Study cohort
Overall, 1123 patients were enrolled (895 as EI, 228 as non-EI). Two persons enrolled as controls reported a SARS-CoV-2 infection before September 30th 2020 but after their test-negative episode of disease; these participants were re-coded as EI participants. Of all enrolled persons, 624 EI (70%) and 175 (77%) non-EI participants had complete data (Figure 2).
Compared to non-EI participants, EI participants were more likely to live in low-income neighbourhoods, less likely to be Caucasian and to have a university education, more likely to be essential workers, and somewhat more likely to be younger (Table 1). The excess of underlying illness in non-EI participants was seen only in the outpatient group and included persons with underlying asthma, other respiratory, and cardiac disease (data not shown).
Table 1: Characteristics of the study cohorts with and without RT-PCR confirmed SARS-CoV-2 infection prior to September 30, 2020 (early infection)
|
With Early Infection1
(N=624)
|
No Early Infection1
(N=175)
|
Standardized difference
|
Sex at birth, female (N, %)
|
327 (52.4%)
|
96 (54.9%)
|
0.05
|
Age in years (median, IQR)
|
55.7y (42.6-65.1)
|
56.4y (39-68.2)
|
0.01
|
Age group (N, %)
|
|
|
|
18-49 years
|
233 (37.3%)
|
69 /(39.4%)
|
0.06
|
50-64 years
|
224 (35.9%)
|
47 (26.9%)
|
|
≥65 years
|
167 (26.8%)
|
59 (33.7%)
|
|
Underlying illness2 (N,%)
|
|
|
|
Any
|
216 (34.6%)
|
76 (43.4%)
|
0.18
|
Diabetes mellitus
|
108 (17.3%)
|
25 (14.3%)
|
0.08
|
Cardiac disease
|
64 (10.3%)
|
28 (16%)
|
0.17
|
Immunocompromised
|
38 (6.1%)
|
11 (6.3%)
|
0.01
|
Severity of eligible illness in 2020 (N, % hospitalized)
|
279 (44.7%)
|
91 (52.0%)
|
|
Children in household (N, % with children<17yr)
|
169 (27.1%)
|
58 (33.1%)
|
0.13
|
Ethnic Background (N, %)
|
|
|
|
Caucasian (White)
|
236 (37.8%)
|
93 (53.1%)
|
0.24
|
Asian (East, Southeast and South)
|
187 (30.0%)
|
42 (24.0%)
|
|
Black
|
79 (12.7%)
|
10 (5.7%)
|
|
Other
|
122 (19.6%)
|
30 (17.1%)
|
|
Education (N, %)
|
|
|
0.28
|
Secondary school or less
|
154 (24.7%)
|
36 (20.6%)
|
|
College level diploma or certificate
|
207 (33.2%)
|
38 (21.7%)
|
|
Undergraduate university degree
|
169 (27.1%)
|
60 (34.3%)
|
|
Graduate or professional degree
|
85 (13.6%)
|
39 (22.3%)
|
|
Neighbourhood income quintile (N, %)
|
|
|
0.30
|
1-lowest
|
198 (31.7%)
|
43 (24.6%)
|
|
2
|
150 (24.0%)
|
33 (18.9%)
|
|
3
|
110 (17.6%)
|
30 (17.1%)
|
|
4
|
74 (11.9%)
|
23 (13.1%)
|
|
5-highest
|
92 (14.7%)
|
46 (26.3%)
|
|
Work status during Omicron wave3 (N,%)
|
|
|
0.36
|
Essential worker
|
334 (53.5%)
|
64 (36.6%)
|
|
Other
|
287 (46.0%)
|
111 (63.4%)
|
|
Hand-to-Face habits4
|
|
|
0.18
|
None
|
175 (28.0%)
|
33 (18.9%)
|
|
≤5 times per day
≥6 times per day
|
342 (54.8%)
101 (16.2%)
|
106 (60.6%)
35 (20.0%)
|
|
COVID-19 vaccine doses received as of 11 Dec 20215
|
|
|
0.25
|
Not vaccinated
|
28 (4.5%)
|
9 (5.1%)
|
|
One dose
|
6 (1.0%)
|
2 (1.1%)
|
|
Two doses
|
466 (74.7%)
|
111 (63.4%)
|
|
Three doses
|
124 (19.9%)
|
53 (30.3%)
|
|
1Note that totals in columns may differ from overall total because of missing data for some variables
2Any underlying illness predisposing to complications of viral respiratory illness (REF NACI). Diabetes mellitus and underlying cardiac disease were the only two conditions present in at least 10% of participants. Immunocompromised includes both immunocompromising illness and therapy (10).
