Drug-event Associations Research Articles

Neuronal toxicity of monoclonal antibodies (mAbs): an analysis of post-marketing reports from FDA Adverse Event Reporting System (FAERS) safety database.

Monoclonal antibodies (mAbs) are pivotal in treating various diseases, including cancers and autoimmune disorders. Despite their therapeutic benefits, mAb therapy has been associated with neurological toxicity. This study aimed to assess the occurrence of neuronal toxicity associated with mAbs, utilizing data from the FDA Adverse Event Reporting System (FAERS) safety database. The study also sought to delineate the medical characteristics of the reported cases. A comprehensive analysis of neurological adverse events reported in the FAERS database was conducted, employing computational methodologies such as proportional relative risk (PRR), information component (IC025), and chi-square (χ2). Individual case safety reports (ICSRs) pertaining to neurological disorders linked to mAbs from the date of first global marketing authorization until June 30, 2023, were meticulously examined. The FAERS safety database contains 79,022 ICSRs linking mAbs to nervous system disorders. Rituximab, bevacizumab, denosumab, nivolumab, and trastuzumab were frequently cited. Reported adverse events include headache, peripheral neuropathy, dizziness, and cerebrovascular accident. Most ICSRs (85.81%) were serious, mainly affecting females (57.04%) with a 14.09% fatality rate. Panitumumab, atezolizumab, bevacizumab, and trastuzumab showed strong drug-event associations. Signal disproportionate reporting (SDR) analysis flagged myasthenia gravis, peripheral neuropathy, and neurotoxicity across multiple mAbs, suggesting potential signals. Interdisciplinary collaboration between oncologists and neurologists is crucial for safe mAb use. Our study enhances understanding of mAb neurological safety. Disproportionality signal analysis provides valuable evidence for risk mitigation.

An innovative method to strengthen evidence for potential drug safety signals using Electronic Health Records

Reports from spontaneous reporting systems (SRS) are hypothesis generating. Additional evidence such as more reports is required to determine whether the generated drug-event associations are in fact safety signals. However, underreporting of adverse drug reactions (ADRs) delays signal detection. Through the use of natural language processing, different sources of real-world data can be used to proactively collect additional evidence for potential safety signals. This study aims to explore the feasibility of using Electronic Health Records (EHRs) to identify additional cases based on initial indications from spontaneous ADR reports, with the goal of strengthening the evidence base for potential safety signals. For two confirmed and two potential signals generated by the SRS of the Netherlands Pharmacovigilance Centre Lareb, targeted searches in the EHR of the Leiden University Medical Centre were performed using a text-mining based tool, CTcue. The search for additional cases was done by constructing and running queries in the structured and free-text fields of the EHRs. We identified at least five additional cases for the confirmed signals and one additional case for each potential safety signal. The majority of the identified cases for the confirmed signals were documented in the EHRs before signal detection by the Dutch Medicines Evaluation Board. The identified cases for the potential signals were reported to Lareb as further evidence for signal detection. Our findings highlight the feasibility of performing targeted searches in the EHR based on an underlying hypothesis to provide further evidence for signal generation.

A reference set of clinically relevant adverse drug-drug interactions

The accurate and timely detection of adverse drug-drug interactions (DDIs) during the postmarketing phase is an important yet complex task with potentially major clinical implications. The development of data mining methodologies that scan healthcare databases for drug safety signals requires appropriate reference sets for performance evaluation. Methodologies for establishing DDI reference sets are limited in the literature, while there is no publicly available resource simultaneously focusing on clinical relevance of DDIs and individual behaviour of interacting drugs. By automatically extracting and aggregating information from multiple clinical resources, we provide a scalable approach for generating a reference set for DDIs that could support research in postmarketing safety surveillance. CRESCENDDI contains 10,286 positive and 4,544 negative controls, covering 454 drugs and 179 adverse events mapped to RxNorm and MedDRA concepts, respectively. It also includes single drug information for the included drugs (i.e., adverse drug reactions, indications, and negative drug-event associations). We demonstrate usability of the resource by scanning a spontaneous reporting system database for signals of DDIs using traditional signal detection algorithms.

Impact of literature reports on drug safety signals.

