tetano
Editor, Senior Moderator
Am J Epidemiol
. 2026 Apr 29:kwag094.
doi: 10.1093/aje/kwag094. Online ahead of print.
Validity of influenza vaccine effectiveness calculations using a test negative design in Thailand
Erica E Zeno[SUP] 1 2 [/SUP], Prabda Praphasiri[SUP] 3 4 [/SUP], Wichan Bhunyakitikorn[SUP] 5 [/SUP], Sutthichai Nakphook[SUP] 4 6 [/SUP], Darunee Ditsungnoen[SUP] 3 [/SUP], Wuttiphong Phakdeekul[SUP] 4 [/SUP], Passakorn Ongarj[SUP] 4 [/SUP], Ratchadaporn Ungchcharoen[SUP] 4 [/SUP], Warinmad Kedthongma[SUP] 4 [/SUP], Seesai Yeesoonsang[SUP] 5 [/SUP], Kanlaya Sornwong[SUP] 6 [/SUP], Chaninan Sonthichai[SUP] 3 [/SUP], Anna N Chard[SUP] 1 [/SUP], William W Davis[SUP] 1 3 [/SUP], Martha P Montgomery[SUP] 1 3 [/SUP], Kriengkrai Prasert[SUP] 4 6 [/SUP]
Affiliations
Thailand uses a test-negative design to calculate influenza vaccine effectiveness (VE) annually for the Southern Hemisphere formulation and shares estimates with the World Health Organization to inform global vaccine strain selection. This study analyzed potential misclassification and confounding factors to evaluate validity of the test-negative design in Thailand. We retrospectively reviewed influenza sentinel surveillance data from June 2023-May 2024 and gathered information on vaccination history and health-seeking behavior. Cases were medically attended illnesses with positive influenza test results; non-cases had negative results for influenza and SARS-CoV-2. We compared VE using the routine approach with six scenarios, representing four potential threats to the validity of test-negative design studies. We assessed exposure (vaccination) misclassification, outcome (influenza) misclassification, and confounding by health-seeking behavior. Unmeasured confounding was evaluated by calculating VE against rhinovirus, an outcome unrelated to influenza vaccination. A total of 4709 patients were included, of which 328 (7.0%) were vaccinated against influenza. Under the routine approach, VE against medically attended influenza was 36.9% (95% confidence interval: 13.5%, 53.9%). In alternative scenarios, VE ranged from 32.4% to 42.8%, with overlapping 95% confidence intervals. These findings support the validity of Thailand's VE estimates and can inform improvements in surveillance procedures.
Keywords: influenza; test-negative design; vaccine effectiveness.
. 2026 Apr 29:kwag094.
doi: 10.1093/aje/kwag094. Online ahead of print.
Validity of influenza vaccine effectiveness calculations using a test negative design in Thailand
Erica E Zeno[SUP] 1 2 [/SUP], Prabda Praphasiri[SUP] 3 4 [/SUP], Wichan Bhunyakitikorn[SUP] 5 [/SUP], Sutthichai Nakphook[SUP] 4 6 [/SUP], Darunee Ditsungnoen[SUP] 3 [/SUP], Wuttiphong Phakdeekul[SUP] 4 [/SUP], Passakorn Ongarj[SUP] 4 [/SUP], Ratchadaporn Ungchcharoen[SUP] 4 [/SUP], Warinmad Kedthongma[SUP] 4 [/SUP], Seesai Yeesoonsang[SUP] 5 [/SUP], Kanlaya Sornwong[SUP] 6 [/SUP], Chaninan Sonthichai[SUP] 3 [/SUP], Anna N Chard[SUP] 1 [/SUP], William W Davis[SUP] 1 3 [/SUP], Martha P Montgomery[SUP] 1 3 [/SUP], Kriengkrai Prasert[SUP] 4 6 [/SUP]
Affiliations
- PMID: 42053054
- DOI: 10.1093/aje/kwag094
Thailand uses a test-negative design to calculate influenza vaccine effectiveness (VE) annually for the Southern Hemisphere formulation and shares estimates with the World Health Organization to inform global vaccine strain selection. This study analyzed potential misclassification and confounding factors to evaluate validity of the test-negative design in Thailand. We retrospectively reviewed influenza sentinel surveillance data from June 2023-May 2024 and gathered information on vaccination history and health-seeking behavior. Cases were medically attended illnesses with positive influenza test results; non-cases had negative results for influenza and SARS-CoV-2. We compared VE using the routine approach with six scenarios, representing four potential threats to the validity of test-negative design studies. We assessed exposure (vaccination) misclassification, outcome (influenza) misclassification, and confounding by health-seeking behavior. Unmeasured confounding was evaluated by calculating VE against rhinovirus, an outcome unrelated to influenza vaccination. A total of 4709 patients were included, of which 328 (7.0%) were vaccinated against influenza. Under the routine approach, VE against medically attended influenza was 36.9% (95% confidence interval: 13.5%, 53.9%). In alternative scenarios, VE ranged from 32.4% to 42.8%, with overlapping 95% confidence intervals. These findings support the validity of Thailand's VE estimates and can inform improvements in surveillance procedures.
Keywords: influenza; test-negative design; vaccine effectiveness.