Implementation fidelity of patient-centered prescription label to promote opioid safe use

Abstract

Purpose:

Patient-centered labels may improve safe medication use, but implementation challenges limit use. We assessed implementation of a patient-centered “PRN” (as needed) label entitled “Take-Wait-Stop” (TWS) with three deconstructed steps replacing traditional wording.

Methods:

As part of a larger investigation, patients received TWS prescriptions (eg, Take: 1 pill if you have pain; Wait: at least 4 h before taking again; Stop: do not take more than 6 pills in 24 h). Prescriptions labels recorded at follow-up were classified into three categories: (1) one-step wording (Take 1 pill every 4 h [without daily limits]), (2) two-step wording (Take 1 pill every 4 h; do not exceed 6 pills/day), and (3) three-step wording. There were three subtypes of three-step wording: (3a) three-step, not TWS (three deconstructed steps, not necessarily TWS wording), (3b) TWS format, employing three steps with leading verbs, but “with additions or replacements” (eg, replaced “do not take” with “do not exceed”), and (3c) verbatim TWS.

Results:

Two hundred eleven participants completed follow-up. Mean age was 44.3 years (SD 14.3); 44% were male. One-step bottles represented 12% (n = 25) of the sample, whereas 26% (n = 55) had two-step wording. The majority (44%, n = 93) had three-deconstructed steps, not TWS (3a); 16% (n = 34) retained TWS structure, but not verbatim (3b). Only 2% (n = 4) displayed verbatim TWS wording (3c). All category three labels (utilizing deconstructed instructions) were considered adequate implementation (62%).

Conclusions:

Exact intervention adherence was not achieved in the majority of cases, limiting impact. Nonetheless, community pharmacies were responsive to new instructions, but higher implementation reliability requires additional supports.

1 |. INTRODUCTION

Challenges with prescription drug labeling are a major contributor to adverse drug events.¹ Complex medication label instructions and low patient health literacy contribute to patient misunderstanding.^2–6 Over a decade ago, a systematic review of prescription labeling “best practices” proposed standards for labeling; these standards were later summarized in an Institute of Medicine (IOM) workshop.^7,8 More recently, a systematic review by Bailey and colleagues reinforced previous findings about the importance of clear labeling, while also calling for additional assessments of labeling in actual use rather than in hypothetical experiments.⁵

Very few studies evaluating the impact of prescription labeling have examined interventions in actual use, rather than in hypothetical scenarios. In 2005, the Target chain of pharmacies changed their label format; two manuscripts describing that system-level change were published.^9,10 In 2016, Wolf and colleagues published the first randomized controlled trial of a patient-centered label using the Universal Medication Schedule (UMS) (one of the IOM recommended best practices) in actual use rather than in hypothetical “demonstrated” scenarios.¹¹ This trial capitalized on a unique clinical setting with a central-fill pharmacy serving eight local clinics. By implementing the intervention at the level of the pharmacy management software and pharmacy label-printing machine, the UMS prescription label was delivered with high reliability.

To our knowledge, no studies have evaluated the reliability of community pharmacies implementing suggested transcription of UMS prescription wording or other patient-centered label changes in the absence of a pharmacy-level intervention. Therefore, it is unknown how patient-centered instructions may be translated onto actual labels in a busy pharmacy practice without system-level supports. This question is particularly salient in the context of opioid prescriptions. As with all sectors of health care right now, inpatient and outpatient pharmacists are trying to figure out ways to address the opioid crisis and prevent opioid-related deaths.

As part of a pragmatic trial to improve patient knowledge and safe use of opioids, we implemented a new patient-centered label change entitled Take-Wait-Stop (TWS). The objective of this change was to improve drug instruction clarity for the most commonly prescribed opioid (hydrocodone-acetaminophen). This label employs three explicit, deconstructed steps to replace traditional one-step “take as needed” wording (ie, PRN, which is the Latin abbreviation for “pro re nata,” meaning “as the circumstance arises”). In this analysis, we evaluate community pharmacy adherence in translating the patient-centered wording on the electronically generated print prescription to the bottle label.