SummaryBackgroundSignal management is considered an important activity in pharmacovigilance and should be performed using any available source of data, including scientific literature. The main aim of this study was to assess the role of scientific literature in both indexed and unindexed journals and compare the relevance of both in the signal management process.MethodsThe study was a retrospective analysis of safety data. For the purposes of the study, drugs for which safety signals were evaluated by European Medicine Agency (EMA) were chosen. A match analysis of data collected in the EudraVigilance (EV) database with data from bibliographic databases such as MEDLINE, Embase or EBSCO (International Pharmaceutical s, IPA and the Allied and the Complementary Medicine Database, AMED) was performed.ResultsA total of 73 drug event associations (DEA) and 4160 individual case safety reports (ICSRs) were analyzed. About 33% of ICSRs were created based on scientific literature. A total of 1196 ICSRs were submitted to the EV database based on journals indexed in global bibliographic databases Embase (86.00%) or MEDLINE (81.96%) or EBSCO (IPA or AMED, 0.66%).ConclusionThis study underlines the importance of scientific literature for the signal management process in addition to other data sources. Most literature ICSRs from this analysis were created based on scientific journals indexed in bibliographic databases; therefore, it can be concluded that a systematic review of bibliographic databases, such as Embase or MEDLINE is highly relevant for the signal management process.

Bayesian model selection in logistic regression for the detection of adverse drug reactions.

Spontaneous adverse event reports have a high potential for detecting adverse drug reactions. However, due to their dimension, the analysis of such databases requires statistical methods. In this context, disproportionality measures can be used. Their main idea is to project the data onto contingency tables in order to measure the strength of associations between drugs and adverse events. However, due to the data projection, these methods are sensitive to the problem of coprescriptions and masking effects. Recently, logistic regressions have been used with a Lasso type penalty to perform the detection of associations between drugs and adverse events. On different examples, this approach limits the drawbacks of the disproportionality methods, but the choice of the penalty value is open to criticism while it strongly influences the results. In this paper, we propose to use a logistic regression whose sparsity is viewed as a model selection challenge. Since the model space is huge, a Metropolis-Hastings algorithm carries out the model selection by maximizing the BIC criterion. Thus, we avoid the calibration of penalty or threshold. During our application on the French pharmacovigilance database, the proposed method is compared to well-established approaches on a reference dataset, and obtains better rates of positive and negative controls. However, many signals (i.e., specific drug-event associations) are not detected by the proposed method. So, we conclude that this method should be used in parallel to existing measures in pharmacovigilance. Code implementing the proposed method is available at the following url: https://github.com/masedki/MHTrajectoryR.

Evidence of Misclassification of Drug-Event Associations Classified as Gold Standard 'Negative Controls' by the Observational Medical Outcomes Partnership (OMOP).

Pharmacovigilance includes analysis of large databases of information on drugs and events using algorithms that detect disproportional frequencies of associations. In order to test such algorithms, attempts have been made to provide canonical reference lists of so-called 'positive controls' and 'negative controls'. Reference sets with even modest levels of misclassification may result in under- or overstatement of the performance of algorithms. We sought to determine the extent to which 'negative control' drug-event pairs in the Observational Medical Outcomes Partnership (OMOP) database are misclassified We searched the medical literature for evidence of associations between drugs and events listed by OMOP as negative controls. The criteria used in OMOP to classify positive and negative controls are asymmetric; drug-event associations published only as case series or case reports are classified as positive controls if they are cited in Drug-Induced Diseases by Tisdale and Miller, but as negative controls if case series or case reports exist but are not cited in Tisdale and Miller. Of 233 drug-event pairs classified in the 2013 version of OMOP as negative controls, 21 failed to meet pre-specified OMOP adjudication criteria; in another 19 cases we found case reports, case series, or observational evidence that the drug and event are associated. Overall, OMOP misclassified, or may have misclassified, 40 (17%) of all 'negative controls.' Results from studies of the performance of signal-detection algorithms based on the OMOP gold standard should be viewed with circumspection, because imperfect gold standards may lead to under/overstatement of absolute and relative signal detection algorithm performance. Improvements to OMOP would include omitting misclassified drug-event pairs, assigning more specific event labels, and using more extensive sources of information.

The effect of exposure misclassification in spontaneous ADR reports on the time to detection of product-specific risks for biologicals: a simulation study.

