Looking at Specialty Training Competition Ratios

- 20 mins

Health Education England’s Medical and Dental Recruitment and Selection website publishes the numbers of applicants and available spaces, and hence the competition ratios for each of the Specialties for the last few years of recruitment.

We will try to extract the data and make some inferences. I have borrowed Hadley Wickham’s approach in scraping data from html files.

#Load the required packages
require(rvest) #This package allows us to scrape data from html into R
require(dplyr) #To wrangle data
require(ggplot2) #To plot graphs

#The data for each year is laid out in its own separate page
x2015 <- read_html("http://specialtytraining.hee.nhs.uk/specialty-recruitment/competition-ratios/2015-competition-ratios/") %>%
  html_nodes("table") %>% 
  .[[1]] %>% 
  html_table()
x2014 <- read_html("http://specialtytraining.hee.nhs.uk/specialty-recruitment/competition-ratios/2014-competition-ratios/") %>%
  html_nodes("table") %>% 
  .[[1]] %>% 
  html_table() %>%
  select(-X3, -X5) #To remove the 2 columns with empty data
x2013 <- read_html("http://specialtytraining.hee.nhs.uk/specialty-recruitment/competition-ratios/2013-competition-ratios/") %>%
  html_nodes("table") %>% 
  .[[1]] %>% 
  html_table() %>%
  select(-X3, -X6, -X7)

There are some inconsistencies in the reporting of some of the Specialties between the different years. For example Medical Microbiology is not reported in 2015, but is reported in 2013 and 2014. Similarly Community Sexual and Reproductive Health is reported in 2014 and 2015, but not in 2013. It may be because there wasn’t any recruitment for some Specialties in some years. Also there is some inconsistency in the nomenclature, eg. Anaesthetics is sometimes reported as Anaesthetics (including ACCS Anaesthetics). Let’s join the data into a big table and clean it a little bit.

x2015[2,1] <- "ACCS EM"
x2014[2,1] <- "ACCS EM"
x2013[2,1] <- "ACCS EM"

x2015[3,1] <- "Anaesthetics"
x2014[3,1] <- "Anaesthetics"
x2013[3,1] <- "Anaesthetics"

x2015[4,1] <- "Broad-Based Training"
x2014[4,1] <- "Broad-Based Training"
x2013[4,1] <- "Broad-Based Training"

x2015[8,1] <- "Core Medical Training"
x2014[8,1] <- "Core Medical Training"
x2013[7,1] <- "Core Medical Training"

x2013[8,1] <- "Core Psychiatry Training"

x2014[18,1] <- "Paediatrics"
x2013[16,1] <- "Paediatrics"

#Join up the data into one dataframe
x <- inner_join(x2013, x2014, by = "X1") %>% inner_join(x2015, by = "X1")

#Rename the columns correctly
colnames(x)[1] <- "Specialty"
colnames(x)[2] <- "Apps2013"
colnames(x)[3] <- "Posts2013"
colnames(x)[4] <- "Ratio2013"
colnames(x)[5] <- "Apps2014"
colnames(x)[6] <- "Posts2014"
colnames(x)[7] <- "Ratio2014"
colnames(x)[8] <- "Apps2015"
colnames(x)[9] <- "Posts2015"
colnames(x)[10] <- "Ratio2015"
#Remove the 1st row
x <- x[-1, ]
row.names(x) <- 1:nrow(x)

Let’s see what we are left with.

require(knitr)
kable(x)
Specialty Apps2013 Posts2013 Ratio2013 Apps2014 Posts2014 Ratio2014 Apps2015 Posts2015 Ratio2015
ACCS EM 534 203 2.6 759 363 2.1 881 363 2.43
Anaesthetics 1189 478 2.5 1262 595 2.1 1294 629 2.06
Broad-Based Training 429 52 8.3 258 42 6.1 363 83 4.37
Clinical Radiology 751 185 4.1 798 227 3.5 917 247 3.71
Core Medical Training 3088 1209 2.6 3065 1468 2.1 2632 1550 1.7
Core Psychiatry Training 650 437 1.5 643 497 1.3 662 466 1.42
Core Surgical Training 1296 676 1.9 1370 625 2.2 1396 604 2.31
General Practice 6447 2787 2.3 5477 3391 1.6 5112 3612 1.42
Histopathology 154 120 1.3 165 93 1.8 189 79 2.39
Neurosurgery 183 37 4.9 159 24 6.6 169 30 5.63
Obstetrics and Gynaecology 591 204 2.9 583 240 2.4 599 238 2.52
Ophthalmology 323 71 4.5 353 82 4.3 374 95 3.94
Paediatrics 793 360 2.2 814 435 1.9 801 446 1.8
Public Health 602 70 8.6 686 78 8.8 724 88 8.23

