The promises and pitfalls of real-time benefit tools: A qualitative study of primary care providers

Abstract

Background:

Real-time benefit tools (RTBT) are federally mandated clinician-facing applications embedded in the electronic health record (EHR) that display the out-of-pocket costs of prescribed medications and lower-cost alternatives, if available. RTBTs may be useful for primary care providers (PCP), whose patients often take numerous medications; however, RTBTs may also add time to visits that are already complex and compressed.

Objective:

To explore PCP experiences with and attitudes toward RTBTs.

Design:

We conducted a qualitative descriptive study of PCPs with access to an RTBT at primary care clinics affiliated with two large academic health systems. Semi-structured interviews elicited 1) attitudes toward medication cost transparency, 2) RTBT training, 3) experiences using RTBTs, and 4) perceived effects of RTBT use on decision-making and administrative burden.

Participants:

Primary care physicians and advance practice providers with ≥1 year of practice.

Measurements:

Thematic analysis of interview transcripts.

Results:

We interviewed 35 PCPs. Most were physicians (25/35), female (23/35), and had ≥10 years’ experience (19/35). We identified three themes. First, PCPs received minimal education on RTBTs and were open to more training. Second, PCPs believe RTBTs have the potential to improve patient care by facilitating cost conversations and reducing administrative burden. Third, however, several pitfalls have hindered use, including incomplete information, inaccurate cost estimates, and clinically inappropriate lower-cost suggestions.

Limitations:

Two study sites, which both use the same EHR.

Conclusion:

While PCPs generally saw the RTBT as a promising tool that can facilitate compassionate, cost-conscious care, they identified problems that limited their desire to use it. For utilization to improve, EHR developers and insurers must ensure that RTBT information is actionable, clinically appropriate, and time saving.

Introduction

About 40% of insured adults in the US say it is difficult to afford their healthcare costs and report not taking their medications due to cost.¹ Cost-related nonadherence can lead to devastating consequences, including faster disease progression and increased mortality.^2-4 Since 2021, Medicare requires Part D plans to report individualized out-of-pocket medication costs to clinicians via real-time benefit tools (RTBTs). RTBTs are applications embedded in the electronic health record (EHR) that display medication costs to clinicians and suggest lower-cost alternatives when available.⁵ While retail pharmacists have long had access to cost and coverage information at the point-of-sale, RTBTs provide this information earlier, at the time of clinical decision-making. Access to RTBTs is increasing nationwide: many private insurers now provide cost estimates to RTBTs⁶ and over two-thirds of US health systems have adopted one in their EHR.⁷

Early evidence suggests that RTBTs may lower medication costs and improve adherence. In one cluster-randomized clinical trial, patients whose clinicians used an RTBT saved an average of $28/month on prescriptions that had available lower-cost alternatives.⁸ In another, medication fill rates were 8% higher for patients whose clinicians queried the RTBT.⁹ However, RTBTs remain inconsistently implemented, poorly integrated, and underutilized by PCPs. PCPs actively query the RTBT in <1% of visits.¹⁰ When the RTBT provides estimates via interruptive alerts, PCPs change their medication orders in response to only 12% of alerts.¹¹

RTBTs may be particularly useful for PCPs, who often care for patients with multiple chronic conditions requiring complex management and numerous medications.^12,13 However, concerns about added time, complexity, and alert fatigue may discourage PCPs from using RTBTs during encounters that already feel rushed. There is limited evidence on the experiences of PCPs who have access to RTBTs. To address this gap, we conducted interviews of PCPs to understand their attitudes toward and experiences with RTBTs. Findings from these interviews are critical to anticipating the impacts of Medicare’s RTBT policies on clinical practice, patient spending, and outcomes.

Methods

Study design, setting, and sample

We conducted a qualitative descriptive study following the Consolidated criteria for reporting qualitative research (Appendix).¹⁴ We conducted semi-structured interviews with PCPs from two large primary care networks affiliated with Duke University Health System (Duke) and UCHealth between October 2023 – April 2024.

