Using Behavioral Economics to Reduce Low-Value Care Among Older Adults: A Cluster Randomized Clinical Trial

Jeffrey T Kullgren; H Myra Kim; Megan Slowey; Joseph Colbert; Barbara Soyster; Stuart A Winston; Kerry Ryan; Jane H Forman; Melissa Riba; Erin Krupka; Eve A Kerr

doi:10.1001/jamainternmed.2023.7703

. 2024 Jan 29;184(3):281–290. doi: 10.1001/jamainternmed.2023.7703

Using Behavioral Economics to Reduce Low-Value Care Among Older Adults

A Cluster Randomized Clinical Trial

Jeffrey T Kullgren ^1,^2,^3,^4,^5,^✉, H Myra Kim ^1,³, Megan Slowey ⁶, Joseph Colbert ⁵, Barbara Soyster ², Stuart A Winston ⁷, Kerry Ryan ⁵, Jane H Forman ¹, Melissa Riba ⁶, Erin Krupka ⁸, Eve A Kerr ^1,^2,³

¹Veterans Affairs Center for Clinical Management Research, Veterans Affairs Ann Arbor Healthcare System, Ann Arbor, Michigan

²Department of Internal Medicine, University of Michigan Medical School, Ann Arbor

³University of Michigan Institute for Healthcare Policy and Innovation, Ann Arbor

⁴Department of Health Management and Policy, University of Michigan School of Public Health, Ann Arbor

⁵University of Michigan Center for Bioethics and Social Sciences in Medicine, Ann Arbor

⁶Center for Health and Research Transformation, Ann Arbor, Michigan

⁷IHA, Ann Arbor, Michigan

⁸University of Michigan School of Information, Ann Arbor

Accepted for Publication: November 22, 2023.

Published Online: January 29, 2024. doi:10.1001/jamainternmed.2023.7703

^✉

Corresponding Author: Jeffrey T. Kullgren, MD, MS, MPH, VA Center for Clinical Management Research, VA Ann Arbor Healthcare System, PO Box 130170, 152, Ann Arbor, MI 48113 (jkullgre@med.umich.edu).

Author Contributions: Drs Kullgren and Kim had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Kullgren, Kim, Forman, Riba, Krupka, Kerr.

Acquisition, analysis, or interpretation of data: Kullgren, Kim, Slowey, Colbert, Soyster, Winston, Ryan, Kerr.

Drafting of the manuscript: Kullgren, Slowey.

Critical review of the manuscript for important intellectual content: All authors.

Statistical analysis: Kullgren, Kim, Slowey.

Obtained funding: Kullgren, Kerr.

Administrative, technical, or material support: Kullgren, Colbert, Soyster, Winston, Ryan.

Supervision: Kullgren, Kim, Colbert, Riba, Kerr.

Conflict of Interest Disclosures: Dr Kullgren reported grants from Donaghue Foundation during the conduct of the study, and support was also provided by the US Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service (HSR&D). In addition, Dr Kullgren reported consulting fees from SeeChange Health, HealthMine, the Kaiser Permanente Washington Health Research Institute, and the Washington State Office of the Attorney General and honoraria from the Robert Wood Johnson Foundation, AbilTo, Inc, the Kansas City Area Life Sciences Institute, the American Diabetes Association, the Luxembourg National Research Fund, the Donaghue Foundation, the National Science Foundation, the University of California Los Angeles, and the University of Pennsylvania outside the submitted work. Mr Colbert reported grants from the Donaghue Foundation and nonfinancial support from Integrated Health Associates during the conduct of the study. Dr Forman reported grants from the Donaghue Foundation during the conduct of the study. Dr Kerr reported grants from the National Institutes of Health and grants from VA HSR&D during the conduct of the study. No other disclosures were reported.

Funding/Support: This study was funded by the Donaghue Foundation. Support for this work was also provided by the US Department of Veterans Affairs HSR&D Service. Dr Kullgren was supported by a US Department of Veterans Affairs Health Services Research and Development Service Career Development Award (CDA 13-267). The project described was also supported by grant No. P30DK092926 (Michigan Center for Diabetes Translational Research) from the National Institute of Diabetes and Digestive and Kidney Diseases.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs or the US government.

Meeting Presentations: Results from this study were previously presented at the 2021 Society of General Internal Medicine Annual Meeting, April 20, 2021, virtual; the 2021 AcademyHealth Annual Research Meeting, June 16, 2021, virtual; and the 2022 International Conference on Deprescribing, September 6, 2022, Kolding, Denmark.

Data Sharing Statement: See Supplement 3.

Additional Contributions: We are grateful for the clinicians and patients from Michigan Medicine and IHA who participated in this study and for the contributions of the trained medical record abstracters who collected blinded outcome data. We also thank Angela Fagerlin, PhD, of the Salt Lake City VA Medical Center and the University of Utah for valuable feedback on intervention design and data collection. She received compensation for these contributions.

^✉

Corresponding author.

PMCID: PMC10825788 PMID: 38285565

This cluster randomized clinical trial tests whether the Committing to Choose Wisely behavioral economic intervention could engage primary care clinicians and older patients to reduce low-value care.

Key Points

Question

How can use of low-value care be decreased among older adults while preserving autonomy in decision-making?

Findings

In this stepped-wedge cluster randomized clinical trial including 81 primary care clinicians and 8030 older adults, a behavioral economic intervention that engaged primary care clinicians and older patients reduced low-value care across 3 common clinical situations and increased deintensification of hypoglycemic medications for diabetes.

Meaning

Use of scalable interventions that nudge patients and clinicians to achieve greater value while preserving autonomy in decision-making should be explored more broadly.

Abstract

Importance

Use of low-value care is common among older adults. It is unclear how to best engage clinicians and older patients to decrease use of low-value services.

Objective

To test whether the Committing to Choose Wisely behavioral economic intervention could engage primary care clinicians and older patients to reduce low-value care.

Design, Setting, and Participants

Stepped-wedge cluster randomized clinical trial conducted at 8 primary care clinics of an academic health system and a private group practice between December 12, 2017, and September 4, 2019. Participants were primary care clinicians and older adult patients who had diabetes, insomnia, or anxiety or were eligible for prostate cancer screening. Data analysis was performed from October 2019 to November 2023.

Intervention

Clinicians were invited to commit in writing to Choosing Wisely recommendations for older patients to avoid use of hypoglycemic medications to achieve tight glycemic control, sedative-hypnotic medications for insomnia or anxiety, and prostate-specific antigen tests to screen for prostate cancer. Committed clinicians had their photographs displayed on clinic posters and received weekly emails with alternatives to these low-value services. Educational handouts were mailed to applicable patients before scheduled visits and available at the point of care.

Main Outcomes and Measures

Patient-months with a low-value service across conditions (primary outcome) and separately for each condition (secondary outcomes). For patients with diabetes, or insomnia or anxiety, secondary outcomes were patient-months in which targeted medications were decreased or stopped (ie, deintensified).

Results

The study included 81 primary care clinicians and 8030 older adult patients (mean [SD] age, 75.1 [7.2] years; 4076 men [50.8%] and 3954 women [49.2%]). Across conditions, a low-value service was used in 7627 of the 37 116 control patient-months (20.5%) and 7416 of the 46 381 intervention patient-months (16.0%) (adjusted odds ratio, 0.79; 95% CI, 0.65-0.97). For each individual condition, there were no significant differences between the control and intervention periods in the odds of patient-months with a low-value service. The intervention increased the odds of deintensification of hypoglycemic medications for diabetes (adjusted odds ratio, 1.85; 95% CI, 1.06-3.24) but not sedative-hypnotic medications for insomnia or anxiety.

