I. Ralph Edwards

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.

Suspected Adverse Drug Reactions Reported For Children Worldwide An Exploratory Study Using VigiBase

AbstractBackground: As a first step towards implementing routine screening of safety issues specifically related to children at the Uppsala Monitoring Centre, this study was performed to explore reporting patterns of adverse reactions in children. Objective: The first aim of this study was to characterize and contrast child reports against adult reports in an overall drug and adverse reaction review. The second aim was to highlight increases in reporting of specific adverse reactions during recent years subdivided by age group. Study Design: This was an exploratory study of internationally compiled individual case safety reports (ICSRs). Setting: Reports were extracted from the WHO global ICSR database, VigiBase, up until 5 February 2010. The reports in VigiBase originate from 97 countries and the likelihood that a medicine caused the adverse effect may vary from case to case. Suspected duplicate and vaccine reports were excluded from the analysis, as were reports with age not specified. The Medical Dictionary for Regulatory Activities (MedDRA®) and the WHO Anatomical Therapeutic Chemical (ATC) classification were used to group adverse reactions and drugs. Patients: In the general review, reports from 1968 to 5 February 2010 were divided into child (aged 0–17 years) and adult (≥18 years) age groups. To highlight increases in reporting rates of specific adverse reactions during recent years, reports from 2005 to February 2010 were compared with reports from 1995 to 1999. The ten adverse reactions with the greatest difference in the proportion of reports between the two time periods were reviewed. In the latter analysis, the reports were subdivided into age groups: neonates ≤27 days; infants 28 days–23 months; children 2–11 years; and adolescents 12–17 years. Results: A total of 3 472 183 reports were included in the study, of which 7.7% (268 145) were reports for children (0–17 years). Fifty-three percent of the child reports were for males, whilst 39% of reports in the adult group were for males. The proportion of reports involving children among Asian reports was 14% and was 15% among reports from Africa and Latin America, including the Caribbean. Among reports from North America, Oceania and Europe, 7% of the reports involved children. For the ATC drug classification groups, the largest difference in percentage units between the child and adult groups was seen for the anti-infective (33 vs 15%), respiratory (11 vs 5%) and dermatological (12 vs 7%) drug groups. Skin reactions were most commonly reported for the children; these were recorded in 35% of all reports for children and 23% of all reports for adults. Medication error-related terms in the younger age groups were reported with an increased frequency during recent years. This was particularly noticeable for the infants aged 28 days–23 months, recorded with accidental overdose and drug toxicity. Reactions reported in suspected connection to medicines used for attention-deficit hyperactivity disorders (ADHD) completely dominated the 2-to 11-year age group and were also common for the adolescents. This study presents variations in the reporting pattern in different age groups in VigiBase which, in some cases, could be due to susceptibilities to specific drug-related problems in certain age groups. Other likely explanations might be common drug usage and childhood diseases in these age groups. Conclusions: Reports in VigiBase received internationally for more than 40 years reflect real concerns for children taking medicines. The study highlights adverse reactions with an increased reporting during recent years, particularly those connected to the introduction of ADHD medicines in the child population. To enhance patient safety, medication errors indicating administration and dosing difficulties of drugs, especially in the younger age groups, require further attention.

A statistical methodology for drug-drug interaction surveillance

Interaction between drug substances may yield excessive risk of adverse drug reactions (ADRs) when two drugs are taken in combination. Collections of individual case safety reports (ICSRs) related to suspected ADR incidents in clinical practice have proven to be very useful in post-marketing surveillance for pairwise drug--ADR associations, but have yet to reach their full potential for drug-drug interaction surveillance. In this paper, we implement and evaluate a shrinkage observed-to-expected ratio for exploratory analysis of suspected drug-drug interaction in ICSR data, based on comparison with an additive risk model. We argue that the limited success of previously proposed methods for drug-drug interaction detection based on ICSR data may be due to an underlying assumption that the absence of interaction is equivalent to having multiplicative risk factors. We provide empirical examples of established drug-drug interaction highlighted with our proposed approach that go undetected with logistic regression. A database wide screen for suspected drug-drug interaction in the entire WHO database is carried out to demonstrate the feasibility of the proposed approach. As always in the analysis of ICSRs, the clinical validity of hypotheses raised with the proposed method must be further reviewed and evaluated by subject matter experts.