2 |. METHODS

2.1 |. Study setting and participants

This study was a planned analysis using data from a three-arm randomized controlled trial¹² investigating a multifaceted intervention to improve patients’ understanding and safe use of prescription opioid pain relievers. Patients were enrolled at an urban academic emergency department (ED) (>88 000 annual visits) in Chicago, Illinois, and were approached by research assistants if they were being discharged with a new prescription for hydrocodone-acetaminophen. Eligible patients were aged 18 years or older, were English speaking, and had not taken hydrocodone-acetaminophen on a regular basis within the last 90 days, on the basis of self-report. Patients were randomized, on the basis of their prescribing physician, to receive standard of care prescription wording and medication information (control arm) or additional medication information and TWS prescription wording (two intervention arms). The additional medication information was in the form of a one-page MedSheet, and a subset of intervention patients also received text messages with safe use information for the week following their ED visit. Patients were included in this analysis if they were assigned to one of the two intervention arms, which both received the TWS label, completed a follow-up interview 7 to 14 days after enrollment, and had their pill bottle available at the time of the assessment. Follow-up and data collection (including pill bottle label data) were similarly collected for control arm participants but were not included in this manuscript because of the focus on patient-centered labeling. Full methods and trial design have been detailed previously.¹²

2.2 |. Take-Wait-Stop label

Members of our team first developed the TWS label as an extension of the UMS,^13,14 following tenets of patient-centered prescription label design.^15–18 The prescription wording intentionally deconstructs the core components of PRN instructions in an effort to more explicitly convey the dose, interval between doses, and maximum daily dose that is not to be exceeded. The wording places emphasis on action terms (TWS) to deconstruct each behavioral step and to support understanding and recall. Additionally, numeric characters are employed instead of words (eg, “1 tab” instead of “one tab”), and “carriage returns” place each section of the instructions on different lines.¹⁸ Simplified text and plain language were used to convey maximum daily dosing. The word “Stop” was used to replace the typical wording “Do not exceed” on the basis of prior qualitative research among patients with limited literacy who found the word “Stop” easier to understand and pronounce than the word “Exceed.”¹⁹

Patients in the intervention arms received the TWS prescription wording on a print prescription requisition. The electronic health record (EHR) was modified for the study to automate the TWS wording on the print prescription, so all prescriptions were generated uniformly and did not require manual entry by the physicians. The automation simply edited the prescription to the TWS format after the physician had chosen the medication strength and frequency. The “Stop” component of the instructions then was preprogrammed on the basis of the dosing interval chosen by the physician. For example, if a physician ordered 5-mg hydrocodone-acetaminophen every 4 hours, the “Stop” criteria would auto-populate as follows: “Stop: Do not take more than 6 pills in 24 hours.”

The study team anticipated the prescription wording might be unfamiliar to pharmacists in the community; thus, three additional steps were taken to increase likelihood of successful translation of the prescription onto the label. First, prior to finalizing the wording for the TWS prescription, the study team worked with a national community pharmacy chain manager to ensure verbatim translation (with carriage returns and formatting maintained) was feasible on a standard label. Second, the default Sig (ie, signetur, which is Latin for “let it be labeled”) on the print prescriptions was changed to say “Special Sig” to draw the dispensing pharmacists attention to the wording change. Third, every print prescription also contained an additional line, which read “Note to Pharmacist: Please print the ‘Take-Wait-Stop’ instructions on the medication label.” The health system did not support e-prescribing of controlled substances from the ED at the time of the study; thus, all prescriptions were printed on tamperproof paper. It was noted that traditional “quick codes” that allow pharmacists keystroke shortcuts when transcribing common prescriptions were not available in the pharmacy software for the TWS instruction.

2.3 |. Study protocol and variables

Sociodemographic variables including age, race, gender, educational attainment, household income, insurance status, self-reported health status, and prior exposure to opioids were obtained via patient self-report at the time of enrollment. Health literacy level was measured using the Newest Vital Sign (NVS).²⁰ The research assistant transcribed the prescription Sig as well as details of the pharmacy where the prescription was filled, during the 7- to 14-day follow-up interview. Because of follow-up data being collected via telephone, we believed we could not reliably capture the presence of “carriage returns,” so the spacing of the label was not evaluated. All data were captured in REDCap (Research Electronic Data Capture), a secure, web-based application for building and managing online data capture for research studies.²¹

After reviewing the final data, study authors met to discuss the types of deviations from the verbatim instructions. Trends in deviations were identified, and authors agreed upon five distinct categories. After category identification, two authors (M.R.E. and A.M.R.) coded each label by category; a third author (D.M.M) adjudicated discrepancies.

