical null hypothesis is that there is no difference in the incidence of reported exposure to a drug of interest between cases and non-cases. Patients with ADRs within a category of interest were more likely to have been exposed to the drug of interest compared to the patients without ADRs in that category, if the lower limit for the 95% CI of the ROR was found to be greater than one. This method was applied to investigate the association after control for a number of covariates.

Exposure Definition

Exposure was defined as the presence of zolpidem in a report, regardless of whether it was suspected of causing the reaction. A number of covariates, including potential confounders were also included as exposure categories. These include demographic variables (age: stratified as pediatric [< 18 years], working age [18–65 years], geriatric [> 65 years] and sex) and drug classes such as anticholinergics, benzodiazepines, selective serotonin and other reuptake inhibitors, tetracyclic and tricyclic antidepressants, antipsychotics, antiepileptics, and a variety of medications associated with the induction of the ADRs of interest (Table S1 in the supplemental material). Drug classes identified as potential exposure categories were established through extensive searches of the MIMS database and reviewed by a clinical team (CW, SH, RF, JH, and BS) for appropriateness.

Statistical Analysis

Statistical analyses were conducted using SAS Enterprise Guide (version 6.1; SAS Institute Inc., Cary, NC, USA). Reporting odds ratios and 95% CIs were determined for each exposure category using univariate logistic regression analysis. Covariates for which RORs were significant were included in multivariate logistic regression analyses.

Year-by-year analyses were conducted for each ADR category by stratifying the records on the basis of reporting for all individual years from 2003 to 2012 and determining adjusted annual RORs and 95% CIs for zolpidem exposure. DEA signals were analyzed over the 10-year period, and we also tested whether there was an interaction between the 3 distinct reporting periods of interest (pre- [2003–2005], during- [2006–2009], and post-media publicity [2009–2012]) for each category of ADR using multivariate logistic regression analyses. If these interactions demonstrated statistical significance, RORs and 95% CIs (adjusted for covariates) for zolpidem exposure for each ADR category were calculated separately for the 3 distinct periods.

A number of sensitivity analyses were also conducted. Firstly, we repeated our analyses using a subset of data consisting of single-drug reports subsequently eliminating the confounding effects of bystander drugs. Secondly, a series of annual multivariate analyses were conducted comparing zolpidem to the benzodiazepine class of hypnotics to control for confounding by indication. All analyses conducted were adjusted for potential exposure categories where possible.

From July 2005, the reporter country was identified in FAERS. This enabled us to repeat the multivariate year-by-year DEA analysis to determine the extent of stimulated reporting in the United States only, for the period July 2005 to August 31, 2012.

RESULTS

The FAERS dataset between January 1, 2003 and August 31, 2012 comprised 2,131,368 reports for analysis.

Univariate Analyses

Demographic and drug-class exposure characteristics of cases and non-cases for each ADR category are presented in Tables 2 and 3. Movement-based parasomnias, hallucinations, and suicidality were reported more often by males, while non-movement-based parasomnias and amnesias were associated more commonly with females. Working age was associated with nonmovement-based parasomnias and amnesias when compared to the pediatric group.

Demographic and drug-class exposure characteristics of cases and non-cases for parasomnia, movement-based parasomnia, and nonmovement-based parasomnia.

jcsm.13.2.223.t02.jpg

table icon
Table 2

Demographic and drug-class exposure characteristics of cases and non-cases for parasomnia, movement-based parasomnia, and nonmovement-based parasomnia.

(more ...)

Demographic and drug-class exposure characteristics of cases and non-cases for amnesia, hallucination, and suicidality.

jcsm.13.2.223.t03.jpg

table icon
Table 3

Demographic and drug-class exposure characteristics of cases and non-cases for amnesia, hallucination, and suicidality.

(more ...)

