A Randomized Controlled Trial of a Behavioral Economic Intervention to Reduce Overuse of Low-Value Services Among Older Adults

From the 2021 HVPAA National Conference

Jeff Kullgren (VA Ann Arbor Healthcare System and University of Michigan), Myra Kim (University of Michigan), Megan Slowey, Joseph Colbert, Barbara Soyster, Stuart Winston, Kerry Ryan, Jane Forman, Melissa Riba, Angela Fagerlin, Erin Krupka, Eve Kerr

Background

The Choosing Wisely® campaign recommends that older patients and their clinicians avoid tests and treatments that do not improve health outcomes and can lead to harms. For example, Choosing Wisely recommends that older patients avoid use of hypoglycemic medications to achieve a hemoglobin A1c < 7.5%, sedative-hypnotic medications for treatment of insomnia or anxiety, and prostate-specific antigen (PSA) tests to screen for prostate cancer. Yet, use of such low-value care remains common among older patients, and it is unclear how to best engage patients and providers to decrease use of these and other low-value services.

Objective

To test whether a behavioral economic intervention that engages both clinicians and older patients can reduce use of hypoglycemic medications for type 2 diabetes, benzodiazepines and sedative-hypnotics for insomnia or anxiety, and PSA tests to screen for prostate cancer.

Methods

We conducted a stepped wedge cluster randomized controlled trial in 8 primary care clinics across 2 health systems. In the intervention, primary care providers (PCPs) were shown the 3 aforementioned Choosing Wisely recommendations and invited to commit themselves to following these recommendations by signing a commitment document. Committed PCPs had their photos displayed on clinic posters and received weekly emails with resources to help them and their patients avoid use of the targeted low-value services. Patient education handouts about these services were mailed to applicable patients prior to scheduled primary care visits and available at the point-of-care.

We conducted chart reviews to collect clinical data on decisions about the targeted low-value services. We used a multivariable generalized linear mixed-effects model with logit link to compare between the control and intervention periods the odds of patient-months in which a low-value service was used across the 3 patient cohorts combined (primary outcome) and separately for each patient cohort (secondary outcome). For patients with diabetes, insomnia, or anxiety, a secondary outcome was the patient-months in which applicable medications were deintensified.

Results

Among older adults with diabetes, with insomnia or anxiety, or who were eligible for prostate cancer screening, a low-value service was used in 20.5% of the 37,116 control period patient-months and 16.0% of the 47,306 intervention period patient-months [adjusted odds ratio (AOR) 0.79, P = 0.03]. For each individual patient cohort, there were no significant differences between the control and intervention periods in the odds of patient-months in which a low-value service was used. The intervention was associated with higher odds of deintensification of hypoglycemic medications for diabetes (AOR 1.85, P = 0.03), but not sedative-hypnotic medications for insomnia or anxiety.

Conclusions

A behavioral economic intervention that engaged both PCPs and their older patients reduced use of low-value services across 3 common conditions and increased deintensification of hypoglycemic medications for diabetes.

Clinical Implications

Use of low-cost, scalable interventions that nudge both patients and providers to achieve greater health care value while preserving autonomy in clinical decision-making should be explored more broadly.