key: cord-0282868-18d4xc53
authors: Steinert, J. I.; Sternberg, H.; Veltri, G. A.; Buethe, T.
title: How Should COVID-19 Vaccines be Distributed between the Global North and South? A Discrete Choice Experiment in Six European Countries
date: 2022-05-19
journal: nan
DOI: 10.1101/2022.05.19.22275055
sha: 5aab2f95aa49c4877ac9de9ea5efd7972870274f
doc_id: 282868
cord_uid: 18d4xc53

Background: The global distribution of COVID-19 vaccinations remains highly unequal. We examine public preferences in six European countries regarding the allocation of COVID-19 vaccines between the Global South and Global North. Methods: We conducted online discrete choice experiments with adult participants in France (n=766), Germany (n=1964), Italy (n=767), Poland (n=670), Spain (n=925), and Sweden (n=938). Respondents were asked to decide which one of two candidates, who varied along four attributes: age, mortality risk, employment, and living in a low- or high-income country, should receive the vaccine first. We analysed the relevance of each attribute in allocation decisions using a conditional logit regression. Results: Across countries, respondents selected candidates with a high mortality and infection risk, irrespective of whether the candidate lived in their own country. All else equal, respondents in Italy, France, Spain, and Sweden gave priority to a candidate from a low-income country, whereas German respondents were significantly more likely to choose the candidate from their own country. Female, younger, and more educated respondents were more favourable of an equitable vaccine distribution. Conclusions: Given these preferences for global solidarity, European governments should promote vaccine transfers to poorer world regions.

In his opening speech to address the United Nations General Assembly in September 2021, Secretary General António Guterres expressed stark discontent with the highly unequal global In contrast, in a discrete choice experiment conducted in Germany and the US, participants in 127 both countries expressed a strong preference for prioritising vaccine allocation to countries 128 with a higher number of COVID-19 deaths and fewer intensive care unit beds, even when they 129 were asked to imagine that they or a vulnerable family member were still waiting for the 130 COVID-19 vaccine. (Klumpp et al., 2021) Notably, no previous study to date has exclusively 131 sampled participants who were still waiting for their first COVID-19 vaccine dose when 132 participating in the survey experiment.

In this paper, we analyse new experimental evidence from six EU countries on citizens' 135 preferences for the distribution of COVID-19 vaccines between the Global South and North. 136 We advance the literature in three ways. First, by covering six countries, we implement the 137 largest survey experiment on international vaccine allocation preferences to date and are thus 138 able to examine differences in citizens' preferences across EU member states. Second, we 139 conduct a discrete choice experiment among participants who are themselves not yet 140 vaccinated, asking them to allocate a COVID-19 vaccine to either a person in their own country 141 or to a person in a country in the Global South. This places specific salience on the notion that 142 donating a vaccine dose to a person in the Global South might mean sacrificing one's own dose 143 or that of a fellow citizen, thus leveraging self-interest-based and nationalistic considerations.

. 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 May 19, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 Third, we specifically examine heterogeneity in participants' preferences along key 145 sociodemographic characteristics. Policymakers can use these insights to anticipate which 146 population groups will be most and least supportive of COVID-19 vaccine donations.

Sample characteristics 149 150 A total of 6,030 eligible participants across all six countries completed the DCE. We excluded 151 participants who did not reply to all of the eight choice tasks in the DCE. To that end, attrition 152 across choice tasks ranged from 0% in Germany and Sweden, roughly 10% in France, Poland 153 and Spain, to almost 15% in Italy. Table 1 presents socioeconomic characteristics of 154 participants in each country. The German sample shows higher proportions of (i) older 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 May 19, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 

. 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 May 19, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 For the attribute of COVID-19 mortality risk, we observe a similar pattern in all surveyed 183 countries: The odds of being chosen to receive the vaccine were between two and almost six 184 times higher for a candidate with an increased COVID-19 mortality risk, relative to a candidate 185 with an average risk. The effect was even more pronounced for a candidate with a strongly 186 increased mortality risk (ranging from OR: 4.35, 95% CI: 3. 187 95% CI: 11.75-17.17 in Sweden; all p-values<0.001).

For employment status and age, we also observed largely similar patterns across countries: 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) Figure 1 illustrates the results of the heterogeneity analysis that was pooled across countries 197 (see Tables S2-3 for exact coefficients by subgroup and interaction terms). The odds of 198 choosing the candidate from the Global South rather than the candidate from the respondents' 199 own country were significantly higher for female (OR of interaction: 1.22, 95% CI: 1.10-1.36, 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) 

. 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 May 19, 2022. ; https://doi.org/10.1101/2022.05.19.22275055 doi: medRxiv preprint Figure S3 and Tables S4-9 summarise the extent to which these patterns prevail in each country.

