Memorandum submitted by the Health Statistics
Users Group
EXECUTIVE SUMMARY
1. Background
The Health Statistics Users Group is a member
of the Statistics Users Forum and this submission complements
that submitted by the Forum. In particular, we support strongly
the comments made about the need for legislation to secure the
independence of the statistical service through accountability
to parliament.
2 and 3. The production of official statistics
and the role of the Office for National Statistics (ONS)
In supporting the Statistics Users Forum's reservations,
especially that it is unrealistic for the proposed Board of the
Office for National Statistics to be simultaneously responsible
for the delivery of statistics and for ensuring quality and adherence
to standards. This model is particularly inappropriate for health
and care statistics. In three of the four countries of the United
Kingdom, much of the responsibility for data lies with organisations
within the National Health Service and this leads to different
lines of accountability. In England the chief executive of the
Information Centre for Health and Social Care is directly accountable
to Parliament.
In view of this, the role of the Statistics
Commission should be retained and strengthened.
We agree that the role of the Registrar General
for England and Wales should be separated from that of the National
Statistician, but the role of the General Register Offices in
administering registers should not be distanced from the production
of statistics derived from their registers. In view of proposed
developments in the use of registers to compile the census, we
see an advantage in the censuses remaining in the same organisations
as their corresponding General Register Offices. Changes in the
role of Registrars General will necessitate amendments to the
Population Statistics Acts and the opportunity should be used
for a long overdue update of statistical aspects of this legislation.
4 and 5. Improving quality and integrity
and increasing public trust
Greater independence for the statistical services
should help improve the jntegrity of the data and a more positive
identity should help counter public mistrust, but they will not
do so in the absence of other changes. Even if taken out of ministerial
control, the designation of selected series and datasets as "national
statistics" is unhelpful as it legitimates suboptimal practice
with respect to "management information" and "departmental
research" and the public is unaware of the distinction between
these. The target culture promotes selective use of statistics
as "good news" which gives an unrecognisable picture
of the positive and negative aspects of the health care system.
Adequate resources are needed to improve the full process of compilation
of statistics including planning, data collection, analysis, interpretation
and dissemination as outputs cannot be viewed in isolation from
this.
6. United Kingdom health statistics
To improve the extent to which health and care
statistics can be constructed for the UK as a whole for use both
internally and internationally, ONS should have a stronger role
in co-ordination and facilitating collaboration between the many
agencies involved, as well as in developing the Code of Practice.
7. Examples from other countries
The Nordic countries, especially Finland, are
seen as leaders in good practice in the production of health and
care statistics and the work of ISD Scotland is seen as being
on a par with this.
1. BACKGROUND
1.1 The Health Statistics Users Group was
established to bring together users and producers of official
health statistics in the four countries of the United Kingdom.
Because of the nature and extent of organisations involved in
the production of statistics about health and health care, many
members are both producers of statistics and users of those produced
by others. In addition, organisations working at a local level
in the National Health Service and local government are both users
of national statistics and contributors of data to national systems.
The group also includes members who do not produce statistics
themselves but have been involving guides for other users. 1-3
1.2 The Group welcomes the Government's
consultation on Independence for Statistics4 and the Treasury's
enquiry. We support the overall aim of introducing legislation
to make official statistics more independent. We are affiliated
to the Statistics Users Forum and broadly support the generic
points made in the Forum's memorandum to the Committee. In particular,
we note the lack of provision for taking account of the needs
of users outside government and recommend that this is rectified
when the legislation is drafted. Like other user groups affiliated
to the Statistics Users Forum, the Health Statistics Users Group
is run by a small group of volunteers. Lines of communication
with government and NHS statistics departments are good, but more
resources would improve our ability to reflect and represent the
range of views held by users.
1.3 In proposing arrangements for greater
independence, the consultation document does not take sufficient
account of differences in the ways in which statistics on particular
subjects are produced. This applies particularly to the arrangements
for statistics about health and care. We therefore wish to make
additional comments and recommendations to be read alongside the
response from the Statistics Users Forum.
1.4 We are also responding to the concurrent
consultation on Informing healthier choices: information and
intelligence for healthy populations,5 which sets out the
Department of Health's strategy for public health information.
It is disappointing that there is no cross-referencing between
the two documents and we recommend that better links are made.
Informing healthier choices lists provide information for
the public as one of its three priorities, along with providing
information support for implementing government policy and supporting
the public health workforce. Despite the welcome acknowledgment
in paragraph 4.9 of Independence for Statistics that "statistics
are a public good, serving a wide range of users", the mechanisms
for doing so are not explicitly identified.
