key: cord-0731264-fdn7sz5x authors: Cobo-Lewis, A. B. title: Equitable Vaccine Access within an Age-Based Framework date: 2021-03-20 journal: nan DOI: 10.1101/2021.03.18.21253915 sha: d3d15743189323e3a193b4f0638d8d813e1a6df6 doc_id: 731264 cord_uid: fdn7sz5x Several states have adopted age-based prioritization for COVID-19 vaccine eligibility (also prioritizing teachers and child care workers) because it is simple (especially when age is quantized by decade) and age is strongly associated with COVID-19 mortality. But this approach raises equity concerns based in law and ethics. Analysis of CDC data reveals that COVID-19 mortality in the U.S. rises 3-fold per decade of life. People with specific conditions (including certain disabilities and certain medical conditions) associated with a 3-fold or 6-fold risk ratio should therefore be granted vaccine eligibility along with people 10 or 20 years older, respectively. Along with additional recommendations on data collection and reporting, this could help address equity within an age-based framework. Main Text: Several U.S. states have moved to age-based prioritization for COVID-19 vaccine eligibility (modified to also prioritize teachers and child care workers, who President Biden has identified as a federal priority). There is substantial variability in how states consider disability (1) , but as of March 8, adults with high risk conditions are not prioritized in 12 states (2) . Agebased prioritization is simple, especially when states quantize age by decade, opening vaccines to 5 people at least in their 70s, then adding people in their 60s, etc. State health departments have justified age-based prioritization by older people's large risks of serious COVID-19 outcomes, including death, and by the need for simple, fast, and transparent systems (e.g., 3). But strictly age-based prioritization has come under assault as being unethical (e.g., 4). Furthermore, the U.S Department of Health and Human Services Office for Civil Rights has advised that, under the 10 Affordable Care Act's nondiscrimination provisions, a state or other entity is "only permitted to consider age as one factor as part of its overall decision-making." (5) How to reconcile the large age-associated effects with ethical and legal demands for equity? The answer is found in a proper quantification of the age-associated effects and a commitment to better data collection and reporting. 15 To properly quantify the age effect on COVID-19 mortality, publicly available data from the U.S. Centers for Disease Control and Prevention (CDC) and U.S. Census Bureau were analyzed (see Supplementary Materials). At the time, the CDC data included data on 507,227 deaths from COVID-19 deaths in the U.S. Fig. 1 plots the results, pooling across the entire population, 20 though data were also analyzed by racial-ethnic category. It is clear that overall COVID-19 mortality in the U.S. increases 3-fold per decade of life. Most other risk factors are associated with COVID-19 mortality less than 3-fold and might be excluded from prioritization with comparatively small effects on health and equity. But a handful of conditions and demographic characteristics are associated with elevation in COVID-19 25 mortality that approaches or even exceeds 3-fold. Fig. 2 plots the COVID-19 mortality risk ratios from the present analysis and for published data on Down syndrome, intellectual or developmental disability, and organ transplant. Early in the pandemic, Down syndrome was reported to be associated with a 10-fold elevation in COVID-19 mortality (6), though an analysis of two more recent datasets has found elevation of 2.9-and 3.5-30 fold, respectively (7). Intellectual or developmental disability (a broader characteristic than Down syndrome) has been reported to be associated with an elevation in COVID-19 mortality of 5.9-(8) to 7.8-fold (9). Organ transplantation has been reported to be associated with an elevation in COVID-19 mortality of 3.5-(10) to 6.5-fold (11). Elevations in COVID-19 mortality by race and ethnicity are not as large in the present analysis 35 as were found earlier in the pandemic (12) , but they are still present. In particular, the point estimate is 2.9-fold for Non-Hispanic Native American Indian and Alaska Natives (see Supplementary Materials for age-specific COVID-19 mortality by racial-ethnic category: the risk ratio rose as high as 6.1-fold for 25-34-year-old Non-Hispanic Native American Indian and Alaska Natives). A 3-fold elevation in COVID-19 mortality is the same risk conveyed by being 10 years older, and a 9-fold elevation in COVID-19 mortality is the same risk conveyed by being 20 years older (because 3 × 3 = 9). In order to begin to address