Multivariate Analyses

The ORs for zolpidem exposure for each ADR category of interest over the 10-year period, adjusted for patient demographics and significant drug-exposure covariates are presented in Table 4. From the table, a significant association between zolpidem exposure and each of the ADRs of interest is observed even after adjusting for patient demographics and signifi-cant drug-exposure covariates. Signals for movement-based parasomnias, in particular, exhibit a marked association with odds ratios of 34.39 (95% CI, 31.39–37.68) and 35.20 (95% CI, 31.65–39.14), adjusted for patient demographics and drug-exposure covariates, respectively.

Odds ratios for zolpidem exposure for each adverse drug reaction category of interest, adjusted for patient demographics and significant drug-exposure covariates.

jcsm.13.2.223.t04.jpg

table icon
Table 4

Odds ratios for zolpidem exposure for each adverse drug reaction category of interest, adjusted for patient demographics and significant drug-exposure covariates.

(more ...)

Annual adjusted ORs with corresponding epidemic curves are illustrated graphically in Figures 1 and 2. Significant ORs for all ADRs of interest with the exception of nonmovement-based parasomnias were observed even prior to the media publicity. During the period of intensified publicity around zolpidem (2006 to 2009), signals were significantly increased for parasomnias (odds ratio, 7.65; 95% CI, 6.98 to 8.39), movement-based parasomnias (odds ratio, 64.92; 95% CI, 56.19 to 75.00), and amnesias (odds ratio, 4.29; 95% CI, 3.97 to 4.64); while signals for nonmovement-based parasomnias (odds ratio, 1.44; 95% CI, 1.22 to 1.70) and hallucinations (odds ratio, 2.05; 95% CI, 1.84 to 2.28) were marginally increased, compared to the pre-publicity period. Post-media publicity, DEA signals for zolpidem for all ADRs except suicidality decreased; although signals in this period remained elevated in comparison to the pre-publicity phase. The trend for reporting suicidality remained unaffected by media coverage. Additionally, adjusted ORs stratified by the 3 distinct periods; pre-, during, and post-media publicity reflect the quantitative DEA signal alterations over the 10-year period.

Odds ratios and epidemic curves for zolpidem exposure when compared to all other medications.

Adjusted reporting odds ratios (and 95% confidence intervals) (left) and epidemic curves (right) for (A) parasomnia (B) movement-based parasomnia (C) nonmovement-based parasomnia. * = p < 0.05.

jcsm.13.2.223a.jpg

jcsm.13.2.223a.jpg
Figure 1

Odds ratios and epidemic curves for zolpidem exposure when compared to all other medications.

(more ...)

Odds ratios and epidemic curves for zolpidem exposure when compared to all other medications.

Adjusted reporting odds ratios (and 95% confidence intervals) (left) and epidemic curves (right) for (A) amnesia (B) hallucination (C) suicidality. * = p < 0.05.

jcsm.13.2.223b.jpg

jcsm.13.2.223b.jpg
Figure 2

Odds ratios and epidemic curves for zolpidem exposure when compared to all other medications.

(more ...)

The effect of stimulated reporting by intense media coverage can also be observed in the corresponding epidemic curves in Figures 1 and 2. During 2006 to 2009 at the height of media publicity, the percentage of zolpidem reports for all ADRs of interest with the exception of suicidality are significantly greater when compared to the pre- and post-publicity periods. In particular, nearly 50% of all movement-based parasomnia cases reported were associated with zolpidem exposure.

Sensitivity Analyses

Multivariate sensitivity analyses were conducted to confirm our primary analyses. Using a subset of data which comprised of reports where a single drug was identified in each report (n = 286,206), we repeated our analyses to remove any confounding bystander drugs. Where ORs could be calculated, our results indicated a marked risk with RORs > 10 for all ADRs of interest. In fact, the risk of reports for movement-based parasomnias was significantly more pronounced with RORs > 100. The second sensitivity analyses involved comparing zolpidem to the benzodiazepine drug-class (n = 172,184) and results from this investigation are displayed in Figures 3 and 4. From the figures, it is evident that zolpidem remains statistically significant with marked RORs for parasomnia, movement-based parasomnia, nonmovement-based parasomnia, amnesia, and hallucination when compared to benzodiazepines. Although benzodiazepines are one of the most popular drug classes used in the treatment of insomnia, they can also be prescribed for a variety of indications including seizures, anxiety, muscle spasms, and alcohol dependence. Furthermore, like zolpidem, the use of benzodiazepines is also linked to the development of the adverse events of interest (parasomnia, movement-based parasomnia, nonmovement-based parasomnia, amnesia, hallucination, and suicidality). As such, the benzodiazepine drug class were the best available comparator in our sensitivity analyses.

