Abstract
Background:
Vaccine effectiveness (VE) estimates vary by population characteristics and circulating variants. North America and Europe have generated many COVID-19 VE estimates but relied heavily on mRNA vaccines. Fewer estimates are available for non-mRNA vaccines and from Latin America. We aimed to estimate the effectiveness of several COVID-19 vaccines in preventing SARS-CoV-2-associated severe acute respiratory infection (SARI) in Paraguay from May 2021 to April 2022.
Methods:
Using sentinel surveillance data from four hospitals in Paraguay, we conducted a test-negative case-control study to estimate COVID-19 vaccine effectiveness against SARI by vaccine type/brand and period of SARS-CoV-2 variant predominance (Gamma, Delta, Omicron). We used multivariable logistic regression adjusting for month of symptom onset, age group, and presence of ≥1 comorbidity to estimate the odds of COVID-19 vaccination in SARS-CoV-2 test-positive SARI case-patients compared to SARS-CoV-2 test-negative SARI control-patients.
Results:
Of 4,229 SARI patients, 2,381 (56%) were SARS-CoV-2-positive case-patients and 1,848 (44%) were SARS-CoV-2-negative control-patients. A greater proportion of case-patients (73%; 95% CI: 71–75) than of control-patients (40%; 95% CI: 38–42) were unvaccinated. During the Gamma variant-predominant period, VE estimates for partial vaccination with mRNA vaccines and Oxford/AstraZeneca Vaxzevria were 90.4% (95% CI: 66.4–97.6) and 52.2% (95% CI: 25.0–69.0), respectively. During the Delta variant-predominant period, VE estimates for complete vaccination with mRNA vaccines, Oxford/AstraZeneca Vaxzevria, or Gamaleya Sputnik V were 90.4% (95% CI: 74.3–97.3), 83.2% (95% CI: 67.8–91.9), and 82.9% (95% CI: 53.0–95.2), respectively. The effectiveness of all vaccines declined substantially during the Omicron variant-predominant period.
Conclusions:
This study contributes to our understanding of COVID-19 VE in Latin America and to global understanding of vaccines that have not been widely used in North America and Europe. VE estimates from Paraguay can parameterize models to estimate the impact of the national COVID-19 vaccination campaign in Paraguay and similar settings.
Keywords: COVID-19, Vaccine effectiveness, Vaccination, Paraguay
1. Introduction
Randomized, placebo-controlled clinical trials have demonstrated high efficacy of COVID-19 vaccines against severe disease [1–3]. However, vaccine effectiveness (VE), the reduction in the risk of infection or disease among vaccinated individuals in real-world conditions, may differ depending on population characteristics, vaccine schedules, and vaccine-variant match with the dominant circulating SARS-CoV-2 variant. VE estimates help to guide vaccination policies and decision making by addressing evidence gaps from clinical trials including effectiveness in specific populations, duration of protection, and protection against variants of concern. These estimates are also essential for evaluating the impact of vaccination at the population level in terms of the burden of disease averted through vaccination campaigns [4].
Following the emergence of SARS-CoV-2, Paraguay collaborated with the Network for Vaccine Effectiveness Evaluation in Latin America and the Caribbean (REVELAC-i) [5] to leverage its severe acute respiratory infection (SARI) sentinel hospital surveillance system to evaluate the effectiveness of COVID-19 vaccines [5,6]. In Paraguay, COVID-19 vaccination started on 22 February 2021, coinciding with the start of an epidemic wave in which the SARS-CoV-2 Gamma variant was the predominant circulating virus. The national COVID-19 vaccination campaign included mRNA vaccines (Pfizer/BioNTech Comirnaty, Moderna Spikevax), recombinant vector vaccines (Oxford/AstraZeneca Vaxzevria, Gamaleya Sputnik V), and inactivated virus vaccines (Bharat Biotech Covaxin, Sinopharm [Beijing] Covilo). By 13 August 2021, when the Delta variant became the predominant circulating strain, 3,278,382 doses of vaccine had been administered, corresponding to vaccine coverage of 39.7% of the population with at least one dose and 18% with a complete 2-dose primary series [7]. In order of priority, target groups for vaccination were healthcare workers and adults aged ≥60 years; adults aged 18–59 years with pre-existing conditions, essential workers, elementary school teachers, and the indigenous population; workers at points of entry, the military, police, firefighters, and incarcerated persons. The general population became eligible for vaccination on March 13, 2021. A booster dose vaccination campaign began on 18 October 2021 for healthcare workers and persons aged >50 years who had received the second dose prior to 30 April 2021. On 17 December 2021, when an Omicron variant wave began, booster doses became available to people who had received the second dose at least four months earlier.
Monitoring VE at the WHO regional and national levels is important because estimates vary by population characteristics and circulating variants; estimates from one population may not be applicable to others [8]. For example, the Gamma variant was predominant in Latin America when the Alpha and Delta variants predominated in Europe and North America [9]. Moreover, North America and Europe relied primarily on mRNA vaccines whereas Latin America used a wider range of vaccines for which effectiveness is less well described [7,10,11]. While VE estimates have been generated elsewhere, particularly in North America and Europe, there are fewer estimates from Latin America and for non-mRNA vaccines [12–24]. We aimed to estimate the effectiveness of COVID-19 vaccines in preventing SARS-CoV-2-associated SARI by type of vaccine and circulating SARS-CoV-2 variant (Gamma, Delta, and Omicron) from May 2021 to April 2022 in Paraguay.