This is still a little messy. The data is humanly readable, but we can’t really use it for analysis in R. We need to try and tidy things up, and express all the rows as observations (the different years and the different specialties), and columns as the variables in question (number of applications, number of available posts, competition ratios). So let’s do this.

#First let's add a column for the year
x2015 <- mutate(x2015, Year = 2015)
x2014 <- mutate(x2014, Year = 2014)
x2013 <- mutate(x2013, Year = 2013)

#Rename the columns (Could definitely be improved into an apply function)
colnames(x2015)[1] <- "Specialty"
colnames(x2015)[2] <- "Applicants"
colnames(x2015)[3] <- "Posts"
colnames(x2015)[4] <- "Ratio"

colnames(x2014)[1] <- "Specialty"
colnames(x2014)[2] <- "Applicants"
colnames(x2014)[3] <- "Posts"
colnames(x2014)[4] <- "Ratio"

colnames(x2013)[1] <- "Specialty"
colnames(x2013)[2] <- "Applicants"
colnames(x2013)[3] <- "Posts"
colnames(x2013)[4] <- "Ratio"

#Let's drop the first and last rows
x2015 <- x2015[-1, ]
x2015 <- x2015[-18, ]
row.names(x2015) <- 1:nrow(x2015)

#Repeat for the 2014 and 2013 dataframes
x2014 <- x2014[-1, ]
x2014 <- x2014[-19, ]
row.names(x2014) <- 1:nrow(x2014)

x2013 <- x2013[-1, ]
x2013 <- x2013[-17, ]
row.names(x2013) <- 1:nrow(x2013)

#Now we can bind the rows together
CompetitionRatios <- bind_rows(x2013, x2014) %>% bind_rows(x2015)

Now let’s look at the resulting dataframe

require(knitr)
kable(CompetitionRatios)
Specialty Applicants Posts Ratio Year
ACCS EM 534 203 2.6 2013
Anaesthetics 1189 478 2.5 2013
Broad-Based Training 429 52 8.3 2013
Cardiothoracic Surgery (Pilot) 68 6 11.3 2013
Clinical Radiology 751 185 4.1 2013
Core Medical Training 3088 1209 2.6 2013
Core Psychiatry Training 650 437 1.5 2013
Core Surgical Training 1296 676 1.9 2013
General Practice 6447 2787 2.3 2013
Histopathology 154 120 1.3 2013
Medical Microbiology & Virology 108 21 5.1 2013
Neurosurgery 183 37 4.9 2013
Obstetrics and Gynaecology 591 204 2.9 2013
Ophthalmology 323 71 4.5 2013
Paediatrics 793 360 2.2 2013
Public Health 602 70 8.6 2013
ACCS EM 759 363 2.1 2014
Anaesthetics 1262 595 2.1 2014
Broad-Based Training 258 42 6.1 2014
Cardiothoracic Surgery (Pilot) 72 7 10.3 2014
Clinical Radiology 798 227 3.5 2014
Community Sexual and Reproductive Health 33 7 4.7 2014
Core Medical Training 3065 1468 2.1 2014
Core Psychiatry Training 643 497 1.3 2014
Core Surgical Training 1370 625 2.2 2014
General Practice 5477 3391 1.6 2014
Histopathology 165 93 1.8 2014
Medical Microbiology 50 14 3.6 2014
Neurosurgery 159 24 6.6 2014
Obstetrics and Gynaecology 583 240 2.4 2014
Ophthalmology 353 82 4.3 2014
Oral and Maxillo Facial Surgery (Pilot) 33 4 8.25 2014
Paediatrics 814 435 1.9 2014
Public Health 686 78 8.8 2014
ACCS EM 881 363 2.43 2015
Anaesthetics 1294 629 2.06 2015
Broad-Based Training 363 83 4.37 2015
Cardiothoracic Surgery 68 8 8.5 2015
Clinical Radiology 917 247 3.71 2015
Community Sexual and Reproductive Health 100 2 50 2015
Core Medical Training 2632 1550 1.7 2015
Core Psychiatry Training 662 466 1.42 2015
Core Surgical Training 1396 604 2.31 2015
General Practice 5112 3612 1.42 2015
Histopathology 189 79 2.39 2015
Neurosurgery 169 30 5.63 2015
Obstetrics and Gynaecology 599 238 2.52 2015
Ophthalmology 374 95 3.94 2015
Oral and Maxillo Facial Surgery 27 5 5.4 2015
Paediatrics 801 446 1.8 2015
Public Health 724 88 8.23 2015