Conclusions and Relevance

In this stepped-wedge cluster randomized clinical trial, the Committing to Choose Wisely behavioral economic intervention reduced low-value care across 3 common clinical situations and increased deintensification of hypoglycemic medications for diabetes. Use of scalable interventions that nudge patients and clinicians to achieve greater value while preserving autonomy in decision-making should be explored more broadly.

Trial Registration

ClinicalTrials.gov Identifier: NCT03411525

Introduction

Low-value services constitute health care that does not improve patient outcomes and can cause unnecessary harms.¹ In the US, delivery of low-value services is common^2,3 and costly,^4,5 particularly among older adults.^6,7,8 One evidence-based effort to raise awareness of low-value services is the Choosing Wisely campaign, which started in the US in 2012 and has spread worldwide.⁹ However, awareness of evidence is often insufficient to change clinical decisions.^10,11 Furthermore, getting clinicians to stop an action may require different approaches than getting clinicians to initiate a practice.^12,13,14 Therefore, complementary strategies are needed to target factors that drive decisions to use low-value services and patient desires to receive them.¹⁵

For example, even when clinicians are aware that specific services may not improve outcomes and could cause harm, the need to make decisions quickly during a busy and cognitively demanding clinical encounter can prompt reflexive decisions to order tests or treatments.¹⁶ Rushed decisions about low-value services can also be more susceptible to patient demands and perceived social norms that more care is better.¹⁷ Furthermore, a busy clinic encounter may not allow time to discuss scaling back tests or treatments.¹⁸

Three promising, complementary solutions to this problem can be found in the field of behavioral economics: supporting deliberative thinking,¹⁹ using nonbinding precommitment,²⁰ and providing social information about what peers decided to do.²¹ First, quick decisions often dominate during busy encounters when clinicians are multitasking and are likely to result in more decisions that diverge from evidence, whereas deliberative thinking may be more common before encounters when clinicians review patient data to form preliminary management plans and is more likely to produce decisions that align with evidence.²² Therefore, if clinicians initiated more decision-making before encounters, they might be more likely to adhere to evidence-based recommendations to avoid low-value services. Second, although initiating clinical decisions before encounters could facilitate avoidance of low-value services, clinicians often need more data from a patient before making their final decision, and patients also want to be involved in the decision process. To overcome these challenges, before encounters, clinicians could be asked to commit their future selves (ie, precommit) to an intended course of action when presented with a particular clinical situation.^20,23,24,25 Adherence to such commitments could be fostered by point-of-care reminders and patient-facing materials to optimize conversations about low-value services. Third, prevailing social norms can influence a range of behaviors.^{26,27,28,29,30,31,32} Thus, signaling social norms of peer clinicians to avoid use of low-value services could further facilitate achievement of this outcome. The objective of this study was to test whether a scalable behavioral economic intervention that supported deliberative thinking, solicited nonbinding precommitment, and provided information about social norms could reduce low-value care.

Methods

Study Design and Recruitment

We conducted this study between December 12, 2017, and September 4, 2019, in 8 primary care clinics of IHA, a private multispecialty group practice, and Michigan Medicine, the health system of the University of Michigan (trial protocol in Supplement 1). We used a stepped-wedge cluster randomized clinical trial design (Figure 1)^33,34 because all clinics could receive the intervention, and implementing the intervention in 1 clinic per month was more feasible than simultaneous implementation in multiple clinics as in a parallel-design cluster randomized clinical trial.

Clinics were selected in collaboration with leaders of each health system to maximize the number of eligible clinicians and applicable patients. All clinics that were invited agreed to participate. Recruitment and the intervention were rolled out to a new clinic each month in a randomly assigned order determined by the study statistician (H.M.K.). Within each clinic, the physicians, nurse practitioners, and physician assistants were invited to participate. Clinicians were excluded if they did not anticipate working in the clinic for the duration of the study. Approximately 2 weeks before each clinic’s intervention period, a study investigator (J.T.K.) visited the clinic to give a presentation about the study and invite participation. Of the 91 clinicians who were invited, 81 provided written informed consent to participate (Figure 2).^34,35 The study was approved by the University of Michigan Medical School Institutional Review Board.

Figure 2. — ^aThe control, transition, and intervention periods followed the approximate schedule shown in the figure with the actual time from the start of the control period (December 15, 2017) to the end of the intervention period (November 1, 2018), spanning roughly 11 months due to scheduling constraints (eg, for recruitment meetings) in each clinic.

^bData for the months of the recruitment and precommitment processes (ie, the transition months) in each clinic were excluded from analyses because clinicians’ behavior could be influenced by (1) prerecruitment knowledge that study recruitment would be starting soon and (2) postcommitment knowledge that the intervention would be starting soon.^34,35 The transition period commenced when the clinic’s medical director was notified that their clinic would be next in the randomly assigned order for recruitment and the intervention. The vertical line in the boxes for these months indicates the approximate time of the recruitment and precommitment processes in each respective clinic.

^cEach clinic (cluster) received 1 to 8 months of the intervention.

Intervention

To promote deliberative thinking about low-value services, clinicians were asked to make a written commitment (eFigure 1 in Supplement 2) to follow 3 Choosing Wisely recommendations: (1) avoid using medications other than metformin to achieve a hemoglobin A_1c (HbA_1c) level less than 7.0% in most older adults with diabetes; (2) do not use benzodiazepines or sedative-hypnotics in older adults as the first choice for insomnia, agitation, or anxiety; and (3) do not routinely perform prostate-specific antigen (PSA)–based screening for prostate cancer in older men. These recommendations were selected because they focused on primary care services that are common in the US,^36,37,38 involved clinician and patient input in decision-making, and represented local opportunities for improvement based on administrative data and feedback from health system leaders.

Before the start of the study, these recommendations were modified slightly based on feedback from clinicians and leaders in each health system. Specifically, the HbA_1c threshold in the American Geriatrics Society (AGS)³⁹ diabetes recommendation was changed from 7.5% to 7.0% to better align with the health systems’ existing quality measures. The American College of Preventive Medicine⁴⁰ recommendation against PSA testing to screen for prostate cancer was narrowed to focus on men age 75 years and older due to a newer recommendation from the US Preventive Services Task Force for shared decision-making around PSA testing for men aged 55 to 69 years⁴¹ and because no professional guidelines recommend PSA screening after age 75 years.⁴² The AGS recommendation around overuse of benzodiazepines and sedative-hypnotics³⁹ was modified to focus on just insomnia and anxiety to better align with the availability of administrative data.

