Jump to ContentJump to Main Navigation
Death in ChildbirthAn International Study of Maternal Care and Maternal Mortality 1800-1950$

Irvine Loudon

Print publication date: 1992

Print ISBN-13: 9780198229971

Published to Oxford Scholarship Online: October 2011

DOI: 10.1093/acprof:oso/9780198229971.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 24 February 2017

(p.518) Appendix 1 Hidden Maternal Deaths

(p.518) Appendix 1 Hidden Maternal Deaths

Source:
Death in Childbirth
Publisher:
Oxford University Press

In the Forty-Fourth Annual Report of the Registrar General (Abstracts of 1881), the Registrar General wrote:

Not rarely, the real cause of death is purposely disguised in order to spare the susceptibilities of friends, or, it may be, to conceal the existence of infectious disease…More frequently, however, the deficiencies in the certificates are not intentional but due to carelessness…An effort was made this year to remedy this evil, and to get greater precision in the statement of causes. With this object, letters of inquiry were sent out to medical men who had given certificates in which the causes of death were imperfectly stated, asking for further particulars. The exigencies of office work did not allow of this being done to more than a very limited extent. Still, a beginning was made…In all some 1,200 letters of inquiry were sent out…

These enquiries, although they dealt with only a sample of unsatisfactory certificates, revealed that out of 183 deaths certified as ‘blood poisoning, pyaemia, septicaemia’ a puerperal cause was found in 89. Out of 321 inquiries concerning deaths due to ‘peritonitis’, a puerperal cause was found in 136. Altogether, a total of 330 ‘hidden’ maternal deaths were discovered, of which 80 per cent were deaths due to puerperal fever. All these deaths which, had there been no enquiry would have been assigned to non-puerperal causes, were allocated to puerperal sepsis. The 330 deaths represented an addition of 8 per cent to total maternal mortality and 12 per cent of deaths due to puerperal fever. It is probable therefore that before 1880 maternal deaths as a whole, and puerperal sepsis deaths were under-recorded to at least the same extent.

Farr's successor at the General Register Office, Dr Ogle, carried out a similar exercise on a larger scale. He instituted a confidential enquiry between the General Register Office and the certifying medical practitioners of England and Wales. The results were published in the Decennial Supplement for the years 1881–90.1

More than 20,000 letters were sent out. About 4,000 related to deaths certified as being due to ‘peritonitis’ occurring in women of childbearing age. Of these, more than 1,000 were eventually transferred to puerperal fever. Of 3,000 deaths certified as ‘pyaemia, septicaemia’ 700 were found to have been due to puerperal causes. Of 272 deaths certified as ‘haemorrhage’, 69 were connected with childbirth, and of 244 attributed to ‘metritis’ (a term which means inflammation of the uterus) more than half were found to have been connected with the puerperal state. In all, nearly 3,000 deaths in the decennium, in which the certifying practitioners had written a vague cause of death, were found to be deaths in childbirth. After the corrections had been made the (p.519) total childbirth deaths for the decade amounted to 42,092. ‘Hidden’ deaths therefore amounted to a minimum of 7 per cent of the total, and the large majority of these were deaths due to puerperal fever.

It was clear that in the 1890s as well as the 1880s, many deaths due to childbirth were, deliberately or accidentally, certified in such a way that they appeared to be deaths from non-puerperal causes. Medical practitioners continued to be resistant to exhortations to improve the accuracy of certification. Ogle told the Select Committee on Death Registration in 1893 that practitioners often replied to his inquiries with abuse, and complained of the absence of a fee for death certification, and, like their successors today, of ‘interfering bureaucracy’, and the tedium of filling up forms.

By 1898, the level of accuracy was still deplorable. ‘The causes of 23,039, or 4.2 per cent of the total deaths, were so unsatisfactorily stated in the registers as to be useless for purposes of classification.’ The largest number of these were deaths under the age of 1, of which 17.9 per cent were unsatisfactory—a point to be remembered by historians of infant mortality. But it was also found that out of 454 deaths returned as ‘peritonitis, pelvic cellulitis, metritis’, and 291 certified as ‘pyaemia and septicaemia’, 101 and 30 respectively were deaths due to puerperal fever. Once again, some 10 per cent of hidden maternal deaths were discovered in the Registrar General's Office and reallocated to puerperal causes.2