The availability of accurate product-specific exposure information is essential in the pharmacovigilance of biologicals, because differences in the safety profile may emerge between products containing the same active substance. In spontaneous adverse drug reaction (ADR) reports, drug exposure may, however, be misclassified, that is, attributed to the incorrect product. The aim of this study was to explore the effect of exposure misclassification on the time to detection of product-specific risks in spontaneous reporting systems. We used data simulations to explore the effect of exposure misclassification. We simulated an active substance-specific subset of a spontaneous reporting system and used the proportional reporting ratio for signal detection. The effect of exposure misclassification was evaluated in three test cases representing product-specific ADRs that may occur for biologicals and studied in relative terms by varying the model parameters (market share and relative risk). We found that exposure misclassification results in the largest delay in identification of risks that have a weak association (relative risk < 2 or 3) with the product of interest and in situations where the product associated with the unique risk has a large (>50%) market share. The absolute public health impact of exposure misclassification, in terms of cases/time to detection, varied considerably across the test cases. Exposure misclassification in ADR reports may result in a delayed detection of product-specific risks, particularly in the detection of weak drug-event associations. Our findings can help inform the future implementation and refinement of product-specific and batch-specific signal detection procedures.

Feasibility of Prioritizing Drug-Drug-Event Associations Found in Electronic Health Records.

Background and ObjectiveSeveral studies have demonstrated the ability to detect adverse events potentially related to multiple drug exposure via data mining. However, the number of putative associations produced by such computational approaches is typically large, making experimental validation difficult. We theorized that those potential associations for which there is evidence from multiple complementary sources are more likely to be true, and explored this idea using a published database of drug–drug-adverse event associations derived from electronic health records (EHRs).MethodsWe prioritized drug–drug-event associations derived from EHRs using four sources of information: (1) public databases, (2) sources of spontaneous reports, (3) literature, and (4) non-EHR drug–drug interaction (DDI) prediction methods. After pre-filtering the associations by removing those found in public databases, we devised a ranking for associations based on the support from the remaining sources, and evaluated the results of this rank-based prioritization.ResultsWe collected information for 5983 putative EHR-derived drug–drug-event associations involving 345 drugs and ten adverse events from four data sources and four prediction methods. Only seven drug–drug-event associations (<0.5 %) had support from the majority of evidence sources, and about one third (1777) had support from at least one of the evidence sources.ConclusionsOur proof-of-concept method for scoring putative drug–drug-event associations from EHRs offers a systematic and reproducible way of prioritizing associations for further study. Our findings also quantify the agreement (or lack thereof) among complementary sources of evidence for drug–drug-event associations and highlight the challenges of developing a robust approach for prioritizing signals of these associations.Electronic supplementary materialThe online version of this article (doi:10.1007/s40264-015-0352-2) contains supplementary material, which is available to authorized users.

The Contribution of National Spontaneous Reporting Systems to Detect Signals of Torsadogenicity: Issues Emerging from the ARITMO Project.

IntroductionSpontaneous reporting systems (SRSs) are pivotal for signal detection, especially for rare events with a high drug-attributable component, such as torsade de pointes (TdP). Use of different national SRSs is rarely attempted because of inherent difficulties, but should be considered on the assumption that rare events are diluted in international databases.ObjectiveThe aim was to describe TdP-related events associated with antipsychotics, H1-antihistamines and anti-infectives in three national SRSs (in Italy, Germany and France) and highlight potential signals of torsadogenicity through a combined literature evaluation.MethodsA common search strategy was applied to extract TdP-related events: (1) TdP, (2) QT interval abnormalities, (3) ventricular fibrillation/tachycardia, and (4) sudden cardiac death. Signals of disproportionate reporting (SDRs) were calculated for TdP + QT interval abnormalities and defined by a lower limit of the 95 % confidence interval of the reporting odds ratio (ROR) >1. Among SDRs with at least three cases without concomitant pro-arrhythmic drugs, we defined potential new signal of torsadogenicity as drugs with no published evidence from (a) the crediblemeds® website (http://www.crediblemeds.com, as of November 1st, 2014); (b) studies on the FDA Adverse Event Reporting System (FAERS); and (c) safety trials or pharmaco-epidemiological studies (as of December 16th, 2014).ResultsOverall, 3505 cases were retrieved (1372, 1468, and 801 for France, Germany and Italy, respectively). Antipsychotics were mainly recorded in Germany (792 cases), whereas antibiotics peaked at 515 and 491 (France and Italy, respectively). Forty-one drugs met criteria for SDRs in at least one single source, of which 31 were detected only from one single SRS: 18, ten and three (French, German and Italian SRS, respectively). By contrast, only five SDRs were detected in all national data sources (amisulpride, aripiprazole, haloperidol, olanzapine, risperidone). Overall, five potential new signals of torsadogenicity were identified: flupentixol, ganciclovir, levocetirizine, oxatomide and tiapride.ConclusionsWe found differences across and within national SRSs in the reporting of drug-induced TdP, which finally resulted in five potential new signals of torsadogenicity. These findings warrant targeted pharmacovigilance studies to formally assess the existence of actual drug–event associations.Electronic supplementary materialThe online version of this article (doi:10.1007/s40264-015-0353-1) contains supplementary material, which is available to authorized users.

