Handing off electronic prescription data from prescribers to community pharmacies: A qualitative analysis of pharmacy staff perspectives

Abstract

Objectives

Understand the process of how community pharmacies handle electronic prescriptions (e-prescriptions) and learn about different errors or potential errors encountered.

Methods

Fifteen remote, semi-structured interviews were conducted with community pharmacy staff. Interview analysis was done with two adapted Systems Engineering Initiative for Patient Safety (SEIPS) methods to understand the workflow and an affinity wall, which led to key words that were tallied to understand the frequency of different issues.

Results

Data entry in community pharmacies is a process that varies based on the different software platforms receiving e-prescriptions. Data entry of a medication product is typically a human reliant process matching an e-prescription with an equivalent medication product. Current automated safety supports focus on matching the dispensed medication to the medication chosen at data entry. Substitutions may be required for a variety of reasons, however, pharmacists’ comfort and permissions in doing so without provider involvement fluctuates.

Conclusions

Prescription errors remain that could be prevented with additional support at the data entry step of e-prescriptions. Few studies demonstrate where these errors originate and what role current technology plays in contributing to or preventing these errors. Future work must consider how these matches between prescribed medications and pharmacy fulfilled medications occur. There is a need to identify potential tools to support data entry and prevent medication errors.

Introduction

Although electronic prescriptions (e-prescriptions) improved patient safety in some ways, they introduced new errors that can cause significant patient harm.^1,2 Despite eliminating handwriting errors from paper prescriptions or transcription errors from verbal prescriptions, new types of errors can occur when implementing electronic sources of medication information for prescribing and dispensing.^1–6 One study found errors pertaining to dispensing the wrong dose, strength, or dose form occur in approximately half of a percent of e-prescriptions.⁷ Given the staggering costs of medical errors, potential harm to patients, and the more than 2 billion e-prescriptions dispensed each year⁸, it is imperative to understand the sources of potential e-prescription errors, how pharmacies receive and handle the information from the e-prescription and what tools are in place to safely facilitate this process.^9,10 Additionally, handoffs from the prescriber to the pharmacy via e-prescription can result in questions and delays when the pharmacy dispensing software cannot automatically process the prescription or the prescription details are not standardized.^11,12

Automated fulfillment aims to prevent data entry errors by eliminating manual selection and increasing the success rate of selecting the prescribed medication product.¹³ Recent evidence suggests 20% of automated data entry failed to appropriately select the right medication product in a study of two pharmacies.¹³ Thus, manual selection occurs regularly, increasing the risk of incorrectly selecting a medication product that is a different ingredient, strength, or dose form than what the prescriber intended for the patient. ¹⁴

While automated safety tools at the product fulfillment stage such as barcode scanning, image detection, and clinical decision support during the drug utilization review (DUR) reduce or detect medication errors, they do not guarantee a safe and accurate medication is dispensed.^15–17 These safety tools rely on matching the dispensed medication to what was previously selected at the data entry stage of filling an e-prescription. If the wrong medication is selected at data entry, none of these tools would alert pharmacy staff to the problem. Little is known about the tools available to support the data entry of e-prescriptions in community retail pharmacies. Previous studies have considered the handoff of information from pharmacist to pharmacist, but less is known about handoff of information through the e-prescription from prescriber to pharmacist with their respective software systems.¹² The objective of this study is to establish the process of handing off e-prescription data to pharmacies and identify potential sources of errors.

Methods

This study performed a qualitative analysis of pharmacy staff perspectives on processing e-prescriptions in retail community pharmacies. Interview guides were created for semi-structured interviews with pharmacists and pharmacy technicians to understand the process for filling an e-prescription. Topics included safety mechanisms, errors for which staff are especially vigilant, common errors, and professional expertise when considering alerts for different substitutions and mismatches of medication products. This research was exempt from IRB oversight by the University of Michigan Institutional Review Board.

