citizens-online / cohealthmapr

A map to show all GP surgeries in England, looking at age and digital engagement. It is designed to help identify areas where there are more people who don’t tend to use digital tools and services, or who are not online. We’ve used NHS data on registration for online GP services to make the map.

This code and resulting map were created by Francis Barton of Citizens Online in March-April 2020.

See:

• this page for RMarkdown output and "behind the scenes"
• our blog post launching the map, and
• the map page and explanatory text.

Data and webpage information last updated January 2021.

Here's a listing of the contents of the data directory in my local repository. As you can see, the file sizes are quite large so I haven't uploaded this folder to GitHub.

library(dplyr)
tibble(files = list.files("data", recursive = TRUE)) %>% 
  mutate(size_mb = round(file.size(paste0("data/", files))/2^20))

# A tibble: 6 x 2
  files                                size_mb
  <chr>                                  <dbl>
1 gp-reg-pat-prac-sing-age-regions.csv       3
2 ons_nspcl_data.Rds                        17
3 POMI_APR2020_to_APR2020.csv               11
4 surgery_data_raw.csv                      21
5 tmp/ons_postcode_data.zip                206
6 tmp/pomi_2021.zip                          1

Look at pomi_data_dictionary.txt for descriptions of the Patient Online data.

This project is written in R using:

• RStudio
• the tidyverse
• janitor
• sf and
• leaflet

cohealthmapr's Issues

Provide more data about practices that aren't data-matched

The creation of full_data involves getting practice data (A) (practice code and patient age data), and then joining that up with the POMI data (B) (practice code, practice name, and patient enabled data).

Currently we do a left_join, so that only POMI data that matches a practice code from A is kept. Some practices in A don't get a match, and are currently excluded from the map.

Two or three job suggestions here:

• Keep details of practices that don't have matching POMI data (show as NA (grey) on map?)
• Check whether there are practices included in POMI that aren't in A (currently no checks for this) - use a full_join instead of left_join?
• Look into whether these can be matched up through a little bit of adjustment (could there even be typos that affect the join?)
• Report the missingness somewhere on the web page