Causes of Bradford Infant Mortality – Investigation with by etssetcf


More Info
									Causes of Bradford Infant Mortality – Investigation
with particular attention to effect of ethnicity

Executive Summary

This section of the report considers the original research commissioned by the
IMC. Because the data included are based on careful appraisals of raw data,
rather than the Office of National Statistics (ONS) classifications of death
registration data, the conclusions differ from those that may be drawn from
ONS data and presented elsewhere in the report.

This section shows that:
   • Bradford Infant Mortality disproportionately affects Pakistani families, at
       least in Airedale. This is consistent with both other areas of the report
       and other published data.
   • Bradford’s high infant mortality is at least partly caused by an excess of
       deaths from congenital anomalies. Babies in an equivalently deprived
       but predominantly white British area outside Bradford are less than half
       as likely to die from congenital anomalies.
   • As a minimum some of Bradford’s excess infant mortality can be
       attributed to autosomal recessive inheritance, and therefore, based on
       published literature, to consanguinity (colloquially “cousin marriage”).
       Bradford babies were more than 5 times more likely to die from
       autosomal recessively inherited diseases than those born in an
       equivalently deprived and predominantly white British area. This
       accounts for around 4 extra infant deaths a year over what might be
       expected for a population as deprived as that of Bradford.
   • That the causes of infant death are not always accurately reflected by
       ONS coding.
   • That further research about exact causes of infant death must underlie
       any future work on Bradford’s excess infant mortality.

Routinely used statistics on infant mortality are derived from the Office for
National Statistics (ONS) who in turn collate information from birth and death
registers. When a death occurs, the attending physician usually completes a
death certificate. A death certificate may be completed before a post-mortem
examination is carried out, or full investigations as to cause of death are
complete. Some deaths are referred to the coroner who ultimately completes
the death certificate. The cause of death recorded on the death certificate is
coded and causes of deaths are usually considered in groups (such as
“infections” and “congenital anomalies”).

By contrast, until the end of 2003, the confidential enquiry into stillbirths and
deaths in infancy were collating information from doctors about infant deaths.
The resulting data may be informed by further diagnostic information and
consideration than was available when the death certificate was completed.

The accuracy of the ONS process in reflecting actual cause of death will be of
interest to all areas investigating their local infant mortality. However these
data may be of particular importance in Bradford.

Bradford has high infant mortality. Bradford’s population is around half a
million people, with a relatively large ethnic minority population. Bradfords
ethnic minority population are predominantly Pakistani in origin. Close
analysis of cause of death information might therefore be expected to provide
clues as to both the causes of death among Bradford’s Pakistani population,
but also potentially generalisable themes for other Pakistani populations

Given its remit to consider Bradford’s excess mortality, the IMC was
concerned to consider causes of death in as much detail as possible. IMC
members noted the concerns that both the quality of death certification in
general and the possibility that the grouping by ONS might not always reflect
the underlying cause of death. IMC noted that research work is ongoing to
further explore the extent to which ONS cause of death relates to the actual
cause of death2.

Sources of Information

The commission considered information about cause of infant death from four

   1)     Information derived from death certificates. This information is
          processed in the manner alluded to above and all the cause of
          death analyses in the “Analysis and Interpretation” paper are based
          on such analyses.
   2)     The Confidential Enquiry into Maternal and Child Health or
          CEMACH (formerly Confidential Enquiry into Stillbirths and Deaths
          in Infancy or CESDI) collected data on infant mortality until the end
          of 2003. Clinicians attending a child who died completed a form
          attributing a cause of death and these were centrally collated.
          CEMACH made some data available to Bradford Health
          Informatics. These data were analysed by comparing Bradford to
          the rest of England and Wales, and are presented below entitled
          “CEMACH data”.
   3)     Further to the analysed data provided by CEMACH, IMC
          commissioned local paediatricians to compare a number of
          individual deaths in Bradford to deaths from elsewhere, using data
          collected by CEMACH and obtained separately from them. This
          work is presented below under the title “Bradford Cause of Death
   4)     Airedale hospital paediatricians consider infant deaths routinely at a
          multidisciplinary team meeting leading to a “clinician attributed
          cause of death”. After the Airedale paediatricians made their data
          available to the IMC, IMC commissioned a reanalysis of the data.
          Comparison of the multidisciplinary meeting attributed cause of
          death with ONS coded cause of death appears below entitled “Do
          ONS causes of death reflect actual cause of death?” The extent to
          which this dataset could inform an overrepresentation of deaths
          attributed to lethal congenital anomalies among ethnic minorities is
          presented below entitled “The Airedale dataset”