The interview guides were structured with five groups of questions. All questions were discussed with a pharmacist in the draft stage to improve the relevance and clarity of the questions. The first group of questions solicited details about participants’ current job role, years of experience, and their major responsibilities. The second group of questions collected information on the current e-prescription process. The goal was to follow an e-prescription from the time it entered the computer system through to fulfillment; looking at how the e-prescription arrived, what steps were automated versus human-reliant, safeguards, and medication product selection. The third set of questions considered errors within the current e-prescription system. We sought to understand common errors, concerning errors, how staff became aware of errors, and what could be done to prevent the repetition of errors. The final two sections elicited feedback on a safety mechanism alerting community pharmacy staff to potential errors when comparing the medication prescribed versus that dispensed.

We shared an overview of the workflow of our concept (Figure 1) to give participants an understanding of how we hoped to use their experiences to improve future support. We discussed three examples of substitution types that are made when dispensing medication products: changes in product ingredients, strengths and dose forms. We discussed dispensing Warfarin as 5 mg vs 1 mg and changing the strength/quantity to allow dosage flexibility (Table 1).

Figure 1.

Proposed Automated Safety Mechanism Workflow

Workflow diagram mapping the concept for an automated tool deciding appropriateness of matches between e-prescriptions and pharmacy medication product selections.

Table 1:

Scenarios shared during the interviews to solicit feedback and thoughts on substitutions.

Scenarios Shared in the Interview
Ingredient Mismatch	Allopurinol 300 MG Oral Tablet (NDC 63304054005) as the prescribed product and Metformin Hydrochloride 500 MG Oral Tablet (NDC 23155010210) as the dispensed product.
Brand to Generic Mismatch	Januvia 100 MG as the prescribed product and SITagliptin 100 MG Oral Tablet as the dispensed product. Januvia’s SBC CUI is 665036 but the generic SITagliptin’s SCD CUI is 665033.
Form Mismatch Salt Mismatch	Hydroxyzine: hydrOXYzine hydrochloride 25 mg as the prescribed product (SCD CUI 995269) and hydrOXYzine pamoate 25 mg (SCB CUI 995253). One is an oral tablet, and the other is an oral capsule.
Quantity/Strength Mismatch	Warfarin 5mg as the prescribed product and Warfarin 1 mg (x5 quantity) as the dispensed product.

Participants

Recruitment fliers were distributed to the University of Wisconsin School of Pharmacy ‘s Pharmacy Practice Enhancement and Action Research Link Network, the University of Michigan College of Pharmacy Preceptor Network, and the University of Minnesota Pharmacy Practice Based Research Network. 28 potential participants emailed the study team, a screening call was completed with 19 individuals, and interviews were scheduled and completed with 15 eligible participants. Interested participants were eligible if they worked in a community pharmacy setting and had at least a year of experience dispensing medication from e-prescriptions. This recruitment process began in January 2023 and all interviews were completed by the end of April 2023.

Data Collection & Analysis

Interviews were recorded on Zoom and were professionally transcribed. The transcripts were verified by the interviewer for accuracy. Affinity walls use square notes of quotes or observations from qualitative interviews that are organized by themes among all interviews to find common issues and potential pain points.^18,19 The affinity wall organized interview data into clusters which led to themes (Figure 2). From these themes, keywords were identified and the number of interviewees that mentioned these keywords were tabulated (Figure 3 and Appendix). Adapting two Systems Engineering Initiative for Patient Safety (SEIPS) methods, Tool 1: People Environment Tools and Tasks Scan and Tool 5: Journey Map, individual workflow diagrams from each interview were created that modeled the route an e-prescription takes through the fulfillment process.²⁰ These individual workflows were analyzed and summarized in a single model within the results section (Figure 4).

Figure 2.

An Overview and Snapshot of One Section of the Affinity Wall

This image shows the general organization of blue sticky notes as data points from the interviews clustered by theme (dark pink sticky notes). One section has been zoomed in on to illustrate a sample theme, data entry.

