Spatial prediction of immunity gaps during a pandemic to inform decision making: A geostatistical case study of COVID-19 in Dominican Republic.

Background: To demonstrate the application and utility of geostatistical modelling to provide comprehensive high-resolution understanding of the population's protective immunity during a pandemic and identify pockets with sub-optimal protection.

Methods: Using data from a national cross-sectional household survey of 6620 individuals in the Dominican Republic (DR) from June to October 2021, we developed and applied geostatistical regression models to estimate and predict Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) spike (anti-S) antibodies (Ab) seroprevalence at high resolution (1 km) across heterogeneous areas.

Results: Spatial patterns in population immunity to SARS-CoV-2 varied across the DR. In urban areas, a one-unit increase in the number of primary healthcare units per population and 1% increase in the proportion of the population aged under 20 years were associated with higher odds ratios of being anti-S Ab positive of 1.38 (95% confidence interval [CI]: 1.35-1.39) and 1.35 (95% CI: 1.32-1.33), respectively. In rural areas, higher odds of anti-S Ab positivity, 1.45 (95% CI: 1.39-1.51), were observed with increasing temperature in the hottest month (per°C), and 1.51 (95% CI: 1.43-1.60) with increasing precipitation in the wettest month (per mm).

Conclusions: A geostatistical model that integrates contextually important socioeconomic and environmental factors can be used to create robust and reliable predictive maps of immune protection during a pandemic at high spatial resolution and will assist in the identification of highly vulnerable areas.