The origins of the data forming the basis of this chapter differ from those
underlying the “Analysis and Interpretation” paper. Causes of death in this
chapter are assigned using pragmatic clinical methodologies rather than the
“grouping of causes” approach taken by ONS. It is therefore to be expected
that differing (but complementary) themes emerge in this chapter to those
identified in the “Analysis and Interpretation” paper.


Data obtained from the Confidential Enquiry into Child and Maternal Health
(CEMACH) on infant deaths between 1994 and 2003 provided a limited
insight into the differences in cause of death among South Asian and non
South Asian babies across the country. These data indicated that congenital
anomalies were identified as a factor in a greater proportion of deaths among
South Asian babies than non South Asian babies in the Bradford District and
across England and Wales. In the Bradford District 50% of South Asian infant
deaths reported to CEMACH mentioned a congenital anomaly as a factor
leading to death compared to 23% of non South Asian infant deaths. In
England and Wales 42% of South Asian infant deaths reported to CEMACH
included a congenital anomaly as a contributory factor compared to 22% of
non South Asian infant deaths.

These data indicates that a greater proportion of South Asian infant deaths
are due to congenital anomalies than non South Asian infant deaths. It also
suggests that South Asian infant deaths in Bradford may be more likely to be
due to congenital anomalies than South Asian infant deaths across England
and Wales. As the majority of Bradford’s South Asian population is of
Pakistani origin, it may reasonably be inferred that congenital anomaly deaths
are more common among Pakistani infants. This has been described4.

Do ONS causes of death reflect actual cause of death?
As part of analysis of all infant deaths in Airedale PCT in years 1993-2003
inclusive, Dr James Yong on behalf of the commission, noted the clinician
attributed cause of death and compared it with the ONS attributed cause of
death. Among a subset of deaths for which both ONS cause of death and a
clinician agreed cause of death was available, 6/ 71 deaths were recorded in
a way that clinicians would regard as seriously inaccurate or misleading.
Clearly a variety of possible reasons including miscoding or incomplete death
certification might underlie such disparities.

Given the limitations of the ONS derived data, the IMC noted that optimal
information about cause of death could only be obtained in a prospective
study and the time frame of IMC was too short to carry out such a study.
Airedale Dataset
A consultant paediatrician from Airedale Hospital NHS trust came to speak to
IMC bringing data about each death known to the Airedale paediatricians1993
-2004. Airedale hospital deals with some patients not resident in Bradford.
There was concern that some Airedale PCT deaths might have been missed.
Therefore, IMC commissioned Dr James Yong (supervised by Dr Sam Oddie)
to review the dataset.

After death certification and any post-mortem report is available, Airedale
paediatricians review each death in a multidisciplinary meeting and attribute a
cause of death. Dr Yong reviewed the ethnicity coding for each death and
confined the analysis to deaths occurring to families resident in Airedale PCT
1993-2004 and ensured that all Airedale PCT resident infant deaths had been
included in the analysis.