Odds ratios and epidemic curves for zolpidem exposure when compared to all benzodiazepines.

Adjusted reporting odds ratios (and 95% confidence intervals) (left) and epidemic curves (right) for (A) parasomnia (B) movement-based parasomnia (C) nonmovement-based parasomnia. * = p < 0.05.

jcsm.13.2.223c.jpg

jcsm.13.2.223c.jpg
Figure 3

Odds ratios and epidemic curves for zolpidem exposure when compared to all benzodiazepines.

(more ...)

Odds ratios and epidemic curves for zolpidem exposure when compared to all benzodiazepines.

Adjusted reporting odds ratios (and 95% confidence intervals) (left) and epidemic curves (right) for (A) amnesia (B) hallucination (C) suicidality. * = p < 0.05.

jcsm.13.2.223d.jpg

jcsm.13.2.223d.jpg
Figure 4

Odds ratios and epidemic curves for zolpidem exposure when compared to all benzodiazepines.

(more ...)

United States-Only Report Analyses

United States-only analyses exhibited a similar signalling pattern with an increased strength of association between zolpidem and the development of these ADRs during the media publicity cluster compared to the primary results displayed in Figures 1 and 2. However, the results in 2005 for parasomnia, nonmovement-based parasomnia, hallucinations, and suicidality were not statistically significant. Furthermore, since reporter country identification only commenced in 2005, there is insufficient information during the pre-publicity period to compare and evaluate the influence of media publicity surrounding zolpidem and the ADRs of interest.

DISCUSSION

To the best of our knowledge, this is the first rigorous analysis of the effect of stimulated reporting on adverse drug reports in the FAERS database. Given the increasing prominence of media, understanding its impact on adverse event reporting is important. In our analyses of the United States FAERS database, zolpidem exhibited an increased association with the development of all ADRs of interest with the exception of nonmovement-based parasomnias prior to the media publicity cluster. However, following extensive negatively charged media coverage and the subsequent FDA boxed warnings and safety alerts, both the number of ADR reports involving zolpidem and the RORs for parasomnias, movement-based parasomnias, nonmovement-based parasomnias, amnesias, and hallucinations were significantly increased.

This stimulated reporting effect demonstrable in our analyses of the FAERS reports appears to be comparable to that observed in our previous investigation of the Australian DAEN system.4 Despite the role of media in stimulating these ADR reports through a “bandwagon effect,” results from this study and our previous Australian study both indicate the existence of elevated signals for the ADRs of interest before the stimulated reporting event. While information on the date of onset of ADRs within the database is incomplete, analyses of the absolute number of reports between zolpidem exposure and the outcomes of interest using onset dates reveal a similar pattern.

Typically, the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM-5) and the third edition of the International Classification of Sleep Disorders (ICSD-3) classifies parasomnias according to the sleep state from which they arise as NREM arousal disorders, REM-related parasomnias, or other parasomnias.2426 However, as ADR reports do not contain electroencephalographic evidence or recordings, the DSM-5 and ICSD-3 classification of parasomnias could not be effectively determined for the ADR reports in the FAERS database. Consequently, in this study, the parasomnia category was differentiated into two distinct categories of movement-based parasomnias and nonmovement-based parasomnias according to the presence or absence of motor activity. By differentiating parasomnias in this way, it is evident that movement-based parasomnias had a marked ROR after extensive sensationalism in the media. Although a strong association between zolpidem exposure and the development of these complex sleep-related behaviors exists, a causal relation cannot be established from spontaneously collected reporting data that can clearly be influenced by the media. Future studies are warranted to establish a causal relation and to investigate possible mechanisms behind the development of these potentially dangerous adverse reactions.

