Diabetes Eight

Introduction

I’m getting into visualisation tools nowadays to help us interpret data more interactively than just static Power BI charts and dashboards. D8 is an embryonic step in this direction.

Background

Due to Covid, there is a huge backlog of long-term condition reviews which need to be performed, of which Diabetes is probably one of the most important and also the hardest to manage. Unlike other conditions which need one or two parameters, for example, Blood Pressure for patients with Hypertension, for Diabetes, in simple terms, you have 3 main cornerstones to optimise treatment: Blood Pressure, Cholesterol and Glucose Control which add complexity to searches and identifying populations.

In my hometown of Sutton, we were tasked to help identify the population to focus on in terms of complete reviews not done and also unoptimised treatment values. In total, there are 9 care processes, but we normally focus on 8 as the 9th one is Retinal Screening and is normally dealt with externally. The 8 Care processes are Cholesterol, Hba1c, Blood Pressure, Foot Checks, Renal Profiling, Checking Urine for Protein, Smoking and BMI. Of those who have had their tests, the focus, as stated above, is on Blood Pressure, Cholesterol and Glucose Control See NICE impact diabetes.

The Problem

So we have 2 problems.

To be able to export from EMIS which patients need which care processes and group by outstanding processes.
For those who have a recording to focus on the abnormal readings and work out the poorly patients.

We looked at this, and it was not easy to manage the above in one search in EMIS to dynamically select patient groups, order them by care processes done and risk score them without complex pivots, so I got to work on creating a tool to help act as a middleman between an EMIS search and export in a format which is more useful.

What is D8?

D8 takes an export from EMIS and parses it into a visualisation tool for the user to select and identify populations they’d like to focus on to help the agenda of getting our Diabetes review done and focusing on the right population of higher-risk patients. If pictures paint 1000 words, here is a video of the application in action.

It works by importing an EMIS Export and using this data for you to interactively choose a population of your choice based on Care Processes not done. So the 0 Care Processes Group are those who need everything done.

The Colours are Rag Rated based on a risk score from the 3 Tiers: Blood Pressure, Cholesterol and Hba1c. The thresholds are

For Hba1c (<53 = low risk, 53-65 = mid risk, >65 = high risk)
For Systolic BP (<145 = low risk, 145-180 = mid risk, > 180 = high risk)
For Diastolic BP (<90 = low risk, 90-105 = mid risk, > 105 = high risk)
For Cholesterol (<5 = low risk, 5-6 = mid risk, >6.0 = high risk)

I really wanted to focus on Non-HDL or Qrisk for Cholesterol risk, but the specification is around cholesterol. Maybe a future version will include this instead, as it’s more accurate. Also, the spec talks about creatinine and urine albumin when we should be looking at eGFR and uACR . Maybe I’ll update if the specs update.

Notice how there are different types of icons. These represent your true population and give you a glimpse of gender and age (e.g. elderly have a walking stick, housebound are in a wheelchair). I might explore this feature further, but I thought just having one icon would be a bit boring.

Installation and Walkthrough

Step 1
You can find the search to import here.
Download the EMIS Search
On the browser, you sometime will get the raw XML displayed, just save it on the desktop from a right click on the text and click on the save as menu option.

Import it into EMIS.

Step 2
Export the search to the desktop as a csv and make sure you exclude the report header, and replicate patient details
The Searches are care processes achieved in the current financial year, excluding Gestational and Type 1 Diabetics. There was an exclusion around patients under the active care of the hospital for their diabetics but we don’t have this information, and type 2’s should really be in the community anyway.

Step 3
Download D8 from here and extract it to the desktop in the default folder

Step 4
Run D8! It will look for the correct file saved on the desktop so don’t change it’s name.
Left-click and drag to select groups
R to Select High Risk, Y to select Med Risk, G to select Low Risk
Black are those where the risk can’t be calculated (ie 0 care processes done)
When selected press E for Export
You’ll find the csv export on the desktop

Step 5
Open the File
You’ll see each patient you’ve selected with columns representing outstanding targets with values if appropriate

You can then process the targets as you see fit in this format. The size of your practice shouldn’t matter. We have it working in Sutton with a total population of 200,000 patients!

IG

I’m not here to help you with IG in your set up I’m afraid but the app should work from extracting the zip file to the desktop and running the executable. If you get issues with Window Defender blocking the executable, right click on DiabetesEight.exe and the properties and click on the Unblock tick box then run again.

In terms of IG

no data from the application goes outside the computer
it only processes data from the EMIS Search and no external data is brought in
The export is in csv format and only on the desktop
We recommend you delete the csv after you’ve processed to maintain use for a long as necessary

Final Thoughts

D8 is free to anyone who would like a bit of help to identify and case manage their Diabetes 8 population in terms of care processes and high-risk patients. I will try to help set it up if you have difficulties so contact me if you need to.

I hope you find it useful in your travels.