PCPs in both health systems had access to an RTBT that could provide cost estimates for ordered medications, information on whether prior authorization was needed, and available lower-cost alternatives (up to 2 alternatives at Duke, 4 at UCHealth).^15,16 Lower-cost alternatives could include lower-cost pharmacies, different quantities (e.g., 90-day vs 30-day supply), different medications, and/or different formulations (e.g., capsule vs tablet). If a lower-cost alternative was available (>$0.15 less at UCHealth, >$2 less at Duke), the RTBT fired an alert at the time the medication order was signed. Clinicians could also voluntarily query the RTBT at any time during the visit by pending a medication order (i.e., placing it in checkout cart) and then clicking “Estimate” (Figure 1).

Figure 1. Real-time benefit tool workflow.

At both institutions, clinicians type a medication order in the medication entry window and then select the clinically appropriate medication in the medication selection window. The medication name and details then appear in the medication order window (i.e., check-out cart). Clinicians have the option to click “Estimate” to view the RTBT window at any time. Clinicians can also sign the prescription without actively querying the RTBT. If a lower-cost alternative is available, the RTBT window pops up, interrupting the clinician’s workflow with cost information. If no lower-cost alternative is available, the RTBT window does not pop up, and the medication order is sent directly to the pharmacy. Lower-cost alternatives can include lower-cost pharmacies, an alternative number of prescription days (e.g., 90-day vs 30-day supply), different medications, and/or different medication formulations (e.g., capsule vs tablet). At the time of the study, the RTBT window was triggered to pop up if an alternative was available for >$0.15 less at UCHealth or >$2 less at Duke.

Study subjects were attending physicians (MD/DO), nurse practitioners (NP), or physician assistants (PA) with ≥1 year of clinical experience at their current institution and providing direct outpatient care ≥1 day/week in General Internal Medicine, Family Medicine, or Geriatrics. Clinicians were eligible regardless of their familiarity with RTBTs. Resident physicians were excluded since they rotate between different specialties. We identified potential participants from publicly available staff directories and sent email invitations with a link to a REDCap survey collecting eligibility information and professional and demographic data (see Appendix). We contacted clinic directors to encourage their staff to participate. Two reminder emails to non-respondents were sent 1 and 2 weeks after the initial invitation. We used purposive sampling to capture a full range of experiences across clinic zip code income-levels (below vs above 2024 US median household income of $83,730),¹⁷ clinician types (physician vs NP/PA), and institutions. Participants were offered $50 as compensation for participation.

Data

The interview guide was informed by the Organizational Theory of Innovation Implementation Effectiveness^18,19 and the Unified Theory of Acceptance and Use of Technology (Appendix).¹⁶ These models posit that clinicians’ perceptions of new clinical decision support systems are influenced by 1) their health system’s implementation policies (e.g., trainings), 2) clinicians’ values (e.g., whether cost should be considered in clinical decisions), and 3) the tool’s performance and integration into clinical workflow. Clinicians’ perceptions in turn influence the tool’s impact on clinical practice and patient outcomes. Interview guides differed based on familiarity with RTBT (reported use approximately >5 times/week vs ≤5 times/week) (Appendix).

A preliminary interview protocol was pilot tested with three PCPs. After refinement, 1 qualitative interviewer (CES, who is a PCP) administered the final interview guide to each participant. All interviews were audio-recorded and immediately sent for transcription. The study team met regularly to review memos and transcriptions. Interviews were conducted until data saturation was reached, meaning no new concepts or ideas emerged during data collection or analysis.²⁰

Analysis

Interview transcripts were analyzed using an applied qualitative thematic approach and NVivo 14.^21,22 Two study team members (DL, WS) coded each transcript, using an a priori codebook developed and iteratively refined by the study team. Discrepancies in double-coded transcripts were reconciled by 2 additional authors (CES, JKD) and the final coded interviews were analyzed, grouped, and summarized by four authors (CES, WS, RS, JKD) into overarching themes.

Human Ethics and Consent to Participate declarations

This study was approved by the Duke Health and UCHealth Institutional Review Boards. Consent was obtained from all participants.

