what causes "long Covid"?

The short version: long COVID is not one disease. It is at least five overlapping mechanisms, and the evidence for each is uneven.

The leading hypothesis that best explains the diversity of symptoms is viral persistence. Multiple 2026 studies found SARS-CoV-2 RNA and spike protein lingering in tissue months after the acute infection cleared from the blood. Colorectal biopsies detected viral RNA in 27% of long COVID patients versus 7.7% of recovered controls, though the small sample (44 vs 13) did not reach statistical significance [1]. In a separate study, spike protein itself was found in long COVID gut tissue alongside localized immune activation, with macrophages, plasma cells, and regulatory T cells clustered around the spike-positive areas [2]. Hamster nasal tissue still harbored nucleocapsid protein and viral RNA 120 days after infection, with ongoing tissue damage and inflammatory gene changes [3]. Together these three studies make a strong circumstantial case that a persistent viral reservoir drives chronic inflammation.

Immune dysregulation is the second mechanism and it lasts a remarkably long time. A three-year longitudinal study found that COVID-19 survivors maintain a Th17-driven inflammatory state that resembles immunosenescence, the immune aging normally seen in the elderly [4]. Two distinct Th17 subsets were identified: one linked to anxiety, chest pain, and smell disorders, the other to an anti-inflammatory recovery signature [4]. This paper followed 47 patients against 25 controls and profiled 925,879 individual cells, making it the most granular long-term immune data available.

The third mechanism is endothelial and microvascular damage. Systematic reviews now cite chronic inflammation, endothelial dysfunction, hypercoagulability, and microthrombosis as central to long COVID's cardiopulmonary consequences [6]. A 2025 case series found vasospastic angina (artery spasm) in 1.8% of 273 long COVID patients with chest pain [5]. Myocardial edema, driven by glycocalyx degradation and pericyte dysfunction, persists in cardiac tissue well after the acute phase [7]. The evidence here is mechanistic but widely documented.

Mitochondrial impairment is the fourth candidate. SARS-CoV-2 proteins - spike, envelope, membrane, nucleocapsid, and several nonstructural proteins - all interact with host mitochondria to disrupt the electron transport chain, fragment mitochondria, increase reactive oxygen species, and suppress ATP production [8]. The spike protein alone reduces maximal oxygen consumption and drives inflammation through the NLRP3 inflammasome [8]. This would explain the profound fatigue and post-exertional malaise.

A fifth mechanism, small fiber neuropathy, was confirmed by skin biopsy in 56.5% of 85 long COVID patients with neuropathic symptoms in a 2025 observational study [9]. This affects both epidermal and autonomic nerve fibers and can explain the burning pain, heart rate variability, and temperature dysregulation many patients report.

My call: The evidence is strong for viral persistence driving immune dysregulation, microvascular damage, and mitochondrial failure as the core mechanisms. The field is converging on a model where one or more of these hits a given patient, which is why symptoms are so variable. Confidence is moderate. The searches did not surface a single large trial comparing treated to untreated patients on hard endpoints, and most studies remain small and cross-sectional. A trial testing an antiviral (like nirmatrelvir or longer-course remdesivir) in early long COVID would settle the persistence hypothesis. That trial has not been run.