NHS-R — Statistical Process Control Charts

class: title-slide, left, bottom

<div>
<style type="text/css">.xaringan-extra-logo {
width: 110px;
height: 128px;
z-index: 0;
background-image: url(img/logo-nhs-blue.png);
background-size: contain;
background-repeat: no-repeat;
position: absolute;
top:1em;right:1em;
}
</style>
<script>(function () {
  let tries = 0
  function addLogo () {
    if (typeof slideshow === 'undefined') {
      tries += 1
      if (tries < 10) {
        setTimeout(addLogo, 100)
      }
    } else {
      document.querySelectorAll('.remark-slide-content:not(.title-slide):not(.inverse):not(.hide-logo)')
        .forEach(function (slide) {
          const logo = document.createElement('div')
          logo.classList = 'xaringan-extra-logo'
          logo.href = null
          slide.appendChild(logo)
        })
    }
  }
  document.addEventListener('DOMContentLoaded', addLogo)
})()</script>
</div>

# NHS-R &mdash; Statistical Process Control Charts
----
## **Examples of how to produce SPCs in R**
### Zoë Turner
### March 2021

---

## Package details

```r
install.packages("qicharts2")
library(qicharts2)
```

</br>

- Available on CRAN

- Comes with sample healthcare datasets

- Created by Jacob Anhøj, a doctor in Denmark

---

# Why I like this package ...

## Reasons to be joyful

* I skim instructions

* I just _get_ healthcare data (not plants and certainly not cars)

* Reproducible examples in the vignette

* More than one type of SPC available

---

## Like this | Snapshot of the vignette:

---

# Nightingale's Crimea...

## Getting data

Copied data manually from this [site](https://understandinguncertainty.org/node/214)

using a GUI (graphical user interface) package called `library(datapasta)` to copy directly to R

Saved in a script for quick retrieval later

```r
source("Crimea.R")
```

---

## Being #rstats

There is a package for that... even for historical data!

```r
install.packages("HistData")
library(HistData)

data("Nightingale")
```

---

## And the coxcomb can be recreated...

### ... and animated as a GIF

Credit: Sian Bladon, Winner of the most [innovative FloViz Competition 2020 ](https://statsyss.wordpress.com/2020/05/11/floviz-challenge-entries/) from the Young Statisticians Section of the Royal Statistical Society

---

# Replot in qicharts2

## Run chart code

Built on ggplot2

```r
qic(startMonth,number,
    data     = crimeaLong,
    chart    = 'run',
    decimals = 0, 
    x.angle  = 90,
    title    = 'Run chart deaths in Crimea War',
    ylab     = 'Number of deaths',
    xlab     = 'Month'
)
```

---

# Replot in qicharts2

## Run chart code

![](data:image/png;base64,#spc-presentation_files/figure-html/unnamed-chunk-2-1.png)

---

# Sprinkle ggplot2 magic...

## Facets

```r
qic(startMonth,number,
    data     = crimeaLong,
    chart    = 'run',
    decimals = 0, 
*   facets   = ~death_cause,
    x.angle  = 90,
    title    = 'Run chart deaths in Crimea War',
    ylab     = 'Number of deaths',
    xlab     = 'Month'
)
```

---

# Sprinkle ggplot2 magic...

## Facets

![](data:image/png;base64,#spc-presentation_files/figure-html/unnamed-chunk-3-1.png)

---

# Just imagine...

Producing...

* 30 incident categories

* 10 wards

* 25 LSOAs

Looking for for signals. That's a lot of individual plots to do in SQL or Excel.

---

## Moving Range SPC

```r
qic(startMonth,number,
    data     = crimeaLong,
*   chart    = 'mr',
    decimals = 0, 
    facet    = ~death_cause,
    x.angle  = 90,
    title    = 'SPC mr chart deaths in Crimea War',
    ylab     = 'Number of deaths',
    xlab     = 'Month'
)
```

---

## mr Chart (or is that Mr Chart!)

As seen on Twitter @Letxuga007

![](data:image/png;base64,#spc-presentation_files/figure-html/unnamed-chunk-4-1.png)

---

# Other SPC charts are available

## Rare events - time between

The code requires a count of days difference.

.footnote[Cabg = coronary artery bypass operations]

```r
# Altered slightly from the qicharts2 vignette

fatalities <- cabg %>% 
  filter(death) %>% 
  mutate(dt = date - lag(date))
```

---

## T-chart

Time between used for rare events like deaths from operations

```r
tchart <- qic(dt,
    data  = fatalities,
*   chart = 't',
    title = 'Days between deaths (T chart)',
    ylab  = 'Days',
    xlab  = 'Death #')

tchart
```

---

## T-chart

A point above the upper control limit is a good thing!

![](data:image/png;base64,#spc-presentation_files/figure-html/unnamed-chunk-5-1.png)

---

# Breaking into the chart

## Getting the data

The numbers generated to create the charts can be viewed/saved by typing:

```r
newObject <- tchart$data
```

---

## When is this useful?

In this blog: https://www.r-bloggers.com/backed-by-data/ @_johnmackintosh shows how to:

* tweak the charts

<br>

And I used this code

* to include only those charts with a signal
* to link back to the data to say which patients/incidents/wards were behind the points. This data can be automatically generated in a report

John Mackintosh has also created 2 specific R packages:

[runcharter](https://github.com/johnmackintosh/runcharter) </br>
--
[spccharter](https://github.com/johnmackintosh/spccharter)

---

# Other interesting things

## Further information

SPC Shiny dashboards are being discussed on the NHS-R Slack site in the channel #proj-shiny-spc

Will be based on some of the work started by [Chris Reading](https://github.com/chrisreading01/SPCwizard) ([patch](https://github.com/ChrisBeeley/SPCwizard/tree/patch-1) by Chris Beeley) and Simon Wellesley-Miller.

- please join us and contribute

link: [nhsrcommunity.slack.com](nhsrcommunity.slack.com)

- Twitter: [@NHSrCommunity](https://twitter.com/NHSrCommunity)
- Me: [@Letxuga007](https://twitter.com/Letxuga007)

.footnote[The **nhsr theme** was designed by [Silvia Canelon](https://silvia.rbind.io) for, and with feedback from, the [NHS-R Community](https://github.com/nhs-r-community), following the [NHS identity guidelines](https://www.england.nhs.uk/nhsidentity/identity-guidelines/).]