Search

Introduction: Hepatitis B (HB) vaccination programs were set up worldwide in the early 1990s. Despite their major focus on reducing the burden of HB infection, they have seldom achieved the targeted population coverage in most countries, including the USA, with around 24.5% of adults being vaccinated against HB. Among proposed reasons for this is the persisting doubt about a possible link between HB vaccination and the occurrence of cases of multiple sclerosis (MS)., Objective: Our objective was to evaluate a potential safety signal between MS and HB vaccination. We conducted a disproportionality analysis (DPA) using the cases reported to the Vaccine Adverse Event Reporting System (VAERS)., Methods: We calculated the proportional reporting rate (PRR) and reporting odds ratio (ROR) of MS having occurred within the 120 days following HB immunization in adults aged 19-49 years when compared with other vaccines using the reports recorded in the VAERS database. Both ratios were estimated globally and then according to the origin of reports (USA vs. non-USA). We then performed a sensitivity analysis using a broader category of demyelinating events., Findings: MS cases following HB vaccination were more likely to originate from outside the USA and to be reported before 2000 than those associated with other immunizations. All computed ratios were found to be statistically significant, with PRRs ranging from 3.48 to 5.56 and RORs ranging from 3.48 to 5.62. When considering the geographical origin, similar RORs were obtained for both US and non-US cases., Conclusion: In VAERS, MS cases were up to five times more likely to be reported after an HB vaccination than after any other vaccination. Since DPA is mainly suited for hypothesis generation, further studies evaluating the nature of the link between MS and HB vaccination would be of considerable importance.

Introduction: Signal detection from healthcare databases is possible, but is not yet used for routine surveillance of drug safety. One challenge is to develop methods for selecting signals that should be assessed with priority., Aim: The aim of this study was to develop an automated system combining safety signal detection and prioritization from healthcare databases and applicable to drugs used in chronic diseases., Methods: Patients present in the French EGB healthcare database for at least 1 year between 2005 and 2015 were considered. Noninsulin glucose-lowering drugs (NIGLDs) were selected as a case study, and hospitalization data were used to select important medical events (IME). Signal detection was performed quarterly from 2008 to 2015 using sequence symmetry analysis. NIGLD/IME associations were screened if one or more exposed case was identified in the quarter, and three or more exposed cases were identified in the population at the date of screening. Detected signals were prioritized using the Longitudinal-SNIP (L-SNIP) algorithm based on strength (S), novelty (N), and potential impact of signal (I), and pattern of drug use (P). Signals scored in the top 10% were identified as of high priority. A reference set was built based on NIGLD summaries of product characteristics (SPCs) to compute the performance of the developed system., Results: A total of 815 associations were screened and 241 (29.6%) were detected as signals; among these, 58 (24.1%) were prioritized. The performance for signal detection was sensitivity = 47%; specificity = 80%; positive predictive value (PPV) 33%; negative predictive value = 82%. The use of the L-SNIP algorithm increased the early identification of positive controls, restricted to those mentioned in the SPCs after 2008: PPV = 100% versus PPV = 14% with its non-use. The system revealed a strong new signal with dipeptidylpeptidase-4 inhibitors and venous thromboembolism., Conclusion: The developed system seems promising for the routine use of healthcare data for safety surveillance of drugs used in chronic diseases.

