“Tonight, @FLOTUS and I tested positive for COVID‐19,” President Trump tweeted on October 2, 2020 (https://twitter.com/realDonaldTrump/status/1311892190680014849). Although that might be a mere coincidence, Donald Trump won in all six states (Alabama, Arizona, Florida, Georgia, Louisiana, and Mississippi) with the highest coronavirus disease 2019 (COVID‐19) incidence (>3000 cases/0.1‐million population at the end of September in 2020 [https://coronavirus.jhu.edu/us-map]) in the presidential election 2016 (https://www.nytimes.com/elections/2016/results/president). Demographic and socioeconomic characteristics are known to play an important role in COVID‐19 epidemics.^1-5 Political orientation (e.g., Republican vs. Democrat or conservative vs. liberal) and voting patterns may be related to a variety of health outcomes including mortality.⁶ Association of political orientation with COVID‐19 transmission, however, has been never analyzed to date. In the present article, state‐level relation of votes in presidential election (the most representative political orientation) to COVID‐19 incidence was investigated.

For each US state, the cumulative number of confirmed COVID‐19 cases until September 30, 2020 was abstracted from “Johns Hopkins Coronavirus Resource Center” (https://github.com/CSSEGISandData/COVID-19/blob/master/csse_covid_19_data/csse_covid_19_daily_reports_us/09-30-2020.csv). Votes in the presidential election 2016 were available on “The New York Times” (https://www.nytimes.com/elections/2016/results/president). Demographic and socioeconomic characteristics were procurable on “American Community Survey” (https://www.census.gov/acs/www/data/data-tables-and-tools/data-profiles/). The COVID‐19 incidence rate was calculated as the cumulative number of cases divided by the population. The "vote ratio" was defined as the votes for Hillary Clinton divided by those for Trump (>1 denotes that Clinton won in the state). The univariable and multivariable random‐effects inverse‐variance (of the COVID‐19 incidence) weighted regression (i.e., meta‐regression where each state was likened to a study in meta‐analysis) was performed using OpenMetaAnalyst (http://www.cebm.brown.edu/openmeta/index.html). The vote ratio, state winner (Clinton or Trump), and demographic or socioeconomic characteristics were entered into the regression as covariates for the logarithmic‐transformed COVID‐19 incidence rate.

Extracted data were summarized in Table S1. The vote ratio in District of Columbia (DC) was 22.23 despite 0.32–2.08 in the other states. Accordingly, we decided to exclude DC (occupying merely 0.21% of the total population and cumulative COVID‐19 cases in the entire US) as an outlier in the analysis of vote ratios. Univariable regression demonstrated that the vote ratio (for Clinton vs. Trump) was significantly and negatively associated with the COVID‐19 incidence rate (coefficient: –0.366; p = .028; Table 1; Figure 1), which would indicate that the COVID‐19 incidence decreases as the vote for Clinton increases (i.e., the COVID‐19 incidence increases as the vote for Trump increases). The state winner (Clinton) was also significantly and negatively related to the COVID‐19 incidence rate (–0.405; p = .004; Table 1). Furthermore, multivariable regression demonstrated independent, significant, and negative association of the vote ratio (–0.999; p < .001; Table 1; multivariable‐I) and state winner (–0.303; p = .044; Table 1; multivariable‐II) (as well as under 18 years, Black/African American, Hispanic/Latino, bachelor's degree or higher, civilian unemployment, and median/mean household income) with the COVID‐19 incidence rate.

The present study investigating relation of political orientation to COVID‐19 epidemics for the first time suggests that the support to Clinton in the presidential election 2016 may be independently and negatively associated with the COVID‐19 transmission, that is, the support to Trump may be independently and positively related to the COVID‐19 epidemic. These findings should be cautiously interpreted. Because of not patient‐level analysis, they never denote that Trump's supporters are at high‐risk of COVID‐19 infection. Due to community‐level investigation, they simply mean that COVID‐19 incidence is higher in states where votes for Trump were greater. A health‐behavior study using the Annenberg National Health Communication Survey (US national survey from 2005 to 2012) data suggests that Republicans and conservatives trend to not be vaccinated against influenza.⁶ Trump's recent unscientific and irresponsible claims, for example, “It's going to disappear. One day—it's like a miracle—it will disappear” on February 27 at a White House meeting, “You don't have to do it (the use of nonmedical cloth face covering)” on April 3 at the White House, etc. (https://www.nytimes.com/2020/10/02/us/politics/donald-trump-masks.html) may also potentially increase the risk of COVID‐19 transmission among his voters. The public‐opinion survey by the Republican pollster Neil Newhouse indicated that “Republican voters were not taking the kinds of precautions that other voters were taking with respect to protecting themselves from the spread of the virus” and “The more we downplay the seriousness of the virus, the more that Republican voters may not take it seriously” (http://transcripts.cnn.com/TRANSCRIPTS/2004/08/cnr.08.html). Further analysis of association of political orientation with COVID‐19 epidemics should be required.

