Retrospective chart review studies: an old dog with some new tricks?

In UBC's latest article, Krista Payne and Dara Stein discuss the use of retrospective chart review studies as a way of capturing real-world patient-level clinical, healthcare and safety data.

Though not a new methodology, retrospective chart reviews remain a viable approach to effectively capture real-world patient-level clinical, healthcare utilization, and safety data. A balance of both good science and pragmatic execution, increasingly, chart review studies are warranted to help fill evidence gaps left unresolved by other methodologies or data sources. Europe, especially, has seen an increase in use of chart reviews because databases to populate burden of illness and other health economic and safety evaluations are not consistently available. In the US, even when secondary sources of healthcare and administrative data are available and access is not a challenge, the level of data specificity required to address the research questions of interest is frequently inadequate. As the evidence requirements of payers and regulators continue to converge, data collection solutions like chart review studies that aid in development of tailored patient-level repositories of data are likely to become more common.

Specific real-world evidence needs addressed by chart reviews are diverse and include:

• Patient characteristics such as demographics, diagnoses, medical history, and concomitant illness

• Patterns of care, such as healthcare visits by type of healthcare provider, medications, diagnostics, procedures, accident and emergency visits, and hospitalizations

• Patterns of medication prescribing and drug utilization

• Effectiveness including clinical outcomes and patient-reported symptoms

• Identification and evaluation of unmet clinical need

• Safety data including documented adverse events, and serious adverse events

"...chart review studies are warranted to help fill evidence gaps left unresolved by other methodologies or data sources."

Burden of illness: defining unmet medical need and quantifying healthcare resource use

Chart review studies are frequently used to inform and populate burden of illness and other health economic analyses. For example, payers want to understand the patterns and costs associated with the current standard of care in order to determine the potential economic and health outcomes impact of a novel therapy, or formulation, once it is available after approval. The challenge is that unlike clinical outcomes and drug tolerability profiles that may be generalized from one geography or healthcare system to another, patterns of healthcare delivery and associated healthcare costs are highly local. Market access and reimbursement activities are managed separately for each country, and must address local evidence requirements. Hence, primary data collection that results in country-specific, patient-level data is often warranted to support local Health Technology Assessment (HTA) dossier submissions. As a stand-alone study, or as a methodology to fill gaps in countries where suitable sources of secondary data do not exist, chart review studies can generate the evidence to support increasingly complex value demonstration, reimbursement and market launch decision-making.

Patterns of drug utilization: evaluating appropriate use of medicines

Drug utilization research emerged in northern Europe in the mid-1960s and began with studies of inter-country and regional differences in antibiotic drug use. It eventually led to the formation of the World Health Organization (WHO) European Drug Utilization Research Group (DURG), and, more than 40 years later, this area of research continues to thrive. As delineated by the WHO the key focus areas include:

• Why are the drugs prescribed?

• Who are the prescribers?

• For whom do the prescribers prescribe?

• Are the patients taking their medicines correctly?

• What are the benefits and risks of the drugs?

"Drug utilization research emerged in northern Europe in the mid-1960s..."

Drug Utilization Studies (DUS) mandated by healthcare regulators, such as the European Medicines Agency (EMA), are increasingly common, and are accepted as a valid Post-Authorization Safety Study (PASS) methodology. Though DUS may not be as encompassing as prospective PASS in terms of the research questions addressed or the scope of data collected, they may be an appropriate add-on, or in some cases even a viable stand-alone alternative, in the assessment of drug safety. DUS objectives frequently include evaluations of off-label or inappropriate drug use, as well as prescriber adherence to US Risk Evaluation and Mitigation Strategy (REMS) or other risk mitigation requirements, making response bias potentially challenging. Retrospective DUS offer an advantage versus the prospective designs because they can provide data void of prescriber response bias, due to the Hawthorne Effect – the phenomena of behavioral changes as a result of participant awareness of being observed.

Chart review study operations: where science and practicality meet

Using a chart review approach, a protocol and case report form (CRF) are developed to describe the study methodology including processes for prescriber or patient identification, case ascertainment, sampling of medical records, and data abstraction. Chart review studies can be characterized by a few key design features:

 
• Detailed inclusion and exclusion criteria based on variables that are documented in routine medical care charts

• A study eligibility period within which patients are eligible to be enrolled in the study

• A defined index event, such as a first prescription of a specific drug , which typically occurs at some time-point within the eligibility period

• A study follow-up period which defines the period of time following the date of the index event within which data can be included for any given patient

The design will also need to determine whether the study will include a single or multiple cohorts of prescribers or patients, common eligibility and study follow-up periods are used or cohort specific ones, and the time period for study execution. If the DUS is designed to evaluate the real-world effectiveness of a REMS or Risk Management Plan (RMP), this must also be considered early in the study conceptualization process, because it is likely to impact various design parameters such as the eligibility period, as well as patient and prescriber sampling frames.

