Older people get less Long COVID. That's what the data says — after you correct for the thing everyone forgot to correct for.
The standard narrative runs like this: aging immune system, weaker recovery, higher risk of post-acute sequelae. It appears in every meta-analysis, every risk factor review, every clinical guidance document. Older patients get Long COVID more often. The signal is consistent. And it is, in a specific statistical sense, wrong.
In February 2026, Alaleh Azhir and colleagues at Mass General Brigham published a preprint that decomposed the age-PASC relationship in 133,792 COVID-19 patients across 12 hospitals and 20 community health centers in Massachusetts. Their question was simple: when you see age predicting Long COVID, are you seeing biology — or are you seeing the diseases that accumulate with time?
The answer rewrites the risk model.
The Reversal
Using generalized estimating equations with cluster-robust variance — and then testing their findings across 768 different analytical specifications to rule out modeling artifacts — Azhir's team found that after adjusting for comorbidity burden, each decade of chronological age was associated with 6% lower odds of developing PASC (OR 0.94, 95% CI: 0.93–0.95).
Not higher. Lower.
Age doesn't increase your Long COVID risk. It decreases it. The comorbidities that come with age are the ones doing the damage.
"Older patients are more vulnerable."
Crude association. Unadjusted.
Age → Comorbidities → ↑↑ PASC risk (indirect)
Indirect path accounts for 145% of total effect
This is the kind of finding that shouldn't surprise epidemiologists but does. It's a textbook example of Simpson's Paradox — an aggregate trend that reverses when you condition on the right variable. In the aggregate, older people get more Long COVID. Within each comorbidity stratum, older people get less.
Where the Shield Breaks
The protection isn't universal. Azhir's causal mediation analysis found a sharp threshold.
Adults younger than 65 retained robust age-based resilience independent of their comorbidity burden. The average direct effect was -0.0042 (p < 0.001) — small per year, but compounding. A healthy 55-year-old had meaningfully lower PASC risk than a healthy 35-year-old, all else equal. The physiological reserve associated with biological aging — whatever combination of immune maturation, inflammatory calibration, and cellular maintenance it represents — provided genuine protection.
At 65, the shield breaks. The average direct effect flipped to +0.0020 (p = 0.14) — statistically indistinguishable from zero. Past 65, age no longer protects. The authors frame this as "exhaustion of age-related protective mechanisms," which is plausible but speculative. The data say only that the protective direct effect vanishes somewhere around the seventh decade.
A 35-year-old with a Charlson Comorbidity Index of 5 — diabetes, moderate kidney disease, COPD — carries higher Long COVID risk than a healthy 60-year-old. The odometer doesn't break the engine. The potholes do.
That Charlson Index number matters. At CCI 5, the odds ratio for PASC was 2.73 — nearly triple the baseline risk. No amount of youth compensates for that level of accumulated organ damage.
The Ratchet Connection
In Post #25, I documented the cumulative biological damage of repeated COVID infections: 37% Long COVID prevalence after three infections, COPD mortality aHR 2.93 after three or more episodes, pediatric PASC doubling with reinfection. The ratchet tightens with each turn.
Azhir's finding reframes the ratchet mechanism. Each COVID infection doesn't just increase Long COVID risk through viral mechanisms — it adds to your comorbidity burden. Post-COVID cardiovascular events, new-onset diabetes, accelerated kidney decline, pulmonary fibrosis. Each infection makes you physiologically older in exactly the way that Azhir's data identifies as the real risk driver.
The ratchet doesn't just tighten grip. It ages you.
A 30-year-old with three COVID infections may carry the comorbidity profile of someone a decade older. Their birth certificate says 30. Their organ systems say otherwise. In Azhir's framework, it's the organ systems that predict Long COVID — not the birth certificate.
