Africa's COVID-19 pandemic, while initially feared, had a milder impact than anticipated, likely due to factors such as a younger population and inadequate testing capabilities. High seroprevalence in several African nations, including Ethiopia, highlights the need for effective diagnostic tools. While qRT-PCR is the gold standard, its cost and turnaround time limit its widespread use. Rapid antigen tests (Ag-RDTs) offer a faster, cheaper alternative but have varying sensitivities and specificities. This study aimed to evaluate the diagnostic accuracy of the Panbio Ag-RDT in comparison to qRT-PCR in Ethiopia, a setting where qRT-PCR testing capacity might be limited and where a rapid test could improve the timeliness of detection and reduce resource use. The WHO recommends Ag-RDTs with at least 80% sensitivity and 97% specificity for symptomatic individuals and high-risk asymptomatic individuals. The study sought to determine the Ag-RDT’s performance in a real-world setting in Ethiopia.

Several studies have explored the performance of Ag-RDTs for SARS-CoV-2 detection, with varying results. Some studies report high sensitivity and specificity, particularly in individuals with high viral loads during early infection, whereas others indicate lower sensitivity, especially in asymptomatic populations. The effectiveness of Ag-RDTs varies depending on factors such as the viral load of the infected individual, the timing of the test relative to symptom onset, and the specific Ag-RDT used. Many studies compared the performance of Ag-RDTs to qRT-PCR, considered the gold standard, to assess their diagnostic accuracy. Studies in Germany and Italy have reported similar findings regarding low sensitivity and high specificity of Ag-RDTs, while studies in other settings have shown contrasting results in sensitivity and specificity. The studies' findings demonstrate the need for careful consideration of the limitations of Ag-RDTs and their appropriate application. Ethiopia’s previous studies on SARS-CoV-2 testing methods highlight the challenges of widespread testing using qRT-PCR, hence the motivation to explore more readily accessible testing methodologies. This sets the stage for a much-needed study to gauge the performance of a specific Ag-RDT under the conditions encountered in the Ethiopian healthcare system during a major outbreak period.

This cross-sectional study involved 1721 patients from six referral hospitals in Oromia Region, Ethiopia, between August 9-23, 2021. Participants underwent both qRT-PCR and Ag-RDT testing using nasopharyngeal swabs. The Panbio™ COVID-19 Ag rapid test was used for Ag-RDT. Data on demographics, clinical characteristics, and test results were collected using electronic data collection tools. The study included suspected COVID-19 cases, inpatients, pre-operative patients, and outpatients. The sample size was determined considering various assumptions, aiming for a large enough sample size to provide sufficient statistical power for the comparisons between qRT-PCR and Ag-RDT results. Data collectors and supervisors were trained extensively on data collection tools, sample collection and handling, and testing procedures. The qRT-PCR tests used the Abbot Real-Time SARS CoV-2 Assay. Data analysis included calculation of sensitivity, specificity, positive and negative predictive values, overall agreement, and Cohen’s kappa. The analysis also included investigation of whether there was an association between clinical characteristics of patients and the accuracy of the tests. Receiver Operating Characteristic (ROC) curve analysis was performed to assess the diagnostic performance of Ag-RDT. Statistical analyses were conducted using Stata 14.0. Ethical approval was obtained from relevant ethical review boards, and informed consent was obtained from all participants.

A total of 1364 participants were included in the final analysis. The Ag-RDT had a sensitivity of 33.3% (95% CI: 30.9%-35.9%) and a specificity of 99.3% (95% CI: 98.8%-99.7%) compared to qRT-PCR. The area under the ROC curve was 0.67 (95% CI: 0.63-0.69). Sensitivity improved in patients with shortness of breath (73.3%), and those with fever, cough, and dyspnea (71.4%). Specificity remained high (>98%) across all scenarios. Ag-RDT positivity showed a strong correlation with viral load (Ct value): 51.7% positive when Ct < 25 (high viral load), 3.4% when Ct 25-30, and 0% when Ct > 30. Overall agreement between Ag-RDT and qRT-PCR was 87.6%, with a Cohen's kappa of 0.66. qRT-PCR detected 91% of Ag-RDT positive cases, while Ag-RDT only detected 33.3% of qRT-PCR positive cases. Positivity rates for both tests were higher among patients with recent contact with confirmed/suspected cases and those with COVID-19 symptoms.

The study's findings indicate that while the Panbio Ag-RDT exhibits high specificity, its low sensitivity limits its use as a primary diagnostic tool for SARS-CoV-2. The lower sensitivity compared to the manufacturer's claims (98.1%) might be due to several factors, including variations in test performance, differences in population characteristics, and the presence of a high proportion of asymptomatic individuals. Improved sensitivity among symptomatic patients is consistent with other studies indicating better performance of Ag-RDTs in individuals with high viral loads. The lower sensitivity in the current study when compared to other studies could be due to differences in population characteristics, differences in the time of testing relative to symptom onset, and variations in the prevalence of infection. The observed low sensitivity is a concern, as relying solely on Ag-RDT for diagnostic and treatment decisions could lead to misdiagnosis and contribute to the spread of infection. The high specificity is valuable, however, especially in low-prevalence settings where false positives are a significant concern. This highlights the importance of using qRT-PCR as a confirmatory test for individuals with suspected COVID-19, particularly those at high risk of severe disease.

While the Panbio Ag-RDT shows promise as a point-of-care screening tool due to its high specificity and ease of use, its low sensitivity limits its use as a primary diagnostic tool. It is most useful in settings with a high prevalence of infection and should be used only for symptomatic patients with a low risk of severe disease. qRT-PCR should remain the primary diagnostic test, especially for high-risk individuals. Future studies could explore the use of Ag-RDTs as part of a tiered testing strategy, combined with other diagnostic approaches or clinical decision rules, to improve detection rates and efficient resource allocation.

The study's limitations include the lack of information on the onset of symptoms when samples were collected, potentially affecting the interpretation of sensitivity across different infection phases. The study was conducted during a specific period and in a particular region of Ethiopia, which might limit the generalizability of the findings to other settings or time points. The exclusion of a subset of samples due to poor quality or labeling errors might have introduced some bias. Despite the large sample size, further investigation is warranted to confirm these findings and potentially refine the use of Ag-RDTs in different clinical scenarios.