ABSTRACT
Aim
Validating an operational algorithm for identifying ventricular arrhythmia and sudden cardiac arrest (VA/SCA) in electronic health record (EHR) data may be useful to minimize measurement bias in studies characterizing real-world VA/SCA risk; however, validation studies require an appropriate reference standard. We aimed to assess if adequate information is documented in unstructured clinical notes of a large EHR database to serve as a reference standard for future validation studies of VA/SCA.
Methods
Twenty potential VA/SCA events were randomly selected from unstructured clinical notes of a large EHR database, TriNetX Dataworks – USA. These notes were reviewed to assess if key clinical elements were documented to confirm the occurrence of VA/SCA and describe their clinical features. These included explicit documentation of an acute event, electrocardiogram (ECG) findings, urgent medical interventions, and other elements.
Results
Explicit documentation of an acute event was recorded for 17 patients (85.0%) and ECG findings were documented for 15 patients (75.0%). Generally, unstructured clinical notes also contained information about signs and symptoms, care setting, medical interventions administered, and event resolution.
Conclusions
The unstructured clinical notes of a large EHR database contained the information necessary to serve as a reference standard for validation studies of a VA/SCA operational algorithm in EHR data.
KEYWORDS: Ventricular arrhythmia, sudden cardiac arrest, electronic health records, unstructured data, validation
Plain Language Summary
Ventricular arrhythmia and sudden cardiac arrest (VA/SCA) accounts for up to 50% of all cardiovascular disease-related deaths and is a potential adverse event of certain medications, making it an important public health and drug safety concern. Studies using large electronic databases such as electronic health records (EHR) data are useful for studying VA/SCA because of their large sample size; however, the ability of EHR databases to accurately identify VA/SCA occurrence is unclear. To assess the accuracy of an operational algorithm (i.e. diagnosis, medication, and other codes) to identify VA/SCA in an EHR database, a validation study must be conducted in which the performance of the algorithm is compared to a reference standard. We aimed to assess if adequate information to serve as a reference standard for future validation studies of VA/SCA is documented in clinical notes written by physicians, nurses, and other healthcare providers contributing to an EHR data network of healthcare organizations with inpatient and outpatient care. We randomly selected 20 potential VA/SCA events from the clinical notes of this EHR data source and reviewed the notes to assess if key clinical elements were documented to confirm the occurrence and describe clinical features of VA/SCA. We found that the clinical notes provided adequate information to confirm VA/SCA event occurrence and had detailed information on other features such as signs and symptoms, care setting, and medical interventions. Future validation studies of a VA/SCA operational algorithm in EHR data may be able to use the clinical notes as a reference standard.
1. Introduction
Sudden cardiac arrest is the most common cause of death in the United States (US) and accounts for up to 50% of all cardiovascular disease-related deaths [1,2]. Given that ventricular fibrillation has been found to be present in 75% of sudden cardiac arrest (SCA) cases [3], ventricular arrhythmias (VA) remain an important risk factor for sudden cardiac arrest and sudden death. Ventricular arrhythmia and sudden cardiac arrest (VA/SCA) is an important public health topic with various causes including coronary artery disease and electrical cardiac diseases, genetic factors, electrolyte imbalances, and medications [1]. Medication-induced VA/SCA is a notable drug safety concern due to numerous medications’ observed effect of prolonging the electrocardiographic QTc interval [4]. Medication-induced QTc prolongation is associated with a ventricular arrhythmia called Torsades de Pointes which can deteriorate into ventricular fibrillation and cause sudden cardiac arrest and death. Over 150 medications have been found to be associated with a risk of QTc prolongation and/or Torsades de Pointes [5].