Results

We sent recruitment emails to 316 PCPs; 39 initiated the e-consent and survey. Of these, 2 did not complete the survey and 2 did not attend their scheduled interview. We stopped recruitment and closed the survey once thematic saturation was reached.²⁰ Interview time averaged 29 minutes (range 19-46). In line with our purposive sampling approach, the final sample was balanced across sites (17 Duke, 18 UCHealth) and populations served (18 worked at clinics in lower-income zip codes). Most participants were physicians (25/35), female (23/35), White (25/35), and of non-Hispanic ethnicity (34/35) (Table 1). Twenty PCPs reported frequent RTBT use while 15 reported rare use. All had seen an RTBT cost estimate at least once.

Table 1.

Participant characteristics

Characteristic	N (%)
Institution
University of Colorado Health	17 (49%)
Duke	18 (51%)
Provider type
Physician	25 (71%)
Nurse practitioner or physician assistant	10 (29%)
Years in practice a
<10	16 (46%)
≥10	19 (54%)
Gender
Female	23 (66%)
Male	12 (34%)
Race
Asian	5 (14%)
Black	2 (6%)
White	25 (71%)
Other b	3 (9%)
Hispanic ethnicity	1 (3%)
Median income of clinic zip code
Below US median income	18 (51%)
Above US median income	17 (49%)
Self-reported frequency of RTBT use
Rare	15 (43%)
Frequent	20 (57%)
Reported patient payer mix
% Commercial, mean (SD) c	52 (16)
% Medicare, mean (SD) d	39 (25)
% Medicaid, mean (SD) c	11 (10)
% Uninsured, mean (SD) c	2 (2)

^aDifference between the year of the interview and the year the participant first began practicing medicine as a resident, physician assistant, or nurse practitioner

^bRace categories available in the survey were: American Indian / Alaska Native, Asian, Black or African American, Native Hawaiian / Other Pacific Islander, White, Other, or Do not wish to answer

^cData provided by 30 participants

^dData provided by 29 participants

We identified 3 themes from clinicians’ experiences, which are described below (see also Table 2 and Appendix Table).

Table 2.

Representative quotations

Themes and Subthemes	Quotations
Theme 1. PCPs receive minimal training on RTBT use and are amenable to more
	“They probably sent us an email, and I may or may not have read it. I don’t remember interacting with any training… I just remember it as showing up one day and being like, “Oh, wow, that’s kind of cool.” (UCHealth, female physician, rare user) “Maybe just a little blurb email or something would be fine, but otherwise – I think people would skip over it.” (UCHealth, female physician, frequent user)
Theme 2. RTBTs have the potential to improve clinical care and reduce administrative burden
2a. RTBT empowers PCPs to engage in cost conversation	“We all have blind spots when it comes to making assumptions about people based on everything from how they dress to what their occupation is to skin color. And I think it's helpful to realize, wow, cost is a much bigger issue for this person than I would have expected.” (Duke, male physician, frequent user) “I don’t know that I am a person that can talk to people about [patients’] finances. Is that really part of my scope of practice?” (UCHealth, female PA, rare user)
2b. PCPs incorporate RTBT cost estimates into their clinical decisions	“It does allow me to have a conversation with patients to know, hey, this might be a higher cost…is this something that you think you can do? If not, we can look at some other options.” (UCHealth, male physician, frequent user) “I feel like it's my responsibility to acknowledge whatever limitations or benefits [patients] have and work within those constraints. If I don't think about their financial well-being as part of the whole picture, I'm not doing my job.” (UCHealth, male PA, frequent user)
2c. RTBT can reduce administrative burden after visits	“An extra couple of minutes in an appointment saves me 30 minutes to an hour of going back and forth with our panel managers, MyChart messages. So, it's really an efficiency perspective that's my primary motivation [for using RTBT].” (Duke, male physician, frequent user) “Getting those pop-ups is just a helpful reminder of, ‘Oh, wait, maybe there is an alternative that I could choose in this moment that better takes care of my patient and saves all these downstream issues.’” (UCHealth, female physician, frequent user) “It seems to be a largely fruitless clicking and time away from the patient, time on the computer.” (UCHealth, male physician, rare user)
Theme 3. PCPs do not fully trust the RTBT, hindering widespread adoption
3a. RTBT cost estimates are not always accurate or available	“There's been times where the cost estimator is not right, it's actually cheaper than what it shows up as being.… I might not go for a medication that I normally would go for because I think it's super expensive.” (Duke, female NP, frequent user) “I did change the medication… to whatever [the RTBT] recommended. And then I got a patient message that this isn’t actually covered…And so, I decided that I wasn’t going to use time on it.” (UCHealth, male physician, rare user)
3b. RTBT recommendations for lower-cost alternatives are not always clinically appropriate	“I wanted to do six weeks of a PPI, and it’s like, well, what if you just did a year supply? Well, that’s not my intention.” (UCHealth, male physician, rare user) “There’s times it’s recommended things that my patient is allergic to. And so then I just override it. Or it doesn’t know that a patient has tried and failed a therapy before.” (Duke, female NP, rare user)
3c. RTBT information is insufficient	“If I knew a CT scan would cost $1,000 at one site versus $100 at another site, that would be incredible.” (UCHealth, female physician, frequent user) “The questions in the back of my mind are OK - prior authorization is required.…What is going to go in that prior authorization? What’s required?” (Duke, male physician, rare user)