2.4 |. Analysis

Labels were categorized into one of three categories: (1) one-step wording (Take 1 pill every 4 h [without maximum daily dose]), (2) two-step wording (eg, Take 1 pill every 4 h; do not exceed 6 pills in a day), and (3) three-step labels. Within the three-step labels, there were three subtypes: (3a) three-step, not TWS (three deconstructed steps, not necessarily using “take” “wait” “stop” wording), (3b) TWS format, employing three steps with leading verbs, but “with additions or replacements” (eg, added “by mouth,” replaced “pills” with “tablets,” and replaced “do not take” with “do not exceed”), and (3c) verbatim TWS (Table 1). Frequency of each label type and examples of wording deviations are reported. Two additional composite end points are reported: labels with three discrete steps (labels in categories 3a, 3b, and 3c) and labels including maximum daily dosage information (labels in categories 2, 3a, 3b, and 3c). Additional label variables analyzed included the frequency of different words used to describe the medication (ie, pill versus tablet versus tab), the frequency of the phrase “by mouth,” the rate of use of numerals versus written numbers, and the accuracy of the prescription technical details (eg, number of pills and interval) to the original prescription. Analyses were performed using SAS 9.4 (SAS Institute Inc, Cary, North Carolina).

TABLE 1.

Label categories

Label Type	Major Changes from Study Label
(1) One-step	● “every” instead of “wait” ● dosing interval not a deconstructed step ● missing “Stop” ● missing maximum daily dose information
(2) Two-step	● “every” instead of “wait” ● dosing interval not a deconstructed step ● missing “Stop” but retains maximum daily dose information
(3) Three-step
(3a) Three-step, not TWS	● three distinct action steps, often missing word “Stop,” but retains maximum daily dose information
(3b) TWS with additions or replacements	● added “by mouth” ● replaced “pills” with “tabs” or “tablets”
(3c) TWS verbatim	Verbatim

Abbreviation: TWS, Take-Wait-Stop.

3 |. RESULTS

Of the 652 participants enrolled in the larger study, 450 participants were enrolled in the intervention arms. Among the 450 intervention arm participants, 248 completed the 7- to 14-day follow-up assessment and were eligible for this analysis. Of those 248, 211 (85.1%) participants had their prescription bottle present and able to be transcribed during the assessment. The mean age of the participants was 44.3 years (SD 14.3), and 43.6% were male (Table 2). The 211 prescription bottles evaluated in this sample were filled at a total of 118 distinct community pharmacy locations, representing seven community pharmacy retail chains. The majority of prescription fills occurred at two major retail pharmacies, referred to hereafter as Retail Chain A and Retail Chain B. A total of 151 individuals filled their prescription at 67 unique locations of Retail Chain A, and 46 individuals filled their prescription at 37 locations of Retail Chain B. The remaining 14 participants filled their prescriptions at 14 locations representing six additional pharmacies (only one of which was not a retail chain pharmacy).

TABLE 2.

Participant demographic, ED visit, and prescription characteristics

Characteristic	N = 211
Demographic characteristics
Age, mean years (SD)	44.3 (14.3)
Male gender, %	43.6
Race, %
White	51.2
African American	27.0
Other	21.8
Education, %
High school grad or less	12.8
Some college	27.0
College graduate	35.1
Graduate degree	25.1
Income level, %
≤$40 000	23.3
>$40 000-$100 000	31.6
>$100 000	45.1
Health literacy, %
Low + marginal	31.2
Adequate	68.7
Primary insurance, %
Medicaid	13.9
Medicare	8.1
Private/managed care	71.8
Self or no insurance	3.4
Other	2.9
Self-reported health status, %
Excellent	15.2
Very good	37.9
Good	33.2
Fair	11.4
Poor	2.4
Previously prescribed hydrocodone, %	37.1
ED visit characteristics
Triage acuity, %
1 and 2	8.7
3	55.6
4 and 5	35.8
Triage pain score, mean (SD)	7.7 (2.3)
Total length of stay (h), median (IQR)	3.9 (2.9-5.2)
Exposure to opioids in the ED, %	87.2
Opioid prescription characteristics
Daily MME prescribed, mean (SD)	30.7 (13.3)
Tabs of opioid prescribed, mean (SD)	15.0 (6.9)

Abbreviations: ED, emergency department; IQR, interquartile range; MME, Morphine Milligram Equivalents.

