Awesome

deprivare

This provides code and data for indices of deprivation, initially supporting the UK.

We know socio-economic deprivation has a significant effect on health outcomes.

This repository provides a simple microservice, embeddable library and command-line tools to allow other software to make use of deprivation indices in the UK.

I use such data in operational electronic health record systems and analytics.

You can use the NHS Postcode Database to lookup a UK postal code and find out the LSOA code for that area. My microservice and library nhspd provides a simple lightweight wrapper around that data product.

I combine medical records with nhspd and the data made available via this repository so I can include deprivation as a factor in analysis using a graph-like API as part of my PatientCare EPR.

Deprivare uses a SQLite database which can be used directly (e.g. if you are using Python which has SQLite bindings built-in, or any other programming language with SQLite bindings available), or via an API on the JVM (e.g. from Java or Clojure), or via a HTTP API through running a HTTP server.

Deprivation indices

Unfortunately, the "Welsh Index of Multiple Deprivation" and the "English Index of Multiple Deprivation" are not easily comparable. Each generates a rank based on a geographical region called an LSOA (Lower Layer Super Output Area) but the top rank in Wales is not equivalent to the top rank in England.

Which dataset should I use?

Individual indices for a geographic region, such as England or for Wales, are available. If you are studying a population from a single region, you can use the index for that region. Indices are available are available for a range of domains, and usually include a composite index providing a view across multiple domains.

However, if your study population is from multiple areas, you cannot use disparate indices and assume rank 1 in, for example England, is the same as rank 1 in Wales. Likewise, you cannot assume the upper quartile in England is equivalent to the upper quartile in Wales.

The work by Abel, Payne and Barclay provides a way to harmonise across England and Wales, but there are also indices based on income and employment, published for England and Wales. Abel, Payne and Barclay published a composite UK index for 2016, and MySociety have used the same approach for 2020.

In practical terms, whichever index used, most studies will not want to use ranks directly, but instead use quartiles or quintiles.

What is this repository?

This repository provides a way to automatically download and make those data available in computing systems. It is designed to be composable with other data and computing services including but not limited to use in a graph-like API.

In essence, it provides a simple way to lookup a deprivation index based on LSOA, in the UK. A LSOA is a small defined geographical area of the UK containing about 1500 people designed to help report small area statistics. You can use nhspd to map from a UK postal code to an LSOA.

Getting started

This description assumes you want to run using source code directly. If there is interest, I can provide pre-built executable files as an alternative.

1. Download and install Clojure.

e.g. on mac os x with homebrew:

brew install clojure/tools/clojure

2. Clone this repository

git clone https://github.com/wardle/deprivare
cd deprivare

3. List available datasets

This will simply list the currently supported datasets. If you need a specific dataset that isn't currently supported, let me know via raising an issue and I'll try to add it for you.

clj -X:available

Result:

|                                                        :name |                         :id |
|--------------------------------------------------------------+-----------------------------|
|                   England Index of Deprivation - ranks, 2019 |      england-imd-2019-ranks |
|                 Welsh Index of Deprivation - quantiles, 2019 |    wales-imd-2019-quantiles |
|                     Welsh Index of Deprivation - ranks, 2019 |        wales-imd-2019-ranks |
| UK composite index of multiple deprivation, 2020 (MySociety) | uk-composite-imd-2020-mysoc |

4. Get information about a dataset

Specify a key value pair, :dataset and the 'id' of the dataset in which you are interested.

e.g.

clj -X:info :dataset uk-composite-imd-2020-mysoc

Result:

UK composite index of multiple deprivation, 2020 (MySociety)
------------------------------------------------------------
A composite UK score for deprivation indices for 2020 - based on England
with adjusted scores for the other nations as per Abel, Payne and Barclay but
calculated by Alex Parsons on behalf of MySociety.

5. Install dataset(s) into a file-based database.

You can ask for all known datasets to be installed.

clj -X:install-all :db depriv.db
=>
Installing dataset:  UK composite index of multiple deprivation, 2020 (MySociety)
Installing dataset:  Welsh Index of Deprivation - ranks, 2019
Installing dataset:  Welsh Index of Deprivation - quantiles, 2019
Installing dataset:  England Index of Deprivation - ranks, 2019

Usually, installing all known datasets is reasonable.

Alternatively, you can install a specific set of datasets:

clj -X:install :db depriv.db :dataset uk-composite-imd-2020-mysoc
clj -X:install :db depriv.db :dataset wales-imd-2019-quantiles
clj -X:install :db depriv.db :dataset wales-imd-2019-ranks

Each dataset will be downloaded and then imported. Each will take a few seconds only.

6. List installed datasets in a database:

clj -X:installed :db depriv.db

7. Run a web service (optional) on port 8080:

clj -X:server :db depriv.db

You can specify a port to use, if you need different to the default:

clj -X:server :db depriv.db :port 8081

You can then request deprivation data by LSOA:

➜  deprivare git:(main) ✗ http -j 127.0.0.1:8080/v1/uk/lsoa/W01001552

The results will vary depending on what datasets are installed, but will be of the form dataset/key.

The keys are a close reproduction of the original source data.

{
  "uk-composite-imd-2020-mysoc/UK_IMD_E_pop_decile": 9,
  "uk-composite-imd-2020-mysoc/UK_IMD_E_pop_quartile": 4,
  "uk-composite-imd-2020-mysoc/UK_IMD_E_pop_quintile": 5,
  "uk-composite-imd-2020-mysoc/UK_IMD_E_rank": 37508,
  "uk-composite-imd-2020-mysoc/UK_IMD_E_score": 6.445237018687946,
  "uk-composite-imd-2020-mysoc/lsoa": "W01001552",
  "wales-imd-2019-quantiles/authority_name": "Monmouthshire",
  "wales-imd-2019-quantiles/lsoa": "W01001552",
  "wales-imd-2019-quantiles/lsoa_name": "Dixton with Osbaston",
  "wales-imd-2019-quantiles/wimd_2019": 1817,
  "wales-imd-2019-quantiles/wimd_2019_decile": 10,
  "wales-imd-2019-quantiles/wimd_2019_quartile": 4,
  "wales-imd-2019-quantiles/wimd_2019_quintile": 5,
  "wales-imd-2019-ranks/access_to_services": 868,
  "wales-imd-2019-ranks/authority_name": "Monmouthshire",
  "wales-imd-2019-ranks/community_safety": 1894,
  "wales-imd-2019-ranks/education": 1894,
  "wales-imd-2019-ranks/employment": 1867,
  "wales-imd-2019-ranks/health": 1882,
  "wales-imd-2019-ranks/housing": 1845,
  "wales-imd-2019-ranks/income": 1905,
  "wales-imd-2019-ranks/lsoa": "W01001552",
  "wales-imd-2019-ranks/lsoa_name": "Dixton with Osbaston",
  "wales-imd-2019-ranks/physical_environment": 442,
  "wales-imd-2019-ranks/wimd_2019": 1817
}

I usually advise thinking of a service as immutable once created. That means you create a service and potentially keep it running. I tend to switch versions by using an API gateway or reverse proxy. Given the nature of deprivation data, you will likely only switch to a different running service as and when new datasets become available. Alternatively, you can, of course, add a new dataset to an existing database. But generally, I advise thinking of these services as disposable and potentially ephemeral. Think cattle not pets.

Running from Docker

An example Dockerfile is provided. This builds an uberjar and then creates a deprivare data file. Running the resulting image runs a HTTP server that returns deprivation data for a given LSOA.

docker build -t deprivare .
docker run --publish 8080:8080 deprivare

Mark