Participants. During the period spanning from August 1, 2021, and April 10, 2022, a total of 1,254 hospitalized patients with COVID-19 from 5 study centers were evaluated for eligibility. Out of this initial group, 1,152 patients were included in the study. In total, 785 patients were aged 60 years or above and thus classified as older adults. The control group encompassed 367 adults younger than 60 years. Patient flow is depicted in Figure 1.

With regard to the whole cohort, 118 patients died, 165 were admitted to an intensive care unit, 47 patients required endotracheal intubation for respiratory support, and 587 patients required oxygen administration. The majority of these cases concerned older adults, who accounted for 112 deaths, 108 ICU admissions, 33 intubations, and 450 instances of oxygen administration. Median duration of hospital stay was 8 days (interquartile range [IQR], 4–16) overall and 10 days (IQR, 5–18) in older adults. Patient characteristics of older and younger adults are outlined in Table 1, Table 2, and Table 3.

Study cohort outcomes and patient characteristics.

Patient characteristics of older adults.

Comparison of patient characteristics between older and younger adults.

Outcome by antibody levels. Anti–SARS-CoV-2 antibodies were significantly lower in older adults who died compared with those who survived (mean 408 binding antibody units [BAU]/mL [95% CI, 242–574], versus mean 1,146 BAU/mL [95% CI, 1,057–1,236]; P < 0.0001).

Older adults whose antibody levels fell below the threshold of 1,200 BAU/mL were more than 4 times more likely to die compared with those with antibody levels above this threshold (OR, 4.41 [95% CI, 2.57–7.56]; P < 0.0001).

With regard to secondary endpoints, older adults who required oxygen administration, endotracheal intubation, or intensive care treatment also exhibited significantly lower antibodies compared with patients who did not require these interventions (oxygen administration: mean 787 BAU/mL [95% CI, 686–889] versus 1,377 BAU/mL [95% CI, 1,249–1,504], P < 0.0001; endotracheal intubation: mean 204 BAU/mL [95% CI, 0–429] versus 1,075 BAU/mL [95% CI, 991–1,159], P < 0.0001; intensive care: mean 587 BAU/mL [95% CI, 391–783] versus 1,111 BAU/mL [95% CI, 1,022–1,200], P < 0.0001).

Figure 2 depicts patient outcomes in percentages with regard to the endpoints all-cause in-hospital mortality, ICU treatment, endotracheal intubation, and oxygen administration by antibody level and vaccination status.

Patient outcomes for older (≥60 years) and younger adults (<60 years) in percentages regarding in-hospital mortality, intensive care treatment, endotracheal intubation, and oxygen administration by antibody level and vaccination status. Bottom row: Outcome by antibody level in vaccinated older adults and older adults infected with the Omicron variant. BAU binding antibody units.

Vaccinated older adults had lower rates of in-hospital mortality, ICU admission, endotracheal intubation, and oxygen administration compared with nonvaccinated patients but higher rates than patients with antibody levels above 1,200 BAU/mL.

Survival over time. Figure 3 shows Kaplan-Meier curves for cumulative survival over time by antibody level for older adults, vaccinated older adults, and older adults infected with the currently prevailing Omicron variant. For comparison, cumulative survival of older adults by vaccination status is also included. Statistical significance was tested by log rank (Mantel Cox) test. Median follow-up time was 90 days after hospital admission (IQR, 48–90 days.)

Cumulative survival over time in older adults. (A–D) Kaplan-Meier curves with 95% CI, for cumulative survival over time in older adults (≥60 years) by high and low anti–SARS-CoV-2 spike antibody level (above and below 1,200 BAU/mL) (A), by vaccination status (B), by antibody level in vaccinated older adults (C), and by antibody level in older adults infected with the Omicron variant (D). Number censored: cumulative number of patients lost to follow-up. Statistical significance was determined by log rank (Mantel-Cox) test. BAU, binding antibody units; nonvacc., nonvaccinated patients.

In older adults, vaccinated patients had better cumulative survival than nonvaccinated patients but had lower cumulative survival than those with antibodies above 1,200 BAU/mL.

While both vaccination status and antibodies above 1m200 BAU/mL are good predictors of protection from in-hospital mortality, patients with spike antibodies above 1,200 BAU/mL had better odds of survival than vaccinated patients (OR, 4.41 [95% CI, 2.57–7.56], P < 0.0001, versus OR, 3.15 [95% CI, 2.09–4.77], P < 0.0001).