In conclusion, the support to Clinton in the presidential election 2016 may be independently and negatively related to the COVID‐19 transmission, that is, the support to Trump may be independently and positively associated with the COVID‐19 epidemic.

“今晚，@ FLOTUS和我对COVID-19进行了阳性测试，”特朗普总统在2020年10月2日发推文（https://twitter.com/realDonaldTrump/status/1311892190680014849）。尽管这可能仅仅是巧合，但唐纳德·特朗普在2019年冠状病毒疾病（COVID-19）发病率最高的六个州（阿拉巴马州，亚利桑那州，佛罗里达州，乔治亚州，路易斯安那州和密西西比州）获胜（> 3000例/十万人口于2020年9月底[https://coronavirus.jhu.edu/us-map]）在2016年总统选举中（https://www.nytimes.com/elections/2016/results/president）。人口和社会经济特征在COVID-19流行中起着重要作用。^1-5政治倾向（例如，共和党与民主党，保守派与自由派）和投票方式可能与包括死亡率在内的各种健康结果有关。⁶迄今为止，从未对政治倾向与COVID-19传播的关联进行过分析。在本文中，调查了总统选举中投票的州级投票（最具代表性的政治方向）与COVID-19发生率的关系。

对于每个美国州，从“约翰霍普金斯冠状病毒资源中心”（https://github.com/CSSEGISandData/COVID-19/blob/master/csse_covid_19_data/ csse_covid_19_daily_reports_us / 09-30-2020.csv）。可在“纽约时报”（https://www.nytimes.com/elections/2016/results/president）上获得2016年总统选举的投票。人口和社会经济特征可通过“美国社区调查”获得（https://www.census.gov/acs/www/data/data-tables-and-tools/data-profiles/）。COVID-19的发生率是根据病例的累计数除以人口得出的。“投票比率”定义为希拉里·克林顿的票数除以特朗普的票数（> 1表示克林顿在该州获胜）。使用OpenMetaAnalyst（http://www.covid-19.com）进行单变量和多变量随机变量反方差（COVID-19发病率）加权回归（即，将每种状态都比作荟萃分析的研究的元回归）。 cebm.brown.edu/openmeta/index.html）。投票率，州优胜者（克林顿或特朗普）以及人口或社会经济特征作为对数转换后COVID-19发生率的协变量输入了回归。

提取的数据总结在表S1中。哥伦比亚特区（DC）的投票率为22.23，尽管其他州为0.32-2.08。因此，我们决定在投票率分析中排除DC（仅占总人口的0.21％，在整个美国仅占累积的COVID-19病例）。单变量回归表明，投票率（克林顿对特朗普）与COVID-19发生率显着负相关（系数：–0.366；p  = .028；表1；图1），这表明COVID ‐19发生率随着对克林顿的投票增加而减少（即，COVID‐19发生率随着对特朗普的投票增加而增加）。州优胜者（Clinton）与COVID-19发生率也呈显着负相关（–0.405; p = .004; 表格1）。此外，多变量回归显示投票率（–0.999; p  <.001;表1; multivariable-I）和州赢家（–0.303; p  = .044;表1; multivariable-II ）的独立，显着和负相关）（以及18岁以下的黑人/非裔美国人，西班牙裔/拉丁美洲人，学士学位或更高学位，平民失业率以及中位数/平均家庭收入）与COVID-19的发生率有关。

本研究首次调查了政治取向与COVID-19流行病的关系，表明2016年总统大选对克林顿的支持可能与COVID-19传播独立且负相关，即对特朗普的支持可能是与COVID-19流行病成正相关。这些发现应谨慎解释。由于没有进行患者水平的分析，他们从不表示特朗普的支持者处于COVID-19感染的高风险中。由于社区级别的调查，他们只是意味着在特朗普投票人数较多的州中，COVID-19的发病率更高。⁶特朗普最近的不科学和不负责任的主张，例如，“它将消失。有一天，这就像一个奇迹，它将消失”，于2月27日在白宫举行的会议上，“没有必要（使用非医用布面罩）”于4月3日在白宫举行，等等。 （https://www.nytimes.com/2020/10/02/us/politics/donald-trump-masks.html）也可能增加选民中传播COVID-19的风险。共和党民意调查员尼尔·纽豪斯（Neil Newhouse）的公开意见调查显示：“共和党选民没有采取其他选民在保护自己免受病毒传播方面采取的各种预防措施”，并且“越是不重视共和党选民的严重性病毒，共和党选民可能就不会认真对待它”（http：//transcripts.cnn。com / TRANSCRIPTS / 2004/08 / cnr.08.html）。应进一步分析政治倾向与COVID-19流行病的关联。

总之，在2016年总统选举中对克林顿的支持可能与COVID-19传播独立且负相关，也就是说，对特朗普的支持可能与COVID-19流行独立且正相关。