Introduction: The consequences of the withdrawal of marketing authorisation of drugs have mostly been studied considering drug prescription patterns for the therapeutic alternatives of the withdrawn drugs. The potential concomitant changes in the reporting of adverse reactions concerning these alternatives have been studied less often., Objective: The objective of this study was to analyse the changes in the reporting of adverse events (AEs) for therapeutic alternatives after the withdrawal of three medicines (dextropropoxyphene, pioglitazone and tetrazepam) from the market for safety reasons., Methods: This study was performed using both the French pharmacovigilance database and the Echantillon Généraliste des Bénéficiaires (a random sample of French health insurance affiliates). For dextropropoxyphene, pioglitazone and tetrazepam alternatives, the number and types of case reports were studied for both the year preceding the first official safety warning and the year following the withdrawal. Reporting rates expressed per 10,000 reimbursements (RR Reimb ) and per 10,000 treated patients (RR Pat ) were also compared for the two periods., Results: After dextropropoxyphene withdrawal, case reports and reimbursements increased for tramadol (case reports: +23%, reimbursements: +13%) and codeine (case reports: +74%, reimbursements: +47%), RR Pat being significantly increased for tramadol (0.92 vs. 1.06, p = 0.02). After pioglitazone withdrawal, case reports increased for dipeptidyl peptidase-4 (DPP-4) inhibitors, glinides, and glucagon-like peptide 1 (GLP-1) analogues (+84%, +22% and +5%, respectively) and reimbursements (+55, +11 and +50%, respectively); both decreased for sulfonylureas (case reports: -6%, reimbursements: -2%). RR Pat increased for DPP-4 inhibitors (1.63 vs. 2.26, p = 0.008). After tetrazepam withdrawal, case reports increased for diazepam, methocarbamol and thiocolchicoside (+110, +86 and +157%, respectively), as lesser did reimbursements. RR Pat increased for diazepam (1.78 vs. 2.41, p = 0.054) and thiocolchicoside (0.14 vs. 0.24, p = 0.013)., Conclusion: For the three drug withdrawals investigated, the number of case reports involving alternatives increased to a larger extent than the numbers of prescriptions. This could relate to a higher occurrence of AEs in new users of alternatives who switched from the withdrawn medicines or to an increased awareness of possible AEs.

Introduction: With increasing availability, the use of healthcare databases as complementary data source for drug safety signal detection has been explored to circumvent the limitations inherent in spontaneous reporting. Areas covered: To review the methods proposed for safety signal detection in healthcare databases and their performance. Expert opinion: Fifteen different data mining methods were identified. They are based on disproportionality analysis, traditional pharmacoepidemiological designs (e.g. self-controlled designs), sequence symmetry analysis (SSA), sequential statistical testing, temporal association rules, supervised machine learning (SML), and the tree-based scan statistic. When considering the performance of these methods, the self-controlled designs, the SSA, and the SML seemed the most interesting approaches. In the perspective of routine signal detection from healthcare databases, pragmatic aspects such as the need for stakeholders to understand the method in order to be confident in the results must be considered. From this point of view, the SSA could appear as the most suitable method for signal detection in healthcare databases owing to its simple principle and its ability to provide a risk estimate. However, further developments, such as automated prioritization, are needed to help stakeholders handle the multiplicity of signals.

Introduction: The case-population approach compares exposure among cases to that of their source population. By using aggregated data to estimate the denominator, this approach can provide a real-time estimate of an association that could be particularly valuable to explore urgent vaccine safety concerns and to generate signals during a vaccine campaign., Objective: Our objective was to present the vaccine case-population method, a method derived from the case-population approach and adapted for vaccine safety surveillance, and to test it using several published examples., Methods: For the vaccine case-population method, exposure in the population is estimated from the sum of at-risk periods using the number of vaccinated individuals, or data of vaccine sales, and the at-risk period considered for the vaccine-event pair. The vaccine case-population method was applied to data from published case-control studies retrieved from the MEDLINE database and having quantified risks associated with vaccines. Odds ratios derived from the vaccine case-population method were compared with those from published case-control studies., Results: A total of 20 vaccine-event pairs were retrieved in which the vaccine case-population method could be applied. For all identified vaccine-event pairs, when a significant association was found using the vaccine case-population method, a significant association was also found in the corresponding case-control study. Conversely, when no association was found by the vaccine case-population method, no association was found in the corresponding case-control study., Conclusion: These results suggest that the vaccine case-population method can produce coherent conclusions and may be used in the future for prospective investigation of urgent vaccine safety concerns or for the prospective generation of vaccine safety signals. This method could also be used to identify selection bias from cases excluded from the case-control study.