Figure 3.

Graphs Tallying the Frequency of Keywords from Interview Sections

The blue graph shows the keywords mentioned when discussing the arrival of the e-prescription to the pharmacy software platform, the green graph shows the keywords mentioned when discussing the critical aspects of fulfilling and e-prescription, and the red graph shows the keywords mentioned when discussing common substitutions or medications that require more awareness. The appendix contains quotes to add context to each bar of the graphs from Figure 3.

Figure 4.

Pharmacy Workflow

A compilation of the different workflows that community pharmacies worked through to fill e-prescriptions.

Results

Participant characteristics

Fourteen pharmacists and one pharmacy technician were interviewed. Years of pharmaceutical dispensing experience ranged from 3 to 40 years with a mean of 18.3 years. Of the pharmacists, nine are pharmacy owners, managers, supervisors or pharmacists in charge of their location. Participants worked in five independent pharmacies, four large chain pharmacies, four health system pharmacies, and three long term care pharmacies. There were seven different dispensing systems used by the pharmacies interviewed, which are used throughout the United States in community pharmacies. One pharmacist worked at both a health system and a large chain pharmacy. Thirteen participants identified as white, one as Asian, and one as Black or African American. Eight participants identified as male and seven as female. Participant ages ranged from 26 to 63 years with a mean age of 40.9 years.

The Process/Problem:

E-prescriptions arrive in community pharmacies in inboxes or queues. The fulfillment process varies by pharmacy, pharmacy software platform, and state regulations. A common scenario is that pharmacy staff see a split screen with the e-prescription data on one side and the pharmacy dispensing software on the other side. One pharmacist (Female, 50 years old) reported, “I go back and forth between the e-script and the documentation that we typed in.” Selection and match validation begins with identifying the patient and then prescriber. Different software platforms offer various support to the pharmacy staff member as the patient is selected. These supports include limiting options to patients living within a specified distance, prioritizing previous patient names as matches, percentage matches based on all patient demographics, and color-coding information fields. An exception was discovered in the interviews; “closed loop” pharmacies only receive prescriptions from local healthcare providers that use the same electronic health records (EHR) platform. Only the closed loop pharmacies demonstrated automated patient selection without needing oversight from a pharmacy staff member due to a single patient database for prescribing and dispensing. Closed loop pharmacies are unique environments where the pharmacy only fills prescriptions from providers within the same healthcare organization, all using an integrated software system from a single vendor.

The same is true for matching the prescribed medication product from the e-prescription to a medication product in the pharmacy for fulfillment. In some instances, community pharmacy dispensing software auto-populates portions of the medication product information from the e-prescription into the pharmacy software, but most of the time, a staff member must make the selection by typing in the prescribed product. “When the data entry technician is entering it, they can search our active products in our formulary and pick something, but there’s nothing in the software that helps,” shared a pharmacist (Female, 38). Another pharmacist (Male, 48) reported, “Either a technician or myself, the pharmacist, will process the prescription, which means linking the information on that script to a real-life drug.” This selection could be from a narrowed list of medication products based on inputs from the pharmacy platform’s analysis of the e-prescription or inventory. Other times this selection provides a menu bar that offers options once the pharmacist begins to type in the medication product information. “It does not try and auto-populate the drug, which I always thought is interesting. Even when we do have the active NDC that the doctor prescribed on the e-prescription. It doesn’t try and fill in the drug. So it forces us to pick the drug every single time. It just leaves it blank. I typically, if I was [going to] enter ciprofloxacin for someone, I would type in CIP,250 to narrow my search results down,” another pharmacist (Female, 30) shared.

Another example shared is that the e-prescription will appear on half of the screen and the pharmacy staff must manually enter all of the data into their software platform by typing in the patient, prescriber, and medication information. A pharmacy manager (Female, 32) explained, “You just go into the line item and then there’s two screens. Left screen is the data entry field, where you enter everything. And then the right-hand side is the image of the electronic prescription. So, essentially you just match it and then you type everything in as ordered and then you push it through.”

