What Yale’s GLP-1 secret-shopper study means for physician-first care
Yale researchers spent five months this year posing as a patient. Same profile, same eligibility, forty-nine different GLP-1 websites. The results, published in JAMA on July 6, read like a stress test the industry failed quietly.
Forty-five of the forty-nine sites approved a prescription. Thirty-four shipped the drug. The median wait: one day. Two compounded orders cleared in under five minutes — faster than it takes to read the intake form.
Only thirteen sites required an actual video visit with a clinician. Three asked for a phone call. The rest ran on a questionnaire and a checkbox.
Picture the gap this creates. A patient with a heart condition, a med list already crowded with five prescriptions, an eating disorder history the intake form skipped entirely — sailing through a system built to approve rather than evaluate. The researchers put it plainly: some platforms optimize for speed instead of care.
RazorMetrics built the opposite system on purpose.
Every medication decision routes through the prescriber who already knows the patient — their history, their other prescriptions, their actual risk profile. A physician who’s spent years earning a 75% response rate makes a genuine clinical call with full context, the kind of judgment years of training produce and a five-minute web form only approximates.
GLP-1s carry real promise for the patients who need them. They also carry real interaction risk for patients managing multiple conditions — exactly the population most likely to get flagged, and exactly the population fastest-tracked through low-oversight telehealth funnels. Speed and safety can coexist, as long as the physician stays in the loop instead of getting routed around.
The Yale study reads as a single data point, and a validating one: the fastest path to a prescription and the safest path to a prescription should be the same path. Anything else is a five-minute promise with a much longer bill attached.
