Appendix

The Problems of Comparative Statistics Certain statistical problems surround the interpreta­tion of the statistics on maternal mortality and inter­national comparisons. In England and Wales before the compulsory registration of stillbirths in 1927, the denominator births meant live births only.

The use of the classification based on the Interna­tional List of Diseases was important when it was adopted by England and Wales in 1911. Previously, deaths due to toxemia of pregnancy (but not deaths from eclampsia) were entered under diseases of the kidney; after 1911 they were entered as maternal deaths. Subsequent revisions of the international list made little difference to the calculation of total maternal mortality, but an appreciable difference in the way maternal deaths were classified.

A major problem in international comparisons was that of associated, or, to use the modem term, indirect deaths. Simply stated, should the death of every pregnant woman be recorded as a maternal death, regardless of the cause? There was no diffi­culty in identifying “true” maternal deaths such as those due to postpartum hemorrhage, puerperal fever, or ectopic gestation; and deaths that had nothing to do with the pregnancy or general health (e.g., the death of a pregnant woman in a road or railway accident) were generally excluded. Difficul­ties arose, however, when a woman died in the course of an otherwise normal pregnancy or labor from influenza, or some preexisting condition such as heart disease, tuberculosis, or nephritis.

Perhaps the most difficult problem is that of “hid­den” deaths - deaths from puerperal fever and abor­tion that are hidden because of the opprobrium sur­rounding them. A death from puerperal fever was, above all other deaths, one for which the physician or nurse was likely to be blamed. There were there­fore strong motives for a physician to register a death like this in such a way that the authorities would not know it was a maternal death. In the late nineteenth and early twentieth centuries, it was not unusual for doctors to certify a death from puerperal sepsis as being due to peritonitis or septicemia with­out mention of childbirth. These conditions were cited because they were, in fact, the immediate causes of death in puerperal sepsis.

The true cause of death in cases such as these could be determined, however, both by asking the offending doctor outright whether childbirth had been involved or by searching the death records for an excess of deaths from peritonitis or septicemia among women of childbearing age. Deaths from both causes (“perito­nitis of unstated origin” and “septicemia”) were more common in men than women once deaths from puer­peral sepsis were excluded. This complicated statisti­cal exercise was carried out by government statisti­cians in Britain and the United States in the 1920s. Indeed, this author has carried out in some detail the same exercise using the lists of causes of death from the 1880s to the 1950s. In England and Wales, in the period from 1900 to 1930, apparently incorrect certifi­cation led to an underestimation of the true number of deaths from puerperal sepsis by about 12 percent. The MMRs described by this essay for the period up to the mid-1930s are therefore likely to be an underesti­mation; they are certainly not an exaggeration of the true figures. Needless to say, the exact MMRs of the past can never be known. Nevertheless, the extent of statistical distortion can to a large extent be esti­mated. It is then reasonable to believe that a gener­ally correct picture of MMRs in most developed coun­tries since the 1880s can be obtained and that the picture becomes increasingly accurate through the present century.

Irvine Loudon