3 Work status was categorized as essential if participants reported working full- or part-time in-person outside of the home. Others comprised those working only at home or who were students, retired or unemployed.
4Hand-to-face habits were defined as rubbing eyes, biting nails or cuticles or habitually putting fingers into the mouth or nose.
5December 11, 2021 represents the start of the Omicron wave in the population area (see text); 50% of participants without EI had received a 3rd dose as of 17 December; while 50% of those with EI had received their 3rd dose by 28 December
At the start of Omicron on 12 December 2021, 94.5% (590/624) EI participants and 93.7% (164/175) non-EI participants had received at least two doses of the vaccine. There was no difference in the timing of the first two doses of vaccine among EI and non-EI participants (median dates of second dose receipt were 21 June and 18 Jun 2021, respectively). Non-EI participants received their third doses slightly earlier than EI participants (median dates of receipt 18 December 2021 and 4 January 2022, respectively P<.0001) such that non-EI recipients were more likely have received a third dose on 12 December 2021 (30.3% versus 19.9%, P=0.003; Table 1). Detailed timing of vaccine doses in the two cohorts is shown in Supplementary Figure 2. At the end of study follow-up, 73.4% (458/624) of EI participants and 76.7% (134/175) of non-EI participants had received a third dose of a COVID-19 vaccine (P=0.26), 32% (200/624) of EI and 51% (89/175) of non-EI participants had received a fourth dose , and 9.9% (62/624) of EI and 22% (38/175) non-EI had received a fifth dose (P<.0005 for both).
SARS-CoV-2 infections during follow-up
Participants were followed for a median of 31.8 (IQR 19.9-33.1) months. Among the 799 participants, 273 individuals had a total of 288 episodes of SARS-CoV-2 infection between 1 October 2020 and 30 January 2023 (Figure 3). Of 624 EI participants, 175 (28%) had one SARS-CoV-2 re-infection episode and 8 (1.3%) had two re-infection episodes during follow-up. Among 175 non-EI participants, 84 (48%) had one infection, 5 (2.9%) had two infections and 1 (0.6%) had 3 infections (P<0.0001 compared to EI participants).
Overall, 269/288 (93.4%) identified infections during study follow-up were symptomatic, and 7 (2.4%) required hospitalization. Symptomatic infections comprised 169/183 (92%) of first re-infections among EI participants, and 87/90 (97%) first infections among non-EI participants (P=0.19); hospitalizations occurred in 6/183 EI (3.3%) re-infections and 1/90 (1.1%) non-EI infections (P=0.32). Overall, 14 infections occurred prior to the Omicron wave: these participants were excluded from the analysis of risk factors for infection during Omicron.
Factors associated with SARS-CoV-2 infection during Omicron
In univariate Cox regression, infection during the Omicron wave was less common in those with EI, in older adults, and in persons who had more vaccine doses during the Omicron period (Figure 4A and Table 2). Infection was more common in persons originally treated as outpatients, in essential workers, in those living in higher income neighbourhoods, and in persons with frequent hand-to-face habits.
