Data607_HW5_WilliamAiken.Rmd

Introduction

This is a creation and analysis of Israeli Vaccination data for Covid 19

Methods

1. First we need to recreate the table of vaccination data

• create a data frame to populate with our vaccination data
• create column names based on the table provided
• create a row name called ‘Age’

covid_df <-
  data.frame(
    Age = as.character(),
    "Population.Not.Vaxed" = as.character(),
    "Population.Fully.Vaxed" = as.character(),
    "Severe.Cases.Not.Vax.per.100K" = as.character(),
    "Severe.Cases.Fully.Vax.per.100K" = as.character(),
    "Efficacy.vs.severe.disease" = as.character(), row.names = "Age"
  )

2. Populate the rows of our csv, with a primary row based on age stratification

• Create a secondary row with the age percentage
• Blank values are represented with ‘NA’

covid_df['<50',]<- c('1,116,834', '3,501,118', '43', '11', NA)
covid_df['<50%',]<- c('23.3%', '73.0%', NA, NA, NA)
covid_df['>50',]<- c('186,078', '2,133,516', '171', '290', NA)
covid_df['>50%',]<- c('7.9%%', '90.4%', NA, NA, NA)

covid_df %>% kbl %>% kable_classic

	Population.Not.Vaxed	Population.Fully.Vaxed	Severe.Cases.Not.Vax.per.100K	Severe.Cases.Fully.Vax.per.100K	Efficacy.vs.severe.disease
<50	1,116,834	3,501,118	43	11	NA
<50%	23.3%	73.0%	NA	NA	NA
>50	186,078	2,133,516	171	290	NA
>50%	7.9%%	90.4%	NA	NA	NA

3. Save out our csv that we are going to commit to github

• We don’t want to run this again

#write.table(covid_df, "israeli_vaccination_data.csv", row.names=TRUE, sep = ",")

4. Load csv from github with readr

file_path <- "https://raw.githubusercontent.com/catfoodlover/Data607/main/israeli_vaccination_data.csv"
covid_df2 <- read_csv(file_path, show_col_types = FALSE)

5. readr saves the data as a tibble which doesn’t support row names so we need to fix our data structrue

• convert to a data frame
• name the rows the first column’s values
• save out a list of column names
• delete the first column
• split the last column which contains a comma separated list
• rename the columns with the saved list
• set all the strings ‘NA’ to NA

covid_df2 <- as.data.frame(covid_df2)
rownames(covid_df2) <- covid_df2[, 1]

## Warning: One or more parsing issues, see `problems()` for details

names_list <- names(covid_df2)
covid_df2[,1] <- NULL

covid_df2 <- covid_df2 %>% separate(Efficacy.vs.severe.disease, sep = ",", c("temp1", "temp2"))

colnames(covid_df2) <- names_list

covid_df2[covid_df2 == 'NA'] <- NA

6. We want to get the information in those secondary rows into columns

• use case_when to detect row names and put the age stratification into a new column called ‘age_group’
• group by age_group and move those population %s into a new column and use the ‘fill’ function to populate all the rows
• remove that no longer needed secondary row

covid_df2 <-
  covid_df2 %>% mutate(age_group = case_when(
    str_detect(row.names(.), "<50") ~ "<50",
    str_detect(row.names(.), ">50") ~ ">50"
  )) %>% group_by(age_group) %>% mutate(
    percent.not.vaxed = str_extract(Population.Not.Vaxed, ".*%"),
    percent.vaxed  = str_extract(Population.Fully.Vaxed, ".*%")
  ) %>% fill(percent.not.vaxed, .direction = c("up")) %>% fill(percent.vaxed, .direction = c("up")) %>% ungroup(.) %>% filter(!is.na(Severe.Cases.Not.Vax.per.100K))

covid_df2 %>% kbl() %>% kable_classic()

Population.Not.Vaxed	Population.Fully.Vaxed	Severe.Cases.Not.Vax.per.100K	Severe.Cases.Fully.Vax.per.100K	Efficacy.vs.severe.disease	age_group	percent.not.vaxed	percent.vaxed
1,116,834	3,501,118	43	11	NA	<50	23.3%	73.0%
186,078	2,133,516	171	290	NA	>50	7.9%%	90.4%

7. Now we can calculate our Efficacy vs severe disease

• we cast our severe cases per 100K to numeric and divide by 100K to get percentages
• we calculate Efficacy as 1 - (% Severe cases vaccinated per 100K)/(% Severe cases not vaccinated per 100K)

covid_df2 <-
  covid_df2 %>% mutate(Efficacy.vs.severe.disease = 1 - ((as.numeric(Severe.Cases.Fully.Vax.per.100K)/100000)/(as.numeric(Severe.Cases.Not.Vax.per.100K)/100000)))

Results

1. Do we have enough information to calculate the total population?

Yes we do, we know what percentages for both the under and over 50

• We divide either treatment arm by the percent population they represent to get the total population
• We sum the totals for both age groups to get the overall total population

(under_50_pop <- 3501118/.73)

## [1] 4796052

(over_50_pop <- 2113516/.904)

## [1] 2337960

(total_pop <- under_50_pop + over_50_pop)

## [1] 7134012

This population represents people in Israel old enough to get vaccinated (12 years and older), total pop ~8.8 million

2. We have already calculated the efficacy

temp <- covid_df2 %>% group_by(age_group) %>% summarise(Efficacy.vs.severe.disease)

kbl(temp) %>% kable_classic()

age_group	Efficacy.vs.severe.disease
<50	0.7441860
>50	-0.6959064

• We can see that vaccination has an efficacy of 74% in the under 50 group
• The vaccine has a surprising -69% efficacy in the over 50 group
• We will dig into explaining why this is in the next question

3. From your calculation are you able to compare the rates of severe disease in unvaccinated and vaccinated individuals?

covid_df2 %>% select(age_group, Population.Not.Vaxed, percent.not.vaxed, Population.Fully.Vaxed, percent.vaxed, Efficacy.vs.severe.disease) %>% kbl() %>% kable_classic()

age_group	Population.Not.Vaxed	percent.not.vaxed	Population.Fully.Vaxed	percent.vaxed	Efficacy.vs.severe.disease
<50	1,116,834	23.3%	3,501,118	73.0%	0.7441860
>50	186,078	7.9%%	2,133,516	90.4%	-0.6959064

My calculation doesn’t allow me to measure vaccine efficacy for a couple of reasons.

• Our percentages don’t add up to 100% so I’m assuming partially vaccinated people and people vaccinated for less than 14 days are excluded?
• Those people should be in another category
• It also unclear to me what ‘per 100k means’, does it mean the entire cohort (>50/<50) or 100k within that age group and treatment arm?
• The fact that such a high percentage of older Israelis have been vaccinated it’s not surprising that there are more breakthrough infections than infections in the unvaccinated