This week the New York Times published a simple infographic showing how many members of Congress have tested positive for COVID-19. According to their reporting there have been 44 Republicans and 17 Democrats who have tested positive (as of January 14, 2021).
They further break it down by Senate and House. In the Senate, there have been 8 out of 51 Republicans and 2 out of 48 Democrats. In the House, it is 36/213 to 15/226.
The article did not provide statistics to compare the two groups, so I figured I would do my own!
Methods
Let’s create this dataset. I used epiR to do the analysis since it automates a lot of the epi process.
library(epiR)## Loading required package: survival## Package epiR 1.0-15 is loaded## Type help(epi.about) for summary information## Type browseVignettes(package = 'epiR') to learn how to use epiR for applied epidemiological analyses## covid <- data.frame(
chamber = rep(c("Senate", "House"), c(51+48, 213+226)),
party = rep(c("Republican", "Democrat", "Republican", "Democrat"), c(51, 48, 213, 226)),
case = rep(rep(c("positive","negative"), 4), c(8, 51-8, 2, 48-2, 36, 213-36, 15, 226-15))
)
covid$case <- factor(covid$case, levels = c("positive", "negative"))
covid$party <- factor(covid$party, levels = c("Republican", "Democrat"))
ftable(covid)## case positive negative
## chamber party
## House Republican 36 177
## Democrat 15 211
## Senate Republican 8 43
## Democrat 2 46With such low numbers of cases in the Senate, it doesn’t make sense to stratify by chamber. I estimated the incidence of coronavirus infection in each cohort and estimated the incidence risk ratio of being Republican vs. Democrat.
I compared the incidence in Congress to the incidence in the general population, currently estimated at 7% (23.6 million divided by 328 million) using a one sample test of proportions.
Results
Incidence and Risk Ratio
table(covid$party, covid$case)##
## positive negative
## Republican 44 220
## Democrat 17 257epi.2by2(table(covid$party, covid$case), method = "cohort.count")## Outcome + Outcome - Total Inc risk * Odds
## Exposed + 44 220 264 16.7 0.2000
## Exposed - 17 257 274 6.2 0.0661
## Total 61 477 538 11.3 0.1279
##
## Point estimates and 95% CIs:
## -------------------------------------------------------------------
## Inc risk ratio 2.69 (1.58, 4.58)
## Odds ratio 3.02 (1.68, 5.44)
## Attrib risk * 10.46 (5.14, 15.79)
## Attrib risk in population * 5.13 (1.22, 9.05)
## Attrib fraction in exposed (%) 62.77 (36.53, 78.17)
## Attrib fraction in population (%) 45.28 (19.43, 62.83)
## -------------------------------------------------------------------
## Test that OR = 1: chi2(1) = 14.640 Pr>chi2 = <0.001
## Wald confidence limits
## CI: confidence interval
## * Outcomes per 100 population unitsAmazingly 16.7% of Republican legislators have had COVID-19. Even 6.6% of Democrats have had it.
Based on the calculation, the incidence risk ratio is 2.69 (95% confidence ratio 1.58-4.58). That is, Republican party affiliation was associated with a 2.69 higher incidence of COVID-19 in Congress compared to being a Democrat.
Test of Proportions
The incidence in the Congress was 61/538 or 11.3%.
p_pop = 23.6/328
prop.test(61, 538, p_pop)##
## 1-sample proportions test with continuity correction
##
## data: 61 out of 538, null probability p_pop
## X-squared = 13.217, df = 1, p-value = 0.0002774
## alternative hypothesis: true p is not equal to 0.07195122
## 95 percent confidence interval:
## 0.08845881 0.14397743
## sample estimates:
## p
## 0.1133829The null hypothesis is that the sample incidence is the same as the population. The two sided probability of finding an incidence of 11.3% with a population incidence of 7.2% is <0.001.
We can also compare the Democrat and Republican rates to the general population.
prop.test(44, 264, p_pop) # Republican test##
## 1-sample proportions test with continuity correction
##
## data: 44 out of 264, null probability p_pop
## X-squared = 34.064, df = 1, p-value = 5.334e-09
## alternative hypothesis: true p is not equal to 0.07195122
## 95 percent confidence interval:
## 0.1248935 0.2183990
## sample estimates:
## p
## 0.1666667prop.test(17, 274, p_pop) # Democrat test##
## 1-sample proportions test with continuity correction
##
## data: 17 out of 274, null probability p_pop
## X-squared = 0.26807, df = 1, p-value = 0.6046
## alternative hypothesis: true p is not equal to 0.07195122
## 95 percent confidence interval:
## 0.0376738 0.0992757
## sample estimates:
## p
## 0.0620438Based on this we reject the null hypothesis that the Republican legislators have the same proportion of infections as the general population. We cannot reject the hypothesis that the Democratic legislation is the same as the general population.
Discussion
Because this is the entire population of US Congress legislators, the confidence ratio is not really helpful. We can say with full certainty that there are more Republican than Democrat legislators in the US Congress who contracted COVID-19.
Comparing the incidence in Republicans and Democrats to the US population, being a Republican legislator results in higher risk for infection than the general population. Being a Democratic legislator was not more risky than being a member of the general population.
Of course believing in small government doesn’t biologically predispose one to getting viral infections. It’s the behaviors that one exhibits that raise the risk, such as not wearing a mask at gatherings with other non-mask wearing people.
The NY Times data did not have the date of onset of disease for each legislator but it would have been interesting to do a survival analysis. I would have liked to see that curve!