2. Methods
2.1. Study design and setting
We conducted a test-negative case-control study to estimate the odds of vaccination in SARS-CoV-2 test-positive SARI cases (case-patient) compared to SARS-CoV-2 test-negative SARI controls (control-patient). SARI was defined using the standard WHO case definition as the presence of an acute respiratory infection with a history of fever or a measured fever of 38 °C or higher, cough, and onset within the past 10 days resulting in a minimum hospital stay of 24 h [25]. Inpatients with SARI were enrolled from four hospitals in Paraguay that conduct SARI sentinel surveillance as part of the Pan American Health Organization’s (PAHO) Severe Acute Respiratory Infections Network (SARInet) [6]. Hospital Central del Instituto de Previsión Social (IPS) and Hospital Integrado IPS Ingavi are respectively 1,300- and 357-bed hospitals located in the capital city, Asunción, whereas Hospital Regional de Ciudad Del Este and Hospital Integrado IPS de Ciudad Del Este are respectively 158- and 122-bed hospitals located in Alto Parana in the southeast of the country (Table 1).
Table 1.
Participating Study Sites in Paraguay that are part of the Pan American Health Organization’s (PAHO) Severe Acute Respiratory Infections Network (SARI-net).
| Hospital Central-IPS | Hospital Integrado IPS Ingavi | Hospital Regional de Ciudad Del Este | Hospital Integrado IPS de Ciudad Del Este | |
|---|---|---|---|---|
|
| ||||
| Setting | Urban - Capital City | Urban - Capital City | Urban - State capital | Urban - State capital |
| Facility type | Tertiary-level referral hospital | Tertiary-level referral hospital | Tertiary-level referral hospital | Tertiary-level referral hospital |
| No. of Beds | 1300 | 357 | 158 | 122 |
| Total annual hospital admissions | 70,950 | 7438 | 7616 | 6735 |
| Population | Inpatient adults and children all ages | Inpatient adults and children all ages | Inpatient adults and children all ages | Inpatient adults and children all ages |
2.2. Study population and study period
We included patients aged ≥5 years with SARI who were hospitalized from 1 May 2021 to 2 April 2022 and who were eligible for COVID-19 vaccination at admission. The study period included three periods of SARS-CoV-2 variant predominance, defined according to genomic surveillance data from Paraguay’s Central Public Health Laboratory: the Gamma variant-predominant period (1 May to 7 August 2021), the Delta variant-predominant period (22 August to 18 December 2021), and the Omicron variant-predominant period (2 January to 2 April 2022).
2.3. Outcome definition
Naso- and oropharyngeal swabs were taken from SARI patients for the detection of SARS-CoV-2 by real-time RT-PCR. Case-patients were defined as SARI patients with a positive SARS-CoV-2 PCR test and control-patients were defined as SARI patients with a negative SARS-CoV-2 PCR test.
2.4. Exposure definition
We used the electronic records of Paraguay’s Expanded Program on Immunization (EPI) to ascertain COVID-19 vaccination status. All facilities can register vaccines in the EPI database under a person’s national ID number. Patients with SARI were matched to vaccination records using unique national ID numbers. In Paraguay, eight vaccines were available during the study period: Bharat Biotech Covaxin, Moderna Spikevax, Pfizer/BioNTech Comirnaty, Oxford/AstraZeneca Vaxzevria, Sinopharm (Beijing) Covilo, Sinopharm Wuhan, Gamaleya Sputnik V, and CoronaVac-Sinovac. However, few patients with SARI (n = 22) received CoronaVac-Sinovac and Sinopharm Wuhan so we could not estimate VE for these vaccines. We considered individuals to be partially vaccinated if they received a single dose of a vaccine with a two-dose schedule at least 14 days before the onset of symptoms. Individuals who received two doses at least 14 days before the onset of symptoms were considered fully vaccinated. Individuals with a complete primary schedule who received an additional dose at least 14 days before the onset of symptoms were considered boosted. Participants who had not received any doses of the vaccine or who were vaccinated after illness onset were classified as unvaccinated.
2.5. Exclusions
We excluded patients with >10 days between illness onset and specimen collection to align with the WHO SARI case definition [26]. SARI cases with missing illness onset or vaccination date(s) as well as those without recorded vaccination status in the EPI database (such as persons without a national ID number) were omitted due to potential vaccination status (exposure) misclassification. Additionally, we excluded patients who were not yet eligible for COVID-19 vaccination at the time of admission or for whom laboratory results were not available. Patients who received a dose of vaccine within 14 days prior to symptom onset were excluded from the analysis. As analyses were stratified by period of variant predominance, patients with admission dates in the two transition weeks between periods of variant predominance were excluded. A 2-week transition period was chosen because the Delta and Omicron variants were predominant (comprised 75% of sequences) after two weeks from initial detection.
2.6. Covariates
We abstracted demographic and clinical information from SARI surveillance databases. Data collected included sex, age, pre-existing medical conditions, pregnancy status, hospital admission date, and date of symptom onset.
2.7. Statistical analysis
We compared demographic and clinical characteristics as well as vaccination status of case-patients and control-patients. For categorical variables, we assessed associations with case-control status using the χ2 test or Fisher’s exact test. For continuous variables, we assessed differences in distributions using the Mann-Whitney test.