The majority of the bottle labels (44%, n = 93) had three-step wording, not TWS (3a) (Table 3). The next most common form of labeling was two-step wording with 26% of the labels blending the instructions for dose and timing as one task and including maximum dose warnings as a second task. The one-step label wording and the TWS label with additions or replacements were seen with low frequency in this sample, occurring in 12% and 16% of the sample, respectively. The least common prescription wording was the actual, verbatim prescribed wording. The TWS wording only appeared in 2% of the sample. Examples from the pill bottles of different variations of the wording are seen in Table 3. For the composite end points, 62% of labels had three discrete steps (category 3a, 3b, and 3c), and 88% of labels included maximum daily dosage information, albeit in different formats. There were no statistically significant differences between retail pharmacy chains in the rate of filling the prescription with three-step wording (Chain A 64.2% vs Chain B 52.2% three-step, P = 0.14; Chain A 64.2% vs Other Pharmacies 71.4% three-step, P = 0.59; Chain B 52.2% vs Other Pharmacies 71.4% three-step, P = −0.20) (Table 4).

TABLE 3.

Frequency of label type and examples

Label Type	n (%)	Examples
(1) One-step	25 (12%)	• Take 1 tablet by mouth every 4 h as needed for pain. • Take 1 pill by mouth every 6 h as needed for pain. • Take 1 tablet by mouth every 6 h before taking it again. • Take 1 tablet every 6 h.
(2) Two-step	55 (26%)	• Take 1 tablet by mouth every 6 h as needed for pain. Do not exceed more than 4 pills in 24 h. • Take 2 tablets by mouth as needed for pain; wait at least 6 h before taking again. • Take 1 tablet by mouth every 4 h as needed; do not take more than 6 tablets per day. • Take 1 tablet by mouth every 6 h as needed for pain; do not take more than 4 tablets in 24 h.
(3) Three-step	131 (62%)
(3a) Three-step, not TWS	93 (44%)	• Take 1 pill by mouth if you have pain; wait at least 6 h before taking again; do not take more than 4 pills in 24 h. • Take 1 pill if you have pain; wait at least 4 h before taking again; take no more than 6 tabs/24 h. • Take 1 tablet by mouth if you have pain; wait at least 4 h before taking again, no more than 6 a day. • Take 1 tablet if you have pain; wait at least 8 h before taking again; do not take more than 3 tablets in 24 h.
(3b) TWS with additions or replacements	34 (16%)	• Take 1 tablet by mouth if you have pain; wait at least 6 h before taking it again; stop do not take more than 4 pills in 24 h. • Take 1 tablet by mouth if you have pain; wait at least 6 h before taking another tablet; stop and do not take more than 4 tablets in 24 h. • Take 1 tablet if in pain; wait at least 6 h before taking again; stop do not exceed more than 4 tablets in 24 h. • Take 1 pill by mouth if you have pain; wait at least 6 h before taking again; stop do not take more than 4 pills within in 24 h.
(3c) TWS verbatim	4 (2%)	Take: 1 pill if you have pain. Wait: at least 6 h before taking again. Stop: do not take more than 4 pills in 24 h.

Abbreviation: TWS, Take-Wait-Stop.

TABLE 4.

Differences in TWS labels by pharmacy

Label Type	Retail Chain A N = 151 n (%)	Retail Chain B N = 46 n (%)	Other Pharmacies N = 14 n (%)
(1) One-step	20 (13.3)	3 (6.5)	2 (14.3)
(2) Two-step	34 (22.5)	19 (41.3)	2 (14.3)
(3) Three-step	97 (64.2)	24 (52.2)	10 (71.4)
(3a) Three-step, not TWS	63 (41.7)	24 (52.2)	6 (42.9)
(3b) TWS with additions or replacements	30 (19.9)	0 (0.0)	4 (28.6)
(3c) TWS verbatim	4 (2.6)	0 (0.0)	0 (0.0)

Abbreviation: TWS, Take-Wait-Stop.

