Addressing the drivers of medical test overuse and cascades: user-centered design to improve patient–doctor communication

Abstract

Background:

Low-value medical testing is a major component of health care overuse, both directly and through the potential for borderline and/or incidental results to trigger cascades (downstream services of uncertain value). The costs and harms from marginal test results and their cascades can add up. It is thus important to both prevent low-value tests at the outset and mitigate cascades when they arise.

Methods:

Informed by a framework for understanding and reducing overuse of care, we employed user-centered design methods (focus groups and 1:1 design meetings) with patients and primary care physicians (PCPs) to understand the problem and iteratively develop an intervention.

Results:

Design meetings with 15 PCPs, 12 patients, and 3 patient focus groups revealed myriad drivers for medical test overuse and cascades. Patients commonly believed that all medical tests yield definitive results and lack downsides. PCPs cited expert recommendations, limited time during visits, fear of lawsuits, and desire to be responsive to patients as reasons for ordering potentially low-value medical tests. We designed an intervention to address these issues using patient pre-visit educational materials, clinician reference materials on test interpretation and incidental findings, and clinician peer comparison on test overuse.

Conclusion:

Overuse of medical testing is driven by a range of factors related to PCPs, patients, and their interactions. Multipronged interventions may have the potential to address these drivers once they are rigorously tested.

Overuse of medical care is a longstanding problem in the United States and a key driver of health care spending.^1,2 Yet, interventions designed to reduce overuse have had limited impact to date,^3–7 prompting a call for renewed efforts to develop and test scalable interventions.^8,9

Low-value medical testing is a major component of overuse, both directly and through the potential for borderline and/or incidental results to trigger downstream “cascades” of additional tests and treatments of uncertain value.^10–12 Unlike with surgeries and other invasive procedures, decisions to order laboratory and imaging tests are not viewed by patients and doctors as major decisions and may not be discussed, yet the costs and harms from marginal tests and their cascades can add up. It is therefore important to both prevent low-value tests at the outset and mitigate cascades when they arise.¹³

To begin to address this problem, we rigorously designed an intervention that requires pilot testing in future work. We examined patient and primary care physician (PCP) perspectives on the drivers of medical test overuse and cascades and used user-centered design methods to iteratively develop a scalable intervention. We designed this intervention with the objective of promoting “productive” conversations about medical tests between patients and PCPs during primary care visits and mitigating cascades of care.¹⁴ To do so, the intervention aims to improve what we call “medical test literacy” in patients and to prepare patients to ask questions about medical tests. For clinicians, we designed the intervention to encourage shared decision making, where appropriate, and effective communication about medical test–ordering decisions.

METHODS

Overview and Conceptual Framework

We conducted a user-centered design process^15,16 guided by Morgan et. al’s Framework for Understanding and Reducing Overuse of Care, which conceptualizes the drivers of overuse through multiple domains.¹⁷ The most proximate domain that influences overuse of medical testing is “patient–clinician interaction,” as that is where the test ordering decision occurs. This domain is directly influenced by other domains discussed in Morgan et al.’s framework, including “clinician attitudes and beliefs,” “patient factors and experiences,” and the “practice environment.” We considered these domains in our data collection, analysis, and intervention design. Farther upstream are the domains of “culture of professional medicine,” which includes the influence of regulations and reimbursement rates, and “culture of health care consumption,” which includes consumer expectations and beliefs that more health care is better health care. We focused on the primary care context as it is founded on strong communication, relationships, and whole-person care.¹⁸

Through iterative design sessions with patients and clinicians and discussions with clinical leadership, we both identified drivers of medical test overuse and cascades guided by Morgan et al.’s framework and developed an educational intervention.^19–22 We deliberately included both driver identification and intervention design in the scope of this investigation so that the intervention design would be informed by the needs of patients and PCPs.^23,24 Furthermore, this scope allows for design session discussions to probe both topics and connections between them. For patients, we designed digital educational materials to be sent prior to scheduled visits to educate patients about medical testing and prepare them to discuss testing with their clinicians. For clinicians, we initially planned to design reference materials informed by user needs to help them conduct conversations around medical testing decisions and interpretation of results. When we learned through physician interviews that such reference materials may be insufficient to change behavior, we added a peer comparison component.²⁵

Settings and Participants

We collected data from four types of participants: (1) practicing PCPs from Brigham and Women’s Hospital (BWH), an academic medical center in Boston, Massachusetts, (2) patients who responded to Rally, an online platform that facilitates recruitment into research studies,²⁶ (3) a patient and family advisory council recruited through the Brigham and Women’s Hospital Center for Patients and Families,²⁷ and (4) informal discussions with clinical leadership. We recruited patients and PCPs from a moderately diverse range of racial/ethnic backgrounds, ages, and education levels. Design sessions occurred between August 2020 and April 2021. Institutional review boards at BWH and the RAND Corporation approved all procedures.

