Women aged 60-64
CHANGING PATTERNS OF MORTALITY IN BAN PONG
3.4 Sources of data
A major problem in the study of mortality in small communities in developing countries is the paucity of data. Thailand is no exception here. Although death rates are available for the country as a whole, there is no official breakdown for smaller administrative areas.
This gap is, to some extent, filled by the Vital Registration system.
However, the degree of under-registration of deaths appears to be even greater than for births. In I960, it was estimated that overall only 60% of deaths were registered, and although this may have
improved to some extent in recent years, it remains a serious problem (Caldwell, 1967). Furthermore, under-registration tends to be biased by sex and by age, with females and young infants being the most likely to be omitted from the records.
Anthropologists who have taken an interest in the demographic patterns prevailing in their field communities, have found various ways of solving the problem. Firth (1957), for example, was able to compare two censuses taken in Tikcpia in 1929 and 1952, to determine the number of deaths which had occurred in the interim. Macfarlane (1976) employed the same method in his study of Gurung mortality, using two village censuses taken 11 years apart, I had hoped to use McDaniel's census data for Ban Pong taken in 1967 for this purpose, but unfor
tunately the original questionnaires had been destroyed and only the cumulative data for the village were available to me (McDaniel, 1967:
Appendix 1). Even so, this method involves a number of serious biases which limit its value. Firstly, it is only feasible for use in a comparatively small, non-mobile population or one in which, like
Tikopia, all migrants can be identified by those remaining in the community at the time of the repeat census. Secondly, it is very likely, particularly if the censuses are taken many years apart, that infants born and dying in the interim will be overlooked.
The reconstruction of mortality patterns using genealogies is possible in small communities, but in a village such as Ban Pong, with a population of over 2000, this would have been an excessively complicated and time-consuming task. Macfarlane uses a restricted version of this method, by recording information on the deaths of parents, and the most recently deceased near relative, of each head of household and of his wife. This method has enabled him to draw some tentative conclusions about adult mortality among the Gurungs, but is by no means conclusive (1976:277).
Another method used by Macfarlane is to extract information about deaths from the fertility histories which he collected for selected Gurung women. His data are limited by the fact that the women he selected were the oldest in each household, thus creating a bias in terms of the age distribution of these women, a factor which, as we shall see shortly, is of some significance in the incidence of infant mortality. In the case of my own fertility survey in Ban Pong, this problem does not arise, since I collected data from every woman in the village who had ever been married, apart from a small number of the oldest women who were too senile to remember the sort of information I was seeking. Even so, there remain a number of biases, inherent in the use of this method, which are impossible to avoid (see Chapter 2,
p. 88 ), but in the absence of any more accurate source of data, this does at least provide some indication of the changing trends in infant and child mortality in the community in recent years.
The final source of data used by Macfarlane and by myself, is the record of all deaths occurring during fieldwork. I was able to extend this method to cover three consecutive years: in my first survey of March 1973, I recorded all deaths reported by the head of each house
hold as having occurred during the preceding 12 months. At the time of my second survey in March 1974, I was able to determine precisely the number of deaths which had occurred in the interim by comparing the successive questionnaires for each household. Finally, a survey conducted by my field assistant in March 1975, enabled me to
calculate the deaths which had occurred in the preceding twelve months. Clearly the data for the middle period (November 1972 to August 1974), when I was actually living in Ban Pong and able to personally record each death as it happened, are likely to be the most accurate.
3.5 Mortality in Ban Pong: The Registration Data
Registration data on deaths occurring in Ban Pong were available for the years 1965 to 1973. The estimated crude death rates for each year are given in Table 18. As in the calculation of crude birth rates for Ban Pong presented in the previous chapter (Table 17, p.119 ) , the total village population for some years has been estimated.
Table 18
Estimated Crude Death Rate, Ban Pong, 1965-1973 Year Total Population Registered Deaths CDR/1000
1965 1885* 16 8.5
The range of annual crude death rates for Ran Pong compare closely with those found for Mae Taeng District and Inthakhin Sub-district for the same period (Chapter 1, Tables 2 and 3, p.79) . Breakdown of deaths by age and sex were, unfortunately, not available for Ban Pong, or for the District and Sub-district, other than for the year 1973.
