The best clinical decision is sometimes not to prescribe. Here’s how physician judgment, triage, and the courage to refer separate real medicine from an algorithm.
By Dr. Teja V. Surapaneni, MD, MS • Board-Certified Internal Medicine • April 2026
I get asked a version of the same question regularly: “Why didn’t you just approve my request right away?”
It’s a fair question. You filled out the intake, you described your situation, you know what you want. From a consumer standpoint, a same-day approval feels like good service. A physician who asks for more information, orders labs, or suggests a referral before prescribing can feel like friction.
I want to explain why that friction is, in many cases, the most important thing I do.
Several large DTC telehealth platforms measure their quality—implicitly or explicitly—by prescription approval rates and time-to-prescription. The faster you approve, the more prescriptions you write. The more prescriptions you write, the more revenue you generate. The incentive structure runs directly toward saying yes.
This is not medicine. It is prescription fulfillment with a license attached.
Board-certified internal medicine is built on a different foundation: comprehensive evaluation before action. That means taking a history, reviewing comorbidities, examining drug interactions, assessing labs, considering the patient’s full clinical picture—and then making a judgment about what is actually indicated. Sometimes the right answer is the prescription the patient requested. Sometimes it is a different medication. Sometimes it is a lab order first. Sometimes it is a referral. And occasionally, the right answer is that telehealth is not the right setting for this patient’s situation at all.
None of those outcomes is a failure. They are all expressions of clinical judgment. The failure is when a physician doesn’t exercise that judgment in the first place.
Algorithmic intake questionnaires are useful tools. I use structured intake to gather information efficiently. But they have a hard ceiling.
A questionnaire asks the questions it was programmed to ask. A physician asks the follow-up question. When a patient reports “occasional headaches” in their hypertension intake, a questionnaire logs the answer and moves on. A physician asks: When did they start? Are they positional? Associated with any vision changes? Have you had your blood pressure checked at home this week? Those answers can completely change the clinical picture—and in some cases, the answer to “schedule a virtual visit” becomes “go to urgent care today.”
The ability to recognize when a patient needs escalation that a questionnaire cannot detect is not a bug in our model. It is the entire point of having a physician in the loop.
Internal medicine training is built around a specific intellectual discipline: knowing the scope of your intervention, and knowing when that scope is exceeded. We refer to cardiology, nephrology, endocrinology, and other specialists not because we cannot manage the medication—often we can—but because the patient’s situation warrants a higher level of evaluation than our setting provides.
For telehealth, this is even more important. Without the ability to perform a physical exam, auscultate the lungs, measure an ankle-brachial index, or observe how a patient breathes when they move, we are working with a more limited information set. Honest telehealth practice acknowledges that limitation and routes patients appropriately when it is exceeded.
A patient with newly diagnosed heart failure presenting for medication management on our platform gets a referral to cardiology—not because I cannot look up the GDMT guidelines, but because a new CHF diagnosis requires a full in-person cardiac evaluation, an echocardiogram, exercise tolerance assessment, and ongoing monitoring that telehealth cannot replicate. Starting carvedilol and furosemide over video for a patient whose ejection fraction I haven’t personally reviewed is not responsible care. It is prescribing at volume.
Similarly, a patient with triglycerides over 500 mg/dL doesn’t get a fenofibrate prescription and a “follow up in three months.” They get an explanation that severely elevated triglycerides can cause pancreatitis, a discussion of the secondary causes I want ruled out, and a referral to a lipid specialist or gastroenterologist if the initial workup is abnormal. That is what the evidence and the guidelines actually recommend.
Every chronic disease intake at YourMD begins with a safety screen. Before we ask about your blood pressure medication history, we ask whether you have chest pain right now. Before we discuss cholesterol management, we ask about sudden weakness or difficulty speaking. Before we continue a diabetes intake, we ask whether your blood sugar is currently above 400 mg/dL.
If those screens return positive results, we do not schedule a telehealth visit. We tell the patient to call 911 or go to the emergency room. Immediately. Without collecting a visit fee.
I have seen DTC platforms that do not have these screens at all. A patient in hypertensive urgency—blood pressure above 180/120 with a headache—completes their intake, gets approved for lisinopril, and their prescription ships in 48 hours. That patient needed an emergency room, not a mail-order pharmacy. The platform optimized for throughput. The physician—or in some cases, the nurse practitioner or PA following a protocol—signed the order anyway.
This is not a hypothetical. Reports to the FDA’s MedWatch database document cases where DTC telehealth prescribing contributed to serious adverse outcomes because basic safety screening was absent or inadequate. The problem is structural, not individual. When the incentive is to approve quickly and the friction of a referral or deferral is measured as a lost conversion, the system produces bad outcomes at scale.
Our intake flows ask safety questions before clinical questions. We have hard stops for emergency presentations that show the patient ER guidance instead of a scheduling calendar. We require labs before initiating certain chronic disease medications—not as bureaucratic gatekeeping, but because prescribing a statin without knowing liver function, or an ACE inhibitor without knowing kidney function and potassium, is how patients get hurt.
We turn some patients away from the platform entirely. A patient with NYHA Class III or IV heart failure, a patient with suspected pheochromocytoma causing refractory hypertension, a patient with familial hypercholesterolemia and an LDL above 250 who has never seen a cardiologist—none of them belong in a DTC telehealth intake flow. They belong with the right specialist, and we tell them so.
We also refer proactively. When a patient’s blood pressure remains uncontrolled on three medications including a diuretic, the clinical guideline for resistant hypertension is specialist evaluation, not a fourth medication added via chat. We generate a referral, explain why, and offer to coordinate care after the specialist has evaluated them. That is a longer path to revenue than pressing approve. It is the right path for the patient.
I am not arguing that high-volume DTC telehealth is wrong as a category. Millions of patients benefit from convenient access to care for conditions that are genuinely straightforward to manage remotely. An otherwise healthy 28-year-old man with mild androgenic alopecia who wants finasteride does not need a comprehensive internal medicine workup. He needs a competent clinician to review his intake, confirm there are no contraindications, and write a prescription. That is appropriate care at appropriate scope.
What I am arguing is that the scope matters, and it shifts significantly when the conditions become complex. Hypertension in a 65-year-old with three prior strokes and chronic kidney disease is not the same clinical problem as hypertension in a 35-year-old with elevated stress and a salty diet. Treating them through the same 5-question intake form and approving both for lisinopril is not good medicine. It is high-throughput pattern matching.
The differentiator at YourMD is that I approach every case as a clinician first. That means some cases take longer. Some cases result in a referral rather than a prescription. Some cases end with “this isn’t the right setting for your situation, and here is where to go instead.” Those outcomes are not failures of our platform. They are evidence that the platform is working correctly.
We recently launched consult-based programs for hypertension, high cholesterol, Type 2 diabetes, and stable heart failure. These are not subscription products. They are physician-led evaluation and management visits, priced per consultation, with lab requirements, structured safety screening, and explicit referral pathways built in from the beginning.
Every one of those programs includes language I insisted on: a list of presentations that should go to the ER, a list of conditions that exceed telehealth scope and require specialist co-management, and a clear statement that receiving a referral from us is not a rejection—it is a clinical service.
The patients who are right for our platform will get excellent care through it. The patients who are not right for our platform will be directed to the care they actually need. Both outcomes reflect a functioning clinical practice.
That is what physician-led telehealth looks like when it is practiced honestly.
This article reflects the clinical philosophy and practice policies of YourMD Telehealth. It is intended as an informational essay and does not constitute medical advice for any individual. If you are experiencing a medical emergency, call 911.
Board-certified internal medicine. $99 initial visit for chronic disease programs.