Radical
Statistics
Health and the gender agenda: what do official statistics tell us?
Alison Macfarlane
The subject of differences between the health of men and women has inspired a considerable amount of research. This article makes no pretensions to review it systematically. Instead I want to raise questions about what official statistics can and cannot tell us about differences between the health of men and women and about problems specific to one sex or the other. It draws on and updates some parts of a much longer article published in 1990 in Women's health counts (Macfarlane, 1990).
The term 'official health statistics' is a loose one. It usually takes in statistics derived from the registration of births and deaths, statutory and voluntary notification of events such as diagnosis of cancer or cases of communicable disease, claims for benefits and statistics about the National Health Service (NHS) . These include both aggregated administrative statistics and statistics derived from individual patients' contacts with the services. Most of these tell us more about death, illness and the use of the NHS than they do about health in a positive sense, but some questions are asked about health in government funded surveys and one will be included in the 2001 census. When data are based on records of individual deaths, birth certificates, hospital stays, consultations with general practitioners or replies to surveys, they can often include comparisons between men and women.
Births and infant mortality
The sex ratio of live births, that is the number of boys divided by the number of girls, was of considerable interest to statisticians and others during the nineteenth century. Numbers of live births have been tabulated by sex since registration began in England and Wales in 1837 and trends are shown in Figure 1. In the nineteenth century, the Registrar General's annual reports often commented on trends in the sex ratio and interest in the topic kept surfacing during the twentieth century (Macfarlane and Mugford, 2000; James, 1987).
Figure 1: Sex ratio of live births, England and Wales, 1838-1998
The sex ratio decreased through the nineteenth century as birth registration became more complete, then increased during the first half of the twentieth with a major peak towards the end of the first world war and an increase in the ratio during the second world war.
Many factors, including the parents' racial origins, the season of birth, birth during wartime, the marital status of the parents, birth order and the father's age are apparently associated with small differences in the sex ratio, although there is no clear agreement about this (James, 1987; Maconochie and Roman, 1997). Other factors seem to be associated with larger differences (James, 1987). Such factors include the timing of insemination within the mother's menstrual cycle, some forms of disease in the parents and exposure to harmful substances in the environment or at work.
Both stillbirth and infant mortality rates are higher for boys than girls as Figures 2 and 3 show, although the differences have all but disappeared for stillbirths. Some congenital anomalies, notably anencephaly, when a baby's brain does not develop so it cannot survive, are more common among female than male foetuses. This led to a narrowing of the gap between boys and girls as numbers of stillbirths attributed to other causes have declined.
Figure 2: Stillbirth rates by sex, England and Wales, 1928-98
Figure 3: Infant mortality rates by sex, England and Wales, 1928-98
The higher infant mortality rate among boys is somewhat paradoxical. On average boys are heavier than girls, as the comparison of cumulative distributions in Figure 4 shows, so they would be expected to have a lower mortality rate. This suggests that the association between mortality and birthweight is different for boys and girls and data about them should be analysed separately more often than they currently are.
Figure 4: Cumulative birthweight distribution by sex, England and Wales, 1998
Mortality at older ages
The differences in infant mortality are not unusual. Mortality rates for males are higher than those for females in all other age groups in England and Wales, as in other developed countries. Figure 5 and Table 1 show that this was the case in the early 1990s. It was also the case among adults a century earlier and in the older age groups before that.
Figure 5: Mortality rates by age and sex, England and Wales, 1891-95 and 1991-95
Click HERE for Table 1: Death rates per 1,000 population and infant death rates per 1,000 live births, England and Wales, selected five year periods,
In younger age groups, the picture was different. Throughout the nineteenth century, and in the first two decades of the twentieth, the death rates for girls in the 10-14 age group were higher than those for boys. This was also true for 5-9 year olds in the 1890s and early 1900s. This excess mortality among girls was associated with deaths from communicable diseases, which were a major cause of death in these age groups at the time. The pattern was similar in other European countries for which data are available and lasted into the 1930s and 1940s in some of them (Tabutin and Willems, 1994). It reflected the relative disadvantage experienced by girls in this age group. Although this mortality difference is past history for those of us who live in developed countries, it is not the case everywhere. In some countries of the world, for example in North Africa, girls are still more likely than boys to die in childhood (Tabutin, 1992).
Pregnancy, including illegal abortion, was a major cause of death among women of childbearing age in England and Wales up to the mid-1930s, as it still is in much of the world today. While it probably contributed to the excess mortality among women aged 25-34 in the mid nineteenth century (see Figure 6), its impact is likely to have been underestimated (Macfarlane and Mugford, 2000).
