Identification Of Adverse Drug Events Research Articles

Validating use of diagnostic codes in Canadian administrative data for identification of adverse drug events.

While diagnostic codes from administrative health data might be a valuable source to identify adverse drug events (ADEs), their ability to identify unintended harms remains unclear. We validated claims-based diagnosis codes for ADEs based on events identified in a prospective cohort study and assessed whether key attributes predicted their documentation in administrative data. This was a retrospective analysis of 3 prospective cohorts in British Columbia, from 2008 to 2015 (n = 13 969). We linked prospectively identified ADEs to administrative insurance data to examine the sensitivity and specificity of different diagnostic code schemes. We used logistic regression to assess which key attributes (e.g., type of event, symptoms and culprit medications) were associated with better documentation of ADEs in administrative data. Among 1178 diagnosed events, the sensitivity of the diagnostic codes in administrative data ranged from 3.4 to 52.6%, depending on the database and codes used. We found that documentation was worse for certain types of ADEs (dose-related: odds ratio [OR]: 0.32, 95% confidence interval [CI]: 0.15, 0.69; nonadherence events (OR: 0.35, 95% CI: 0.20, 0.62), and better for those experiencing arrhythmias (OR: 4.19, 95% CI: 0.96, 18.28). ADEs were not well documented in administrative data. Alternative methods should be explored to capture ADEs for health research.

MtADENet: A novel interpretable method integrating multiple types of network-based inference approaches for prediction of adverse drug events

Identification of adverse drug events (ADEs) is crucial to reduce human health risks and accelerate drug safety assessment. ADEs are mainly caused by unintended interactions with primary or additional targets (off-targets). In this study, we proposed a novel interpretable method named mtADENet, which integrates multiple types of network-based inference approaches for ADE prediction. Different from phenotype-based methods, mtADENet introduced computational target profiles predicted by network-based methods to bridge the gap between chemical structures and ADEs, and hence can not only predict ADEs for drugs and novel compounds within or outside the drug-ADE association network, but also provide insights for the elucidation of molecular mechanisms of the ADEs caused by drugs. We constructed a series of network-based prediction models for 23 ADE categories. These models achieved high AUC values ranging from 0.865 to 0.942 in 10-fold cross validation. The best model further showed high performance on four external validation sets, which outperformed two previous network-based methods. To show the practical value of mtADENet, we performed case studies on developmental neurotoxicity and cardio-oncology, and over 50 % of predicted ADEs and targets for drugs and novel compounds were validated by literature. Moreover, mtADENet is freely available at our web server named NetInfer (http://lmmd.ecust.edu.cn/netinfer/). In summary, mtADENet would be a powerful tool for ADE prediction and drug safety assessment in drug discovery and development.

A holistic AI-based approach for pharmacovigilance optimization from patients behavior on social media

In this paper, we propose a holistic AI-based pharmacovigilance optimization approach using patient’s social media data. Instead of focusing on the detection and identification of Adverse Drug Events (ADE) in social media posts in single time points, we propose a holistic approach that looks at the evolution of different user behavior indicators in time. We examine various NLP-based indicators such as word frequency, semantic similarity, Adverse Drug Reactions mentions, and sentiment analysis. We introduce a classification approach to identify normal vs. abnormal time periods based on patient comments. This approach, along with user behavior indicators, can optimize the pharmacovigilance process by flagging the need for immediate attention and further investigation. We specifically focus on the Levothyrox® case in France, which sparked media attention due to changes in the medication formula and affected patient behavior on medical forums. For classification, we propose a deep learning architecture called Word Cloud Convolutional Neural Network (WC-CNN), trained on word clouds from patient comments. We evaluate different temporal resolutions and NLP pre-processing techniques, finding that monthly resolution and the proposed indicators can effectively detect new safety signals, with an accuracy of 75%. We have made the code open source, available via github.

