Prioritisation of population groups with the most interactions for COVID-19 vaccination can substantially reduce total fatalities

Abstract The rapid development of vaccines against the SARS-CoV-2 virus is an unprecedented achievement. Once vaccines become mass produced, they will have to be distributed to almost the entire population to prevent deaths and permit prompt economic recovery. The necessity to vaccinate a large number of people in a short period of time, and possibly with insufficient vaccine doses to cover most, creates in itself a new challenge for governments and health authorities: which population groups (by age or other criteria) should be targeted first and what sequence must be followed, if any at all, to achieve the minimum number of fatalities? In this work, we demonstrate the importance and impact of optimally planning the priorities for vaccine deployment by population groups using a modified SEIR-type model for the COVID-19 outbreak considering age-related groups. Finding the absolute guaranteed best solution of the mathematical optimisation problem may be hard, if even possible, and would likely require intense computational resources for every possible case study scenario. In this work, several strategies are evaluated and compared, in an attempt to approach the most effective possible vaccination priority sequence in an example case study using demographic and epidemiological data from Spain. The minimum total fatalities at the end of the vaccination campaign is the objective pursued. The population groups classifications are established based on relevant differences in mortality (due to their age) and risk-related behaviour such as their number of daily person-to-person interactions. Assuming a capacity limited constant vaccination rate, vaccination distribution strategies were evaluated for different vaccine effectiveness levels and different percentages of final vaccine population coverage. Our results unambiguously show how planning vaccination by priority groups can achieve dramatic reductions in total fatalities (more than 70% in some cases) compared to no prioritisation. The results also indicate in all cases, for all vaccine effectiveness and coverage values evaluated, that the criteria for groups vaccination priority should not be those with the highest mortality but rather those the highest number of daily person-to-person interactions. Strikingly, our results show in all cases, that prioritisation of groups with the highest mortality but less social interactions, may lead to significantly larger numbers of final total fatalities, even higher as if no group priorities were established at all. The explanation, clearly displayed by the mechanistic model, is that vaccination avoids infections that reduce mortality not only from the vaccinated group itself but also from the projected secondary and subsequent infections inflicted on the rest of the population by those vaccinated in that group. Precisely this amplification effect (exponential nature of the curve) appears to cause the larger reduction in total fatalities if the groups with the most interactions are vaccinated first. The possible contradiction of these results with some published recommendations highlight the importance of conducting an open comprehensive and rigorous analysis of this problem leaving behind any subjective preconceptions.