Now we can start doing something with the data.

Let’s now plot some graphs to visualise the data

#Need to first convert the Specialty column from a vector of characters to factors
CompetitionRatios$Specialty <- as.factor(CompetitionRatios$Specialty)
CompetitionRatios$Ratio <- as.numeric(CompetitionRatios$Ratio)
CompetitionRatios$Year <- as.factor(CompetitionRatios$Year)
CompetitionRatios$Applicants <- as.numeric(CompetitionRatios$Applicants)
CompetitionRatios$Posts <- as.numeric(CompetitionRatios$Posts)

ggplot(data=CompetitionRatios, aes(x=Year, y=Ratio, group = Specialty, colour = Specialty)) +
  geom_line()

center

So this is a little busy, it might be worth reducing the number of Specialties to just GP, CMT, CST, Paediatrics and Anaesthetics, the top 5 Specialties by number of applicants in all three years.

#Need to first convert the Specialty column from a vector of characters to factors
CompetitionRatios2 <- CompetitionRatios %>% filter(Specialty == "Anaesthetics" | Specialty == "Core Medical Training" | Specialty == "Core Surgical Training" | Specialty == "General Practice" | Specialty == "Paediatrics")

ggplot(data=CompetitionRatios2, aes(x=Year, y=Ratio, group = Specialty, colour = Specialty)) +
  geom_line() +
  geom_point(aes(shape = factor(Specialty)))

center

We can immediately see that generally speaking for this group of specialties, other than Core Surgical Training, the Competition Ratio is reducing over these 3 years. Why is this so? Is it due more to dropping applicant numbers or an increase in posts?

Let’s plot the graphs of Applicants over time and Posts over time.

#Need to first convert the Specialty column from a vector of characters to factors
CompetitionRatios2 <- CompetitionRatios %>% filter(Specialty == "Anaesthetics" | Specialty == "Core Medical Training" | Specialty == "Core Surgical Training" | Specialty == "General Practice" | Specialty == "Paediatrics")

ggplot(data=CompetitionRatios2, aes(x=Year, y=Applicants, group = Specialty, colour = Specialty)) +
  geom_line() +
  geom_point(aes(shape = factor(Specialty)))

center

ggplot(data=CompetitionRatios2, aes(x=Year, y=Posts, group = Specialty, colour = Specialty)) +
  geom_line() +
  geom_point(aes(shape = factor(Specialty)))

center

We can see that for GP, and CMT the number of applicants have been dropping year-on-year. Whereas for CST and Anaesthetics, the applicant numbers have been rising slightly. Paediatrics has been relatively static.

However, despite the dropping number of applicants for GP and CMT, there have also been a significant increase year-on-year in the number of available posts for these specialties. Anaesthetics and Paediatrics have also had some increase in their number of posts.

In conclusion, GP and CMT in particular have much less uptake in applicants, seemingly indicating that they are becoming less popular, despite the presumably increasing workforce demands for these two specialties.

Danny Wong

Danny Wong

Anaesthetist & Health Services Researcher

comments powered by Disqus
rss facebook twitter github youtube mail spotify instagram linkedin google pinterest medium