Clinicians were asked to make this written commitment (eFigure 1 in Supplement 2) immediately after a brief presentation about the 3 Choosing Wisely recommendations and written informed consent. To remind clinicians of their commitment and signal social norms to avoid use of low-value services, photographs of committed clinicians appeared on posters in public waiting areas and examination rooms (example in eFigure 2 in Supplement 2). To support point-of-care deliberative thinking and adherence to written commitments, committed clinicians’ scheduled patients for whom any of the 3 recommendations might apply were identified in administrative data (see Study Cohorts) and in advance of their appointment were mailed an applicable Consumer Reports⁴³ Choosing Wisely patient education handout (eFigures 3-5 in Supplement 2) along with brief information about the study. These mailed handouts were also available to clinicians and patients at the point of care. To further support point-of-care deliberative thinking and adherence to written commitments, committed clinicians received a weekly email with strategies to avoid use of low-value services during patient encounters (eFigures 6-9 in Supplement 2). All intervention components continued throughout each clinic’s intervention period. Clinicians who consented to study participation but chose not to make the written commitment received no further intervention.

Study Cohorts

The patient cohorts for mailing of handouts and data analysis were defined using administrative data and diagnostic codes assigned to any encounter in the 12 months before the start of the control period (eTable 1 in Supplement 2). For the diabetes cohort, inclusion criteria were being 65 years or older and having an International Classification of Diseases, Ninth Revision (ICD-9) or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) code for type 2 diabetes. For the insomnia/anxiety cohort, inclusion criteria were being 65 years or older and having an ICD-9 or ICD-10 code for insomnia or anxiety. For the prostate cancer screening cohort, inclusion criteria were being male and 75 years or older, and the exclusion criterion was having an ICD-9 or ICD-10 code for prostate cancer.

Outcomes

The primary outcome was patient-months with low-value care across the 3 patient cohorts as defined in the 3 Choosing Wisely guidelines. We chose patient-months of low-value care as the main outcome measure because use of low-value care can be dynamic over time for individual patients, particularly for management of chronic conditions. Secondary outcomes were within-cohort patient-months with low-value care and, for the diabetes and insomnia/anxiety cohorts, patient-months with deintensification and intensification of the targeted medications. Outcome data were collected by trained medical record abstracters who had access to the full electronic health records (EHRs) and were blinded to the control and intervention periods.

Diabetes Cohort

Low-value care was defined as a patient-month in which a hypoglycemic medication (ie, any medication for diabetes other than metformin; eTable 1 in Supplement 2) was continued at the same dose, newly started, or increased after an HbA_1c level less than 7.0%. If a hypoglycemic medication was decreased or stopped after an HbA_1c level less than 7.0% but another hypoglycemic medication was started or increased, this was considered a substitution and classified as low-value care. Patient-months without any preceding HbA_1c value were excluded from analyses.

Deintensification was defined as a patient-month in which a hypoglycemic medication was decreased or stopped after an HbA_1c level less than 7.0%, without a start or increase of a hypoglycemic medication in the same patient-month. Patient-months were considered eligible for deintensification whenever the most recent HbA_1c level was less than 7.0% and the patient was receiving a hypoglycemic medication that had not been deintensified. When deintensification occurred, that and all subsequent patient-months were considered ineligible for deintensification until another HbA_1c level less than 7.0% or the start or increase of a hypoglycemic medication.

Intensification was defined as a patient-month in which a hypoglycemic medication was started or increased after an HbA_1c level was less than 7.0%. All patient-months in which the most recent HbA_1c level was less than 7.0% were considered eligible for intensification.

Insomnia/Anxiety Cohort

Low-value care was defined as a patient month in which a patient was receiving a benzodiazepine or sedative-hypnotic medication (eTable 1 in Supplement 2). Deintensification was defined as a dose decrease or stoppage of a benzodiazepine or sedative-hypnotic medication without a dose increase or start of a similar medication. Intensification was defined as a dose increase or start of a benzodiazepine or sedative-hypnotic medication without a dose decrease or stoppage of another similar medication.

Prostate Cancer Screening Cohort

Low-value care was defined as a patient-month in which a patient had a screening PSA test. Whether a PSA test was screening or diagnostic (ie, ordered to evaluate symptoms) was determined by the blinded medical record abstracters using criteria developed by research team clinicians (J.T.K., E.A.K.). When a medical record abstracter was unable to determine whether a PSA test should be classified as a screening or diagnostic test, the classification was made by a research team clinician. All patient-months were considered eligible for low-value care, but once a patient had a diagnostic PSA test, that and all subsequent patient-months were excluded from analyses.

Statistical Analysis

Patient demographic characteristics were summarized for each cohort and health system. Because the composition of the diabetes and prostate cancer screening cohorts could change over time as patients met or did not meet inclusion criteria for analyses, patient characteristics were also summarized separately for the control and intervention periods.

To examine effects of the intervention on the primary outcome of patient-months with low-value care across the 3 cohorts, we fit a generalized linear mixed-effects model with logit link in which the dependent variable was patient-months with low-value care. We used a mixed-effects model to assess, across and within the 3 study cohorts, participant-specific differences in the odds of receipt of low-value care during intervention months compared to control months, rather than population-based differences.⁴⁴ The model included clinics as fixed effects and patients as random intercepts to account for correlation of longitudinal data within patients. Additional independent variables included patient cohort, age, gender, race and ethnicity (based on EHR data),⁴⁵ and time in months since study initiation to account for secular trends. The primary predictor variable was the binary time-dependent indicator for an intervention month. The prespecified study protocol called for inclusion of data from clinician surveys in models, but these data were not ultimately used to maximize study power in the setting of incomplete collection of clinician survey data.

Secondary analyses compared the proportion of patient-months with low-value care separately for each of the 3 study cohorts, as well as the proportion of patient-months with deintensification and intensification separately for the diabetes and insomnia/anxiety cohorts. For secondary analyses, we fit similar separate models for each of the 3 study cohorts in which the dependent variable was low-value patient-months or, for the diabetes and insomnia/anxiety cohorts, patient-months with deintensification and intensification. To examine whether intervention effects differed over time, in exploratory analyses we added to the models an interaction term of intervention period by months since study start.

We conducted 3 sensitivity analyses for the diabetes cohort to evaluate the robustness of findings to prespecified analytic decisions. In the first, an HbA_1c level of 7.5% was used as the threshold for outcomes. In the second, an HbA_1c level of 7.0% was used, and patient-months without a preceding HbA_1c value were included and classified as not having low-value care. In the third, demographic characteristics were omitted as independent variables in the model.

We designed the study to have greater than 80% power to detect with .05-level 2-sided tests a reduction in patients-months with low-value care from 20.0% to 15.3% for diabetes, 15.0% to 10.7% for insomnia and anxiety, and 3.0% to 1.0% for prostate cancer screening, with an average of 11 clinicians in each of the 8 clinics and .05 within-clinician correlation. Analyses were conducted using Stata, version 16.0 (StataCorp LLC). Reporting followed the Consolidated Standards of Reporting Trials Extension (CONSORT Extension) reporting guideline for stepped-wedge cluster randomized clinical trials.

Results

Patient Characteristics

Of the 81 clinicians who participated, all made a written commitment to following the 3 Choosing Wisely recommendations, and 72 had applicable encounters during the study period. For the 8030 patients that composed the 3 cohorts, the mean (SD) age was 75.1 (7.2) years; 4076 (50.8%) were men and 3954 (49.2%) were women. For the diabetes cohort, there were 2680 patients, and the mean age and percentage of patients receiving any hypoglycemic medication at baseline (regardless of whether this was classified as low-value care) differed across the 2 health systems (Table 1). For the insomnia/anxiety cohort, there were 4226 patients, and the mean age and percentage of patients who were taking a benzodiazepine or sedative-hypnotic medication at baseline differed across the 2 health systems. For the prostate cancer screening cohort, there were 2342 patients, and the mean age differed across the 2 health systems. For all 3 cohorts, the percentages of patients in racial and ethnic groups differed across the 2 health systems.