The methods for causality assessment of adverse drug reactions were developed in the 1970s and 1980s, alongside the development of pharmacovigilance. The French method is one of the earlier of these, following on from the pioneering works by Irey and Karch and Lasagna. Initially published in 1978, it was updated in 1985, and again in 2011. The main alterations to the original method are presented in tables annexed to this paper. The successive versions improved the presentation, provided more formalised definitions of the criteria for assessing causality, while at the same time ensuring the method remained easy to use. Causality assessment enables the causal link between a drug and the occurrence of an adverse reaction to be formalised and explained. It contributes to diagnosis, and to determining the action to be taken in case of an adverse drug reaction. It can contribute to the quality and the relevance of the data stored in pharmacovigilance databases., (Copyright © 2016 Société française de pharmacologie et de thérapeutique. Published by Elsevier Masson SAS. All rights reserved.)

Introduction: The two methods for minimizing competition bias in signal of disproportionate reporting (SDR) detection--masking factor (MF) and masking ratio (MR)--have focused on the strength of disproportionality for identifying competitors and have been tested using competitors at the drug level., Objectives: The aim of this study was to develop a method that relies on identifying competitors by considering the proportion of reports of adverse events (AEs) that mention the drug class at an adequate level of drug grouping to increase sensitivity (Se) for SDR unmasking, and its comparison with MF and MR., Methods: Reports in the French spontaneous reporting database between 2000 and 2005 were selected. Five AEs were considered: myocardial infarction, pancreatitis, aplastic anemia, convulsions, and gastrointestinal bleeding; related reports were retrieved using standardized Medical Dictionary for Regulatory Activities (MedDRA(®)) queries. Potential competitors of AEs were identified using the developed method, i.e. Competition Index (ComIn), as well as MF and MR. All three methods were tested according to Anatomical Therapeutic Chemical (ATC) classification levels 2-5. For each AE, SDR detection was performed, first in the complete database, and second after removing reports mentioning competitors; SDRs only detected after the removal were unmasked. All unmasked SDRs were validated using the Summary of Product Characteristics, and constituted the reference dataset used for computing the performance for SDR unmasking (area under the curve [AUC], Se)., Results: Performance of the ComIn was highest when considering competitors at ATC level 3 (AUC: 62 %; Se: 52 %); similar results were obtained with MF and MR., Conclusion: The ComIn could greatly minimize the competition bias in SDR detection. Further study using a larger dataset is needed.

Purpose: To test an automated method to decrease the number of false-positive (FP) signals of disproportionate reportings (SDRs) generated by co-prescription., Methods: Automated backward stepwise removal of reports concerning the drug associated with the highest ranked SDR for an event was tested for gastric and oesophageal haemorrhages (GOH), central nervous system haemorrhages and cerebrovascular accidents (CNSH), ischaemic coronary artery disorders and muscle pains (MP) using the reporting odds ratio in the French spontaneous reporting research database. After ranking SDRs detected in the complete dataset on the lower limit of the reporting odds ratio 95% confidence interval, reports concerning the drug with the highest ranked SDR were removed. In the dataset thus generated, SDRs were again identified, ranked and reports related to the drug involved in the newly highest ranked SDR removed. The process was repeated until no signal was detected. Initially detected SDRs eliminated using this technique were assessed regarding the summary of products characteristics and the literature to determine their FP nature., Results: Seventeen SDRs were successively eliminated for GOH, 37 for CNSH, 15 for ischaemic coronary artery disorders, and 36 for MP. Four were FP for GOH, 29 for CNSH, 7 for ACI and none were FP for MP. The positive predictive value of the backward stepwise removal procedure in identifying FP SDRs ranged from 0% (MP) to 78.4% (CNSH)., Conclusions: Although further adjustment is needed to improve the method presented herein, our results suggest that numerous FP signals because of co-prescription bias could be eliminated using an automated method., (Copyright © 2013 John Wiley & Sons, Ltd.)