Given the rare occurrence of VA/SCA (60 per 100,000 population) [6], randomized clinical trials may lack a sample size large enough to characterize the risk of VA/SCA associated with a specific drug, warranting the conduct of large safety studies in the postmarketing setting. Such studies are also helpful in characterizing the VA/SCA risk in a real-world patient population unrestricted by clinical trial eligibility criteria (e.g., patients with preexisting VA/SCA risk factors may be excluded from participating in clinical trials but can still be treated in clinical practice). Large, electronic healthcare databases such as electronic health records (EHR) and health insurance claims are commonly used to support postmarketing studies in order to assess the real-world safety of medications, especially of those that recently received regulatory approval [7,8]. For example, EHR databases have been used to investigate real-world safety outcomes in patients treated with COVID-19 vaccines [9]. Furthermore, electronic healthcare databases can be leveraged to characterize the epidemiology of VA/SCA in patients with other etiologies, such as cardiac diseases or genetic factors.
Though electronic healthcare databases contain diagnosis, medication, and procedure codes needed to characterize VA/SCA in real-world patient populations, the validity of such codes to accurately identify and describe VA/SCA events may vary depending on the data source. Validation studies, in which the validity of an operational algorithm for identifying an event of interest is estimated against a chosen reference standard [10], are useful to ensure that misclassification in a planned postmarketing study is minimized.
While an operational algorithm for identifying VA/SCA has been validated previously in Medicaid health insurance claims [11], it is unclear if such an algorithm performs similarly in a structured EHR data-derived database. EHR databases provide a good opportunity for algorithm validation given the potential availability of unstructured clinical notes, a common reference standard for validating algorithms in large electronic healthcare databases [10]. However, the ability to serve as a reference standard for a given operational algorithm depends on the information documented by healthcare providers in clinical notes during routine medical practice.
In order to better understand if unstructured clinical notes can serve as a reference standard for validating an algorithm that identifies VA/SCA events in structured EHR data, we reviewed the information documented in the unstructured clinical notes of a large, EHR-derived database. Specifically, we aimed to assess if adequate information is available in the clinical notes to confirm the occurrence and describe the severity, interventions administered, and other features of acute VA/SCA events.
2. Materials and methods
2.1. Study overview and data source
We conducted a descriptive pilot analysis of potential VA/SCA events identified in the unstructured clinical notes of a large EHR-derived database in order to assess the availability of key clinical elements necessary for confirming the occurrence of and describing acute VA/SCA events. We used the repository of unstructured clinical notes available in the TriNetX Dataworks – USA network, a de-identified, longitudinal EHR-derived dataset that includes outpatient and inpatient EHRs for over 120 million patients from 70 healthcare organizations across the US. Network members include academic medical centers, integrated delivery networks, specialty hospitals, and large specialty physician practices. Sixteen of the 70 healthcare organizations (22.9%) comprising the network readily provide unstructured clinical notes written by physicians, nurses, and other healthcare professionals. Clinical notes include a variety of types (e.g., admission and discharge notes, progress notes, specialty consult notes, laboratory results, radiology and biopsy results) from both the outpatient and inpatient settings.
2.2. Selection of and review of potential events
We used keyword search in the repository of unstructured clinical notes to identify patients experiencing a potential VA/SCA event between the years 2010 and 2023 (structured data elements such as International Classification of Diseases, Tenth Revision, Clinical Modification [ICD-10-CM] diagnosis codes were not used to identify potential events). Keywords included ‘ventricular arrhythmia,’ ‘ventricular tachycardia,’ ‘ventricular fibrillation,’ ‘Torsades de pointes,’ ‘sudden cardiac arrest,’ and ‘cardiac arrest.’ We then randomly selected 20 potential events (in 20 unique patients) and reviewed the clinical notes to identify the availability of key clinical elements necessary to confirm their occurrence and to describe clinical features of the selected potential events. The sample size of 20 potential events was chosen in order to optimize the collection of relevant details from the clinical notes while being logistically feasible with the resources available (i.e., the effort required in manually reviewing the clinical notes).