Table 1. Patient Characteristics.

Condition	Patients, No. (%)				P value^a
	Health system 1		Health system 2
	Control	Intervention	Control	Intervention
Diabetes cohort
Sample size, No.^b	458	781	1415	1899	NA
Hypoglycemic medication^c	183 (40.0)	278 (35.6)	664 (46.9)	834 (43.9)	.009
Age, mean (SD), y	73.5 (6.5)	73.5 (6.5)	74.3 (6.8)	74.5 (6.9)	.03
Gender
Men	227 (49.6)	390 (49.9)	718 (50.7)	976 (51.4)	.66
Women	231 (50.4)	391 (50.1)	697 (49.3)	923 (48.6)	.66
Race and ethnicity group^d
Asian, non-Hispanic	35 (7.6)	68 (8.7)	41 (2.9)	52 (2.7)	<.001
Black, non-Hispanic	50 (10.9)	85 (10.9)	237 (16.8)	292 (15.4)
Hispanic	7 (1.5)	9 (1.2)	12 (0.9)	15 (0.8)
White, non-Hispanic	346 (75.6)	590 (75.5)	1007 (71.2)	1381 (72.7)
Other race, non-Hispanic	10 (2.2)	16 (2.1)	23 (1.6)	36 (1.9)
Unknown race and ethnicity	10 (2.2)	13 (1.7)	95 (6.7)	123 (6.5)
Insomnia/anxiety cohort
Sample size, No.	2185	2185	2041	2041	NA
Sedative-hypnotic medication^e	392 (17.9)	392 (17.9)	784 (38.4)	784 (38.4)	<.001
Age, mean (SD), y	72.6 (6.3)	72.6 (6.3)	73.9 (7.4)	73.9 (7.4)	<.001
Gender
Men	683 (31.3)	683 (31.3)	588 (28.8)	588 (28.8)	.08
Women	1502 (68.7)	1502 (68.7)	1453 (71.2)	1453 (71.2)	.08
Race and ethnicity group^d
Asian, non-Hispanic	78 (3.6)	78 (3.6)	34 (1.7)	34 (1.7)	<.001
Black, non-Hispanic	87 (4.0)	87 (4.0)	173 (8.5)	173 (8.5)
Hispanic	19 (0.9)	19 (0.9)	7 (0.3)	7 (0.3)
White, non-Hispanic	1965 (89.9)	1965 (89.9)	1686 (82.6)	1686 (82.6)
Other race, non-Hispanic	25 (1.1)	25 (1.1)	36 (1.8)	36 (1.8)
Unknown race and ethnicity	11 (0.5)	11 (0.5)	105 (5.1)	105 (5.1)
Prostate cancer screening cohort^f
Sample size, No.	751	736	1591	1543	NA
Age, mean (SD), y	79.9 (4.3)	80.0 (4.3)	81.2 (5.3)	81.3 (5.3)	<.001
Race and ethnicity group^d
Asian, non-Hispanic	74 (9.9)	71 (9.7)	29 (1.8)	28 (1.8)	<.001
Black, non-Hispanic	27 (3.6)	27 (3.7)	116 (7.3)	112 (7.3)
Hispanic	13 (1.7)	12 (1.6)	14 (0.9)	14 (0.9)
White, non-Hispanic	625 (83.2)	614 (83.4)	1302 (81.8)	1264 (81.9)
Other race, non-Hispanic	11 (1.5)	11 (1.5)	39 (2.5)	38 (2.5)
Unknown race and ethnicity	1 (0.1)	1 (0.1)	91 (5.7)	87 (5.6)

Open in a new tab

Abbreviation: NA, not applicable.

^{^a}

Comparison between 2 health systems during control period.

^{^b}

Excluding patient-months without any preceding hemoglobin A_1c values, resulting in different numbers of patients during the control and intervention periods.

^{^c}

Use of at least 1 hypoglycemic medication (defined as any antidiabetic medication other than metformin) at the start of the corresponding study period, regardless of whether this medication use was ultimately classified as low-value care.

^{^d}

Race and ethnicity groups as classified from electronic health record data. The unknown race and ethnicity category included individuals for whom race and ethnicity data were not available.

^{^e}

Use of 1 or more benzodiazepine or sedative-hypnotic medications at the start of the corresponding study period.

^{^f}

Patient-months are censored at the time of a diagnostic prostate-specific antigen test, resulting in different numbers of patients during the control and intervention periods.

Low-Value Care

Across the 3 patient cohorts, we observed low-value care in 7627 of the 37 116 control patient-months (20.5%) and 7416 of the 46 381 intervention patient-months (16.0%) (Table 2 and Figure 3⁴⁶). Overall, 554 patients in the diabetes cohort (20.7%), 1362 patients in the insomnia/anxiety cohort (32.2%), and 143 patients in the prostate cancer screening cohort (6.1%) received low-value care.

Table 2. Outcomes for Diabetes, Insomnia/Anxiety, and Prostate Cancer Screening Cohorts.

Cohort	Control months		Intervention months		aOR (95% CI)
Cohort	%	No. of patient-months meeting outcome criteria/total No. of patients-months at risk for outcome	%	No. of patient-months meeting outcome criteria/total No. of patients-months at risk for outcome	aOR (95% CI)
All 3 cohorts combined, low-value care	20.5	7627/37 116	16.0	7416/46 381	0.79 (0.65-0.97)^a
Diabetes cohort^b
Low-value care^c	19.6	1345/6864	14.1	1769/12 516	0.69 (0.45-1.05)^d
Deintensification^e	5.6	90/1611	5.4	152/2791	1.85 (1.06-3.24)^d
Intensification^f	1.2	56/4515	0.6	55/8527	0.71 (0.32-1.58)^d
Insomnia/anxiety cohort
Low-value care	32.1	6180/19 223	24.9	5603/22 470	0.77 (0.53-1.13)^d
Deintensification^g	2.7	174/6354	2.8	161/5767	0.84 (0.53-1.33)^d
Intensification^h	0.9	167/19 223	0.7	155/22 470	1.04 (0.66-1.66)^d
Prostate cancer screening cohort, low-value care	0.9	102/11 029	0.4	44/11 395	0.69 (0.34-1.40)^d

Open in a new tab

Abbreviations: aOR, adjusted odds ratio from multivariable model; HbA_1c, hemoglobin A_1c.

^{^a}

Odds ratio associated with intervention period overall for the outcome of interest based on a hierarchical logistic regression model, adjusting for study cohort, intervention period, time in months, age, racial and ethnic minority status, study clinic, and patients as random intercepts.

^{^b}

Excluding patient-months without preceding HbA_1c values and using an HbA_1c threshold of 7%.

^{^c}

Patient-months in which, after an HbA_1c level was less than 7.0%, a hypoglycemic medication (ie, any medication for diabetes other than metformin) was continued at the same dose, newly started, or increased; or a hypoglycemic medication was decreased or stopped but another hypoglycemic medication was started or increased.

