Marie Lindquist

A Bayesian neural network method for adverse drug reaction signal generation

AbstractObjective: The database of adverse drug reactions (ADRs) held by the Uppsala Monitoring Centre on behalf of the 47 countries of the World Health Organization (WHO) Collaborating Programme for International Drug Monitoring contains nearly two million reports. It is the largest database of this sort in the world, and about 35 000 new reports are added quarterly. The task of trying to find new drug–ADR signals has been carried out by an expert panel, but with such a large volume of material the task is daunting. We have developed a flexible, automated procedure to find new signals with known probability difference from the background data. Method: Data mining, using various computational approaches, has been applied in a variety of disciplines. A Bayesian confidence propagation neural network (BCPNN) has been developed which can manage large data sets, is robust in handling incomplete data, and may be used with complex variables. Using information theory, such a tool is ideal for finding drug–ADR combinations with other variables, which are highly associated compared to the generality of the stored data, or a section of the stored data. The method is transparent for easy checking and flexible for different kinds of search. Results: Using the BCPNN, some time scan examples are given which show the power of the technique to find signals early (captopril–coughing) and to avoid false positives where a common drug and ADRs occur in the database (digoxin–acne; digoxin–rash). A routine application of the BCPNN to a quarterly update is also tested, showing that 1004 suspected drug–ADR combinations reached the 97.5% confidence level of difference from the generality. Of these, 307 were potentially serious ADRs, and of these 53 related to new drugs. Twelve of the latter were not recorded in the CD editions of The physicians Desk Reference orMartindales Extra Pharmacopoea and did not appear in Reactions Weekly online. Conclusion: The results indicate that the BCPNN can be used in the detection of significant signals from the data set of the WHO Programme on International Drug Monitoring. The BCPNN will be an extremely useful adjunct to the expert assessment of very large numbers of spontaneously reported ADRs.

Antipsychotic drugs and heart muscle disorder in international pharmacovigilance: data mining study

Abstract Objectives: To examine the relation between antipsychotic drugs and myocarditis and cardiomyopathy. Design: Data mining using bayesian statistics implemented in a neural network architecture. Setting: International database on adverse drug reactions run by the World Health Organization programme for international drug monitoring. Main outcome measures: Reports mentioning antipsychotic drugs, cardiomyopathy, or myocarditis. Results: A strong signal existed for an association between clozapine and cardiomyopathy and myocarditis. An association was also seen with other antipsychotics as a group. The association was based on sufficient cases with adequate documentation and apparent lack of confounding to constitute a signal. Associations between myocarditis or cardiomyopathy and lithium, chlorpromazine, fluphenazine, haloperidol, and risperidone need further investigation. Conclusions: Some antipsychotic drugs seem to be linked to cardiomyopathy and myocarditis. The study shows the potential of bayesian neural networks in analysing data on drug safety.

VigiBase, the WHO Global ICSR Database System: Basic Facts

The main aim of the WHO International Drug Monitoring Programme, started in 1968, is to identify the earliest possible pharmacovigilance signals. The program now has more than 80 member countries from all parts of the world contributing individual case safety reports (ICSRs) to the WHO Global ICSR Database System, VigiBase. VigiBase is maintained and developed on behalf of WHO by the Uppsala Monitoring Centre (UMC), situated in Uppsala, Sweden. The database system includes the ICH E2B compatible ICSR database, the WHO Drug Dictionaries (WHO-DD and -DDE), and the medical terminologies WHO Adverse Reaction Terminology (WHO-ART), International Classification of Diseases (ICD), and the Medical Dictionary for Regulatory Activities (MedDRA). Apart from data management and quality assurance tools, the VigiBase system includes a web-based reporting tool, an automated signal detection process using advanced data mining, and search facilities, available to the member countries and, on request, to other stakeholders.

A Retrospective Evaluation of a Data Mining Approach to Aid Finding New Adverse Drug Reaction Signals in the WHO International Database

AbstractBackground: The detection of new drug safety signals is of growing importance with ever more new drugs becoming available and exposure to medicines increasing. The task of evaluating information relating to safety lies with national agencies and, for international data, with the World Health Organization Programme for International Drug Monitoring. Rationale: An established approach for identifying new drug safety signals from the international database of more than 2 million case reports depends upon clinical experts from around the world. With a very large amount of information to evaluate, such an approach is open to human error. To aid the clinical review, we have developed a new signalling process using Bayesian logic, applied to data mining, within a confidence propagation neural network (Bayesian Confidence Propagation Neural Network; BCPNN). Ultimately, this will also allow the evaluation of complex variables. Methods: The first part of this study tested the predictive value of the BCPNN in new signal detection as compared with reference literature sources (Martindale’s Extra Pharmacopoeia in 1993 and July 2000, and the Physicians Desk Reference in July 2000). In the second part of the study, results with the BCPNN method were compared with those of the former signalling procedure. Results: In the study period (the first quarter of 1993) 107 drug—adverse reaction combinations were highlighted as new positive associations by the BCPNN, and referred to new drugs. 15 drug—adverse reaction combinations on new drugs became negative BCPNN associations in the study period. The BCPNN method detected signals with a positive predictive value of 44% and the negative predictive value was 85%. 17 as yet unconfirmed positive associations could not be dismissed with certainty as false positive signals.Of the 10 drug—adverse reaction signals produced by the former signal detection system from data sent out for review during the study period, 6 were also identified by the BCPNN. These 6 associations have all had a more than 10-fold increase of reports and 4 of them have been included in the reference sources. The remaining 4 signals that were not identified by the BCPNN had a small, or no, increase in the number of reports, and are not listed in the reference sources. Conclusion: Our evaluation showed that the BCPNN approach had a high and promising predictive value in identifying early signals of new adverse drug reactions.