The Paxlovid Mirror
The same Mass General Brigham group published a companion paper that extends the age-specificity finding in a different direction. In 19,413 patients and 22,094 COVID episodes, Paxlovid's effect on PASC was strikingly uneven:
Paxlovid protected against GI Long COVID but nearly doubled eye/ear sequelae. It produced no overall PASC reduction across all ages — the benefit appeared only in non-hospitalized patients over 65. The same age threshold that Azhir's first paper identified as the break point reappears here: younger patients with intact physiological reserve don't benefit from Paxlovid's PASC protection; only those whose age-related resilience has been exhausted see a signal.
This echoes the timing-matters pattern from Post #8 and Post #28: interventions that seem broadly promising are actually narrow in their window — narrow in timing, narrow in age, and now narrow in organ system.
The Heterogeneity Problem
The sharpest critique of this paper comes from the Long COVID patient research community. As noted in the S4ME discussion: Long COVID is many things. Post-exertional malaise, POTS, brain fog, fatigue, pain, and organ damage may each have different relationships with age and comorbidity.
Azhir's study defines PASC using the P2RC precision phenotyping system — a significant improvement over simple ICD code searches — but it still treats PASC as a single outcome. If fatigue-dominant Long COVID has a different age-comorbidity profile than dysautonomia-dominant Long COVID, the pooled analysis would mask both. Gao et al.'s ultrasensitive profiling in Nature Immunology (2025) found that even the breathlessness subtype alone has a distinct biomarker signature — CCL3, CD40, IL-18, IRAK1 — suggesting that different manifestations run on different biology. The 6% per-decade protection might be real for some phenotypes and absent for others.
The study is also Massachusetts-specific, retrospective, and based on EHR coding. These are standard limitations for this type of work, and the 768-specification robustness analysis goes further than most to address them. But generalizability to other populations — different comorbidity profiles, different healthcare access patterns, different COVID variant exposures — remains untested.
What This Changes
If Azhir's finding replicates — and the statistical robustness gives reason for cautious confidence — it demands a shift in how Long COVID risk is assessed.
Current risk tools weight age heavily. The CDC prevalence models, the BRFSS ecological analyses, the clinical screening tools — all treat age as an independent risk factor. Azhir's data suggests this is backwards for anyone under 65. For younger adults, age is actually working in their favor, and the thing driving their risk is comorbidity burden — specifically, the Charlson Index score that captures diabetes, kidney disease, liver disease, heart failure, COPD, and cancer.
The clinical implication is concrete. A 40-year-old presenting post-COVID with a Charlson Index of 3 deserves closer monitoring than a 55-year-old with a CCI of 0. Current practice often does the opposite — flagging older patients for PASC screening while younger patients with significant comorbid disease are assumed to be lower risk.
For the Long COVID research enterprise, this finding adds a methodological requirement: any study that reports age as a risk factor for PASC without adequately controlling for comorbidity burden is likely reporting a confounded estimate. The 145% mediation means that unadjusted age effects don't just overestimate risk — they point in the wrong direction entirely.
And for the 44 million Americans living with Long COVID, it offers a reframe. The disease didn't target you because you were old enough to be vulnerable. It targeted the damage your body had already accumulated — from prior infections, from chronic conditions, from the physiological toll of living. Your age was the proxy. Your body's wear was the cause.
Sources
Azhir A, et al. The age paradox in post-infectious sequelae: physiological reserve outweighs chronological age in Long COVID susceptibility. medRxiv, February 2026. DOI: 10.64898/2026.02.24.26346989v1
Azhir A, et al. Paxlovid shows organ-specific and age-specific impacts on risk of developing post-acute sequelae of COVID-19. Communications Medicine, 2026.
Gao Z, et al. Ultrasensitive profiling identifies monoclonal-origin biomarkers of Long COVID subtypes. Nature Immunology, April 2025. DOI: 10.1038/s41590-025-02135-5
Lim WW, et al. Multi-systemic health risks following SARS-CoV-2 reinfection. BMC Global Public Health, 2025.
Bramante CT, et al. Outpatient treatment of COVID-19 and incidence of post-COVID-19 condition. The Lancet Infectious Diseases, 2023. DOI: 10.1016/S1473-3099(23)00299-2