Case Objectives

• List the most common errors associated with computerized provider order entry.
• Discuss standards for the transmission of electronic prescriptions to pharmacies.
• Appreciate how complex medication regimens can be difficult to enter into computerized provider order entry systems.
• Describe interventions that can help mitigate errors associated with computerized provider order entry systems.

The Case

A 71-year-old man underwent resection of a colorectal cancer. Unfortunately, his hospitalization was complicated by an acute pulmonary embolism (PE), which was treated with rivaroxaban (a new oral anticoagulant).

At the time of his discharge home, the physician electronically prescribed his medications; the prescription was sent to an outpatient pharmacy. He required two prescriptions for rivaroxaban (per protocol), one for 15 mg twice a day for 10 more days, and then 20 mg daily after that. The discharging nurse reviewed the full medication list (13 medications) with the patient and his wife. The prescriptions were filled by the pharmacy.

Ten days later, the patient's wife returned to the pharmacy requesting a refill of the rivaroxaban 15 mg twice a day. On re-reviewing the medications, the wife explained the patient had been taking both prescriptions at the same time (a total daily dose of 50 mg daily). This overdose had placed him at very high risk for bleeding complications. Fortunately, he did not experience any adverse events. The outpatient pharmacist worked with the physician, wife, and patient to clarify the proper dosing.

The patient safety committee at the hospital performed a full review of the case. The hospitalist had appropriately entered the prescriptions in the electronic health record, including the appropriate start and stop dates for the rivaroxaban. However, when they reviewed the prescriptions in the outpatient pharmacy system, there was no start date associated with the 20 mg rivaroxaban. Further testing revealed that neither the start nor stop dates were transmitted to the outpatient pharmacy.

Moreover, the patient safety committee learned that transmission of start and stop date fields in electronic prescriptions is not required, and there is no national standard. The committee realized that every patient discharged from the hospital with an electronic prescription was at risk for adverse events because of this issue. They wondered what other important information was not transmitted and what individuals and institutions could do to try to prevent such adverse events.

The Commentary

by Adam Wright, PhD, and Gordon Schiff, MD

Two decades ago, at the dawn of the patient safety movement, there was optimism that electronic prescribing could eliminate many types of medication errors. Indeed, early studies at our institution and others showed that moving from poorly legible handwritten prescriptions to easily readable (by humans and computers) prescriptions significantly decreased the rate of medication errors.(1,2) Electronic prescriptions were populated by more structured, standardized, readily available dosing information and enhanced by thoughtful and timely decision support alerts (related to drug allergy, drug interactions, drug lab precautions). Although on balance computerized provider order entry (CPOE) has been a positive safety intervention, we now know that (i) it has not fulfilled its promise of preventing many types of errors and (ii) it can introduce new types of errors—as illustrated by this patient's potentially harmful rivaroxaban error.(3-6)

In a series of studies funded by the US Food and Drug Administration and the National Patient Safety Foundation, we sought to identify ways that electronic prescribing can prevent error, fail to detect or prevent errors it was designed to minimize, or introduce novel problems.(6,7) We examined medication error and safety reports, computer help-desk logs, and orders that were subsequently discontinued that were designated as erroneous by clinicians entering the discontinuation orders. We also watched clinicians attempt to perform a series of standardized orders on different outpatient systems to observe potential errors by the prescriber and/or computer.(8-10) We found that the types and frequency of errors differed depending on the mechanism by which they were being identified and on the systems being studied; this finding strongly suggested we were looking at only the tip of a larger iceberg. We also heard from clinicians about frustrations in learning and using CPOE systems, including difficulty ordering complex medication regimens, inflexible systems, and excessive alerts—these challenges led to error-prone workarounds and alert fatigue, which in turn resulted in clinicians overriding potentially critical warnings. We developed a detailed taxonomy of the types of errors that were seen and reported, and we summarized the most common and serious errors (Table 1).(6) Most concerning was a surprising lack of systematic and proactive efforts to collect and learn from CPOE errors. Errors often went unreported and uninvestigated. In the rarer cases where errors were fully investigated, the results were not widely and transparently shared. Fortunately for this patient's rivaroxaban error, the error was identified before any harm occurred, and the error was investigated and shared in the health care system in which it occurred.