Background: An updated probabilistic causality assessment method and the Liverpool algorithm presented as an improved version of the Naranjo algorithm, one of the most used and accepted causality assessment methods, have recently been proposed., Objective: In order to test the validity of the probabilistic method in routine pharmacovigilance, results provided by the Naranjo and Liverpool algorithms, as well as the updated probabilistic method, were each compared with a consensual expert judgement taken as reference., Methods: A sample of 59 drug-event pairs randomly sampled from spontaneous reports to the French pharmacovigilance system was assessed by expert judgement until reaching consensus and by members of a pharmacovigilance unit using the updated probabilistic method, the Naranjo and Liverpool algorithms. Probabilities given by the probabilistic method, and categories obtained by both the Naranjo and the Liverpool algorithms were compared as well as their sensitivity, specificity, positive and negative predictive values., Results: The median probability for drug causation given by the consensual expert judgement was 0.70 (inter-quartile range, IQR 0.54-0.84) versus 0.77 (IQR 0.54-0.91) for the probabilistic method. For the Naranjo algorithm, the 'possible' causality category was predominant (61 %), followed by 'probable' (35 %), 'doubtful', and 'almost certain' categories (2 % each). Category distribution obtained with the Liverpool algorithm was similar to that obtained by the Naranjo algorithm with a majority of 'possible' (61 %) and 'probable' (30 %) followed by 'definite' (7 %) and 'unlikely' (2 %). For the probabilistic method, sensitivity, specificity, positive and negative predictive values were 0.96, 0.56, 0.92 and 0.71, respectively. For the Naranjo algorithm, depending on whether the 'possible' category was considered in favour or in disfavour of drug causation, sensitivity was, respectively, 1 or 0.42, specificity 0.11 or 0.89, negative predictive value 1 or 0.22 and positive predictive value 0.86 or 0.95; results were identical for the Liverpool algorithm., Conclusion: The logistic probabilistic method gave results closer to the consensual expert judgment than either the Naranjo or Liverpool algorithms whose performance were strongly dependent on the meaning given to the 'possible' category. Owing to its good sensitivity and positive predictive value and by providing results as continuous probabilities, the probabilistic method seems worthy to use for a trustable assessment of adverse drug reactions in routine practice.

Background: 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., Objective: The objective of this study was to explore the effects of event-competition bias on safety signals generated from spontaneous reporting databases., Methods: 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., Results: 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., Conclusion: 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.

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.

Objective: Updating a logistic causality assessment method to improve its agreement with consensual expert judgment (CEJ)., Study Design and Setting: A random sample of 53 drug-event pairs from a pharmacovigilance database were evaluated independently by CEJ and by a group of experts in pharmacovigilance using the logistic method. Causes of disagreement between both approaches were analyzed, and changes in the assessment of some criteria of the logistic method were proposed and tested in models. The model giving results closest to the CEJ was retained and compared with the initial version on another set of drug-event pairs., Results: Finally, only the criterion "Search for nondrug cause" was changed into "Search for other causes." The assessment not investigated, possible other cause decreased the probability of drug causation instead of being neutral, whereas the assessment not applicable, not required remained neutral. This new version presents much improved specificity (0.56 vs. 0.33), relatively good sensitivity (0.96), and positive and negative predictive values (0.92 and 0.71)., Conclusion: The updated logistic method presented here improves the initial version that had poor specificity and tended to overestimate drug causation. This new version presents satisfactory characteristics to be used in routine pharmacovigilance., (Copyright © 2012 Elsevier Inc. All rights reserved.)

Background: Improving the detection of drug safety signals has led several pharmacovigilance regulatory agencies to incorporate automated quantitative methods into their spontaneous reporting management systems. The three largest worldwide pharmacovigilance databases are routinely screened by the lower bound of the 95% confidence interval of proportional reporting ratio (PRR₀₂.₅), the 2.5% quantile of the Information Component (IC₀₂.₅) or the 5% quantile of the Gamma Poisson Shrinker (GPS₀₅). More recently, Bayesian and non-Bayesian False Discovery Rate (FDR)-based methods were proposed that address the arbitrariness of thresholds and allow for a built-in estimate of the FDR. These methods were also shown through simulation studies to be interesting alternatives to the currently used methods., Objective: The objective of this work was twofold. Based on an extensive retrospective study, we compared PRR₀₂.₅, GPS₀₅ and IC₀₂.₅ with two FDR-based methods derived from the Fisher's exact test and the GPS model (GPS(pH0) [posterior probability of the null hypothesis H₀ calculated from the Gamma Poisson Shrinker model]). Secondly, restricting the analysis to GPS(pH0), we aimed to evaluate the added value of using automated signal detection tools compared with 'traditional' methods, i.e. non-automated surveillance operated by pharmacovigilance experts., Methods: The analysis was performed sequentially, i.e. every month, and retrospectively on the whole French pharmacovigilance database over the period 1 January 1996-1 July 2002. Evaluation was based on a list of 243 reference signals (RSs) corresponding to investigations launched by the French Pharmacovigilance Technical Committee (PhVTC) during the same period. The comparison of detection methods was made on the basis of the number of RSs detected as well as the time to detection., Results: Results comparing the five automated quantitative methods were in favour of GPS(pH0) in terms of both number of detections of true signals and time to detection. Additionally, based on an FDR threshold of 5%, GPS(pH0) detected 87% of the RSs associated with more than three reports, anticipating the date of investigation by the PhVTC by 15.8 months on average., Conclusions: Our results show that as soon as there is reasonable support for the data, automated signal detection tools are powerful tools to explore large spontaneous reporting system databases and detect relevant signals quickly compared with traditional pharmacovigilance methods.