We used multivariable logistic regression to estimate the odds ratio of vaccination in case-patients versus control-patients, adjusting for month of symptom onset, age group (5–29, 30–59, 60–79, 80+ years), and presence of at least one comorbidity (asthma, hypertension, diabetes, cardiovascular disease, immunocompromised, lung disease, or obesity). We identified potential confounders a priori and a series of bivariate models stratified by potential confounders were used to assess confounding. We estimated VE for partial vaccination during the Gamma variant predominant period (when few SARI cases had been fully vaccinated) and VE for complete primary series vaccination during the Delta and Omicron variant-predominant periods. We estimated VE by vaccine brand but combined Pfizer and Moderna into one VE estimate for mRNA vaccines due to small numbers. As booster doses were not introduced before October 2021, we estimated VE for receipt of a booster dose during the Omicron variant-predominant period only. We estimated VE as 1- adjusted odds ratio of vaccination in case-patients versus control-patients × 100 for each model as well as their 95% confidence intervals. Estimates with confidence interval widths >200 were not shown due to sparse data bias [27].
We performed all analyses using Stata software (version SE 14.2) and R version 4.2.2.
2.8. Ethical considerations
The Pan American Health Organization Ethical Review Committee reviewed the protocol and determined that it does not constitute human subjects research.
3. Results
3.1. SARI patient enrollment
From 1 May 2021 to 2 April 2022, 7,350 SARI patients were admitted to the four participating sentinel surveillance hospitals; of these, 4,229 (58%) met the eligibility criteria and were included in the analysis (Fig. 1). Among the 4,229 study participants, 2,381 (56%) were SARS-CoV-2-positive case-patients and 1,848 (44%) were SARS-CoV-2-negative control-patients (Table 2). The majority (n = 1,624, 68%) of SARS-CoV-2-positive case-patients had illness onset during the Gamma variant-predominant period (May to August 2021) when the proportion (n = 64, 4%) of SARI patients with complete primary series vaccination was lowest (Fig. 1, Fig. 2). Most (n = 1,526, 83%) of the test-negative control-patients were hospitalized from August 2021 onwards. Starting in week 52 of 2021, the number of test-positive case-patients increased with the emergence of the SARS-CoV-2 Omicron variant, with incidence peaking in week 3 of 2022 (Fig. 2).
Fig. 1.
Selection of severe acute respiratory infection (SARI) patients for inclusion in COVID-19 vaccine effectiveness analyses at four hospitals, REVELAC-i, Paraguay, 1 May 2021– 2 April 2022.
Table 2.
Demographic and clinical characteristics of enrolled patients with COVID-19–associated severe acute respiratory infections (SARI) at four hospitals, by case status and variant-predominant period,* REVELAC-i, Paraguay, 1 May 2021–2 April 2022 (n = 4229)†.
| Gamma |
Delta |
Omicron |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| Characteristic | SARS-CoV-2 positive SARI cases | SARS-CoV-2 negative SARI controls | p-value | SARS-CoV-2 positive SARI cases | SARS-CoV-2 negative SARI controls | p-value | SARS-CoV-2 positive SARI cases | SARS-CoV-2 negative SARI controls | p-value |
| (N = 1,624) | (N = 322) | (N = 200) | (N = 846) | (N = 557) | (N = 680) | ||||
|
| |||||||||
| Median Age — yr (IQR) | 57 (43–69) | 63 (46–76) | <0.001 | 59 (46–71) | 72 (59–81) | <0.001 | 72 (60–81) | 67 (44–78) | <0.001 |
| Age group — no. (%) | <0.001 | <0.001 | <0.001 | ||||||
| 5–29 yrs | 90 (5.5) | 40 (12.4) | 10 (5.0) | 60 (7.1) | 36 (6.5) | 114 (16.8) | |||
| 30–59 yrs | 791 (48.7) | 99 (30.7) | 92 (46.0) | 157 (18.6) | 103 (18.5) | 135 (19.9) | |||
| 60–79 yrs | 620 (38.2) | 128 (39.8) | 72 (36.0) | 382 (45.2) | 267 (47.9) | 287 (42.2) | |||
| ≥80 yrs | 123 (7.6) | 55 (17.1) | 26 (13.0) | 247 (29.2) | 151 (27.1) | 144 (21.2) | |||
| Sex — no. (%) | 0.054 | 0.073 | 0.568 | ||||||
| Male | 925 (57.0) | 164 (50.9) | 110 (55.0) | 403 (47.6) | 289 (51.9) | 365 (53.7) | |||
| Female | 699 (43.0) | 158 (49.1) | 90 (45.0) | 443 (52.4) | 268 (48.1) | 315 (46.3) | |||
| Pregnant — no. (%)** | 10 (1.4) | 5 (3.2) | 0.133 | 4 (4.4) | 2 (0.5) | 0.009 | 6 (2.2) | 8 (2.5) | 0.813 |