Detecting Potential Safety Issues in Large Clinical or Observational Trials by Bayesian Screening When Event Counts Arise from Poisson Distributions

Patients in large clinical trials and in studies employing large observational databases report many different adverse events, most of which will not have been anticipated at the outset. Conventional hypothesis testing of between group differences for each adverse event can be problematic: Lack of significance does not mean lack of risk, the tests usually are not adjusted for multiplicity, and the data determine which hypotheses are tested. This article describes a Bayesian screening approach that does not test hypotheses, is self-adjusting for multiplicity, provides a direct assessment of the likelihood of no material drug–event association, and quantifies the strength of the observed association. The criteria for assessing drug-event associations can be determined by clinical or regulatory considerations. In contrast to conventional approaches, the diagnostic properties of this new approach can be evaluated analytically. Application of the method to findings from a vaccine trial yields results similar to those found by methods using a false discovery rate argument or a hierarchical Bayes approach. [Supplemental materials are available for this article. Go to the publisher's online edition of Journal of Biopharmaceutical Statistics for the following free supplemental resource: Appendix R: Code for calculations.]

A Potential Event-Competition Bias in Safety Signal Detection: Results from a Spontaneous Reporting Research Database in France

In spontaneous reporting databases, reports of well-established drug-event associations may mask alerts that arise from other drugs (drug competition bias). However, a symmetrical event-competition bias has not yet been explored whereby known events may mask an association with new events for a given drug or drug class. The objective of this study was to explore the effects of event-competition bias on safety signals generated from spontaneous reporting databases. The drug classes tested included statins, oral anticoagulants, antipsychotics and HIV antiretrovirals. For each, a type A reaction was selected, and its potential competitive effect on the generation of other safety signals for the drug was explored. These were rhabdomyolysis/myopathy for statins, haemorrhage for oral anticoagulants, extrapyramidal syndrome for antipsychotics and lipodystrophy for HIV antiretrovirals. Signals of disproportionate reporting (SDRs) were detected using the case/non-case approach in the French research spontaneous reporting database (which contains reports from 1 January 1986 to 31 December 2001), before and after removing all reports concerning these competitor events. SDRs were considered as potential signals if not reported in the literature before 1 January 2002 but confirmed since. The whole database included 207,236 reports, 4,355 of which included statins as one of the suspected drugs. The removal of reports of rhabdomyolysis/myopathy concerned 8,425 reports among which 867 involved statins. After this removal, 11 new SDRs appeared for statins that had not been detected initially. Similarly, 15 SDRs were unmasked for oral anticoagulants, six for antipsychotics and nine for HIV antiretrovirals. After literature-based assessment, five of the 41 unmasked SDRs appeared related to potential safety signals confirmed after 2002. This study demonstrated that a masking phenomenon resulting from an event-competition effect could occur when performing signal detection using disproportionality analyses of spontaneous reporting databases. This should be taken into account when routine signal detection is performed.

