Mathematical models can help understand SARS-CoV-2 transmission and assess the efficacy of measures to mitigate viral spread. Our model highlights that the non-pharmaceutical interventions (both social distancing measures and increase in ICU bed capacity) were highly successful in the Republic of Cyprus – an island state of the European Union without land borders but a popular tourist destination – as they kept the number of COVID-19-related deaths at low levels and the need for ICU beds within the capacity of the local health-care system.
The IFR is a critical indicator for the evaluation of the consequences of the COVID-19 pandemic. A recent modeling study on different settings estimated an IFR ranging from 0.5 to 1.4% [24]. Compared to those estimations, Cyprus experienced a low IFR during the first pandemic wave (central estimation: 0.71%). Possible explanations for the country’s low IFR include a) timely public health interventions, since NPIs launched before the first death, b) opportunity to improve the preparedness of the country’s health-care system, since Cyprus was not among the first affected countries, and c) the geographic position of Cyprus, which is a peripheral country of the European Union and thus less vulnerable to the effect of external seeds.
According to our estimates, only a relatively small proportion of people in the Republic of Cyprus have been infected during the first epidemic wave (IAR: 0.31% (95% CrI: 0.15, 0.54%), which is among the lowest in Europe [3]. This means that, at the end of the first wave, Cyprus was far from achieving collective immunity (the collective immunity threshold for Cyprus was estimated at 1-(1/R0) =1-(1/2.66) =62.4%). Under the counterfactual scenario, which resembles a strategy without viral suppression based on herd immunity from natural infections, 86,600 individuals (95% CrI: 46,700, 122,100) would have been infected, which corresponds to a prevalence of 9.9% (95% CrI: 5.3, 13.9%), still well below the target of 62.4%. Therefore, collective immunity could only be achieved in Cyprus through a combination of vaccination and infection, although the duration of immunity following infection remains unknown [25].
Globally, COVID-19 true infections are more than the laboratory-confirmed cases [3, 22]. However, case ascertainment in the Republic of Cyprus during the first epidemic wave was relatively high (33.2%). This means that, during the first epidemic wave, each individual diagnosed with COVID-19 in Cyprus corresponded to about two undiagnosed cases. The corresponding ratios for Germany, Spain, Sweden, and Greece were 23.5, 9.8, 6.1, and 24%, respectively [3, 6, 26]. It is important that the efficacy of the testing strategy to control COVID-19 is strongly dependent on the efficiency of backward contact tracing and adherence of the infected individuals to isolation.
According to our model estimates, the implemented public health measures averted 715 (95% CrI: 339, 1235) deaths. Comparing the different interventions, social distancing measures were more important to flattening the epidemic peak and reducing the pressure on the health care system than the increase of the ICU bed capacity only. Consistent with previous work [6, 7], our results highlighted that any intervention to boost only ICU bed capacity would not have been an effective healthcare policy, as the demand for ICU beds would rapidly outrun availability. Figure 2B displays that after April 10th, 2020, the demand for ICU beds would have increased exponentially.
Several studies have underlined the value of the NPIs to limit the circulation of the virus during the first COVID-19 wave [2, 3, 27,28,29]. Despite the efficacy of NPIs, however, it is important to take into consideration their indirect negative effects, including reductions in economic activities, mental issues due to isolation, and difficulties in accessing health care for chronic and other diseases [30, 31]. In particular, the behavioral fatigue associated with adherence to COVID-19 restrictions (the pandemic fatigue) has been a vital issue that affects the response to the pandemic, since long-period strict lockdowns may face a reduction in their effectiveness [31]. Pandemic fatigue appeared in the Republic of Cyprus too, after the end of the first wave, and increased gradually in all the subsequent waves, although the NPIs were lifted for some time. Understanding pandemic fatigue is challenging since it has several causes [32, 33]. Future work that examines the factors that drive compliance and identify those that could be modifiable are needed. Furthermore, research that seeks to identify the optimal duration of NPIs policies will be useful. All these data could inform targeted, tailored, and effective policies, interventions, and communications.
Study implications
Our study contributes to the discussion regarding the effectiveness of NPIs during the first epidemic wave. Our results provide theoretical support that fast and accurate interventions minimized the first COVID-19 wave and prevented the overload of the healthcare system. These findings could guide both the confrontation of next waves of COVID-19 and policies during the interim period between epidemic waves when social distancing restrictions are lifted. Additionally, in periods when social distancing measures are lifted, improved surveillance and active contact tracing are very important since they are capable of timely detecting any COVID-19 outbreak very quickly. Finally, our results could be useful in designing policies in the case of an emergence of a novel disease without effective treatment or vaccines that could mimic SARS-CoV-2 transmission.
Comparison with other studies
Our results are consistent with previous modeling studies that examined the first COVID-19 wave in the Republic of Cyprus showing that NPIs managed to limit the burden of the first wave of COVID-19 [8, 9]. However, there are two main differences between these studies and the present work. The first is that our study used the trajectories of COVID-19-related deaths, hospitalized cases, and ICU bed use as calibration points, while the other two studies used the number of reported cases. The second difference lies in the nature of the models. More specifically, in our analysis, we have used a stochastic, individual-based mathematical model while the other two studies have used compartmental mathematical models. Overall, however, all three studies underline the significance of NPIs to minimize the burden of COVID-19.
Limitations
As with any modelling study, there are also limitations. First, the model ignores the impact of social networks in the population and assumes that it is randomly mixed. Second, an important assumption is that all deaths due to SARS-CoV-2 infection have been identified and reported (e.g., no deaths under the status quo scenario occurred before admission to the health-care facility). Third, we assumed that post-infection immunity completely protects again reinfection over the duration of simulations. Fourth, we assumed that all deaths occurred in the ICU. Notwithstanding, the effect of this assumption on our projections is likely to be marginal, since the simple-bed mortality is relatively low and no deaths outside the hospital system have been reported [5]. The above was supported also by analyses from Euromomo that showed that overall mortality in Cyprus remained within the expected range [34]. Fifth, due to the relatively low number of COVID-19 related deaths during the first epidemic wave, the model does not capture the initial numbers of deaths very precisely. Finally, in the counterfactual scenario, public health interventions were removed, while assuming that everything else remained exactly as in the status quo scenario, i.e., there would be no changes in the duration that a patient stays in an ICU bed or in hospital.