Best ways to improve data quality

Originally published in GP magazine .

Written for GP magazine around 2002. This is the version I submitted - the published article may have been edited.

Data quality is as hard to define as clinical governance, and to some degree the two overlap. One of the classical computer axioms is “garbage in, garbage out”. Data quality addresses this on a practice-wide and individual level. What is the point of using the computer to record consultations and data if we cannot trust it? Good data quality helps in many ways; not just in consulting, but in enabling easy computer-based audit, simplifying administration, enabling automated PMA reports and making the practice look attractive for clinical trial work, attracting medical students etc.

Data quality is a perfect topic for a practice development plan. It involves all clinical and non-clinical staff. You can have meetings of all staff to discuss the topic generally, and then working parties could be set up from each group and across groups to tackle certain issues. It is easy to set definable goals, and as this is all done on the computer it is then easy to audit them.

Some basic tips are:

As stated, develop a practice-wide approach, not just a project for the doctors. If the nurses are recording blood pressures differently to you, how is a computer-generated PMA form going to pick them up?

Everyone needs to have a basic understanding of Read codes. What they are. What they mean. How to find them quickly. Read code courses are easily available and relatively cheap, and make a good course for staff to go on as part of their professional development. It then becomes easy to explain why everyone should code the same thing the same way, and this makes subsequent data extraction simple.

Don’t go it alone; there is outside help. PRIMIS is an NHS-funded organisation that helps large groups achieve good data quality. They don’t tend to help one practice, but you might be able to join a PRIMIS group or even set one up. Some university research departments have interests in this area. Your local MAAG may be able to offer courses or help. There are independent gurus who have been championing this, who may be able to come and give you some pointers.

Don’t do too much at once, and think things through. Concentrate on doing things correctly and not wasting your time and money. The classic anecdote is the practice that computer-summarised all their paper records but didn’t code their incoming letters, which were filed into notes that had been summarised, thereby invalidating all the work. Similarly, don’t rush things; it may take years to get all records summarised, and different people adapt at different speeds. One idea is to pick a topic like cardiovascular medicine and concentrate on accurately coding these patients, making sure blood pressures, cholesterols, smoking status and aspirin usage are coded the same way each time by everyone, then expand to diabetics, asthmatics etc. We have found a lot of these patients are attending hospital clinics and never come to the surgery, so it makes sense to take extra care to code their letters and extract things like the above to make us look better when we are audited.

Understand why you are doing it. You are doing it to improve your care to patients, to be more efficient and possibly to make more money. If everybody accepts and understands this it will be easier to implement.