In this case, it appears that the inpatient discharging clinicians made laudable efforts to clarify the precise dosing sequencing and to educate the patient about this fairly complex initiation regimen. However, despite these efforts, critical information was not effectively transmitted to the pharmacy or the patient. This problem is common with both health information technology (IT) and patient education.(11) Increasingly, staff enter information into the computer, assuming it will be transmitted and viewed by the appropriate individuals.(4) However, without well-designed and rigorously tested systems that include effective feedback loops, reliable communication cannot be assumed. Like this case, there have been multiple prior reports of critical information being entered in CPOE systems by an ordering clinician (often in free-text fields) that turned out to be fields that were not transmitted to or seen by the pharmacist.(12,13)

The National Council for Prescription Drug Programs (NCPDP), a nonprofit multistakeholder forum, is responsible for developing standards for transmitting electronic prescriptions to pharmacies. In the United States, most electronic prescriptions are sent through the Surescripts network. Surescripts has built a network to link prescribers, pharmacy benefit managers, pharmacies, and many other organizations.

Implementation guidance for NCPDP's SCRIPT standard recommends that physicians order structured taper or changed regimens (as this patient's clinical team did) and discourages using free-text notes for tapers.(14) Even though the clinicians in this case did what they should have done, the pharmacy could not see critical details of the prescription. It is unclear whether the pharmacy system receiving this patient's prescription was the hospital's own ambulatory pharmacy or an outside community pharmacy. We also don't know if the prescription was electronically transmitted through Surescripts or another connection using the NCPDP SCRIPT standard, or whether it was through an internal connection to the hospital's pharmacy system or even via a paper prescription faxed or hand-carried by the patient to their local pharmacy. Regardless of how the prescription was sent, there are often multiple opportunities for information "voltage drops"—in this case resulting in loss of details about the rivaroxaban dosing instructions.

However, it is worth noting that the complexity of the regimen for initiating rivaroxaban predisposes to confusion and errors, some of which have been previously reported in the literature.(15) The patient is supposed to start with a higher dose (30 mg/day), but achieves that by initiating treatment with a lower tablet strength (15 mg) taken twice daily, and then switches to a lower daily dose by switching to a higher tablet strength (20 mg taken once daily) after 21 days. In addition to cases like this patient's, the literature reports patients erroneously continuing the twice daily routine after the 21 days, thus taking the 20 mg pills not once but twice (40 mg) daily. There are even differing recommendations about whether to take with food (instructions for the initial higher strength tablets) versus without needing to eat (which is acceptable at lower strength of the medication).(15)

To enter this order in the computer, a clinician might consider using a "ramp up," "taper down," or "combination" regimen feature. Our hospital uses a widely deployed electronic health record/CPOE system (Epic) that helpfully allows for tailoring regimens for either type of serial dosing changes. But with elements of both here, this regimen lends itself to confusion in placing the order. Further, when we tested our ability to enter this regimen using these advanced ordering features in our Epic system, we were unable to accommodate the different strengths required, leading to a taper that involved taking 1.33 of the 15 mg tablets (which the system rightly warned was not recommended), or as many as six 5 mg tablets. All of these "Swiss Cheese" holes predispose to errors, suggesting the need for different methods to prevent such errors. We have summarized some of the ways institutions and vendors could better approach anticipating and preventing some of these errors (Box).

Fortunately, in 2014 the manufacturer (Janssen Pharmaceuticals) introduced a "starter pack" to simplify dosing for patients initiating rivaroxaban (Xarelto) for acute deep vein thrombosis (DVT) or pulmonary embolism (PE).(16) Thus, a patient can more reliably start on the higher daily dose (30 mg) for the first 3 weeks (based on the higher risk of recurrence during this initial period), and then transition to the 20 mg daily dose for the next 7 days (presumably to be followed by a subsequent prescription for the additional course of treatment). Ordering this starter pack seems much less error-prone, and we were able to generate an error-free prescription in our Epic system. This assumes the ordering clinician is aware of this option, and there are no insurance or excess copay barriers to ordering this product this way. Moreover, the exact calculation of how many remaining days for the starter pack might vary depending on how many days of inpatient treatment a patient had already received in the hospital before discharge (which could be offset by initiating the starter pack while in the hospital, removing pills from the starter pack before dispensing, or instructing the patient not to take all of the pills at the beginning of the starter pack).

