What's Wrong With One-Size-Fits-All Medicine

Clinical guidelines are built on randomized controlled trials, and RCTs are genuinely among the most reliable tools in medicine. When a well-designed trial tells us that intensive blood pressure control reduces cardiovascular events, that finding is real. The problem comes when we apply population-level averages — built on heterogeneous patient groups, averaged across vastly different biological profiles — as if they were prescriptions for individuals.

A blood pressure trial might enroll 10,000 patients and find that treating to 130 systolic reduces events by 18%. That's meaningful science. Embedded in that finding, though, is enormous individual variation: some patients drive the effect strongly, others receive little benefit, and some may actually experience harm from aggressive treatment. The guideline that follows flattens all of that variation into a single threshold.

“This isn't a failure of guidelines. It's a limitation of them — one that becomes a failure only when clinicians stop at the guideline rather than using it as a starting point.”

The ACCORD trial illustrated this directly. Intensive glycemic control in older diabetic patients with elevated cardiovascular risk was associated with increased mortality compared to standard care — a population for whom the guideline's logic, applied without individual context, generated the wrong answer (ACCORD Study Group, N Engl J Med, 2008).[1] The average benefit masked a harm signal in a specific subgroup.

This isn't a failure of guidelines. It's a limitation of them — one that becomes a failure only when clinicians stop at the guideline rather than using it as a starting point.

The strongest evidence for precision approaches is in oncology. Tumor genomics now routinely guide treatment selection for lung, breast, colon, and other cancers, with targeted therapies producing outcomes that population-level chemotherapy regimens couldn't match. An analysis published in JAMA Oncology estimated that approximately 7% of U.S. cancer patients currently benefit from genomically guided therapy — modest in absolute terms, but representing hundreds of thousands of people receiving treatment that would otherwise have been unavailable or inappropriate (Marquart et al., JAMA Oncology, 2018).[2]

Pharmacogenomics — how genetic variation affects drug metabolism — has established clinically actionable results for drugs including warfarin, clopidogrel, codeine, and several antidepressants. The FDA's pharmacogenomic biomarker labeling now covers over 300 drugs, though translation to routine prescribing practice remains inconsistent.

In metabolic medicine, large nutrition studies have demonstrated wide individual variation in glycemic responses to identical foods (Zeevi et al., Cell, 2015)[3] — findings that support personalized dietary approaches, though whether acting on those differences improves hard clinical outcomes meaningfully better than standard dietary guidance is still being actively studied.

Precision medicine will not provide clean answers for every clinical question. Some genetic risk variants are small in effect size and not reliably actionable. Reference ranges on standard labs are derived from populations that may not represent the person in your chair. Some commercially available biomarker tests — epigenetic age panels, multi-cancer early detection, comprehensive metabolomics — are scientifically interesting but have outcomes data that is still maturing. Acting on every result without acknowledging what we know and don't know is not precision medicine — it's testing for the sake of testing.

There is also an access problem worth naming directly: more comprehensive workups cost more, and a model that serves only people who can afford concierge care risks deepening health disparities rather than reducing them. This is a real structural challenge for the field.

The version of precision medicine I believe in isn't about rejecting population evidence or treating guidelines as noise. It's about recognizing that guidelines are the beginning of a clinical thought process, not the end of one.

The person sitting across from me is not the average enrolled trial participant. They have a specific genome, a specific history, specific goals, and a specific time horizon. The guideline gives me a well-evidenced starting point. What I'm supposed to bring to the table — what the 15-minute annual wellness visit doesn't leave room for — is the interpretation of that starting point through the lens of who that person actually is.

That's what one-size-fits-all medicine fails to do. Not because the evidence it's built on is wrong, but because it stops short of the part that actually requires a physician.