Relevance to current UK practice

Integrated software infrastructure is still a fairly recent development, but in addition to the potential benefits discussed above (e.g. monitoring of prescribing, safety, quality assurance, and contracting), the availability of integrated information provides for a

more efficient system of targetinginterventions towards the neediest individuals. This

approach potentially benefits recipients, providers and commissioners. Identifying at

risk individuals or groups using health data, including practice based data collected

during routine care, increases the efficiency of this process, because the effectiveness of preventive interventions is generally greatest when aimed at those at highest risk. This specific issue is the focus of this thesis, applied to the areas of cardiovascular and diabetes risk reduction.

From the advent of electronic records, cardiovascular disease data were generally well supported by the Read coding system discussed in the next chapter. For each risk factor, an appropriate Read code is usually available and there is often a selection of alternative codes with similar meanings. For other conditions, where relevant Read codes did not exist, the facility was available for practices to create their own electronic codes. This enabled primary care teams to undertake audits of practice specific activities for which no relevant Read code existed. During the latter half of the

1990s this practice began to be discouraged as the NHS prepared for integration above the practice level. Such ‘home-grown’ codes might be meaningless when retrieved outside the original practice context and so were seen as a barrier to standardisation and data integration. The most recent development in this process has been the Directed Enhanced Service (DES) for Information Management and Technology of 2008, through which practices rationalised their use of such codes. In 2002, a new international coding system SNOMED CT (Systematised Nomenclature of Medicine Clinical Terms) was created through a merger of the Clinical Terms Version 3 (a subgroup of Read codes used in the NHS) with the SNOMED RT (Reference Terminology) system (18), the latter already in use by the College of American Pathologists (19). This expanded system is becoming increasingly adopted into the NHS and is the basis for the new EMIS-Web electronic record system to be discussed at the end of this thesis. It is designed to support the integration of health data at an international level.

Practice based audit activity led on later in the 1990s to Primary Care Organisations (PCOs) carrying out audits remotely and providing comparisons with similar practices in the region, adjusted for demographic confounders such as age and deprivation distributions. This was only possible if appropriate data were standardised across practices in the region under study. Before long, this concept applied to the NHS as a whole.

The focus on cardiovascular disease contrasts with that of other medical conditions including malignancy, whose risk factors or symptom profiles (particularly with regard to electronically coded data) are often still poorly defined at least in such a way that would facilitate early detection (20). Social health variables are poorly recorded in primary care. Even a factor as important for cardiovascular outcomes as ethnicity has only very recently started to be recorded systematically by practices. Risk factor information is often incomplete, and its adequacy for cardiovascular disease will be explored later in this thesis. There are many problems in health care that require

identification and targeting of those at greatest risk, as suggested above. Of these, cardiovascular disease has several advantages as a topic for primary care research.

Firstly, cardiovascular risk variables have benefited above all others in the data standardisation process described above. Prior to the Quality and Outcomes Framework, other initiatives provided incentives (including financial) for the collection of cardiovascular risk factor information. This particularly applied to the ‘Banding’ system of the early 1990s, through which practices would collect data on such variables as smoking status, blood pressure, and cholesterol levels. Different levels of activity (‘Bands’) would attract different levels of payment. This system was later dissolved, but the resulting electronic data were saved in the practice systems, available for future access. The result of this was that by the end of this decade a tradition had become established to promote cardiovascular risk factor recording in coded, standardised form. Secondly, cardiovascular risk is a well researched area and the relative importance of the various risk factors is known quite well, as a result of several large cohort studies and epidemiological surveys to be discussed in Chapter 2. The result of this is that risk algorithms that weight the main independent factors and produce risk estimates, are widely available and in common use among primary care practitioners all over the UK. Targeting patients at raised cardiovascular risk has become an area of intense interest over the past few years, because of the availability of these algorithms and of the opportunity to modify risk through a variety of interventions.

Finally, the study of cardiovascular risk benefits from the fact that the outcomes (cardiovascular events) are to a large extent recorded electronically in the same databases as the risk factor data. The same is not true of outcomes such as hospital admission, which may occur without any coded entry into the practice based record, or of fall risk, where the fall itself might (occasionally) be recorded in primary care but the risk data (such as whether there are stairs at home, whether the patient uses a walking aid) are, if recorded at all, more likely to be found in a database at social

services, at a local occupational therapy provider, or in housing data of the local council.

The importance of cardiovascular risk reduction to current UK practice is reflected in its inclusion in a number of recent guidelines and recommendations by expert bodies. These include the National Framework for Coronary Heart Disease of 2000 (21), the Fourth report of the British Hypertension Society of 2004 (22), the Second report of the Joint British Societies on prevention of cardiovascular disease in clinical practice of 2005 (23), the NICE Technology Appraisal 24: Statins for prevention of cardiovascular events of 2006 (24), the SIGN Clinical Guideline 97: Risk estimation and the prevention of cardiovascular disease of 2007 (25) and the NICE Clinical Guideline 67: Lipid Modification of 2008 (26). All of these documents recommend the identification of individuals for preventive interventions based on estimated cardiovascular risk, generally drawing on the availability of standardised risk factor data in UK general practice.