A Data Mining Approach for Signal Detection and Analysis

The WHO database contains over 2.5 million case reports, analysis of this data set is performed with the intention of signal detection. This paper presents an overview of the quantitative method used to highlight dependencies in this data set.The method Bayesian confidence propagation neural network (BCPNN) is used to highlight dependencies in the data set. The method uses Bayesian statistics implemented in a neural network architecture to analyse all reported drug adverse reaction combinations.This method is now in routine use for drug adverse reaction signal detection. Also this approach has been extended to highlight drug group effects and look for higher order dependencies in the WHO data.Quantitatively unexpectedly strong relationships in the data are highlighted relative to general reporting of suspected adverse effects; these associations are then clinically assessed.

Signal selection and follow-up in pharmacovigilance.

The detection of unknown and unexpected connections between drug exposure and adverse events is one of the major challenges of pharmacovigilance. For the identification of possible connections in large databases, automated statistical systems have been introduced with promising results. From the large numbers of associations so produced, the human mind has to identify signals that are likely to be important, in need of further assessment and follow-up and that may require regulatory action. Such decisions are based on a variety of clinical, epidemiological, pharmacological and regulatory criteria. Likewise, there are a number of criteria that underlie the subsequent evaluation of such signals. A good understanding of the logic underlying these processes fosters rational pharmacovigilance and efficient drug regulation. In the future a combination of quantitative and qualitative criteria may be incorporated in automated signal detection.

An ABC of drug-related problems.

The problems relating to the use of medicines are manifold. They may differ in pharmacological, pathological, epidemiological and legal respects, and may have different consequences, for example, as regards scientific study, regulation or rational use. Pharmacovigilance is concerned with all such problems: adverse effects and interactions as well as problems relating to ineffectiveness, inappropriate use, counterfeiting, dependence or poisoning.Practically all medicine-related problems can be classified in one basic system, taking into account their characteristics and distinctions. This system distinguishes between appropriate and inappropriate drug use, dose-related and dose-unrelated problems, and types A (‘drug actions’), B (‘patient reactions’) and C (‘statistical’) adverse effects. This classification may serve as an educational tool and may be useful in when choosing a study method and for the design of effective strategies in pharmacovigilance.

The association between antidepressant use and disturbances in glucose homeostasis: evidence from spontaneous reports.

ObjectivesDepression is common in patients with diabetes, and the use of antidepressants may impair glycaemic control. We assessed the association between antidepressant use and hyper- and hypoglycaemia.MethodsBased on spontaneous reports listed in the World Health Organization (WHO) Adverse Drug Reaction Database, a case-control study was conducted. The study base consisted of all adverse drug reactions (ADRs) ascribed to antidepressants, antipsychotics and benzodiazepines between 1969 and 2005. Cases were defined as reported ADRs classified as hyper- or hypoglycaemia and separated in different study populations. All other reports were considered as controls. Exposure to antidepressants was the primary determinant investigated. Benzodiazepines and antipsychotics were chosen as reference groups. Potential confounding factors, namely, age, gender, use of antidiabetic medication, use of hyper- or hypoglycaemia-inducing comedication and reporting year, were determined on the index date. Multivariate logistic regression was used to evaluate the strength of the association, which was expressed as reporting odds ratios (RORs) with 95% confidence intervals (95% CI).ResultsOverall, the use of antidepressants was associated with hyperglycaemia [ROR 1.52 (95% CI: 1.20–1.93)] and of hypoglycaemia [ROR 1.84 (95% CI: 1.40–2.42)]. The association with hyperglycaemia was most pronounced for antidepressants with affinity for the 5-HT2c receptor, histamine-1 receptor and norepinephrinic (NE) reuptake transporter. The association with hypoglycaemia was most pronounced for antidepressants with affinity for the serotonin reuptake transporter.ConclusionThe results of this study strengthen the findings in individual case reports that the use of antidepressants is associated with disturbances in glucose homeostasis.

Bayesian neural networks with confidence estimations applied to data mining

An international database of case reports, each one describing a possible case of adverse drug reactions (ADRs), is maintained by the Uppsala Monitoring Centre (UMC), for the WHO international prog ...

Withdrawal reactions with selective serotonin re-uptake inhibitors as reported to the WHO system

AbstractObjective: The present study was performed both to investigate whether there might be a difference between the selective serotonin re-uptake inhibitors, (SSRIs) with regard to the incidence of withdrawal reactions, and to describe the associated symptoms. From the WHO database, therefore, all case reports from the year of introduction for each of the SSRIs, fluoxetine, paroxetine and sertraline, were retrieved. Sales figures were obtained from Intercontinental Medical Statistics International. The reporting rates were calculated as the number of reports per million defined daily doses (DDDs) sold per year. Results: The reporting rate of withdrawal reactions for paroxetine was found to be higher than that for sertraline and fluoxetine in each of the countries selected for detailed analyses (US, UK and Australia), as well as for all 16 countries combined. Moreover, using the WHO system of organ classification, the ratio of central nervous system to psychiatric withdrawal symptoms was 1.9 and 2.1 for paroxetine and sertraline, respectively, whereas that for fluoxetine was 0.48, indicating a possible qualitative difference between the SSRIs with respect to the nature of the withdrawal syndrome.