We included clinical elements that were derived from those used to validate a previously-published algorithm for identifying VA/SCA events in Medicaid health insurance claims from 5 US states [8], as well as additional elements deemed important by the authors. Clinical elements necessary to confirm event occurrence included explicit documentation of an acute VA or SCA event by the healthcare provider, documentation of the location of event onset, documented findings from electrocardiogram (ECG) reports, telemetry readings, or electrophysiology reports, and documentation of signs and symptoms consistent with VA or SCA presentation.
Clinical elements used to describe the features and discern severity of the event included documented risk factors for or the suspected cause of the VA or SCA event, presence of a preexisting pacemaker or implantable cardioverter-defibrillator (ICD), administration of urgent medical interventions, administration of electrolyte supplementation, setting of care for the VA/SCA event, and resolution of event. Urgent medical interventions included performance of cardiopulmonary resuscitation, performance of defibrillation, insertion of a pacemaker or ICD, and initiation of an antiarrhythmic medication.
2.3. Statistical analysis
For each clinical element, we calculated the frequency and proportion of potential events that had documentation of the clinical element in the clinical notes. For clinical elements used to describe the potential event, we calculated the frequency and proportion for the specific categories documented in the clinical notes (e.g., for administration of an urgent medical intervention, the frequency and proportion of potential events reporting each type of medical intervention was calculated). This pilot study was approved by the BRANY institutional review board (BRANY study number 24-08-484-1821).
3. Results
3.1. Clinical elements to confirm event occurrence
Twenty patients with a potential VA/SCA event from 11 healthcare organizations were included in this analysis. Explicit documentation of acute VA or SCA by a healthcare provider was recorded for 17 patients (85.0% of patients with a potential event) (Table 1). The remaining three patients without explicit documentation of an acute event in the clinical notes had the mention of a past medical history of VA or SCA or were suspected to have VA but were ultimately diagnosed with atrial fibrillation. The location of the onset of the event was documented for 17 patients (85.0%), with the majority initially occurring outside of the healthcare setting (i.e., in the community). Findings from ECG reports, telemetry readings, or electrophysiology reports were documented for 15 patients (75.0%). ECG findings were typically more comprehensive for patients with a known cardiac medical history (e.g., patients who already had a pacemaker/ICD) or for events that initially occurred in the hospital. Signs and symptoms of VA/SCA were documented in 11 patients (55.0%), with syncope or unresponsiveness being the most frequently documented (10 patients).
Table 1.
Information documented in unstructured clinical notes of an electronic health record database for potential VA/SCA events: clinical elements to confirm event occurrence.
| Element derived from unstructured clinical notes, n (%) | Potential VA/SCA events N = 20 |
|---|---|
| Explicit documentation of acute VA or SCA by healthcare provider | 17 (85.0%) |
| Documentation of location of event onset | 17 (85.0%) |
| Outside of hospital | 10 (58.8%)* |
| During hospital admission | 7 (41.2%)* |
| Documented findings on ECG reports, telemetry readings, or electrophysiology reports | 15 (75.0%) |
| Documentation of signs and symptoms of VA or SCA | 11 (55.0%) |
| Syncope or unresponsiveness | 10 (90.9%)† |
| Palpitations | 1 (9.1%)† |
| Dizziness or lightheadedness | 2 (18.2%)† |
| Shortness of breath | 3 (27.3%)† |
| Nausea | 1 (9.1%)† |
SCA: sudden cardiac arrest; VA: ventricular arrhythmia.
*Proportion calculated with the denominator as the number of patients who had a documented location of event onset.
†Proportion calculated with the denominator as the number of patients who had documented signs and symptoms of VA or SCA.