equity, in any jurisdiction with an age-based framework for COVID-19 vaccine prioritization, people with conditions associated with a 3-fold elevation in COVID-19 mortality should become eligible for vaccines along with people 10 years older, and people with conditions associated with a 9-fold elevation in COVID-19 mortality should become eligible for vaccines along with people 20 year older. For example, a 5 40-year-old with Down syndrome or organ transplantation should become eligible for COVID-19 vaccines at the same time as 50-or 60-year-olds from the general population. Important for speeding vaccine delivery, Down syndrome is easy to identify. This recommendation continues to be relevant in the U.S. until all adults become vaccine-eligible on May 1. It will also remain relevant several months or more into the future for countries that continue to experience vaccine 10 shortage and that are considering an age-based eligibility framework. Down syndrome and organ transplantation are both found on the CDC's list of medical conditions with sufficient evidence to conclude they put people at increased risk of severe illness from COVID-19 (13) . Molecular and genetic investigation has begun to suggest mechanisms that 15 contribute to elevated COVID-19 risk in Down syndrome (14) , and the immune suppressants associated with organ transplantation make the risk obvious. On the other hand, the strong association of intellectual or developmental disabilities overall with COVID-19 mortality may be at least partially explained by congregate living-90% of Americans with intellectual or developmental disabilities live alone, with a roommate, or with family caregivers (15) , but 20 people living in congregate settings are overrepresented in the published analyses about COVID-19 in people with intellectual or developmental disabilities, and some states that do not prioritize people with intellectual or developmental disabilities per se may prioritize a sub-population indirectly by prioritizing people living in congregate settings. It is therefore important to obtain unbiased estimates of the COVID-19 mortality risk among people with intellectual or 25 developmental disabilities who live in less segregated and less congregate settings and to quantify the additional COVID-19 mortality risk that segregated and congregate settings may pose for this population. In addition, outside the U.S., data should be brought to bear to potentially modify the age-related 3-fold threshold according to local conditions of the pandemic. 30 In the U.S., state developmental disabilities agencies, for the most part, have COVID-19 data about people with disabilities-at least for people receiving services from state institutions or home-and community-based services funded through Medicaid-because COVID-19 cases and deaths constitute "critical incidents" that must be reported to the agency. But the data collection systems for people with intellectual developmental disabilities are largely segregated from the 35 general public health data systems. Consequently, while COVID-19 dashboards, which are common throughout the U.S., track and report data by several risk factors (age and race among them), they rarely if ever report data on disability, even though states could link their public health and vaccine databases with databases supporting their developmental disabilities agencies or Medicaid agencies. This makes it impossible to assess the COVID-19 mortality associated 40 with intellectual or developmental disability per se, and it also makes it impossible for the public to track vaccine progress in this marginalized population-important considerations even after vaccine eligibility is broadened to the general adult population. This must change. . CC-BY-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. . CC-BY-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. . CC-BY-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 March 20, 2021. 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 March 20, 2021. ; https://doi.org/10.1101/2021.03.18.21253915 doi: medRxiv preprint All data were downloaded from the CDC and the Census Bureau, where they are publicly available. Because the data are aggregated and de-identified, no IRB review was required. Census data was from Vintage 2019 population estimates (the most recent available at time of analysis in March 2021). To analyze the overall increase of COVID-19 mortality by age, COVID-19 data were downloaded from https://data.cdc.gov/resource/9bhg-hcku.json. To analyze associations involving race or ethnicity, additional COVID-19 data were downloaded from https://data.cdc.gov/resource/ks3g-spdg.json. Race and ethnicity groups were constructed