Duplicate detection in adverse drug reaction surveillance

The WHO Collaborating Centre for International Drug Monitoring in Uppsala, Sweden, maintains and analyses the world’s largest database of reports on suspected adverse drug reaction (ADR) incidents that occur after drugs are on the market. The presence of duplicate case reports is an important data quality problem and their detection remains a formidable challenge, especially in the WHO drug safety database where reports are anonymised before submission. In this paper, we propose a duplicate detection method based on the hit-miss model for statistical record linkage described by Copas and Hilton, which handles the limited amount of training data well and is well suited for the available data (categorical and numerical rather than free text). We propose two extensions of the standard hit-miss model: a hit-miss mixture model for errors in numerical record fields and a new method to handle correlated record fields, and we demonstrate the effectiveness both at identifying the most likely duplicate for a given case report (94.7% accuracy) and at discriminating true duplicates from random matches (63% recall with 71% precision). The proposed method allows for more efficient data cleaning in post-marketing drug safety data sets, and perhaps other knowledge discovery applications as well.

Reasons for Reporting Adverse Drug Reactions—Some Thoughts Based on an International Review

A pilot study was made to explore positive reasons for physicians and pharmacists taking time to report adverse reactions, rather than reasons for failing to report which has been studied by many authors. The 34 national drug monitoring centres participating in the international programme at the time of the study were asked by letter from the WHO Collaborating Centre for International Drug Monitoring, Uppsala to investigate the reasons why adverse reactions were reported. National Centres were asked to write to 20 consecutive reporters, sending each a copy of their own report, asking why they had chosen to report that particular reaction, and asking for more general comment. Twelve countries responded with information about the habits and views of the reporters of 177 cases. Since this was an explorative pilot study the letter to reporters deliberately had only an open question about reason for reporting. Categories were developed by the WHO Centre from the responses given. Reasons for reporting fell into a total of 14 categories with the great majority in the top six:

Statins, neuromuscular degenerative disease and an amyotrophic lateral sclerosis-like syndrome: an analysis of individual case safety reports from vigibase.

AbstractBackground: The WHO Foundation Collaborating Centre for International Drug Monitoring (Uppsala Monitoring Centre [UMC]) has received many individual case safety reports (ICSRs) associating HMG-CoA reductase inhibitor drug (statin) use with the occurrence of muscle damage, including rhabdomyolysis, and also peripheral neuropathy. A new signal has now appeared of disproportionally high reporting of upper motor neurone lesions. Aim and Scope: The aim of this paper is to present the upper motor neurone lesion cases, with other evidence, as a signal of a relationship between statins and an amyotrophic lateral sclerosis (ALS)-like syndrome. The paper also presents some arguments for considering that a spectrum of severe neuromuscular damage may be associated with statin use, albeit rarely. The paper does not do more than raise the signal for further work and analysis of what must be regarded as a potentially very serious and perhaps avoidable or reversible adverse reaction, though it also suggests action to be taken if an ALS-like syndrome should occur in a patient using statins. Methods: The 43 reports accounting for the disproportional reports in Vigibase (the database of the WHO Programme for International Drug Monitoring) are summarised and analysed for the diagnosis of an ALS-like syndrome. The issues of data quality and potential reporting bias are considered. Results: ‘Upper motor neurone lesion’ is a rare adverse event reported in relationship to drugs in Vigibase (a database containing nearly 4 million ICSRs). Of the total of 172 ICSRs on this reported term, 43 were related to statins, of which 40 were considered further: all but one case was reported as ALS. In 34/40 reports a statin was the sole reported suspected drug. The diagnostic criteria were variable, and seven of the statin cases also had features of peripheral neuropathy. Of a total of 5534 ICSRs of peripheral neuropathy related to any drug in Vigibase, 547 were on statins. The disproportional reporting of statins and upper motor neurone lesion persisted after age stratification, and such disproportionality was not seen for statins and Parkinson’s disease, Alzheimer’s disease, extrapyramidal disorders, or multiple sclerosis-like syndromes. Discussion: Because the cases were sometimes atypical we propose the use of the term ‘ALS-like syndrome’ and speculate whether this is part of a spectrum of rare neuromuscular damage. The diagnosis of ALS is often problematic, and the insidiousness and chronicity of the disease make causality with a drug difficult to assess. The disproportionally high reporting makes this an important signal nevertheless, since ALS is serious clinically and statins are so widely used. Wide use of the statins also makes a chance finding more probable, but is unlikely to cause disproportional reporting when there are no obvious biases identified. Conclusion: We emphasise the rarity of this possible association, and also the need for further study to establish whether a causal relationship exists. We do advocate that trial discontinuation of a statin should be considered in patients with serious neuromuscular disease such as the ALS-like syndrome, given the poor prognosis and a possibility that progression of the disease may be halted or even reversed.

