Help for package NHSRdatasets

Type:

Package

Title:

NHS and Healthcare-Related Data for Education and Training

Date:

2021-03-09

Version:

0.3.0

Maintainer:

Chris Mainey <c.mainey@nhs.net>

Description:

Free United Kingdom National Health Service (NHS) and other healthcare, or population health-related data for education and training purposes. This package contains synthetic data based on real healthcare datasets, or cuts of open-licenced official data. This package exists to support skills development in the NHS-R community: https://nhsrcommunity.com/.

License:

CC0

Language:

en-GB

Encoding:

UTF-8

LazyData:

true

RoxygenNote:

7.1.1

Depends:

R (≥ 3.5.0)

BugReports:

https://github.com/nhs-r-community/NHSRdatasets/issues

Imports:

tibble

Suggests:

dplyr, caret, e1071, readr, lattice, magrittr, varhandle, rsample, ggplot2, lme4, MASS, ModelMetrics, lmtest, rmarkdown, scales, ggrepel, lubridate, tidyr, forcats, knitr, janitor, stringi, synthpop

VignetteBuilder:

knitr

URL:

https://github.com/nhs-r-community/NHSRdatasets, https://nhs-r-community.github.io/NHSRdatasets/

NeedsCompilation:

Packaged:

2021-03-11 08:09:48 UTC; Christopher

Author:

Gary Hutson [aut], Tom Jemmett

[aut], Chris Mainey

[aut, cre], Zoë Turner [aut], NHS-R community [cph]

Repository:

CRAN

Date/Publication:

2021-03-14 00:00:06 UTC

Hospital Length of Stay (LOS) Data

Description

Artificially generated hospital data. Fictional patients at 10 fictional hospitals, with LOS, Age and Date status data Data were generate to learn Generalized Linear Models (GLM) concepts, modelling either Death or LOS.

Usage

data(LOS_model)

Format

Data frame with five columns

ID: A fictional patient ID number
Organisation: A factor representing one of ten fictional hospital trusts, e.g. Trust1
Age: Age in years of each fictional patient
LOS: In-hospital length of stay in days. The difference between admission and discharge date in dates
Death: Binary for death status: 0 = survived, 1= died in hospital

Source

Generated by Chris Mainey chris.mainey@uhb.nhs.uk, Feb-2019

Examples

data(LOS_model)

model1 <- glm(Death ~ Age + LOS, data=LOS_model, family="binomial")
summary(model1)

# Now with an Age, LOS, and Age*LOS interaction.
model2<- glm(Death ~ Age * LOS, data=LOS_model, family="binomial")
summary(model2)

NHS England Accident & Emergency Attendances and Admissions

Description

Reported attendances, 4 hour breaches and admissions for all A&E departments in England for the years 2016/17 through 2018/19 (Apr-Mar). The data has been tidied to be easily usable within the tidyverse of packages.

Usage

data(ae_attendances)

Format

Tibble with six columns

period: The month that this data relates to
org_code: The ODS code for this provider
type: The department type. either 1, 2 or other
attendances: the number of patients who attended this department in this month
breaches: the number of patients who breaches the 4 hour target in this month
admissions: the number of patients admitted from A&E to the hospital in this month

Details

Data sourced from NHS England Statistical Work Areas which is available under the Open Government Licence v3.0

Source

NHS England Statistical Work Areas

Examples

data(ae_attendances)
library(dplyr)
library(ggplot2)
library(scales)

# Create a plot of the performance for England over time
ae_attendances %>%
  group_by(period) %>%
  summarise_at(vars(attendances, breaches), sum) %>%
  mutate(performance = 1 - breaches / attendances) %>%
  ggplot(aes(period, performance)) +
  geom_hline(yintercept = 0.95, linetype = "dashed") +
  geom_line() +
  geom_point() +
  scale_y_continuous(labels = percent) +
  labs(title = "4 Hour performance over time")