3.2. Clinical elements to describe event
A suspected cause or relevant risk factor of the VA/SCA event was documented for 9 patients (45.0% of patients with a potential event). The most commonly documented suspected cause or risk factor was significant cardiac medical history (4 patients; 44.4% of patients with a documented suspected cause or risk factor), followed by acute severe illness (3 patients; 33.3% of patients with a documented suspected cause or risk factor), acute use of a medication with Torsades de Pointes risk (3 patients; 33.3% of patients with a documented suspected cause or risk factor), and chronic use of a medication with Torsade de Pointes risk (1 patient; 11.1% of patients with a documented suspected cause or risk factor). A non-physiologic extrinsic cause (e.g., motor vehicle accident) was not documented for any patient. An urgent medical intervention was administered in 16 patients (80.0% of patients with a potential event). Defibrillation (12 patients; 75.0% of patients with an urgent medical intervention) and initiation of an antiarrhythmic medication (12 patients; 75.0% of patients with an urgent medical intervention) were the most common, followed by cardiopulmonary resuscitation (8 patients; 50.0% of patients with an urgent medical intervention), and insertion of a pacemaker/ICD (2 patients; 12.5% of patients with an urgent medical intervention). Electrolyte supplementation was administered in 16 patients (80.0% of patients with a potential event) (Table 2).
Table 2.
Information documented in unstructured clinical notes of an electronic health record database for potential VA/SCA events: clinical elements to describe event.
| Element derived from unstructured clinical notes, n (%) | Potential VA/SCA events N = 20 |
|---|---|
| Documented suspected cause or risk factor of VA and SCA | 9 (45.0%) |
| Chronic | |
| Cardiac medical history (e.g., long QT syndrome, cardiomyopathy, heart failure) | 4 (44.4%)‡ |
| Chronic use of medications with Torsades de Pointes risk (e.g., loperamide) | 1 (11.1%)‡ |
| Acute | |
| Acute severe illness (e.g., septic shock, acute renal failure, severe hypokalemia) | 3 (33.3%)‡ |
| Acute use of medications with Torsades de Pointes risk (e.g., levofloxacin) | 3 (33.3%)‡ |
| Non-physiologic extrinsic cause of event (e.g., motor vehicle accident, blunt trauma) | 0 (0.0%)‡ |
| Pre-existing pacemaker/ICD | 4 (20.0%) |
| Urgent medical intervention administered | 16 (80.0%) |
| Cardiopulmonary resuscitation | 8 (50.0%)§ |
| Defibrillation | 12 (75.0%)§ |
| Pacemaker/ICD insertion | 2 (12.5%)§ |
| Antiarrhythmic medication | 12 (75.0%)§ |
| Electrolyte supplementation administered (e.g., magnesium, potassium) | 16 (80.0%) |
| Healthcare setting | |
| Outpatient | 3 (15.0%) |
| Emergency department (without inpatient admission) | 0 (0.0%) |
| Inpatient (without escalation to intensive care) | 7 (35.0%) |
| Intensive care unit | 10 (50.0%) |
| Resolution of event | |
| Patient expired | 2 (10.0%) |
| Patient discharged from hospital or healthcare facility | 18 (90.0%) |
ECG: electrocardiogram; SCA: sudden cardiac arrest; VA: ventricular arrhythmia.
‡Proportion calculated with the denominator as the number of patients who had a documented suspected cause or risk factor of VA and SCA.
§Proportion calculated with the denominator as the number of patients who had an urgent medical intervention administered.
Ten patients (50.0%) were treated in the intensive care/critical care unit, 7 patients (35.0%) were treated in the inpatient setting without escalation to intensive care/critical care, and 3 patients (15.0%) were treated in the outpatient setting. No patients were treated in the emergency department without admission to the inpatient setting. Two patients (10.0%) died during care for the event and 18 patients (90.0%) were discharged from the healthcare facility (Table 2). The two patients that died were also being managed for acute serious infection and sepsis and/or organ failure in addition to VA/SCA.
4. Discussion
In this descriptive pilot analysis of unstructured data in a large EHR database, we found that the unstructured clinical notes provided adequate information to confirm the occurrence of acute VA/SCA events and additional detailed information on the course of care needed to describe the severity and other features of these events. In the past, challenges with accessing unstructured clinical notes related to resource availability (e.g., ability to manually abstract information from patient charts on paper) and legal requirements (e.g., institutional processes in place to preserve patient privacy) have limited the ability to conduct large-scale validation studies [10]. Our findings demonstrate that conducting a validation study for an operational algorithm identifying VA/SCA events in a large EHR database, such as Dataworks – USA, may be feasible using unstructured clinical notes as a reference standard.
