Opposite Directions

Which sex gets worse Long COVID?

The question has an obvious answer. Women. Every survey says so. The CDC Household Pulse Survey, the National Health Interview Survey, clinic cohorts from Northwestern to Mount Sinai — women report Long COVID at 1.5 to 2 times the rate of men. The epidemiological consensus is settled.

Except it isn't. The consensus is an artifact of which instrument you used to ask.

The Finding That Should Unsettle Everything

Hamlin et al. (Science Translational Medicine, November 2024, Stanford, n=45) tracked immune responses during acute COVID infection and followed patients to determine who developed Long COVID. They measured TGF-β1 — transforming growth factor beta, a master regulator of immune tolerance and tissue remodeling — across innate and adaptive immune cells.

In males who went on to develop Long COVID: TGF-β1 expression increased during acute infection. Elevated across multiple cell types. Elevated LAP TGF-β1 protein in plasma. The signal was clear and directionally consistent.

In females who went on to develop Long COVID: TGF-β1 expression decreased during acute infection. Reduced in NK cells, CD8+ effector memory T cells, CD4+ central memory T cells. The signal was equally clear — and pointed the other way.

This is not a magnitude difference — not "women have more" or "men have less." This is a directional reversal. The same molecular pathway, measured at the same point in the disease, predicts the same outcome (Long COVID), by moving in opposite directions depending on the sex of the patient.

Hamlin's group found more. Males who developed Long COVID showed reduced DDX3Y — a Y-linked gene involved in interferon response. Females who developed Long COVID showed elevated XIST — the X-inactivation escape gene linked to autoimmunity. NK cell signaling differed: males activated through TGF-β/IFN-γ, females through CCL5/CD99/ADAM10/IL-15. Monocyte communication pathways diverged: males received CD40L signals (associated with recovery) versus IFNG/IL15 (associated with Long COVID); females received CCL5, with enhanced interferon-stimulated gene subnetworks.

But NF-κB and AP-1 — the transcription factors that sit at the core of the three-locks model — were elevated in both sexes. The upstream biology diverges. The downstream convergence point is shared. Two roads to the same locked door.

The Monocyte Contradiction

If only one study showed sex-specific reversal, it could be an artifact. But the same pattern reappears across independent labs using different methods.

Elahi et al. (Cell Reports Medicine, November 2025, University of Alberta, n=78 LC + 62 controls) studied immune profiles in Long COVID and ME/CFS. Women with Long COVID showed a shift toward myelopoiesis — increased neutrophils and monocytes, depleted regulatory T cells. The female immune system was overproducing myeloid cells.

Silva and Iwasaki (medRxiv 2024, Yale) found the apparent opposite: males with Long COVID had decreased monocyte and dendritic cell frequencies, with elevated NK cells.

Read Silva alone: males have the monocyte problem (depletion). Read Elahi alone: females have the monocyte problem (overproduction). Read both: the answer depends on which lab, which method (CyTOF versus flow cytometry versus scRNA-seq), which cohort (Yale versus Alberta), and which case definition (Long COVID versus Long COVID/ME/CFS overlap).

These might be complementary — male depletion and female overproduction as two halves of a sex-dimorphic response to the same viral trigger. Or they might be contradictory — artifacts of different measurement platforms applied to different populations. The data cannot currently distinguish between these interpretations. That is not a limitation of the science. It is the science telling us that the construct "monocyte dysfunction in Long COVID" fractures along sex lines before it fractures along any other axis.

The Instruments Disagree About Everything

The TGF-β reversal and monocyte contradiction are not isolated. They are instances of a broader pattern: every instrument used to measure sex differences in Long COVID gives a different answer. Some give opposite answers.

Ten instruments. No consistent answer. The sex difference in Long COVID is not a fact waiting to be measured. It is a construct that changes shape depending on how you look at it.

Cross-Sex Hormone Depletion

One of Elahi's findings deserves its own attention because it defies the expected pattern so completely.

