Table 2. Evidence on clinical and demographic correlates of T cell response to SARS-CoV-2 infection from studies included in this review (* disease severity was defined in various ways in included studies; for some according to intensive care unit admission; a number used the Chinese National Health Commission definition [66]).

Category	Correlate	Dimension or sub-population	Findings
Clinical	Disease severity*	Asymptomatic or pauci-symptomatic	• One study evaluated T cell responses in asymptomatic patients (n = 20) and found little change in the circulating T cell frequencies within this group [51].
		Moderate disease	• Reduced numbers of both CD4+ and CD8+ T cells in moderate and severe cases, alongside increased numbers of activated CD4+ and CD8+ T cells expressing PD-1 or Tim-3; as well as potential reductions in cytotoxic potential and polyfunctionality were reported in one narrative review [3].
		Severe or critical disease	Cell counts
			• A medium quality meta-analysis found that patients with severe disease had statistically significant, two-fold decreases in both CD4+ and CD8+ T cells, as well as in CD3+ T cells (1.7-fold) and overall lymphocyte number (1.44-fold), alongside statistically significant increases in neutrophils (1.33-fold) and overall leukocytes (1.2-fold) [67].
			• A large study (N = 599) reported reduced total, CD4+, and CD8+ T cells being associated with more severe disease, comparing n = 43 ICU-admitted patients with non-ICU-admitted patients, and comparing critical/severe non-ICU patients with mild/moderate non-ICU patients (as per Chinese national definitions*) [37].
			• Other large studies [11, 18, 32, 36] showed comparable findings, and 3 studies also reported reduced CD3+ cells in more severe disease [18, 32, 36]; however, one only found significant cell count differences for critical vs severe disease, and not for severe vs moderate disease [18].
			Cell ratios
			• Six studies reported marked increases in CD4/CD8 ratio (due to increases in CD4+ but reductions in CD8+ cells) in severe and critical patients compared to those with moderate disease [22, 25, 36, 62, 65]. The last of these also showed CD8+ T cell counts were much slower to normalise than CD4+ in patients with severe disease [25].
			• Two studies however, reported significant reductions in CD4+, but not in CD8+, T cells in severe disease (n = 452), or ‘aggravated’ disease, defined as clinically progressive at 7 days (n = 17) [26, 34].
			• A small study from Iran reported increased CD8 expression in ICU patients relative to healthy controls, quantified by flow cytometry as mean fluorescence intensity (MFI), with no significant differences seen in CD4/CD8 ratio, or CD4+ T cell MFI [40].
	Clinical endpoint	Survival vs non-survival	• Two studies with large cohorts followed up COVID-19 patients until death or discharge, both conducting multivariate analysis. Luo et al. (n = 1018), reported significantly lower CD3+, CD4+ and especially CD8+ counts in non-survivors than survivors, and found that CD8+ T cell counts <165 cells/μL (OR 5.93) were independently associated with mortality after adjustment for age, sex and comorbidities [21]. Liu et al. (n = 340) reported that lower helper T cells (OR 0.22) and higher CD4/CD8 ratio (OR 4.8) were highly significant predictors of mortality [63].
			• Whilst also reporting lower CD8+ counts in non-survivors throughout the disease course, Wang et al. (n = 157) also found that non-survivors had lower CD4+ counts only evident in middle and late stages of disease, and that non-survivors had a lower CD4/CD8 ratio [29].
			• Based on 28 deaths amongst 187 patients, Xu et al. demonstrated that total T cell counts <500/μl, CD3+ counts <200/μl, CD4+ or CD8+ counts <100/μ as well as B cell counts <50/μL, were significantly associated with risk of in-hospital death, however this is only on univariate analysis [32].
			• In a cohort of n = 548, Chen et al. reported significantly elevated neutrophil-to-lymphocyte ratio (NLR), platelets-to-lymphocytes ratio (PLR), reduced peripheral CD3+, CD4+ and particularly CD8+ counts in non-survivors [36]. He et al. (n = 204) reported that T cell levels continued to fall until death in non-survivors, whilst in survivors with severe disease, levels increased after 15 days and normalised after 25 days of treatment [11].
		RNA persistence	• Four small but high or medium quality clinical cohort studies from China showed that slower resolution of PCR-positivity is associated with reductions in peripheral T cells.
			• Jiang et al. (n = 23) found that the baseline abnormalities in CD3+, CD4+ and CD8+ T cells underwent robust recovery in patients who became RNA negative 2 weeks after diagnosis, whilst they did not do so in those who remained persistently positive [12].
			• Liu et al. compared 37 cases who remained positive at day 20, with 37 patients at their point of diagnosis, as well as 54 healthy controls, and showed that both the persistently positive and control groups had higher CD3+ and CD4+ levels, suggesting that these subsets do normalise despite viral persistence [45].
			• In a similar study, though with a persistence threshold of 15 days, Dong et al. (n = 18) also found global reductions across CD3+, CD4+ and CD8+ subsets for persistent positives, which increased between admission and discharge; as well as significant negative correlation between overall T cell count and duration of positive nucleic acid test [38].
			• Liu et al. (n = 39) also reported higher global T and B cells in patients becoming RT-PCR negative within 14 days [20].
	Co-morbid disease status		• Three studies considered the effect of comorbid status, all originating from China and spanning patients with non-severe, severe and critical clinical presentations [11, 20, 39]. Two had significant methodological limitations [20, 39].
			• One study (n = 204) found significantly lower total lymphocyte and lymphocyte subset counts in patients with comorbidities compared with those without (though “comorbidities” not defined) [11].
			• The second (n = 39) found statistically significant differences in CD8+ counts between patients with comorbid disease and those without (p = 0.046), but no difference in CD4+ counts—although here again the range of comorbidities considered was not defined [20].
			• The final study compared outcomes in a paediatric cohort with or without “allergic disease” (not clearly defined) and showed no effect on clinical course, total lymphocyte or lymphocyte subset counts [39].
Demographic	Age	Older adults	• A high-quality clinical cohort study and a medium-quality case-control study, both from China, reported lower T cell total and subset counts, including CD3+, CD4+, CD8+ subsets, for older patients aged 60 or over [11, 37].
		Children	• Four medium-quality studies—1 case control and 3 case series—considered cellular responses in children in samples from China, all showing comparable CD3+, CD4+ and CD8+ counts to healthy paediatric controls, or where the comparison group was adults, higher T cell counts across subsets [19, 33, 35]. However, potential confounders such as disease severity or comorbidities were not controlled for in these studies.
	Sex		• One medium-quality case series (n = 27) from China examined differences in cytokine secretion by sex of cases, showing reductions in CD4+ and CD8+ count for all patients irrespective of gender but more generalised cytokine responses were observed among male participants than females, for IL-6, TNF-α and procalcitonin–although the statistical significance of these differences was not tested [24].