Unlocking meaning from private data… Visualising lifestyles that lead to poor coronary health
Filed in: General News, Health Mapping
March 15, 2012
Having recently completed some work for the NHS (more on that soon), we’re on a bit of a health kick here at BD. Today I decided to use the TGI, our people classification P2 and some work we did with the NWPHO[1] to try and visualise the link between healthy eating and heart disease.
Consider the below maps. They show the frequency of people that eat pizza more than once per week, the number of people that think eating their ‘five-a-day’ is very important and the density of lifestyles likely to lead to coronary disease.
Note: The darker the red, the higher the reading / frequency. The maps display anonymised and aggregated data and not real cases.
- Map A – People that believe in eating fruit and veg 5 times a day
- Map B – Lifestyles that may lead to coronary disease
- Map C – People that eat pizza more than once a week.
It’s not hard to see that maps B and C show a similar pattern.
The conclusion that people who eat pizza frequently are more prone to heart disease is no surprise nor is the conclusion that people who get their ‘five-a-day’ are likely to have healthier hearts.
What is interesting is that a source of data which is rightly very private has been compared with another data set. As this is modelled data, it is important to note that it does not reveal any individual patient records.
My question is: Is this a useful technique?
This graph has a correlation of 0.61. A similar chart for five-a-day and heart disease gives a negative correlation of -0.42.
These are no surprises. We have made the hole and dug up what was expected but we have joined very private health data with another dataset. Should we keep digging?
Posted by Geoff Beacon, Chairman at Beacon Dodsworth
[1] As described in An inverse correlation between coronary disease and skin cancer.
[2] TGI data sourced from Kantar Media.
March 16, 2012 by alec fraher
Well done BD.
As a former health and social care information and performance manager I’d endorse your efforts.
If health reforms are going to be meaningful then getting at the heuristics in geodemographic mapping is mighty powerful.
Your work can only add value – narrowing the gap between what the stats say and the reality, a bit more closely, has to be welcomed.
It is that simple.
Good intelligence can add life to years and years to lives as Alf Morris, now Lord Morris of Manchester, once said.
And, perhaps more importantly, enable the identification of the the health and social services’ black swans and blue oceans much needed to shift the balance of care towards prevention and good public health.
It’s a No Brainer.
Once again, well done – I know its detailed and frustrating work.