While an operational algorithm for identifying VA/SCA has been validated in health insurance claims data [11], one has yet to be validated in a large EHR database. The development and validation of a VA/SCA algorithm in EHR data would allow studies to leverage key clinical variables typically available in EHR data but not health insurance claims, such as laboratory data (e.g., serum electrolyte measures) and genetic risk factors, while ensuring that misclassification of VA/SCA events is minimized. These potential strengths can improve a study’s ability to accurately measure VA/SCA risk in specific populations and better inform clinical decision-making, e.g., comparing the VA/SCA risk between two medications while adjusting for confounding bias from key factors such as electrolyte abnormalities or describing the clinical characteristics of patients experiencing VA/SCA from different etiologies.
For all potential events in this pilot analysis, the information documented in the unstructured clinical notes could confirm whether or not an acute VA/SCA event occurred. Our review of the clinical notes was able to discern between true acute events and other scenarios such as (1) the patient having a medical history of VA/SCA but not experiencing an acute event during the current visit, and (2) the patient initially being suspected of having VA/SCA but ultimately being diagnosed with new-onset atrial fibrillation. Additionally, 75% of potential events had documentation of ECG or telemetry findings. More comprehensive electrophysiology data were typically available for patients with a known cardiac medical history during a routine outpatient cardiology visit and patients who experienced an event during hospitalization. Inpatient electrophysiology/cardiology consult notes provided the most detailed interpretation of ECG and telemetry findings.
Generally, signs and symptoms were noted on admission for patients who initially developed an acute VA/SCA event outside of the healthcare setting, either reported by the patient or a bystander who witnessed the event (e.g., a family member witnessing the patient collapsing). The availability of signs and symptoms can be useful to confirm the location of onset of the event if explicit documentation of the location of onset is not available (e.g., signs and symptoms consistent with VA/SCA on presentation to the hospital may indicate that the event initially occurred outside of the healthcare setting) [11].
The unstructured clinical notes also provided details on the course of care for potential events that can be helpful to discern potential causes or risk factors, severity, and resolution of the event. The suspected cause or relevant risk factors were noted in the clinical notes for many potential events, which can help characterize the baseline characteristics of a population of interest experiencing VA/SCA. Information on the intensity of the care setting and medical interventions performed may be useful in discerning severity of VA/SCA events. For example, clinical trials using the Common Terminology Criteria for Adverse Events (CTCAE) define severe, non-fatal ventricular arrhythmia events (grade 3 and 4) as events that require urgent medical intervention or are life-threatening [12]. For all potential events, we were able to identify whether the patient died or was discharged from the healthcare facility. However, the specific cause of death was unclear for the two patients who died due to the complexity of the clinical cases. These patients were being treated for severe cases of infection, sepsis, and/or organ failure in addition to VA/SCA; thus, the cause of death for these patients may have been multifactorial. While the structured EHR data include codes that can identify these clinical features (e.g., diagnosis codes for baseline risk factors and in-hospital death, procedure codes for setting of care and medical interventions), these codes are prone to misclassification. Supplementing the structured data with information from the unstructured clinical notes may improve the clinical characterization of VA/SCA events.
The strength of this pilot analysis is our ability to assess the availability of information in the unstructured clinical notes from 11 separate healthcare organizations which contribute to a large EHR-derived database consisting of healthcare organizations across the US. The limitation of this exercise is that there may have been some clinical notes that were missing for individual patients in the repository of available clinical notes. Also, findings may not be generalizable to all patients included in the EHR database given the small sample selected from the subset of healthcare organizations that provide clinical notes. Last, we did not specifically identify a study population of interest using an operational algorithm in the structured EHR data, as one would do in a formal validation study. This would allow for more comprehensive description of demographics, comorbidities, and other patient characteristics, and evaluation of algorithm performance. Future validation studies should be conducted to assess the performance metrics, such as positive predictive value and negative predictive value, of an operational algorithm that identifies VA/SCA in EHR data.