^{^d}

Odds ratio associated with intervention period overall for the outcome of interest based on a hierarchical logistic regression model, adjusting for intervention period, time in months, age, gender (except for prostate cancer screening cohort analysis), racial and ethnic minority status, study clinic, and patients as random intercepts.

^{^e}

Patient-months in which the most recent HbA_1c level was less than 7.0% and there was a discontinuation or dose decrease of a hypoglycemic medication without any new start or dose increase of another hypoglycemic medication.

^{^f}

Patient-months in which the most recent HbA_1c level was less than 7.0% and there was a new start or dose increase of a hypoglycemic medication without any discontinuation or dose decrease of another hypoglycemic medication.

^{^g}

Patient-months in which there was a discontinuation or dose decrease of a benzodiazepine or sedative-hypnotic medication without any new start or dose increase of another benzodiazepine or sedative-hypnotic medication.

^{^h}

Patient-months in which there was a new start or dose increase of a benzodiazepine or sedative-hypnotic medication without any discontinuation or dose decrease of another benzodiazepine or sedative-hypnotic medication.

Figure 3. — Vertical bars represent separately for control and intervention clinics the monthly percentage of patients in the diabetes, insomnia/anxiety, and prostate cancer screening cohorts who received low-value care as defined for that cohort.

Across cohorts, the adjusted odds of patient-months with low-value care were lower in the intervention period compared to the control period (adjusted odds ratio [aOR], 0.79; 95% CI, 0.65-0.97). In an exploratory analysis (eTable 2 in Supplement 2), the intervention was less effective over time (aOR for the intervention by time interaction, 1.06; 95% CI, 1.01-1.12).

For each patient cohort, the point estimate for the aOR was similar to the pooled estimate, but there were no statistically significant differences between the control and intervention periods in the odds of patient-months with low-value care (Table 2; eFigures 10-12 in Supplement 2). In exploratory analyses (eTable 2 in Supplement 2), the intervention was less effective over time for the insomnia/anxiety cohort (aOR for the intervention by time interaction, 1.22; 95% CI, 1.11-1.33) but not for the diabetes or prostate cancer screening cohorts.

Medication Deintensification and Intensification

In the diabetes cohort, the percentage of patient-months with deintensification was slightly higher during control months (90 of 1611 [5.6%]) compared to intervention months (152 of 2791 [5.4%]). However, in our data we observed a secular trend toward less deintensification over the study period (eFigure 13 in Supplement 2). In multivariable analyses that adjusted for this secular trend (Table 2), the odds of deintensification of hypoglycemic medications were higher during intervention patient-months than control patient-months (aOR, 1.85; 95% CI, 1.06-3.24). For the insomnia/anxiety cohort, medication deintensification did not differ between the intervention and control periods (Table 2; eFigure 14 in Supplement 2). For both the diabetes and insomnia/anxiety cohorts, there were no significant differences in medication intensification between the intervention and control periods (Table 2; eFigures 15 and 16 in Supplement 2).

Sensitivity Analyses

In the diabetes cohort, when an HbA_1c level of 7.5% was used as the threshold for outcomes, the adjusted odds of patient-months with low-value care were lower in the intervention period compared to the control period, but the difference was not significant (eTable 3 in Supplement 2). The adjusted odds of patient-months with low-value care were lower in the intervention period compared to the control period when patient-months without a preceding HbA_1c value were included and classified as not having low-value care (aOR, 0.67; 95% CI, 0.51-0.89) and when patient demographics were omitted as independent variables (aOR, 0.63; 95% CI, 0.42-0.95). In all sensitivity analyses, the odds of deintensification of hypoglycemic medications were higher during intervention patient-months than control patient-months.

Discussion

The multicomponent Committing to Choose Wisely behavioral economic intervention engaged primary care clinicians and older patients to reduce low-value care across 3 common clinical situations and led to more deintensification of hypoglycemic medications for diabetes. The intervention did not lead to statistically significant reductions in use of low-value care for any individual patient cohort, likely due to the lower power to find relatively small within-cohort intervention effects, although results for the diabetes cohort were sensitive to prespecified analytic decisions. The intervention did not significantly change use of low-value medications for insomnia or anxiety.

Interventions to reduce use of low-value care often use clinician and patient education⁴⁷ but are most likely to be effective when combined with other elements in multicomponent interventions.⁴⁸ Our findings align with this evidence and also build on previous studies of behavioral economic interventions to reduce use of low-value care. For example, accountable justification, peer comparison, and poster-sized commitment letters have significantly reduced inappropriate antibiotic prescriptions.^49,50 Peer comparison and individual audit and feedback have reduced the number of pills per opioid prescription for acute pain.⁵¹ Clinician precommitment with point-of-care reminders has reduced unnecessary imaging tests for low back pain.³⁴ Our approach incorporated some of these strategies yet provides a broader intervention template (ie, matching local improvement opportunities with professional guidelines; soliciting clinician precommitment to these guidelines; and point-of-care supports to promote deliberative thinking, commitment adherence, and social norms) that could be adapted to target different types of low-value care in other settings. Because some exploratory analyses showed that the intervention was less effective over time, more research is needed to determine how interventions that seek to reduce use of low-value care can remain salient in busy practice environments.

Although in our prespecified analyses the intervention did not lead to a significant reduction in use of low-value care for older patients with diabetes, our prespecified and sensitivity analyses all showed that the intervention led to more deintensification of hypoglycemic medications for diabetes. This increased deintensification is important because overtreatment of diabetes is common among older adults^36,52 and infrequently addressed despite its potential for harms.^53,54,55,56 Other studies have sought to reduce diabetes overtreatment among older adults through behavioral economic interventions embedded within an EHR.^57,58,59,60 Our findings highlight the promise of behavioral economic strategies to address this prevalent clinical problem but offer an alternative approach that engages both clinicians and patients without requiring EHR modifications.

In contrast, the Committing to Choose Wisely intervention did not significantly change use of benzodiazepines and sedative-hypnotic medications for insomnia or anxiety. This finding could indicate the greater difficulty that primary care clinicians and patients may have in avoiding use of low-value services for symptomatic conditions relative to asymptomatic conditions.⁶¹ Interventions that used clinician education, patient education, or both have significantly reduced use of benzodiazepines among older patients.^{62,63,64,65,66,67} Little research has tested behavioral economic approaches to reduce overuse of benzodiazepines,⁶⁸ and more studies are needed to examine whether such interventions can effectively reduce this type of low-value care.

Limitations

Our study had limitations. We selected and slightly modified for this study 3 Choosing Wisely nonacute care recommendations for older adults and targeted these recommendations with a relatively scalable intervention in academic and community health systems. However, findings may not generalize to all other Choosing Wisely recommendations or practice settings and could be less scalable in some environments. Because cohorts for mailing of patient handouts and data analysis were defined using administrative data and diagnostic codes from encounters in the 12 months before the start of the control period, we were unable to examine effects of the intervention on patients who were newly diagnosed with an applicable condition during the study. We measured outcomes through detailed medical record reviews but lacked data to evaluate the true value of each service for each individual patient. We tested a multicomponent intervention and were not able to isolate the effects of each component. Although the examined medications included both sodium–glucose cotransporter-2 inhibitors and glucagonlike peptide-1 receptor agonists (eTable 1 in Supplement 2), at the time of the study, use of these medications was much less common than it is now.^69,70 This analysis does not use data from surveys and interviews that were conducted with patients and clinicians, but future analyses of these data will yield key insights into experiences with efforts to reduce use of low-value care.