By 1923, the accuracy of certification had gready improved. A total of 8,788 inquiries were sent to medical practitioners and replies were received from 7,905. Because of expense and the considerable amount of work involved, this was, as before, only a sample of imperfect certificates. Out of 182 inquiries concerning deaths certified as ‘pyaemia and septicaemia’ only 8 were due to puerperal sepsis. Out of 183 inquiries concerning deaths certified as ‘peritonitis’ only 5 were due to childbirth causes.3 The results of continuing enquiries during the years 1927–32 can be seen in Table A1.1. The Registrar General managed to restore a certain proportion of maternal deaths to their rightful place by carefully searching deaths attributed to septicaemia and peritonitis. But this was not the only way that maternal deaths were hidden.

We saw in Chapter 2 that deaths due to puerperal fever were sometimes attributed to a non-septic puerperal cause such as ‘haemorrhage’ to avoid the accusation of a poor standard of antisepsis. Hospitals and obstetricians tended to be sensitive to such an accusation and sometimes yielded to the temptation to transfer a death from a septic to a non-septic category to improve the look of the annual report. While the total mortality from childbirth would be unaltered, the mortality from sepsis would be reduced. If ‘haemorrhage’ was chosen, however, the deceit could be exposed by the time of the death. Deaths due to puerperal fever usually occur between the seventh to fourteenth postnatal day, deaths due to haemorrhage in the first few hours or days at the most (Appendix Table 27). However, around the fourteenth day is the period in which most deaths from deep vein thrombosis and pulmonary embolism occur, deaths which used to be classified as ‘phlegmasia alba dolens, embolism and sudden death’. This provided a much safer classification to anyone who wanted to hide a death due to puerperal fever. (p.520)

Table A1.1 England and Wales, 1927–1932. Showing the Reallocation of Certain Causes of Death as a Result of Enquiries into Death Certificates in which the Recorded Cause of Death was Indefinite in the Two Categories in which ‘Hidden’ Deaths from Puerperal Sepsis were Found, Namely ‘Pyaemia and Septicaemia’ and ‘Peritonitis’

Causes of death

1927

1928

1929

1930

1931

1932

Pyaemia/Septicaemia

Replies received

170

168

202

181

216

197

Cause of death reallocated as:

Diseases teeth and gums

7

14

5

11

13

10

Tonsillitis

6

8

6

7

19

22

Diseases of skin

18

15

20

21

32

29

Disease of umbilicus

5

Puerperal sepsis

9

9

3

6

5

6

Peritonitis

Replies received

177

122

100

85

78

83

Cause of death reallocated as:

TB of peritoneum

3

4

5

3

3

Syphilis

1

1

Cancer

6

2

4

2

3

1

Gastric and duodenal ulcer

9

5

6

7

10

1

Appendicitis

21

21

11

12

5

15

Intestinal obstruction

3

8

6

8

4

6

Disease of the female generative organs:

6

7

7

15

5

Bacilliary dysentry

1

Puerperal sepsis

4

10

5

4

3

3

Total recorded deaths from puerperal sepsis for each year:

1,026

1,184

1,157

1,243

1,050

1,596

Deaths reallocated to puerperal sepsis (totals):

13

19

8

10

8

9

Deaths reallocated to puerperal fever as a percentage of total deaths from puerperal fever (percentage):

1.3

1.6

0.7

0.8

0.8

0.6

Source: Registrar General's Statistical Reviews for the appropriate years.

In 1923, the Registrar General investigated this possibility, revealing a Scotland Yard capacity for detective work. He knew there was a seasonal fluctuation in the death rate from puerperal fever, with a winter maximum and a summer minimum. It was a consistent pattern in all countries from which records could be obtained. He showed that a slight but similar fluctuation in deaths from non-septic puerperal causes also occurred which turned out to be confined to a well-marked seasonal fluctuation in (p.521) deaths assigned to the group of deaths assigned to ‘puerperal phlegmasia alba dolens, embolism and sudden death’. Death from this cause is not seasonal, so there was every reason to believe it was due to a substantial number of ‘hidden’ deaths due to puerperal fever, falsely certified as sudden death and embolism.4 Table A1.1 shows that by the early 1930s the number of hidden deaths due to puerperal sepsis which were discovered and reallocated by the Registrar General's Office had fallen to no more than about 1 per cent of total deaths in that category.