Since only 18 deaths were recorded for Ban Pong in that year, analysis by age and sex has not been attempted, the numbers being too small to be of any significance. However, deaths recorded in 1973 for Inthakhin Sub-district have been analysed in this way, since it might reasonably be assumed that mortality patterns over time would not vary greatly within this comparatively small population . Figures are approximate, 1 since a breakdown of the total sub-district population by age and sex was not available, so I have used the proportions given for the
District population as reported in the 1970 Census.
^ In fact crude death rates for the thirteen villages in Inthakhin Sub
district in 1973 ranged from 3.6/1000 to 15.6/1000'. Although this could reflect actual variations in mortality from one village to another, or differences in their age structure, I think it is to a large extent the result of eratic reporting of deaths, and random fluctuations in communities whose population sizes vary from under 300 to over 2000.
Table 19
Estimated.Age/Sex-Specific Death Rates, Inthakhin Sub-district, 1973 (rates per
Several points emerge from an examination of Table 19. First, the age- specific death rates do tend to follow the U-shaped trend mentioned earlier, with rates lowest between ages 10 and 39. Second, although the overall average rates for males and females are more or less the same, there appears to be a bias against reporting deaths of female infants, since it is unlikely that the actual death rate for males under 10 is as much as twice that for females in the same age group.
Third, I would suggest that the rates for all children under 10 are lower than might be expected, indicating a considerable degree of
under-reporting of infant deaths1 . I will return to this point shortly.
Finally, although the overall crude death rate of 7.4 is mere or less the same as the rates for Mae Taeng District and Chiengmai Province for the same year (Chapter 1, Tables 2 and 3, p.79), it would seem to be rather low, given the poor quality of health services in such rural areas of Thailand. This assertion will be discussed further in the following section.
1 Unlike in the case of registration of births, where registration is encouraged by the obligation to provide a birth certificate for all children entering school, there is little motivation to register deaths.
Although officially there are penalties for non-registration of deaths, they are rarely, if ever, imposed and where registration involves
relatives of the deceased making a journey to the home of the sub
district kamnan, the inconvenience can often lead to considerably late, or non-reporting of such events.
A final factor of interest which emerged from analysis of the
Registration data, was the seasonal distribution of deaths. The highest proportion of deaths (37.3%) occurred during the four months of the rainy season, July to October, and the least (27.1%) in the four dry, cool months from November to February, with "35.5% occurring during the hot dry months between March and June. Although it is likely that there
is some degree of error in the recording of the actual months in which deaths occurred, the possible association between climatic conditions and mortality is worthy of note. An increase in death rates during the rainy season was also noted by Macfarlane in his study of the Gurung, a factor which he would attribute primarily to drainage problems 1
(personal communication). However, Macfarlane also reports a consider
able decline in mortality rates from water-borne diseases among the
Gurung in Thak following the installation of a water-pipe (1976:269-270).
At the time of my fieldwork, in Ban Pong there was no running water in the village; all households in the community obtained their water from wells, of which there were 277 in 1973. Considerable progress had been made in the village with the introduction of latrines in recent years .2
In 1973, more than 80% of households in Ban Pong had a latrine either within, or beside their homes, and it was the aim of the local
government-trained sanitation officer to install them in all households in the area.
1 Another possible factor in the seasonal variation in mortality rates could be the greater inaccessibility of medical services for populations isolated during the monsoons. Although Ban Pong is comparatively well- situated in this respect, it was, for example, completely cut off for more than a week in July 1973 when the'bridge at Mae Taeng was swept away by exceptionally violent flood waters.
2 Cassen (1978:208) mentions the considerable benefits to public health which can be obtained by this single measure. However, I would suggest that in areas such as Ban Pong, where water for domestic use is drawn from quite shallow wells, and where latrines are constructed over cesspits, the risk of pollution of the water supply, particularly during the monsoon season, is high.
140.