Finally, it must be mentioned that this error occurred in the context of two high-risk clinical domains: anticoagulation and transition from hospital to home. Interactions between the various components of this system require a situational awareness of the various things that can go wrong and potential for new errors in these domains. Direct oral anticoagulants (such as rivaroxaban) were developed to overcome the myriad of known risks and complexities in oral anticoagulation with warfarin (Coumadin) and may be easier and safer to use than warfarin.(17) Ironically, had the patient been on warfarin, he likely would have been more closely monitored by a pharmacist, nurse, or physician, and an error (on the part of the patient, pharmacist, or computer) might have been detected earlier. Computerized provider order entry systems have become increasingly sophisticated in their capabilities to handle complex ramping or tapering regimens; yet, our state-of-the-art commercial system still had trouble handling this rivaroxaban starting regimen. This dosing regimen, intended to lower recurrence risk (based on an untested assumption that this tiered regimen is safer and more effective) spawned a "starter pack" to simplify initiation for treating PE or DVT. However, there is now a risk that a physician might erroneously use it to start prophylaxis for stroke in a patient with atrial fibrillation or at risk for DVT or pulmonary embolism (prophylaxis requires lower dosing). Ordering by indication can help ensure that clinicians order appropriate dosing for medications where the dose differs by indication.(18,19)

This leads to one final, but in some ways most critical, question related to anticoagulation in this case. Could his PE have been prevented had he been given DVT/PE prophylaxis in the hospital? A cancer inpatient undergoing abdominal surgery is at high risk, and guidelines suggest starting prophylaxis for such patients.(20) Rather than just thinking about "starting packs," we should also consider "starting lacks" that might have prevented this serious, potentially preventable, complication.

Take-Home Points

• Anticoagulation, transitions from inpatient to home, and complex regimens are high-risk situations calling for extra attention and safeguards to avoid known risks and ensure patient education and error-free use of these medications. Safeguards should include both heightened situational awareness on the part of the clinicians caring for the patient as well as improved systems to prevent errors.
• Improvements to computerized provider order entry systems, including indication-based ordering (that would allow tailoring orders to different clinical indications) and better testing of the integration between CPOE and pharmacy systems, are needed.
• For several of the newer oral anticoagulation drugs (e.g., rivaroxaban, privaroxaban), initiating treatment for deep vein thrombosis or pulmonary embolism requires complex dosing regimens, in which the patient begins with a higher dose, followed by treatment at a lower dose. Manufacturers of these medications now market "starter packs" that appear to make this easier and less error-prone.
• Medication errors should be routinely reported and investigated, with attention to the role electronic ordering can play in both preventing future errors as well as introducing novel ones.

Adam Wright, PhD

Professor of Biomedical Informatics

Vanderbilt University Medical Center

Gordon Schiff, MD

Associate Director, Center for Patient Safety Research and Practice

Division of General Internal Medicine, Brigham and Women's Hospital

Associate Professor of Medicine Harvard Medical School

Quality and Safety Director Harvard Medical School Center for Primary Care 

Faculty Disclosures: Drs. Wright and Schiff have declared that neither they, nor any immediate member of their families, have a financial arrangement or other relationship with the manufacturers of any commercial products discussed in this continuing medical education activity. In addition, the commentary does not include information regarding investigational or off-label use of pharmaceutical products or medical devices.