| ≥ 1 pre-existing condition — no. (%) | 1004 (61.8) | 211 (65.5) | 0.234 | 129 (64.5) | 654 (77.3) | <0.001 | 394 (70.7) | 466 (68.6) | 0.460 |
| COVID-19 vaccination — no. (%) | <0.001 | <0.001 | 0.298 | ||||||
| Unvaccinated | 1388 (85.5) | 234 (72.7) | 128 (64.0) | 266 (31.4) | 216 (38.8) | 235 (34.6) | |||
| Partial vaccination | 195 (12.0) | 65 (20.2) | 20 (10.0) | 144 (17.0) | 49 (8.8) | 62 (9.1) | |||
| Full vaccination (primary series) | 41 (2.5) | 23 (7.1) | 52 (26.0) | 422 (49.9) | 221 (39.7) | 304 (44.7) | |||
| Booster vaccination (third dose) | 0 | 0 | 0 (0.0) | 14 (1.7) | 71 (12.7) | 79 (11.6) | |||
| Type of vaccine (1st dose) — no. (%)╪ | 0.036 | 0.228 | 0.001 | ||||||
| Oxford/ AstraZeneca Vaxzevria | 86 (36.4) | 39 (44.3) | 24 (33.3) | 236 (40.7) | 148 (43.4) | 155 (34.8) | |||
| Covaxin/Bharat | 97 (41.1) | 26 (29.5) | 22 (30.6) | 131 (22.6) | 80 (23.5) | 81 (18.2) | |||
| Pfizer/BioNTech Comirnaty | 2 (0.8) | 3 (3.4) | 3 (4.2) | 67 (11.6) | 29 (8.5) | 71 (16.0) | |||
| Gamaleya Sputnik V | 28 (11.9) | 8 (9.1) | 8 (11.1) | 70 (12.1) | 44 (12.9) | 57 (12.8) | |||
| Sinopharm (Beijing) Covilo | 18 (7.6) | 7 (8.0) | 13 (18.1) | 65 (11.2) | 33 (9.7) | 57 (12.8) | |||
| Moderna/Spikevax | 2 (0.8) | 4 (4.5) | 1 (1.4) | 7 (1.2) | 6 (1.8) | 12 (2.7) | |||
| Sinovac/Coronavac | 3 (1.3) | 0 (0.0) | 1 (1.4) | 3 (0.5) | 1 (0.3) | 12 (2.7) | |||
| Sinopharm Wuhan | 0 (0.0) | 1 (1.1) | 0 (0.0) | 1 (0.2) | 0 | 0 | |||
| Outcome— no. (%) | <0.001 | <0.001 | <0.001 | ||||||
| Died | 676 (41.6) | 90 (28.0) | 73 (36.5) | 157 (18.6) | 220 (39.5) | 159 (23.4) | |||
| Discharged/under treatment | 733 (45.1) | 168 (52.2) | 91 (45.5) | 473 (55.9) | 295 (53.0) | 378 (55.6) | |||
| Unknown | 215 (13.2) | 64 (19.9) | 36 (18.0) | 216 (25.5) | 42 (7.5) | 143 (21.0) | |||
| Days between symptom onset and sampling — median (IQR) | 6 days (3–8) | 5 days (3–7) | 0.104 | 6 days (4–8) | 3 days (1–6) | <0.001 | 5 days (3–7) | 4 days (2–6) | <0.001 |
| Days between vaccination (1st dose) and symptom onset — median (IQR) | 33 (21–49) | 57 (28–77) | <0.001 | 142 (91–186) | 143 (113–183) | 0.391 | 243 (204–265) | 247 (202–277) | 0.036 |
| Days between last vaccination and symptom onset — median (IQR) | 28 (20–42) | 44 (23–64) | <0.001 | 98 (76–137) | 99 (62–126) | 0.209 | 152 (76–203) | 163 (87–215) | 0.022 |
The Gamma predominant period is defined as 1 May to 7 August 2021. The Delta predominant period is defined as 22 August to 18 December 2021. The Omicron predominant period is defined as 2 January to 2 April 2022.
SARI patients with symptom onsets during the 2-week transition periods between variant-predominant periods (8–21 August 2021 and 19 December 2021 – 1 January 2022) were excluded.
The numbers of female SARI cases were used as the denominators for percentages of pregnant women.
Two participants received different vaccine types for their first and second doses; all other participants received the same vaccine type. The numbers of vaccinated participants were used as the denominators for the percentage that received each vaccine type.
Fig. 2.
Distribution of patients with SARS-CoV-2 positive and negative severe acute respiratory infections (SARI) by week of hospital admission and variant predominant period, Paraguay, May 2021 to April 2022 (n = 4,512).
3.2. Characteristics of SARI patients
During the study period, 1,063 (25%) patients with SARI had completed the primary vaccine series, 535 (13%) were partially vaccinated, 164 (4%) were boosted, and 2,467 (58%) were unvaccinated (Table 2). The most frequently received vaccine was Oxford/AstraZeneca Vaxzevria (n = 688; 39%), followed by Bharat Biotech Covaxin (n = 437; 25%), Gamaleya Sputnik V (n = 215; 12%), Sinopharm (Beijing) Covilo (n = 193; 11%), Pfizer/BioNTech Comirnaty (n = 175; 10%), Moderna Spikevax (n = 32; 2%), Sinovac (n = 20; 1%), and Sinopharm Wuhan (n = 2; 0%). However, the proportion of SARI patients whose first dose was an mRNA vaccine increased over time from 3% (n = 11) in the Gamma variant-predominant period to 12% (n = 78) and 15% (n = 118) in the Delta and Omicron variant-predominant periods, respectively. The proportion that had received Bharat Biotech Covaxin as a first dose declined from 35% (n = 123) in the Gamma period to 24% (n = 153) and 20% (n = 161) in the Delta and Omicron periods respectively.
