Our matched analysis dataset consisted of 22,841 veterans infected during the Omicron period and 22,841 matched veterans infected during the Delta period. (Fig. 1) The median age (IQR) was 62.0 years (49.0, 72.0), 91.9% were male, and 82.4% were white. The majority of veterans in our cohort were multimorbid, and more than 75% (34,492 / 45,682) had two or more pre-existing chronic health conditions. The median Charlson Comorbidity Index score was 3 (IQR 2, 4).
Among both groups, 7393 (32.4%) had received two doses of an mRNA vaccine, and 990 (4.3%) had received an additional booster dose (Table 1). Infection was diagnosed ≥14 days after booster dose in 910 (4.0%) of those infected during the Omicron period and 403 (1.8%) of those infected during the Delta period.
Severity of the disease
Among those infected in the Omicron period, 20,681 (90.5%) had mild disease, whereas 1308 (5.7%) met moderate disease criteria and 852 (3.7%) met criteria for severe / critical illness. Among those infected in the Delta period, 19,356 (84.7%) had mild illness, 1467 (6.4%) met moderate illness, and 2018 (8.8%) severe / critical criteria, respectively. Overall, a significantly lower proportion of veterans met moderate or severe / critical illness criteria in the Omicron period than in the Delta period (9.5% vs. 15.3%; s<0.001; Table 2).
Of the 2160 moderate or severe / critical infections during the Omicron period, 48 (2.2%) occurred in those who had received a booster dose ≥14 days before. Of the 3485 moderate or severe / critical infections during the Delta period, 58 (1.7%) occurred among individuals who had received a booster mRNA vaccine at least 14 days before their infection. (P<0.001; Table 3)
In the multivariable logistic regression model, infection in the Omicron period was associated with lower odds of moderate or severe disease (unadjusted odds ratio: 0.58, 95% CI 0.55 to 0.62, P<2 × 10−16; adjusted odds ratio (aOR): 0.56; 95% CI 0.53-0.59, P<2 × 10−16).
Baseline characteristics of the entire cohort before matching (72,492 in the Omicron period and 35,848 in the Delta variant period) are presented in the supplementary table 2. Veterans with confirmed COVID-19 disease in the Omicron period were younger, and a higher proportion were women and non-whites. Fewer veterans were unvaccinated during the Omicron period (33.5% vs. 48.8%), and a significantly higher proportion had received a booster vaccine (19.4% Omicron vs. 4.1% Delta).
We calculated the odds of the severity of the disease stratified by the dominant variant. Vaccination was associated with significant protection against moderate or severe / critical illness. The unadjusted OR was 0.99 (95% CI 0.89-1.09, P= 0.786), and aOR was 0.51 (95% CI 0.46-0.57, P<2 × 10−16) for Omicron variant infection among vaccinated with 2nd dose ≥ 3 months before infection. The unadjusted OR was 0.68 (95% CI 0.51–0.87) and the aOR was 0.26 (95% CI 0.20–0.34, P<2 × 10−16) for recipients of a booster dose ≥ 14 days before infection. The corresponding unadjusted and adjusted ORs for the Delta variant period were 0.85 (95% CI 0.78-0.92, P= 5.47 × 10−5 and 0.47 (95% CI 0.43-0.51, P<2 × 10−16) for those vaccinated with 2nd dose ≥ 3 months before infection, respectively. For recipients of a booster dose ≥ 14 days before infection, the unadjusted OR for severe / critical illness was 0.82 (95% CI 0.60–1.09, P= 0.177), and aOR was 0.34 (95% CI 0.25-0.46, P= 3.46 × 10−12).
The proportion of veterans in need of organ support measures during Omicron and Delta variant periods is summarized in the supplementary table 3. A significantly lower proportion of individuals in the Omicron variant period required supplemental low-flow oxygen (36.3% vs. 63.4%), high-flow oxygen (8.8% vs. 25.9%), and mechanical ventilation (6.5 % vs. 10.0%). (P<0.001 for all comparisons). The need for accidental renal replacement therapy and vasopressor support was not different between the dominant Omicron and Delta periods.
We recalculated the proportion of individuals in each disease severity category using a more stringent definition of variant obesity, which limited time periods where each variant accounted for> 98% of all reported variants (October 1 to December 4, 2021 for the Delta variant and January 2 to 15 January 2022 for the Omicron variant). Results from our sensitivity analysis of 19,874 matched pairs corresponded to our primary results (Supplementary table 4).
To rule out a potentially confusing effect of previous treatment with monoclonal antibodies or nitravelmir / ritonavir (paxlovid) on the severity of the disease, we performed susceptibility studies and excluded 3861 patients who had received these treatments after a positive SARS-CoV-2 PCR test. Estimates of disease severity of 21,231 matched pairs corresponded to our primary analysis (Supplementary table 5).
Similarly, we performed additional analyzes to exclude confounding due to different hospital bed capacities during Omicron and Delta periods. It is challenging to assess the bed capacity in this context because the number of authorized beds does not necessarily correspond to the number of manned beds, ie. beds with nurse and other staff available to accommodate patients. We had information on acute medical and surgical beds in operation for 107 of 129 VA facilities included in our pre-matching data set. The average number of daily admissions for these facilities was well below the capacity of the operating bed and is therefore unlikely to confuse our estimates. The average number of admissions per day in the 24-day Omicron period was 1.62 (IQR 0.87, 2.67). In contrast, the median number of daily admissions during the 72-day Delta period was 0.49, IQR: 0.29, 0.76), resulting in a median daily admissions ratio (Omicron: Delta) of 3.25 ( IQR 1.74, 5.25). We performed sensitivity analysis on the matched data and calculated disease severity estimates for facilities with a median daily Omicron: Delta hospitalization ratio below and above the 50% percentile (<3.25 vs. ≥3.25). The results were similar in both strata, suggesting that the bed capacity for emergency care did not significantly affect our severity assessments.