Compared to the Australian TGA results, zolpidem DEA signals for all ADRs of interest in the FAERS database were not as elevated over the ten-year period. This may be attributable to the inclusion of different confounders and covariates, as well as the effect of direct-to-consumer public advertising of medications. In the previous study utilizing spontaneously collected Australian TGA reports, only predominantly psychoactive agents such as benzodiazepines, antidepressants, anticholinergics, antipsychotics, and opiates were included as potential confounders. However, recent studies have revealed a number of associations between bizarre sleep-related behaviors with non-psychoactive agents such as statins,27 varenicline,28 β-blockers,29 and leukotriene antagonists.30 As such, these agents were analyzed and included in the FAERS data analysis as potential drug-exposure covariates. Secondly, in contrast to Australia, the FDA permits direct-to-consumer advertising of pharmaceuticals.31 Whilst the FDA have established guidelines to regulate the advertising of prescription preparations, advertisements themselves are not subject to mandatory review or approval prior to their release.31,32 In a review by Wilkes et al.,31 the authors remarked that direct-to-consumer ads tend to magnify positive benefits of the drug and downplay the negative adverse reactions, which are typically discussed last and buried in the narrative. This is further supported by a number of studies which conclude that adverse effects and contraindications are rarely ever mentioned in the ads and frequently play down the risks and side effects of the medication.33 Although it is uncertain as to how these advertisements impact on the generation of the signals detected, such advertisements have been recognized to provide biased medical information often promoting the beneficial effects of the medication.34 By highlighting the beneficial effects, the negative impression conveyed in media reports may be somewhat alleviated.

Whilst SRS remains the mainstay of drug safety monitoring, there are numerous weaknesses and limitations in ADR data collected by these pharmacovigilance systems including underreporting and variable data quality such as the absence of clinical information, e.g., dose and timing information. Given the key role of dose in the development of adverse effects from hypnotics, the lack of this information in the FAERS database remains a crucial problem. Furthermore, case reports of zolpidem induced amnestic complex sleep-related behaviors as well as hallucinations have been characterized to be dose-dependent,35 although reactions have occurred at the standard therapeutic dose. More recently, the dose of zolpidem was reviewed and subsequently revised by the FDA based upon blood concentration data. In 2013, the FDA released a safety announcement revising labelling requirements and dosing recommendations on all zolpidem preparations to reflect the increased risk of next-morning cognitive impairment and residual daytime effects.36 In particular, the recommended initial dosage for zolpidem in women has been halved following evidence of increased zolpidem blood levels in females due to gender variability in the elimination of the drug. Despite these recommendations, there is inadequate evidence to associate blood concentration of zolpidem to the development of adverse reactions. Moreover, recent studies undertaken have failed to reveal a relation between benzodiazepine blood levels and driving impairment.37 Although difficult to accomplish, there is an inevitable need for evidence-based medicine and regulatory action to determine the effects of dose, blood concentration levels, and the subsequent development of these neuropsychiatric reactions.

Most importantly, the use of alcohol and other recreational or illicit substances are poorly recorded in ADR data. Although alcohol has the ability to interfere with normal intrinsic sleep physiology,38 its role in the development of parasomnias is far from clear.39 While prescribing guidelines recommend against co-administration of alcohol and zolpidem, there are numerous cases where patients have disregarded this advice and continued to use these agents concomitantly. As such, the potential role of alcohol as a confounder cannot be disregarded. There is also a lack of clinical information on concurrent conditions such as depression, restless legs syndrome, epilepsy, and dementia, all of which are associated with the ADRs of interest.

