EHR alerts fail to curb low-value test orders

Clinical takeaway: Order blood counts and ECGs for cataract patients only when a specific clinical question calls for them. 

Guidelines that recommend against routine preoperative testing for cataract surgery have done little to change practice, with most patients still receiving at least one test. Health systems that did cut testing paid for it with resource-intensive changes, such as a clinic staffed by associate providers or a redesign of preoperative visits and clinician education. 

Electronic alerts promise a cheaper route, and behavioral nudges have reduced low-value prostate-specific antigen testing, antibiotic prescribing, and opioid prescribing. This pragmatic randomized trial tested whether EHR alerts framed around different patient harms could do the same for preoperative testing. 

Testing remained common, occurring in about eight in 10 preoperative encounters in both arms, with ordering rates declining slightly under both alert and usual-care conditions. No single alert produced a statistically significant change in overall testing versus usual care. But some alerts were more effective than others. A stress-focused alert, which warned physicians that testing can distress patients, was linked to roughly 60% lower odds of blood count and ECG orders, while a financial-harm alert warning of costs to patients was linked to lower blood count orders alone. 

Blood count orders also fell under usual care, to 34.7% from 54.6%. The authors consider these findings hypothesis generating. The reduction in ECG ordering was seen only among physicians who ordered tests at every baseline visit, an exploratory finding. 

UCLA Health, a single health system, randomized 776 physicians to one of three EHR alerts citing society recommendations against routine testing, or to usual care. The 229 physicians with an eligible visit within 30 days of cataract surgery held 1045 encounters with 967 patients. 

The primary outcome was the change in encounters with any signed testing order, a year before versus a year after, in alert arms versus usual care. The authors say the trial may have been underpowered. Emergency department visits, hospitalizations, and deaths within 90 days did not differ between arms. 

Physicians could cancel an order with one click or override the alert by providing a justification, and the investigators propose that the ease of overriding limited the effect. They call for trials that pair alerts with workflow changes, or with a justification that cannot be easily overridden or that enters the permanent record. Payment may matter as well. The health system is mostly fee-for-service, and reductions in low-value care may come more readily under accountable care and other shared-risk models. 

“From a clinical and policy perspective, our findings suggest that while EHR-based nudges can influence moment-to-moment decision-making, they are unlikely to be sufficient as a standalone strategy to eliminate low-value preoperative testing,” the authors conclude. 

Source: Sarkisian CA, et al. (2026 Nov 5) JAMA Ophthalmol. Effect of Electronic Behavioral Nudges on Low-Value Preoperative Testing: A Pragmatic Randomized Clinical Trial