A Reference Standard for Evaluation of Methods for Drug Safety Signal Detection Using Electronic Healthcare Record Databases

The growing interest in using electronic healthcare record (EHR) databases for drug safety surveillance has spurred development of new methodologies for signal detection. Although several drugs have been withdrawn postmarketing by regulatory authorities after scientific evaluation of harms and benefits, there is no definitive list of confirmed signals (i.e. list of all known adverse reactions and which drugs can cause them). As there is no true gold standard, prospective evaluation of signal detection methods remains a challenge. Within the context of methods development and evaluation in the EU-ADR Project (Exploring and Understanding Adverse Drug Reactions by integrative mining of clinical records and biomedical knowledge), we propose a surrogate reference standard of drug-adverse event associations based on existing scientific literature and expert opinion. The reference standard was constructed for ten top-ranked events judged as important in pharmacovigilance. A stepwise approach was employed to identify which, among a list of drug-event associations, are well recognized (known positive associations) or highly unlikely ('negative controls') based on MEDLINE-indexed publications, drug product labels, spontaneous reports made to the WHO's pharmacovigilance database, and expert opinion. Only drugs with adequate exposure in the EU-ADR database network (comprising ≈60 million person-years of healthcare data) to allow detection of an association were considered. Manual verification of positive associations and negative controls was independently performed by two experts proficient in clinical medicine, pharmacoepidemiology and pharmacovigilance. A third expert adjudicated equivocal cases and arbitrated any disagreement between evaluators. Overall, 94 drug-event associations comprised the reference standard, which included 44 positive associations and 50 negative controls for the ten events of interest: bullous eruptions; acute renal failure; anaphylactic shock; acute myocardial infarction; rhabdomyolysis; aplastic anaemia/pancytopenia; neutropenia/agranulocytosis; cardiac valve fibrosis; acute liver injury; and upper gastrointestinal bleeding. For cardiac valve fibrosis, there was no drug with adequate exposure in the database network that satisfied the criteria for a positive association. A strategy for the construction of a reference standard to evaluate signal detection methods that use EHR has been proposed. The resulting reference standard is by no means definitive, however, and should be seen as dynamic. As knowledge on drug safety evolves over time and new issues in drug safety arise, this reference standard can be re-evaluated.

Effect of Competition Bias in Safety Signal Generation

Background: Automated disproportionality analysis of spontaneous reporting is increasingly used routinely. It can theoretically be influenced by a competition bias for signal detection owing to the presence of reports related to well-established drug-event associations. Objective: The aim of the study was to explore the effects of competition bias on safety signals generated from a large spontaneous reporting research database. Methods: Using the case/non-case approach in the French spontaneous reporting research database, which includes data of reporting in France from January 1986 to December 2001, the effects of the competition bias were explored by generating safety signals associated with six events of interest (gastric and oesophageal haemorrhages, central nervous system haemorrhage and cerebrovascular accidents, ischaemic coronary disorders, migraine headaches, muscle pains, and hepatic enzymes and function abnormalities) before and after removing from the database reports relating to drugs known to be strongly associated with these events, whether they constituted cases or non-cases. As this study was performed on a closed database (last data entered 31 December 2001), potential signals unmasked by removal were considered as real signals if no or only incomplete knowledge about the association was available from the literature before 1 January 2002. Results: For gastric and oesophageal haemorrhages, after removing reports involving antithrombotic agents or NSAIDs, three potential signals were unmasked (prednisone, rivastigmine and isotretinoin). For central nervous system haemorrhage and cerebrovascular accidents, after removing reports involving antithrombotic agents, three potential signals were unmasked (ethinylestradiol, interferon-α-2B and methylprednisolone). For ischaemic coronary disorders, after removing reports involving anthracyclines, bleomycine, anti-HIV drugs or triptans, one potential signal was unmasked (ondansetron). For migraine headaches, after removing reports involving nitrates, calcium channel blockers, opioid analgesics or intravenous immunoglobulins, six potential signals were unmasked (ammonium chloride, leflunomide, milnacipran, montelukast, proguanil and pyridostigmine). For muscle pains, after removing reports involving statins or fibrates, seven potential signals were unmasked (hydroxychloroquine, lactulose, levodopa in combination with dopadecarboxylase inhibitor, nevirapine, nomegestrol, ritonavir and stavudine). Finally, for hepatic enzymes and function abnormalities, after removing reports involving NSAIDs, anilides, antituberculosis drugs, antiepileptics, ketoconazole, tacrine, or amineptine, two potential signals were unmasked (caffeine, metformin). Of all these unmasked potential signals, ten appeared non/incompletely documented as at 1 January 2002 and were considered as real signals, with three of these later being confirmed by the literature and finally considered as true positives (isotretinoin, methylprednisolone and milnacipran). Conclusion: This study confirms that a competition bias can occur when performing safety signal generation in spontaneous reporting databases. The minimization of this bias could lead to previously masked signals being revealed.

