The importance of citizenship for deserving COVID-19 treatment

The study investigates whether the boundary of citizenship shapes perceptions of who deserves life-saving healthcare during the COVID-19 pandemic in Germany. Building on debates about welfare chauvinism, the authors ask if similar dynamics extend to healthcare, specifically triage decisions for ICU treatment. The context is the scarcity of intensive respiratory care during COVID-19, which forced difficult allocation decisions. The paper compares the salience of citizenship to utilitarian considerations (e.g., survival chances, age, social contribution) commonly used in triage guidelines, using realistic, pandemic-based scenarios rather than abstract moral dilemmas. The study aims to assess the depth and societal salience of the citizenship boundary and its implications for social integration and resource allocation debates in Western Europe.

The paper situates its inquiry within research on welfare chauvinism and discrimination against immigrants across domains such as labor, housing, politics, and interactions with bureaucracies. Prior studies indicate public willingness to restrict resources to non-citizens and to discriminate against immigrants in access to COVID-19 treatment in Denmark and Switzerland. The authors contrast ascriptive boundaries (citizenship, national origin) with utilitarian criteria prioritized in medical ethics and German ICU guidelines (e.g., survival probabilities). The literature also highlights broad public support for healthcare access and the rise of post-national rights in the EU, which might reduce the salience of formal citizenship. Given that healthcare is a universal need and typically judged less on personal responsibility than benefits like unemployment insurance, the authors consider healthcare a least likely case for detecting discrimination. They thus formulate competing hypotheses: H1 that non-citizen immigrants are deprioritized relative to citizens for COVID-19 treatment, and H2 that citizens and non-citizens are treated equally.

Design: A preregistered, paired, between-subjects conjoint experiment with forced choice was conducted. Each respondent evaluated one pair of hypothetical patients and selected who should receive priority ICU admission and ventilation.

Attributes: Seven randomized attributes described each patient: (1) Chance of survival in ICU (20%, 50%, 80%); (2) Age (30, 44, 64 years); (3) Occupation (medical doctor, professor, nurse, cook), varying in social status and system relevance; (4) Children (has school-aged children vs has no children); (5) Criminal record (has vs has no criminal record); (6) Gender (male, female); (7) Citizenship/migration status (German citizen vs migrant with residence permit). In later waves they also varied EU vs non-EU non-citizen immigrants for exploratory analyses.

Instructions: Respondents were informed that listed survival chances already accounted for age and gender.

Sample and fieldwork: 23 online survey waves fielded by Respondi between mid-April 2020 and March 2021. Waves 1–13 (mid-April to mid-August 2020) had ~500 respondents each; waves 14–23 (early November 2020 to March 2021) had N=700 or N=1000. The sample was nationally representative by gender, age, and education; about 20% had a migration background. The study covered the first peak (April 2020), the calmer summer of 2020, the second major wave (fall/winter 2020/21), and the start of the third wave (March 2021).

Case context: Germany resembles other Western European countries—non-citizen immigrants face discrimination but can access robust welfare and healthcare services; citizenship laws have liberalized, making the citizen/non-citizen boundary comparable across Western Europe.

Randomization details: In waves 1–14, paired profiles could not share the same category on any attribute to avoid identical profiles. In waves 15–23, 40% followed the same constraint; for 60%, attributes were fully randomized. Analyses indicate programming differences did not affect findings.

Analysis: Estimated average marginal component effects (AMCEs) and average component interaction effects (ACIEs) via OLS-equivalent models; standard errors clustered by survey wave; two-tailed tests; no multiple-comparison corrections; analyses on unweighted data with listwise deletion of missing values.

Ethics and preregistration: Approved by the Ethics Commission of the WZB (study 2020-0-88); GDPR compliant; informed consent obtained; preregistered on April 21, 2020 (OSF link provided in paper).

