Select Committee on Science and Technology Fourth Report

CHAPTER 6: Role of the NHS

6.1 Against the background of the previous Chapter's consideration of general questions concerning data availability, quality and linkage, we examine in this Chapter the development of a national capacity in human genetic databases from present NHS records and systems.

The vision

6.2 As previously noted, a key way of understanding the interaction between genetic make-up and environment is to track large numbers of individuals over long periods of time, note their disease and other outcomes (or the lack of them), and look for correlations between those health outcomes and those individuals' genetic make-up and other life circumstances.

6.3 As indicated in paragraph 2.5, we were impressed by the vision of Dr George Poste (then of SmithKline Beecham) about the scientific and medical advances possible from using the wealth of patient information in the NHS coupled with genetic data. He saw several advantages in a population-based approach to the collection of health and genetic information through the delivery of routine health care[38]. These included:

(a)  large numbers, with minimal selection bias;

(b)  representative, community-based cohorts; and

(c)  standardisation of approach.

6.4 Dr Poste also identified a number of benefits that would accrue to the NHS (and the United Kingdom as a whole) from such an approach, including:

(a)  progress in understanding disease at the public health level;

(b)  provision of new resources to support NHS research and development;

(c)  stimulating production of novel therapeutics, diagnostics and the better targeting of treatment; and

(d)  attracting inward investment by companies.

6.5 This vision suggested that the United Kingdom was perhaps uniquely positioned to contribute to this kind of human genetic database research, since:

(a)  the NHS provided near complete coverage of the population;

(b)  it had high levels of clinical expertise and the necessary infrastructure through primary, secondary and tertiary care to obtain data on individual characteristics and DNA;

(c)  in comparison with some other countries (Iceland, Estonia, Denmark) the population is large and genetically diverse; and

(d)  in theory at least, tracking of the health of the population through health service records, linked to information on genetic make-up and lifestyle, should be achievable, since the NHS provides a national system for the management and delivery of health care, and for holding and handling clinical data.

6.6 Dr Poste recognised that his picture of the future NHS was some way from present reality, a view that chimes with the concerns expressed elsewhere in this Report. He particularly noted the need to build a large-scale computational infrastructure that went beyond what the NHS currently had (which related mainly to clinical care and governance) and would also go beyond what researchers were building with other genomic databases.

Current position


6.7 As noted in paragraph 5.15, there were real problems of data quality in clinical records, especially in general practice, where even basic data such as postcodes were often wrongly recorded.

6.8 In its written evidence, the Department of Health observed that the 1998 Information for Health[39] strategy document had noted that the current inability to aggregate data in consistent and comparable ways made for difficulty in the use of patient-based information at clinical, operational and health authority levels (p 32).

6.9 In a supplementary memorandum, the Department noted that data quality varied between Trusts and between different health data sets. This was particularly a problem with Hospital Episode Statistics data. Specific knowledge and expertise were needed to analyse these data sets properly, even to the extent of excluding some data where necessary (P 11).


6.10 As noted in paragraph 5.12 - and in contrast to the position in England and Wales - the Health Service in Scotland (covering 5 million people) had developed a system of data linkage between the various data sets for which it had responsibility. This had proved to be an invaluable resource both for clinical audit purposes and health service management as well as for epidemiological research.

6.11 At our Imperial College seminar (see Appendix 4), we heard from Dr Aylin, Professor Purves, Professor Forman and others about the great difficulties and uncertainties in linking different data sets within the NHS in England and Wales. Some of the problems were:

(a)  lack of a common computer platform and standard software in the NHS, so that often the different systems could not 'talk' to each other - for example, data from the maternity system were often not downloaded into the child health system but had to be copied by hand from one computerised database to the other[40];

(b)  absence of a unique identifier (such as the NHS number[41]) common to all records[42], with allocation at birth[43] in order to capture all relevant data[44];

(c)  lack of an agreed protocol within the NHS with which to carry out data linkage[45];

(d)  absence of a standard coding system to capture clinical data effectively, especially in general practice; and

(e)  the fact that textual information was very difficult to deal with from a computing and database perspective.