A greater proportion of test-positive case-patients (73%; 95% CI: 71–75) than of test-negative control-patients (40%; 95% CI: 38–42) were unvaccinated (Table 2). Primary series vaccination increased over time as vaccines became more widely available. Specifically, among case-patients, primary series vaccination increased from 2.5% in the Gamma period to 26% in the Delta period and 39.7% in the Omicron period. Among control-patients, primary series vaccination was 7.1% in the Gamma period, 49.9% in the Delta period, and 44.7% in the Omicron period.
Compared to control-patients, a greater proportion of case-patients were male (56% [95% CI: 54–58] vs. 50% [95% CI: 48–53]) and a smaller proportion had an underlying medical condition (64% [95% CI: 62–66] vs. 72% [95% CI: 70–74]). Case-patients were younger than control-patients (median age 61 [IQR: 46–73] vs. 69 [IQR: 52–79] years). The median age of case-patients, as well as the proportion with at least 1 pre-existing medical condition, increased over time (Table 2). Case fatality was high across all periods of variant predominance. Overall, among those with known outcome status, nearly half (46%, 95% CI: 44–48) of case-patients died compared to 29% (95% CI: 26–31) of control-patients. The median time between symptom onset and sample collection was greater in case-patients (6 [IQR: 3–8] vs. 4 [IQR: 2–6] days). The median time between receipt of last vaccine dose and symptom onset increased over time in both case-patients and control patients but was shorter in case-patients than control patients in the Gamma (28 [IQR: 20–42] vs 44 [IQR: 23–64] days) and Omicron (152 [IQR: 76–203] vs 163 [IQR: 87–215] days) variant-predominant periods. Overall, we did not observe significant differences in pregnancy status between case-patients and control-patients.
3.3. VE for partial vaccination during the Gamma variant-predominant period
From 1 May to 7 August 2021, the median time between receipt of last vaccine dose and onset of symptoms was 31 days (IQR: 21–47). Adjusting for age group and presence of ≥ 1 pre-existing condition, the effectiveness of a single dose (partial vaccination) was 90.4% (95% CI: 66.4–97.6) for mRNA vaccines and 52.2% (95% CI: 25.0–69.0) for Oxford/AstraZeneca Vaxzevria. VE estimates for Gamaleya Sputnik V (42.6%, 95% CI: −39.7; 73.8) and Sinopharm (Beijing) Covilo (47.0%, 95% CI: −52.4; 79.0) were lower with wide confidence intervals (Table 3, Fig. 3).
Table 3.
Effectiveness of partial COVID-19 vaccination against hospitalization by vaccine type during the period of SARS-CoV-2 Gamma variant predominance, 1 May – 7 August 2021, Paraguay (n = 1,882).
| SARS-CoV-2-positive SARI cases | SARS-CoV-2-negative SARI controls | Vaccine effectiveness, % (95% CI)* | |
|---|---|---|---|
|
| |||
| Gamma predominant period mRNA vaccine** | 4 | 7 | 90.4 (66.4–97.6) |
| Oxford/AstraZeneca Vaxzevria | 84 | 35 | 52.2 (25.0–69.0) |
| Gamaleya Sputnik V | 25 | 8 | 42.6 (−39.7–73.8)† |
| Sinopharm (Beijing) Covilo | 16 | 6 | 47.0 (−52.4–79.0)† |
| Bharat Biotech Covaxin | 65 | 9 | -*** |
| Sinovac-Coronavac | 1 | - | - |
| Unvaccinated | 1388 | 234 | ref |
Adjusted by age group and presence of at least one comorbidity.
Includes Pfizer/BioNTech Comirnaty and Moderna/Spikevax vaccines.
95% CI widths greater than 50.
Estimates with confidence interval widths > 200 were not shown due to sparse data.
Fig. 3.
Estimates of COVID-19 vaccine effectiveness against hospitalization by vaccine type/brand and period of variant predominance, 1 May 2021–2 April 2022, Paraguay Caption. Adjusted by month of symptom onset, age, and presence of at least one comorbidity. VE estimates during the Gamma period are for partial (one dose) vaccination; estimates for the Delta and Omicron periods are for complete primary series vaccination. The estimate for the effectiveness of partial vaccination with Bharat Biotech Covaxin is not shown due to sparse data.
3.4. VE during the Delta variant-predominant period
From 22 August to 18 December 2021, the median time between receipt of last vaccine dose and onset of symptoms was 99 days (IQR: 63–129). The VE in patients who completed the primary series was 90.4% (95% CI: 74.3%–97.3%) for mRNA vaccines, 83.2% (95% CI: 67.8%–91.9%) for Oxford/AstraZeneca Vaxzevria, and 82.9% (95% CI: 53.0–95.2) for Gamaleya Sputnik V, adjusting for age group, month of illness onset, and presence of ≥ 1 pre-existing condition (Table 4, Fig. 3). Estimates for Sinopharm (Beijing) Covilo (VE: 29.7, 95% CI: −42.6–67.3) and Bharat Biotech Covaxin (VE: 24.9, 95% CI: −36.9–59.8) were lower with wide confidence intervals.
Table 4.