User-Center Design Process

We used user-center design methods both to understand possible reasons for medical test overuse and cascades and to develop intervention materials.^15,16 Three members of the research team [RSR, NT, IG] conducted design meetings with individual patients and PCPs, as well as focus groups with members of the patient advisory committee. Meetings were conducted via video conference and typically lasted 60 minutes with patients and 45 minutes with physicians. We alternated between patients and physicians to balance both perspectives throughout the process. We continued to recruit patients and physicians until we reached saturation for themes and intervention components.^28,29

Initial patient design meetings focused on patients’ experiences with medical tests, how medical testing decisions were made, challenges when talking with clinicians about medical tests, and what they wanted to hear from their clinicians to help them feel comfortable asking questions about why a test should or should not be ordered. Initial physician design meetings focused on how physicians approached conversations with patients about medical tests, as well as their experiences with and views on cascades.

Findings from these initial meetings informed the content we drafted for patient- and clinician-facing intervention materials. We also employed literature-informed best practices in development of patient education materials, such as using a fifth grade reading level, short sentences, chunked information with headings, summary of key points that are repeated, questions, and stories.^30–33 In later design sessions, patients were asked to provide feedback on the content, design, interpretation, and mode of delivery of patient-facing materials. Physicians were asked to review and provide feedback on both the patient- and clinician-facing materials. When reviewing the draft materials, participants were asked to “think aloud” about their reactions to the materials.³⁴

Data Analysis

All interviews and focus group discussions were audio-recorded and transcribed. At least two team members reviewed each transcription and independently summarized key points on current practices, content priorities, and design preferences using conventional content analysis techniques.³⁵ After reviewing independently, all research team members met to discuss major findings to reach a consensus on key themes, how they fit within the domains of the Framework for Understanding and Reducing Overuse of Care, and design implications. The patient- and clinician-facing prototypes were revised between design meetings based on these analyses and informed the content of subsequent design meetings.

RESULTS

We conducted 1:1 design meetings with 12 patients and 15 PCPs and held 3 patient focus groups (3–9 members). Participant characteristics are shown in Table 1. Within the six domains of Morgan et al.’s overuse framework, we identified 11 themes that help explain the reasons for overuse of medical testing (Table 2). Through our user-centered design process and analysis of data, we identified intervention design components to allow health care institutions to address these themes (Figures 1 and 2).

Table 1:

Participant Characteristics

Patients (for interviews and focus groups)	N = 22 (%)
Age (years)	Range: 23–82
< 44	7 (32%)
45–65	8 (36%)
> 65	7 (32%)
Ethnicity
Not Hispanic or Latinx	22 (100%)
Race
White	15 (68%)
Black or African American	4 (18%)
Asian or Pacific Islander	3 (14%)
Gender
Male	7 (32%)
Female	15 (68%)
Level of Education
High school diploma or GED or some college but no degree	3 (13.5%)
2-year or 4-year college degree	9 (40.5%)
Post-graduate degree	10 (45%)
PCPs	N = 15 (%)
Gender
Male	11 (73%)
Female	4 (27%)
Racial/ethnic group
Asian	4 (27%)
Non-Hispanic white	11 (73%)
Years since Residency
6–15	3 (20%)
> 15	12 (80%)

Table 2.

Summary of Themes Explaining the Reasons for Overuse of Medical Test Ordering and Intervention Design Implications by Domain of Framework for Understanding and Reducing Overuse of Care¹³

