Challenges of Medical Record Abstraction in a Long-Term Follow-up Study

• Published in: Journal of registry management Vol. 48; no. 3; pp. 88 - 91
• Main Authors: Ahmed, Muhammad F, Galfo, Andrea, Huggins, Wendy, Paddock, Lisa E, Stroup, Antoinette M, Malhotra, Jyoti
• Format: Journal Article
• Language: English
• Published: United States 2021
• Subjects: Abstracting and Indexing; Data Accuracy; Follow-Up Studies; Humans; Medical Records; Pilot Projects; chart abstraction; electronic health records; long‐term study; medical record abstraction
• ISSN: 1945-6123

Medical records are a rich source of information and have tremendous value in epidemiological research. Nevertheless, the process of obtaining and abstracting medical records for a long-term follow-up study is complicated, time-consuming, and resource intensive. We identified the following major challenges during this process. First, widely varying infrastructure of electronic health record systems used by different organizations makes it difficult to ensure that all medical charts from all sources for a particular patient have been received. Second, extensive use of free text by health care providers requires a manual line-by-line search for relevant information, which may result in some missing data due to human error. Third, there are often discrepancies between patients' provided lists of health care providers and the registry data, which may affect the data-collection process. Fourth, providers have varied requirements for medical record release of their patients, which might entail multiple patient contacts. This, in turn, can frustrate patients and discourage them from participating in current or future research studies. Fifth, the use of inconsistent medical terminology by different providers complicates conversion of unstructured text into categorical data for analysis. We have the following recommendations for any future study with similar design to overcome the above challenges. First, the source of medical records best suited for the research objectives should be identified from the beginning. Second, the abstractors should be appropriately trained to accomplish research-specific tasks. Third, a quality data-tracking system for the abstracted elements should be employed to ensure data integrity. Fourth, the abstracted cases should be reviewed by one other abstractor. We also recommend a pilot study with a smaller number of patients to evaluate the required resources before any large-scale study.