Only 17 (8%) labels used the term “pills,” as indicated on the TWS print prescription, rather than “tabs” or “tablets.” The majority of bottles 166 (79%) included the phrase “by mouth” in the instructions. Very few bottles 11 (5%) employed written numbers (eg, “one”) rather than numerals on the labels. On two of the bottles, the pharmacist changed the prescription to state “1-2 tablets” rather than the prescribed “1 pill,” and in one instance, the pharmacist changed the dosing interval from the prescribed 6 hours to provide a dosing range of “4-6 hours.”

Demographics and pill bottle label data for participants in the control sample may be seen in Appendix S1. Although not focused on patient-centered labeling, these data are most useful for demonstrating the current state of labeling in the absence of the intervention with 96% of labels in the control arm being labeled with one-step wording.

4 |. DISCUSSION

These data demonstrate lower than expected verbatim adherence to TWS prescription wording at the level of the dispensing pharmacy, however, an overall reasonable implementation of change when looking at the composite outcomes. Although fewer than one in 50 patients received the verbatim wording requested on their print prescription, three in five patients received a version of the label that employed deconstruction of action steps. Numerous studies and high-profile cases of errors from the era of handwritten prescriptions demonstrated errors that occurred at the point of prescription transcription of instructions.²² E-prescribing has been touted as a solution to improve patient safety for many reasons; one of the hypothesized drivers of improvement is eliminating errors related to illegible prescriber handwriting.²³ To our knowledge, this is the first study to demonstrate variability of interpretation of computer-generated instructions and specifically of patient-centered label wording. Notably, this deviation from the requested instructions persisted, despite the TWS prescription being accompanied by a request to label as written.

Patients receiving two-step and one-step wordings represented a combined 38% in this sample. For those patients, who were intended to receive the patient-centered TWS label, the lack of success in filling the label as written resulted in a change from clear directions to those that were less detailed. Although only 2% of the sample received the requested wording, the intervention may have nonetheless positively influenced the wording of the labels. Looking at the TWS with replacements and the three-step wording, 62% combined were recipients of labels that separate the dose from the timing of administration, with inclusion of instructions about maximum daily dose. While not technically adherent to our preferred patient-centered wording, this deconstruction appears to be an improvement over the common default one-step wording. Evaluating the differential impact of the deconstructed steps versus the exact wording was beyond the scope of this manuscript; while we believe deconstruction of any form is most likely an improvement, it could be the case that in some instances, practical construction could lead to additional safety problems.

This low level of verbatim implementation fidelity is likely multifactorial. Our team had hoped to conduct interviews with local pharmacists filling the highest volume of prescriptions; however, we were not permitted to contact the pharmacists by the local community pharmacy chains, resulting in the inability to provide a more in-depth analysis of the barriers to implementation. It is possible that time limitations relating to the need to fully transcribe without the assistance of quick codes made the TWS a workflow interruption that suggested implementing this best practice to be inefficient in pharmacy workflow. Community pharmacies often rely upon preprogrammed computer codes for their most commonly used drugs and sigs. With these quickly typed codes, a pharmacist could quickly type a drug ID code and instruction code (eg, TAB. 6 could automatically populate a label with “Take one tablet every 6 hours” or PRN PA could insert “as needed for pain”). Using these examples, one can imagine that it would be a timely workflow interruption to manually enter the TWS label instructions rather than type “TAB. 6 PRN PA” and be finished with the label.

At first glance, it may appear that pharmacists used quick codes for labels in categories 1, 2, and 3a; however, on review of the quick code lists for Retail Chains A and B (who filled prescriptions for 93% of the sample), neither system has quick codes for “wait,” “at least,” “stop,” “do not take more than,” or “exceed.” While all of the one-step labels and some of the two-step labels were likely generated from quick codes, it appears that other two-step labels (eg, those that utilized “do not take more than” and “24 hours”) required nonquick code instruction entry. We believe that a majority of the pharmacists were taking time to transcribe the special instructions; however, not all were doing so verbatim. In some cases, they were likely deploying the quick codes to start the instructions and then adding or replacing text. Further support for this belief that the majority of pharmacists made changes is the data that 96% of the control arm bottles had one-step labels that match exactly with known quick codes. This information suggests that there may be willingness at the pharmacist level to deviate from the use of quick codes; however, it remains unclear why there was a lack of verbatim transcription.