2. THE PRODUCTION
OF OFFICIAL
HEALTH STATISTICS
2.1 The ways in which official health statistics
are compiled in the four countries of the United Kingdom diverge
in a number of ways from the model assumed in the document as
well as from each other. This should be taken into account when
discussing the infrastructure to support proposals for independence
and is outlined below.
2.2 Devolution has a long history in this
area and long predates recent legislative changes. For most of
its history, the General Register Office for England and Wales,
established in 1837, has combined the administration of civil
registration with analysis and publication of data derived from
this and the conducting of the population census in England and
Wales. Since 1948, it has been responsible for the NHS Central
Register. The General Register Office for Scotland, established
in 1855, has similar responsibilities, as does the General Register
Office for Northern Ireland, set up in 1922, although the NHS
register is organised differently in Northern Ireland.
2.3 Over time, the General Register Office
for England and Wales developed further areas of health and population
statistics in England and Wales, for example statistics on cancer
registration and congenital anomalies. Some areas of data collection,
for example communicable disease and hospital in-patient statistics
were initially developed within the General Register Office and
subsequently passed on to other agencies. After it became part
of the Office of Population Censuses and Surveys in 1970 and the
Office for National Statistics in 1996, the administration of
registration was separated from primary analysis, that is the
routine production of annual tables. This in turn has been split
from the more exploratory secondary analyses which make fuller
use of the data. In contrast, the General Register Offices for
Scotland and Northern Ireland have retained responsibility for
annual publications, but in Northern Ireland secondary analyses
are undertaken by other parts of the Northern Ireland Statistics
and Research Agency.
2.4 The administration of publicly funded
health care, and the accompanying collection of statistics about
this has developed separately in each of the four countries of
the United Kingdom since the Ministry of Health covering England,
the Welsh Board of Health and the Scottish Board of Health were
set up in 1919 and separate arrangements were made for Northern
Ireland on partition of Ireland in 1922.
2.5 Since the latter half of the 20th century,
statistical activities have moved out of government departments
and into the National Health Service. The precedent was set in
1974 when the Research and Intelligence Unit established in 1965
in the Scottish Home and Health Department became the Information
Services Division (ISD) within the Scottish Health Service, although
continuing to provide support to what is now the Scottish Executive
Health Department.6 ISD has a strong reputation nationally and
internationally for the quality of its data and analyses.
2.6 This model has influenced the formation
within the NHS of Health Solutions Wales, part of the Health of
Wales Information Service and the Information Centre for Health
and Social Care established in England in April 2005. Although
England and Wales have retained an analytical capacity within
the Department of Health and the Welsh Assembly Government, the
responsibility for most primary data collection lies in the NHS
outside direct accountability to ministers and the Information
Centre is a special health authority whose director is directly
responsible to parliament. In addition, in England, some data
collection formerly undertaken by the Department of Health is
now undertaken by other agencies such as the National Patient
Safety Agency and the Healthcare Commission.
2.7 In contrast, in Northern Ireland, the
Information and Analysis Directorate, whose work encompasses both
statistical and economic analyses, is part of the Planning and
Resources Group of the Department of Health, Social Services and
Public Safety.
2.8 The Channel Islands and the Isle of
Man have different health care systems from the rest of the United
Kingdom and have their own arrangements for health care statistics
and also for civil registration. Because of their small size,
their residents make use of some specific and mainly specialist
services in England. These islands therefore take part in some
relevant health information activities, for example the Confidential
Enquiry into Maternal and Child Health.
3. THE ROLE
AND GOVERNANCE
OF THE
OFFICE FOR
NATIONAL STATISTICS
3.1 In supporting the Statistics Users Forum's
views about the proposed role of the Board of the Office for National
Statistics, we would point out that the assumptions on which the
proposals are based conflict with the ways in which health and
care statistics are collected.
3.2 As mentioned above, the Chief Executive
of the Information Centre for Health and Social Care is directly
accountable to parliament. The Information Centre also has a co-ordinating
role with respect to other agencies with a role in health statistics
in England. This cuts across the arrangements proposed in paragraph
4.20 for the Office for National Statistics to take responsibility
for the quality and integrity of national statistics which it
does not itself produce. Although it is appropriate for ONS to
take the lead on updating and maintaining the Code of Practice,
and in consulting users of statistics on changes, it cannot, by
definition, have a monopoly in implementing it.