Adverse Drug Events in Hospitalized Patients in Chile: Prevalence & Implications

To study the prevalence of adverse drug events (ADE) in hospitalized patients in Chile. As part of our research, we also assessed the validity of the method used to identify the occurrence of an ADE based on the discharge diagnoses of the patient. The study included 1,7 million patients hospitalized during 2019-2020. We analyzed the following variables for each patient: ICD-10 discharge diagnoses, sociodemographic information, and clinical outcome indicators, i.e., diagnosis-related group (DRG) weight, length of stay, and mortality. To validate the method for the identification of ADEs, first, we generated a random representative sample of patients, stratified by sex and medical specialty, hospitalized in a Chilean public hospital in 2019, and then we compared the outcome of the method with the opinion of a group of clinical experts that reviewed each patient's discharge summary retrospectively. The prevalence of ADEs in hospitalized patients in Chile during 2019 and 2020 was 2,7% and 3,1%, respectively. The precision of the method used to identify ADEs was 83,3% or higher (CI 90%). This paper uses nationwide data to describe the prevalence of ADEs and their correlation with different factors associated with the patient, the patient's disease, and the health service. These studies are essential to designing public health policies that effectively address healthcare quality and patient safety.

DrugCentral 2023 extends human clinical data and integrates veterinary drugs.

DrugCentral monitors new drug approvals and standardizes drug information. The current update contains 285 drugs (131 for human use). New additions include: (i) the integration of veterinary drugs (154 for animal use only), (ii) the addition of 66 documented off-label uses and iii) the identification of adverse drug events from pharmacovigilance data for pediatric and geriatric patients. Additional enhancements include chemical substructure searching using SMILES and 'Target Cards' based on UniProt accession codes. Statistics of interests include the following: (i) 60% of the covered drugs are on-market drugs with expired patent and exclusivity coverage, 17% are off-market, and 23% are on-market drugs with active patents and exclusivity coverage; (ii) 59% of the drugs are oral, 33% are parenteral and 18% topical, at the level of the active ingredients; (iii) only 3% of all drugs are for animal use only; however, 61% of the veterinary drugs are also approved for human use; (iv) dogs, cats and horses are by far the most represented target species for veterinary drugs;(v) the physicochemical property profile of animal drugs is very similar to that of human drugs. Use cases include azaperone, the only sedative approved for swine, and ruxolitinib, a Janus kinase inhibitor.

Monitoring User Opinions and Side Effects on COVID-19 Vaccines in the Twittersphere: Infodemiology Study of Tweets.

BackgroundIn the current phase of the COVID-19 pandemic, we are witnessing the most massive vaccine rollout in human history. Like any other drug, vaccines may cause unexpected side effects, which need to be investigated in a timely manner to minimize harm in the population. If not properly dealt with, side effects may also impact public trust in the vaccination campaigns carried out by national governments.ObjectiveMonitoring social media for the early identification of side effects, and understanding the public opinion on the vaccines are of paramount importance to ensure a successful and harmless rollout. The objective of this study was to create a web portal to monitor the opinion of social media users on COVID-19 vaccines, which can offer a tool for journalists, scientists, and users alike to visualize how the general public is reacting to the vaccination campaign.MethodsWe developed a tool to analyze the public opinion on COVID-19 vaccines from Twitter, exploiting, among other techniques, a state-of-the-art system for the identification of adverse drug events on social media; natural language processing models for sentiment analysis; statistical tools; and open-source databases to visualize the trending hashtags, news articles, and their factuality. All modules of the system are displayed through an open web portal.ResultsA set of 650,000 tweets was collected and analyzed in an ongoing process that was initiated in December 2020. The results of the analysis are made public on a web portal (updated daily), together with the processing tools and data. The data provide insights on public opinion about the vaccines and its change over time. For example, users show a high tendency to only share news from reliable sources when discussing COVID-19 vaccines (98% of the shared URLs). The general sentiment of Twitter users toward the vaccines is negative/neutral; however, the system is able to record fluctuations in the attitude toward specific vaccines in correspondence with specific events (eg, news about new outbreaks). The data also show how news coverage had a high impact on the set of discussed topics. To further investigate this point, we performed a more in-depth analysis of the data regarding the AstraZeneca vaccine. We observed how media coverage of blood clot–related side effects suddenly shifted the topic of public discussions regarding both the AstraZeneca and other vaccines. This became particularly evident when visualizing the most frequently discussed symptoms for the vaccines and comparing them month by month.ConclusionsWe present a tool connected with a web portal to monitor and display some key aspects of the public’s reaction to COVID-19 vaccines. The system also provides an overview of the opinions of the Twittersphere through graphic representations, offering a tool for the extraction of suspected adverse events from tweets with a deep learning model.