Other potential barriers are related to familiarity with the TWS wording and data supporting the impact of small wording changes on patient understanding, particularly for those patients with limited literacy. One additional reason for lack of full verbatim adherence may be the addition of medication route (eg, by mouth) to the label, which may have been added by pharmacists in the attempt to improve medication safety or related to regulatory requirements. There are possible options available to mitigate these barriers including increased education about health literacy and prescription labeling or technological supports such as customized prescription codes.

Although we cannot conclusively determine the barriers to implementation and causes of variability in this sample, the findings nonetheless raise concern about the ability to effect change in medication labeling from the “bottom-up.” A white paper published by the National Council for Prescription Drug Programs (NCPDP) in 2013 encouraged prescribers and dispensers to start implementing the UMS into clinical practice; however, acknowledged many challenges.¹⁵ Our data indicate that starting at the level of the prescriber would not be a reliable implementation strategy for the UMS or other patient-centered labeling changes. Additional challenges, as highlighted in the NCPDP paper, relate to implementing an intervention on a larger scale, such as capacity of the industry in light of other initiatives, enabling technology to support effective delivery, cost-effectiveness, and the role of state boards of pharmacy.¹⁵ Although further research is needed to determine the efficacy of the TWS label and to formalize a universal set of standards for prescription labeling, enacting lasting change in prescription drug labeling will likely require a “top-down” approach guided by national- or state-level legislation. The State of California passed legislation implementing many of the UMS best practices and serves as one example of how such change can be implemented.²⁴

This study is not without limitations. Data were collected in a single city in the Midwest and there was a large drop in patient retention at follow-up. Findings may not be generalizable to other locales with different community pharmacy chains; however, the patients in our sample filled their prescriptions at a wide range of pharmacy chains and locations. Prior to the study, the team tested the ability of the TWS wording to fit on the label of the pharmacy chain with the highest market share in the urban area; however, this testing only occurred within one pharmacy chain. We present the demographic data of the patients to describe the population; however, the analysis is more focused on the pill bottle as the unit of analysis, and we do not have access to the demographics of the pharmacist who read and transcribed the instructions. We were not able to conduct follow-up interviews with the pharmacists as originally planned to gain further information on the limitations to implementation. It is possible that the low level of verbatim implementation would have been ameliorated by e-prescribing (rather than the EHR printed prescription); however, even when medications are e-prescribed, the pharmacist often still needs to manually enter medication instructions.²⁵ Future efforts working directly with retail pharmacy chains to change quick codes may result in better implementation.

In a pragmatic trial of a larger intervention promoting safe opioid use, we found that pharmacies infrequently filled the prescription with the exact wording requested; however, the wording on majority of the prescriptions was altered, and we believe represent an improvement over normal prescription wording. There are many hypothesized barriers that may have prevented more full, verbatim implementation, and these findings highlight challenges in testing patient-centered labels in actual practice without central-fill pharmacies. Multiple groups, including the National Academies of Medicine (formerly IOM) and the Food and Drug Administration, are continuing to discuss ways to improve drug labeling and information provided to consumers, both prescription and over the counter.^26,27 These findings underscore the importance of any of their forthcoming recommendations being implemented on a state or national scale, supported by legislation, rather than being in the format of recommendations to providers or dispensers.

Supplementary Material

Additional supporting information may be found online in the Supporting Information section at the end of the article.

KEY POINTS.

• Patient-centered label design has potential to improve patient understanding and safe use of medication, but implementation challenges may limit use.

• We evaluated community pharmacy implementation of a new patient-centered label for PRN (as needed) opioid analgesics and found moderate adherence to the deconstruction of prescription steps but low adherence to the exact patient-centered language.

• The findings of this study suggest that community pharmacies are flexible to change, but higher reliability in implementation likely requires additional administrative or regulatory supports.