Discontinuation of outpatient medications: implications for electronic messaging to pharmacies using CancelRx

Abstract

Electronic communication of prescription discontinuation, or CancelRx, has the potential to improve medication safety. We aimed to describe the proportion of discontinued outpatient medications that would result in a CancelRx message to understand its impact on medication safety. We used a data report to identify all outpatient medications discontinued in the electronic health record (EHR) of an academic health system in 1 month (October 2018). Among all 63 485 medications discontinued, 23 118 (36.4%) were e-prescribed, 25 982 (40.9%) were patient-reported or reconciled, and the remainder prescribed nonelectronically. Discontinued high-risk medications were more likely to be e-prescribed (2768 of 5896, 47.0%). A discontinuation reason was specified in 37 353 (58.9%) of all discontinued medications. Approximately one-third to one-half of discontinued medications were e-prescribed within the same EHR and would result in a CancelRx message to the pharmacy. Extension of this functionality to medications reconciled in the EHR could significantly expand the impact of CancelRx on medication safety. In addition, complete and accurate discontinuation reasons are needed to optimize CancelRx implementation.

INTRODUCTION

Without effective electronic communication between prescribers and pharmacists, outpatient medications may be dispensed to patients after intended discontinuation, resulting in patient harm.^1–3 CancelRx is a transaction within the NCPDP SCRIPT standard which allows an electronic health record (EHR) to send an electronic cancellation message to pharmacy management software to communicate discontinuation of a prescription. Like e-prescribing, a cancellation message is sent from the EHR through a health information network, most commonly Surescripts, to a specified pharmacy. If the pharmacy can match the cancellation message to a prescription, that prescription is deactivated so it can no longer be filled. The pharmacy then sends a cancellation response message back to the EHR, indicating whether the cancellation was completed.

CancelRx has been identified as an important tool for communication between prescribers and pharmacists to improve medication safety and has been demonstrated to result in successful deactivation at the pharmacy of over 90% of e-prescriptions.^4–7 The ability to cancel a prescription has been required of EHR vendors as part of the US Medicare Access and CHIP Reauthorization Act of 2015 (MACRA) and the 2017 Stage 3 Meaningful Use criteria.⁸ However, there have been barriers to implementation by health care providers and institutions, leading to slow adoption,⁸ as each institution must enable this transaction within their EHR in partnership with their health information network.

Studies of implementation of CancelRx by health systems have been limited.^6,7,9,10 Within a widely used EHR, CancelRx messages are sent when e-prescriptions written within the same instance of the EHR are discontinued.⁷ When CancelRx is enabled, sending a cancellation message to the pharmacy is incorporated into the workflow of medication discontinuation, rather than as an additional, discrete choice by the provider. This ensures that CancelRx messages are sent, when possible, but may result in CancelRx transactions when no notification is needed. In addition, because patients often receive care, and prescriptions, from multiple prescribers using different EHRs, patients’ outpatient medication lists in an EHR include prescriptions that did not originate within the EHR, including those reconciled from pharmacies and pharmacy benefit managers using an intermediary, such as Surescripts, and medications manually entered as “patient-reported” without linkage to the original prescription. Although prescribers in various settings, including medical offices, emergency departments, and at hospital discharge may instruct patients to discontinue or change outpatient medications that were originally written outside their EHR, a CancelRx message will not be sent to the pharmacy for these prescriptions. The degree to which this limits the functionality of CancelRx is unknown.

OBJECTIVE

The objective of this study was to describe the proportion of discontinued outpatient medications that would result in a CancelRx message to pharmacies to further understand the potential impact of this functionality on medication safety.

MATERIALS AND METHODS

We identified all outpatient medications which were discontinued from patients’ medication lists in the EHR of 2 hospitals and their affiliated outpatient practices from October 1–31, 2018. Outpatient medications were classified into 3 categories: e-prescriptions, nonelectronic prescriptions (eg, printed, faxed), and patient-reported or reconciled medications. Medications were identified as reordered, rather than discontinued, if reorder was specified as the reason for discontinuation, which is automatically populated by the reorder function in the EHR. We defined a set of high-risk medications based on the ISMP List of High-Alert Medications in Community/Ambulatory Healthcare by the pharmaceutical class or therapeutic class designated in the EHR. These included antiretrovirals, anticoagulants, hypoglycemic agents, antineoplastics, and immunosuppressants. We did not include narcotics as these could not be electronically prescribed during the period of this analysis. We categorized a medication as having an associated pharmacy if there was a pharmacy listed in the order details within the EHR. We conducted data analysis using SAS 9.4 (SAS Safety Corp., Cary, SC, USA). This study was reviewed by the Johns Hopkins School of Medicine Institutional Review Board.