Since the Registrar General's office had always stressed it was impossible to inquire into more than a small proportion of unsatisfactory certificates, it is virtually certain that a certain number of maternal deaths allocated to other causes remain hidden in the lists of the Registrar General to this day. I doubt if it is possible to estimate the number hidden by placement in the wrong category of maternal mortality—in other words the number transferred from one puerperal category to another as described above. At least these did no harm to the estimation of total maternal mortality. But it seemed to me that it should be possible to discover at least some of the maternal deaths hidden in the non-puerperal categories of ‘septicaemia’ and ‘peritonitis’, and that it was worth trying because these would give an estimate of the extent to which the MMR as a whole was distorted. (Incidentally, the enquiries of successive Registrars General suggest that when maternal deaths were hidden, a large majority found a home in ‘septicaemia’ and ‘peritonitis’, and not elsewhere.) The hunt for these hidden deaths was conducted by the following method.

The method is based on searching for an excess of deaths amongst women of childbearing age in the two categories, ‘peritonitis’ and ‘septicaemia’, and it was carried out by searching the Registrar General's Decennial Reviews from the 1880s to 1950. In both categories, if there were no ‘hidden’ maternal deaths one would expect to find either equal death rates for both sexes, or an excess of male deaths. In particular, there should have been an excess of male deaths due to septicaemia because of greater exposure to accidents and injuries which turned septic. There was in fact a considerable excess of male over female deaths due to septicaemia at all ages and in every decennium. In the 1920s, for example, the number of female deaths in this category expressed as a percentage of male deaths was 52 per cent in ages 5–14, 59 per cent in ages 15–44, and 50 per cent in ages 45–64. I drew a blank in this part of the hunt, and I concluded there was no evidence to suggest that a substantial number of puerperal sepsis deaths remained hidden in the columns allocated to this category.

When the distribution of deaths in the category ‘peritonitis of unstated origin’ was examined, the result was quite different and much more complex. First, however, it is necessary to note that the total deaths allocated to this category fell very steeply in the early part of this century, as Table A1.2 shows. There is a simple explanation for the enormous fall in deaths allocated to this category. The three most common conditions which could lead to death from peritonitis were appendicitis, gastric ulcer, and duodenal ulcer. Appendicitis and gastric ulcer were only described and recognized as common diseases at the end of the nineteenth century; duodenal ulcer not until the early twentieth. Appendicitis and gastric ulcer were first tabulated as a separate cause of death in 1901, duodenal ulcer in 1911. Previously, deaths due to these diseases which were accompanied by signs of peritonitis (as many were) were certified simply as (p.522)

Table A1.2 Total Deaths in the Category ‘Peritonitis of unstated Origin’ from 1881–90 to 1950 Showing the Fall in the Number of Deaths in this Category as a Result of the Recognition of Appendicitis and of Gastric and Duodenal Ulcer as Disease Categories in the Early Years of the Twentieth Century

Decennium

Total deaths

1881–90

23,123

1891–1900

22,247

1901–10

8,884

1911–20

4,807

1921–30

3,851

1931–40

3,265

1941–50

1,592

Source: Registrar General's Reports, Decennial Supplements.

deaths due to ‘peritonitis’. This fall was therefore a reflection of the change in nomenclature, not a change in the prevalence of diseases which caused peritonitis.5 Before 1901, buried in this huge category of deaths labelled as ‘peritonitis’, was a small number of deaths due to puerperal sepsis. They might be more numerous then than later; but they would be less obvious because they were obscured by the size of this category. After 1911—and gradually rather than suddenly because it takes time for new diagnostic categories to be accepted—deaths due to puerperal sepsis would be more obvious as an excess. This can be seen in Table A1.3 which shows an excess of deaths amongst women of childbearing age in the category ‘peritonitis of unstated origin’. The column for the age-group 15–44 shows a steep fall in the total number of deaths (here expressed as total deaths per million living in that age-group) but a rise in the excess of female deaths over male. In the older age-groups, there is a constant excess of male deaths except for ages 45–64 in the 1890s which was probably due to the well-known excess of gastric ulcer in women in this group; when gastric ulcer became a certifiable cause of death, that excess disappeared.

In the decade 1901–10, the excess of female deaths per million population was forty-seven. Because the category as a whole was so large, however, the excess expressed as a ratio of female deaths divided by male was only 1.34, By the decade 1921–30, the excess had fallen to twenty per million, but total deaths in this category had become so small that the ratio had increased to 1.47.