References

1. Bates DW, Teich JM, Lee J, et al. The impact of computerized physician order entry on medication error prevention. J Am Med Inform Asso. 1999;6:313-321. http://www.ncbi.nlm.nih.gov/pubmed/11529801
2. Gandhi TK, Weingart SN, Seger AC, et al. Outpatient prescribing errors and the impact of computerized prescribing. J Gen Intern Med. 2005;20:837-841. http://www.ncbi.nlm.nih.gov/pubmed/16117752
3. Koppel R. What do we know about medication errors made via a CPOE system versus those made via handwritten orders? Crit Care. 2005;9:427-428. https://www.ncbi.nlm.nih.gov/pubmed/16277728
4. Nanji KC, Rothschild JM, Salzberg C, et al. Errors associated with outpatient computerized prescribing systems. J Am Med Inform Assoc. 2011;18:767-773. http://www.ncbi.nlm.nih.gov/pubmed/21715428
5. Westbrook JI, Baysari MT, Li L, Burke R, Richardson KL, Day RO. The safety of electronic prescribing: manifestations, mechanisms, and rates of system-related errors associated with two commercial systems in hospitals. J Am Med Inform Assoc. 2013;20:1159-1167. http://www.ncbi.nlm.nih.gov/pubmed/23721982
6. Schiff GD, Hickman TT, Volk LA, Bates DW, Wright A. Computerised prescribing for safer medication ordering: still a work in progress. BMJ Qual Saf. 2016;25:315-319. http://www.ncbi.nlm.nih.gov/pubmed/26515444
7. Computerized Prescriber Order Entry Medication Safety (CPOEMS): Uncovering and Learning From Issues and Errors. Brigham and Women's Hospital, Harvard Medical School, Partners HealthCare. Silver Spring, MD: US Food and Drug Administration; December 15, 2015. Available at http://www.fda.gov/Drugs/DrugSafety/MedicationErrors/ucm477360.htm
8. Schiff GD, Amato MG, Eguale T, et al. Computerised physician order entry-related medication errors: analysis of reported errors and vulnerability testing of current systems. BMJ Qual Saf. 2015;24:264-271. http://www.ncbi.nlm.nih.gov/pubmed/25595599
9. Amato MG, Salazar A, Hickman TT, et al. Computerized prescriber order entry–related patient safety reports: analysis of 2522 medication errors. J Am Med Inform Assoc. 2017;24:316-322. http://www.ncbi.nlm.nih.gov/pubmed/27678459
10. Hickman TT, Quist AJL, Salazar A, et al. Outpatient CPOE orders discontinued due to 'erroneous entry': prospective survey of prescribers' explanations for errors. BMJ Qual Saf. 2018;27:293-298. http://www.ncbi.nlm.nih.gov/pubmed/28754812
11. Kripalani S, LeFevre F, Phillips CO, Williams MV, Basaviah P, Baker DW. Deficits in communication and information transfer between hospital-based and primary care physicians: implications for patient safety and continuity of care. JAMA. 2007;297:831-841. http://www.ncbi.nlm.nih.gov/pubmed/17327525
12. Ai A, Wong A, Amato M, Wright A. Communication failure: analysis of prescribers' use of an internal free-text field on electronic prescriptions. J Am Med Inform Assoc. 2018;25:709-714. http://www.ncbi.nlm.nih.gov/pubmed/29474659
13. Dhavle AA, Yang Y, Rupp MT, Singh H, Ward-Charlerie S, Ruiz J. Analysis of prescribers' notes in electronic prescriptions in ambulatory practice. JAMA Intern Med. 2016;176:463-470. http://www.ncbi.nlm.nih.gov/pubmed/26954486
14. SCRIPT Implementation Recommendations. Scottsdale, AZ: National Council for Prescription Drug Programs; June 2015. Available at https://www.ncpdp.org/NCPDP/media/pdf/SCRIPT-Implementation-Recommendat…
15. Gaunt MJ. Caution when prescribing Xarelto to first-time patients. Pharmacy Times. October 14, 2015. Available at https://www.pharmacytimes.com/publications/issue/2015/october2015/cauti…
16. Chu A, Limberg J. Rivaroxaban program for acute venous thromboembolism upon ED discharge, with focus on utility of commercially available dose pack. Am J Emerg Med. 2017;35:1910-1914. https://www.ncbi.nlm.nih.gov/pubmed/28869100
17. Lowenstern A, Al-Khatib SM, Sharan L, et al. Interventions for preventing thromboembolic events in patients with atrial fibrillation: a systematic review. Ann Intern Med. 2018;169:774-787. https://www.ncbi.nlm.nih.gov/pubmed/30383133
18. Schiff G, Seoane-Vazquez E, Wright A. Incorporating indications into medication ordering—time to enter the age of reason. N Engl J Med. 2016;375:306-309. http://www.ncbi.nlm.nih.gov/pubmed/27464201
19. Garabedian PM, Wright A, Newbury I, et al. Comparison of a prototype for indications-based prescribing with 2 commercial prescribing systems. JAMA Netw Open. 2019;2:e191514. http://www.ncbi.nlm.nih.gov/pubmed/30924903
20. Kuderer NM, Lyman GH. Guidelines for treatment and prevention of venous thromboembolism among patients with cancer. Thromb Res. 2014;133(suppl 2):S122-S127. https://www.ncbi.nlm.nih.gov/pubmed/24862132