RESULTS

We identified 63 485 (58.4%) discontinued and 45 168 (41.6%) reordered medications. Among all discontinued medications, 23 118 (36.4%) were electronically prescribed within the EHR, so discontinuation could be communicated to the pharmacy via CancelRx. However, among discontinued e-prescriptions, 2806 (12.1%) were written >12 months prior and could not be filled at the pharmacy. Among all discontinued medications, an additional 25 982 (40.9%) were patient-reported or reconciled from outside sources, and 14 385 (22.7%) were ordered in the EHR but not electronically prescribed. Of patient-reported or reconciled medications, 10 808 (41.6%) had a pharmacy associated with the medication record in the EHR.

Discontinued medications from highest-risk medication classes were more likely to have been e-prescribed within the EHR. Among 5896 discontinued highest-risk medications, 2768 (47.0%) were electronically prescribed and could be discontinued at the pharmacy through CancelRx. An additional 1717 (29.1%) were patient-reported or reconciled, while 1411 (23.9%) discontinued medications were ordered in the EHR but not electronically prescribed.

A reason for discontinuation was specified in 37 353 (58.9%) of all discontinued medications. Among those with discontinuation reasons, a majority were nonspecific, indicating that the prescription was discontinued or completed (24 104, 64.5%); clinically significant reasons were less frequently selected, including error (4586, 12.3%), dose adjustment (2886, 7.7%), alternate therapy (1574, 4.2%), ineffective or unavailable (1013, 2.7%), and adverse drug reaction or contraindication (384, 1.0%). The proportion that was e-prescribed varied by discontinuation reason (range, 28.2%–48.9%), with an additional 22.9%–56.7% patient-reported or reconciled (Figure 1).

Figure 1.

Proportion of outpatient medications by order type for each clinically meaningful discontinuation reason.

DISCUSSION

Approximately one-third of outpatient medications discontinued for clinically significant reasons and one-half of high-risk medications were originally prescribed in the EHR and would trigger a CancelRx message upon discontinuation. A substantial proportion of remaining discontinued medications were patient-reported or reconciled from outside sources. When a reason for discontinuation was indicated, only one-third specified a clinically significant reason that should result in communication with the pharmacy.

CancelRx has been demonstrated to be highly effective in communicating discontinuation of e-prescriptions from EHRs to pharmacies.^6,7 However, over the past 20 years, growth in specialty care in the US has resulted in care fragmentation; a majority of outpatient visits now occur with specialists and nearly one-third of Medicare patients see 5 or more physicians.¹¹ As a result, patients’ medications are prescribed from multiple EHRs, with medications entered as patient-reported or reconciled into EHRs without the original prescription. The substantial proportion of patient-reported and reconciled medications in our analysis particularly illustrates the challenge in hospital-based settings, where patients may be more likely to receive care in multiple health systems, resulting in fewer discontinued medications e-prescribed within the EHR, and a larger gap in CancelRx functionality. Although the exact proportion of outpatient medications that are e-prescribed in an EHR will likely vary by institution or setting, our results demonstrate that in addition to maximizing e-prescribing, extension of CancelRx functionality to include medications reconciled from outside sources would significantly expand its impact on medication safety.

Our results Illustrate the complexity of medication management, as indicated by the volume of discontinued medications, the diversity of the types of outpatient medications (eg, e-prescriptions, patient-reported medications), and the reasons for discontinuation.

These results support several additional recommendations relevant to CancelRx implementation.

