Population and study design

SARI surveillance was established in Albania in 2009, primarily focusing on influenza and other respiratory pathogens. However, since 2022, the SARI surveillance system has undergone significant extensions. It is now integrated into an electronic infectious disease surveillance system (SISI), allowing identification and comprehensive testing of SARI samples for influenza, SARS-CoV-2, and other relevant pathogens. This enhanced integration provides a more robust understanding of respiratory infections in the country.

In Albania, there are eight sentinel sites for SARI surveillance located across seven regions as well as two national hospitals in capital city are enrolled in this system. These locations were chosen to provide a representative geographical sample covering 72% of the country’s population.

We conducted a test-negative case control study among SARI patients over 18 years old who meet WHO case definition [8] and were hospitalized at one of the eight SARI sentinel sites from July 2022 to July 2023. This design involved analyzing data on vaccinated and unvaccinated patients based on the “test-negative design” (TND), where cases are SARI patients who tested positive for SARS-CoV-2, and controls are SARI patients, tested negative for SARS-CoV-2. Due to limited viral typing capacities, COVID-19 variants were not included in this study design. Variant-specific data were not available at the individual case level because genomic typing was not performed systematically for all SARS-CoV-2–positive SARI patients. Therefore, all laboratory-confirmed SARS-CoV-2 cases were analysed together as COVID-19–positive cases without stratification by variant or sub-variant. During the study period, the dominant circulating variant was omicron.

Data collection

The SARI surveillance study participants were swabbed and tested for COVID-19 using real-time reverse transcription–polymerase chain reaction (RT-PCR) tests. Respiratory specimens were taken from all SARI patients within 48 hours of hospital admission, and PCR testing for COVID-19 and influenza was carried out at the Institute of Public Health (IPH). SARI cases were identified across a range of hospital wards within sentinel sites, including the Intensive Unit Care (ICU)/critical care unit, general/internal medicine, pediatric medicine, infectious disease ward, respiratory disease ward, and the maternity ward within the regional hospital. Required information on study participants includes demographic information, clinical and treatment details, and vaccination status for both COVID-19 and influenza infections. All the data collected were entered into the SISI system, while COVID-19 and influenza vaccination details were gathered from the electronic immunization system. Data were retrospectively retrieved from both systems for research purposes on 01.01.2024. For this analysis, authors had access only to de-identified datasets, and no directly identifiable personal information was available to the authors during or after data extraction.

Inclusion and exclusion criteria

SARI patients aged ≥18 years with an available SARS-CoV-2 RT-PCR result and documented COVID-19 vaccination status were eligible for inclusion. As shown in Fig 1, of 3348 hospitalized SARI patients tested for SARS-CoV-2 between 1 July 2022 and 1 July 2023, 1490 were excluded: 1031 were aged <18 years; 13 did not meet the WHO SARI case definition; 35 had missing vaccine dose dates; 108 were swabbed >7 days after hospitalization; 54 were swabbed >10 days after symptom onset; 215 were influenza-positive [10]; and 34 had an unknown COVID-19 test result. The final analytic cohort therefore included 1858 eligible patients aged ≥18 years, comprising 410 SARS-CoV-2–positive cases and 1448 SARS-CoV-2–negative controls (Fig 1).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart for SARI patient inclusion in the VE study, SARI sentinel surveillance, 1^st July 2022- 1^st July 2023.

https://doi.org/10.1371/journal.pgph.0006418.g001

Definitions

Patients meeting the WHO case definition [11] for SARI and who were eligible according to the inclusion-exclusion criteria were enrolled in the study. The study participants were divided into two groups – cases and controls. The cases were defined as confirmed COVID-19 cases, identified as patients admitted to the hospital with SARI symptoms and having a positive SARS-CoV-2 PCR test result obtained within 48 hours of admission or documented within 14 days prior to admission. The control group comprised SARI patients testing negative for SARS-CoV-2, within 48 hours, and were not tested positively 14 days prior to admission.

