Jun 10, 2019
Part 5: Selection, Protection, and Abstraction
Published in: Manipulative and Physiological Therapeutics, Volume 42, Issue 5, pages 327–334 (June 1, 2019). doi: 10.1017/j.jmpt.2019.02.008
Posted on RAND.org on June 11, 2020
This article was published outside of RAND. The full text of the article can be found at the link above.
The purpose of this paper is to describe the 4-step process (consent, selection, protection, and abstraction) of acquiring a large sample of chiropractic patient records from multiple practices and subsequent data abstraction.
From April 2017 to December 2017, RAND acquired patient records from 99 chiropractic practices across the United States. The records included patients enrolled in a survey e-study (prospective sample) and a random sample of all clinic patients (retrospective sample) with chronic back or neck pain. Clinic staff were trained to collect the sample, scan, and transfer the records. We designed an online data collection tool for abstraction. Protocols were instituted to protect patient confidentiality. Doctors of chiropractic were selected and trained as abstractors, and a system was established to monitor data collection.
In compliance with data protection protocols, 3603 patient records were scanned, including 1475 in the prospective sample and 2128 in the random sample. A total of 1716 patients (prospective sample) consented to having their records scanned, but only 1475 could be retrieved. Of records scanned, 19% were unusable owing to illegibility, no care during the period of interest, or poor scanning. The abstractor interrater reliability for appropriateness of care decisions was fair to moderate (κ .38-.48).
The acquisition, handling, and abstraction of a large sample of chiropractic records was a complex task with challenges that necessitated adapting planned approaches. Of the records abstracted, many revealed incomplete provider documentation regarding the details of and rationale for care. Better documentation and more standardized record keeping would facilitate future research using patient records.