ADENet: a novel network-based inference method for prediction of drug adverse events.

Identification of adverse drug events (ADEs) is crucial to reduce human health risks and improve drug safety assessment. With an increasing number of biological and medical data, computational methods such as network-based methods were proposed for ADE prediction with high efficiency and low cost. However, previous network-based methods rely on the topological information of known drug-ADE networks, and hence cannot make predictions for novel compounds without any known ADE. In this study, we introduced chemical substructures to bridge the gap between the drug-ADE network and novel compounds, and developed a novel network-based method named ADENet, which can predict potential ADEs for not only drugs within the drug-ADE network, but also novel compounds outside the network. To show the performance of ADENet, we collected drug-ADE associations from a comprehensive database named MetaADEDB and constructed a series of network-based prediction models. These models obtained high area under the receiver operating characteristic curve values ranging from 0.871 to 0.947 in 10-fold cross-validation. The best model further showed high performance in external validation, which outperformed a previous network-based and a recent deep learning-based method. Using several approved drugs as case studies, we found that 32-54% of the predicted ADEs can be validated by the literature, indicating the practical value of ADENet. Moreover, ADENet is freely available at our web server named NetInfer (http://lmmd.ecust.edu.cn/netinfer). In summary, our method would provide a promising tool for ADE prediction and drug safety assessment in drug discovery and development.

Development and Adjustment of an Algorithm for Identifying Drug-Related Hospital Admissions in Pediatrics.

Adverse drug events (ADEs) in the outpatient pediatric pharmacotherapy can be serious and lead to inpatient admissions. Recent research only focused on ADE identification during hospitalization. The aim of the present study was to develop an algorithm to identify drug-related hospital admissions in pediatrics. A systematic literature research was performed, and a pediatric trigger tool for identifying drug-related inpatient admissions was built. The initial version was tested in a sample of 292 patients admitted to a German university children's hospital. Subsequently, the tool was further improved by combining different modules as a novel approach. The obtained algorithm with 39 triggers in 5 modules identified drug-related inpatient admissions at a sensitivity of 95.5% (95% confidence interval [CI], 89.3%-100%) and a specificity of 16.5% (95% CI, 11.9%-21.2%), respectively. After modifications including trigger activation requiring a combination of different modules, specificity increased to 56.9% (95% CI, 50.7%-63.0%). Identifying 36 of 44 ADEs leading to admission, sensitivity remained high (81.8% [95% CI, 70.4%-93.2%]). The overall positive predictive value was 25.2% (95% CI, 18.1%-32.3%). The algorithm is the first trigger tool to identify ambulant acquired ADEs leading to hospital admission in pediatrics. However, the underlying patient sample is small.Using a larger population for refinement will allow further specifications and reduction in the total amount of triggers and thus signals.

Identifying adverse drug events in patients at a pediatric ward in a Brazilian hospital: application and performance of the triggers

Abstract Objectives: to evaluate the performance of a trigger tool in identifying adverse drug events (ADE) in hospitalized children. Methods: a retrospective cohort study review on 133 medical records at a federal maternal and child reference hospital in Rio de Janeiro in 2016. A list of 14 triggers was developed to detect ADE in the pediatric population. Three steps were performed: (1) search for triggers; (2) selection of suspected cases of ADE and (3) final determination of ADE by experts’ consensus. Results: 360 triggers were identified in 100 hospitalizations (75.2%), with an average of 2.7 triggers/ hospitalization. The most frequent triggers were “abrupt medication stop” (79.7%); “antiemetics use” (8.9%) and “laxatives use” (7.2%); while the “diphenhydramine use”, “phytomenadione use” and “excessive sedation/lethargy/fall/hypotension” obtained the highest performance indicating ADE every time they occurred. Thirty-one ADE were identified in 12.8% of the hospitalizations; 11 (35.5%) ADE were detected without the aid of the triggers thus, pruritus and diarrhea were the most frequent. Conclusion: the trigger tool proved to be useful in identifying ADE in hospitalized children, especially if high performance and high frequency triggers are used in identifying the events. The inclusion of the triggers “diarrhea” and “pruritus”, may favor the identification of ADE in patients at pediatric wards.