Framework Domain	Themes	Implications for intervention design (authors’ analysis)
Culture of professional medicine	Clinicians attribute overuse and cascades to factors out of their control, such as radiologist recommendations, culture of ensuring nothing is missed, fear of malpractice lawsuit.	Requires changes in expert recommendations, regulations, medical education; materials should be embedded within existing resources and address local cultures
Clinician attitudes and beliefs	Clinicians recognize that incidental findings and cascades are burdensome, time consuming, and rarely clinically important. Clinicians largely believe they already use best practices when communicating with patients about medical test ordering decisions.	Interventions should appeal to clinicians’ existing beliefs on low-value of test ordering following incidental findings. Educational materials will be insufficient to change behavior; other strategies such as peer comparison with test ordering rates may be needed.
Practice environment	Clinicians have limited time to discuss tradeoffs in decision to order medical tests. Fear of lawsuits, especially if going against advice of specialist	Interventions should be designed to save clinicians time and provide proven ways to discuss tradeoffs with scripted language and data available at point of care. Create institutional guidelines/reference material for clinicians to give them cover to avoid medical test overuse and cascades.
Culture of health care consumption	Patients may believe that most or all medical tests yield definitive results, are always good, and lack any downsides. Patients may perceive all discovery of incidental findings and resulting cascades as beneficial.	Educate patients about the “medical test literacy,” the limits of medical tests, and potential benefits and downsides. Educating patients about how incidental findings and cascades are often a waste of time and can be harmful.
Patient factors and experiences	Patients vary in their decision-making preferences. Some prefer to let doctor make all decisions. Patients vary in their interest in getting medical tests. Some prefer to get as many tests as possible. Patients may assume any educational materials about medical tests are designed to encourage them to get tests.	Design educational materials to be broadly applicable to a range of preferences, provide insight into how doctors think about medical tests and questions that patients can ask their doctor. Educate patients on test benefits and downsides. Start educational materials by putting medical tests in context, as “one tool in the toolbox.” Avoid emphasizing reasons to get a medical test.
Patient-clinician interaction	Clinicians want to be responsive to patient needs.	Train clinicians to offer patients alternatives to testing (e.g., close follow-up) to demonstrate caring.

Figure 1:

Screenshot of patient educational materials available via web (https://testinfo.bwh.harvard.edu/) to prime them for conversations about medical testing. Prior to a visit, patients will receive an email and/or text message with a link to the materials. [Color in print and online]

Figure 2: Clinician reference materials.

Panel A shows a screenshot from the medical test interpretation reference. Panel B shows a screenshot from the incidental findings reference. [Color in print and online]

Drivers and Potential Interventions for Medical Test Overuse and Cascades by Framework Domain

Within the “culture of professional medicine” domain, clinicians identified several factors that drove their decisions to order potentially low-value medical tests and pursue cascades. These factors included the following:

• Discomfort with contradicting recommendations from radiologists: “At the end of the day, it’s like, if the radiologist is telling me to do it, then, like, I’m not gonna kind of challenge their expertise, even though I may not be 100% confident.” (Physician 6)

• The desire to avoid missing something: “I just saw a healthy 65-year-old woman for physical. And I did her white cell count last year, and she didn’t have any complaints. And I was scratching my head, should I repeat it or not? And I decided, what the heck, she has good insurance, why don’t I check it. And it came [back as] 35,000. And I scratched my head, and then I repeated it, and 44,000, and I started to dig deeper and she has a myeloid leukemia. And yes. She is most likely an outlier, and most likely we would have caught it later, but it makes you think twice whether or not to do a test.” (Physician 14)

Within the “clinician attitudes and beliefs” domain, clinicians reported extensive experience with incidental findings and cascades. They described cascades as burdensome, time consuming, and most often of limited clinical value.

“I know that the more I order the more I find. And then it will be a reason for the patient to worry and for me to repeat unnecessary tests.”

(Physician 14)

“We’re subjecting people to a lot of testing, you know, it costs more, it’s not great for them, it’s something, one more thing, that we have to keep track of that I really feel is unlikely to be clinically significant.”

(Physician 11)

This suggested that clinicians would be receptive to an intervention to address these issues. However, clinicians also largely believed that they already use best practices when communicating with patients about medical test decisions, which suggested they may not believe they have room for improvement.

“I think most of us, all of us, use [proposed language for communicating incidental findings to patients] during those conversations, that there are incidental findings, what do we do with, you know, how do we proceed with those incidental findings, what it means… what kind of work up it needs.”

(Physician 2)

Assessment of the “practice environment” domain revealed additional reasons for possible over-testing and cascades: time constraints and fear of lawsuits.

“There’s a lot of teaching that goes into the telling the patient why you think this test is not necessary… If they’re not convinced, and I don’t have time, more time, to explain [it to] them or convince them, I usually order that [test].”

(Physician 2)

“You know, everybody is afraid of being sued. It’s much easier for me to order [the] unnecessary mammogram than to miss something.”

(Physician 14)

The role of the “culture of health care consumption” domain was evident in design meetings with patients. Patients were often surprised to learn that medical tests could be inconclusive or have other downsides.

“No doctor can tell me by looking at me what my cholesterol level is, what my sugar level is, it’s only via test that I will know it, and if you tell me that test itself is not conclusive by any chance, or rarely conclusive, rather, I’m in a mess.”