COVID-19 vaccine effectiveness against hospitalization by vaccine type and variant predominance period, 22 August 2021 to 2 April 2022, Paraguay (n = 1844).
| SARS-CoV-2-positive SARI cases | SARS-CoV-2-negative SARI controls | Vaccine effectiveness, % (95% CI)* | |
|---|---|---|---|
|
| |||
| Delta predominant period mRNA vaccine** | 4 | 56 | 90.4 (74.3–97.3) |
| Oxford/AstraZeneca Vaxzevria | 11 | 156 | 83.2 (67.8–91.9) |
| Gamaleya Sputnik V | 4 | 37 | 82.9 (53.0–95.2) |
| Sinopharm (Beijing) Covilo | 12 | 61 | 29.7 (−42.6–67.3)† |
| Bharat Biotech Covaxin | 20 | 112 | 24.9 (−36.9–59.8)† |
| Sinovac-CoronaVac | 1 | 0 | - |
| Unvaccinated | 128 | 266 | ref |
| Omicron predominant period mRNA vaccine** | 28 | 64 | 45.3 (5.8–68.7)† |
| Oxford/AstraZeneca Vaxzevria | 94 | 107 | 10.0 (−31.3–38.2)† |
| Gamaleya Sputnik V | 30 | 42 | 25.4 (−30.2–57.5)† |
| Sinopharm (Beijing) Covilo | 28 | 44 | 42.3 (−2.0–67.6)† |
| Bharat Biotech Covaxin | 41 | 44 | 13.0 (−46.1–48.0)† |
| Sinovac-CoronaVac | 0 | 3 | - |
| Booster dose (any vaccine)*** | 71 | 79 | 21.0 (−16.1–46.4)† |
| Unvaccinated | 216 | 235 | ref |
Adjusted effectiveness by month of symptom onset, age group, and presence of at least one comorbidity.
Includes Pfizer/BioNTech Comirnaty and Moderna/Spikevax vaccines
95% CI widths>50.
adjusted for age group and presence of ≥1 comorbidity only. Booster doses were with Pfizer/BioNTech Comirnaty (n = 46), Moderna Spikevax (n = 1), and Oxford/AstraZeneca Vaxzevria (n = 103).
3.5. VE during the Omicron variant-predominant period
From 2 January to 2 April 2022, the median time between receipt of last vaccine dose and onset of symptoms was 157 days (IQR: 84–210). Adjusting for age group, month of illness onset, and presence of ≥ 1 pre-existing condition, VE estimates for individual vaccines ranged from 10.0% (95% CI: −31.3–38.2) for Oxford/AstraZeneca Vaxzevria to 45.3% (95% CI: 5.8–68.7) for mRNA vaccines. Adjusting for age group and presence of ≥ 1 pre-existing condition, VE for receipt of a booster dose with an mRNA or Oxford/AstraZeneca Vaxzevria was 21.0% (95% CI:−16.1–46.4) (Table 4).
4. Discussion
We found that VE in preventing SARS-CoV-2-associated SARI varied by vaccine, the number of doses received, and variant-predominant period. Vaccination with mRNA vaccines consistently provided the greatest protection across all variant-predominant periods. Adenovirus viral vector vaccines (Oxford/AstraZeneca Vaxzevria and Gamaleya Sputnik V) also provided protection. Specifically, during the Gamma variant-predominant period, partial vaccination with Oxford/AstraZeneca Vaxzevria provided moderate (52.2%) protection, and during the Delta variant-predominant period, primary series vaccination with Oxford/AstraZeneca Vaxzevria and Gamaleya Sputnik V provided strong (>80%) protection from COVID-19-associated SARI. However, the effectiveness of all vaccines declined substantially during the Omicron variant-predominant period. We did not find evidence for the effectiveness of vaccination with either Sinopharm (Beijing) Covilo or Bharat Biotech Covaxin during any period.
Several factors may have contributed to the decline in VE from the Delta to the Omicron variant-predominant period. Immune evasion by the Omicron variant likely played a role. Recent studies have shown that VE is lower against infection, symptomatic disease, and hospitalization caused by the Omicron variant than by previous variants [24]. Waning immunity may have also contributed to declining VE because the time between vaccination and symptom onset increased from the Delta to the Omicron variant-predominant period. In addition, over time, underlying population immunity increases as immune-naïve people become infected in the community. Although we did not have information on prior infection, we would expect that a greater proportion of SARS-CoV-2-negative control-patients had immunity from prior infection during the Omicron predominant period than in the Delta predominant period, lowering VE estimates. It is also possible that a greater proportion of unvaccinated individuals had prior infection compared to vaccinated individuals during the Omicron variant-predominant period as those with prior infection may have been less likely to seek vaccination. This difference in vaccination rates would have biased odds ratios towards the null, resulting in lower VE estimates and could partially explain the lower VE observed during the Omicron predominant period. Finally, influenza circulated in Paraguay during the period of Omicron variant predominance, and previous studies have shown that the inclusion of influenza controls underestimates true COVID-19 VE [28,29]. However, the magnitude of bias tends to be low at the level of influenza circulation (positivity ≤ 25%) observed in Paraguay during the study period.
Our study has several limitations. First, our analysis used real-world sentinel surveillance data to evaluate the effectiveness of many vaccines administered as part of an emergency response. The high number of vaccine types used in Paraguay contributed to sample sizes that were insufficient to precisely estimate the effectiveness of individual vaccines in each period of variant predominance, particularly when VE estimates declined. For example, the confidence intervals of estimates during the Omicron variant-predominant period exceeded 50 and several lower bounds were negative, making point estimates difficult to interpret.
Second, although the surveillance hospitals aimed to test all SARI patients for SARS-CoV-2, this was not possible due to resource limitations, especially during periods of high COVID-19 incidence. SARS-CoV-2 test results were required for inclusion in the analysis, and systematic differences between those tested and those not tested could have introduced selection bias. For example, it is conceivable that clinicians might have prioritized SARS-CoV-2 testing in unvaccinated patients with a high suspicion of COVID-19, biasing VE estimates upwards.
