Information requested
Please may you give me the information on the current methods you use for detecting and also preventing dental contractor fraud for the NHS?
NHSCFA response
In response to your request, the principal responsibility for the investigation of dental fraud within the NHS lies with NHS England as a primary care issue.
In addition the management of claims and payment systems for dentistry is also part of the NHS England remit, although the NHS Business Services Authority undertake payment processing and provider assurance processes (see NHS Dental Services | NHSBSA, ) as such the majority of preventative measures will be in that area.
The NHSCFA relationship with the sector is defined in the sharing agreement with the General Dental Council:
The NHSCFA do however detect dental contractor fraud through reporting, analysis of data (National Dental Contract data, Personal Dental Services (PDS) contracts, Community Services (CS) contracts) and undertake analysis of the sector as a whole via the Strategic Intelligence Assessment.
There is information on our website about our approaches:
Dental contractor fraud | NHS fraud reference guide | NHS Counter Fraud Authority (cfa.nhs.uk)
Where fraud is found as part of investigations and a weakness is identified, then recommendations are made to the appropriate authorities to amend their processes.
Current methods for detecting fraud are based on a distinctive problem being addressed, the data and its structure / availability and the tool applied alongside the organisation outcome required.
Therefore, currently rule based analysis is the main approach applied, using domain knowledge to add context to any records identified as irregular from the rules applied. It must be noted no method/algorithm can 100% confirm fraud without an investigation taking place.
In addition, our approach will include several strands:
- We undertake proactive data projects using dental contractor data – accessing and analysing external datasets in partnership with the NHSBSA, in order to produce identification of outliers through a variety of analytical techniques.
- The purpose of these exercises is to identify quantifiable outliers within which fraud can be found, leading to investigations and subsequent criminal sanctions and a greater understanding of the impact of fraud prevention activity.
- We have also undertaken some exercises using Machine learning (Unsupervised Learning, K means Clustering) to learn how behaviour has changed alongside how this can be applied in the future to identify further clusters outside of normal behaviour