Data quality is a recognised issue within the NHS. Indeed, when talking at the HETT Conference 2018 about the potential benefit that new technologies such as predictive analytics could have on healthcare, Sarah Wilkinson, CEO of NHS Digital, said, “…the greatest challenge is not the technology or the IG, but the quality of data in NHS systems. This needs to be addressed.”

So, the questions arise, why is data quality an issue and how can we fix it?

An audit conducted by leading healthcare intelligence provider CHKS in 2014 stated that reference cost submissions at 33% of trusts audited were materially inaccurate, with ‘at risk’ trusts having the most unreliable submissions. Only 12% of trusts had good-quality costing data across all services. These findings demonstrate the challenges that using this information presents at both local and national levels and highlight the data quality problems many trusts are experiencing.

Since 1982, when the Korner report was published, there has been a commitment to collect a national set of data for management of the NHS. At the time, the president of the Royal College of Physicians (RCP) indicated support for the fact that clinical data from every inpatient episode was to be coded. It was recommended that every clinician should ensure that clinical coding was as accurate as possible.

There has, however, been strong evidence that the majority of clinicians have not engaged with the process. A 2011 HSCIC report, ‘Hospital Episode Statistics (HES): Improving the quality and value of hospital data’, states:

“They have not been concerned about the accuracy of the data, the many ways that it is used, nor have they used the data to support their own clinical practice or service developments. Many have little or no knowledge of the large database Secondary Uses Service (SUS) into which trusts are required to submit data from their Patient Administration Systems (PAS) or of the Hospital Episode Statistics (HES) database which provides a repository of data for secondary uses”.

A further disconnect arises in many trusts because the coding of clinical diagnoses (including comorbidities) and procedures, using the International Classification of Diseases (ICD-10) and the OPCS Classification of Interventions and Procedures (OPCS-4), is often undertaken by specialist clinical coders who have very little contact with frontline clinicians. They work from clinical notes which are neither structured nor standardised and which often prove to be inadequate for coding purposes. This leads to errors and omissions.

Since the onset of PbR, clinicians have become more aware of the financial consequences of inaccurate and incomplete clinical coding, and in a few NHS trusts they have worked closely with clinical coders to maximise income. However, a recent report from the Audit Commission on PbR has shown that in general, coded data is still poorly implemented in clinical practice and that many clinicians remain uninterested.

Operational and system change

Healthcare data migration projects during a period of transformation to a new Electronic Patient Records (EPR) system or PAS are some of the most critical initiatives to perform correctly. This is a time when safeguarding the continuity and integrity of data is crucial. Ensuring the highest levels of patient data quality is obviously key, but delays can often impact the delivery of healthcare services – so a robust data migration delivery strategy is vital.

Although the process of data migration is a complex – and at times a high-risk – activity, support from the right partners and suitable planning for the journey does not need to be as daunting as many would have you believe. Solutions such as the Stalis CareXML data migration service are designed to be highly automated and to minimise user intervention and reduce the burden on scarce resources. The system offers rules-based identification of data discrepancies and, through access to your originating patient record, enables fully validated data correction.

Referral to Treatment (RTT)

Another area where poor data quality has been shown to be prevalent is in the reporting of RTT, associated pathway and waiting list data. Many trusts have been failing to meet national targets in this respect.

According to NHS England’s latest waiting times statistics, the proportion of patients waiting more than 18 weeks to begin their treatment increased to 9.4% in September 2016. This means that there were more than 347,800 patients still waiting to begin their treatment after 18 weeks. More than 1,181 had been waiting for more than a year. In September 2016, the total waiting list increased to 3.7 million. Due to lack of (or under-) reporting, the true figure is estimated to be 3.9 million. This is the highest level since December 2007.

Without a doubt, the NHS is facing extraordinary demands – but in part, the inability to reduce waiting lists and meet RTT targets is caused by poor-quality and incorrect data. This is often due to NHS staff not fully understanding RTT pathways and appropriate codes, and it leads to reduced capacity through duplicate pathways and bookings, missed or delayed appointments and inappropriate prioritisation of patients.

Getting to the root cause

Validating and cleaning up data is all very well, but unless the root causes of the problems are identified and an action plan is put in place to correct the issues, they will just resurface.

Undertaking a root-cause analysis and reviewing the information flow from the data creation point through all intermediate processes is the only way to fully understand what is occurring. This will enable the data analyst to review the rectification options in order to ensure future data integrity and long-term sustainable savings.

Many data quality issues occur due to process failures, so correcting Standard Operating Procedures (SOPs) where the error is introduced is significantly better than updating poor data downstream.

Stalis are long established experts in health and care data and we are passionate about unlocking the value of data in the NHS. To this end we have teamed up with Draper and Dash to co-sponsor this year’s Healthcare Data Assurance Summit.