Methods for drug safety signal detection in longitudinal observational databases: LGPS and LEOPARD

There is a growing interest in using longitudinal observational databases for drug safety signal detection, but most of the existing statistical methods are tailored towards spontaneous reporting. Here a sequential set of methods for detecting and filtering drug safety signals in longitudinal databases is presented. Longitudinal GPS (LGPS) is a modification of the Gamma Poisson Shrinker (GPS) that uses person time rather than case counts for the estimation of the expected number of events. Longitudinal Evaluation of Observational Profiles of Adverse events Related to Drugs (LEOPARD) is a method that can be used to automatically discard false drug-event associations caused by protopathic bias or misclassification of the dates of the adverse events by comparing prior event prescription rates to post event prescription rates. LEOPARD can generate a single test statistic, or a visualization that can be used for more qualitative information on the relationship between drug and event. Both methods were evaluated using data simulated using the Observational medical dataset SIMulator (OSIM), including the dataset used in the Observational Medical Outcomes Partnership (OMOP) cup, a recent public competition for signal detection methods. The Mean Average Precision (MAP) was used for performance measurement. On the OMOP cup data, LGPS achieved a MAP of 0.245, and the combination of LGPS and LEOPARD achieved a MAP of 0.260, the highest score in the competition. The sequential use of LGPS and LEOPARD have proven to be a useful novel set of methods for drug safety signal detection on longitudinal health records.

A potential competition bias in the detection of safety signals from spontaneous reporting databases

To study whether reports related to known drug-event associations could hinder the detection of new signals by increasing the detection thresholds when using disporportionality analyses in spontaneous reporting (SR) databases. The French SR database (2005-2006 data) was used to test this hypothesis for the following events: bleeding, headache, hepatitis, myalgia, myocardial infarction, stroke, and toxic epidermal necrolysis (TEN). For each of these, using the Proportional Reporting Ratio (PRR) and the Reporting Odds Ratio (ROR), the number of cases needed to trigger a signal out of 50, 100, and 200 reports for a hypothetical newly introduced drug were computed before and after removing from the database reports involving drugs known to be associated with the event. For bleeding and stroke, removing potentially competitive data resulted in a decrease of the number of cases needed to trigger a signal for a newly introduced drug for both PRR and ROR (e.g., from 9 to 4, and 5 to 3 cases out of 50 reports for bleeding and stroke, respectively using the PRR). They were not or only slightly modified for the other studied events. Removing reports related to known drug-event associations could increase the sensitivity of signal detection in SR databases. This should be considered when using SR databases for signal detection as it could result in earlier identification of new drug-event associations.

The EU-ADR project: preliminary results and perspective.

The EU-ADR project aims to exploit different European electronic healthcare records (EHR) databases for drug safety signal detection. In this paper we describe the project framework and the preliminary results. As first step we created a ranked list of the events that are deemed to be important in pharmacovigilance as mining on all possible events was considered to unduly increase the number of spurious signals. All the drugs that are potentially associated to these events will be detected via data mining techniques. Data sources are eight 8 databases in four countries (Denmark, Italy, the Netherlands, and the United Kingdom) that are virtually linked through harmonisation of input data followed by local elaboration of input data through custom-built software (Jerboa). All the identified drug-event associations (signals) will be thereafter biologically substantiated and epidemiologically validated. To date, only Upper gastrointestinal bleeding (UGIB) event has been used to test the ability of the system in signal detection. An initial ranked list comprising 23 adverse events was identified. The top-ranking events were: cutaneous bullous eruptions, acute renal failure, acute myocardial infarction, anaphylactic shock, and rhabdomyolysis. Regarding the UGIB test, a total of 48,016 first-ever episodes were identified. The age-standardized incidence rates of UGIB varied between 40-100/100,000 person-years depending on country and type of healthcare database. A statistically significant association between use of NSAIDs and UGIB was detected in all of the databases. a dynamic ranked list of 23 adverse drug events judged as important in pharmacovigilance was created to permit focused data mining. Preliminary results on the UGIB event detection demonstrate the feasibility of harmonizing various health care databases in different European countries through a distributed network approach.