6.12 Aware of such difficulties, Sir George Radda and Dr Dexter told us (as noted in paragraph 5.16) that the UK Population Biomedical Collection would develop its own system of follow-up (mainly through general practice). This might act as a model for future developments in the health service (Q 77). Sir John Pattison also took the view that the MRC/Wellcome study would be a useful test-bed for developing the necessary systems within the health service to be able to link genetic information to health service patient data (Q 39).

6.13 Nonetheless, the NHS does have a long history of successfully carrying out long-term epidemiological follow up of individuals through health, survey and census records. Examples (among many possible) include:

(a)  the Office for National Statistics Longitudinal Study linking census data with death, cancer registration and other records on a 1 per cent sample of the population[46];

(b)  the seminal work by Professors Sir Richard Doll and Sir Richard Peto that correlated the smoking and drinking habits of male British doctors with their mortality from lung cancer, heart disease and other causes[47]; and

(c)  the use of cancer registries for a range of clinical, genetic and epidemiological research, as noted in paragraph 5.13.

6.14 Data collected on the 500,000 people to be included in the proposed national cohort study could potentially be linked in the same way with data on mortality and cancer incidence. However, in the absence of specific registers, data for other clinical outcomes, such as heart disease, diabetes, stroke and other common conditions could not currently be captured unless they resulted in a fatality.

Future challenges and prospects


6.15 Dr Drury of the Department of Health noted that Information for Health had made it very clear that, in the years up to 2005, the NHS needed to undertake a major programme of co-ordinated work focussing on good quality information to support the delivery of care. This work was being taken forward by the NHS Information Authority (Q 1).

6.16 Sir John Pattison agreed that the systems available in the NHS had been disappointing in the past, but thought that Information for Health provided a coherent strategy for improvement with clear targets. He had no evidence that those targets were not being met (Q 19). He agreed that there was a major issue about the completeness and accuracy of records, but said that use of a data dictionary and manual would go some way towards obtaining the necessary standardisation, while local (and currently unmet) needs would provide a push for improved quality (Q 35). There was also an NHS Strategic Tracing Service which allowed the central checking of certain administrative (non-clinical) patient information for accuracy and completeness (Q 35).


6.17 In oral (Q 19) and supplementary written evidence (P 11), the Department of Health told us that 35 per cent of Acute Trusts would have electronic patient records by 2002, with full coverage by 2005 - although Sir John Pattison was circumspect as to whether or not this target could be achieved (Q 27). To facilitate record linkage, the NHS number (see paragraph 6.11(b)) would be used as an identifier throughout the health service (QQ 32 & 33).

6.18 It was, however, clear that the current strategy would not yield a single national system. Trusts would still be in a position to develop their own systems. To provide an electronic health record which facilitated follow-up (longitudinal) studies, data would need to be extracted from the primary file and then held for data analysis in secondary files (for example, hospital episode statistics) in standard form (QQ 20-25, 27, 28 & 31).

6.19 Sir George Radda thought that it would be highly desirable to have electronic health information available in the NHS (Q 85). Professor Bell felt it would take ten years to achieve the much-needed coherent and systematic approach to computerised record keeping within the NHS (QQ 358 & 359). Sir John Pattison thought it could be useful to have a 'demonstrator' site for developing the interface between the electronic patient record and genetic data (Q 29).


6.20 For the purposes of clinical audit, governance, health service management and research, it is important to develop and maintain high quality disease registers (with very high - and, desirably, total - population coverage) to facilitate epidemiological follow up. In its supplementary memorandum, the Department of Health said that, following on from recommendations in the White Paper, Saving Lives: our Healthier Nation[48], developing such registers was a priority for the health service.

6.21 Professor Forman informed us at our Imperial College seminar (see Appendix 4) that the whole cancer registry system would collapse if current GMC guidelines requiring informed consent and/or anonymisation prior to registration were to be enforced (see paragraphs 7.29 ff). In this regard, the UK Association of Cancer Registries[49] saw Clause 67 of the Health and Social Care Bill introduced to the House of Lords[50] as our Inquiry was drawing to a close as crucial in proposing special provision for disease registers to function without the need for specific consent from the patient.


6.22 Because of its almost universal population coverage and the potential to follow the health of the population over time through capture of routinely collected clinical data, the NHS is potentially a very valuable source of data for medical research that, in time, would feed back into improved health for the nation. However, this potential will not be realised unless there is considerable investment in systems to collect standardised and comparable data on clinical history, consultations and investigations wherever they occur in the health service (including general practice), and to allow linkage of clinical data across different data sets.