Pharmacovigilance and biosimilars: considerations, needs and challenges

Introduction: Biosimilars are biologic medicines that are highly similar to approved biologics, notwithstanding minor differences in clinically inactive components. Since 2007, biosimilars have been approved for use in patients in the European Union (EU) and other regions. European experience provides several lessons as the United States (US) healthcare system prepares for biosimilar approvals. These lessons emphasize the need for adequate efficacy and safety studies, post-marketing surveys and a robust pharmacovigilance system that can accurately track and trace biologics, including biosimilars and their reference products, from the patient to the manufacturer. Areas covered: We review the EU experience with biosimilar pharmacovigilance and discuss the implications for biosimilar pharmacovigilance in the USA. Furthermore, we review several aspects of biosimilar pharmacovigilance, including cohort event monitoring, traceability, biosimilar interchangeability, pharmacovigilance system development, nomenclature and counterfeit tracking. Expert opinion: The availability of biosimilars as lower-cost biologics must carefully consider issues of safety, efficacy and traceability. Stringent pharmacovigilance procedures are required to detect potential differences in safety signals between biosimilars and their reference products. Pharmacovigilance of biologics should include processes that are easily used by prescribing practitioners to ensure that data are consistent and new safety signals are properly reported and assigned to the correct product.

Patient-Reported Outcome Measures in Safety Event Reporting: PROSPER Consortium Guidance

The Patient-Reported Outcomes Safety Event Reporting (PROSPER) Consortium was convened to improve safety reporting by better incorporating the perspective of the patient. PROSPER comprises industry, regulatory authority, academic, private sector and patient representatives who are interested in the area of patient-reported outcomes of adverse events (PRO-AEs). It has developed guidance on PRO-AE data, including the benefits of wider use and approaches for data capture and analysis. Patient-reported outcomes (PROs) encompass the full range of self-reporting, rather than only patient reports collected by clinicians using validated instruments. In recent years, PROs have become increasingly important across the spectrum of healthcare and life sciences. Patient-centred models of care are integrating shared decision making and PROs at the point of care; comparative effectiveness research seeks to include patients as participatory stakeholders; and industry is expanding its involvement with patients and patient groups as part of the drug development process and safety monitoring. Additionally, recent pharmacovigilance legislation from regulatory authorities in the EU and the USA calls for the inclusion of patient-reported information in benefit–risk assessment of pharmaceutical products. For patients, technological advancements have made it easier to be an active participant in one’s healthcare. Simplified internet search capabilities, electronic and personal health records, digital mobile devices, and PRO-enabled patient online communities are just a few examples of tools that allow patients to gain increased knowledge about conditions, symptoms, treatment options and side effects. Despite these changes and increased attention on the perceived value of PROs, their full potential has yet to be realised in pharmacovigilance. Current safety reporting and risk assessment processes remain heavily dependent on healthcare professionals, though there are known limitations such as under-reporting and discordant perspectives between patient reports and clinician perceptions of adverse outcomes. PROSPER seeks to support the wider use of PRO-AEs. The scope of this guidance document, which was completed between July 2011 and March 2013, considered a host of domains related to PRO-AEs, including definitions and suitable taxonomies, the range of datasets that could be used, data collection mechanisms, and suitable analytical methodologies. PROSPER offers an innovative framework to differentiate patient populations. This framework considers populations that are prespecified (such as those in clinical trials, prospective observational studies and some registries) and non-prespecified populations (such as those in claims databases, PRO-enabled online patient networks, and social websites in general). While the main focus of this guidance is on post-approval PRO-AEs from both prespecified and non-prespecified population groups, PROSPER has also considered pre-approval, prespecified populations. The ultimate aim of this guidance is to ensure that the patient ‘voice’ and perspective feed appropriately into collection of safety data. The guidance also covers a minimum core dataset for use by industry or regulators to structure PRO-AEs (accessible in the online appendix) and how data, once collected, might be evaluated to better inform on the safe and effective use of medicinal products. Structured collection of such patient data can be considered both a means to an end (improving patient safety) as well as an end in itself (expressing the patient viewpoint). The members of the PROSPER Consortium therefore direct this PRO-AE guidance to multiple stakeholders in drug safety, including industry, regulators, prescribers and patients. The use of this document across the entirety of the drug development life cycle will help to better define the benefit–risk profile of new and existing medicines. Because of the clinical relevance of ‘real-world’ data, PROs have the potential to contribute important new knowledge about the benefits and risks of medicinal products, communicated through the voice of the patient.