These 15 selected examples represent exemplars from 13 different participants, 2 of whom had two quotations.

PA: physician assistant. UCHealth: University of Colorado Health. RTBT: real-time benefit tool. NP: nurse practitioner

Theme 1. PCPs receive minimal training on RTBT use and are amenable to more

No PCPs could recall having received formal RTBT training and PCPs demonstrated numerous knowledge gaps. Some were unsure what the cost estimates represented (e.g., average projections versus patient-specific out-of-pocket costs). Others were unaware of the RTBT’s basic functionalities. For example, while all PCPs had seen the RTBT interruptive alert, about half (15/35) did not know they could voluntarily request an estimate.

PCPs were mostly receptive to receiving education on how to use the RTBT effectively, either informally from colleagues or via unobtrusive trainings such as clinic-based “lunch & learns” or tutorials. PCPs were also amenable to health system-led programs that would encourage – but not require – RTBT use.

Theme 2. RTBTs have the potential to improve clinical care and reduce administrative burden

2a. RTBT empowers PCPs to engage patients in cost conversations

Almost all PCPs (n=33) found the RTBT valuable and liked that it acted as a reminder that medications may be unaffordable. The RTBT also seemed to prompt PCPs to initiate cost conversations: “It opens a door for that conversation or just spurs it a little bit more when [the cost] is in your face.” When the RTBT popped up, some PCPs described turning their computer screen to the patient or narrating what they saw in the RTBT window and asking, “Is that something that’s affordable to you?” They appreciated that the RTBT provided enough information to discuss the financial tradeoffs of different treatment options.

Of note, a minority of PCPs (n=2) stated that cost conversations were outside their scope of practice and that other clinic staff such as clinic-based pharmacists and/or nurses should be responsible for using and acting on the RTBT.

2b. PCPs incorporate RTBT cost estimates into shared decision-making

Almost all PCPs (33/35) had considered RTBT cost estimates when prescribing medications, typically after selecting the best clinical option. PCPs were more likely to switch to a lower-cost alternative if the switch was straightforward (e.g., different pharmacy), if the original order was “expensive,” and/or if the potential cost savings were “high.” PCP definitions of “expensive” differed widely: for some, any amount below $100 for a 30-day supply was considered affordable, while others thought that $5-10 was significant. PCPs generally agreed, however, that it was not worth having the RTBT interrupt their workflow if the savings potential was only $1-2.

Some PCPs regularly queried the RTBT voluntarily, generally for new brand-name medications and/or to determine whether prior authorization was required. They tended to query the RTBT most for patients who mentioned that affordability was an issue, took multiple prescription medications (to help with medication prioritization), or specifically asked about a medication’s cost.

A minority of PCPs (n=2) routinely ignored RTBT alerts, in part because they worried that patients would ration their care.

Domain 2c. RTBTs may reduce clinician administrative burden and save time after visits

Using the RTBT could add several minutes to the visit, by prompting cost conversations or potentially lengthy searches for affordable medications. RTBT alerts also regularly interrupted workflows at inopportune times (e.g., after a new medication had already been discussed). This extra time spent on the RTBT was viewed by some as worthwhile, however, because it could reduce time-consuming post-visit communications about costs and could prevent delays in access to needed medications.