5. Conclusions
Studies using large electronic databases are valuable in characterizing the risk of VA/SCA with specific medications due to their large sample sizes; however, it is unclear if an operational algorithm for identifying VA/SCA performs well in an EHR-derived dataset. We found that the unstructured clinical notes included in a large US EHR database may contain the information necessary to serve as a reference standard for validation studies focused on operational algorithms identifying VA/SCA. Future studies should be conducted to assess the performance of such an algorithm in EHR data.
Acknowledgments
The authors thank Lucy Smigiel for her project management support.
Funding Statement
This study was funded by Daiichi Sankyo, Inc.
Article highlights
Large studies using electronic health record (EHR) data can be useful for characterizing the real-world occurrence of ventricular arrhythmia and sudden cardiac arrest (VA/SCA), an important public health and drug safety concern.
An operational algorithm for identifying VA/SCA in EHR data has yet to be validated; therefore, we aimed to assess if unstructured clinical notes of a large EHR database has the information necessary to serve as a reference standard for a future validation study.
We randomly selected 20 potential VA/SCA events from the unstructured clinical notes and found that the notes had adequate information documented to confirm the occurrence of an acute event and describe the severity, interventions administered, and other features.
Future validation studies focused on operational algorithms identifying VA/SCA in EHR data may be able to use the unstructured clinical notes as a reference standard.
Author contributions
Neil Dhopeshwarkar, Charles Dharmani, and K. Arnold Chan designed the study. Neil Dhopeshwarkar analyzed the data. Neil Dhopeshwarkar, Charles Dharmani, Oluwatosin Fofah, Nora Tu, Nasser Khan, Tzuyung Douglas Kou, and K. Arnold Chan interpreted the data. Neil Dhopeshwarkar drafted the manuscript. Neil Dhopeshwarkar, Charles Dharmani, Oluwatosin Fofah, Nora Tu, Nasser Khan, Tzuyung Douglas Kou, and K. Arnold Chan critically revised the manuscript, approved the final version, and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Disclosure statement
Drs. Dharmani, Fofah, Tu, Khan and Kou are full-time employees of Daiichi Sankyo Inc. All own restricted stock units of Daiichi Sankyo. Drs. Dhopeshwarkar and Chan are full-time employees and shareholders of TriNetX, LLC. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
Ethical declaration
This study was approved by the Biomedical Research Alliance of New York (BRANY) institutional review board (IRB) (BRANY study number 24-08-484-1821). The TriNetX team submitted the study protocol for IRB review and received approval on 28 October 2024. BRANY IRB serves as the IRB of record for TriNetX research projects. To ensure compliance with human subject protection, all key personnel have completed Human Subjects Research training through The Collaborative Institutional Training Initiative (CITI) Program.
Informed consent is not required as there were no human participants involved.
Data availability statement
The data that support the findings of this study were made available to Daiichi Sankyo, Inc. under license for the current study and restrictions apply to the availability of these data. These data are not publicly available. However, the data are commercially available from the authors with permission of Daiichi Sankyo, Inc. and TriNetX, LLC.
References
Papers of special note have been highlighted as either of interest (•) or of considerable interest (••) to readers.
- 1.Zeppenfeld K, Tfelt-Hansen J, de Riva M, et al. 2022 ESC guidelines for the management of patients with ventricular arrhythmias and the prevention of sudden cardiac death. Eur Heart J. 2022;43(40):3997–4126. doi: 10.1093/eurheartj/ehac262 [DOI] [PubMed] [Google Scholar]; •• The European Society of Cardiology guidelines for managing VA/SCA, which highlights the severity of VA/SCA complications and its public health importance.