to match those used in a previous study of racial and ethnic disparities (12): Hispanic, Non-Hispanic American Indian and Alaska Native, Non-Hispanic Asian or Pacific Islander (including Native Hawaiian), and Non-Hispanic Black. The three racial categories were for people reporting those races alone (but results were robust to including people of mixed race in those racial categories). Analysis was conducted in R 4.0.4. COVID-19 mortality was calculated for 9 age categories that spanned a decade each (except for the youngest [0-14 years] and oldest [85+ years]). The logarithm base 10 of COVID-19 mortality per 100k population was regressed on age as well as on age and racial-ethnic category. In the regressions, age was treated as a linear predictor from 1 for the lowest group to 9 for the oldest group. (Linear regression was used for simplicity, but results were robust to the use of general linear models.) Before log transform, COVID-19 mortality spanned four orders of magnitude. Although there is some suggestion in Fig. 1 that might suggest a slight quadratic trend of log mortality vs age, the linear trend alone provided an excellent fit (adjusted R 2 > 0.98). Fig. S1 presents the data for each racial-ethnic category in a separate panel, while also repeating the data and fit for the overall population in every panel. These data were analyzed with overall COVID-19 mortality as a reference category, and the difference in intercept between racial-ethnic category and reference was largest for Non-Hispanic American Indian and Alaska Native (coefficient = 0.4568 [equivalent to risk ratio of 10 0.4568 = 2.86], t = 4.240, df = 39, P = 0.0001), then Hispanic (coefficient = 0.2872, risk ratio = 1.94, t = 2.666, df = 39, P = .01), then Non-Hispanic Black (coefficient = 0.2207, risk ratio = 1.66, t = 2.048, df = 39, P = 0.047). The difference in intercept was not significant between Non-Hispanic Asian or Pacific Islander and overall population (coefficient = -0.1113, risk ratio = 0.77, t = -1.003, P = 0.31). COVID-19 mortality among racial and ethnic minorities also appeared especially elevated from levels in overall population in early adulthood through midlife, especially for Non-Hispanic American Indian and Alaska Native (note how the data in Fig. S1 appear to follow a stronger quadratic trend in three of the four racial-ethnic groups than they do in the general population). This was not quantified with additional regressions in order to simplify the analysis-and especially to simplify the presentation of results in order to clarify their policy implications. Instead, for each racial-ethnic category, the raw COVID-19 mortalities in age group were compared directly to the corresponding values in the overall population (a model-free approach). The maximum risk ratio by age was similar to the regression-derived risk ratios for three racialethnic categories, but for the Non-Hispanic American Indian and Alaska Native, the risk ratio was as large as 6.08 for age 25-34, compared to a risk ratio of 2.86 for that racial-ethnic group's regression-derived value. . CC-BY-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 March 20, 2021. ; https://doi.org/10.1101/2021.03.18.21253915 doi: medRxiv preprint Fig. 3 displays risk ratios deriving from the regressions, along with risk ratios from other studies for Down syndrome, intellectual or developmental disability, and organ transplant (except that, in one case (8), an odds ratio is displayed, but COVID-19 mortality is low enough that odds ratios and risk ratios can be treated similarly for those data). The full contents of the R are included in supplementary material (in an R file). In order for the code to run for the first time, a census API must be obtained as described at https://api.census.gov/data/key_signup.html and installed using the census_api_key() function in the tidycensus R package. For a low-bandwidth analysis such as this, no API key is required from the CDC. . CC-BY-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. . CC-BY-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 March 20, 2021. ; https://doi.org/10.1101/2021.03.18.21253915 doi: medRxiv preprint COVID-19 Vaccine Prioritization Dashboard How America's vaccine system makes people with health problems 5 fight for a place in line Indicators of Severe COVID-19 Illness Variation in racial/ethnic disparities in COVID-19 mortality by age in the United States: A cross-sectional study Network analysis of Down syndrome and SARS-CoV-2 identifies risk and protective factors for COVID-19 The State of the States in Intellectual 10 and Developmental Disabilities: United States Acknowledgments: Thanks to Angie Claussen for comments on a previous draft.