Third, vaccinated persons often differ from unvaccinated persons in their risk of infection, independent of vaccination. Our test-negative design minimized bias due to healthcare seeking behavior but the risk of SARS-CoV-2 exposure (e.g., masking) might have differed between vaccinated and unvaccinated persons. Fourth, while all levels of Paraguay’s healthcare system can register vaccines in the EPI database, it is possible that some doses were not recorded, especially in rural areas with limited internet connectivity. Such omissions would have led to misclassification of truly vaccinated persons as unvaccinated, introducing non-differential exposure misclassification and an expected bias towards the null, thus reducing VE estimates. Finally, our study population included the catchment populations of four surveillance hospitals in two regions of Paraguay and may not be generalizable to populations with different sociodemographic characteristics or SARS-CoV-2 circulation patterns.
Notwithstanding these limitations, our study exemplifies the importance of leveraging existing surveillance platforms to estimate and systematically monitor VE during a pandemic in a middle-income country. Our findings are consistent with studies in other populations demonstrating that mRNA vaccines provide the greatest effectiveness against severe COVID-19 disease [30,31]. While VE has been estimated elsewhere, country-specific estimates are important because population characteristics can affect estimates; thus, estimates from one setting might not be transportable to others. Moreover, this study contributes to global understanding of vaccines that have not been widely used elsewhere. VE estimates from Paraguay can also be used to parameterize models to estimate the impact of the national COVID-19 vaccination campaign in Paraguay and similar settings.
Acknowledgments
Special thanks to the professionals who work in the participating sentinel centers (Hospital Central-IPS, Hospital Integrado IPS Ingavi, Hospital Regional de Ciudad Del Este, and Hospital Integrado IPS de Ciudad Del Este), as well as officials at DGVS, the Expanded Program on Immunization, the Central Public Health Laboratory (LCSP), the Regional Laboratory of Alto Paraná, and Cyrlab (private laboratory). We would also like to thank Eva Leidman, Laura Zambrano, Ashley Fowlkes, and Daniella Malave for their review and feedback on the manuscript. Special thanks to Angel Rodríguez, Juliana Leite and Daniel Salas (PAHO Regional Office) for their support and supervision.
Funding
This project is partially funded through a Collaborative Agreement between PAHO and the U.S. Centers for Disease Control and Prevention (Grant no. IP000940). The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
Footnotes
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
5. Data sharing
Surveillance data collected for the study are not publicly available but can be requested from Dr Guillermo Sequera, Director de la Dirección General de Vigilancia de la Salud, at the following address: mspdgvs@gmail.com.
Data availability
Data will be made available on request.
References
- [1].Polack FP, Thomas SJ, Kitchin N, Absalon J, Gurtman A, Lockhart S, et al. Safety and Efficacy of the BNT162b2 mRNA Covid-19 Vaccine. N Engl J Med 2020;383(27):2603–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Voysey M, Clemens SAC, Madhi SA, Weckx LY, Folegatti PM, Aley PK, et al. Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-CoV-2: an interim analysis of four randomised controlled trials in Brazil, South Africa, and the UK. Lancet 2021;397(10269):99–111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Logunov DY, Dolzhikova IV, Shcheblyakov DV, Tukhvatulin AI, Zubkova OV, Dzharullaeva AS, et al. Safety and efficacy of an rAd26 and rAd5 vector-based heterologous prime-boost COVID-19 vaccine: an interim analysis of a randomised controlled phase 3 trial in Russia. Lancet 2021;397(10275):671–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Tokars JI, Rolfes MA, Foppa IM, Reed C. An evaluation and update of methods for estimating the number of influenza cases averted by vaccination in the United States. Vaccine 2018;36(48):7331–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Pan American Health Organization. Network for the evaluation of vaccine effectiveness in Latin America and the Caribbean—influenza, (REVELAC-i). 2022. December 19]; Available from: https://www.paho.org/en/network-evaluation-vaccine-effectiveness-latin-america-and-caribbean-influenza-revelac-i.
- [6].Pan American Health Organization. Severe Acute Respiratory Infection Network. [cited 2022 December 25]; Available from: https://www.sarinet.org/.
- [7].Pan American Health Organization. COVID-19 Vaccination in the Americas. 2022. 12/22/2022 [cited 2022 25 December]; Available from: https://ais.paho.org/imm/IM_DosisAdmin-Vacunacion.asp.
- [8].Ioannidis JPA. Factors influencing estimated effectiveness of COVID-19 vaccines in non-randomised studies. BMJ Evid Based Med 2022;27(6):324–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Schwalb A, Armyra E, Méndez-Aranda M, Ugarte-Gil C. COVID-19 in Latin America and the Caribbean: Two years of the pandemic. J Intern Med 2022;292(3):409–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].European Centre for Disease Prevention and Control. COVID-19 Vaccine Tracker. 2022. December 19, 2022]; Available from: https://vaccinetracker.ecdc.europa.eu/public/extensions/COVID-19/vaccine-tracker.html#distribution-tab. [Google Scholar]
- [11].Centers for Disease Control and Prevention. COVID Data Tracker. 2022. [cited 2022 December 25.]; Available from: https://covid.cdc.gov/covid-data-tracker.