For the control group (<60 years), there was a trend toward lower mortality in patients with higher antibody levels, albeit not at a statistically significant level. Mortality rates in younger adults did not differ by vaccination status. Due to the low number of deaths in younger adults, these results need to be interpreted with care.

Risk estimation and adjustment for potential confounders. In order to assess the risk associated with lower anti–SARS-CoV-2 spike antibodies, we built multiple logistic regression models for all endpoints. We further calculated Cox proportional hazard models for the primary endpoint in-hospital mortality to provide a second measure of risk. To limit the influence of potential confounders, these models were then adjusted for the covariates age, BMI, SARS-CoV-2 variant, T2D, hypertension, coronary artery disease (CAD), heart failure, stroke/transient ischemic attack (TIA)/cerebrovascular disease (CVD), and renal disease.

Figure 4 shows risk of outcome by antibody level and vaccination status as both unadjusted and adjusted ORs for all endpoints and as hazard ratios for the primary endpoint in-hospital mortality. Results for vaccinated older adults and older adults infected with the Omicron variant are also presented. Unadjusted and adjusted risk ratios by antibody level and vaccination status for all endpoints are shown in Supplemental Figure 1 (supplemental material available online with this article; https://doi.org/10.1172/jci.insight.183913DS1).

Risk of outcome in older adults. (A–D) Risk of outcome in older adults, aged 60 years or older, by antibody level above versus below 1,200 BAU/mL. (A) and vaccination status (B); risk of outcome by antibody level in vaccinated older adults (C) and in older adults infected with the Omicron variant (D). Unadjusted and adjusted odds ratios are shown for the outcomes oxygen administration, endotracheal intubation, intensive care admission, and in-hospital mortality. Unadjusted and adjusted hazard ratios are shown for in-hospital mortality. Adjusted odds and hazard ratios were calculated by multiple logistic and Cox regression analyses and adjusted for age, BMI, SARS-CoV-2 variant, type 2 diabetes, hypertension, CAD, heart failure, stroke/TIA/CVD, and renal disease.

After adjusting for potential confounders, older adults with antibody levels below 1,200 BAU/mL exhibited almost 5 times the mortality risk of patients above this threshold (adjusted OR [aOR], 4.92 [95% CI, 2.59–9.34], P < 0.0001). In addition, they were approximately 2.6 times more likely to be admitted to an intensive care unit (aOR, 2.64 [95% CI, 1.52–4.62], P = 0.00062). The odds for endotracheal intubation were 6.5 times higher and patients were more than twice as likely to require oxygen if antibody levels were below 1,200 BAU/mL (endotracheal intubation aOR, 6.50[ 95% CI, 1.48–28.47], P = 0.013; oxygen administration aOR, 2.34 [95% CI, 1.60–3.43], P < 0.0001).

In the Cox proportional hazards analysis, risk of death for older adults was more than 3 times higher if antibody levels were found to be lower than 1,200 BAU/mL (hazard ratio, 3.92 [95% CI, 2.34–6.56], P < 0.0001). Analogous to the logistic regression model, the results remained stable after adjusting for potential confounders (aHR 4.27 [95% CI, 2.34–7.81], P < 0.0001).

Older adults infected with the currently prevailing Omicron variant were more than 6 times more likely to die if antibody levels were below 1,200 BAU/mL (aOR, 6.31 [95% CI, 2.43–16.40], P = 0.00016).

In comparison with antibody levels below and above 1,200 BAU/mL, vaccination status was a weaker predictor of our primary endpoint, in-hospital mortality (aOR, 4.92 [95% CI, 2.59–9.34], P < 0.0001, vs. aOR, 3.68 [95% CI, 2.26–6.01], P < 0.0001).

Mortality risk estimation by antibody titer increment. In order to further quantify a possible dose-effect relationship between anti–SARS-CoV-2 spike antibodies and mortality risk in older adults, we calculated the increase in mortality risk with decreasing antibody levels in steps of 100 BAU/mL and 250 BAU/mL.

After adjusting for potential confounders, mortality risk increased by approximately 1.1 for each 100 BAU/mL decrease (aOR, 1.08 [95% CI, 1.05–1.11], P < 0.0001) and by 1.2 for each 250 BAU/mL decrease (aOR, 1.21 [95% CI, 1.13–1.30], P < 0.0001).

Results were comparable for older adults infected with the currently prevailing Omicron variant (100 BAU/mL steps: aOR, 1.08 [95% CI, 1.04–1.13], P = 0.00011; 250 BAU/mL steps: aOR, 1.22 [95% CI, 1.10–1.35], P = 0.00011).