Table 1 shows that there is a large actual excess of lethal congenital
anomalies in families of Asian ethnicities over the number of such deaths to
“non Asian” families despite the predominantly white population in Airedale
PCT. “Asian” population groups comprise around 20% of Airedale PCTs
Table 1
Infant Deaths in Airedale PCT by clinician attributed cause of death, divided
by ethnicity

                  Non        Asian        Total
                  Number of Number of
                  deaths (%) deaths (%)
No       Lethal 32           11           43

Lethal            16         27           43

Total             48         38

Of the 38 “asian” families, 29 had self reported ethnicity as Pakistani. This
suggests that Pakistani infant mortality in Bradford, or at least in Airedale
PCT, is high, as it represents 35% (29/86) of Airedale deaths despite
Airedale’s low Pakistani birthrate. That Pakistani infant mortality is higher in
Bradford than that in the white British population has been shown in other
areas of the report. Numbers of other ethnic groups are too small to permit
meaningful analysis.
Cause of Death Study
The IMC commissioned a group led by Dr Sam Oddie to review routinely
collected data about infant deaths in Bradford and compare it with areas of
similar deprivation.

Using anonymised data, without data on ethnicity, drawn from that gathered
nationwide by the Confidential Enquiry into Maternal and Child Health
(CEMACH) we compared all 323 Bradford Infant deaths in years 1999-2003
inclusive with 120 deaths in a similarly deprived but predominantly white area
(aggregated data from Doncaster and Mansfield)) and 174 deaths in a
similarly deprived but ethnically mixed area (Leicester). Leicester has a
relatively low Pakistani population 3. The time periods for the deaths from all
areas were the same. Deprivation comparisons were made using Index of
Multiple Deprivation 2004.

Three trained clinicians reviewed the available information on each death
without being aware of where the death originated from, family name or
ethnicity and made judgements as to the significance to the deaths of
prematurity, anomalies etc. The intention was to investigate the hypothesis
that causes of infant death in Bradford were distinguishable fro those in the
comparator areas.

Table 2
Rates and Relative Risk (95% confidence interval) of Infant Deaths in
Bradford compared to areas of comparable deprivation and differing ethnic
makeup 1999 – 2003

                                  Bradford Metropolitan   Doncaster           and   Leicester
                                  District                Mansfield

Total Births 1999 - 2003          35 823                  21 190                    19 968

Total Infant Deaths 1999 -        323                     120                       174

Infant Mortality                  9.02                    5.66                      8.71
(rate/ 1000 livebirths)

Deaths        attributed     to   3.92                     3.27                     4.89
(rate/ 1000 livebirths)

Relative Risk and          95%    1                       0.83 (0.63, 1.11)         1.26 (0.97, 1.61)
confidence interval

Deaths        attributed     to   4.89                    1.45                      2.55
Congenital Anomalies
(rate/ 1000 livebirths)

Relative Risk and          95%
confidence interval               1                       0.42 (0.29, 0.63)         0.74 (0.53, 1.02)

Deaths        attributed    to    0.78                    0.16                      0.30
Autosomal            Recessive
(rate/ 1000 livebirths)

Relative Risk and 95%             1                       0.18 (0.06, 0.60)         0.38 (0.16, 0.93)
confidence interval
Deaths from diseases that         2.40                    0.58                      2.10
were potentially diagnosable
in the antenatal period
(rate/ 1000 livebirths)

Relative Risk and 95%
confidence interval               1                       0.24 (0.13, 0.43)         0.88 (0.61, 1.27)
Deaths from diseases where        5.93                    2.25                      6.56
death would be medically
anticipated in first 2 years of
(rate/ 1000 livebirths)

Relative Risk and          95%    1                       0.38 (0.28, 0.53)         1.11 (0.89, 1.38)
confidence interval
Results (table 2) show that the rate of infant death attributed to congenital
anomaly in Bradford is more than twice that in Doncaster and Mansfield
combined and higher than in Leicester though this latter finding just misses
the conventional criteria for statistical significance

The comparison populations were selected to have ethnic mixes contrasting
with those of Bradford. Bradford’s substantial Pakistani population is one
notable difference between Bradford’s population and that of the comparison
areas. Evidence exists that Pakistani infants living in the UK are more likely
than non Pakistani infants to die from congenital anomalies4. Given these
findings, that Bradford’s infant mortality attributed to congenital anomalies
exceeds that of other areas, was arguably predictable.

