The next few chapters describe the diseases peculiar to women. Women produce more than half the food in the non-Muslim parts of India and Nepal, and up to 80% of it in Africa. Yet their health has been much neglected. Half a million of them die in childbirth each year, 99 per cent in the developing world. The chances of this happening depend on how often a woman becomes pregnant (the concern of this chapter), and how dangerous each pregnancy is, as measured by the maternal mortality ratio (MMR). This is the number of maternal deaths expressed per 100,000 live births. A maternal death is: ''The death of a woman while she is pregnant, or within 42 days of the termination of pregnancy, irrespective of the duration and site of her pregnancy, from any cause related to, or aggravated by her pregnancy, or its management, but not from any accidental or incidental causes'. (ICD[nd]9) In the Maternal Mortality Rate, which is not discussed here, the denominator is 10, 000 women in the 15[nd]49 reproductive age group.
The MMR varies widely. It used to be high all over the world. In some communities in Africa it is still 1000 or more, which means that a mother has a 1 per cent chance of dying from pregnancy-related causes, or a 10 per cent chance if she has 10 pregnancies in her lifetime. In the developing world as a whole it is about 450, but in the developed world it has fallen to 30. The result is that a mother who delivers in Bangladesh has a 400 times greater chance of dying than a mother in Scandinavia.
The deaths of mothers are more difficult to prevent than those of their children. Apart from family planning, there are no simple ways of preventing a mother's death in the way that oral rehydration and immunization can save a child. Mothers die from abortions and ectopic pregnancies, eclampsia, anaemia, haemorrhage, obstructed labour, ruptured uteri, or sepsis, and often from more than one of these causes. Mostly, they either die at home, or present late in hospital as unbooked emergencies who have had no antenatal care. Tragically, those who need care are least likely to get it. Saving their lives requires improved education for women and services at three levels: (1) In the community. (2) In clinics and health centres. (3) In adequately equipped and staffed district hospitals. Especially, it needs plenty of well-trained midwives. Establishing all this needs political will. To raise this you will need to know your local MMR. Here is a simple new way of measuring it. Ask adults what happened to their adult sisters and whether or not they died in childbirth. Their mortality experience will be a measure of that of the community as a whole.
SAMPLE SIZE is important, and depends on the expected level of maternal mortality, the average number of 15[+] sisters that respondents will have, and the amount of error you are willing to tolerate. If the expected level of maternal mortality is high (an MMR of about 1000), and most respondents come from large families, you will probably be able to get a reasonable estimate from 2000 respondents, which is the absolute minimum, and is only applicable for least developed rural areas. 4000[nd]6000 is suitable for most other situations, but you will need at least 8000 for urban populations in more advanced developing countries. In the developing world, each household often has three adults who can be interviewed, so that to find 2000 respondents, you may only need to visit 667 households (2000[/]3).]][+2] TOTAL FERTILITY RATE (TFR). You will need this to calculate the MMR. It is the average number of live births that a mother could expect to have in her lifetime if she experienced the same age-specific fertility rates as mothers now living. You can get it from the UN Demographic Yearbook, or from your Central Statistical Office. Or, much less satisfactorily, you can work it out. You can ask women aged 15[nd]49 how many live births they have had in the last 12 months, and work out the TFR as described below. Alternatively, you can use a range of TFRs, from 4 to 7, which will cover most developing-country populations, and see what answers you get.]][+2] PRESENT AND PAST FATALITY AND FERTILITY. The older your respondents, the longer ago their sisters will have died, the longer ago they will have had babies, and the less ''up to date' your estimate. By using the information from all your respondents aged 15[nd]45, you will get an estimate which reflects the level or maternal mortality over the last 10[nd]12 years. This estimate is likely to be similar to the current one, because in the populations in which this method is used, the MMR is unlikely to have changed much over the previous decade.]][+2]
THE SISTERHOOD METHOD [s7]FOR DETERMINING THE LEVEL OF MATERNAL MORTALITY You will need some interviewers and a calculator. Carefully translate the questions into the local language, and stencil them with separate answer boxes for each respondent. You should be able to get all the answers for each household on a sheet of A4 paper.
Use interviewers with a reasonable level of literacy and numeracy. Pupil nurses are ideal; make the survey an educational project for them. They will need to question at least 2000 respondents (see above), so it will take them several days. Take them into a classroom and explain the method. Then, ask them to visit a reasonably random selection of households, which you think will reflect the population in your district. In each house ask the head of the household to list all adults aged 15[+]; then question each in turn. Try to talk directly to all adults over 15, rather than using another household member to respond on his or her behalf.
(a) ''How old are you?'' Or, ''What is your date of birth?'' Interviewer: check that the respondent is over 15 years old.
(b) Interviewer: What is the sex of the respondent? If the respondent is female, and you are going to calculate the TFR, ask question (c):
(c) ''How many live births have you had in the last 12 months, even if the baby is no longer alive?''
(d) For all respondents, male and female ask: ''How many sisters have you ever had who were born to your mother?''
(e) ''How many of these sisters ever reached the age of 15, including those who are now dead?''
(f) ''How many of these sisters, who reached the age of 15, are alive now?''
(g) ''How many of these sisters are dead?''