Evaluation of Performance of IHI Global Trigger Tool in Identification of Adverse Drug Events: A Prospective Observational Study

Aims: The study aims to evaluate the performance of the Institute for Healthcare Improvement (IHI) global trigger tool in the identification of adverse drug events.
 Study design: Prospective observational study.
 Place and duration of study: The study was conducted in a General Medicine department of a secondary care referral hospital located in rural, resource-limited settings of Bathalapalli, Anantapur district, Andhra Pradesh, India. The study was conducted for a period of six months, from June 2019 to November 2019.
 Methodology: A pre-designed data collection form was used to collect the data from the study participants. The required data was obtained from the patient case sheet, lab reports, treatment charts, daily nursing notes, daily physician notes, and direct patient interviews. The global trigger tool developed by IHI was used for the rapid review of inpatient medical records and to generate clues for the identification of ADEs. Descriptive statistics were used to represent the findings of demographics, clinical characteristics, ADE profiles, IHI triggers, and clinical alterations. ADE incidence was shown in a measure of ADE per 1000 patient days. IHI global trigger tool performance in detecting ADE was measured by sensitivity, specificity, positive predictive value, and negative predictive value.
 Results: A total of 192 patients were enrolled in the study. Among them, 225 triggers and 123 ADEs were detected. The incidence of ADEs in the inpatients estimated by the IHI method was estimated as; 20.2 ADEs per 1000 patient days,64.0ADEs per 100 admissions, 56.2percent of admissions with ADE. Majority of ADEs are shown possible relationship with drug (60; 48.7%), level-2 severity (49; 39.8%), and not preventable (52; 42.2%). Most IHI global triggers showed high sensitivity, specificity, positive predictive value, and negative predictive value for detecting ADEs.
 Conclusion: The study shows that most of the triggers in the IHI Trigger tool have shown good accuracy in identifying ADEs. Thus, using the IHI Trigger Tool to identify ADEs can help to improve patient safety. Therefore, the study recommends incorporating IHI global trigger tool in routine, conventional ADE screening techniques to improve the detection rate and promote drug safety.

Accuracy of triggers in the identification of adverse drug events in hospitalized elderly

Objective: Evaluate trigger accuracy for identifying adverse drug events (ADEs) in hospitalised elderly. Methods: Two hundred patients 60 years old from a medical clinic within a private hospital were followed-up. For ADE identification the adapted Global Trigger Tool tracker methodology was used. Causality was determined using the Naranjo Algorithm. Results: Of the 200 elderly patients included in the study, 106 were females (53%), the average age was 79 years, and the average length of hospital stay was approximately ± 10 days. Selected triggers were identified 1,457 times. The group of triggers with the best performance regarding its analysis accuracy was evolution triggers, with sensitivity of 69% and positive predictive value of 68%. In the individual performance analysis, the evolution tracker allergy, allergic reaction, pruritus achieved 100% performance for both sensitivity and positive predictive value. A total of 165 ADs were identified. Of these, 18% were phlebitis and 16% were hypoglycaemia. Drugs associated with ADE included insulin (15%) and Clarithromycin (9%). Conclusion: The triggering methodology has been effective for identifying ADEs. In addition, determination best trigger for constructing an ADE identification tool for hospitalised elderly was performed.

Use of ICD-10-CM T codes in hospital claims data to identify adverse drug events in Taiwan.

