Disease register validation work - a case study in the words of a facilitator

The problem

The prevalence calculation in QMAS was due to change and this could potentially have an adverse effect on practice payments, particularly for practices with small disease registers. With the removal of the square rooting formula in 2009-10 and the removal of the 5% cut-off due in 2010-11 practices were looking to ensure their registers were as accurate as possible during the 2009-10 QOF year. While practices had always looked to maintain their registers it was felt that some pro-active work was necessary in order for everyone to feel confident that registers were accurate and complete, not least for the good of patient care.

My role

Part of my role as Improvement & Development Manager is to help practices make improvements in primary care which can lead to improved outcomes for the practice and the patient. I have recently been involved in helping to improve immunisation uptake across the city and also achievement of various QOF indicators, so it was natural for me to get involved in working on prevalence. I got involved in producing comparative data, assisting with running the queries, analysing the results and sharing hints and tips.

What we did

We analysed the prevalence rates for each practice and provided comparative graphs based on the last three years data, allowing practices to compare against other local practices. This helped identify where prevalence might be lower than anticipated and also highlighted a fairly wide variance across the city in some cases. We also tied this in with 'expected prevalence' rates for each practice by utilising the NHS Comparators tool. This work gave us a starting point and enabled the PCT and the practices to focus on areas identified as a priority.

We then decided to go back to basics and requested the old RPDQ queries from PRIMIS+ which help to identify patients that are potentially missing from a disease register. These queries look for medication and other codes associated with a particular disease (including CHD, Diabetes and Hypertension) and then pull out those patients who do not have a diagnosis of that disease. In some cases this did indeed identify patients that were missing from the register, whilst in many other cases it highlighted a data quality issue, for example where a wrong code had been entered. We had also used some of the queries from the IM&T DES in this exercise as these also did a similar job (eg. from the Group 1 query: Levothyroxine versus Hypothyroid diagnosis). Practices were also encouraged to use the clinical system functionality, including system searches, to help address prevalence and data quality issues.

As a result of this work some READ code refresher training was provided and this was based on the PRIMIS+ training module. A 'prevalence pack' was also produced and shared with practices. This gave hints and tips on validating registers including sections on the processes for capturing data, removing patients from registers, reviewing prevalence figures and understanding the QOF Business Rules. Elements of the PRIMIS+ modules on Data Accreditation and Electronic patient records were also brought in.

The outcome

Many practices have seen their prevalence rates improve and the variance across the city is now not so apparent for many disease areas. The PCT average is more in line with national figures than before. I believe there is more confidence now that registers are complete and accurate and this has also led to consideration of other more general data quality issues. Whilst there is no hard evidence to support this claim it is our understanding that this on-going work has led to more patients coming in for chronic disease reviews and a reduction in unnecessary exception reporting for QOF where patients have been taken off registers as appropriate. The plan is for interested practices to run the queries again later this year and make use of any new queries that are produced by PRIMIS+.

Steve McInnes.