- [12].Castelli JM et al. Effectiveness of mRNA-1273, BNT162b2, and BBIBP-CorV vaccines against infection and mortality in children in Argentina, during predominance of delta and omicron covid-19 variants: test negative, case-control study. Bmj, 2022; 379: e073070. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].González S, et al. Effectiveness of the first component of Gam-COVID-Vac (Sputnik V) on reduction of SARS-CoV-2 confirmed infections, hospitalisations and mortality in patients aged 60–79: a retrospective cohort study in Argentina. EClinicalMedicine 2021;40:101126. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Costa Clemens SA, et al. Effectiveness of the Fiocruz recombinant ChadOx1-nCoV19 against variants of SARS-CoV-2 in the Municipality of Botucatu-SP. Front Public Health 2022;10:1016402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Arregocés-Castillo L, Fernández-Niño J, Rojas-Botero M, Palacios-Clavijo A, Galvis-Pedraza M, Rincón-Medrano L, et al. Effectiveness of COVID-19 vaccines in older adults in Colombia: a retrospective, population-based study of the ESPERANZA cohort. Lancet Healthy Longev 2022;3(4):e242–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Cerqueira-Silva T, Andrews JR, Boaventura VS, Ranzani OT, de Araújo Oliveira V, Paixão ES, et al. Effectiveness of CoronaVac, ChAdOx1 nCoV-19, BNT162b2, and Ad26.COV2.S among individuals with previous SARS-CoV-2 infection in Brazil: a test-negative, case-control study. Lancet Infect Dis 2022;22(6):791–801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Cerqueira-Silva T, de Araujo Oliveira V, Paixão ES, Florentino PTV, Penna GO, Pearce N, et al. Vaccination plus previous infection: protection during the omicron wave in Brazil. Lancet Infect Dis 2022;22(7):945–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].González S, et al. Effectiveness of BBIBP-CorV, BNT162b2 and mRNA-1273 vaccines against hospitalisations among children and adolescents during the Omicron outbreak in Argentina: A retrospective cohort study. Lancet Reg Health Am 2022;13:100316. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Más-Bermejo PI, et al. Cuban Abdala vaccine: Effectiveness in preventing severe disease and death from COVID-19 in Havana, Cuba; A cohort study. Lancet Reg Health Am 2022;16:100366. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Paternina-Caicedo A, et al. Effectiveness of CoronaVac and BNT162b2 COVID-19 mass vaccination in Colombia: A population-based cohort study. Lancet Reg Health Am 2022;12:100296. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Jara A, Undurraga EA, González C, Paredes F, Fontecilla T, Jara G, et al. Effectiveness of an Inactivated SARS-CoV-2 Vaccine in Chile. N Engl J Med 2021; 385(10):875–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Florentino PTV, Alves FJO, Cerqueira-Silva T, Oliveira VdA, Júnior JBS, Jantsch AG, et al. Vaccine effectiveness of CoronaVac against COVID-19 among children in Brazil during the Omicron period. Nat Commun 2022;13(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Hitchings MDT, Ranzani OT, Dorion M, D’Agostini TL, de Paula RC, de Paula OFP, et al. Effectiveness of ChAdOx1 vaccine in older adults during SARS-CoV-2 Gamma variant circulation in São Paulo. Nat Commun 2021;12(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Feikin DR, Abu-Raddad LJ, Andrews N, Davies M-A, Higdon MM, Orenstein WA, et al. Assessing vaccine effectiveness against severe COVID-19 disease caused by omicron variant. Report from a meeting of the World Health Organization. Vaccine 2022;40(26):3516–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Fitzner J, Qasmieh S, Mounts AW, Alexander B, Besselaar T, Briand S, et al. Revision of clinical case definitions: influenza-like illness and severe acute respiratory infection. Bull World Health Organ 2018;96(2):122–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].World Health Organization, Maintaining surveillance of influenza and monitoring SARS-CoV-2 – adapting Global Influenza surveillance and Response System (GISRS) and sentinel systems during the COVID-19 pandemic: Interim guidance. Geneva; 2020. [Google Scholar]
- [27].Greenland S, Mansournia MA, Altman DG. Sparse data bias: a problem hiding in plain sight. BMJ 2016;352:i1981. [DOI] [PubMed] [Google Scholar]
- [28].Doll MK et al. Effects of Confounding Bias in Coronavirus Disease 2019 (COVID-19) and Influenza Vaccine Effectiveness Test-Negative Designs Due to Correlated Influenza and COVID-19 Vaccination Behaviors. Clinical Infectious Diseases, 2022; 75(1): e564–e571. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Pan American Health Organization. FluNET Home Page. [cited August 18 2023]; Available from: https://ais.paho.org/phip/viz/ed_flu.asp.
- [30].Premikha M et al. Comparative Effectiveness of mRNA and Inactivated Whole-Virus Vaccines Against Coronavirus Disease 2019 Infection and Severe Disease in Singapore. Clin Infect Dis, 2022; 75(8): 1442–1445. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Rotshild V, Hirsh-Raccah B, Miskin I, Muszkat M, Matok I. Comparing the clinical efficacy of COVID-19 vaccines: a systematic review and network meta-analysis. Sci Rep 2021;11(1). [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Surveillance data collected for the study are not publicly available but can be requested from Dr Guillermo Sequera, Director de la Dirección General de Vigilancia de la Salud, at the following address: mspdgvs@gmail.com.
Data will be made available on request.