Theme 3. The RTBT’s flaws lead to mistrust and may hinder widespread adoption

Despite considerable optimism about RTBTs, PCPs did not completely trust the information they saw. Overall, the RTBT’s flaws were felt to prolong clinical encounters and made PCPs less willing to use it.

3a. RTBT cost estimates are not always accurate or available

RTBT cost estimates were sometimes inaccurate. PCPs usually learned that the information was erroneous days or even months after clinic visits – via patient portal messages or conversations at follow-up appointments. RTBT alerts also occasionally popped up with cost estimates that were grayed out and listed as “unknown.” While many PCPs appreciated that the alerts reminded them to address medication costs, they became frustrated when the RTBT interrupted their workflow to provide information that was of no use.

All PCPs expressed some level of distrust in the RTBT’s accuracy, which affected how they used it. Some considered its information “with a grain of salt,” double-checking costs in online formularies or asking patients to verify estimates at the pharmacy. These extra steps took time. Others dismissed all cost estimates outright after as few as one bad experience. One PCP who frequently double checked the RTBT thought that the information was becoming more accurate over time.

3b. RTBT recommendations for lower-cost alternatives are not always clinically appropriate

PCPs tended to value the RTBT more for its cost estimates (despite inaccuracies) than for its lower-cost alternative recommendations, which they reported were frequently inappropriate. Examples of inappropriate recommendations included lengthening a 6-week course of a proton pump inhibitor for peptic ulcer disease to 90 days; switching to a strongly anticholinergic medication for an 80-year-old patient; or sending an urgently needed antibiotic to a mail-delivery pharmacy. When recommended alternatives were clinically inappropriate, some PCPs searched for appropriate lower-cost alternatives on their own, either via additional RTBT queries or by searching online. One PCP described this process as a lengthy “fishing expedition” that took time and caused frustration. Others systematically ignored all recommendations for lower-cost alternatives.

3c. RTBT information is insufficient

PCPs were disappointed that the RTBT did not provide cash prices (i.e., medication costs without insurance), information on coupons, or cost estimates for more items, such as over-the-counter medications, laboratory tests, imaging, medical supplies, and vaccines (Table 3). Without more comprehensive information, they found it difficult to provide counseling on financial tradeoffs.

Table 3.

Clinician-suggested additions to information provided by real-time benefit tools

Medication costs	Over-the-counter medications
	Prescription medications obtained without using insurance
	Information on available coupons and pharmaceutical assistance programs
	Coverage by secondary insurance plans
	Future changes in out-of-pocket costs due to deductibles or reaching out-of-pocket maximums
	Monthly or annual estimates for an entire medication regimen
Non-medication costs	Laboratory tests
	Imaging
	Medical supplies (especially glucose monitors and test strips / sensors)
	In-clinic orders (e.g., EKG)
	Vaccines
Insurer coverage restrictions	Maximum allowable number of days' supply for a given medication
	Formulary information (e.g., list of which medications are preferred within a drug class)
	Step therapy requirements for prior authorization
	Documentation requirements for prior authorization

PCPs appreciated that the RTBT provided information on whether prior authorization was required, but that information was also incomplete. The RTBT provided no information on the step therapy or documentation requirements needed for prior authorization to be approved. So, clinicians continued to receive prior authorization denials that required time-consuming appeals and medication switches that they thought could have been prevented if the RTBT had provided more complete coverage and formulary information upfront.

Discussion

In this qualitative study, we interviewed 35 PCPs from 2 academic medical centers to explore their attitudes toward and experiences with the federally mandated RTBT. We found that PCPs valued the RTBT for its ability to increase awareness of patients’ cost concerns, improve cost-informed shared decision-making conversations, and reduce administrative burden. However, the RTBT’s shortcomings prevented PCPs from fully trusting it, including experiences with inaccurate and inconsistently available estimates, clinically inappropriate recommendations, and incomplete information. Our findings represent early implementation difficulties that could improve as RTBT technology and regulations evolve.