Given the impact of stimulated reporting on the generation of DEA signals, drug regulatory bodies should consider this notoriety bias when making drug related decisions based on adverse event risk. Importantly, historical cases offer valuable lessons. The “trial by media”18 of triazolam, a short-acting benzodiazepine that underwent intensive adverse event reporting following media coverage, and subsequently lead to the banning of the drug in some countries in the early nineties, is one such case. While careful attention to intensified reporting is essential, regulatory decisions must also consider how such reporting diverts use away from the agent of interest to less tested or less safe medicines, often used off-label. In the case of triazolam, bans in certain countries possibly resulted in the use of longer acting benzodiazepines40; some researchers have referred to this as “out of the frying pan into the fire.”41 Following the zolpidem saga, there is a reported trend in off-label use of anti-psychotics such as quetiapine for insomnia.41,42 Thus, careful analyses to identify adverse event signals must be undertaken prior to regulatory decisions and clear dissemination made to both professionals and the public regarding the level of risk in using a medicine that has been the subject of stimulated reporting.

CONCLUSIONS

A significant stimulated reporting phenomenon can be observed following negative media publicity surrounding the adverse effect profile of zolpidem in Australia and the United States. Results from both the previous Australian study4 and this study indicate a marked increase in the absolute number of reports and the strength of association between zolpidem and the development of the ADRs of interest. While spontaneously collected ADR data demonstrate a significant association between zolpidem and the induction of these bizarre sleep-related behaviors, causality related to zolpidem cannot be conclusively concluded. Nevertheless, caution is warranted by physicians and healthcare professionals when prescribing zolpidem to ensure patients are adequately informed about potential neuropsychiatric adverse events.

DISCLOSURE STATEMENT

This was not an industry supported study. This study was performed at the Faculty of Pharmacy, The University of Sydney. Dr. Marshall reports non-financial support from Teva Cephalon outside the submitted work. The authors have indicated no financial conflicts of interest.

ABBREVIATIONS

ASCII

American Standard Code for Information Interchange

ADR

adverse drug reaction

ATC

Anatomical Therapeutic Chemical

CI

confidence intervals

DAEN

Database of Adverse Event Notifications

DEA

drug-event association

DSM

Diagnostic and Statistical Manual of Mental Disorders

FAERS

FDA Adverse Event Reporting System

FDA

United States Food and Drug Administration

ICSD

International Classification of Sleep Disorders

MedDRA

Medical Dictionary for Regulatory Activities

MIMS

Monthly Index of Medical Specialties

OR

odds ratio

PT

preferred term

ROR

reporting odds ratio

SGML

standard generalized markup language

SRS

spontaneous reporting systems

TGA

Australian Therapeutic Goods Administration

WHO

World Health Organization

REFERENCES

1 

Kang DY, Park S, Rhee CW, et al. Zolpidem use and risk of fracture in elderly insomnia patients. J Prev Med Public Health. 2012;45(4):219–226. [PubMed Central][PubMed]

2 

Olson LG. Hypnotic hazards: adverse effects of zolpidem and other z-drugs. Aust Prescr. 2008;31:146–149.

3 

Victorri-Vigneau C, Feuillet F, Wainstein L, et al. Pharmacoepidemiological characterisation of zolpidem and zopiclone usage. Eur J Clin Pharmacol. 2013;69:1965–1972. [PubMed]

4 

Ben-Hamou M, Marshall NS, Grunstein RR, Saini B, Fois RA. Spontaneous adverse event reports associated with zolpidem in Australia 2001-2008. J Sleep Res. 2011;20(4):559–568. [PubMed]

5 

Collins LM. Ambien-amnesia link no surprise to doctors. Desert News Utah Web site. http://www.deseretnews.com/article/635192102/Ambien-amnesia-link-no-surprise-to-doctors.html. Published March 16, 2006. Accessed March 31, 2015.

6 

DeNoon DJ. Ambien Linked to “Sleep Eating.” WebMD Web site. http://www.webmd.com/sleep-disorders/news/20060315/ambien-linked-to-sleep-eating. Published March 15, 2006. Accessed March 27, 2015.

7 

ABC News. American Broadcasting Company News Web site. ‘Sleepdriving’ Wreaking Havoc on Roads. http://abcnews.go.com/GMA/Health/story?id=1704089. Published March 9, 2006. Accessed March 27, 2015.

