5 minutes to read this insight
We were recently commissioned by Bradford Districts CCG to investigate respiratory conditions across the three CCGs in the area as part of the “Bradford NHS Open for Innovation” event by Digital Catapult, Medipex at DHEZ.
Initially we focussed on “NHS RightCare – Commissioning for value” packs. The packs identify ten other CCGs across England that have similar characteristics, enabling comparison. Focussing on respiratory conditions we created a Chord Diagram to transform the data to visualise the three lowest performing CCGs for each sub-topic within the Respiratory Pack:
Res 01 Asthma – Emergency admissions by children
Res 02 Asthma – Emergency admissions by adults
Res 03 Acute lower Resiratory infections – Emergency child admissions
Res 04 Acute lower Resiratory infections – Emergency adult admissions
Res 05 Asthma patients (8yrs+) with measures of variability or reversibility
Res 06 Asthma patients who have had a review (12 months) (%)
Res 07 Emergency admission rate for children with asthma, 0-18yrs
Res 08 Asthma patients, 14-19, where smoking status is recorded
Res 09 Smokers- support and treatment offered
Res 10 Resiratory conditions – Total non-elective spend
Res 11 Asthma – Non-elective spend
Res 12 Reported to estimated prevalence of COPD
Res 13 COPD patients who have had flu immunisation
Res 14 COPD patients who have had a review & breathlessness assessment
Res 15 COPD patients with a record of FeV1 in the preceding 12 months
Res 16 <75 Mortality from bronchitis, emphysema & COPD
With this knowledge, we then looked at 6 years of monthly Prescription Data from NHS Digital. We used almost 1 billion rows of data to select data from the BNF that related to respiratory conditions (Beclometasone, Dipropionate, Budesonide, Ciclesonide, Fluticasone Propionate (Inh), and Mometasone Furoate). For this we used Amazon Web Services to run our algorithm – only selecting practices within the Bradford CCG’s, before filtering the data so we only got prescriptions for respiratory. This took around 6 hours once we hit the “go” button. Whilst we could see seasonal trends in the data, we could also view which practices were outliers over time:
Socio-demographic variables explain around 45 per cent of the variation in emergency admissions between GP practices, with deprivation more strongly linked to emergency than to elective admission. Duffy R, Neville R, Staines H (2002).
Speaking with the CCGs, they were concerned about why so many people with respiratory conditions attend A&E rather than going to their local GP. This was a particular concern for childhood attendances as parents were taking their child into A&E rather than their GP – causing a much larger strain on the hospital (in terms of cost, and time).
We started to build up the picture, visualising the A&E location, before adding deprivation levels at the Lower Super Output Area (LSOA), and adding asthma prevalence (taken from QOF) onto a map. This gave us some interesting (albeit expected) patterns. We wondered if there were any other factors at play – why are so many people attending A&E in these areas. After further investigation we found that there was a correlation between patient satisfaction of a GP and A&E attendance rates. The GP Patient survey contains a multitude of questions however we wanted to focus on the factors that we believe might explain why people go to A&E rather than their GP practice. We analysed the following questions to find which GP practices had the least satisfaction for patient access to services:
- Overall, how would you describe your experience of your GP surgery?
- Generally, how easy is it to get through to someone at your GP surgery on the phone?
- Overall, how would you describe your experience of making an appointment?
- How satisfied are you with the hours that your GP surgery is open?
- Would you recommend your GP surgery to someone who has just moved to your local area?
- How long after initially contacting the surgery did you actually see or speak to them?
We could then “bin” the these questions to identify where the lowest satisfaction levels in the area:
Explore the map in further detail here
Creating a solution to reduce A&E attendances
There are 88 different languages spoken in Bradford
As well as patient dissatisfaction with their access to GP’s, we were also told that there are often language barriers so we looked at ways to tackle that. Analysing Census data, we found that there are 88 different languages spoken in Bradford – the main languages being English, Panjabi, Urdu, Polish, and Bengali (with Sylheti and Chatgaya).
A simple, low cost solution could be to create a Facebook campaign to target areas within the city.
Facebook has functionality that can not only target specific areas, but can also select languages spoken, age, and many other metrics. If the CCGs were to create content specific to the areas and languages spoken by the population, patients could be better educated in what to do if they, or their children suffer from respiratory conditions. The campaigns are relatively inexpensive, enabling CCGs to control their daily budget. Over time, this strategy could reduce A&E attendances, pushing users to get the treatment they need in the most appropriate location.
Contact us to find out how we can create insights into your data.
Duffy R, Neville R, Staines H (2002). ‘Variance in practice emergency medical admission rates: can it be explained?’ British Journal of General Practice, vol 52, no 474, pp 14–17.
Contains public sector information licensed under the Open Government Licence v3.0:
“NHS RightCare – Commissioning for value” packs, published by NHS England
Monthly Prescription Data, published by NHS Digital
Annual Quality Outcomes Framework data, published by NHS Digital
The GP Patient survey published by NHS England
LSOA boundary shapefiles, published by Office for National Statistics.
National Census data, published by Office of National Statistics