Adverse drug events (ADEs) are a major public health concern worldwide and may prolong hospital stays, causing a burden on the healthcare system and increasing the associated costs. Therefore, optimizing medication use and reducing ADEs are priorities for public health. Medication safety can be monitored and improved by identifying ADEs. The utilization of diagnoses coded according to the International Statistical Classification of Diseases and Related Health Problems (ICD) system for the identification of ADEs has been firmly established. In Taiwan, however, the validity of recording ADEs on the basis of inpatient ICD-10-CM T codes has not been evaluated. Therefore, this study investigated the potential usefulness of ICD-10-CM T codes in routine hospital data for the identification of ADEs and for increasing the rate of reporting. We use hospital claims data of hospitalized patients from one medical centre in northern Taiwan between 1 July 2016 and 30 June 2018. We defined an ADE to have taken place if an ICD-10-CM T code was present among the primary or secondary diagnosis codes. The inpatients who were discharged with T codes in a primary or secondary diagnosis were identified by the computerized T code information platform, and the retrospective review of the medical charts was performed by pharmacists to confirm the ADEs. 1384 inpatients who were discharged with the relevant T codes in a primary or secondary diagnosis were identified during the study period. Code T36 (poisoning by, adverse effect of or underdosing of systemic antibiotics) was the most common code, accounting for 56.6%, followed by T42 (17.7%; poisoning by, adverse effect of or underdosing of antiepileptic, sedative-hypnotic or antiparkinsonism drug). Overall, 789 clinically significant ADEs were identified after medical chart review. The dermatologic system was the most commonly involved. The overall positive predictive value for a flagged code representing an ADE was 57%. Furthermore, the use of T codes to confirm the number of ADE cases increased the ADE reporting rate by 9.17%. The PPV of ICD-10-CM T codes analysed in our study was insufficient for identifying ADEs during hospitalization. The sensitivity and specificity of this were inadequate. However, the T code system can be used as an auxiliary resource for pharmacists to identify potential ADEs and report the information as prompts on the physician order entry system. When a physician prescribes a drug that may cause an ADE in a patient, an alert is issued to ensure medication safety. In conclusion, the T codes did not perform well in our study and caution should be exercised in their use to identify ADEs on their own.

ADE Eval: An Evaluation of Text Processing Systems for Adverse Event Extraction from Drug Labels for Pharmacovigilance.

IntroductionThe US FDA is interested in a tool that would enable pharmacovigilance safety evaluators to automate the identification of adverse drug events (ADEs) mentioned in FDA prescribing information. The MITRE Corporation (MITRE) and the FDA organized a shared task—Adverse Drug Event Evaluation (ADE Eval)—to determine whether the performance of algorithms currently used for natural language processing (NLP) might be good enough for real-world use.ObjectiveADE Eval was conducted to evaluate a range of NLP techniques for identifying ADEs mentioned in publicly available FDA-approved drug labels (package inserts). It was designed specifically to reflect pharmacovigilance practices within the FDA and model possible pharmacovigilance use cases.MethodsPharmacovigilance-specific annotation guidelines and annotated corpora were created. Two metrics modeled the experiences of FDA safety evaluators: one measured the ability of an algorithm to identify correct Medical Dictionary for Regulatory Activities (MedDRA®) terms for the text from the annotated corpora, and the other assessed the quality of evidence extracted from the corpora to support the selected MedDRA® term by measuring the portion of annotated text an algorithm correctly identified. A third metric assessed the cost of correcting system output for subsequent training (averaged, weighted F1-measure for mention finding).ResultsIn total, 13 teams submitted 23 runs: the top MedDRA® coding F1-measure was 0.79, the top quality score was 0.96, and the top mention-finding F1-measure was 0.89.ConclusionWhile NLP techniques do not perform at levels that would allow them to be used without intervention, it is now worthwhile exploring making NLP outputs available in human pharmacovigilance workflows.Electronic supplementary materialThe online version of this article (10.1007/s40264-020-00996-3) contains supplementary material, which is available to authorized users.

