Additional considerations for assessing COVID-19 impact on dengue transmission – Authors' reply

Christina Yek and colleagues raise two additional considerations when interpreting our recent findings that COVID-19 interventions reduced dengue incidence in 2020.¹ First, whether administrative delays might be an additional, unconsidered dimension to under-reporting and, second, whether the inclusion of abnormal data from 2019 might bias our predictions of cases averted.

Disruption-induced administrative delays in reporting are plausible and would have led to fewer dengue cases being reported in 2020. To minimise this, we restricted our analysis to January–December, 2020, despite more recent data being available from 2021. Searches for data were last updated on Feb 2, 2022, and no delay-related changes were identified compared with the original searches from Feb 23, 2021. If administrative delays did occur in 2020, they were probably quickly rectified before early 2021. Furthermore, our case fatality-based under-reporting analysis would probably have detected under-reporting due to administrative delays if they had occurred. Many countries with dengue endemics (eg, Sri Lanka) have separate reporting procedures for suspected dengue deaths that involve distinct, rapid reporting channels that are regularly audited.² Delays in reporting dengue cases but not deaths would result in higher case fatality rates, which we did not detect for any country.

We also agree that 2019 was an abnormally high incidence year for dengue and this would have resulted in below average incidence in 2020, similar to previous post-outbreak years (eg, 2017 in Brazil), even in the absence of COVID-19 interventions. These post-outbreak reductions are probably due to a combination of viral (eg, genotype replacement, as suggested), mosquito (eg, successful vector control), and host (eg, rising immunity to circulating viruses) factors that might differ between outbreaks but have a consistent effect of suppression.3, 4 The annual anomaly term in our model estimates this expected post-outbreak reduction. Although 2019 was an unprecedented year for dengue globally, many countries have had similar outbreaks previously (see the appendix [p 31] in our Article¹), allowing annual anomaly effects to be appropriately estimated. Inclusion of this term decreases predicted cases in 2020 and, thus, cases averted by COVID-19 interventions. Therefore, removing 2019 dengue data from the historical model-fitting dataset, as suggested, marginally increases our estimate of dengue cases averted by COVID-19 interventions but also substantially increases prediction uncertainty (0·76 million [95% credible interval 0·00–2·23] vs 0·72 million [0·12–1·47]). We therefore believe the original estimates we presented¹ offer the best overall estimate of the protective effects of COVID-19 interventions against dengue.

We declare no competing interests.