Meanwhile, a director of retail pharmacy (Male, 32) at a closed loop pharmacy reported that, “So if I get it from [the same e-prescribing software] customer, I don’t actually have to enter the prescription at all. It pre-enters everything for me. If it’s coming from a [different e-prescribing software], [(the e-prescription network) and (dispensing software) will communicate and they’ll try to pre-enter, but I’ll have to come behind and clean up more than what was anticipated.”

Safety Mechanisms

Numerous important safety mechanisms are built into the process of filling a prescription. A trained staff member is the first safety mechanism, applying their experience and training to make the appropriate medication selection based on the e-prescription information. Communication between pharmacy staff becomes a secondary check if there is any concern about what is the appropriate medication to fulfill the e-prescription. As one pharmacist (Male, 59) shared, “I have technicians who are senior techs, who have been working with me for 15 years or more, and so I rely on their... They have expertise, they bring a lot of experience, and so I rely on their wisdom.”

Community pharmacies primarily rely on humans to double-check the work of data entry without much support from computerized tools. One human-reliant mechanism is at the pharmacist verification step as the pharmacist considers if the correct medication product has been selected. In some states, such as Minnesota, this is a separate verification step from verifying that the prescribed medication matches what has been chosen from the shelf, but in others it can be a combined step in the process of both medication selection and product fulfillment.

Individual customization settings and techniques of retrieving the medication off the shelf before entering the data are human-reliant mechanisms to help the staff choose the correct medication. One staff member shared (Female, 30), “Again, super customizable. Every technician here can choose to see more or less information.” A staff pharmacist (Male, 63) shared, “They might grab the bottle off the shelf and enter in the National Drug Code (NDC) numbers because it’s faster to get an actual product than to search alphanumerically across multiple different manufacturers.”

Post data entry, all pharmacy staff shared examples of safety mechanisms that are part of the fulfillment process. These included clinical decision supports: allergy alerts, drug-drug interactions, and graphs visualizing normal dose ranges versus the prescribed strength. Optical counting machines, barcode scanners, and optical medication scanning programs help verify that the medications selected off the shelf match what was entered.

Medication Product Substitutions

Pharmacy staff might substitute a similar medication for the written prescription. It is within a pharmacist’s domain and professional capacity to do so. Reasons shared for making substitutions were medication shortages, availability of inventory, flexibility in dosing, and insurance approvals. Closed loop pharmacies allow pharmacists more leeway to make changes directly to the e-prescription data than in other practices where one organization is prescribing and another is dispensing. A pharmacist using a closed loop software platform (Male, 32) explained that “I am the prescriber and the pharmacy in that instance. So I can adjust anything that I need to on the prescription because I am also the providing entity.” The medication product that is listed as prescribed automatically becomes what is dispensed and all changes to the e-prescription are tracked in the EHR.

Pharmacists in non-closed loop settings had a variety of responses to when and if they felt comfortable making substitutions. Some pharmacists shared that to make a change to an e-prescription, they must speak with the prescriber’s office. “If we’re changing anything, we’re looping the doctor in and we’re not taking any liberties” and another went on to clarify that, “I technically legally cannot substitute capsules versus tablets” (Female, 50). Twelve interviewees (80%) mentioned an example where they would make a substitution without prior approval or a new prescription from the provider. A common substitution involved dose forms: tablets to capsules, liquids to tablets, etc. Pharmacies may adjust the quantity and strength based on stock availability or insurance requirements. The resulting changes are equivalent to the original prescription. “If you can’t get the 20 milligram tablets or capsules, but you got plenty of the tens, yes, it’s not uncommon to either overwrite or make a computerized note that said, use the tens and then adjust the ‘take one’ to ‘take two’ and adjust the final quantity. Yeah, that’s another example of where no, I would not contact the physician to get an okay for that. That’s silly.” (Male, 63).