Additionally in Bradford more deaths could be attributed to autosomal
recessive disease than in either comparator area. The risk of such an infant
death in Bradford was five times the risk in Doncaster and Mansfield, and
more than twice the risk in Leicester. This fits with work by Bundey et al, but
is very important due to its relatively contemporaneous nature4. Although the
dataset analysed contained no information about the ethnicity of each infant
death, the comparison areas were chosen to provide contrasting ethnic mixes,
with the aim of investigating whether Bradford’s infant mortality might be
influenced by its ethnic composition.

Interpretation of Bradford cause of death study

We have shown that infant mortality attributed to congenital anomaly occurs
more commonly in Bradford than elsewhere in England. This is consistent
with the findings, particularly relating to Pakistani infants, of the analysis
section of the report. Additionally taken with the results of the CEMACH data,
and the “Airedale dataset” this suggests that increased rates of congenital
anomaly deaths in Bradford may be at least partly explained by the unique
nature of Bradford’s ethnic minority population.

A minimum of 5 Bradford infant deaths a year have been shown to be
attributable to autosomal recessive inheritance, which is not a large figure
compared to the approximately 70 infant deaths a year seen in Bradford.

If we make the assumption that Leicester, Doncaster and Mansfield are
representative of the England and Wales population as a whole, then around
4 Bradford infant deaths a year (over what could be expected were it not for
Bradford’s ethnic mix) can be attributed to autosomal recessive inheritance.

However this figure should be regarded as a minimum estimate. The
available source data were very limited in nature with only limited fields to
describe the circumstances and cause of each death. Post mortem rates are
low, particularly in Bradford. Additionally, as it is frequently not possible to
make exact genetic diagnoses, clear diagnoses may not be recorded on
death certificates or CEMACH report forms. Although these caveats will apply
to all areas, as the incidence in Bradford was higher, increased ascertainment
might be expected to lead to a still more obvious difference between rates of
death attributed to autosomal recessive inheritance in Bradford and

In other words the otherwise unexplained excess anomaly rates among the
Bradford population is likely to be partially explained by underascertainment of
autosomal recessive inheritance.

The contribution of diseases with the potential for antenatal diagnosis to infant
mortality was also considered. We showed (table 2) that infants in Bradford
were more likely than those in a predominantly white British area to die of
conditions that were, at least theoretically, antenatally diagnosable. This has
implications for delivery of fetal medicine and antenatal genetic services to the
people of Bradford.

An excess of Bradford infant deaths were judged to have occurred on account
of conditions which inevitably lead to death, by comparison to a predominantly
white British area.

How does this section’s data compare with other IMC

This section has contributed new information to the IMC report. It shows
unequivocally that autosomal recessive inheritance explains at least a part of
Bradfords’ infant mortality excess and this may well relate to consanguinity4.
That consanguinity is associated with increased infant mortality is known,
regardless of ethnicity7.

The excess of deaths from congenital anomalies is slightly greater in this
analysis than that shown in the analysis section.. This may reflect that
clinicians reviewed data concerning each death rather than depending on the
accuracy of the death certification process and ONS’s classification of the
resultant data. We think the Bradford cause of death study, taken with the
other work in this chapter, provides evidence that the ONS system may
underascertain deaths truly attributable to anomalies.
   1. Bradford       Metropolitan        District     Council        website:
      pulation_as_per_census_2001%2f accessed 14th February 2006

   2. CEMACH       website         accessed    12th  February   2006

   3.         (accessed      10th
      February 2006)

   4. Bundey S, Alam H. A five year prospective study of the health of
      children in different ethnic groups, with particular reference to the effect
      of inbreeding. Eur J Hum Genet. 1993;1(3):206-19
   5. Health Services Quarterly. 28. Winter 2005.
   6. Source, Bradford Health Informatics
   7. Stoltenberg C, Magnus P et al. Influence of consanguinity and
      maternal education on risk of still birth and infant death in Norway
      1967-1993. Am J Epid 1998;148 (5)452-458

To top