key: cord-1051209-hvsnrrka authors: Banholzer, N.; van Weenen, E.; Lison, A.; Cenedese, A.; Seeliger, A.; Kratzwald, B.; Tschernutter, D.; Salles, J. P.; Bottrighi, P.; Lehtinen, S.; Feuerriegel, S.; Vach, W. title: Estimating the effects of non-pharmaceutical interventions on the number of new infections with COVID-19 during the first epidemic wave date: 2021-01-20 journal: nan DOI: 10.1101/2021.01.15.21249884 sha: 497b3d9d5385fea577d08ee212db338ffb302a97 doc_id: 1051209 cord_uid: hvsnrrka The novel coronavirus (SARS-CoV-2) has rapidly developed into a global epidemic. To control its spread, countries have implemented non pharmaceutical interventions (NPIs), such as school closures, gathering bans, or even stay-at-home orders. Here we study the effectiveness of seven NPIs in reducing the number of new infections, which was inferred from the reported cases of COVID-19 using a semi-mechanistic Bayesian hierarchical model. Based on data from the first epidemic wave of n=20 countries (i.e., the United States, Canada, Australia, the EU-15 countries, Norway, and Switzerland), we estimate the relative reduction in the number of new infections attributed to each NPI. Among the NPIs considered, event bans were most effective, followed by venue and school closures, whereas stay-at-home orders and work bans were least effective. With this retrospective cross-country analysis, we provide estimates regarding the effectiveness of different NPIs during the first epidemic wave. The implementation date of an NPI refers to the first day a measure went into action. The 48 implementation dates were thoroughly checked to ensure correctness and consistency across countries. 49 Overall, eight authors were involved in collecting, categorizing, and checking the data. Furthermore, 50 local residents and/or native speakers were recruited from some countries in order to verify our 51 encoding in cases where the interpretation of legal terms was not ambiguous or where it was difficult 52 to distinguish, for instance, whether an NPI was enforced or recommended. Sources and details on 53 data collection for NPIs are provided in Supplement 6. Cancellation of mass gatherings (i.e., 50 people or more) School closure Closure of schools (for primary schools) Venue closure Closure of venues for recreational activities and/or shops, bars, and restaurants Border closure Closure of national borders for individuals Gathering ban Prohibition of small gatherings in public or private spaces of people from different households Stay-at-home order Prohibition of movement without valid reason (e.g., restricting mobility except to/from work, local supermarkets, and pharmacies) Work ban Closure of non-essential business activities (i.e., all businesses except supermarkets, food suppliers, and pharmacies), thus prohibiting corresponding mobility Table 1 . List of non-pharmaceutical interventions (NPIs). Our selection of Western countries defined a sample that followed a similar and comparable 55 overall strategy in controlling the COVID-19 outbreak. On the one hand, the national strategies 56 consisted of similar NPIs, which we can expect to work in a similar manner, despite cultural and 57 organisational differences between countries. On the other hand, the national strategies differed and sequencing of NPIs across countries after considering regional variation within countries. For most countries, there was no regional variation and the NPIs were implemented at one day across the entire country. 66 In this section, we provide a short description of our model. A detailed description, including all 67 modeling and prior choices is given in Supplement 1. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 20, 2021. infections I jt in country j at day t. In the absence of any measure, this would be µ I jt = C jt δ j , where C jt are the number of contagious subjects and δ j is the country-specific daily transmission rate. In the presence of NPIs, we multiply this with the reduction due to avoided infections, resulting in where a mjt denotes the fraction of avoided infections due to NPI m = 1, . . . , M in country j at day t. In case NPI m is implemented and fully effective, a mjt is set equal to the value θ m . However, an NPI may not be fully effective in a country due to regional differences or because it may take a few days until subjects respond to the new measures. Hence, the general structure of a mjt is . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 20, 2021. ; https://doi.org/10.1101/2021.01.15.21249884 doi: medRxiv preprint where p rj is region's r = 1, . . . , R j proportion of the total population in country j, where T mrjt denotes the number of days since the implementation of NPI m in region r of country j at time 77 t (such that T mrjt = 1 denotes the first day at which a reduction in the number of new infections 78 could be expected), and where f (T mrjt ) is a time-delayed response function, which is specified such 79 that the response to an NPI increases from zero on day T mrjt = 0 to one on day T mrjt = 3 (Fig. 3a) , 80 reflecting our expectation that NPIs typically require a few days until they are fully effective. The 81 choice for the time-delayed response is varied as part of the sensitivity analysis. The effect of an NPI when fully implemented is equal to θ m . Within our Bayesian framework, distribution. This distribution is assumed to be known and our choice is based on an estimate by a 105 recent study using data on the exposure for both the index and secondary case 19 . . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 20, 2021. ; We collected data from publicly available data sources (Johns Hopkins Coronavirus Resource Center 16 126 for epidemiological data; news reports and government resources for policy measures). All the 127 public health information that we used is documented in the main text, the extended data, and 128 supplementary tables. A preprocessed data file together with reproducible code is available from 129 https://github.com/nbanho/npi_effectiveness_first_wave. Using data from the first epidemic wave, we estimated the relative reduction in the number of 133 new infections for each NPI (Fig. 4a ). Event bans were associated with the highest reduction in 134 the number of new infections (37 %; 95% CrI 21 % to 50 %). The reduction was lower for venue 135 closures (18 %; 95% CrI −4 % to 40 %) and school closures (17 %; 95% CrI −2 % to 36 %), followed 136 by border closures (10 %; 95% CrI −2 % to 21 %) and gathering bans (9 %; 95% CrI −4 % to 23 %). 137 stay-at-home orders (4 %; 95% CrI −6 % to 17 %) and work bans on non-essential business activities (1 %; 95% CrI −8 % to 12 %) appeared to be the least effective among the NPIs considered in this 139 analysis. The estimates for the individual effects suggest a particular strong effect of event bans. This result 141 is further supported by analyzing the posterior ranking of the effects (Fig. 4b) , which indicates that 142 we could be at least 98 % sure that event bans were among the two most effective NPIs. Conversely, 143 we could be at least 76 % sure that work bans were among the two least effective NPIs. All NPIs together lead to an estimated relative reduction in the number of new infections by 145 67 % (95% CrI 64 % to 71 %). The combined effectiveness of NPIs was also analyzed (Fig. 4c-d) . 146 Thereby, we could be at least 97 % sure that at least five NPIs simultaneously lead to a reduction in is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 20, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 20, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 20, 2021. ; https://doi.org/10.1101/2021.01.15.21249884 doi: medRxiv preprint to further investigation whether the closure of primary schools is less effective than the closure of 205 secondary schools and universities 28 . A small effect was estimated for stay-at-home orders. This seems to contradict findings about 207 the high effectiveness of the lockdown from Flaxman et al. 12 . However, one should consider that 208 their definition of a lockdown encompasses multiple NPIs that we differentiated (e.g., gathering bans, 209 venue closures, and stay-at-home orders). Taken together, our estimated "lockdown effect" would 210 therefore also be large. Finally, although our findings suggest that work bans were not effective, 211 it should be considered that our definition of a work ban referred to a strict ban of non-essential 212 business activities, while many countries only issued recommendations. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 20, 2021. ; https://doi.org/10.1101/2021.01.15.21249884 doi: medRxiv preprint NPIs explicitly into account. We approached this by incorporating the share of the country's total 235 population that is affected by active NPIs in our model. The other studies ignored this variation 236 or restricted the analysis to countries with no or very little regional variation. A further specific 237 property was to allow for a gradual increase in the response to NPIs over the first few days, whereas 238 the other studies assumed a full response on the first day NPIs were implemented. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 20, 2021. ; For instance, it is possible that additional measures or an increasing general awareness encouraged social distancing and hence lead to less infections. If this is the case, such effects will erroneously be 264 assigned to the NPIs and possibly overstate their overall impact. 265 Third, it is challenging to distinguish between the effects of single NPIs due to their concurring 266 introduction in many countries (Supplement Tbl. 4 ). This is reflected by wide credible intervals 267 and a negative association between effects (Supplement Fig. 3.2) , suggesting that the effect of one 268 NPI may be attributed partially to another. Note that the effect of event bans could be estimated . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted January 20, 2021. a first level of maturity. This may be sufficient to justify the use of these models in analyzing the 292 effects of NPIs. In principle, these models could also be used to study the effect of lifting NPIs. A natural next step would be to apply these models to data from the second wave, which is still 294 ongoing. An interesting question would then be whether we can expect similar effects in the first and 295 the second wave. It is likely that some effects may have changed, as the situations are not necessarily 296 comparable and experience from the first wave may have helped in dealing with the second wave. Of course, there is still room for improvement and refinement of the models. If information 298 on new cases is also available at a regional level, regional variation in implementing NPIs can be 299 used as an additional source of information using a two-level hierarchical approach. Furthermore, 300 between-country and regional variation may be linked to certain characteristics which may help to 301 understand the conditions under which an NPI is most effective. Similarly, information on cases 302 stratified by patient characteristics may provide new insights. Also the modeling of the outcome can 303 be improved, e.g., by taking weekday effects into account. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 20, 2021. an observational study. The Lancet Public Health. 2020 5;5(5):e279-e288. Available from: https://linkinghub.elsevier.com/retrieve/pii/S2468266720300906. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted January 20, 2021. ; https://doi.org/10.1101/2021.01.15.21249884 doi: medRxiv preprint World Health Organization. Coronavirus diseases (COVID-19) advice for the public COVID-19 outbreak in Lombardy The timing of COVID-19 377 transmission. medRxiv. 2020 9 Prior Choice Recommendations Acknowledgements We thank various people around the world for checking our data on non-