6.23 In any case, a robust and coherent information strategy for the NHS is essential for the proper functioning of the health service. We are unconvinced that the current strategy to develop an electronic patient record nationally by 2005 will deliver on time, nor are we convinced that a workable system will result. We recommend that the Government should review the strategy for instituting electronic patient records throughout the NHS, to include clinical information contained in GP, hospital and other health records. Delivering a fully functioning national system by 2005 will require firmness of purpose to drive forward the development of robust and standardised systems. This must be supported by appropriate funding, including proper investment in the NHS skills base.

6.24 For this information to be useful for human genetic database research, the fullest feasible data on the incidence and prevalence of disease are essential. Disease registers play a vital part here. We fully endorse the intention in Saving Lives: our Healthier Nation to extend and strengthen disease registers, and recommend that the Government should give this high priority.

6.25 We were concerned to learn that NHS Trusts continue to have autonomy to decide on their own computing strategy and protocols, making for unnecessary complications in the essential task of extracting and linking data across Trusts and across different data sets. To facilitate proper communication and data linkage throughout the NHS, we recommend that the Government should urgently develop and implement a unified information system specifying and requiring adherence to completely compatible common standards. This must be backed with sufficient resources. Without such standards and the necessary resources, projects relying on large-scale use of NHS data will not succeed.

6.26 Effective data linkage can be achieved only if individual patients can be uniquely identified. That has been the position over many years in Scotland. The NHS in England and Wales needs to follow suit. Accordingly, we recommend that the Government should urgently make use of the NHS number mandatory as a common identifier. It should appear on all health records in the health service, including death certification. The NHS number must be assigned at birth (or on arrival in the United Kingdom for those born elsewhere). Furthermore, we recommend that the Government should give appropriate publicity to this change in practice, ensuring that patients not only know their NHS numbers but also understand why it is important to use them in their interactions with the health service.

38   See page 15 of Meeting with Health and Science Ministers, 2nd Report Session 1999-2000, HL Paper 11. Back

39   Subtitled An information strategy for the modern NHS, 1998-2005, Department of Health, September 1998. Back

40   Chapple J, Golightly S, Charles Z. "Linking data for better health in pregnancy and childhood." Proceedings of a conference held at The King's Fund, Cavendish Square, London, January 12-13, 1999. London: CASPE, 2000. Back

41   Different from and with wider coverage than the National Insurance number, which would not be suitable for this purpose. Back

42   A February 2001 briefing statement from the UK Association of Cancer Registries indicated that the NHS number was currently used in only 50-60 per cent of NHS records and was not supplied at all on death certificates (although this was currently under consideration by the Home Office Review of Death Certification). Moreover, the wrong number was occasionally recorded. Back

43   Or as soon as possible for those without an NHS number - or on arrival for those born outside the UK - giving effectively complete population coverage. Back

44   Chapple J, Golightly S, Charles Z. "Linking data for better health in pregnancy and childhood." Proceedings of a conference held at The King's Fund, Cavendish Square, London, January 12-13, 1999. London: CASPE, 2000Back

45   Professor Forman told us (see Appendix 4) that different algorithms for data linkage, developed in-house and of variable quality, were being used by the different cancer registries. Back

46   Hattersley L, Creeser R. Longitudinal Study 1971-1001. History, organisation and quality of data. London: HMSO, Office of Population, Censuses and Surveys, Series LS No. 7, 1995. Back

47   Doll R, Peto R, Wheatley K, Gray R, Sutherland I. "Mortality in relation to smoking: 40 years' observations on male British doctors." Brit Med J 1994; 309: 901-911. Doll R, Peto R, Hall E, Wheatley K, Gray R. "Mortality in relation to alcohol: 13 years' observations on male British doctors." Brit Med J 1994; 309: 911-918. Back

48   Cm 4386, July 1999. Back

49   In a February 2001 briefing statement. Back

50   Clause 59 in the Bill as originally introduced to Parliament. Back

previous page contents next page

House of Lords home page Parliament home page House of Commons home page search page enquiries index

© Parliamentary copyright 2001