Validating the Chinese geriatric trigger tool and analyzing adverse drug event associated risk factors in elderly Chinese patients: A retrospective review.

The aim was to evaluate the performance of the initial Chinese geriatric trigger tool to detect adverse drug events (ADEs) in Chinese older patients, to attempt to shorten this list for improving the efficiency of the trigger tool, and to study the incidence and characteristics of ADEs in this population. A sample of 25 cases was randomly selected per half a month from eligible patients who aged 60 years and older, hospitalized more than 24 hours, and discharged or died between January 1, 2015 and December 31, 2017 in West China hospital. A two-stage retrospective chart review of the included inpatients were conducted. ADEs were detected using a list of 42 triggers previously selected by an expert panel by means of a Delphi method. The number of triggers identified and ADEs detected were recorded and the positive predictive value (PPV) of each trigger was calculated to select the most efficient triggers. Several variables were recorded, including age, sex, number of diseases, length of hospital stay and so on, to analyze the risk factor of ADEs. Among 1800 patients, 1646 positive triggers and 296 ADEs were detected in 234 (13.00%) patients. Older patients who were younger, had more medications, longer stays or more admission, and did not experience surgical operation more likely experienced ADEs. Triggers with PPV less than 5% were eliminated, which resulted in the upgraded version of Chinese geriatric trigger tool of 20 triggers with a PPV of 28.50%. This upgraded tool accounted for 99.66% of all ADEs detected. The upgraded version of Chinese geriatric trigger tool was an efficient tool for identifying ADEs in Chinese older patients. Future, the trigger tool could be incorporated into routine screen systems to provide real-time identification of ADEs, thereby enabling timely clinical interventions.

Specific Toxicity of Maintenance Olaparib Versus Placebo in Advanced Malignancies: A Systematic Review and Meta-analysis.

We performed a systematic review and meta-analysis to investigate the safety of maintenance with olaparib after platinum-based chemotherapy in cancer patients. Eligible studies included randomized controlled trials (RCTs) regarding the clinical role of olaparib maintenance therapy versus placebo in BRCA-mutated, advanced cancers. Safety profile from each selected study was investigated for all-grade and G3-G4 haematological and non-haematological adverse drug events (ADEs). Four RTCs that involved 1099 patients were included in the analysis. Overall incidences of all-grade and G3-4 ADEs in olaparib group were 97.6% and 41%, respectively. Patients treated with maintenance olaparib showed higher risk of all-grade and G3-G4 anaemia, all-grade neutropenia and thrombocytopenia. Moreover, all-grade and G3-G4 fatigue, all-grade vomiting, diarrhoea, nausea and decreased appetite were more common in the olaparib group compared to placebo. Despite an increased risk and incidence of several haematological and non-haematological toxicities, olaparib is a relatively safe agent for the treatment of advanced solid tumors. Prompt identification of ADEs is mandatory to avoid therapy discontinuation and optimize treatment.

Efficiency of an interactive program for enhancing patient reporting of adverse drug events in primary care