Table 2: Univariate and multivariable analysis of factors associated with SARS CoV-2 infection in study participants from 12 December 2021 to end of the study (31 January 2023)
|
Rate per 1000 participant days
|
Univariate Cox
|
Multivariable Cox
|
|
Rate (95% CI)
|
Hazard ratio, 95% CI
|
P value
|
Hazard ratio
95% CI
|
P value
|
Early SARS-CoV-2 infection:1 No (non-EI)
|
1.75 (1.41-2.18)
|
Reference
|
|
Reference
|
|
Yes (EI)
|
0.86 (0.74-0.99)
|
0.49 (0.38-0.64)
|
<.0001
|
0.56 (0.43-0.74)
|
<.001
|
Sex at birth: Male
|
0.94 (0.78-1.13)
|
Reference
|
|
|
|
Female
|
1.09 (0.93-1.29)
|
1.16 (0.90-1.48)
|
0.25
|
-
|
|
Age group: 18-49 years
|
1.40 (1.17-1.67)
|
Reference
|
|
|
|
50-64 years
|
0.91 (0.73-1.13)
|
0.66 (0.50-0.87)
|
0.0001
|
NE6
|
0.022
|
≥65 years
|
0.72 (0.55-0.93)
|
0.52 (0.38-0.71)
|
|
|
|
Underlying illness: Any2
|
0.94 (0.76-1.15)
|
0.87 (0.68-1.13)
|
0.31
|
-
|
|
Diabetes mellitus
|
0.84 (0.61-1.15)
|
0.79 (0.56-1.12)
|
0.18
|
|
|
Cardiac disease
|
0.80 (0.54-1.18)
|
0.76 (0.50-1.15)
|
0.20
|
|
|
Immunocompromised
|
1.27 (0.79-2.04)
|
1.24 (0.76-2.03)
|
0.38
|
|
|
Severity of 2020 illness: Hospitalized
|
0.82 (0.68-1.00)
|
Reference
|
|
-
|
|
Not hospitalized
|
1.21 (1.03-1.41)
|
1.46 (1.14-1.87)
|
0.003
|
|
|
Children (≤17yrs) in household: No
|
0.85 (0.73-1.00)
|
Reference
|
|
Reference
|
|
Yes
|
1.53 (1.25-1.86)
|
1.77 (1.37-2.27)
|
<.0001
|
1.60 (1.19-2.15)
|
0.002
|
Ethnic Background: Caucasian
|
1.29 (1.08-1.53)
|
Reference
|
|
Reference
|
|
Asian (East/SE/South)
|
0.88 (0.69-1.12)
|
0.68 (0.51-0.92)
|
0.006
|
0.57 (0.41-0.79)
|
<.001
|
Black
|
0.63 (0.40-0.99)
|
0.50 (0.31-0.80)
|
|
0.57 (0.34-0.95)
|
0.03
|
Other
|
0.94 (0.70-1.25)
|
0.73 (0.52-1.03)
|
|
0.70 (0.49-1.00)
|
0.05
|
Education: Secondary school or less
|
0.64 (0.47-0.87)
|
Reference
|
|
Reference
|
|
College/diploma/certificate
|
0.94 (0.75-1.18)
|
1.46 (1.00-2.14)
|
0.0002
|
1.32 (0.09-1.95)
|
0.16
|
Undergraduate university degree
|
1.31 (1.07-1.61)
|
2.03 (1.40-2.94)
|
|
1.59 (1.08-2.32)
|
0.02
|
Graduate or professional degree
|
1.42 (1.08-1.85)
|
2.19 (1.46-3.29)
|
|
1.52 (0.99-2.32)
|
0.06
|
Neighbourhood income quintile: 1-lowest
|
0.71 (0.54-0.92)
|
Reference
|
|
Reference
|
|
2
|
0.92 (0.71-1.20)
|
1.30 (0.89-1.88)
|
0.001
|
1.27 (0.88,1.83)
|
0.20
|
3
|
1.15 (0.87-1.51)
|
1.61 (1.10-2.36)
|
|
1.46 (0.99-2.14)
|
0.06
|
4
|
1.57 (1.17-2.11)
|
2.19 (1.47-3.25)
|
|
1.95 (1.31-2.91)
|
0.001
|
5 – highest
|
1.26 (0.97-1.65)
|
1.77 (1.22-2.58)
|
|
1.43 (0.96-2.12)
|
0.08
|
Work status, Omicron3: Other
|
0.88 (0.73-1.06)
|
Reference
|
|
Reference
|
|
Essential worker
|
1.17 (1.00-1.38)
|
1.33 (1.04-1.70)
|
0.02
|
1.29 (0.98-1.68)
|
0.07
|
Hand-to-face habits4: None
|
0.74 (0.56-0.97)
|
Reference
|
|
|
|
≤5 times per day
|
1.12 (0.96-1.31)
|
1.51 (1.10-2.06)
|
0.02
|
-
|
|
≥6 times per day
|
1.25 (0.95-1.64)
|
1.68 (1.14-2.47)
|
|
|
|
COVID-19 vaccine doses before Omicron5