"Despite these challenges, under an increasing number of circumstances, multinational chart review burden of illness or DUS may be warranted."

A few key challenges should be considered when executing burden of illness or DUS studies based on a retrospective chart review approach:

• Although informed consent is not required in most countries for retrospective chart review studies, there are some exceptions. The need for informed consent can prolong study timelines as well as contribute to patient selection bias.

• In a database analysis of secondary health data, patients are sampled using an automated algorithm, hence, a full denominator of all "eligible" cases can be ascertained quite rapidly. In contrast, chart review studies require each site-based study team to first build a sampling frame through identification of potentially eligible subjects. In the absence of electronic medical records – often the case in Europe – this can be laborious and time consuming.

• Given the time and costs associated with implementing a study, including identifying and enrolling medical sites and primary data collection, chart review DUS typically include fewer sites and patients than database studies. Primary data collection using charts for sample sizes as large as those of DUS using secondary sources of data are usually prohibitive from both cost and logistics perspectives. Depending on the research objectives, especially if the aim is to document the occurrence of rare events or outcomes, the DUS chart review approach may be inadequate.

Despite these challenges, under an increasing number of circumstances, multinational chart review burden of illness or DUS may be warranted. Especially in the context of an EMA or US Food and Drug Administration (FDA) post-market DUS requirement when secondary data sources are lacking, this approach to scientifically rigorous evaluations of drug utilization patterns may be a strategic and worthwhile investment.

About the authors:

Krista Payne Med, Executive Director of Value Demonstration; Safety, Epidemiology, Registries and Risk Management and Principal Scientific Consultant

Email: krista.payne@ubc.com

Krista Payne is a Principal Scientific Consultant and Executive Director of Value Demonstration, within the Safety, Epidemiology, Registries and Risk Management group at UBC in Dorval, Quebec. This department is responsible for the conceptualization, and design of strategic evidence gathering programs, and real-world observational studies in support of market access and product reimbursement. Routinely, value demonstration studies have been designed to provide tailored datasets to populate health economic models or other peri- and post-approval quantitative and qualitative burden of illness evaluations.

Since 1996, her observational pharmacoeconomic research efforts have included projects in relation to oncology, Alzheimer's disease, depigmentation, hypercholesterolaemia, hypertension, stroke, pertussis, COPD, Thalassemia, haematological disorders, C difficile, stem cell mobilization, and chronic liver disease, amongst others. Ms. Payne has designed, planned, and coordinated a number of international health economic assessments as well as implemented international burden of illness studies, and health and social service resource use and patterns of care. She has a keen interest in global studies of actual practice and has designed and overseen prospective and retrospective data collection projects in more than 14 countries around the world.

Dara Stein, MSc, Senior Research Associate, Value Demonstration, Safety, Epidemiology, Registries and Risk Management

Ms. Stein is a Senior Research Associate in Value Demonstration within Safety, Epidemiology, Registries and Risk Management at UBC in Montreal, Quebec. Ms. Stein holds a BSc in Human Kinetics from the University of Ottawa, and a MSc in Human Nutrition from McGill University.

Prior to joining UBC, Ms. Stein worked as a Research Associate in the Division of Clinical Epidemiology of the Research Institute of the McGill University Health Centre. Ms. Stein has experience designing and managing observational research studies including retrospective chart reviews, and prospective and cross-sectional cohort studies. Ms. Stein has worked in various therapeutic areas including oncology, cardiovascular diseases (e.g. atrial fibrillation), intensive care, infectious diseases (e.g. peritonitis, hepatitis), and acute bleeding events. Ms Stein research focuses on drug utilization, safety, burden of illness, treatment patterns and resource utilization (including direct and indirect cost of healthcare). She also has wide experience working on mandated (i.e. by FDA or EMA) and non-mandated chart review applications of post-authorization safety studies (PASS) in North America and Europe including studies that have been registered with ENCePP.

Are retrospective chart review studies a strategic and worthwhile investment?