8 

Marshall NS, Glozier N, Grunstein RR. Thirty-fold spike in adverse event reporting associated with zolpidem use in australia was most likely caused by the media. Sleep. 2009;32(Abstract Suppl):A393.

9 

Robertson J, Walkom EJ, Bevan MD, Newby DA. Medicines and the media: news reports of medicines recommended for government reimbursement in Australia. BMC Public Health. 2013;13:489. [PubMed Central][PubMed]

10 

Faasse K, Gamble G, Cundy T, Petrie KJ. Impact of television coverage on the number and type of symptoms reported during a health scare: a retrospective pre-post observational study. BMJ Open. 2012;2(4):e001607. [PubMed Central][PubMed]

11 

Pariente A, Gregoire F, Fourrier-Reglat A, Haramburu F, Moore N. Impact of safety alerts on measures of disproportionality in spontaneous reporting databases: the notoriety bias. Drug Saf. 2007;30(10):891–898. [PubMed]

12 

Hoffman KB, Dimbil M, Erdman CB, Tatonetti NP, Overstreet BM. The Weber effect and the United States Food and Drug Administration's Adverse Event Reporting System (FAERS): analysis of sixty-two drugs approved from 2006 to 2010. Drug Saf. 2014;37(4):283–294. [PubMed Central][PubMed]

13 

Hartnell NR, Wilson JP. Replication of the Weber effect using postmarketing adverse event reports voluntarily submitted to the United States Food and Drug Administration. Pharmacotherapy. 2004;24(6):743–749. [PubMed]

14 

Edwards IR, Aronson JK. Adverse drug reactions: definitions, diagnosis, and management. Lancet. 2000;356(9237):1255–1259. [PubMed]

15 

Martin RM, Hilton SR, Kerry SM. The impact of the October 1995 “pill scare” on oral contraceptive use in the United Kingdom: analysis of a general practice automated database. Fam Pract. 1997;14(4):279–284. [PubMed]

16 

Godlee F, Smith J, Marcovitch H. Wakefield's article linking MMR vaccine and autism was fraudulent. BMJ. 2011;342:c7452. [PubMed]

17 

Medawar C, Herxheimer A, Bell A, Jofre S. Paroxetine, Panorama and user reporting of ADRs: consumer intelligence matters in clinical practice and post-marketing drug surveillance. Int J Risk Saf Med. 2002;15(4):161–169.

18 

Lasagna L. The Halcion story: trial by media. Lancet. 1980;1(8172):815–816. [PubMed]

19 

Cresswell A. Ledger's death puts the focus on sleeping pill. The Australian Web site. http://www.theaustralian.com.au/news/health-science/ledgers-death-puts-the-focus-on-sleeping-pill/story-e6frg8y6-1111115402576. Published January 26, 2008. Accessed March 27, 2016.

20 

Swartz A. Oscars News: Heath Ledger's Academy Award Will Go to His Daughter Someday. Mic Web site. https://mic.com/articles/136249/oscars-news-heath-ledger-s-academy-award-will-go-to-his-daughter-someday. Published February 25, 2016. Accessed April 14, 2016.

21 

Roy AN, Smith M. Prevalence and cost of insomnia in a state Medicaid feefor-service population based on diagnostic codes and prescription utilization. Sleep Med. 2010;11(5):462–469. [PubMed]

22 

United States Food and Drug Administration Web site. Questions and Answers on FDA's Adverse Event Reporting System (FAERS). http://www.fda.gov/Drugs/GuidanceComplianceRegulatoryInformation/Surveillance/AdverseDrugEffects. Updated May 5, 2016. Accessed December 19, 2016.

23 

Wong CK, Ho SS, Saini B, Hibbs DE, Fois RA. Standardisation of the FAERS database: a systematic approach to manually recoding drug name variants. Pharmacoepidemiol Drug Saf. 2015;24(7):731–737. [PubMed]

24 

Sateia MJ. International classification of sleep disorders-third edition highlights and modifications. Chest. 2014;146(5):1387–1394. [PubMed]

25 

American Academy of Sleep Medicine. International Classification of Sleep Disorders. 3rd ed. Darien, IL: American Academy of Sleep Medicine; 2014.