First, EHR vendors and institutions implementing CancelRx should balance the potential safety benefit of CancelRx messages with the impact of these transactions on pharmacies. In Epic (Epic Corporation, Verona, WI), a leading EHR, a CancelRx message is sent by default when an e-prescription is discontinued and does not require a specific action by the discontinuing user to initiate it. While this ensures that important CancelRx are not missed, it also leads to CancelRx messages when prescriptions cannot be dispensed, including expired prescriptions and medications that have been dispensed and do not have refills (ie, “clean up” of medication lists during medication reconciliation). In our study, 12% of discontinued e-prescriptions were more than 12 months old and could not be dispensed; sending CancelRx messages for older e-prescriptions and those dispensed and without refills may reduce unwanted medication refill requests but have more limited safety benefits and add work at pharmacies, leading to perceptions that CancelRx messages have “low-value”.⁹

There are some configurations that can restrict when CancelRx messages are sent, including the function used to discontinue the e-prescription (eg, reorder, change, discontinue) and the reason for discontinuation. While sending a CancelRx message to pharmacies when a medication is reordered would reduce potentially fillable duplicate prescriptions at the pharmacy, these data suggest this would result in an approximately 70% increase in CancelRx transactions from our EHR. Further work is needed to determine the optimal strategy to identify when CancelRx transactions should be sent to pharmacies.

Second, EHRs should make it easy to document an appropriate, clinically meaningful reason for medication discontinuation. A reason for discontinuation was not selected in nearly half of medication discontinuations. In addition, a majority of reasons were nonspecific (eg, “stopped at discharge”). While the discontinuation reason is not currently transmitted in the CancelRx message, it will be included in a future version of the NCPDP SCRIPT standard to communicate this information to pharmacies. Therefore, it will be important to have a taxonomy of discontinuation reasons that meets the information needs of pharmacy staff. Several studies have classified discontinuation reasons, including a mnemonic of ABCDE for alternate therapy, brand changes, completion of therapy, dose change, erroneous order,¹² with other reasons including lack of efficacy, intolerance or adverse drug event, including laboratory abnormalities, no longer indicated, contraindication, and patient preference.^1,13,14 Further understanding of the ideal set of discontinuation reasons and accuracy of selection is required to ensure the selection of the appropriate reason to minimize missing data and facilitate effective communication across healthcare providers.

Third, when a CancelRx message is only sent for e-prescriptions written in the discontinuing EHR, and not all pharmacies accept CancelRx transactions, it is critical for EHR vendors to ensure that prescribers understand whether a CancelRx message will be sent to the pharmacy.⁷ Clear indication in real-time whether a CancelRx transaction will be sent—and display of the outcome of the CancelRx transaction—would allow prescribers to use alternative communication methods, when needed.

This study was limited to medications discontinued from patients’ medication lists in the EHR in 1 month at an academic medical center. As we did not have access to pharmacy data, we cannot determine which discontinued medications had an existing prescription at the pharmacy, and whether they were fillable. We only captured medication discontinuations that were documented in the medication list and would not identify medications that were verbally discontinued in communications with patients or pharmacies, but not removed from the medication list. Some discontinued outpatient medications may not have been identified as reordered if prescribers created a new prescription rather than using the reorder function in the EHR. Only about half of discontinued medications had a reason identified. Because of health system workflows, some of the reasons may be over-represented, including “discontinue at discharge” which is automatically populated in discharge medication reconciliation, and “error” which is used by nonprescribers to make corrections to data entry errors in patient-reported medications. As a result, some discontinuations due to error might not be clinically meaningful. Finally, the accuracy of pharmacy information associated with outpatient medications in the EHR other than e-prescriptions is unknown. However, this study still provides insights into the complexity of e-prescribing and identifies a gap in CancelRx functionality with important implications for CancelRx implementation. Additional studies with pharmacy data could confirm these results and further quantify the added benefit of extension of CancelRx functionality.

CONCLUSION

Approximately one-third of outpatient medications discontinued for clinically meaningful reasons and one-half of high-risk medications would result in a CancelRx message to the pharmacy. Extension of CancelRx functionality to prescriptions reconciled from other EHRs could significantly expand the impact of CancelRx on medication safety. Complete and accurate discontinuation reasons are needed to inform when CancelRx transactions should be sent and to provide situational awareness to other stakeholders, including pharmacies.

FUNDING

SIP was supported by a Doris Duke Early Clinician Investigator Award. SIP has also received funding from the NCPDP Foundation and AHRQ for research on CancelRx.

AUTHOR CONTRIBUTIONS

The authors confirm contribution to the paper as follows: study conception and design: SIP and ARC; data analysis: SIP; all authors contributed to interpretation of results, draft manuscript revision, and approved the final version of the manuscript.

CONFLICT OF INTEREST STATEMENT

None declared.

Data Availability

Individual participant data are not available for data sharing, but aggregate data may be requested by communication with the corresponding author.