RTBTs have been promoted by the Centers for Medicare and Medicaid Services since 2021 as a potential mechanism for improving access to medications for Medicare enrollees. Two recent regulations have the potential to increase clinician trust in and use of RTBTs in coming years. First, Part D sponsors will be required to adhere to specific interoperability standards when transmitting cost and coverage information to RTBTs starting in 2027, which may improve the accuracy and availability of cost estimates.²³ Second, all EHRs will be required to include an RTBT in their base product by January 2028.²⁴ The scope of cost estimates will expand to include medications, medical supplies and vaccines, in line with what PCPs want. As Part D sponsors and EHR vendors prepare to comply with these regulations and improve on RTBT technology, findings from our interviews can inform which aspects of RTBT design and content they should prioritize. Our findings can also help policymakers anticipate the future impacts of federal RTBT policy on clinical practice.

Prior studies have found that RTBT use is associated with improved adherence and lower out-of-pocket costs, suggesting RTBT regulations are having their intended effect.^8,9,25 Only one study has found no association between RTBT availability and adherence, but it lacked data on whether clinicians had actually viewed the RTBT information.⁶ Efforts to increase clinician utilization could increase the RTBT’s effects on medication adherence and costs. For example, one recent study found that PCPs are far more likely to view cost estimates and make medication changes when health systems implement interruptive alerts.^11,26 EHR and RTBT developers should work on improving RTBT integration into clinician workflow in order to maximize utilization, for example via user-centered design interventions.

Our findings suggest that RTBT utilization is also unlikely to improve unless it is viewed as saving – or at least not wasting – time. PCPs consistently report having insufficient time during clinical encounters and spending too much time on EHR work instead of direct clinical face time with patients.^27,28 In our study, some PCPs were so concerned about wasting time that a single bad experience with the RTBT could lead them to click out of every alert. This finding is consistent with Everson and colleagues’ 2022 qualitative study, where physicians expressed concern that RTBTs could disrupt their workflow and that erroneous or extraneous information could waste time.²⁹ Future user-centered design interventions should aim to maximize high-yield alerts and minimize low-yield alerts.

EHR and RTBT developers should also work to incorporate information that PCPs want to see. For example, RBTs could have the capability to consider clinical indications, medication history, comorbidities, drug-drug interactions, and patient age when providing recommendations for lower-cost alternatives. Individualized recommendations that incorporate the full clinical picture would likely be more acceptable to PCPs and lead to greater use. Payers and EHR vendors could collaborate to make the RTBT’s prior authorization information more comprehensive and actionable, so that clinicians can select the most preferred medication and complete all necessary paperwork in real-time. Finally, as RTBTs evolve, they could expand to showing additional, clinically relevant information (Table 3). They should avoid, however, showing cost information that could create confusion for clinicians who may be unfamiliar with cost-of-care terminology (e.g., Average Wholesale Prices, insurer-negotiated prices). Future research should assess whether PCPs’ experiences with RTBTs change as the industry and regulatory landscape evolves.

Our study has several limitations. First, all participants used the Epic EHR (Verona, WI). Our findings do not necessarily represent experiences with RTBTs in other EHRs. Second, participant characteristics likely differed from those of the overall PCP population, because we used purposive sampling to ensure balanced numbers of participants across patients served, clinician type, and location. Third, all participants practiced in clinics affiliated with large health systems so their experiences may not reflect the views of PCPs in smaller organizations. Additionally, PCPs who chose to participate in the study may differ systematically from non-respondents. For example, they may have had more memorable experiences with the RTBT. Fourth, we did not interview clinical pharmacists, who are often central to medication decision-making in primary care. Future research should assess the perspectives of pharmacists and other staff including nurses and case managers. Finally, all interviews were conducted by a PCP so there is a potential for social desirability bias in participants’ responses. However, as a clinician, the interviewer was also able to ask informed follow-up questions about clinical scenarios and the RTBT itself.

Conclusion

PCPs generally saw the RTBT as a promising clinical tool but identified a number of pitfalls that limited their desire to use it. RTBT developers, EHR developers, and payers should harness PCPs’ early experiences to redesign RTBTs with the goal of providing accurate, actionable, and clinically meaningful information, while minimizing administrative burden.

Data Sharing Statement:

Deidentified data collected for the study can be made available upon reasonable request from Dr. Sloan (corresponding author).