How do we know that this excess was due to hidden deaths from puerperal sepsis and not some other cause? We can never, of course, be certain. But the hypothesis is supported by an analysis of the age-distribution of the excess of female deaths over male within the age-group 15–44 which showed that the excess was greatest between the ages of 20 and 35, the ages in which most births occurred. (p.523)

Table A1.3 Deaths from ‘Peritonitis of Unstated Origin’. Death Rates per Million Living in each Sex and Certain Age Groups, and the Ratio of Female Deaths to Male. Decennial Periods 1891–1900 to 1931–40. England and Wales

Age groups

15–44

45–64

65 +

1891–1900

Total deaths per million living

Males

502

1,168

2,530

Females

511

1,193

1,995

Ratio, female deaths/male deaths

1.02

1.02

0.79

1901–10

Total deaths per million living

Males

139

379

989

Females

186

326

759

Ratio, female deaths/male deaths

1.34

0.86

0.77

1911–20

Total deaths per million living

Males

63

204

387

Females

94

158

269

Ratio, female deaths/male deaths

1.49

0.77

0.69

1921–30

Total deaths per million living

Males

44

115

219

Females

64

86

143

Ratio, female deaths/male deaths

1.45

0.75

0.65

1931–40

Total deaths per million living

Males

37

104

174

Females

46

67

94

Ratio, female deaths/male deaths

1.24

0.64

0.54

Source: Registrar General's Decennial Supplements.

Assuming that the total excess of female deaths over male in the category ‘peritonitis’ consisted of hidden puerperal sepsis deaths, how many were there in comparison to total deaths from puerperal sepsis? The answer is given in Table A1.4. This suggests that hidden deaths due to puerperal sepsis—that is, deaths due to puerperal sepsis hidden and not discovered by the Registrar General in the category ‘peritonitis’—may have been no more than 3 per cent at the end of the nineteenth century, falling to 2 per cent by the 1930s. By 1950, there were very few hidden in this manner. The excess of female deaths in the category ‘peritonitis’ during the quinquennium 1946–50 was only fifteen and fell to nil soon afterwards.

It is unlikely that the tables to this appendix show the whole number of hidden (p.524)

Table A1.4 Total Deaths from Puerperal Sepsis (including Septic Abortion), and the Excess Deaths of Females over Males in Age Groups 15–44 in the Category ‘Peritonitis of Unstated Origin’ England and Wales, for all decades from 1891–1900 to 1931–40

Decade

Total deaths from purerperal sepsis

Excess deaths in females aged 15–44 from ‘Peritonitis of unstated origin’a

2nd Column as a percentage of 1st Column

1891–1900

21,605

639

2.9

1901–10

16,270

595

3.6

1911–20

12,460

406

3.2

1921–30

11,612

290

2.5

1931–40

7,722

158

2.0

(a) Expressed as total deaths and as a percentage of registered total deaths from puerperal sepsis. Source: Registrar General's Decennial Supplements.

maternal deaths. At best one can guess that the total number of ‘hidden’ puerperal sepsis deaths in the twentieth century seldom exceeded 5 per cent of the correctly certified number. Since deaths from puerperal sepsis amounted, in round figures, to about half of total maternal mortality, the extent to which the MMR was understated because of hidden maternal deaths was probably around 2.5 per cent by the 1920s, but may have been higher—possibly nearer to 5 per cent in the late nineteenth century. There is therefore no reason to believe that deception and carelessness by certifying doctors led to a gross distortion of the MMR as a whole. Up to the mid-1930s, puerperal fever was usually the most common cause of maternal deaths (possible exceptions include certain southern States of the USA such as Virginia). Usually the published records show puerperal fever as the cause of about 40 per cent (or a little less) of total maternal deaths. Because of the phenomenon of hidden puerperal fever deaths I am inclined to suspect that 50 per cent would be nearer the truth.

Notes:

(1) Supplement to the 55th Annual Report of the Registrar General (1895).

(2) 61 st Annual Report of the Registrar General for 1898.

(3) Registrar General's Statistical Review for 1925 (1927).

(4) Registrar General's Statistical Review for 1923 (1925), 90–6.

(5) Registrar General's Statistical Review for 1926 (1928).