Introduction Adverse drug events (ADE) are frequents in primary care. They could often be prevented and their complications avoided. One of the main causes is that patients rarely know how to detect ADE. Another cause is inadequate communication between patients and physicians. If patients were effectively trained in identifying and reporting ADE, this should enhance ADE management (and avoid consequences on patient's health). The aim of the InPAct study was to assess an interactive program promoting patients to be active in ADE identification and their reporting to their general practitioners (GP), in order to enhance ADE management in primary care. We focused on patients treated by anti-hypertensive drugs, mainly prescribed in primary care and known to be often responsible for ADE. Methods/design We conducted a cluster randomized controlled stepped wedge trial, with eight clusters of GP. The InPAct program was implemented in the clusters in random order along five successive three-month periods. The program featured: – a patient booklet including information on cardiovascular risks and management of antihypertensive drugs, care plans and patient ADE report forms; – standardized training of physicians in how to present the booklet to the patient. The primary outcome was the active reporting of ADE by patients to their GP within three months of inclusion, after the identification of the ADE by the patients themselves. Total ADE identification (i.e. ADE identified by physician or patient, independently of patient active reporting to their practitioner), knowledge of patient about cardiovascular risks and management of antihypertensive treatments and patient satisfaction about their care management by their general practitioner were evaluated as secondary outcomes. The effect of the intervention was quantified and tested using a mixed logistic model to integrate cluster and time effects. Results We analyzed data for 1095 patients included by 60 different GP belonging to the 8 clusters. Multivariate analyses showed that InPAct program efficiently enhanced active reporting of ADE by patients to their GP (OR = 3.5, 95 CI [1.2–10.1], P = 0.02), but also total ADE identification (OR = 2.6, 95 CI [1.3–5.1], P = 0.01). The anti-hypertensive prescription status (initiation, modification or renewal) was strongly associated with patient active ADE reporting and total ADE identification. Moreover, patient satisfaction about their care management by their GP was higher in the intervention group, as the communication or the given information by the practitioner were concerned. Conclusion The InPAct intervention constituted an original program that was intended to improve ADE management in primary care by enhancing patient involvement in their own care. Finally, the InPAct intervention was efficient not only to enhance patient ADE identification associated with an active reporting, but also total ADE identification and communication between patients and their practitioners. Our study suggests that a better communication between patients and GP could improve patient safety in primary care.

Pharmacovigilance from social media: An improved random subspace method for identifying adverse drug events

ObjectiveRecent advances in Web 2.0 technologies have seen significant strides towards utilizing patient-generated content for pharmacovigilance. Social media-based pharmacovigilance has great potential to augment current efforts and provide regulatory authorities with valuable decision aids. Among various pharmacovigilance activities, identifying adverse drug events (ADEs) is very important for patient safety. However, in health-related discussion forums, ADEs may confound with drug indications and beneficial effects, etc. Therefore, the focus of this study is to develop a strategy to identify ADEs from other semantic types, and meanwhile to determine the drug that an ADE is associated with. Materials and methodsIn this study, two groups of features, i.e., shallow linguistic features and semantic features, are explored. Moreover, motivated and inspired by the characteristics of explored two feature categories for social media-based ADE identification, an improved random subspace method, called Stratified Sampling-based Random Subspace (SSRS), is proposed. Unlike conventional random subspace method that applies random sampling for subspace selection, SSRS adopts stratified sampling-based subspace selection strategy. ResultsA case study on heart disease discussion forums is performed to evaluate the effectiveness of the SSRS method. Experimental results reveal that the proposed SSRS method significantly outperforms other compared ensemble methods and existing approaches for ADE identification. Discussion and conclusionOur proposed method is easy to implement since it is based on two feature sets that can be naturally derived, and therefore, can omit artificial stratum generation efforts. Moreover, SSRS has great potential of being applied to deal with other high-dimensional problems that can represent original data from two different aspects.

Adverse drug events in German hospital routine data: A validation of International Classification of Diseases, 10th revision (ICD-10) diagnostic codes.

ObjectiveAdverse drug events (ADEs) during hospital stays are a significant problem of healthcare systems. Established monitoring systems lack completeness or are cost intensive. Routinely assigned International Statistical Classification of Diseases and Related Health Problems (ICD) codes could complement existing systems for ADE identification. To analyze the potential of using routine data for ADE detection, the validity of a set of ICD codes was determined focusing on hospital-acquired events.Material and methodsThe study utilized routine data from four German hospitals covering the years 2014 and 2015. A set of ICD, 10th Revision, German Modification (ICD-10-GM) diagnoses coded most frequently in the routine data and identified as codes indicating ADEs was analyzed. Data from psychiatric and psychotherapeutic departments were excluded. Retrospective chart review was performed to calculate positive predictive values (PPV) and sensitivity.ResultsOf 807 reviewed ADE codes, 91.2% (95%-confidence interval: 89.0, 93.1) were identified as disease in the medical records and 65.1% (61.7, 68.3) were confirmed as ADE. For code groups being predominantly hospital-acquired, 78.5% (73.7, 82.9) were confirmed as ADE, ranging from 68.5% to 94.4% dependent on the ICD code. However, sensitivity of inpatient ADEs was relatively low. 49.7% (45.2, 54.2) of 495 identified hospital-acquired ADEs were coded as disease in the routine data, from which a subgroup of 12.1% (9.4, 15.3) was coded as drug-associated disease.ConclusionsICD codes from routine data can provide an important contribution to the development and improvement of ADE monitoring systems. Documentation quality is crucial to further increase the PPV, and actions against under-reporting of ADEs in routine data need to be taken.