(Patient 8)

“Patient factors and experiences” influenced patients’ reasons for asking for or agreeing to a medical test. We found a range in decision-making preferences. Some preferred to let the doctor make all decisions and would get any test their doctor recommends.

“I don’t like to necessarily be in the driver’s seat when I go into my visits… I kind of come here because I want you to, in a way, take that burden from me.”

(Patient 9)

Others prefer playing a more active role.

“The doctor will recommend and decide and discuss pros and cons with you, but ultimately, don’t be too, you know, concerned, because at the end of the day, it’s really your decision.”

(Patient in focus group)

Patients also varied in their interest in getting medical tests. PCPs described experiences with patients preferring to get many tests. Other patients had negative experiences with cascades, so their views on medical tests changed over time.

“…some people don’t wait—they’d rather do a test to get a prognosis rather than wait to find out whether you’re right or wrong.”

(Physician 7)

“That MRI really convinced me that there was something seriously wrong with my back when probably, there wasn’t, or if there was, it wasn’t anything that wasn’t going to get better… and so it gave me an attitude… I had a kind of return of those symptoms, not as bad, recently, within the past couple years, and I told them ‘don’t give me an MRI, I don’t want to know.’”

(Patient 10)

Within the “patient-clinician interaction” domain, we found that clinicians sometimes ordered medical tests to demonstrate that they care about their patient and want to help, even when they know the tests may be low-value.

“I mean, I’ve ordered X-rays almost as placebos… ‘It’s radiation, you’re giving it to the patient, it’s letting the patient manipulate you’ is what you hear… To me, that’s not what this is about. It’s working with the patient.”

(Physician 1)

Intervention Components to Prime Conversations About Medical Tests Between Patients and Clinicians

Using the findings above and through iterative design with patients and clinicians, we developed an intervention addressing factors contributing to overuse and cascades that would be feasible for health systems to implement. The intervention involves three components: patient education materials, clinician reference materials, and clinician peer comparison. We designed the intervention for use before annual physicals because it is common for medical tests, including low-value tests, to be ordered during such visits.^36,37

Patient Educational Materials.

To prepare patients for conversations about medical tests, we highlighted three key points in patient, pre-visit education materials (Figure 1). The first point is that medical tests are one of several tests in a doctor’s toolbox. (We use “doctor” because the clinicians involved in our initial study included only doctors. These points should work for other clinicians as well.) This point attempts to put medical tests in their proper context within medical decision making so that patients don’t view tests as essential in all circumstances. We tested a version of the materials that began with explaining the reasons one might get a medical test but found that patients then assumed that the purpose of the materials was to convince them to get a medical test.

“I feel like after reading [an early draft of the prototype educational materials], it’s like, ‘well this [test result] could be true, or not, and you have to conduct another test.’”

(Patient 11)

The second point is that medical tests have benefits and downsides. In our design meetings, we found many patients were unaware of the potential downsides and limitations of medical tests. Several patients were surprised and uncomfortable at this notion as they were more accustomed to being encouraged to get medical tests without discussion of false positives or the risk of incidental findings. We learned that overemphasizing downsides might scare patients from getting any medical tests, but some discomfort was unavoidable for patients who had assumed medical tests were always conclusive and had no downsides.

“I feel actually a little, I don’t want to say less confident, but a little more concerned about what, you know, the point of the test, you know, if it’s not, you know, 100% reliable.”

(Patient 9)

Finally, our education materials reminded patients that they can always talk with their doctor and ask them question about the purpose of the medical tests and alternatives.

“I feel more empowered to ask questions, because if there’s a flyer that’s being sent or, you know, a text that I’m getting that’s linked to this website, it shows to me that this is like, you know, a health care office that’s interested in me, you know, asking questions and being informed.”

(Patient 9)

Although these messages addressed patient needs, some PCPs expressed concern that such pre-visit education would raise patient questions or concerns that would require more work from them during already busy visits.

“…and my thought is that you will open a Pandora’s box.”

(Physician 14)

Clinician Reference Materials.

For clinicians, we developed two reference materials in partnership with our health system’s Primary Care Office Insight (PCOI) reference source to ensure easy, point of care accessibility via the electronic health record. Based on feedback that physicians learned medical test interpretation in medical school but do not have formal training in this area afterward, we developed a reference on medical test interpretation that includes primers on Bayesian reasoning, pre- and post-test probability, sensitivity, and specificity, as well as flowcharts and scripted language to decide and discuss when to order a medical test (Figure 2a).

Based on feedback from physicians that they were sometimes flummoxed by how to follow up or talk about incidental findings, we developed a reference on incidental findings that included scripted language, statistics on frequency of these findings, interpretations of abnormal findings on lab results, and for imaging incidentalomas, follow-up plans with links to more detailed references (Figure 2b). Although PCPs agreed with the content of these materials, most said that they believe they already follow the suggestions outlined in the materials.

