What Is a Chart Review?
A chart review, also referred to as a chart abstraction or medical record review, is a specific method of RWD collection commonly used to gain insights in healthcare research. Chart reviews involve systematically reviewing the medical records, or charts, of a patient population to gather information on a specific clinical issue, treatment, or outcome.
Chart review relies on the collection of structured and unstructured data from within the electronic medical records (EMR) or electronic health records (EHR) which will be analyzed for research purposes. The method of chart review is considered only when a research need exists for supplementing structured patient data with unstructured data.
What Is Unstructured Data?
Unstructured data is information that does not fit neatly into a predefined data model or schema and in the case of chart reviews, frequently takes the form of free-text provider notes. Unstructured data may originate from anyone interacting with and recording information within the patient chart, including the patient themselves, treating physicians, nurses, radiologists, etc.
Many important details regarding a patient’s genetics, experience, and outcomes exist within the EHR as unstructured data including genomics reports, over-the-counter (OTC) medication use, treatment lines of therapy, reason(s) for medication discontinuation, and patient-generated information (eg, patient-reported outcomes [PROs], wearables, digital therapeutics [DTx], etc).
How Is Unstructured Data Accessed in Chart Review?
A trained professional, often a nurse or a physician, reviews the patient chart and abstracts relevant information into a research database that has been specifically designed in accordance with research protocols and needs. For example, an abstractor may read the following within a patient chart “John Smith complained of nausea and migraines over the last 3 days” and record both the nausea and migraine within the research database. As the abstractor is performing data entry, this patient data is independently reviewed, cleaned, and monitored for quality. Once all data entry is complete, patient information is deidentified in accordance with research protocols and analyzed for insight generation.
Types of Chart Review: Retrospective Vs Prospective
Chart reviews are frequently observational in nature and may be performed as either retrospective or prospective.
Retrospective Chart Review
A retrospective review is performed when patient chart information is being gathered using data from the past, with no influence on provider-captured information within the context of the patient’s visit.
Prospective Chart Review
A prospective review is performed when patient data gathering is future-focused, and data are abstracted from the patient chart as it becomes available. Aside from timeframe, one key difference between the two is prospective reviews allow for the possibility of influencing the information captured within the patient’s treatment.
Chart Review Considerations
While chart review has a great power to access expanded insights into the real-world patient experience, projects using chart review as a data collection method have historically been avoided, or used as last-case resort, due to their notoriously lengthy timelines and high cost of implementation. Once researchers determine the necessity for unstructured data insights, they need to carefully consider partners for implementation. Choosing a research partner with a deep understanding of chart review implementation, management, and data sources is critical to ensuring projects stay on timeline, on budget, and have high-quality data.
AI Vs Human Chart Review
With the recent boom of artificial intelligence (AI) use in the healthcare industry, many data scientists are seeking to outsource some or all their abstraction activities, significantly reducing the time burden of manual abstraction. While this use of AI may revolutionize the landscape of chart review projects, it is important to note that we have not reached a place where humans may be removed from the process. When evaluating data vendors using AI to abstract unstructured data, it is important to consider data source, training, and review of AI algorithms being utilized.
To summarize the considerations for chart review, regardless of abstraction method, researchers must carefully consider the following factors and how these will affect insight acceptance among targeted stakeholders:
- Data Source: Ensure that patient data undergoing abstraction or review is relevant, heterogeneous, and diverse
- Data Collection Design: Data structure, and data collection instruments, should be built in a manner that is easily analyzable, limits the use of free-text fields, and provides options that indicate missing information to minimize or remove missingness in the final dataset
- Training: When using human abstractors, training guidelines should be developed and implemented that are comprehensive and clear, with continuous monitoring for any need for retraining or guideline updates. When using AI abstraction, it is critical to ensure that the algorithms or models being used are consistently and specifically trained and validated on the data of interest (here, patient medical data)
- Data Quality and Review: As abstraction is ongoing, incoming data feeds should be consistently monitored and reviewed for any errors. With human abstraction, researchers will often implement either a double data entry (DDE) or source data verification (SDV) system to manage quality and simultaneously manage any discrepancies in abstractor data entry processes. In the use of AI abstraction, human review of algorithm-abstracted data is required in order to comply with regulatory standards. Another related consideration here would be the use and maintenance of data audit trails, which becomes particularly important when the research goal is Food and Drug Administration (FDA) submission
- Ethics: Given the review and transfer of potentially sensitive patient data, researchers should continuously manage data flow to ensure patient privacy is protected and the research remains compliant with all applicable regulatory standards
Use Cases of Chart Review
Use cases for the abstraction of patient chart data are varied, with abstracted information providing benefits across the market access spectrum. Given the importance of unstructured data in a chart review, some examples of potential insights include:
- Clinical Evaluation of Treatment (side effect burden, outcomes, etc)
- Revealing Patient Characteristics (sentiment, satisfaction, unmet need, etc)
- Improving Market Access Strategy (treatment adoption, pricing strategies, patient support programs, etc)
- Identifying Emerging Trends (changes in standard of care, physician adherence to treatment guidelines, off-label use of medications, etc)
Chart Review at Magnolia Market Access
We specialize in the collection, management, and analysis of unstructured data, including chart reviews, with a strong focus on process optimization and efficiency. Reach out to learn how we can help you integrate both structured and unstructured RWD into your market access strategy.
Magnolia Market Access Authors: Beni Turner, Pamela Landsman-Blumberg