Comparing Methods for Estimating Direct Costs of Adverse Drug Events

ObjectivesTo estimate how direct health care costs resulting from adverse drug events (ADEs) and cost distribution are affected by methodological decisions regarding identification of ADEs, assigning relevant resource use to ADEs, and estimating costs for the assigned resources. MethodsADEs were identified from medical records and diagnostic codes for a random sample of 4970 Swedish adults during a 3-month study period in 2008 and were assessed for causality. Results were compared for five cost evaluation methods, including different methods for identifying ADEs, assigning resource use to ADEs, and for estimating costs for the assigned resources (resource use method, proportion of registered cost method, unit cost method, diagnostic code method, and main diagnosis method). Different levels of causality for ADEs and ADEs’ contribution to health care resource use were considered. ResultsUsing the five methods, the maximum estimated overall direct health care costs resulting from ADEs ranged from Sk10,000 (Sk = Swedish krona; ~€1,500 in 2016 values) using the diagnostic code method to more than Sk3,000,000 (~€414,000) using the unit cost method in our study population. The most conservative definitions for ADEs’ contribution to health care resource use and the causality of ADEs resulted in average costs per patient ranging from Sk0 using the diagnostic code method to Sk4066 (~€500) using the unit cost method. ConclusionsThe estimated costs resulting from ADEs varied considerably depending on the methodological choices. The results indicate that costs for ADEs need to be identified through medical record review and by using detailed unit cost data.

Leveraging graph topology and semantic context for pharmacovigilance through twitter-streams.

BackgroundAdverse drug events (ADEs) constitute one of the leading causes of post-therapeutic death and their identification constitutes an important challenge of modern precision medicine. Unfortunately, the onset and effects of ADEs are often underreported complicating timely intervention. At over 500 million posts per day, Twitter is a commonly used social media platform. The ubiquity of day-to-day personal information exchange on Twitter makes it a promising target for data mining for ADE identification and intervention. Three technical challenges are central to this problem: (1) identification of salient medical keywords in (noisy) tweets, (2) mapping drug-effect relationships, and (3) classification of such relationships as adverse or non-adverse.MethodsWe use a bipartite graph-theoretic representation called a drug-effect graph (DEG) for modeling drug and side effect relationships by representing the drugs and side effects as vertices. We construct individual DEGs on two data sources. The first DEG is constructed from the drug-effect relationships found in FDA package inserts as recorded in the SIDER database. The second DEG is constructed by mining the history of Twitter users. We use dictionary-based information extraction to identify medically-relevant concepts in tweets. Drugs, along with co-occurring symptoms are connected with edges weighted by temporal distance and frequency. Finally, information from the SIDER DEG is integrate with the Twitter DEG and edges are classified as either adverse or non-adverse using supervised machine learning.ResultsWe examine both graph-theoretic and semantic features for the classification task. The proposed approach can identify adverse drug effects with high accuracy with precision exceeding 85 % and F1 exceeding 81 %. When compared with leading methods at the state-of-the-art, which employ un-enriched graph-theoretic analysis alone, our method leads to improvements ranging between 5 and 8 % in terms of the aforementioned measures. Additionally, we employ our method to discover several ADEs which, though present in medical literature and Twitter-streams, are not represented in the SIDER databases.ConclusionsWe present a DEG integration model as a powerful formalism for the analysis of drug-effect relationships that is general enough to accommodate diverse data sources, yet rigorous enough to provide a strong mechanism for ADE identification.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-016-1220-5) contains supplementary material, which is available to authorized users.