Patients were considered fully vaccinated with a primary series if they had received two doses of a COVID-19 vaccine at least 14 days before SARI symptom onset. Those who had received both doses followed by a booster dose at least 14 days before symptom onset were considered fully vaccinated with a booster [12]. Unvaccinated patients included those who had not received any COVID-19 vaccine or had received only one dose at least 14 days before symptom onset.

Comorbidities were defined as the presence of one or more chronic conditions documented in hospital medical records at the time of admission, including cardiovascular disease, immunocompromised conditions, diabetes, liver disease, neurological conditions, chronic kidney disease, chronic lung disease, malignant neoplasms, asthma bronchialis, and obesity.

Individuals with a previous infection were defined as any eligible individual who tested positive for SARS-COV-2 by RT-PCR from January 2020 and up to 14 days before the current positive test.

Severe outcomes were defined as fatal outcome of patients or having one of the following requirements: the need for admission to the intensive care unit (ICU) for supplemental oxygen or mechanical ventilation.

Statistical analysis

We summarized the demographic and clinical characteristics of both cases and test-negative controls, as well as the vaccinated and unvaccinated SARI patients. Discrete variables were expressed as frequencies and proportions, while continuous variables were summarized using medians and interquartile ranges (IQR). Comparisons of baseline characteristics between groups were made using Fisher’s exact test for categorical variables and the Wilcoxon rank-sum test for continuous variables. Vaccine effectiveness was estimated for patients with a confirmed prior infection, severe outcomes, and across different age groups. For the age group analysis, patients were stratified into categories 18–59, 60–79, and ≥80 years to derive age-specific VE estimates. Each subgroup analysis was performed on a dedicated cohort restricted to the population of interest. Within each subgroup-specific cohort, the test-negative design was applied, designating SARS-CoV-2 positive patients as cases and negative patients as controls. The models were structured using multivariable logistic regression, where the SARS-CoV-2 test result served as the dependent variable and vaccination as the primary independent variable. Crude VE was calculated as VE = (1−OR) × 100%, with 95% confidence intervals (CI) provided for precision. Adjusted VE estimates were obtained by adjusting for age, sex, presence of at least one chronic condition (i.e., hypertension, heart disease, chronic respiratory disease, diabetes, liver disease, renal disease, immunodeficiency), and duration of hospital stay as a proxy for unmeasured confounding. All analyses were performed using R Statistical Software (v4.2.1; R Core Team 2021).

Ethics statement

Based on the Albanian Law on Communicable Diseases No. 15/2016, which permits studies related to communicable disease surveillance to be conducted using existing surveillance systems, the IPH determined that the activities described in the study protocol are limited to public health surveillance and meet the criteria for routine surveillance as defined in this law. The data used in this analysis are derived from national surveillance systems that are established and regulated by law. In accordance with this legal framework, obtaining informed consent from patients is not required for public health surveillance activities. For the present analysis, only anonymized surveillance data were used.

Vaccination and severe outcomes

Point estimates of VE were higher among fully vaccinated individuals particularly those with an additional booster, against severe outcomes among SARI patients. Despite the relatively low overall VE of 30.6% with just the primary course, this effectiveness increases to 37.8% with a booster dose, although confidence intervals were wide and overlapped zero Comparatively, other studies report higher VEs, but consistently show increased effectiveness with the administration of booster doses. For instance, a study in Spain using data from 2021 reported VE estimates between 70% and 80% for fully vaccinated individuals, while a UK study conducted between December 8, 2020, and October 1, 2021, found that VE against hospitalization and death remained around 80% at 20 weeks post-vaccination. Similarly, an Israeli study analyzing data from July 30 through August 31, 2021 demonstrated that a third Pfizer-BioNTech dose significantly reduced rates of severe illness, particularly in older adults [13–15]. In addition, our study period coincided with Omicron-dominant circulation in Europe, during which VE against hospitalization has generally been lower than in earlier variant periods‌‌ [7,8,16]. This aligns with comparator studies conducted during the Omicron wave between December 2021 and July 2022 Albania’s use of a heterogeneous vaccine product mix, including viral vector and inactivated vaccines, together with the low vaccination coverage observed in this study population—where more than half of hospitalized SARI patients were unvaccinated and only around one fifth had received a booster dose (Table 1)—may also have contributed to the modest VE estimates.