Interviewer: Check that the sum of (f) [+] (g) = (e), and sort out any discrepancies. (e) should be equal to or less than (d), it cannot be greater than (d).
(h) ''How many of these dead sisters died while they were pregnant, or during childbirth, or in the 6 weeks after the end of pregnancy?''
For practice, do a pilot study on about 10 households, and adjust your questions as necessary.
THE TFR based on question (c) above, can be derived from the age-specific fertility like this:
(1) Group the female respondents aged 15[nd]49 by 5-year age group, that is 15[nd]19, 20[nd]24, [...], 45[nd]49.
(2) For each 5-year age-group, divide the number of live births in the last 12 months by the number of all women for that age (not just those giving birth). This will give you an estimate of the age-specific fertility rate for each age-group as a proportion.
(3) Add the proportions together and multiply the total by 5 (for the 5 years in each age-group) to get the TFR, which will probably be between 4 and 7.
MMR. Work out a table like that in Table 15-1. You will need to make two corrections:
(A) Respondents in the age-groups 15[nd]19, and 20[nd]24, can expect some more sisters to reach the age of 15. So work out the average number of sisters reaching age 15 for the respondents over 25. In Fig. 15-1 this was 1.54, that is each respondent above the age of 25 had 1.54 sisters reaching age 15. Multiply the number of respondents in the age-groups 15[nd]19 and 20[nd]24 by this figure to fill in the first two entries in column (3). These have been marked with a star. For example, in Fig. 15-1 there were 320 respondents aged 15[nd]19. When questioned they said they had 325 sisters who reached age 15. They could however expect some more, so 320 was multiplied by 1.54 to give 493, the first figure in Column (3).
(B) Column (5) is an adjustment factor based on a typical developing-country population model. Respondents aged 50, or more, will be referring to sisters who will have been subjected to the full risks of a lifetime of childbirth. Younger respondents will be referring to sisters who have only been exposed to part of that risk, so a correction has to be made to the reported number of sisters reaching the age of 15 to calculate ''sister units of risk exposure' in column (6). To get column (6), multiply column (3) by column (5). Column (5) is ''given' and is always the same.
Sum column (4), which is the total number of maternal deaths reported by your respondents, and sum column (6).
Divide the sum of (4), by the sum of (6), to give you the total lifetime risk of maternal death. It was 0.048 in Fig. 15-1, which means that a woman reaching the age of 15 has nearly a 5%, or about a 1 in 21 (1[/]21 = 0.048) chance of dying of pregnancy- related causes during her reproductive life.
To calculate the MMR divide the estimate of the total lifetime risk of maternal death by your estimate of the TFR, and multiply by 100,000.
For example, the TFR for the population in Fig. 15-1 was 6. So, 0.048[/]6[*]100,000 = 800. The MMR calculated by this method is therefore about 800. As noted above, strictly speaking, this figure refers to period 10[nd]12 years before the data were collected. However, in situations where this method is used, it is unlikely to have altered significantly.
Try to publish your findings somewhere. Inform the appropriate government office about your results, and include details of your study design, sample size, the questions you used, and any problems you found.
DIFFICULTIES [s7]IN MEASURING THE LEVEL OF MATERNAL MORTALITY If a RESPONDENT IS UNAVAILABLE, ask the household member who is most likely to know about the respondent's sisters.
Alternatively, omit missing respondents, and visit more households to collect the number of respondents you need. But make sure that the time of day you call would not exclude certain groups, such as women working in the fields.
If for any reason you CANNOT GET AN ADEQUATE SAMPLE, consider carrying out the study jointly with a neighbouring health unit.
If a HOUSEHOLD SURVEY IS IMPRACTICAL, consider questioning all adults attending a fixed health facility, such as mothers attending an immunization clinic. They will be a selected group, but their sisters will probably be less so.
If questions on pregnancy-related matters are CULTURALLY SENSITIVE, confine your questions to female respondents aged 15[+].
If you are worried about the RELIABILITY of your data: (1) Calculate the MMR separately for males and female respondents. It should be similar. If it is not, the sample may be too small, or the responses may be biased between the sexes. Women may be better informed about the circumstances of the death of their sisters than men. (2) Compare the MMR for the 15[nd]49 age group (who will be younger and better able to remember past deaths) with that for all 15[+] respondents. The figures should not be very different. The higher of the two is likely to be nearer the truth. The figure from the 15[nd]49 age group will probably be the most reliable, provided more than [3/4] of your respondents are in this age group. Try to get at least half your sample between 25[nd]49, without too many [lt]25 or [mt]50.
Fig. 15-1 MATERNAL MORTALITY BY THE THE SISTERHOOD METHOD, as determined for the Gambia in 1987. Kindly contributed by Wendy Graham.
Graham W, Brass W, Snow R, ''Estimating maternal mortality in developing countries', Lancet 1988;i:416[nd]7.[-3] Graham W, et al. ''Indirect Estimation of Maternal Mortality; the Sisterhood Method'. Centre for Population Studies. The London School of Hygiene and Tropical Medicine. London WC1E 6AZ.[-3] Graham W, and Airey P, ''Measuring the maternal mortality, sense and sensitivity'; Health Policy and Planning 1987 (2);4:323[nd]333.