# Now produce a plot showing the performance of each trust
ae_attendances %>%
  group_by(org_code) %>%
  # select organisations that have a type 1 department
  filter(any(type == "1")) %>%
  summarise_at(vars(attendances, breaches), sum) %>%
  arrange(desc(attendances)) %>%
  mutate(performance = 1 - breaches / attendances,
         overall_performance = 1 - sum(breaches) / sum(attendances),
         rank = rank(-performance, ties.method = "first") / n()) %>%
  ggplot(aes(rank, performance)) +
  geom_vline(xintercept = c(0.25, 0.5, 0.75), linetype = "dotted") +
  geom_hline(yintercept = 0.95, colour = "red") +
  geom_hline(aes(yintercept = overall_performance), linetype = "dotted") +
  geom_point() +
  scale_y_continuous(labels = percent) +
  theme_minimal() +
  theme(panel.grid = element_blank(),
        axis.text.x = element_blank()) +
  labs(title = "4 Hour performance by trust",
       subtitle = "Apr-16 through Mar-19",
       x = "", y = "")

Deaths registered weekly in England and Wales, provisional

Description

Provisional counts of the number of deaths registered in England and Wales, by age, sex and region, in the latest weeks for which data are available.

Usage

data(ons_mortality)

Format

Data frame with five columns

category_1: character, containing the names of the groups for counts, e.g. "Total deaths", "all ages".
category_2: character, subcategory of names of groups where necessary, e.g. details of region: "East", details of age bands "15-44".
counts: numeric, numbers of deaths in whole numbers and average numbers with decimal points. To retain the integrity of the format this column data is left as character.
date: date, format is yyyy-mm-dd; all dates are a Friday.
week_no: integer, each week in a year is numbered sequentially.

Details

Source and licence acknowledgement This data has been made available through Office of National Statistics under the Open Government Licence http://www.nationalarchives.gov.uk/doc/open-government-licence/version/3/

Source

Collected by Zoë Turner zoe.turner2@nottshc.nhs.uk, Apr-2020 from https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/datasets/weeklyprovisionalfiguresondeathsregisteredinenglandandwales

Examples

data(ons_mortality)

library(dplyr)
library(tidyr)

wideForm <- ons_mortality %>%
 select(-week_no) %>%
 pivot_wider(names_from = date,
             values_from = counts
 )

Stranded Patient (Patients flagged as having a greater than 7 day LOS) Model

Description

This model is to be used as a machine learning classification model, for supervised learning. The binary outcome is stranded vs not stranded patients.

Usage

data(stranded_data)

Format

Tibble with nine columns (1 x outcome and 8 predictors)

stranded.label: Outcome variable - whether the patient is stranded or not
age: Patient age on admission
care.home.referral: Whether than have been referred from a care home
medicallysafe: Medically safe for discharge - means the patient is assessed as safe, but has not been discharged yet
hcop: Indicates whether they have been triaged from a Health Care for Older People specialty
mental_health_care: Flag to indicate whether they need mental health support and care
periods_of_previous_care: Count of the number of previous spells of care
admit_date: Date they were admitted to hospital
frailty_index: An initial index assessment to say if the patient is frail or not. This is needed for alignment of service provision.

Source

Synthetically generated by Gary Hutson g.hutson@nhs.net, Mar-2021.

Examples

library(magrittr)
library(dplyr)
data("stranded_data")
stranded_data %>%
 glimpse()

Synthetic National Early Warning Scores Data

Description

Synthetic NEWS data to show as the results of the NHSR_synpop package. These datasets have been synthetically generated by this package to be utilised in the NHSRDatasets package.

Usage

data(synthetic_news_data)

Format

Tibble with twelve columns

male: character string containing gender code
age: age of patient
NEWS: National Early Warning Score (NEWS)
syst: Systolic BP - Systolic BP result
dias: Diastolic Blood Pressure - result on NEWS scale
temp: Temperature of patient
pulse: Pulse of the patient
resp: Level of response from the patient
sat: SATS(Oxygen Saturation Levels) of the patient
sup: Suppressed Oxygen score
alert: Level of alertness of patient
died: Indicator to monitor patient death

Source

Generated by Dr. Muhammed Faisal and created by Gary Hutson g.hutson@nhs.net, Mar-2021

Examples

library(magrittr)
library(dplyr)
data("synthetic_news_data")
synthetic_news_data %>%
 glimpse()