Conclusions

The results of this cluster randomized clinical trial have important implications for clinicians and policymakers. Overuse of low-value care is a worldwide phenomenon^35,71 in need of solutions that can be scaled and sustained across practice environments. Our Committing to Choose Wisely behavioral economic intervention engaged primary care clinicians and older patients to reduce low-value care across 3 common clinical situations and provides a multicomponent template that could be adapted to target use of low-value care in a range of settings. Use of such scalable interventions that nudge patients and clinicians to achieve greater value while preserving autonomy in decision-making should be explored broadly.

Supplement 1.

Trial Protocol

jamainternmed-e237703-s001.pdf^{(226.6KB, pdf)}

Supplement 2.

eFigure 1. Commitment Opportunity Form

eFigure 2. Example Clinic Poster

eFigure 3. Diabetes Patient Education Handout

eFigure 4. Insomnia/Anxiety Patient Education Handout

eFigure 5. Prostate Cancer Screening Patient Education Handout

eFigure 6. Clinician Weekly Resource Email 1

eFigure 7. Clinician Weekly Resource Email 2

eFigure 8. Clinician Weekly Resource Email 3

eFigure 9. Clinician Weekly Resource Email 4

eFigure 10. Low-Value Care by Study Month for Diabetes Cohort

eFigure 11. Low-Value Care by Study Month for Insomnia/Anxiety Cohort

eFigure 12. Low-Value Care by Study Month for Prostate Cancer Screening Cohort

eFigure 13. Deintensification by Study Month for Diabetes Cohort

eFigure 14. Deintensification by Study Month for Insomnia/Anxiety Cohort

eFigure 15. Intensification by Study Month for Diabetes Cohort

eFigure 16. Intensification by Study Month for Insomnia/Anxiety Cohort

eTable 1. ICD-9 Codes, ICD-10 Codes, and Prescription Medications

eTable 2. Study Outcomes with Intervention*Time Interaction in Regression Models

eTable 3. Sensitivity Analysis for Diabetes Cohort

jamainternmed-e237703-s002.pdf^{(6MB, pdf)}

Supplement 3.

