The Rational Doctor: Why 96.7% of Primary Care Physicians Won't Follow Testosterone Guidelines

3.3%

Primary care guideline adherence for testosterone prescribing
Pederson et al., ENDO 2025 | Scripps Clinic | n=269 prescriptions, 51 providers

That number was presented at ENDO 2025 as evidence of a problem. The Endocrine Society guideline asks for two morning blood draws confirming low testosterone, documented symptoms, exclusion of reversible causes, and a specific diagnostic threshold — before initiating TRT. At Scripps Clinic, overall adherence was 5.9%. Among specialists, 11.4%. Among primary care physicians: 3.3%.

The framing was predictable: providers need more education. The survey confirmed it — 76.5% of the same providers asked for more educational tools. And 74.5% viewed direct-to-consumer testosterone marketing as a public health problem.

But 3.3% is not a failure rate. It's a consensus.

When 96.7% of trained physicians independently reach the same conclusion — that the guidelines as written are not followable in their clinical reality — the diagnostic question inverts. Not why don't doctors follow guidelines? but what do the guidelines ask that rational physicians consistently refuse to do?

This is the view from behind the desk.

The Patient They Can't Recognize

The first obstacle isn't knowledge. It's visibility.

A UK Testosterone Awareness Survey (2024) found that 49% of men scored as likely testosterone-deficient on the qADAM screening tool. Five percent had been diagnosed. That's a 10:1 gap between probable disease and recognized disease — not because the patients weren't seeing doctors, but because neither party connected the symptoms to the diagnosis.

El-Osta et al. (2025) surveyed men receiving TRT and found 81% had relied on online sources rather than their GP for information about testosterone deficiency. 86% delayed seeking treatment for more than a year. 24% waited more than five years. The GP wasn't the entry point — the GP was the last resort, after the patient had already self-diagnosed.

Liu et al. (2025, qualitative) interviewed GPs directly and found a consistent pattern: physicians attributed symptoms of testosterone deficiency to aging, stress, or depression. Not because they were ignorant of hypogonadism, but because the symptom profile — fatigue, low mood, decreased libido, reduced muscle mass — maps identically onto half a dozen other conditions that a primary care physician sees daily. Without a screening trigger, the testosterone pathway never opens.

A Warwick University Delphi consensus (55 healthcare professionals) concluded with six recommendations, the first of which was simply: awareness of testosterone deficiency is lacking among healthcare professionals.

This isn't a training problem with a training solution. It's a construct recognition problem. The disease doesn't present as a disease. It presents as aging.

The Test They Can't Interpret

Suppose the GP does suspect testosterone deficiency. They order a blood test. The result comes back: 285 ng/dL.

Is this patient hypogonadal?

The answer depends entirely on which of thirteen institutional voices the GP chooses to follow.

Sources: Endocrine Society 2018; AUA 2018; ICSM 2024 (Khera et al., SMR 2025); Tsampoukas et al., IJIR 2025 (nine-guideline comparison); EAU 2025 (Salonia et al., European Urology).

One patient. Eleven institutional opinions. Three different diagnostic terms. Thresholds spanning from 231 to 346 ng/dL — a 50% range. At 285 ng/dL, this man is simultaneously hypogonadal and normal, depending on which guideline the GP opens.

The field describes this as a threshold disagreement. It is not. It is a construct disagreement.

The 30% of men who normalize on retest aren't measurement noise. They're the diagnostic signature of a construct boundary fuzzy enough that individual patients drift across it between blood draws. Add the 426% inter-lab variability documented in Mexico City (134 labs, same samples), and the same patient is hypogonadal or not depending on which lab processes the sample.

The instrument is defining the disease. Not detecting it.

A primary care physician facing this landscape has two options: pick a guideline arbitrarily and pretend the others don't exist, or acknowledge the incoherence and hesitate. The Pederson data tells us which option 96.7% choose.

The Tool That Doesn't Exist

Even if the thresholds converged, primary care lacks a basic screening instrument.

Jenkins and Kini published a GP audit using an EMIS clinical protocol for testosterone screening. Compliance reached 100% when the structured tool was available within the electronic health record. But there was no distribution mechanism. The tool existed in one practice. It never propagated.

The Warwick Delphi reached the same conclusion from the expert side: structured diagnostic guidance for primary care is needed but does not exist. Six recommendations, no implementation pathway.

One candidate has emerged. A 2025 study in PLOS ONE proposed TyG-WHtR (triglyceride-glucose index multiplied by waist-to-height ratio) as a metabolic screening marker for testosterone deficiency, reporting an AUC of 0.776 and an odds ratio of 6.61. No blood draw required for the screening step. But the study has not been independently replicated. One dataset, one population.