|
|
|
|
|
|
Not vaccinated
|
1.28 (0.73-2.25)
|
Reference
|
|
-
|
|
One dose
|
0.53 (0.13-2.13)
|
0.42 (0.09-1.89)
|
0.46
|
|
|
Two doses
|
0.99 (0.86-1.13)
|
0.78 (0.43-1.39)
|
|
|
|
Three doses
|
1.18 (0.89-1.57)
|
0.93 (0.49-1.75)
|
|
|
|
Number of vaccine doses during Omicron5
|
|
|
|
|
|
None
|
1.67 (1.40-1.99)
|
Reference
|
|
NE7
|
|
One dose
|
1.03 (0.85-1.24)
|
0.62 (0.48-0.81)
|
<.0001
|
|
|
Two doses
|
0.38 (0.25-0.60)
|
0.23 (0.15-0.38)
|
|
|
|
Three doses
|
0.21 (0.08-0.55)
|
0.13 (0.05-0.34)
|
|
|
|
Time since most recent vaccine dose
|
|
|
|
NE7
|
|
Not vaccinated
|
1.32 (0.73-2.38)
|
NA
|
|
|
|
≥120 days
|
1.12 (0.97-1.30)
|
Reference
|
|
|
|
75 - 119 days
|
0.97 (0.69-1.37)
|
0.95 (0.64-1.41)
|
0.78
|
|
|
14-74 days
|
0.75 (0.56-1.00)
|
0.71 (0.50-1.01)
|
0.06
|
|
|
Abbreviations: SE southeast, FT Full-time, PT part time, CI Confidence Interval, NE not estimable
1Early infection is defined as an RT-PCR confirmed SARS-CoV-2 infection between January 24 and September 30, 2020.
2Any underlying illness predisposing to complications of viral respiratory illness (REF NACI). Diabetes mellitus and underlying cardiac disease were the only two conditions present in at least 10% of participants. Immunocompromised includes both immunocompromising illness and therapy (10).
3Work status was categorized as essential if participants reported working full- or part- time in-person outside of the home. Other comprised those working only at home or who were students, retired or unemployed
4Hand-to-face habits were defined as rubbing eyes, biting nails or cuticles or habitually putting fingers into the mouth or nose.
5The start of the Omicron wave in the population area was December 12, 2021 (see text); vaccine dates are lagged by 14 days.
6In multivariable analysis, age was modelled with a restricted cubic spline; the hazard ratio per decade of age was 0.82 (95% CI 0.66-0.90).
7Vaccine doses were modelled as time varying in the multivariable analysis such that direct estimates are not available.
In multivariable analysis, the hazard ratios associated with early COVID-19 infection did not change relative to univariate analysis (Figure 4B and Table 2). In Cox regression analysis the HR (95%CI) for those with early infection versus those without was 0.56 (0.43-0.74), in the piecewise exponential model, the incidence rate ratio (95%CI) was 0.57 (0.44-0.75). In multivariable sensitivity analyses, the hazard ratio (95%CI) associated with prior infection was 0.52 (0.40-0.69) when only symptomatic infections were included, 0.55 (0.37-0.82) when only PCR-confirmed infections were included, and 0.54 (0.39-0.74) when only EIs occurring before 31 May 2020 were included. Other factors associated with infection during the Omicron wave included younger age, being ethnically White, higher neighbourhood income, and living in a household with children (Table 2).