Data Sharing Statement

jamainternmed-e237703-s003.pdf^{(16.6KB, pdf)}

References

1.Blumenthal-Barby JS. “Choosing Wisely” to reduce low-value care: a conceptual and ethical analysis. J Med Philos. 2013;38(5):559-580. doi: 10.1093/jmp/jht042 [DOI] [PubMed] [Google Scholar]
2.Colla CH, Morden NE, Sequist TD, Schpero WL, Rosenthal MB. Choosing Wisely: prevalence and correlates of low-value health care services in the United States. J Gen Intern Med. 2015;30(2):221-228. doi: 10.1007/s11606-014-3070-z [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Chua KP, Schwartz AL, Volerman A, Conti RM, Huang ES. Differences in the receipt of low-value services between publicly and privately insured children. Pediatrics. 2020;145(2):e20192325. doi: 10.1542/peds.2019-2325 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Schwartz AL, Landon BE, Elshaug AG, Chernew ME, McWilliams JM. Measuring low-value care in Medicare. JAMA Intern Med. 2014;174(7):1067-1076. doi: 10.1001/jamainternmed.2014.1541 [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Carter EA, Morin PE, Lind KD. Costs and trends in utilization of low-value services among older adults with commercial insurance or Medicare Advantage. Med Care. 2017;55(11):931-939. doi: 10.1097/MLR.0000000000000809 [DOI] [PubMed] [Google Scholar]
6.Mafi JN, Reid RO, Baseman LH, et al. Trends in low-value health service use and spending in the US Medicare fee-for-service program, 2014-2018. JAMA Netw Open. 2021;4(2):e2037328. doi: 10.1001/jamanetworkopen.2020.37328 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Park S, Jung J, Burke RE, Larson EB. Trends in use of low-value care in traditional fee-for-service Medicare and Medicare Advantage. JAMA Netw Open. 2021;4(3):e211762. doi: 10.1001/jamanetworkopen.2021.1762 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Oronce CIA, Fendrick AM, Ladapo JA, Sarkisian C, Mafi JN. The utilization and costs of Grade D USPSTF services in Medicare, 2007-2016. J Gen Intern Med. 2021;36(12):3711-3718. doi: 10.1007/s11606-021-06784-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Bhatia RS, Levinson W, Shortt S, et al. Measuring the effect of Choosing Wisely: an integrated framework to assess campaign impact on low-value care. BMJ Qual Saf. 2015;24(8):523-531. doi: 10.1136/bmjqs-2015-004070 [DOI] [PubMed] [Google Scholar]
10.Cabana MD, Rand CS, Powe NR, et al. Why don’t physicians follow clinical practice guidelines? a framework for improvement. JAMA. 1999;282(15):1458-1465. doi: 10.1001/jama.282.15.1458 [DOI] [PubMed] [Google Scholar]
11.Grol R, Grimshaw J. From best evidence to best practice: effective implementation of change in patients’ care. Lancet. 2003;362(9391):1225-1230. doi: 10.1016/S0140-6736(03)14546-1 [DOI] [PubMed] [Google Scholar]
12.van Bodegom-Vos L, Davidoff F, Marang-van de Mheen PJ. Implementation and de-implementation: two sides of the same coin? BMJ Qual Saf. 2017;26(6):495-501. doi: 10.1136/bmjqs-2016-005473 [DOI] [PubMed] [Google Scholar]
13.Davidson KW, Ye S, Mensah GA. Commentary: de-implementation science: a virtuous cycle of ceasing and desisting low-value care before implementing new high value care. Ethn Dis. 2017;27(4):463-468. doi: 10.18865/ed.27.4.463 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Steinman MA, Boyd CM, Spar MJ, Norton JD, Tannenbaum C. Deprescribing and deimplementation: time for transformative change. J Am Geriatr Soc. 2021;69(12):3693-3695. doi: 10.1111/jgs.17441 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Armstrong D. Clinical autonomy, individual and collective: the problem of changing doctors’ behaviour. Soc Sci Med. 2002;55(10):1771-1777. doi: 10.1016/S0277-9536(01)00309-4 [DOI] [PubMed] [Google Scholar]
16.Baron RJ. What’s keeping us so busy in primary care? a snapshot from one practice. N Engl J Med. 2010;362(17):1632-1636. doi: 10.1056/NEJMon0910793 [DOI] [PubMed] [Google Scholar]
17.Hardin G. The tragedy of the commons: the population problem has no technical solution; it requires a fundamental extension in morality. Science. 1968;162(3859):1243-1248. doi: 10.1126/science.162.3859.1243 [DOI] [PubMed] [Google Scholar]
18.Kerr EA, Hofer TP. Deintensification of routine medical services: the next frontier for improving care quality. JAMA Intern Med. 2016;176(7):978-980. doi: 10.1001/jamainternmed.2016.2292 [DOI] [PubMed] [Google Scholar]
19.Kahneman D. Maps of bounded rationality: psychology for behavioral economics. Am Econ Rev. 2003;93(5):1449-1475. doi: 10.1257/000282803322655392 [DOI] [Google Scholar]
20.Ariely D, Wertenbroch K. Procrastination, deadlines, and performance: self-control by precommitment. Psychol Sci. 2002;13(3):219-224. doi: 10.1111/1467-9280.00441 [DOI] [PubMed] [Google Scholar]
21.Krupka E, Weber RA. The focusing and informational effects of norms on pro-social behavior. J Econ Psychol. 2009;30:307-320. doi: 10.1016/j.joep.2008.11.005 [DOI] [Google Scholar]
22.Sheeran P, Gollwitzer PM, Bargh JA. Nonconscious processes and health. Health Psychol. 2013;32(5):460-473. doi: 10.1037/a0029203 [DOI] [PubMed] [Google Scholar]
23.Hsiaw A. Goal-setting and self-control. J Econ Theory. 2013;148(2):601-626. doi: 10.1016/j.jet.2012.08.001 [DOI] [Google Scholar]
24.Hoch SJ, Loewenstein GF. Time-inconsistent preferences and consumer self-control. J Consum Res. 1991;17(4):492-507. doi: 10.1086/208573 [DOI] [Google Scholar]
25.Duckworth AL. The significance of self-control. Proc Natl Acad Sci U S A. 2011;108(7):2639-2640. doi: 10.1073/pnas.1019725108 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Aarts H, Dijksterhuis A. The silence of the library: environment, situational norm, and social behavior. J Pers Soc Psychol. 2003;84(1):18-28. doi: 10.1037/0022-3514.84.1.18 [DOI] [PubMed] [Google Scholar]
27.Kallgren CA, Reno RR, Cialdini RB. A focus theory of normative conduct: when norms do and do not affect behavior. Pers Soc Psychol Bull. 2000;26(8):1002-1012. doi: 10.1177/01461672002610009 [DOI] [Google Scholar]
28.Goldstein NJ, Cialdini RB, Griskevicius V. A room with a viewpoint: using social norms to motivate environmental conservation in hotels. J Consum Res. 2008;35(3):472-482. doi: 10.1086/586910 [DOI] [Google Scholar]
29.Schultz PW. Strategies for promoting proenvironmental behavior. Eur Psychol. 2014;19(2):107-117. doi: 10.1027/1016-9040/a000163 [DOI] [Google Scholar]
30.Neighbors C, Larimer ME, Lewis MA. Targeting misperceptions of descriptive drinking norms: efficacy of a computer-delivered personalized normative feedback intervention. J Consult Clin Psychol. 2004;72(3):434-447. doi: 10.1037/0022-006X.72.3.434 [DOI] [PubMed] [Google Scholar]
31.Larimer ME, Neighbors C. Normative misperception and the impact of descriptive and injunctive norms on college student gambling. Psychol Addict Behav. 2003;17(3):235-243. doi: 10.1037/0893-164X.17.3.235 [DOI] [PubMed] [Google Scholar]
32.Nolan JM, Schultz PW, Cialdini RB, Goldstein NJ, Griskevicius V. Normative social influence is underdetected. Pers Soc Psychol Bull. 2008;34(7):913-923. doi: 10.1177/0146167208316691 [DOI] [PubMed] [Google Scholar]
33.Zwarenstein M, Treweek S, Gagnier JJ, et al. ; CONSORT group; Pragmatic Trials in Healthcare (Practihc) group . Improving the reporting of pragmatic trials: an extension of the CONSORT statement. BMJ. 2008;337(2):a2390. doi: 10.1136/bmj.a2390 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Kullgren JT, Krupka E, Schachter A, et al. Precommitting to choose wisely about low-value services: a stepped wedge cluster randomised trial. BMJ Qual Saf. 2018;27(5):355-364. doi: 10.1136/bmjqs-2017-006699 [DOI] [PubMed] [Google Scholar]
35.Kerr EA, Kullgren JT, Saini SD. Choosing Wisely: how to fulfill the promise in the next 5 years. Health Aff (Millwood). 2017;36(11):2012-2018. doi: 10.1377/hlthaff.2017.0953 [DOI] [PubMed] [Google Scholar]
36.Lipska KJ, Ross JS, Miao Y, Shah ND, Lee SJ, Steinman MA. Potential overtreatment of diabetes mellitus in older adults with tight glycemic control. JAMA Intern Med. 2015;175(3):356-362. doi: 10.1001/jamainternmed.2014.7345 [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Maust DT, Kim HM, Wiechers IR, Ignacio RV, Bohnert ASB, Blow FC. Benzodiazepine use among Medicare, commercially insured, and veteran older adults, 2013-2017. J Am Geriatr Soc. 2021;69(1):98-105. doi: 10.1111/jgs.16825 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Kim DD, Daly AT, Koethe BC, et al. Low-value prostate-specific antigen test for prostate cancer screening and subsequent health care utilization and spending. JAMA Netw Open. 2022;5(11):e2243449. doi: 10.1001/jamanetworkopen.2022.43449 [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Choosing Wisely Workgroup AGS; AGS Choosing Wisely Workgroup . American Geriatrics Society identifies five things that healthcare providers and patients should question. J Am Geriatr Soc. 2013;61(4):622-631. doi: 10.1111/jgs.12226 [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Livingston CJ, Freeman RJ, Mohammad A, et al. ; Choosing Wisely(®) Task Force . Choosing Wisely® in preventive medicine: the American College of Preventive Medicine’s top 5 list of recommendations. Am J Prev Med. 2016;51(1):141-149. doi: 10.1016/j.amepre.2016.03.009 [DOI] [PubMed] [Google Scholar]