Background: Anti-tumour necrosis factor (TNF) agents, through their intense immunoregulatory effect, have been suspected to increase the risk of malignant lymphoma. However, the classical epidemiological approaches conducted over about the last 10 years have not totally succeeded in addressing the question of a causal or artifactual association. Therefore, the analysis of a substantial set of case reports, although usually considered as poorly generalizable to the general population, could be particularly informative. Two main sources of case reports in postmarketing settings are available; publications in medical journals and reports to pharmacovigilance systems., Objective: The aim of the study was to compare the characteristics of case reports from both these sources in order to understand whether they provided the same information for the investigation of the causal link between lymphoma and anti-TNF agents., Methods: All case reports of malignant lymphoma in patients treated with an anti-TNF agent published in MEDLINE and all reports to the French pharmacovigilance system up to 1 February 2010 were identified. Cases of malignant lymphoma identified in postmarketing surveillance from both sources were compared regarding the following variables: age, sex, anti-TNF agent involved, indication for use, type of lymphoma, prior or concomitant immunosuppressive drugs and time to onset of lymphoma., Results: A total of 81 published case reports and 61 cases reported to the French pharmacovigilance system were compared. In published reports, patients were younger (p = 0.03) and more frequently receiving a first anti-TNF treatment (p = 0.03), particularly infliximab (p = 0.03). Conversely, in the pharmacovigilance system reports, a succession of different anti-TNFs (p = 0.03) and adalimumab (p < 0.0001) were more frequently reported. Lymphomas in patients treated with anti-TNF agents for Crohn's disease were more prevalent in published cases than in pharmacovigilance reports (p < 0.0001), and in particular involved hepatosplenic T-cell lymphoma. Conversely, rheumatoid arthritis was the main indication for anti-TNF agents in pharmacovigilance reports (p = 0.01). Time to onset was markedly shorter in published cases (median 12 months) than in pharmacovigilance reports (median 30 months; p = 0.0001)., Conclusions: Characteristics of published cases and those reported to the French pharmacovigilance system differed markedly for all characteristics tested, except sex and the use of prior or concomitant immunosuppressive drugs. Published case reports favoured convincing arguments for drug causation whereas cases reported to the pharmacovigilance system were more disparate but could describe more accurately the reality of lymphoma occurrence in this particular population. These results argue for the use of the pharmacovigilance reports when case reports are used to investigate the causal link between lymphoma and anti-TNF agents at the population level. Data from cases notified to the French pharmacovigilance system did not indicate an increased risk of lymphoma during the early phase of anti-TNF treatment. To confirm this hypothesis, a study combining pharmacovigilance reports from several countries, or, if feasible, a cohort study both with a large sample size and a long duration of follow-up would be required.