Figure 6: Sex ratio of mortality by age, England and Wales
In some cases, pregnancy may not have been mentioned on the death certificate, despite checks made by the General Register Office from 1881 onwards. In others, the death may have been classified in other ways. For example, until the mid 1920s, deaths from illegal abortion were classified with deaths attributed to accidents and violence.
Even if pregnancy is mentioned on a death certificate, it may not be coded as the 'underlying cause' of death. In England and Wales, only this 'underlying cause' was included on the computerised death record up to 1992. Since 1993, all the causes written by the doctor on the death certificate have been computerised. This technical change has made it possible to identify deaths attributed to other causes, including suicides and violent deaths among pregnant women, provided of course that the information was written on the death certificate in the first place (Department of Health et al, 1998).
Use of hospital services
Conditions related to reproduction are much more readily visible among causes of hospital admission among young women. Figure 7 compares rates of episodes of in-patient or day case care in NHS hospitals in England for men and women. An episode is a period of care under a consultant, so the same person can have several episodes in different departments during the same in-patient admission and a number of successive in-patient and day case admissions for the same condition. Figure 7 shows much higher rates for women than men in the 14-19 and 20-44 age groups, but much higher rates for men in the older age groups. To what extent might these reflect differences in health? In Figure 8, admissions for abortion, miscarriage and conditions related to pregnancy and childbirth are excluded from the data for women. This suggests much smaller differences between men and women for the remaining conditions, which are more likely to be related to ill-health.
Figure 7: Hospital episode rates, all diagnoses, by sex and age, England, 1995/96
Of course, use of hospital services is not a measure of ill health, particularly in view of the extent to which the one person can be counted several times if they go into hospital more than once for the same illness. There has been an increase since the early 1990s in the extent of data collected about the health of the population through the government programme of health surveys in each of the four countries of the United Kingdom and through surveys funded from other sources. These avoid the biases which are inevitable in data based on the use of services. The fact that they are based on questionnaires plus in some cases clinical measurements enable very much more focussed questions to be asked about public health.
On the other hand, there is no guarantee that gender bias will be eliminated from the ways in which the questions are framed. As they are based on relatively small samples, this means that they cannot cover rare conditions which are specific to one sex or the other. In addition, as they are cross-sectional surveys, they include few pregnant women in their sample. This means that they are not the best way of collecting data about the immediate health consequences of pregnancy. It would also not be surprising if sensitive issues, such as domestic violence, were under-reported here as in many other data collection systems.
Thus despite the fact that sex is usually recorded in data about health and health service utilisation, the data cannot necessarily be taken at their face value. Despite the considerable amount of information available to compare men and women, there are still important gaps in our knowledge, even in developed countries such as those which make up the United Kingdom.
REFERENCES
Department of Health, Welsh Office, Scottish Office Department of Health, Department of Health and Social Services, Northern Ireland (1998), Why mothers die. Report on confidential enquiries into maternal deaths in the United Kingdom, 1994-1996, London: The Stationery Office.
James. W. H. (1987), 'The human sex ratio. Part 1. A review of the literature', Human Biology; 59: 721-752.
Macfarlane A. J. (1990), 'Official health statistics and women's health and illness', in Roberts H, ed. Women's health counts, London: Routledge.
Macfarlane A. J. and Mugford M. (2000), Birth counts: statistics of pregnancy and childbirth, Text, second edition, London: TSO.
Maconochie N. and Roman E. (1997), Sex ratios: are there natural variations within the human population? British Journal of Obstetrics and Gynaecology; 104: 1050-1053.
Office for National Statistics, (1999), Mortality statistics, general, England and Wales, Series DH1, no.30, London: TSO.
Tabutin D. (1992), 'Excess female mortality in Northern Africa since 1965: a description', Population: An English Selection; 4: 187-208.
Tabutin D.and Willems M. (1994), 'La surmortalité des filles jusqu'en 1940', in: Masuy-Stroobant G, Gourbin C, Buekens P, eds. Santé et mortalité des enfants en Europe: inégalités sociales d'hier et d'aujourd'hui. Actes de la Chaire Quetelet, Louvain-la-Neuve, 12-14, Sept. Louvain-la-Neuve: Academia-Bruylant, 1996.
Alison Macfarlane
40 Warwick Road
St Albans
Herts AL1 4DL
Tel/fax 01727 852111
E-mail Alison.Macfarlane@perinat.ox.ac.uk