26 

American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 5th ed. Arlington, VA: American Psychiatric Association; 2013.

27 

Takada M, Fujimoto M, Yamazaki K, Takamoto M, Hosomi K. Association of statin use with sleep disturbances: data mining of a spontaneous reporting database and a prescription database. Drug Saf. 2014;37(6):421–431. [PubMed]

28 

Raidoo BM, Kutscher EC. Visual hallucinations associated with varenicline: a case report. J Med Case Rep. 2009;3:7560. [PubMed Central][PubMed]

29 

Goldner JA. Metoprolol-induced visual hallucinations: a case series. J Med Case Rep. 2012;6:65. [PubMed Central][PubMed]

30 

Alkhuja S, Gazizov N, Alexander ME. Sleeptalking! Sleepwalking! Side effects of montelukast. Case Rep Pulmonol. 2013;2013:813786. [PubMed Central][PubMed]

31 

Wilkes MS, Bell RA, Kravitz RL. Direct-to-consumer prescription drug advertising: trends, impact, and implications. Health Aff. 2000;19(2):110–128.

32 

United States Food and Drug Administration Web site. Prescription Drug Advertising: Questions and Answers. http://www.fda.gov/Drugs/ResourcesForYou/Consumers/PrescriptionDrugAdvertising/ucm076768.htm. Updated June 19, 2015. Accessed December 19, 2016.

33 

Lexchin J, Mintzes B. Direct-to-consumer advertising of prescription drugs: the evidence says no. Journal of Public Policy & Marketing. 2002;21(2):194–201.

34 

Stange KC. Time to ban direct-to-consumer prescription drug marketing. Ann Fam Med. 2007;5(2):101–104. [PubMed Central][PubMed]

35 

Toner LC, Tsambiras BM, Catalano G, Catalano MC, Cooper DS. Central nervous system side effects associated with zolpidem treatment. Clin Neuropharmacol. 2000;23(1):54–58. [PubMed]

36 

United States Food and Drug Administration Web site. FDA Drug Safety Communication: Risk of next-morning impairment after use of insomnia drugs; FDA requires lower recommended doses for certain drugs containing zolpidem (Ambien, Ambien CR, Edluar, and Zolpimist). http://www.fda.gov/Drugs/DrugSafety/ucm334033.htm. Published January 10, 2013. Updated January 16, 2016. Accessed December 19, 2016.

37 

Verster JC, Roth T. Blood drug concentrations of benzodiazepines correlate poorly with actual driving impairment. Sleep Med Rev. 2013;17(2):153–159. [PubMed]

38 

Roehrs T, Roth T. Sleep, sleepiness, sleep disorders and alcohol use and abuse. Sleep Med Rev. 2001;5(4):287–297. [PubMed]

39 

Pressman MR, Mahowald MW, Schenck CH, Bornemann MC. No scientific evidence that alcohol causes sleepwalking. J Sleep Res. 2008;17(4):473–474.

40 

Rosenthal E. Sleeping pill Halcion banned by Britain. The Baltimore Sun Web site. http://articles.baltimoresun.com/1991-10-03/news/1991276026_1_halcion-food-and-drug-drug-administration. Published October 3, 1991. Accessed October 19, 2015.

41 

Huthwaite M, Cleghorn M, MacDonald J. “Out of the frying pan”: the challenges of prescribing for insomnia in psychiatric patients. Australas Psychiatry. 2014;22(3):288–291. [PubMed]

42 

Carton L, Cottencin O, Lapeyre-Mestre M, et al. Off-label prescribing of antipsychotics in adults, children and elderly individuals: a systematic review of recent prescription trends. Curr Pharm Des. 2015;21(23):3280–3297. [PubMed]


Supplemental Material

Supplemental Material

(.pdf | 259 KB)