3.3 We agree that the post of Registrar
General for England and Wales should be separated from that of
head of the Government Statistical Service but we do not think
it should be separated from the analysis and dissemination of
data derived from registration currently undertaken by ONS. It
also should not be separated from the role currently played by
ONS of overseeing the very extensive use of the National Health
Service Central Register for research purposes and the changes
likely to take place in the latter with the implementation of
the National Programme for IT in the NHS in England. In view of
proposals to replace the census with a register-based system,
it is more appropriate for it to remain with the organisation
responsible for registers. Closer links should be made with the
General Register Offices for Scotland and Northern Ireland, given
their responsibilities for vital statistics and censuses.
3.4 Surveys make an important contribution
to information about the health of the population and its use
of health and social care. In general, they are commissioned by
health ministries from survey organisations, including both ONS'
social survey division and private sector organisations, which
in some cases work in partnership with academic departments. Health
surveys are commissioned for each country separately but other
surveys relevant to health or including questions about health
may cover more than one country. For example the General Household
Survey, now incorporated into the Continuous Population Survey,
covers Great Britain and the five-yearly Infant Feeding Survey
covers the whole United Kingdom.
3.5 The diversity of the arrangements for
producing health and care statistics reinforces the view expressed
by the Statistics Users Forum that it is unrealistic for the proposed
Board of the Office for National Statistics to be simultaneously
responsible for the delivery of statistics and for ensuring quality
and adherence to standards. We therefore support the retention
and strengthening of the role of the Statistics Commission.
4. ENSURING THE
QUALITY AND
INTEGRITY OF
NATIONAL STATISTICS
4.1 The integrity of national statistics
is dependent not only on making adequate arrangements to ensure
their independence but also on the availability of adequate resources
to do work of a sufficient standard and develop statistics to
meet the changing needs of society. We are concerned at the recent
cuts in resources available for health statistics within ONS,
compounded by the division of analytical activities between London
and Newport and the loss of skilled staff who have not moved from
London when their posts have been relocated elsewhere. As a result
there seems little scope for new analyses developments in data
dissemination, professional contacts have been lost and the use
of data has been impeded. A parallel loss of skills has arisen
from the relocation of statistical posts in the Information Centre
from London to Leeds. Examples can be provided.
4.2 An important consideration in the production
of health statistics is the need to strike a balance between the
need to protect individuals from identification and the ability
to use data to provide information to inform decisions about public
health and health care. Over the past few years, ONS has implemented
heavy-handed and cumbersome disclosure control measures. These
have restricted both the availability of data, particularly those
relating to health, for further analysis and also the integrity
of some data. The long tradition of ONS by which it increased
its ability for secondary analysis by collaborating with outsiders
has increasingly been replaced by a situation where ONS staff
spend time policing disclosure control measures designed to impede
potential collaborators and outside users from accessing the data.
4.3 The problems arising from disclosure
control are compounded by the outdated provisions of the Population
Statistics Acts. The need for updating these has long been recognised,7
but legislative time has not been made available to do so. As
these Acts will have to be revised to change the role of the Registrars
General, the opportunity should be taken to revise and update
their other provisions at the same time.
4.3 Both Independence for statistics
and Informing healthier choices refer to the need to reduce
the burden of data collection. This should be counterbalanced
considering the extent to which this could be counterbalanced
by feeding back useful analyses and information to people, especially
those in the NHS who provide data. This would also contribute
to improving quality by giving them the opportunity to identify
possible errors in the data.
4.4 Particularly in England, where there
is substantial investment in the National Programme for IT, statisticians
and data analysts have had little involvement in decisions about
how data are collected. These have largely been made by informatics
specialist and IT suppliers. There is now a belated but welcome
move towards greater public health and statistical involvement.
This should be strengthened, particularly in view of the many
problems arising in implementing the National Programme for IT,
despite massive investment.
5. INCREASING
PUBLIC TRUST
IN STATISTICS
5.1 Concerns about ministerial and other
political interference relate not only to release protocols, but
also to the whole series of choices that are made about which
statistics are and are not produced. Added to this, the prevailing
target culture concentrates attention on limited areas of activity
and related statistics and lead to neglect of other areas. As
a result, target-based statistics present a picture at odds with
users' own experience. The agenda can also be restricted by cutting
budgets. When departmental ministers decide the scope of National
statistics and departmental statistics programmes and resources,
this allows scope to restrict the agenda for the statistics to
be collected to those which are likely show the government of
the day in a favourable light and exclude those which are likely
to be unflattering. Even if this does not occur in practice, it
gives rise to the perception that political interference may occur.
We therefore strongly agree that these decisions should not be
taken by departmental ministers.
5.2 We agree with the Statistics Users Forum
that the definition of "national statistics" is unhelpful
and gives a licence for bad behaviour in release/non-release of
"management information" and "departmental research".
The public does not appreciate the difference so all these activities
should abide by the Code of Practice, which should apply not only
to outputs, but to the whole process of collection, analysis,
interpretation and dissemination of official statistics.
5.3 Giving a higher profile to statistical
organisations in government and NHS and a more active use of statistical
press releases will help to reinforce a separate and independent
identity for official statistics and help to reduce mistrust and
counter the selective use of statistics by ministers as "good
news".
5.4 Mistrust in ONS has been fuelled by
a growing perception that ONS mistrusts outsiders. The use of
disclosure control to impede access to data for analysis and exclusion
of outside authors from ONS press conferences has compounded this.
This is a matter to be addressed by the new National Statistician,
rather than a subject for legislation.
5.5 Measures taken for disclosure control
can lead to distrust among specific groups of users. For example,
the medical profession distrusts cause of death data based on
the initial causes of death written on death certificates rather
than the revised causes modified in the light of pathologists'
and coroners' investigations. Another example is local infant
mortality rates which are based on small numbers of events but
have a high policy profile. Published data analysed by registration
year are inaccurate and lead to mistrust and should be replaced
by more accurate data based on years of occurrence.
5.6 The launch of the Information Centre
in England has been overshadowed by issues arising from contracting
out a large tranche of its analytical work to a joint venture
with a private sector company. The Dr Foster affair has clouded
launch of the Information Centre for Health and Social Care in
England. The Information Centre has retained control of data sources
and primary outputs, but the resulting secondary analyses are
not publicly available. This is despite a £20 million investment
of public funds without a tendering process, a matter which has
attracted considerable criticism. There are concerns about the
lack of accountability of the Dr Foster organisation and we shall
attempt to monitor the statistical quality of the work it produces.
Meanwhile we recommend that analyses produced with public funds
should be publicly available and commercial providers should be
required to tender for public funds, as is usually the case with
work commissioned from both private companies and academic organisations.
6. UK HEALTH
STATISTICS
6.1 Because of the ways health and care
statistics have developed separately in the countries of the United
Kingdom documented above, it is difficult to derive statistics
for the United Kingdom as a whole. There has been more harmonisation
of registration and census statistics, as a result of collaboration
between the organisations concerned. Similar collaboration is
needed in respect of health and care statistics, although some
differences are likely to continue as a result of differences
in the ways in which care is organised. The gaps in statistics
for the United Kingdom as a whole pose problems internally, but
also lead to problems in contributing to international organisation
and health monitoring activities in Europe.
6.2 The Office for National Statistics has
recently produced a second volume of United Kingdom Health statistics8.
This has been useful in drawing together compatible data where
possible but has also illustrated the problems involved and the
patchy coverage. We recommend that the ONS' co-ordinating role
be continued and expanded.
7. EXAMPLES FROM
OTHER COUNTRIES
7.1 In many other countries, health and
care statistics are collected outside the central statistical
office. Even in Canada, where Statistics Canada is very centralised,
Health Canada collects a considerable amount of data.
7.2 There are a number of examples of good
practice, particularly in the Nordic countries which have a strong
tradition of compiling good quality health data through the use
of linked registers, while maintaining confidentiality. STAKES
in Finland could be cited as a particular example of good practice.
In this context, ISD Scotland is well respected internationally
and has been considered as an "honorary" Nordic country.
REFERENCES
1 Leadbeter D, Rigby M, eds. Harnessing official
statistics. Abingdon: Radcliffe Press. 2000.
2 Macfarlane AJ, Mugford M. Birth counts:
statistics of pregnancy and childbirth. Volume 1, Text. Second
edition. London: The Stationery Office, 2000.
3 Radical Statistics Health Group. Official health
statistics: an unofficial guide. Kerrison S, Macfarlane AJ, eds.
London: Edwin Arnold. 2000.
4 HM Treasury. Independence for statistics: a
consultation document. London: TSO, 2006.
5 Department of Health. Informing healthier choices:
Information and intelligence for healthy populations: London:
Department of Health, 2006.
6 Bryden JS. Posterity planting. How ISD has
been nurtured over 40 years. Edinburgh: ISD Scotland, 2005.
7 Office of Population Censuses and Surveys.
Registration: proposals for change. Cm 939. London: HMSO,
1990.
8 Office for National Statistics. United Kingdom
Health Statistics. Series UKHS No. 2. Newport: Office for National
Statistics, 2006.
May 2006
| 
