NHS Business Services Authority (NHSBSA) Data Analytics Programme Rob Bain, Data Analytics Manager
NHSBSA Portfolio The NHS Business Services Authority is a Special Health Authority and an Arms Length Body of the Department of Health which provides a range of critical central services to NHS organisations, NHS contractors, patients and the public.
NHS allocations 2013/14 Source: Department of Health corporate plan 2013 to 2014
The Data Analytics Learning Lab (DALL): Purpose To pilot a different approach to analytics removing existing data access constraints, enabling different data sets from across our business streams to be brought together to enhance our understanding, providing a powerful environment for analysing transaction data rather than aggregated data To proactively seek and test business improvement opportunities for the NHSBSA and its clients such as: Insight Generation Performance, benchmarking and reporting development Estimating levels of fraud & error/ Finding the perpetrators of fraud, error and debt Understanding service behaviour & characteristics Supporting audit activities Understanding and improving Data Quality What if modelling Optimisations Impact analytics Demonstrating compliance
DALL Virtual Team
2015/16 Purpose and Strategic Goals
DALL Timeline
Examples of Completed projects
Dental Provision of Inlays Product: Insight Report Customer: Dental Services National rate has decreased by a fifth from 10 to 8 (inclusion of inlay treatment per 100 Band 3 FP17s) over the six years from 2009/10 to 2014/15. Inlay provision in London has fallen by 27 per cent over the same period. Reduction across London means that 6 million available for other treatment in 2014/15 compared with 2009/10 Further 8 million a year would be available for other treatment if inlay rates in London matched those outside the capital. Trend analysis 2009/10 to 2014/15
Dental Provision of Inlays In 2014/15, 10 per cent of contracts provided half of all inlays Around a third of all contracts (34.4 per cent) provided 90 per cent of all inlays Outlier Contracts In contrast a similar proportion of contracts (35.4 per cent) did not provide any inlays. It was estimated that outlier contracts provided an excess equivalent to 34,800 courses of Band 3 treatment nominally equivalent to 10.5 million in 2014/15. Around 100 contracts were identified with unusually high volumes of inlay treatments but where there was an unusually low radiograph rate.
Dental Provision of Inlays 2014/15 Regional analysis shows that rates can be reduced even with a lower starting point: Midlands and East of England was below the national average in 2009/10 but its rate decreased in step with the national trend falling some two percentage points by 2014/15. Map of the inlay rate across the 351 English local authorities.
Polypharmacy the use of multiple medications by a patient Product: Insight Report Customer: NHS England Medicines Optimisation General insight: Demographic picture of Polypharmacy in the UK (within electronic prescribing system). Specific study of patients receiving ten or more distinct items per calendar month, including: Age spread. Prescription payment exemption reason. Net Ingredient Cost associated with these patients. Percentage of patients in each CCG who were dispensed ten or more items in March 2015. Central, South and North Manchester are the 1 st 2 nd and 4 th worst offenders, respectively.
Polypharmacy Number of patients receiving one to twenty distinct items per month in March 2015. Seven percent of patients receive ten or more distinct items per month. Age demographics of patients on 10+ distinct items per month in EPS system, for the whole of 2014. This is before normalising for underlying patient age distribution, which moves the peak to the right, reflecting that the elderly population are more likely to consume many medicinal products concurrently.
Polypharmacy Specific insights: Identification of commonly co-occurring medicinal items (Market Basket Analysis). In particular, highest co-occurring items are multiple doses of Warfarin. Food items frequently occur in multiple varieties together. Identification of co-occurrences of medicines with potentially harmful interactions. Patient-level prescription data in electronic system allows us to detect potentially harmful combinations of drugs being dispensed to patients (shown right, March 2015). With clinical input, this analysis could be repeated for many potentially harmful combinations of drugs, and dangerous prescribing practices identified. Potentially harmful NSAID combinations in March 2015
Prescribing for cancer patients Why? Improve Public Health England understanding of a cancer patient s journey and interactions with the NHS which in turn will improve the information and care provided to patients. What? Prescription data transferred from an NHSBSA dataset covering data processed by the NHSBSA between the 1st February 2014 and the 30th April 2014 to the National Cancer Registration Service There are 9,890,226 rows of data in the pilot prescription dataset. If linking on Patient ID, there are 2,229,236 rows of prescriptions data which link to a cancer patient. This represents 22.5% of the total prescription dataset. If linking by tumour, there are 2,587,474 matches (26.1%) with the prescription data set. This is likely to be due to patients who have more than one tumour. How? By Encrypted External Hard drive
Prescribing for cancer patients continued Next Steps? Agreement for regular transfer of the prescriptions data, using the method trialled in the pilot, ideally to occur once per quarter. Agreement on methods of feeding back analysis to the NHSBSA and any publication of future findings if regular data transfer is agreed. The prescriptions data is modified to include drug code. It is discussed whether more information about the non-cancer patients may be available under a new data sharing agreement, for example the age/gender of non-cancer patients to enable standardised rates to be calculated.
Provider Management Purpose To aid Area Teams in managing the dispensing contractors contract by identifying areas for further investigation regarding Medicine Use Reviews, New Medicine Service, Out of Pocket Expenses and Staff Hours. Requirement A toolbox that visually shows cumulative outlier activity e.g. MUR activity.
Potential Savings FY2015/16 The DALL has a target of highlighting at least 200 million of potential savings for the NHS FY2015/16 The thermometer below shows what we have achieved so far towards that target. Potential Savings 138m 0.0m 20.0m 40.0m 60.0m 80.0m 100.0m 120.0m 140.0m 160.0m 180.0m 200.0m
Business Requirements The Data Lab operating model must be flexible to respond to the needs of the customer. Tracker created to record and track new and existing initiatives Scoring mechanism developed to ensure the team concentrate on the vital few Lab coordinator ensures initiatives are assigned appropriately and are meeting agreed deadlines
What have we learned Keep Information Governance and Information Technology onside Free the prisoners Manage expectations Deliver some benefits early on Get various people from across the business involved Take all the help you can get from the experts Data Governance/Data Quality
Positives so far The solution is fast queries are returned in seconds. The team has had no issues learning to use the software and are developing new skills that will help the business in the future. There has been no impact on the network or existing NHSBSA Systems Structured data loading has been relatively easy from Oracle and MS SQL platforms. Unstructured data can now be loaded to Endeca easily to allow quick initial investigations and analysis to take place. 24/7 access to the system from Newcastle and Eastbourne offices. Access from other sites possible in theory (not tested). Analysis of data from across the business is undertaken by qualified staff within a quality controlled environment.
Data Quality The Challenge We did not have a way to measure the quality of our data and ensure standards were raised The Solution Data Asset Registers are being completed by the Data Quality Leads across the business. Data Dictionaries will be completed for all items of data across the business. These will help the DALL team and the business streams understand our data in the future.
Big data is not about the data, the real value is in the analytics!