41.Grossman DC, Curry SJ, Owens DK, et al. ; US Preventive Services Task Force . Screening for prostate cancer: US Preventive Services Task Force recommendation statement. JAMA. 2018;319(18):1901-1913. doi: 10.1001/jama.2018.3710 [DOI] [PubMed] [Google Scholar]
42.Jaramillo E, Tan A, Yang L, Kuo YF, Goodwin JS. Variation among primary care physicians in prostate-specific antigen screening of older men. JAMA. 2013;310(15):1622-1624. doi: 10.1001/jama.2013.277514 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Santa JS. Communicating information about “what not to do” to consumers. BMC Med Inform Decis Mak. 2013;13(suppl 3):S2. doi: 10.1186/1472-6947-13-S3-S2 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Li F, Hughes JP, Hemming K, Taljaard M, Melnick ER, Heagerty PJ. Mixed-effects models for the design and analysis of stepped wedge cluster randomized trials: an overview. Stat Methods Med Res. 2021;30(2):612-639. doi: 10.1177/0962280220932962 [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Kullgren JT, Malani P, Kirch M, et al. Older adults’ perceptions of overuse. J Gen Intern Med. 2020;35(1):365-367. doi: 10.1007/s11606-019-05434-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Volandes AE, Zupanc SN, Lakin JR, et al. Video intervention and goals-of-care documentation in hospitalized older adults: the VIDEO-PCE randomized clinical trial. JAMA Netw Open. 2023;6(9):e2332556. doi: 10.1001/jamanetworkopen.2023.32556 [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Ingvarsson S, Hasson H, von Thiele Schwarz U, et al. Strategies for de-implementation of low-value care—a scoping review. Implement Sci. 2022;17(1):73. doi: 10.1186/s13012-022-01247-y [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Colla CH, Mainor AJ, Hargreaves C, Sequist T, Morden N. Interventions aimed at reducing use of low-value health services: a systematic review. Med Care Res Rev. 2017;74(5):507-550. doi: 10.1177/1077558716656970 [DOI] [PubMed] [Google Scholar]
49.Meeker D, Linder JA, Fox CR, et al. Effect of behavioral interventions on inappropriate antibiotic prescribing among primary care practices: a randomized clinical trial. JAMA. 2016;315(6):562-570. doi: 10.1001/jama.2016.0275 [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Meeker D, Knight TK, Friedberg MW, et al. Nudging guideline-concordant antibiotic prescribing: a randomized clinical trial. JAMA Intern Med. 2014;174(3):425-431. doi: 10.1001/jamainternmed.2013.14191 [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Navathe AS, Liao JM, Yan XS, et al. The effect of clinician feedback interventions on opioid prescribing. Health Aff (Millwood). 2022;41(3):424-433. doi: 10.1377/hlthaff.2021.01407 [DOI] [PubMed] [Google Scholar]
52.Maciejewski ML, Mi X, Sussman J, et al. Overtreatment and deintensification of diabetic therapy among Medicare beneficiaries. J Gen Intern Med. 2018;33(1):34-41. doi: 10.1007/s11606-017-4167-y [DOI] [PMC free article] [PubMed] [Google Scholar]
53.McAlister FA, Youngson E, Eurich DT. Treatment deintensification is uncommon in adults with type 2 diabetes mellitus: a retrospective cohort study. Circ Cardiovasc Qual Outcomes. 2017;10(4):e003514. doi: 10.1161/CIRCOUTCOMES.116.003514 [DOI] [PubMed] [Google Scholar]
54.Sussman JB, Kerr EA, Saini SD, et al. Rates of deintensification of blood pressure and glycemic medication treatment based on levels of control and life expectancy in older patients with diabetes mellitus. JAMA Intern Med. 2015;175(12):1942-1949. doi: 10.1001/jamainternmed.2015.5110 [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Aubert CE, Lega IC, Bourron O, Train AJ, Kullgren JT. When and how to deintensify type 2 diabetes care. BMJ. 2021;375:e066061. doi: 10.1136/bmj-2021-066061 [DOI] [PubMed] [Google Scholar]
56.Caverly TJ, Fagerlin A, Zikmund-Fisher BJ, et al. Appropriate prescribing for patients with diabetes at high risk for hypoglycemia: national survey of Veterans Affairs health care professionals. JAMA Intern Med. 2015;175(12):1994-1996. doi: 10.1001/jamainternmed.2015.5950 [DOI] [PubMed] [Google Scholar]
57.Rowe TA, Brown T, Lee JY, et al. Development and pilot testing of EHR-nudges to reduce overuse in older primary care patients. Arch Gerontol Geriatr. 2023;104:104794. doi: 10.1016/j.archger.2022.104794 [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Belli HM, Chokshi SK, Hegde R, et al. Implementation of a behavioral economics electronic health record (BE-EHR) module to reduce overtreatment of diabetes in older adults. J Gen Intern Med. 2020;35(11):3254-3261. doi: 10.1007/s11606-020-06119-z [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Belli HM, Troxel AB, Blecker SB, et al. A behavioral economics-electronic health record module to promote appropriate diabetes management in older adults: protocol for a pragmatic cluster randomized controlled trial. JMIR Res Protoc. 2021;10(10):e28723. doi: 10.2196/28723 [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Brown T, Rowe TA, Lee JY, et al. Design of behavioral economic applications to geriatrics leveraging electronic health records (BEAGLE): a pragmatic cluster randomized controlled trial. Contemp Clin Trials. 2022;112:106649. doi: 10.1016/j.cct.2021.106649 [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Zikmund-Fisher BJ, Kullgren JT, Fagerlin A, Klamerus ML, Bernstein SJ, Kerr EA. Perceived barriers to implementing individual Choosing Wisely^® recommendations in two national surveys of primary care providers. J Gen Intern Med. 2017;32(2):210-217. doi: 10.1007/s11606-016-3853-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Tannenbaum C, Martin P, Tamblyn R, Benedetti A, Ahmed S. Reduction of inappropriate benzodiazepine prescriptions among older adults through direct patient education: the EMPOWER cluster randomized trial. JAMA Intern Med. 2014;174(6):890-898. doi: 10.1001/jamainternmed.2014.949 [DOI] [PubMed] [Google Scholar]
63.Vicens C, Sempere E, Bejarano F, et al. Efficacy of two interventions on the discontinuation of benzodiazepines in long-term users: 36-month follow-up of a cluster randomised trial in primary care. Br J Gen Pract. 2016;66(643):e85-e91. doi: 10.3399/bjgp16X683485 [DOI] [PMC free article] [PubMed] [Google Scholar]
64.Del Giorno R, Ottini A, Greco A, et al. Peer-pressure and overuse: the effect of a multimodal approach on variation in benzodiazepine prescriptions in a network of public hospitals. Int J Clin Pract. 2020;74(3):e13448. doi: 10.1111/ijcp.13448 [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Melo TAR, Bezerra CO, Fernandes BD, Rotta I, Reis WCT, Aguiar PM. Pharmacists’ contribution to benzodiazepine deprescribing in older outpatients: a systematic review and meta-analysis. Int J Clin Pharm. 2023;45(5):1037-1049. doi: 10.1007/s11096-023-01637-2 [DOI] [PubMed] [Google Scholar]
66.Van der Linden L, Hias J, Liesenborghs A, et al. The impact of a pharmacist intervention on post-discharge hypnotic drug discontinuation in geriatric inpatients: a before-after study. BMC Geriatr. 2023;23(1):407. doi: 10.1186/s12877-023-04139-y [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Niznik JD, Collins BJ, Armistead LT, et al. Pharmacist interventions to deprescribe opioids and benzodiazepines in older adults: a rapid review. Res Social Adm Pharm. 2022;18(6):2913-2921. doi: 10.1016/j.sapharm.2021.07.012 [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Marti J, Bachhuber M, Feingold J, Meads D, Richards M, Hennessy S. Financial incentives to discontinue long-term benzodiazepine use: a discrete choice experiment investigating patient preferences and willingness to participate. BMJ Open. 2017;7(10):e016229. doi: 10.1136/bmjopen-2017-016229 [DOI] [PMC free article] [PubMed] [Google Scholar]
69.Essien UR, Tang Y, Figueroa JF, et al. Diabetes care among older adults enrolled in Medicare Advantage versus traditional Medicare fee-for-service plans: the Diabetes Collaborative Registry. Diabetes Care. 2022;45(7):1549-1557. doi: 10.2337/dc21-1178 [DOI] [PMC free article] [PubMed] [Google Scholar]
70.Mahtta D, Ramsey DJ, Lee MT, et al. Utilization rates of SGLT2 inhibitors and GLP-1 receptor agonists and their facility-level variation among patients with atherosclerotic cardiovascular disease and type 2 diabetes: insights from the Department of Veterans Affairs. Diabetes Care. 2022;45(2):372-380. doi: 10.2337/dc21-1815 [DOI] [PMC free article] [PubMed] [Google Scholar]
71.Brownlee S, Chalkidou K, Doust J, et al. Evidence for overuse of medical services around the world. Lancet. 2017;390(10090):156-168. doi: 10.1016/S0140-6736(16)32585-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials