Vol. 18, No. 1, pp. 59-66
Printed in Great Britain
Journal of Public Health Medicine
'Inaccuracy' in death certification - where
are we now?
G. Maudsley and E. M. I. Williams
Abstract
Background This review aims to document and analyse
aspects of death certification that are relevant to public
health.
Methods A literature review on death certification primarily used the computerized Index Medicus (1981 to mid1995), and concentrated on completing death certificates,
accuracy, standards, education and procedural requirements. Further sentinel publications pre-dating this were
identified from the main literature base.
Results The uses of mortality data, historical and
procedural context for recording death, the philosophy
of Underlying Cause of Death and its relationship to 'the
truth', the extent and impact of 'inaccuracy', the
certificate and the certifier, and possible ways forward
are discussed. It is argued that the question 'How
inaccurate are cause of death data?' is harder to answer
than the literature suggests. Deriving a useful estimate is
difficult because of inter-study differences in (1) definition, measurement (how and by whom?) and practical
importance of error, and standards used; (2) focus (e.g.
death certificate or mortality data), observing everyday
practice or simulation exercises, diagnostic and/or
semantic issues.
Conclusion The traditional perspective on improving the
quality of death certification has not worked. There is a
need for reorientated thinking rather than just urging
more education. Evidence-based educational interventions are needed. The flaws in the theoretical framework
of cause of death and the routine nature of death
certification are unavoidable, but require consideration.
Certifiers need practical feedback mechanisms, integral to
continuing quality assurance at all levels and fostering an
understanding of the construction of mortality data.
Continued development should be a core public health
medicine role.
Keywords: medical education,
Underlying Cause of Death
semantics,
standards.
suggests, and that a change of perspective is required to
improve standards.
Although invaluable for many public health-related
activities (Fig. 1), mortality data are often maligned
('available but rarely sufficient', 'the most complete, but
possibly the least specific'1 outcome measure).
nMdoriBS The Health of tin Nation* taifctl
writing pnbiic health —"*^*< rcpoctideckfiaf public beatUi policy
flUx^u DOOQB S
cui
ealth priorities
allocating i
n r s s n g health aervice quality
participating iocUnktl •"«**
improving case asxnainmnB and allowing tarvhi) tmijtit
m
FIGURE 1 Overlapping areas of public health-related activity
informed by mortality data.
When infectious diseases were rife and rapidly fatal,
mortality data were a better 'window' on morbidity3
but, in contemporary usage, the diversity of applications should not be underestimated. The death
certificate provides, for example:
(1) important personal information for relatives;
(2) a 'safety net' for cancer registration (Fig. 1),
which is an increasingly valuable information source
for purchasers4 and providers given the cancer service
reorganization;5
Background
This review aims to document and analyse aspects of
death certification that are relevant to public health. It
is argued that the question 'How inaccurate are cause
of death data?' is harder to answer than the literature
Department of Public Health, University of Liverpool, Whelan
Building, P.O. Box 147, Liverpool L69 3BX.
G. MAUDSLEY, Lecturer in Public Health Medicine
E. M. I. WILLIAMS, Senior Lecturer in Public Health Medicine
Address correspondence to Dr Maudsley.
© Oxford University Press 1996
60
JOURNAL OF PUBLIC HEALTH MEDICINE
(3) data on avoidable mortality that, expertly
interpreted, remain worthwhile indicators of health
service performance - a warning signal for health care
deficiency.6 Interpretation is crucial, however; hence
the controversy over league tables.7
As the National Health Service strives for knowledge-based decision-making,8 the integrity of the
information underpinning policy should be questioned.
Death certificate diagnoses need to be complete,
accurate and current.9 Inaccuracies will ultimately
affect health-related decisions concerning both individuals and populations, but the numerous estimates
available are difficult to reconcile. Non-standard
approaches mean that subtly different aspects of
'inaccuracy' are studied, the impact of which is often
discussed but not quantified.
Setting the context
History of death certification
Death certification was first introduced, nationally, in
England and Wales in 1837. The Registrar General was
created, and death registration became compulsory to
prove death legally and improve mortality data.10 The
complex legislation concerning certification, registration and disposal of the dead was reviewed in the
Brodrick report (1971):10
(1) Originally, anybody knowing the deceased
could declare cause of death, but by 1845 the
Registrar General was dispatching forms to registered
doctors.
(2) By 1874 penalties operated for not registering
death, and only registered medical practitioners could
officially certify death. 'Violent' or 'suspicious' deaths
(and from 1885, those of 'unknown cause' or 'sudden
and uncertified') were to be referred by the Registrar of
Births and Deaths to the coroner. If no inquest was held
and an official certificate was unavailable, death
registration used the best available information.
(3) Certification by unregistered medical practitioners, unqualified midwives and chemists caused
continuing public concern about hidden deaths from
'poison, violence or criminal neglect'. It was not,
however, until administrative then legislative changes
in 1914 and 1926, respectively, that the coroner referral
system finally dealt with deaths not certified by an
attendant registered medical practitioner.
Recording death
Although the doctor attending the deceased during the
last illness is legally required to certify cause of death,
there is no specific statutory requirement for the cause
of every death to be medically certified.10 Ultimately,
however, very few deaths remain uncertified (007 per
cent in 1988-199211).
In a Belfast study of everyday conversations concerning the last days of deceased people, cause of death
was high on the agenda and reduced to its physical
origins.12 As for elsewhere in Western Europe,'. . .one
dies from one's diseases rather than say of old age,
malfeasance or misfortune'. Of interest, certifying
death from old age is allowable in this country, but
not in the United States, prompting the confession: 'I'm
convinced that plenty of people do die of old age.
Whatever scientific diagnoses I have been scribbling on
my state's death certificates to satisfy the Bureau of
Vital Statistics, I know better.'13
The 'death certificate'
The term 'death certificate' is not statutorily defined
and can mean the certificate from the doctor or
coroner, and the copy of the death register entry for
the 'qualified informant' - usually a relative.10 Here, it
refers to the document from the certifier. In England
and Wales, besides coroner's certificates, the certificates
to be completed by registered medical practitioners are
the Medical Certificate of Cause of Death, Form 66
(Med A); the Medical Certificate of Cause of Death of
live-born children dying within the first 28 days of life,
Form 65 (Med B); and the Certificate of still-birth,
Form 34 (SB) - this can also be completed by a
registered midwife. The most common certificate, Form
66, is discussed here.
Underlying Cause of Death
The Cause of Death statement is in the internationally
agreed format14 based on the philosophically enigmatic
concept of Underlying Cause of Death 15~17 (consideration of which revisits nineteenth-century controversies
about classifying cause of death18). Defined pragmatically as the entity initiating the causal chain leading to
death (i.e. a single-cause basis), questions arise such as:
'What about the multifactorial nature of disease?
Where does the causal sequence start (e.g. lung cancer
or smoking19)?'
The death certificate originally invited a single entry,
and even though multiple entries were commonplace,
only one would be coded. An international standard
Cause of Death statement was then introduced that
allowed multiple entries, but encouraged the certifier to
identify Underlying Cause of Death in the sequence.13
Strict international rules for 'single-cause coding'20
were agreed in 1948:14
(1) the General Rule selects the lowermost entry in
Part I as the Underlying Cause of Death;
'INACCURACY' IN DEATH CERTIFICATION
(2) supplementary rules attempt to retrieve the
probable Underlying Cause of Death from incorrectly
constructed Cause of Death statements.
'Multiple-cause coding' would lose less information from
the Cause of Death statement, but could be cumbersome
to analyse and still open to misinterpretation.
Official procedure
As well as being required to certify cause of death, the
doctor attending the deceased during the last illness has
a common law duty (but no more than 'any [other]
person about the deceased') to refer deaths requiring an
inquest to the coroner.10 The coroner has a statutory
duty to enquire into violent or unnatural deaths,
sudden deaths of unknown cause, deaths in prison
and other deaths as required by certain Acts. Nevertheless, legally, even if the doctor refers a death to the
coroner, this does not remove the duty to complete and
issue a certificate.21 Often, however, normal death
certification is omitted in coroner's cases to avoid
unnecessary paperwork and distress for relatives, thus
exposing a tension between legal requirements and
custom and practice.
Most coroner's referrals will undergo autopsy and
certification of death via Form 99 or Part B of Form
100 ('Pink form B') depending on whether or not an
inquest is held.10 Of certified d e a t h s ^ 99 per cent of
total), doctors certify approximately three-quarters (the
coroner certifies the remainder).11
The Registrars of Births and Deaths, appointed by
local authorities, register deaths as a public record, and
issue disposal orders. A draft death return (Form 310)
is coded by the Office of Population Censuses and
Surveys (OPCS). Traditionally, this was copied weekly
to the relevant district health authority for information
on district residents ahead of official publications.
From 1995, this record has been available monthly
in computerized format (including deaths of nonresidentsregisteredlocally) (letter to District Directors
of Public Health from OPCS, April 1995). Increasing
accessibility might increase data use and the pressure
for quality improvements.
Public and statistical records
The qualified informant supplies information for the
public record of death. The only item that the Registrar
of Births and Deaths should normally transcribe from
the death certificate is cause of death. Subsequent
revision of the Cause of Death statement, using further
information requested from the certifier (e.g. from
autopsy), alters the statistical record but not the public
record. Some details, e.g. marital status, are published
only in aggregate, anonymized form.22
61
Automated cause of death coding was introduced in
1993.23 The non-medical OPCS staff who previously
coded the data manually (with a 1 per cent error rate24),
now code those Cause of Death statements rejected by
the computer program - initially 25 per cent, predicted
to be 10 per cent by 1996.23 Coding error is likely to be
greater with complex statements (because diagnostic
disagreement increases25), and greater with badly
constructed statements (because manual coding and
supplementary coding rules are used).
Establishing cause of death
The inevitability, unambiguous nature and statutory
recording of death probably explain the pre-eminence
of mortality data in public health investigations.
Validity is, however, often questioned:22
(1) A weakness is that although death is factual,
cause of death involves opinion. Different diagnostic
acumen, opinion and style affect its wording considerably. A minimalist style, although legal, is epidemiologically unhelpful.26 Also, autopsy information does
not remove inter-observer variability and the less
studied intra-observer variability in death certification.
(2) It is also possible that quality suffers because
death certification is perceived to be unglamorous
routine paperwork, a 'burdensome task',16 of low
priority.
Diagnosis versus wording
The doctor should certify death 'to the best of his
knowledge and belief.27 Accuracy of cause of death is
influenced by diagnostic and 'semantic' considerations,
themselves influenced by various characteristics of the
certifier, the certificate, the deceased and the 'true'
cause of death.28 How accuracy of coded Underlying
Cause of Death relates to the actual or 'true' cause of
death is affected by the deductive and recording
processes (Fig. 2).
Declining autopsy rates cause concern29 mainly
because diagnostic information is lost. The autopsy is
allegedly undervalued because it is retrospective and
appears primarily educational.30 Consequently, the
pathologist's capacity to influence Cause of Death
statements, besides providing ante-mortem diagnostic
information, is reduced but potentially includes:
(1) 'Consent' or 'hospital' autopsy. As this autopsy is
for investigating the extent of causal pathology (those
of'unknown cause' requiring referral to the coroner), it
particularly contributes to the further information
offered to OPCS by certifiers. Annually, OPCS follows
up 3 per cent of deaths for such information.22
JOURNAL OF PUBLIC HEALTH MEDICINE
62
instructions ('[containing] a wealth of information'21)
are often ignored by certifiers,36 and international
material14'37 might be inaccessible. The Brodrick
report10 recommended that a doctor should certify
death only if, with no indications for referral to the
coroner, 'he is confident on reasonable grounds that he
can certify the medical cause of death with accuracy
and precision'. This proposed change from 'to the best
of his knowledge and belief27 was not implemented.
The intention had been to encourage appropriate
referral to the coroner, but resources were unavailable
for increased referrals.38
Joint recommendations, in 1982, from the Royal
College of Physicians and the Royal College of
Pathologists stated that:39
DEATH
'TRUE' Cause
of Death
> I
Decided upon using
information available:
'ObUeal'*
•W-
clinicopalhological correlaUon
'Pathotogieal' after autopsy'
DIAGNOSIS
of cause of death
±.
SEMANTICS (worting)
of came of death
(1) provisionally registered house officers should not
usually complete death certificates (also recommended
in the Brodrick report10);
(2) undergraduate and postgraduate education
should be provided for hospital doctors and general
practitioners.
Certified
cause of death
AL.
Coded
Underiymg Cauie of Death
(for mortality data)
I
FIGURE 2 Relating the 'true cause of death' to the Underlying Cause of Death recorded in the mortality data.
'Including special investigations. "Including macroscopic
and microscopic findings and special investigations.
(2) Coroner's autopsy. The coroner's certificate is
assumed to reflect the pathologist's wording of cause of
death31 - to what extent is unknown. Also, until 1993 it
did not even have to be in the standard format.23
Criticism of the non-standard approach to stating
cause of death in autopsy reports32"33 has been
addressed in recommendations from the Royal College
of Pathologists. These urge 'the standard form required
by the OPCS'.34 Ideally, particularly after autopsy,34
clinicians and pathologists should liaise over cause of
death. Although the clinical view of autopsies is not
invariably favourable,35 the macroscopic and microscopicfindingsfrom autopsy need to be reconciled with
clinical findings (including special investigations).
Finally, even with autopsy information, semantic
inaccuracies (i.e. word choice and/or layout) in death
certification are no less relevant and should be
remediable.
Recommendations and guidance about death
certification
Consistency in stating cause of death is encouraged
by various guidelines. The death certificate book
Allegedly, this joint report received more attention
from the national than the medical press,40 and house
officers continue to certify death, as is evident from
their use as study subjects.36'41'42
Extent of inaccuracy in
death certification
Deriving a useful estimate of inaccuracy for mortality
data is difficult because of inter-study differences in:
(1) definition, measurement (how and by whom?)
and practical importance of error, and standards used;
(2) focus (e.g. death certificate or mortality data),
observing everyday practice or simulation exercises,
diagnostic and/or semantic issues.
Some death certificate errors might lose relevance after
coding, e.g. being rectified by the supplementary coding
rules. Also, the OPCS 'query system' allows nonspecific causes to be clarified with the certifier (2-5 per
cent of deaths),22 selectively addressing errors of
omission. Despite the potential of such systems,43'44
however, their long-lasting 'educational' impact on the
certifier is questionable45 given the numbers involved.
For many years, diagnosing cause of death has been
extensively investigated against an autopsy 'gold
standard',46"52 but clinicopathological correlation is
variably defined and the relative contribution of poorly
formulated wording is not explicit:
(1) In a Dublin quality assurance programme,
the 'major clinical diagnosis' (undefined) was not
confirmed at autopsy in 25 per cent of deaths.50
'INACCURACY' IN DEATH CERTIFICATION
(2) A Norwegian autopsy study placed Underlying
Cause of Death in a different International Classification of Diseases (ICD) chapter (e.g. neoplasia is one
of 17) in 12 per cent of cases - a non-significant
decrease from a decade earlier (19 per cent).51 In a wellcited Scottish study,48 22 per cent of cases showed such
discrepancy between chapters. The circulatory disease
chapter was sub-divided into 'cardiovascular' and
'cerebrovascular' - inaccuracy was worst for cerebrovascular disease and infections. For disease codes (i.e.
at a finer level of classification) the overall discrepancy
was 39 per cent,48 compared with 55 per cent in another
large British study.47
(3) Over one-half of major discrepancies found at
autopsy are clinically significant.48
(4) Clinicopathological correlation decreases with
decreasing clinical diagnostic certainty48'53 and with
increasing patient age47 (probably because of multiple
pathology). Cause of death itself also affects clinicopathological correlation; e.g. in this country, lung
cancer is underdiagnosed and cerebrovascular disease
overdiagnosed.53
(5) Some estimates are difficult to interpret because
of confused terminology.52
(6) Work in the United States suggests that underrecording of the fact of autopsy on death certificates (by
27 per cent) could bias any autopsy studies selecting
cases this way.54
Other studies unrestricted to autopsied deaths use
expert review of all information, but autopsy rates
differ.55"57 Disagreement with certified Underlying
Cause of Death, for disease codes, affected 20 per
cent in a large, Spanish, population-based study.56 This
was comparable to British estimates.55 Duration of
professional relationship with the deceased and place of
death56 have little effect, but disagreement is greatest
for coroner's certificates and least when there is no
autopsy.55 More information in complex cases apparently allows both clarification and confusion.
Alternatively, cause of death wording is examined
alone,58"62 but criteria differ subtly. In Sri Lankan, US
and Canadian work, 62 per cent,59 59 per cent58 and 32
per cent,61 respectively, had unacceptable Cause of
Death statements. In Australia, 'major errors' (when
Underlying Cause of Death is not conveyed appropriately) affect 16 per cent.60 Where death certificate
counterfoils have been similarly studied in this country,
in hospital 25 per cent are inaccurately worded for
cause of death (i.e. no cause given; multiple causes
given - sequence unclear; single cause given - relevant
details absent; single cause given - error in layout).62 In
general practice, 6 per cent omit age, sex or place of
death63 (and each general practitioner certifies on
63
average 12 deaths a year39). How closely counterfoils
mirror certificates is unknown.
Inconsistency is also inherent in manual coding
because coding rules require interpretation.64 Formerly, underestimation of male pneumoconiosis
deaths by 48 per cent was blamed mainly on 'the
ignorance of pathologists and coroners' of death
certification, and on OPCS slavishly following coding
rules rather than the coroner's intention.31 Cause of
Death statements implying but not stating 'pneumoconiosis' were highlighted for being coded otherwise, even
though accompanied by industrial disease verdicts
sufficient for widows' pensions.
As shown by vignette-based simulation exercises,
international diagnostic and coding variations are
considerable,65'66 e.g. for European mortality differences, diagnostic transfer mostly explains differences in
uterine and cervical cancer65 and 35 per cent of the
variation for diabetes mellitus is explained by coding
practices.66 International67 and UK interregional41
mortality differences are unlikely to be explained by
diagnostic transfer between broad disease categories,
e.g. ICD chapters.
Impact of inaccuracy in death
certification
Quantified examples of the impact of inaccuracy are
needed; a rare example involved fractured neck of
femur, under-recording of which might have produced
a revenue loss of up to £2 million by Leicestershire
Health Authority for 1988-1989.68 (The calculations
involved condition-specific standardized mortality
ratios.)
Inaccurate mortality data have extensive, but
unquantified, international implications beyond the
more obvious health care uses. Bringing mortality data
into conventional analysis of the economic organization of society is probably no less important in
developed countries such as the United States than in
sub-Saharan Africa.69
The certifier and the certificate
Poor medical knowledge about referring deaths to the
coroner,70 and 'the apparent inability of doctors to
complete a death certificate accurately'71 are worldwide concerns.72 Evidence of deficient death certificate
wording accumulates, e.g. 29 per cent of 500 death
certificate counterfoils were reported by a pathologist
to show wording inaccuracies.73 Ironically, however,
the instructions about 'modes of dying' were misinterpreted - the true estimate was lower, albeit important,
at 14 per cent.71 (Modes of dying are 'mechanisms', e.g.
64
JOURNAL OF PUBLIC HEALTH MEDICINE
acute cardiac failure, which are discouraged altogether
on US certificates,74 but are allowed in this country if
qualified, i.e. not used alone in Part I.) Knowledge of
coroners and pathologists about constructing Cause of
Death statements has yet to be measured.
To be well written, Cause of Death statements
require attention to detail. Exhortations to be specific
abound.75 Some certifiers appear 'lazy', favouring
particular diagnoses; e.g. a hospital doctor completed
46/79 (58-2 per cent) death certificates in one year with
'bronchopneumonia' in Part I, but an apparent Underlying Cause of Death in Part II.31
Specific
education
is commonly
recommended.24'25'36'39'40'44'76'77 This should recognize
current constraints.36 These include the sequential
'diagnostic' logic and single-cause coding of Cause of
Death statements, which generate error in themselves,78"80 and the criteria for Part II entries, which
are difficult to fulfil.80 The statement format is
'hallowed by tradition' but some would like an
addendum for multiple pathology,81 because cause of
death is 'more a matter of philosophy than fact'.82 In a
US randomized controlled trial, reversing the ordering
of the statement was ineffective, and requesting only the
Underlying Cause of Death was detrimental.83'84 Such
work is unusual. Using the Underlying Cause of Death,
extracted from an internationally agreed statement
format, is a risky but pragmatic compromise. Loss of
information, error and misinterpretation are possible,
yet this efficient and simple summary measure is
invaluable if interpreted critically.
World wide, knowledge and attitudes of certifiers36'42-45'74'76 have been less studied than clinicopathological correlation. House officers and general
practitioners appreciate the importance of accurate
death certification, and are interested in doing something about improving performance, but knowledge
varies.36 Some certifiers modify cause of death to
avoid distress,36 e.g. not recording alcohol.85 A major
factor in deficient death certification is, however, its
'routinized' orientation.45 When interviewed, certifiers
contradict themselves in explaining the main concepts
of cause of death45'74 - routine tasks are not usually
discussed or reflected upon.
Where to next?
The legal aspects of dying have long since required
review,71 but organizational and technical 'fixes' are
unlikely, alone, to improve cause of death data. Some
view the current system as more about 'policing the
dead' than producing worthwhile data.64
It is known that inaccurate death certification is a
problem, but not how to reconcile the relative
contributions of diagnostic and semantic errors and
relate these to impact. Beyond more robust evidence, a
different educational perspective (to apply the available
evidence more effectively) is required, advancing from
merely urging educational input to evidence-based
interventions. Certifiers are receptive to more education
about death certification,36 but it is not yet known
which interventions are best. The flaws in the theoretical framework of cause of death64 and the routine
nature of death certification are unavoidable, but
necessary considerations. Certifiers need practical feedback mechanisms to improve understanding of the
construction of mortality data.
Mortality data are essential for many aspects of
everyday public health practice. More meaningful
estimates of accuracy are thus required, as are
evidence-based interventions, educational commitment
and continuing quality assurance (integrated within the
existing audit framework) at all levels. Autopsy is not
the only answer36'77 to inadequate cause of death
wording, but the potential contribution of pathologists
cannot be underestimated, e.g. assisting the death
certification of autopsied54 and unautopsied deaths,
and facilitating good practice.87 Given the epidemiological origins of death certification, and the clear
potential for further improvement through dialogue
and sharing expertise with clinicians, continued
development should be a core public health medicine
role.
'Die, my dear Doctor! That's the last thing I shall
do.'88
Viscount Palmerston (Henry John Temple), 1784-1865,
on being told that he was dying
References
1
National action to improve
death certification
A continuing OPCS initiative to improve the quality of
death certification8* has been updating the instructions,
revising the certificate format within the limits of
current legislation, and producing an educational
video, the effects of which will need to be evaluated.
2
3
McPherson K, Bunker J. Health information as a guide to
the organization and delivery of services. In: Holland
WW, Detels R, Knox G, eds. Oxford textbook of public
health. Vol. 2, Methods of public health, 2nd edn. Oxford:
Oxford University Press, 1991: 71-73.
Secretary of State for Health. The health of the nation: a
strategy for health in England. Cm 1986. London:
HMSO, 1992.
Donaldson RJ, Donaldson LJ. Essential public health
medicine. Lancaster: KJuwer Academic, 1993: 2-34.
' I N A C C U R A C Y ' IN DEATH C E R T I F I C A T I O N
4
Pollock A. The future of cancer registries: purchasers need
to recognise how important they are for monitoring
services. Br Med J 1994; 309: 821-822.
5
Department of Health, Welsh Office. A policy framework
for commissioning cancer services. A report of the Expert
Advisory Group on Cancer to the Chief Medical Officers
of England and Wales. London: Department of Health,
1995.
6
Holland WW, Fitzgerald AP, Hildrey SJ, Phillips SJ.
Heaven can wait. / Publ Hlth Med 1994; 16(3): 321330.
7
McKee M, Hunter D. Mortality league tables: do they
inform or mislead? Qual Hlth Care 1995; 4: 5-12.
8
Smith R. Towards a knowledge based health service:
priorities are set for health technology assessment.
BrA/e</J1994;309:217-218.
9
Ludwig J. About death certificates (editorial). Mayo Clinic
Proc 1980; 55: 347-348.
10
Report of the Committee on Death Certification and
Coroners. Cmnd 4810. London: HMSO, 1971.
" Office of Population Censuses and Surveys. Mortality
statistics, general: reviews of the Registrar General on
deaths in England and Wales [for] 1988,1989,1990,1991,
1992. Series DH1 No. 21,23,24,26,27. London: HMSO,
1990-94: Table 10.
12
Prior L. The social organization of death: medical discourses
and social practices in Belfast. London: Macmillan, 1989:
25-46.
13
Nuland SB. How we die. London: Chatto & Windus, 1994:
43-44.
14
World Health Organization. Manual of the International
Classification of Diseases, Injuries and Causes of Death,
9th revision. London: HMSO, 1977: I and II.
15
Nordenfelt L. Causes of death - a philosophical essay.
Stockholm: Swedish Council for Planning and Coordination of Research, Committee for Future Oriented
Research, 1983.
16
Reay DT. Certification of death. Trans Assoc Life Insurance
Med Directors Am 1990; 73: 155-164.
17
Ashley J, Devis T. Death certification from the point of
view of the epidemiologist. Population Trends 1992; 67:
22-28.
18
Hamlin C. Could you starve to death in England in 1839?
The Chadwick-Farr controversy and the loss of the
'social* in public health. Am J Publ Hlth 1995; 85(6): 8 5 6 866.
19
Anonymous. Certification of smoking related deaths.
Health Trends 1992; 24(2): 46.
20
Arrundale J, Cole SK. Death certification and mortality
statistics. Health Bull 1991; 49(1): 82-90.
21
Knight B. Legal aspects of medical practice, 5th edn.
Edinburgh: Churchill Livingstone, 1992: 123-134.
22
23
24
25
Ashley JSA, Cole SK, Kilbane M P J . Health information
resources: United K i n g d o m - health a n d social factors mortality statistics. In: H o l l a n d W W , Detels R, K n o x G,
eds. Oxford textbook of public health, Vol. 2, Methods of
public health, 2nd edn. Oxford: Oxford University Press,
1991: 3 2 - 3 8 .
Birch D . A u t o m a t i c coding of causes of death. Population
Trends 1993; 7 3 : 1-3.
Balarajan R. Certification of d e a t h a n d the clinician. Health
Trends 1983; 15: 53-56.
G a u DW, Diehl A K . Disagreement among general practitioners regarding cause of death. Br Med J 1982; 284:
239-241.
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
65
Evans P M . Death certificates in patients with chronic
diseases. Med Audit News 1991; 1(2): 24-26.
Births and Deaths Registrations Act (1953): Public General
Acts and Church Assembly Measures of 1953. London:
H M S O , 1953: 1 and 2 Eliz. II, Chapter 20, Part II,
Section 22(1): 176.
Maudsley G, Williams E M I . Death certification - a sad
state of affairs (Letter). J Publ Hlth Med 1994; 16(3):
370-371.
Report of the Joint Working Party of the Royal College of
Pathologists, the Royal College of Physicians of London
and the Royal College of Surgeons of England. The
autopsy and audit. London: Royal College of Pathologists, 1991.
Hunt A C . The appropriate use of diagnostic services: (Lx)
The autopsy: a personal view. Health Trends 1986; 18:
27-28.
Cochrane A L , Moore F. Death certification from the
epidemiological point of view. Lancet 1981; ii: 742-743.
Uncertain certificates (Editorial). Lancet 1981; ii: 2 2 - 2 3 .
Busuttil A, Kemp IW, Heasman MA. The accuracy of
medical certificates of cause of death. Health Bull 1981;
39: 146-152.
Royal College of Pathologists. Guidelines for post mortem
reports. Bull R Coll Pathol 1993; 84: 11-14.
Paton A. Personal view. Br Med J 1972, 3: 287.
Maudsley G, Williams E M I . Death certification by house
officers and general practitioners - practice and
performance. J Publ Hlth Med 1993; 15(2): 192-201.
World Health Organization. Medical certification of cause
of death: instructions for physicians on the use of
International Form of Medical Certificate of Cause of
Death. Geneva: World Health Organization, 1979.
Knight B. Brodrick exhumed again. Br Med J 1982; 285:
393-394.
A joint report of the Royal College of Physicians and the
Royal College of Pathologists. Medical aspects of death
certification.//? Coll Phys London 1982; 16(4): 206-218.
Alderson M R , Bayliss RJS, Clarke CA, Whitfield A G W .
Death certification. Br Med J 1983; 287: 444-445.
Diehl A K G a u D W . Death certification by British doctors:
a demographic analysis. J Epidemiol Commun Hlth 1982;
36: 146-149.
Bell G, Cremona A. Alcohol and death certification: a
survey of current practice and attitudes. Br Med J 1987;
295: 95.
Hopkins D D , Grant-Worley JA, Bollinger TL. Survey of
cause-of-death query criteria used in the US and the
efficacy of the criteria used by the Oregon vital statistics
program. Am J Publ Hlth 1989; 79(5): 570-574.
Rosenberg H M . Improving cause-of-death statistics. Am J
Publ Hlth 1989; 79(5): 563-564.
Bloor M. A minor office: the variable and socially
constructed character of death certification in a Scottish
city. J Hlth Social Behav 1991; 32(3): 273-287.
James G, Patton R E , Heslin AS. Accuracy of Cause-ofDeath statements on death certificates. Publ Hlth Rep
(Washington) 1955; 70(1): 3 9 - 5 1 .
Heasman M A , Lipworth L. Accuracy of certification of
cause of death: a report on a survey conducted in 1959 in 75
hospitals of the National Health Service to obtain
information on the extent of agreement between clinical
and post-mortem
diagnoses. General Register Office;
Studies on Medical and Population Subjects N o . 20.
London: H M S O , 1966.
66
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
63
66
67
68
J O U R N A L OF PUBLIC HEALTH MEDICINE
Cameron HM, McGoogan E. A prospective study of 1152
hospital autopsies; (I) Inaccuracies in death certification.
JPathol 1981; 133: 273-283.
Cameron H M , McGoogan E. A prospective study of 1152
hospital autopsies: (II) Analysis of inaccuracies in clinical
diagnosis and their significance. J Palhol 1981; 133:285300.
Harrison M, Hourihane DO'B. Quality assurance programme for necropsies. J Clin Palhol 1989; 42: 1190—
1193.
Karwinski B, Hartveit F. Death certification: increased
clinical confidence in diagnosis and lack of interest in
confirmation by necropsy is not justified. J Clin Palhol
1989; 42: 13-17.
Nielsen G P , Bjornsson J, Jonasson J G . The accuracy of
death certificates: implications for health statistics.
Virchows Arch A, Palhol Anal Hislopalhol 1991; 419:
143-146.
Heasman M A . Accuracy of death certification. Proc R Soc
Med 1962; 55: 733-736.
Hanzlick R, Parrish R G . The failure of death certificates to
record the performance of autopsies (Letter). J Am Med
Assoc 1993; 269(1): 47.
Medical Services Study G r o u p of the Royal College
of Physicians of London. Death certification and
epidemiological research. Br Med J 1978; U: 1063-1065.
Benavides F G , Bolumar F , Peris R. Quality of death
certificates in Valencia, Spain. Am J Publ Hlth 1989;
79(10): 1352-1354.
Mousse M A A , Shafie MZ, Khogali M M , et al. Reliability
of death certificate diagnoses. J Clin Epidemiol 1990;
43(12): 1285-1295.
Zumwalt R E , Ritter M R . Incorrect death certification: an
invitation to obfuscation. Postgrad Med 1987; 81(8):
245-254.
F e r n a n d o R. Medical certification of cause of death in the
General Hospital, Colombo. Ceylon MedJ 1990; 3 5 : 7 1 74.
Weeramanthri T, Beresford B. Death certification in
Western Australia - classification of major errors in
certificate completion. AuslJ Publ Hlth 1992; 16(4): 4 3 1 434.
Jordan J M , Bass MJ. Errors in death certificate completion
in a teaching hospital. Clin Invest Med: Mid Clin Exp
1993; 16(4): 249-255.
Leadbeatter S. Semantics of death certification. J R Coll
Phys London 1986; 20(2): 129-132.
Black D A , J a c h u c k SJ. D e a t h certification in general
practice: review of records. Br Med J 1984; 288: 1 1 2 7 1129.
P r i o r L. M a k i n g sense of mortality. Social Hlth Illness 1985;
7: 167-190.
Kelson M, Farebrother M. The effect of inaccuracies in
death certification and coding practices in the European
Economic Community (EEC) on international cancer
mortality statistics. Int J Epidemiol 1987; 16(3): 411-414.
Balkau B, Jougla E, Papoz L, and the E U R O D I A B
Subarea C Study Group. European study of the
certification and coding of causes of death of six clinical
case histories of diabetic patients. Int J Epidemiol 1993;
22(1): 116-126.
Reid D D , Rose G A . Assessing the comparability of
mortality statistics. Br MedJ 1964; V. 1437-1439.
D o n a l d s o n LJ, Parsons L, C o o k A J . Death certification in
fractured neck of femur. Public Hlth 1989; 103: 2 3 7 - 2 4 3 .
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
84
87
88
Sen A. T h e economics of life a n d death. Sci Am 1993;
268(5): 18-25.
Start R D , Delargy-Aziz Y, D o m e s C P , Silcocks P B ,
Cotton B W K . Clinicians a n d the coronial system: an
assessment of the ability of clinicians to recognize
reportable deaths. Br MedJ 1993; 20: 273-275.
Leadbeatter, Knight B. Reporting deaths to coroners: all
the legal aspects of dying need re-examining. Br Med J
1993; 306: 1018.
Still A, Koelmeyer T D , Vuletic J C . Death certification in
Auckland. NZ MedJ 1992; 105; 330-332.
Slater D N . Certifying the cause of death: an audit of
wording inaccuracies. J Clin Pathol 1993; 46: 232-234.
Smith A F , Mingay DJ, Jobe JB, Weed JA, Clark G. A
cognitive approach to mortality statistics. In: Proceedings of the American Statistical Association
Meeting,
survey research methods section; 1992; Boston. Alexandria, V A : A m e r i c a n Statistical Association, 1992: 8 1 2 817.
Adelstein A M . Certifying cause of d e a t h . Health Trends
1977; 9: 7 8 - 8 1 .
Freedman M A , Rushford G A , McQuillen EN, Baron K P .
Issues related to the medical certification of death: a
physician survey. NY State J Med 1988; 88(10): 522-525.
Moriyama I M . Problems in measurement of accuracy of
cause-of-death statistics. Am J Publ Hlth 1989; 79(10):
1349-1350.
Gittelsohn A M . O n the distribution of Underlying Causes
of D e a t h . Am J Publ Hlth 1982; 72(2): 133-140.
Kielkowski D , Steinberg M , B a r r o n P M . Life after d e a t h mortality statistics a n d the public health. S Afr Med J
1989; 76(12): 6 7 2 - 6 7 5 .
Lindahl BIB, Glattre E, Lahti R, Magnusson G, Mosbech
J. The W H O principles for registering causes of death:
suggestions for improvement. J Clin Epidemiol 1990;
43(5): 4 6 7 - 4 7 4 .
Ashworth T G . Inadequacy of death certification: proposal
for change. J Clin Pathol 1991; 44: 265-268.
Emery JL. Certification of death by the pathologist. Proc R
Soc Med 1962; 55: 738-740.
Jobe JB, Smith AF, Mingay DJ, Weed JA, Roberts RM.
Cognitive issues in improving the quality of mortality
statistics. In: Proceedings of the 1993 public health
conference on records and statistics: towards the year
2000 - refining the measures; 1993; Washington, DC.
Washington, DC: US Government Printing Office, 1993;
PHS 94-1214: 255-260.
Smith A F , M i n g a y D J , Weed J A , J o b e J B . Cognitive
processes in completing d e a t h certificates. In: H e r r m a n n
D , Chair. Contributions of cognitive and social psychology
to answering survey questions. Symposium presentation at
6th A n n u a l C o n v e n t i o n of the A m e r i c a n Psychological
Society 1994; W a s h i n g t o n .
Bell G, Cremona A. Alcohol and death certification:
influencing current practice and attitudes. Br J Addiction
1989; 84: 1523-1525.
Anonymous. OPCS to pilot revised medical certificate of
cause of death. OPCS Newslett 1995; 3(Sep): 3.
Kelly S A , G r a d w e l l E. T h e teaching of d e a t h certification
(Letter). J Clin Pathol 1992; 45(10): 9 4 2 - 9 4 3 .
Viscount Palmerston 1865. Cited in: Wintle J, Kenin R, eds.
The Penguin concise dictionary of biographical quotations.
Harmondsworth: Penguin, 1981: 484.
Accepted on 17 October 1995