Background: Different methods have been proposed for assessing a possible causal link between a drug treatment and an adverse event in individual patients. They approximately belong to three main categories: expert judgement, operational algorithms and probabilistic approaches., Objective: To compare, in a set of actual drug adverse event reports, three different methods for assessing drug causality, each belonging to one of the three main categories: expert judgement, the algorithm used by the French pharmacovigilance centres since 1985, and a novel method based on the logistic function., Methods: Fifty drug-event pairs were randomly sampled from the database of the Bordeaux pharmacovigilance centre, France. To serve as the gold standard, the probability for drug causation, from 0 to 1, was first determined for each drug-event pair by a panel of senior experts until consensus was reached. Causality was then assessed by members of the Bordeaux pharmacovigilance centre by using the French algorithm and the logistic method. Results expressed as a probability with the logistic method and as a score from 0 to 4 with the French algorithm were then compared with consensual expert judgement, as were the sensitivity, specificity and positive and negative predictive values., Results: Probabilities ranged from 0.08 to 0.99 (median 0.58; mean 0.60) for experts versus 0.18-0.88 (median 0.73; mean 0.67) for the logistic method. Consensual expert judgement was not discriminant (p = 0.50) in ten cases. For the algorithm, only three of five causality scores were found, doubtful scores being clearly predominant (74%) followed by possible (16%) and probable (10%) scores. Sensitivity and specificity were 0.96 and 0.42, respectively, for the logistic method versus 0.42 and 0.92 for the algorithm. Positive and negative predictive values were 0.78 and 0.83, respectively, for the logistic method versus 0.92 and 0.42 for the algorithm., Conclusions: Agreement between the three approaches was poor, and only satisfactory for drug events judged as drug-induced by consensual expert judgement. The logistic method showed high sensitivity at the expense of poor specificity. Conversely, the algorithm had poor sensitivity but good specificity. The comparatively good sensitivity and positive predictive values of the logistic method suggest that it may be more useful in the routine or automated assessment of case reports of suspected but still unknown adverse drug reactions. With a substantial rate of false positives relative to true negatives (low specificity), the logistic method does not replace, but can be complemented by, critical clinical assessment of individual cases in evaluating drug-related risk.

Pharmacovigilance spontaneous reporting systems are primarily devoted to early detection of the adverse reactions of marketed drugs. They maintain large spontaneous reporting databases (SRD) for which several automatic signalling methods have been developed. A common limitation of these methods lies in the fact that they do not provide an auto-evaluation of the generated signals so that thresholds of alerts are arbitrarily chosen. In this paper, we propose to revisit the Gamma Poisson Shrinkage (GPS) model and the Bayesian Confidence Propagation Neural Network (BCPNN) model in the Bayesian general decision framework. This results in a new signal ranking procedure based on the posterior probability of null hypothesis of interest and makes it possible to derive with a non-mixture modelling approach Bayesian estimators of the false discovery rate (FDR), false negative rate, sensitivity and specificity. An original data generation process that can be suited to the features of the SRD under scrutiny is proposed and applied to the French SRD to perform a large simulation study. Results indicate better performances according to the FDR for the proposed ranking procedure in comparison with the current ones for the GPS model. They also reveal identical performances according to the four operating characteristics for the proposed ranking procedure with the BCPNN and GPS models but better estimates when using the GPS model. Finally, the proposed procedure is applied to the French data., (Copyright (c) 2009 John Wiley & Sons, Ltd.)

Aims: To evaluate agreement between five senior experts when assessing seven causality criteria and the probability of drug causation., Methods: A sample of 31 adverse event-drug pairs was constituted. For each pair, five experts separately assessed (i) the probability of drug causation, which was secondarily divided into seven causality levels: ruled out (0-0.05), unlikely (0.06-0.25), doubtful (0.26-0.45), indeterminate (0.46-0.55), plausible (0.56-0.75), likely (0.76-0.95), and certain (0.96-1); and (ii) seven causality criteria. To test discrepancies between experts, the kappa index was used., Results: The agreement of the five experts was very poor (kappa = 0.05) for the probability of drug causation. Among the seven levels of causality, only 'doubtful' showed a significant rate of agreement (kappa = 0.32, P < 0.001). For all criteria, the kappa index was significant except for the item 'risk(s) factor(s)' (kappa = 0.09). Agreement between experts was good (0.64, P < 0.001) only for the criterion 'reaction at site of application or toxic plasma concentration of the drug or validated test'. However, the rate of agreement with kappa indices of the causality criteria ranged from 0.12 to 0.38., Conclusions: This study confirms that in the absence of an operational procedure, agreement between experts is low. This should be considered when designing a causality assessment method. In particular, criteria inducing a low level of agreement should have their weight reduced.

Pharmacovigilance aims at detecting the adverse effects of marketed drugs. It is generally based on the spontaneous reporting of events thought to be the adverse effects of drugs. Spontaneous Reporting Systems (SRSs) supply huge databases that pharmacovigilance experts cannot exhaustively exploit without data mining tools. Data mining methods; i.e., statistical association measures in conjunction with signal generation criteria, have been proposed in the literature but there is no consensus regarding their applicability and efficiency, especially since such methods are difficult to evaluate on the basis of actual data. The objective of this paper is to evaluate association measures on simulated datasets obtained with SRS modeling. We compared association measures using the percentage of false positive signals among a given number of the most highly ranked drug-event combinations according to the values of the association measures. By considering 150 drugs and 100 adverse events, these percentages of false positives, among the 500 most highly ranked drug-event couples, vary from 1.1% to 53.4% (averages over 1000 simulated datasets). As the measures led to very different results, we could identify which measures appeared to be the most relevant for pharmacovigilance.

Objective: To analyse the nature of adverse drug reactions, their frequency, the drugs incriminated, their outcome, the different ways to treat them and their cost., Methods: Descriptive, prospective and non-comparative study, based on spontaneous reporting. Each patient was followed-up over a 6-month period., Results: One hundred and fifteen adverse events were reported in 113 patients (58 men and 55 women), with a mean age of 59.5 years. Cutaneous and haematological reactions predominated, with a majority of cardiovascular and psychotropic agents incriminated. Sixty-three percent of adverse reactions were serious: 49% led to (or prolonged) hospitalisation, 6% led to death, 4% to after effects, 3% were life-threatening and 1% led to congenital malformation. The mean duration of an adverse reaction was of 37.4 days. The total cost of these 115 adverse reactions was an estimated 610.110 Euros, with a mean global cost of 5.305 Euros per event., Conclusion: This study shows the individual and collective extent of the impact of adverse individual and collective reactions.

Background: The French pharmacovigilance system is based on a network of 31 regional centres located in teaching hospitals and coordinated by the French Medicines Agency ("Agence Française de Sécurité Sanitaire des Produits de Santé" [Afssaps]). Since 1984, they have shared a common database of adverse drug reactions (ADRs) that are spontaneously reported by healthcare professionals. The objective of this study is to describe the characteristics of the reports and the reporting trends in the French pharmacovigilance spontaneous reporting database from 1986 to 2001., Methods: All the reports from January 1986 to December 2001 were included. Drugs and ADRs were translated to anatomical therapeutic chemical (ATC) codes and MedDRA classifications, respectively., Results: The total number of reports was 197 580 over the 16-year period, with linearly increase over time. The median (interquartile range [IQR]) age of the patients was 53 (34-70) and the male/female ratio was 0.82. The median (IQR) time between the date of occurrence of the ADR and the date of report was 73 days (34-166). The reporter was a specialist in 74% of the reports and a general practitioner in 17%. The annual rate of reporting according to medical demography strongly increased for the specialists, especially since 1994. At least one ADR was considered as serious in 44.8% of the reports. The ADRs were most frequently related to nervous system drugs (23%), followed by cardiovascular drugs (19%) and systemic anti-infectives (17%). The latter class had the fastest progression mostly due to antiretroviral therapy since 1996. According to the Medical Dictionary for Regulatory Activities (MedDRA) coding, the system organ most often reported was skin and subcutaneous tissue disorders (29%), followed by nervous system disorders (19%), gastrointestinal disorders (12%), blood and lymphatic system disorders (12%), vascular disorders (12%) and general disorders and administration site conditions (12%)., Discussion: All spontaneous reporting systems are affected by under-reporting. One of their goals is to generate early signals, which might be more affected by reporting bias than by under-reporting. Some improvements should be made in the design of the French database, but data collected since 1986 constitute an essential tool for the routine work of the 31 pharmacovigilance centres., Conclusion: This first description of the data of the French pharmacovigilance database involving all drugs and ADRs shows an increasing tendency to reporting over time, especially in specialists and for systemic anti-infective drugs. The database that uses hierarchical international classifications for drugs and adverse reactions may be used for further studies and could be the basis for an automatic signal generation system.