Antipsychotics, Glycemic Disorders, and Life-Threatening Diabetic Events: A Bayesian Data-Mining Analysis of the FDA Adverse Event Reporting System (1968-2004)

This analysis compared diabetes-related adverse events associated with use of different antipsychotic agents. A disproportionality analysis of the US Food and Drug Administration (FDA) Adverse Event Reporting System (AERS) was performed. Data from the FDA postmarketing AERS database (1968 through first quarter 2004) were evaluated. Drugs studied included aripiprazole, clozapine, haloperidol, olanzapine, quetiapine, risperidone, and ziprasidone. Fourteen Medical Dictionary for Regulatory Activities (MedDRA) Primary Terms (MPTs) were chosen to identify diabetes-related adverse events; 3 groupings into higher-level descriptive categories were also studied. Three methods of measuring drug-event associations were used: proportional reporting ratio, the empirical Bayes data-mining algorithm known as the Multi-Item Gamma Poisson Shrinker, and logistic regression (LR) analysis. Quantitative measures of association strength, with corresponding confidence intervals, between drugs and specified adverse events were computed and graphed. Some of the LR analyses were repeated separately for reports from patients under and over 45 years of age. Differences in association strength were declared statistically significant if the corresponding 90% confidence intervals did not overlap. Association with various glycemic events differed for different drugs. On average, the rankings of association strength agreed with the following ordering: low association, ziprasidone, aripiprazole, haloperidol, and risperidone; medium association, quetiapine; and strong association, clozapine and olanzapine. The median rank correlation between the above ordering and the 17 sets of LR coefficients (1 set for each glycemic event) was 93%. Many of the disproportionality measures were significantly different across drugs, and ratios of disproportionality factors of 5 or more were frequently observed. There are consistent and substantial differences between atypical antipsychotic drugs in the disproportionality reporting ratios relating to glycemic effects, especially life-threatening events, in the AERS database. The relative associational rankings of drugs are similar in reports from younger and older patients. These results agree with several other reports in the literature, do not support a "class effect" hypothesis, and provide a strong rationale for further studies to clarify the issue.

Accounting for Multiplicity in the Evaluation of “Signals” Obtained by Data Mining from Spontaneous Report Adverse Event Databases

Surveillance of drug products in the marketplace continues after approval, to identify rare potential toxicities that are unlikely to have been observed in the clinical trials carried out before approval. This surveillance accumulates large numbers of spontaneous reports of adverse events along with other information in spontaneous report databases. Recently developed empirical Bayes and Bayes methods provide a way to summarize the data in these databases, including a quantitative measure of the strength of the reporting association between the drugs and the events. Determining which of the particular drug-event associations, of which there may be many tens of thousands, are real reporting associations and which random noise presents a substantial problem of multiplicity because the resources available for medical and epidemiologic followup are limited. The issues are similar to those encountered with the evaluation of microarrays, but there are important differences. This report compares the application of a standard empirical Bayes approach with micorarray-inspired methods for controlling the False Discovery Rate, and a new Bayesian method for the resolution of the multiplicity problem to a relatively small database containing about 48,000 reports. The Bayesian approach appears to have attractive diagnostic properties in addition to being easy to interpret and implement computationally.

Effect of Well Established Drug-Event Associations on the Generation of New Signals in Spontaneous Reporting Databases

Effect of Well Established Drug-Event Associations on the Generation of New Signals in Spontaneous Reporting Databases

Data mining in pharmacovigilance: lessons from phantom ships

Pharmacovigilance experts devote considerable effort to post-marketing surveillance of adverse drug reactions (ADRs). Although the prepared mind of the pharmacovigilance expert remains the cornerstone of this process [1], statistical algorithms, also known as data mining algorithms (DMAs), are being promoted as supplementary tools for safety reviewers. Opinions vary on their utility and optimum deployment mainly because their use has not been completely validated for various reasons, including a lack of consensus on gold standards for causality. True positive associations may be inherently more interesting, but constructing reference sets for validation also require identification of “true negatives” for measuring performance of DMAs. Occasionally, drug-event associations (DEAs), originally considered credible based on traditional pharmacovigilance monitoring, are discounted with various levels of certitude after further investigation. We refer to these DEAs “phantom ships” [2]. Phantom associations may be discounted through epidemiological evidence, careful clinical analysis of the individual cases, and/or based on fundamental clinical pharmacological principles [3–9].