[Skip to Content]
[Skip to Content Landing]
Invited Commentary
April 2016

Free-Text Notes as a Marker of Needed Improvements in Electronic Prescribing

Author Affiliations
  • 1Hospitalist Service, Division of General Internal Medicine and Primary Care, Brigham and Women’s Hospital, Boston, Massachusetts
  • 2Department of Medicine, Harvard Medical School, Boston, Massachusetts
JAMA Intern Med. 2016;176(4):471-472. doi:10.1001/jamainternmed.2015.8562

In this issue of JAMA Internal Medicine, Dhavle and colleagues1 present the results of a study indicating that, of the 14.9% of electronic prescriptions with free-text notes, 66.1% contained inappropriate content for which a structured data field (ie, a field intended for certain specific pieces of information, such as days’ supply or indication for a medication) exists in the most commonly used national e-prescribing standard. Most concerning, almost 1 in 5 (19.0%) of these inappropriate notes contained conflicting medication directions from the structured fields intended for this purpose. Of the notes with appropriate content, almost half (47.3%) contained information that could be communicated using structured fields approved in a version of the e-prescribing standard that has yet to be implemented. An additional 9.6% were prescription cancellation requests for which a separate e-prescribing message exists but is not widely supported in most currently available e-prescribing software.

The implications of these various types of free-text notes can range from the merely distracting for the community pharmacist to those that could be severely harmful for patients (Table). However, even distractions can lead indirectly to patient harm if they result in lapses of attention that increase the rate of dispensing errors. Several of these categories of notes are time consuming for pharmacists and sometimes for prescribers (eg, if telephone calls are needed to clarify a prescription), therefore wasting valuable health care resources. Moreover, the final category in this Table (ie, conflicting information), as illustrated in the phenytoin sodium (Dilantin) example in the Discussion section by Dhavle and colleagues1 (and included in the present Table) is not infrequent: conflicting information apparently occurred in 14.6% of all prescriptions with free-text codes sampled by Dhavle et al, or approximately 2.2% of all e-prescriptions. If extrapolated nationally, assuming SureScripts captures two-thirds of all new e-prescriptions in the United States, this would mean that more than 34 million such errors occur each year.

Table.  
Implications of Free-Text Notes in Electronic Prescriptions
Implications of Free-Text Notes in Electronic Prescriptions

As with most medication safety issues, the causes of the problem are multifactorial. Failure of prescribers to use available structured fields likely reflects a combination of poor usability of electronic health records and e-prescribing tools plus a lack of adequate education, training, and feedback among prescribers. Indeed, in common practice, few such educational opportunities exist: prescribers often have a single training session when a new system is implemented or when they join a new practice and have no opportunities for feedback, especially from pharmacists. Failure of approved standards to be widely implemented reflects deficiencies in regulation and in the health information technology development process. The types of information in these free-text notes also reveals a wide communication gulf between prescribers and pharmacies, a gulf that can only be partially bridged with current tools, technologies, and health care structures.

Because the causes of this problem are multifactorial, the solutions must be as well. Vendors of electronic health records and e-prescribing software should view the findings of Dhavle and colleagues1as a guide to improvements that must be made in the usability of their products. For example, many vendors often make it difficult to efficiently and accurately prescribe sliding scales, tapers, or other complicated regimens (eg, “take 1 capsule in the morning and 2 capsules in the evening,” which was likely the intention of the prescriber of the above-mentioned phenytoin example). Updates in currently approved e-prescribing standards should be expedited and integrated into the next generation of tools; such changes should receive the highest priority by regulators and vendors. Future updates to the standards will also need to be developed and expedited. New structured fields should be considered when there is risk for this type of information to be omitted or incomplete when not prompted. Prescribers need adequate training in the use of e-prescribing tools. Moreover, ongoing surveillance for inappropriate free-text notes could be used as a trigger for feedback and further training as needed.

More ambitiously, changes to our health care system are needed to break down barriers between those who prescribe medications and those who dispense them. Widespread use of the Cancel Prescription Request option would clearly be an improvement over the status quo given the frequency with which discontinued medications are likely still dispensed by community pharmacies. However, a solution with even greater impact would be the automatic transmission of complete, accurate, and up-to-date medication regimens to community pharmacies. In current practice, pharmacies receive electronic prescriptions one medication at a time as they are prescribed or renewed, with little information regarding a patient’s overall regimen. Transmission of the full regimen would improve the concordance between prescribed and dispensed medication regimens, allow better detection of nonadherence, and provide greater opportunities for patient education. Switching from single transmission of medications to full regimens would require large changes in practice for both prescribers and pharmacies. It would also require accurate documentation of medication regimens, which is a well-known challenge.2 However, the potential benefits are numerous, especially during transitions of care, such as recent hospitalization, when medication discrepancies are common and often harmful.3,4

Not all clinical information in an electronic health record needs to be structured and coded. For example, systems that structure every point of clinical history or every physical examination finding are often more burdensome than they are helpful. However, medications (and likely allergies and problem lists) will likely remain structured because of the many benefits of doing so. When well-designed, electronic health record and e-prescribing systems can ensure that complete and accurate medication information is documented and transmitted and can provide clinical decision-making support, including dose suggestions, alerts, and warnings, that can improve patient safety.5 Responding appropriately to this decision support is another important responsibility of prescribers as they learn to use e-prescribing tools.

The findings of the study by Dhavle and colleagues1 should be viewed as a wake-up call to health information technology vendors, prescribers, regulators, and health care systems that several changes need to be made to improve the safety of electronic prescribing and that these changes need to be made soon.

Back to top
Article Information

Corresponding Author: Jeffrey L. Schnipper, MD, MPH, FHM, Hospitalist Service, Division of General Internal Medicine and Primary Care, Room BC3-2Y, Brigham and Women's Hospital, 1620 Tremont St, Boston, MA 02120 (jschnipper@partners.org).

Published Online: March 7, 2016. doi:10.1001/jamainternmed.2015.8562.

Conflict of Interest Disclosures: None reported.

References
1.
Dhavle  AA, Yang  Y, Rupp  MT, Singh  H, Ward-Charlerie  S, Ruiz  J.  Analysis of prescribers’ notes in electronic prescriptions in ambulatory practice [published online March 7, 2016].  JAMA Intern Med. doi:10.1001/jamainternmed.2015.7786.Google Scholar
2.
Tam  VC, Knowles  SR, Cornish  PL, Fine  N, Marchesano  R, Etchells  EE.  Frequency, type and clinical importance of medication history errors at admission to hospital: a systematic review.  CMAJ. 2005;173(5):510-515.PubMedGoogle ScholarCrossref
3.
Pippins  JR, Gandhi  TK, Hamann  C,  et al.  Classifying and predicting errors of inpatient medication reconciliation.  J Gen Intern Med. 2008;23(9):1414-1422.PubMedGoogle ScholarCrossref
4.
Salanitro  AH, Kripalani  S, Resnic  J,  et al.  Rationale and design of the Multicenter Medication Reconciliation Quality Improvement Study (MARQUIS).  BMC Health Serv Res. 2013;13:230.PubMedGoogle ScholarCrossref
5.
Bates  DW, Gawande  AA.  Improving safety with information technology.  N Engl J Med. 2003;348(25):2526-2534.PubMedGoogle ScholarCrossref
×