Women with Long COVID had reduced testosterone. Men with Long COVID had reduced estradiol. Each sex lost the hormone typically associated with the other — cross-sex depletion. This is not what standard endocrine models predict. Stress-axis suppression (the HPA-driven mechanism most commonly invoked for fatigue) would suppress sex hormones along their own axis: testosterone down in men, estrogen down in women. Cross-sex depletion implies something different — possibly disrupted aromatase activity (the enzyme that converts testosterone to estradiol), possibly direct viral or inflammatory damage to hormone-producing tissues operating through sex-specific pathways.

Elahi also found elevated CD71+ erythroid cells in women with Long COVID — a marker of disrupted erythropoiesis, the process of red blood cell production. Combined with the myelopoiesis shift (the bone marrow tilting toward white blood cell production at the expense of other lineages), this suggests that female Long COVID may involve a fundamental reallocation of hematopoietic resources. The factory is making the wrong things.

The Clinical Dimension

Kopinsky et al. (Annals of Clinical and Translational Neurology, July 2026, Northwestern, n=2,329) provide the largest sex-stratified neurological assessment. Women reported four or more neurological symptoms 79.6% of the time versus 64.9% for men. Women scored worse on both subjective quality of life and objective cognitive function tests.

But Feliz et al. (Biology of Sex Differences, 2026, Lisbon, n=34 LC + 26 controls) complicate this by showing that the immune mechanism behind cognitive symptoms may differ by sex. Perforin — the molecule NK cells use to kill infected or damaged cells — was reduced in Long COVID, but this reduction was driven entirely by the female patients. Male patients showed elevated TNF-α instead. NK cell and B cell populations were intact in both sexes.

Same clinical presentation (cognitive impairment). Different immune mechanism underneath (cytotoxic failure in women, inflammatory cytokine elevation in men). If a treatment trial enrolls "Long COVID patients with cognitive symptoms" without stratifying by sex, it enrolls two immunologically distinct populations into one arm.

This is the taxonomic inertia problem compressed into a single variable. Not "Long COVID contains multiple diseases" in the abstract — but specifically: the female version and the male version may require different interventions for the same symptom.

The Gap That Should Not Exist

Here is the fact that makes all of the above more than academic:

No published Long COVID treatment trial has reported sex-stratified efficacy data.

Not RECOVER-AUTONOMIC. Not RECOVER-NEURO. Not STOP-PASC. Not IAMPOCO. Not TOGETHER. The trials enrolled mixed populations, reported aggregate results, and left the sex-specific question unanswered. Some collected the data. None published it.

If the TGF-β finding is real — if males and females develop Long COVID through immunologically opposite mechanisms that converge on shared downstream pathology — then every aggregate treatment result in the literature is averaging two different responses. A drug that works for one sex and fails for the other would show a modest aggregate effect or none at all. This is exactly the failure mode the field has been producing.

The Lindberg MIRACLE-S data (eClinicalMedicine, 2026, Karolinska, n=1.2 million) showed that premenopausal women aged 40-54 had the highest cardiovascular risk (RR 1.45), while postmenopausal women had no excess risk. If hormonal status modifies the cardiovascular phenotype this sharply, it almost certainly modifies treatment response. But no trial has tested this.

What I Don't Know

The Hamlin TGF-β finding is from n=45 with sex-stratified subgroups even smaller. The directional reversal is striking enough that it demands replication — and no replication exists. If a larger study finds the same TGF-β pattern, it reshapes the field. If it doesn't, the centerpiece of this analysis dissolves.

Whether the 12-month reversal Hamlin observed — females with ongoing Long COVID switching to increased TGF-β1 — means female biology converges with male biology over time, or represents a different mechanism entirely, is unknown. If it's convergence, then early-stage and late-stage female Long COVID may be different conditions.

The Silva and Elahi monocyte disagreement may be a genuine biological contradiction, a measurement artifact (different platforms, different panels, different populations), or complementary observations of different monocyte subsets that happen to use the same word. I cannot currently distinguish these.

The XIST elevation in females — Hamlin notes it but doesn't test the autoimmunity connection mechanistically. X-inactivation escape is associated with lupus and other female-predominant autoimmune conditions. The connection to Long COVID autoimmunity is plausible but speculative at this stage.

Most fundamentally: whether sex-stratified treatment would actually produce different results in Long COVID trials is an untested prediction. It follows logically from the biology. But the data to confirm or refute it has been collected and not published, or not collected at all.