Early infection and risk of infection with SARS-CoV-2 variants
Incidence rates of COVID-19 in EI and non-EI participants during periods of activity due to different variants, as well as the hazard ratio associated with infection in different subvariant periods in multivariable Cox regression analysis are shown in Table 3. Protection was highest against infections in the pre-Omicron era; estimates of reductions in IRR and adjusted infection hazard ratios during periods of activity with different predominant Omicron variants were not statistically significantly different. Early infection and infection during the BA.1/2 wave appeared to independently increase protection against later Omicron infections (Table 4).
Table 3: Incidence of Omicron infection by time period1 in those with and without RT-PCR confirmed SARS-CoV-2 infection prior to September 30, 2020 (Early Infection)
|
Early Infection
(EI)
|
No Early Infection
(non-EI)
|
EI vs non -EI
IRR (95% CI)
|
EI vs. Non-EI Hazard Ratio2 (95% CI)
|
Prior to Omicron
|
|
|
|
|
Number of COVID-19 infections
|
5
|
9
|
|
|
Participant-days
|
369651
|
101273.7
|
|
|
COVID-19 rate/1000 participant days
|
0.014
|
0.089
|
0.15 (0.05-0.45)
|
|
Omicron BA.1/BA.2
|
|
|
|
|
Number of COVID-19 infections
|
91
|
40
|
|
|
Participant-days
|
107331
|
30142
|
|
|
COVID-19 rate/1000 participant days
|
0.85
|
1.33
|
0.64 (0.44-0.93)
|
0.66 (0.45-0.97)
|
Omicron BA.4/BA.5
|
|
|
|
|
Number of COVID-19 infections
|
68
|
38
|
|
|
Participant-days
|
96844
|
26221
|
|
|
COVID-19 rate/1000 participant days
|
0.70
|
1.45
|
0.48 (0.33-0.72)
|
0.44 (0.28-0.68)
|
Mixed BA.5/BQ1.1/XBB
|
|
|
|
|
Number of COVID-19 infections
|
27
|
10
|
|
|
Participant-days
|
44859
|
11817
|
|
|
COVID-19 rate/1000 participant days
|
0.60
|
0.85
|
0.71 (0.34-1.47)
|
0.71 (0.32-1.56)
|
1Time periods are defined as follow: prior to Omicron, 25 January 2020 – 11 December 2021; Omicron BA1/BA.2, 12 December 2021 – 3 Jun 2022; Omicron BA4/BA5, 4 June 1 - 13 November 2022; and mixed BA.5/BQ1.1/XBB, 14 November 2022 - 31 January 2023 (end of study).
2Hazard ratios from multivariable analysis adjusted for age, COVID-19 vaccination, severity of 2020 illness, neighborhood income quantile, ethnicity, education, work status during omicron and presence of children in the household.
Table 4: Incidence of SARS-CoV-2 from June 4, 2022 to January 31, 2023 (due to BA.4/5 and later subvariants) in those with and without RT-PCR confirmed SARS-CoV-2 infection prior to September 30, 2020 (Early Infection)
|
Early Infection
(EI)
|
No Early Infection
(non-EI)
|
EI vs non -EI
IRR (95% CI)
|
EI vs. Non-EI Hazard Ratio1 (95% CI)
|
No BA.1/BA.2 infection
|
|
|
|
|
Number of COVID-19 infections
|
89
|
42
|
|
|
Participant-days
|
121623
|
28779
|
|
|
COVID-19 rate/1000 participant days
|
0.73
|
1.46
|
0.50 (0.35-0.72)
|
0.47 (0.32-0.68)
|
With BA.1/BA.2 infection
|
|
|
|
|
Number of COVID-19 infections
|
4
|
5
|
|
|
Participant-days
|
20665
|
9413
|
|
|
COVID-19 rate/1000 participant days
|
0.19
|
0.53
|
0.36 (0.10-1.36)
|
0.20 (0.03-1.21)
|
1Hazard ratios from multivariable analysis adjusted for age, COVID-19 vaccination, severity of 2020 illness, neighborhood income quantile, ethnicity, education, work status during omicron and presence of children in the household.