Pharmacists frequently mentioned the need to take care of their patients by getting their treatment started promptly versus waiting for the prescriber to approve the substitution. The same pharmacist continued, “If I call ‘em, it might make it easier, but if it’s a weekend and it’s hard or it’s in the evening and I can’t, I’d rather dispense a product to a patient now rather than make them wait until tomorrow to get what I think is not gonna be a problem.” Another pharmacist similarly stated, “But say it’s unavailable, it’s a weekend or something, if I need to [I’ll] get it started. And then notify the office the next week” (Female, 26). Warfarin, a medication used to prevent blood clots, is an example of a medication where some pharmacists prefer to adjust the strength/quantity to allow for easy patient modification. A pharmacist (Female, 30) explained, “We are typically going to use a 1mg tablet even though the 2mg was prescribed, so that if all of a sudden they get an INR and the doctor goes, ‘Oh, wait, their INR was a little high. Let’s do 1 mg Monday, Tuesday, Wednesday, but 1.5 mg the rest of the week,’ we can more easily make that adjustment for the patient without having to change out the tablet strength itself.”

Common Selection Errors

The pharmacist staff shared examples of medications that are easily mistaken for other medications (Table 2). Scenarios include medications that are the same ingredient with different strengths, medications that are the correct ingredient and strength but a different release form, medications that have very similar spellings or tall man lettering, or dealing with options on the drop-down menus where the length of the information about the medication is so long that a critical component is cut off from view.

Table 2:

Examples of issues that were shared by pharmacists and pharmacy technicians interviewed.

Examples of Common Errors and Concern for Errors
Scenario Type	Description of Scenario	Examples Provided by Pharmacists
Correct Ingredient, Wrong Strength	The ingredient in the medication is the same in both the prescribed drug product and the filled drug product, but the strength is different.	Dextroamphetamine, which could be filled as Adderall (i.e., mixed amphetamine salts) or Dexedrine (i.e., pure dextroamphetamine).
Correct Ingredient and Strength, Wrong Release Form	The ingredient and strength of the medication is the same in both the prescribed drug product and the filled drug, but the release forms are different.	“Adderall XR vs Adderall immediate release will happen.”
Correct Medication and Strength, Wrong Form	The correct ingredient and strength but the form is different.	Ophthalmic (eye) vs OTIC (ear) Creams vs Ointments Tablets vs Capsules
Correct Ingredient and Release Form, Wrong Strength	The ingredient and release form are correct, but the strength is different.	“Verapamil ER 240 vs Verapamil ER 360”
Wrong Ingredient	The drug name looks or sounds like another drug and is mis-selected.	Hydroxyzine and Hydralazine, Ropinirole and Risperdal, or Fluvoxamine and Fluoxetine

Other medications are known to need a more thorough human check because of their potency or potential to cause serious harm to a patient. These medications include narcotics, antipsychotics, anti-seizure medications, methotrexate, insulin, and anticoagulants. Pharmacists are careful to double check high-price medications due to the financial implications of dispensing the incorrect amount. Some pharmacies have added safety stops in their pharmacy dispensing software to trigger an extra verification step by the pharmacist when filling a prescription with these medications.

Discussion

With more than 2 billion e-prescriptions filled each year in the U.S., even a small percentage of errors is significant.⁸ Much work has been done to increase the accuracy in product fulfillment (e.g., optical scanning, barcode scanning) but less has been done at the data entry phase. A multitude of platforms exist for sending and receiving e-prescriptions. Without a standardized system or requirements in place amongst all parties involved in the e-prescription process, there remains a need for automated tools or safety mechanisms to prevent errors at this step. While no universal software platform exists for e-prescriptions, the National Council for Prescription Drug Program (NCPDP) developed standards for e-prescription information transmission.²¹ The NCPDP’s SCRIPT standard for e-prescribing requires the inclusion of drug product descriptions and NDCs. However, SCRIPT does not require that drug compendia are used to populate these data fields, resulting in each organization creating drug product records on e-prescriptions.

