Data quality and the IMT DES

Originally published in GP magazine .

Written for GP magazine around 2006. This is the version I submitted - the published article may have been edited.

Data accreditation for paper-light practices is the hardest part of the IMT directed enhanced service to achieve. The intention is to create an electronic record and summary that are good enough to be uploaded to the Spine and shared around the NHS. In theory you can ask to be assessed at any time and as often as you like, though resources will limit this. It is a prerequisite that you are signed up to be a paper-light practice and that you are entering all consultations electronically and contemporaneously where appropriate. It is assumed there will be a preparation phase followed by the assessment.

The assessment

This will consist of a practice submission, an electronic audit and a visit by an assessor, usually a clinician, and there is talk they should be able to view notes. The e-audit will use several techniques that have been around for some time, analysing prevalence and coding. A lot of the audit will look for outliers to expected levels. It is my understanding that the audits are being trialled on 50 practices and the rates from these practices will define normal.

Analysing activity

The audit will look to see that appropriate numbers of consultations have been recorded as home visits, telephone and surgery. It will also look at referral rates.

Prevalence

The prevalence of common significant diagnoses will be assessed, and to check for depth of coding the prevalence of minor conditions such as haemorrhoids will also be looked for, as well as the prevalence of allergies or reactions and of family history.

Drug to disease audits

Drug data tends to be accurate, so looking to see if each drug has a corresponding diagnosis is one way to look for quality. There are numerous searches that can be run, e.g. lamotrigine and epilepsy, tamoxifen and breast cancer, finasteride and prostate disease, amiodarone and cardiac diagnoses, etc.

Inconsistencies

Checking for male codes that appear on women and vice versa, and also validating that dates and procedures are sensible, are easy searches to do - e.g. smears on men, or smears post total hysterectomy.

Primis has been engaged to help practices achieve this and to help construct the e-audit. Several areas have already worked with them for some time and will be familiar with these techniques. We are part of the Keele University GP research network who do a lot of this work with practices.

What needs to be done

This is easy - sort out all your existing data and make sure all new data going in is good quality!

Tidy up existing data and seek help where you can. Depending on your system there are numerous tools available, with more coming. I have used MSD Informatics, which has some interesting searches looking for drugs with no appropriate diagnoses. There are numerous searches and audits you can also do yourself. I have no doubt some drug companies will be offering their services.

Training - especially Read coding. A good working knowledge of Read codes and the differences between symptom codes, diagnostic codes and procedural codes is essential for everyone entering data onto your system, not just doctors. Implementing a practice Read code formulary may be useful. You may need to do mini audits of Read code usage and educate people to use better, more appropriate codes. Often small interventions can have an effect without having to name and shame.

Problem coding. Using a problem code with each consultation and minimising the use of free text as the problem will help score some of the points. Templates that facilitate rapid code entry should be used where possible.

Coding letters. The dreaded job of highlighting incoming letters and coding any relevant new diagnoses or procedures needs to be taken seriously. We use Docman, a document management system, and have a specially trained clerk.

Liaise with your PCT and Primis if available. A good PCT should be a useful resource to help you achieve this. They want you to get it.