The screening gap means the GP's default diagnostic pathway is: patient complains of symptoms that map to many conditions → GP has no structured trigger to test testosterone → testosterone is not tested → the disease is not recognized. Not ignorance. Architecture.

The Fear That's Half Right

Suppose the GP does diagnose testosterone deficiency. Suppose they navigate the guideline maze and reach a treatment decision. Now they face monitoring — and a safety literature that produces rational caution for partially wrong reasons.

The TRAVERSE trial (n=5,246) resolved the primary safety question: TRT is not associated with increased major adverse cardiovascular events (HR 0.96, noninferiority confirmed). But the trial surfaced secondary signals: atrial fibrillation 3.5% vs 2.4% (P=0.02), acute kidney injury 2.3% vs 1.5% (P=0.04).

A GP reading this literature would reasonably conclude: testosterone therapy carries cardiovascular and renal risks that require careful monitoring. That conclusion is half right.

The GP monitoring fear is simultaneously rational (AF appears to be a real clinical concern, validated across both endogenous and exogenous testosterone data) and artifact-driven (AKI is likely a pandemic-era confound that disappeared when COVID-period enrollment was accounted for). The GP doesn't have access to this decomposition. They see a composite "safety profile" that conflates a real drug signal with a probable pandemic artifact into a single risk narrative.

Layer on the monitoring guidelines themselves: the AUA recommends checking hematocrit and flagging above 50%. The EAU uses 54%. The difference matters — Morgentaler and Clift (n=9,537) documented 24.15% polycythemia at 12 months. At the AUA threshold, that's roughly one in four patients requiring intervention. At the EAU threshold, far fewer. Which guideline does the GP follow? The one that generates more alarm, or the one that generates less?

The rational response to this confusion is caution. That's what 96.7% chose.

What the System Assumes vs. What the GP Faces

Pederson found that 17.8% of TRT prescriptions at Scripps lacked a confirmed low testosterone level before initiation. Among PCPs, 21%. The instinct is to call this dangerous. But consider the alternative: a GP with a symptomatic patient, a test result in the guideline grey zone (231–346 ng/dL, where the same number is pathological or normal depending on the source), and a patient who has already been told by an online clinic that they need treatment. The GP has three choices: refuse (and lose the patient to an unmonitored DTC provider), test again (and get a different number 30% of the time), or treat and monitor closely. Twenty-one percent chose the third option without waiting for a second confirmatory draw. Not reckless. Pragmatic.

The Silence

On April 16, 2026, the FDA announced it was encouraging TRT manufacturers to pursue a supplemental indication for low libido in men with idiopathic hypogonadism. The Federal Register notice (FR 2026-07615) asked NDA holders to contact FDA by April 30, 2026.

Tomorrow is April 30.

No manufacturer has publicly committed to filing a supplemental NDA. Fifteen days of silence.

The structural reason is the one this entire article describes. An sNDA requires defining the patient population. "Idiopathic hypogonadism" is defined negatively — low testosterone without a known structural or genetic cause. But which low testosterone threshold defines who qualifies? The Endocrine Society's 300 ng/dL? The EAU's 346? The VA's 275? The BSSM's 231?

Different thresholds produce different denominators. Different denominators produce different outcome data. Different outcome data from different sponsors using different constructs would produce contradictory results — not because the drug works differently, but because the patient populations were defined differently. This is aggregation reversal built into the regulatory architecture.

The FDA is expanding access to a construct that thirteen institutional voices have failed to define. The DEA has extended telehealth prescribing through December 2026 without resolving the diagnostic infrastructure underneath. And the GP — the person who would actually prescribe this expanded indication in primary care — has been telling us for years, in the clearest behavioral signal available, that the system doesn't cohere.

3.3% adherence was never the problem.

It was the diagnosis.

Sources

Pederson et al., "Adherence to Male Hypogonadism Treatment Guidelines" — ENDO 2025, Medscape coverage
Tsampoukas et al., nine-guideline comparison — IJIR 2025
Khera, Torres, Grober, Morgentaler et al. — ICSM 2024, Sexual Medicine Reviews (Oct 2025)
Salonia et al. — EAU 2025 guidelines, European Urology (May 2025)
Bhasin & Snyder — NEJM 2025 review
VA Clinical Recommendations — January 2026
Morgentaler & Clift — WJMH 2026 (n=9,537)
TRAVERSE — NEJM 2023
ASPREE AF data — eClinicalMedicine 2024 (n=4,570 men ≥70)
PaTeR/Zitzmann — Andrology, January 2026
El-Osta et al. (2025) — patient experience with testosterone deficiency
Liu et al. (2025) — qualitative GP interview study
UK Testosterone Awareness Survey (2024)
FDA — Press announcement, April 16, 2026
Federal Register — FR 2026-07615