E-prescriptions arrive with varying levels of details to pharmacy platforms that process the data in different ways. Most often, data entry is not fully automated and is human reliant to select the medication to dispense. Some platforms link the dispensed medication to a drop-down menu of suggested medications in stock at the pharmacy. Unfortunately, these drop-down menus can be limited in character length, cutting off details about the product from staff view. Currently, automated safety efforts are focused on appropriate product selection after the data entry has occurred. All pharmacies have clinical safety processes as part of the Drug Utilization Review, which provides screening for drug allergies, duplicate therapies, drug-drug interactions and more. Each pharmacy used barcode scanning to verify that dispensed items matched the selected medication. Others had additional automated tools for optical scans, dispensing, and counters. All of these automated safety mechanisms were checking the fulfillment against the medication selected during the data entry process. Tools like barcode and optical scanners are matching the fulfilling medication to what was entered at the data entry point and are only accurate if the data entered was accurate. This creates an opening for potential errors to be missed and can lead to adverse events.

Pharmacists and pharmacy staff are irreplaceable experts in the field. This study found that human reliance varies by e-prescription, pharmacy platform, organization, and state. Different organizations split the verification into separate data entry and product fulfillment verification steps based on staffing or state regulations. Separating these steps can reduce the amount of cognitive work done at a single point, which may improve patient safety. However, expertise is not without limits or lapses and why automated safety tools in the medication dispensing process exist. The pharmacist verification step(s) aims to recover from errors at data entry, but is not failsafe due to high workloads, fatigue, and limitations of human cognition.^6,22–24 One technique to support human selection of the correct medication is the use of Tall Man lettering to differentiate look-alike, sound-alike medications by capitalizing letters in the text (e.g., hydrOXYzine vs. hydrALAzine).²⁵ Even when platforms attempt to completely automate data entry, errors still can occur, as noted in a study at a rate of 20%.¹³ A human still must verify what the automation has entered.

Additionally, pharmacy staff have the autonomy to make substitutions. Substitutions may be necessary when a medication is out of stock, the medication form is denied by insurance, or a similar form is less expensive. Different states and organizations provide varying levels of support for their pharmacists to do so without contacting the provider. Waiting for a new prescription from the prescriber can take time that prevents prompt treatment. Pharmacists wish to help patients begin their medication therapy and prevent delays in care. In light of this, dispensing software platforms must be flexible enough to allow for necessary product substitutions when the prescribed medication product is not the most appropriate.

Finally, consideration should also be given to studying how the e-prescription process could be updated and standardized to eliminate ambiguity between medications, prescriptions for out-of-date medications, and how software platforms process the information on the e-prescriptions. Using tools like Natural Language Processing, Machine Learning, and leveraging National Drug Databases, an automated tool could be built to support pharmacy staff at data entry. Strengthening or standardizing the requirements of how and what information comes from the e-prescription to create a seamless handoff, a normalized interoperability, of the prescription to the pharmacy platform could also prevent data entry errors.

Limitations

These findings are limited by the small pool of pharmacists that were interviewed, though major themes still emerged. Interviewees were geographically limited to pharmacies in Michigan, Minnesota, and Wisconsin. Their experiences may not be representative of pharmacies on a national scale due to differences in pharmacy practice norms and state-level regulations. Information gathered was reliant on pharmacist/technician understanding of how their software works. It would be valuable to learn more about the components of linking information from software developers and conduct a formal evaluation of software platforms.

Conclusion

There is a gap in automated safety support in community pharmacies at the data entry level of processing e-prescriptions. Current tools support matching the product dispensed to the medication entered from the e-prescription, but leave room for introducing errors during product selection. Knowing that the right medication was selected at data entry further strengthens the other safety tools used throughout the rest of the fulfillment process.