Impact of previous SARS-CoV-2 infection

Patients with a prior history of SARS-CoV-2 infection had higher estimated VEwhen a booster was added to their primary vaccination course, with VE reaching 76.2%. This finding is consistent with observed benefitsof combining natural and vaccine-induced immunity [17] and further supports the role of booster doses in improving estimated protection particularly among previously infected individuals [18]. These findings align with results from the US Flu VE Network, which reported enhanced VE against outpatient illness during the Omicron variant’s circulation using data up to August 8, 2021 [19]. Additionally, study from Israel analyzing data from August and September 2021 demonstrated that patients with hybrid immunity had reduced risks of reinfection and severe disease compared to those with only vaccination or natural infection, with both studies emphasizing that hybrid immunity provides broader and more durable protection, particularly against variants of concern [17].

From a policy perspective, these findings suggest that in settings with high levels of prior infection but low booster uptake, such as Albania [4], risk-based booster strategies targeting older adults and individuals with comorbidities may yield substantial public health benefit even when population-wide booster coverage remains limited. Consistent with this, WHO SAGE’s 2023 guidance, synthesizing evidence collected since January 2021, emphasizes focusing booster efforts on high-risk groups (e.g., older adults and those with comorbidities) given widespread hybrid immunity in populations [5].

Age-specific vaccine effectiveness

The age-specific analysis revealed variations in VE. Patients over 80 years demonstrated the highest VE with booster doses, indicating that booster doses, suggesting greater estimated protection in older populations, who are generally more vulnerable to severe disease. Booster doses have been associated with enhanced immune protection in elderly populations, thereby reducing the risks associated with severe outcomes as demonstrated in studies using data from February through September 2021 and earlier periods up to mid-2021 [20,21]. In contrast, point estimates in the 60–79 age group were lower and confidence intervals were wide, limiting definitive interpretation in this age stratum. For patients aged 18–59 years, the VE estimates were 35.5% for the primary course and 42.8% after booster vaccination, compared with 15.5% and 15.0%, respectively, in the 60–79 group. This age group generally displayed a higher VE compared to the 60–79 group, likely due to fewer comorbidities and a stronger immune response. Booster doses among adults aged 18–49 in the United States, studied during December 2021, were associated with a reduced risk of hospitalization [22]. Furthermore, data spanning August 26, 2021, to January 22, 2022, indicated that while VE tends to decline over time post-booster, it still indicates meaningful estimated protection among younger adults [23]. Although VE in the 18–59 and 60–79 age groups is lower than in the 80 + group, it remains consistent with protection, underscoring the different dynamics of immune responses across age groups and the role of booster doses in enhancing protection. Overall, our wide confidence intervals across age groups necessitate caution, but the point estimates align with the biological expectation that booster doses are associated with greater protection.

Limitations of the study

Despite these significant findings, our study has several limitations. First, as an observational study the test-negative design it is inherently susceptible to biases, including selection and information bias. Reliance on hospital-based SARI data may miss severe COVID-19 cases among patients who did not seek hospital care, introducing potential selection bias. Hospital length of stay was included in adjusted models to partially account for differences in clinical course between cases and controls, however this is an imperfect severity proxy and may be influenced by non-clinical factors (e.g., discharge/transfer practices, bed availability) and competing outcomes such as early death, so residual confounding may remain. Moreover, the exclusion of patients with incomplete vaccination records could influence the representativeness of our cohort. Vaccine product–specific analyses were not performed and therefore, VE could not be estimated by vaccine type or platform. The study did not include pediatric populations or patients below 18 years old, limiting the generalizability of the findings to younger age groups. Additionally, the study was limited by a lack of comprehensive genomic sequencing, which prevented stratification of vaccine effectiveness (VE) by SARS-CoV-2 variants or Omicron sub-variants. Omicron sub-variants differ in their degree of immune escape, with sub-variants such as BA.4 and BA.5 showing greater resistance to vaccine-induced antibodies [16]. Consequently, VE estimates in this study likely represent an average across multiple sub-variants, potentially attenuating overall effectiveness during periods dominated by more immune-evasive strains.