“It’s like telling people something they already know, even though they don’t necessarily, like, practice it, right? And I think people, like, believe they’re doing this…”

(Physician 6)

Clinician Peer Comparison.

We designed an email template that compares a given clinician’s test-ordering rates with those of their peers for ten potentially low-value tests. The intent of these emails would be to “nudge” the clinicians to self-examine their medical test ordering habits and motivate them to review the reference materials we supplied. In design meetings, we showed PCPs mockup emails made using both the templates and fictional medical testing data. These mockups included graphical peer comparison for common medical tests: complete blood counts, basic and complete metabolic panels, liver function tests, thyroid tests, urinalyses, and electrocardiograms. The PCPs interviewed seemed receptive to this approach.

“Three times worse than everyone else, I’m like ‘that’s not good’ so I’m thinking, okay. This graph hits that home further… I mean, I would be, I personally would be welcoming of help. I don’t want to send tests that aren’t needed… I think I would be happy to make sure that I’m doing guideline concordant care.”

(Physician 10)

DISCUSSION

Based on input from patients and PCPs, we identified key themes that highlight causes and potential interventions for medical test overuse and cascades with the help of Morgan et. al’s Framework for Understanding and Reducing Overuse.¹⁷ Patients tended to believe all medical tests yielded definitive results and lacked downsides. Clinicians largely recognized that over-testing was a problem and understood the harms from resulting cascades. They attributed these issues to expert recommendations, limited time during visits, fear of lawsuits, and their desire to be responsive to patients’ needs. Clinicians generally believed that they already follow best practices for discussing medical tests, which suggests changing their behavior will require more than receiving reference materials.

To address these issues, we designed an intervention that, with further testing, has potential to address these barriers. The intervention involves educating patients, giving clinicians reference materials, and sharing peer comparison data with clinicians on their testing rates so that they recognize their own role in reducing testing and cascades. Unlike prior work on shared decision making that focuses on one or a set of specific tests,^38–45 we focused on conversations around any kind of medical test, such as those commonly ordered by a PCP during an annual physical. This broader approach has the potential to complement other approaches by increasing what might be called “medical test literacy” and preparing patients for more difficult shared decision-making conversations when they come up. This work highlights the contributions of patients and clinicians in an iterative design process to understand the drivers and possible solutions to low-value test ordering and cascades.

Limitations

Our approach has several limitations. First, our sample of patients and providers was somewhat modest, with 15 physicians and 22 patients participating in design meetings. Patients and providers may thus hold additional insights on the drivers of low-value medical tests and cascades that were not captured by our sample. Second, patients in our sample may not be representative of all types of patients. Specifically, the majority of patient participants were white and/or female, and no participants were Hispanic. Power dynamics may also make it difficult for some patients to ask questions about medical testing. Moreover, patients volunteered to participate in our design meetings; research volunteers may be more positively inclined toward medical care compared with others. However, patients more positively inclined toward medical care are likely the patients who would be best targeted for the intervention. Third, although we attempted to design the intervention to be maximally scalable, the peer comparison components would be facilitated by a quality measurement infrastructure that small or under-resourced clinics may lack. Fourth, sending patients educational materials about the limits of medical tests risks unintended consequences of confusing patients and diminishing their confidence in the health care system, or prompting them to request medical tests where they otherwise would not. While we explored these concerns during user testing, empirical evaluation is needed to assess this risk formally. Fifth, our intervention did not attempt to address all aspects of the Morgan et al. framework. Rather, it provides one building block. Additional, complementary interventions are needed to address other drivers of low-value care, especially those related to the domains we did not address, such as the culture of professional medicine. Evidence from interventions based on the Choosing Wisely campaign suggests that multifaceted approaches work best.^5–7,20,46

CONCLUSIONS

Overuse of medical tests is widely acknowledged by PCPs and driven by a range of factors, including patient beliefs that all medical tests yield definitive results and lack downsides, expert recommendations, limited time during visits, fear of lawsuits, and clinician desires to be responsive to patients. Many, but not all, of these factors have the potential to be addressed by clinical institutions in the form of general education for patients about medical testing and a combination of reference materials and peer comparison for clinicians. Further testing is needed to assess this intervention’s potential to address medical test overuse and cascades.

CONFIRMATION OF CONFIDENTIALITY

We confirm that all personal identifiers have been removed or disguised so the patients and physicians described are not identifiable and cannot be identified through the details of the story.