The heterogeneities we identify were strongest in the German, Spanish, and Swedish samples.

In the French and Italian sample, these patterns were prevalent, too, but not statistically 225 significant. Interestingly, in terms of age, we found that older participants were relatively more 226 supportive of distributing the vaccine to a candidate from the Global South in both countries.

In the Italian sample, this effect was even statistically significant (OR of interaction: 1.34, 95% 228 CI: 1.00-1.78, p-value=0.049). In the Polish survey, heterogeneity patterns were less clear, and . 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 May 19, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 In our DCE conducted online in six EU countries, we found widespread global solidarity and 241 support for a more equitable distribution of COVID-19 vaccines between the Global North and 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 May 19, 2022. ; https://doi.org/10.1101/2022.05.19.22275055 doi: medRxiv preprint At the same time, our findings were inconclusive for Polish respondents and diverged for the 264 German DCE: German survey participants were significantly less willing to allocate the 265 COVID-19 vaccine to a person living in the Global South than to a person living in their own 266 country, thus revealing preferences consistent with vaccine nationalism. 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 May 19, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 evidence of a substantially higher reported COVID-19 threat perception among German 288 respondents (see Table S10 and Figure S4 ). Italy, France, Spain, Poland, and Sweden, the vaccination rate in those countries exceeded 50% 297 (30%) for first (second) doses whereas in Germany it was only at 25% (8%) (see Table S9 ). (3) migration background. Third, our analysis points to a number of predictors of variation. 308 However, they are not susceptible to experimental manipulation and should therefore not be 309 interpreted as causal. Fourth, at the country level (except in Germany), we were not able to 310 achieve our initial target sample sizes due to budget constraints. Thus, we likely only have 311 adequate statistical power in the pooled analysis. Lastly and relatedly, sample sizes varied 312 . 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 May 19, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 across countries and statistical power was higher in the German survey than in the other five 313 surveys. However, since we find statistically significant effects of the main attributes in all 314 countries except from Poland, lack of statistical power might have been less of a concern. 315 316 Policymakers and global health scholars have condemned the unequitable distribution of 317 COVID-19 vaccines between high-and low-income countries as "vaccine 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 May 19, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 Study Sample 336 We conducted an online survey experiment in six EU countries: France, Germany, Italy, 337 Poland, Spain, and Sweden. The German survey was launched at the height of the third wave 338 in April 2021; the other five surveys were carried out in June 2021, coinciding with a phase of 339 low case numbers in each country (see Figure S1 ). In each country, we recruited respondents 340 aged 18 years and older, drawing on online panels of the survey provider Bilendi-Respondi. 341 We sampled participants based on quotas that were matched to the census population of each 342 target country in terms of (1) gender, (2) age, (3) education, and (4) geographic location (e.g.,

state or province within each country) (see Table S1 for the census statistics of the sampled 344 countries). Participants were given an individual link to the survey, where they first received 345 information about the study's purpose, data protection regulations, and voluntary participation.

After completing the survey, participants received a voucher worth three to five Euros, which 347 was distributed by the survey company. 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 May 19, 2022. ; https://doi.org/10.1101 https://doi.org/10. /2022 We used the DCE methodology to elicit participants' preferences on the allocation of a 361 COVID-19 vaccine dose to a person in the Global North or to a person in the Global South. 362 We presented eight different choice sets and asked respondents to choose whether Person A or 363 Person B should receive the COVID-19 vaccine first. Respondents were told that the other 364 candidate of each pair would have to wait substantially longer to receive their first vaccine 365 dose. In each of the eight choice sets, one candidate was described as living in the country of 366 residence of the respondenti.e. a high-income country with high healthcare system capacity, 367 and the other candidate was described as living in a low-income country with a low healthcare 368 system capacity. The healthcare system capacity was explicitly mentioned in the two candidate 369 profiles. Across choice sets, candidates' characteristics varied along three additional attributes, 370 namely (1) age (20; 40; 60; 80 years), (2) individual COVID-19 mortality risk due to 371 comorbidity and/or lifestyle (no increased risk; increased risk; strongly increased risk), and (3) 372 employment status (not employed; employed and guaranteed income; employed and income 373 losses due to COVID-19 restrictions; employed in essential services).

The specific combination of candidate profiles in the eight choice sets, i.e., the experimental 376 design (presented in Table 4 ) was selected for statistical efficiency (referred to as "D- 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) 

Parochial altruism in humans

Vaccine Equity -Finding a Path Forward

We need people's WHO to solve vaccine inequity, and we need it now