• Main effect of citizenship: Non-citizen immigrants are about 10 percentage points less likely than citizens to be prioritized for COVID-19 treatment. This penalty is sizeable and robust, and larger than effects for occupation, age, or gender.
• Utilitarian attributes dominate but citizenship remains salient:
  • Survival chance: Relative to 80% (reference), 50% survival reduces selection probability by ~15 pp; 20% survival reduces it by ~26 pp.
  • Family responsibilities: Having no school-aged children reduces selection probability by ~21 pp compared to having school-aged children.
  • Criminality: Having a criminal record reduces selection probability by ~18 pp compared to no record.
  • Occupation: System relevance matters (nurses prioritized over professors), but differences are smaller than the citizenship gap.
  • Age: Older patients (e.g., 64 years) are less likely to be prioritized than younger ones, but these differences are smaller than the citizenship effect.
• Temporal stability: Across 23 waves over 12 months, non-citizen immigrants are consistently seen as less deserving than citizens. The penalty is smaller in waves 14–23 than in waves 1–13 (statistically significant), but not clearly tied to pandemic severity; EU vs non-EU manipulation does not explain the change.
• Heterogeneity across respondents:
  • National attachment: The penalty increases with respondents’ felt closeness to Germans; only those at the extreme low end (feeling distant from Germans) show no bias.
  • Immigration attitudes: Pro-immigration respondents show a slight bias in favor of non-citizen patients; as criticism of immigration increases, bias shifts against non-citizens. Across most of the scales, the non-citizen penalty is statistically significant.
  • Citizenship of respondents: Respondents without German citizenship do not exhibit a penalty against non-citizen immigrant patients.
• Patient-immigrant subtypes: No difference in prioritization between EU and non-EU non-citizen immigrants in this design.

The findings directly address the research question, demonstrating a meaningful and persistent citizenship boundary in public judgments about deservingness for life-saving COVID-19 treatment. Even when utilitarian criteria such as survival probability, family responsibilities, absence of criminal record, and system-relevant occupations strongly shape triage preferences, citizenship independently and substantially affects choices. The citizenship penalty exceeds differences associated with age and occupational status, indicating that national membership acts as a salient social boundary in Germany. These results suggest broad healthcare chauvinism among German residents, with notable variation aligned with national attachment and immigration attitudes. The consistency across diverse pandemic phases underscores the depth of the boundary, even as its magnitude varies over time. The absence of differences between EU and non-EU non-citizens implies that formal citizenship per se, rather than region of origin within/outside the EU, structures these judgments. The results have implications for social cohesion, integration debates, and the ethics of resource allocation in Western Europe, revealing that ascriptive membership can compete with or complement utilitarian principles in public opinion about triage.

The study provides robust evidence that German respondents prioritize citizens over non-citizen immigrants for access to life-saving COVID-19 treatment, extending welfare chauvinism to healthcare contexts. The citizenship penalty is substantial relative to other patient characteristics and persists across time and respondent subgroups, though it is absent among non-German-citizen respondents. These findings highlight the depth of the citizenship boundary in public judgments and have implications for integration and policy debates on resource allocation. Future research could examine more granular out-group categories (e.g., religion, specific national origins) to test whether certain subgroups elicit stronger biases, explore dynamic relationships between threat/anxiety and healthcare chauvinism, investigate professional decision-makers’ judgments, and assess generalizability across countries and crises.

• The experiment reflects public opinion in hypothetical scenarios; real-world clinical decision-making may differ.
• Analyses use unweighted data with listwise deletion; no adjustment for multiple comparisons across conjoint attributes.
• Only one paired choice per respondent may limit within-respondent reliability; attribute randomization constraints varied across waves, though checks suggest no effect on results.
• Generalizability is limited to Germany, though authors argue similarities with Western Europe.
• The design does not distinguish immigrant-origin citizens from native-origin citizens, potentially attenuating estimated biases.
• Temporal variation in the non-citizen penalty is observed but its causes (e.g., anxiety, pandemic severity) are not identified within this design.