- 2.Kong MH, Fonarow GC, Peterson ED, et al. Systematic review of the incidence of sudden cardiac death in the United States. J Am Coll Cardiol. 2011;57(7):794–801. doi: 10.1016/j.jacc.2010.09.064 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Greene HL. Sudden arrhythmic cardiac death—mechanisms, resuscitation and classification: the Seattle perspective. Am J Cardiol. 1990;65(4):4B–12B. doi: 10.1016/0002-9149(90)91285-E [DOI] [PubMed] [Google Scholar]; • This article describes the relationship between ventricular arrhythmia and sudden cardiac arrest and highlights how commonly ventricular arrhythmia is observed in sudden cardiac arrest cases.
- 4.Schwartz PJ, Woosley RL. Predicting the unpredictable: drug-induced QT prolongation and Torsades de Pointes. J Am Coll Cardiol. 2016;67(13):1639–1650. doi: 10.1016/j.jacc.2015.12.063 [DOI] [PubMed] [Google Scholar]
- 5.Arizona Center for Education and Research on Therapeutics. CredibleMeds®: risk categories for drugs that prolong QT & induce torsades de pointes (TdP). 2024. [cited 2025 Mar 10]. Available from: https://crediblemeds.org/index.php/druglist; • The Arizona Center for Education and Research on Therapeutics CredibleMeds list of drugs that prolong QT and induce Torsades de Pointes, which summarizes the large number of drugs that are associated with QT prolongation/Torsades de Pointes.
- 6.Stecker EC, Reinier K, Marijon E, et al. Public health burden of sudden cardiac death in the United States. Circ Arrhythm Electrophysiol. 2014;7(2):212–217. doi: 10.1161/CIRCEP.113.001034 [DOI] [PMC free article] [PubMed] [Google Scholar]; • This article reports the rare occurrence of VA/SCA, which underscores the need for large postmarketing studies leveraging electronic healthcare data sources.
- 7.Alipour-Haris G, Liu X, Acha V, et al. Real-world evidence to support regulatory submissions: a landscape review and assessment of use cases. Clin Transl Sci. 2024;17(8):e13903. doi: 10.1111/cts.13903 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Prilla S, Groeneveld S, Pacurariu A, et al. Real-world evidence to support EU regulatory decision making—results from a pilot of regulatory use cases. Clin Pharmacol Ther. 2024;116(5):1188–1197. doi: 10.1002/cpt.3355 [DOI] [PubMed] [Google Scholar]
- 9.European Medicines Agency . Incidence rates of vulval ulceration following comirnaty vaccine. 2024. [cited 2025 Apr 11]. Available from: https://catalogues.ema.europa.eu/node/3612/administrative-details
- 10.Ehrenstein V, Hellfritzsch M, Kahlert J, et al. Validation of algorithms in studies based on routinely collected health data: general principles. Am J Epidemiol. 2024;193(11):1612–1624. doi: 10.1093/aje/kwae071 [DOI] [PubMed] [Google Scholar]; •• This article comprehensively reviews considerations for validating algorithms in electronic healthcare data sources and emphasizes the importance of algorithm validation for accurately measuring outcomes of interest like VA/SCA.
- 11.Hennessy S, Leonard CE, Freeman CP, et al. Validation of diagnostic codes for outpatient-originating sudden cardiac death and ventricular arrhythmia in medicaid and medicare claims data. Pharmacoepidemiol Drug Saf. 2010;19(6):555–562. doi: 10.1002/pds.1869 [DOI] [PMC free article] [PubMed] [Google Scholar]; •• A previously-validated algorithm for identifying VA/SCA in healthcare claims data; key clinical elements were partly derived from this reference.
- 12.U.S. Department of Health and Human Services . Common terminology criteria for adverse events (CTCAE) version 5.0. Bethesda (MD): National Cancer Institute; 2017. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study were made available to Daiichi Sankyo, Inc. under license for the current study and restrictions apply to the availability of these data. These data are not publicly available. However, the data are commercially available from the authors with permission of Daiichi Sankyo, Inc. and TriNetX, LLC.