Data and Methods

Canada in the mid-1980s, we know virtually nothing about contraceptive use among Canadians during the 1990s. 2 The 1995 General Social Survey (GSS-95) collected detailed information on contraceptive use, as well as fertility and union histories, from a nationally representative sample of Canadian women and men aged 15 and older. The main objective of this article is to use these data to update our knowledge of contraceptive use in Canada in the 1990s. Thus, we provide a descriptive account of contraceptive practices among women and men of all marital statuses and focus on changes in their contraceptive choices over the past decade. Moreover, as few fertility surveys have collected information on men's contraceptive use, the GSS-95 provides a rare opportunity to study men's contraceptive behavior. Data The 1984 CFS was the only in-depth national fertility survey conducted in Cana-da. The target population was all women aged 18–49, regardless of their marital status. The survey excluded women living in the Yukon and Northwest Territories,* institutionalized women, women living C ontraception should meet the physical , mental and social health needs of individuals throughout their lives. 1 While the wide availability and low cost of contraception have contributed to the decline in Canadian fertility over the past two decades, recent concerns regarding sexually transmitted diseases (STDs), including HIV and AIDS, have complicated contraceptive issues. Thus, contemporary contraceptive practice is often undertaken both to regulate fertility and to provide protection from STDs. There has been relatively little scientific scrutiny of Canadian contraceptive practices , however. The 1984 Canadian Fertility Survey (CFS) was the first national study of the fertility of women in their childbearing years in Canada. Prior to this survey, patterns of contraceptive use had not been extensively researched in Cana-da, nor had the issues of HIV and AIDS been brought to general public attention. While the 1984 CFS provided an extensive account of women's contraceptive use in in households without telephones and those unable to speak either English or French. Telephone survey methods were used to collect the data. A total of 5,315 women aged 18–49 completed interviews, a 70% response rate. In the CFS, women were asked to provide detailed information about their childbearing intentions and experiences, contraceptive practices and other socioeconomic characteristics. Information regarding sexual activity was not collected, as it was deemed too sensitive for telephone interviews. 3 The GSS-95, which was administered by Statistics Canada, was the …


RESULTS
The correlation coefficient for Northern Ireland male births with female births, r = 0.9886, was exceptionally high and significant at p < 0.05. The linear relationship between the number of female births (y) and the number of male births (x) was y = 0.901084x + 598.6523 with a standard deviation from regression of 196.52. From this equation the predicted number of female births in 1978 for an x value of 13168 is 12464 with 95 per cent confidence limits of 12416 to 12514. These limits do not include the value 13071 (Table) observed in 1978, and so we conclude that the observed number of female births significantly exceeds that expected from the overall relationship. The correlation coefficient for Northern Ireland sex ratio with year, r = 0.2570, was small, but was nevertheless significantly different from zero at 5 per cent level, suggesting an increasing sex ratio (y) over the period observed. The scatter diagram showed that this trend was somewhat erratic but was nonetheless positive. The relationship was y = 0.206868x + 1055.13 where x was a code for the year, with code 1 for 1913, code 2 for 1914, etc., up to code 66 for 1978. The standard deviation from regression was 15.05. It was very obvious from the scatter diagram that the value of the sex ratio for 1978 should be treated as an outlier. The predicted ratio for 1978 is 1069 with 99.9 per cent confidence limits of 1018 to 1119. The observed ratio, 1007, is therefore significantly different from the predicted value at the 0.1 per cent level.
The sex ratio of births in the 26 Local Government Districts in Northern Ireland shows considerable variation between districts and within districts from year to year. A sex ratio less than 1000 occurred in 3 districts in 1974, 4 in 1975, 5 in 1976, and 4 in 1977; in 1978 a ratio less than 1000 occurred in 10 districts, 8 of which, including Belfast, were in the eastern half of the country.
In parallel with other parts of the British Isles, Northern Ireland experienced a falling birth rate from a peak of 23.6 live births per 1,000 of the population in 1964 to a trough of 16.5 in 1977. A rise was expected in 1978, and the rate did rise to 17.1. In the event, however, between 1977 and 1978 the number of male births rose by only 14 while the number of female births rose by 788. It seems that the low sex ratio in 1978 may reflect a deficit of male births rather than an excess of females. The fall in the sex ratio did not begin during 1977 when the ratio in the last quarter was 1076, nor extend into 1979 when the ratio in the first quarter was 1067.
In view of reports by James7 for England and Wales and Renkonen8 for the United States of America that sex ratios decline regularly with increasing maternal age, we looked at the distributions of maternal age in Northern Ireland in 1977, when the ratio was high, and in 1978 when it was low. We found that the distributions were significantly different, 1978 having proportionately fewer mothers aged under 30 and more over 30. We then correlated the sex ratio with maternal age for each of the five years for which data were available, 1974 to 1978. Only one year, 1976, showed declining ratios with increasing age, the other four years having correlation coefficients opposite in sign and supporting the contrary trend. We also looked at the distributions of duration of marriage and of parity in 1977 and 1978. Although these variables differed significantly in distribution there was no evidence that either was correlated with the sex ratio of live births.
The mean sex ratio of Northern Ireland stillbirths from 1961 to 1977 was 975 males per 1,000 females. In 1978 the ratio, 1113, was the highest recorded except for 1123 in 1977. The numbers of stillbirths in 1978 were 128 males and 115 females, the lowest ever recorded, and so the high ratio accounts for only a small part of the deficit of males. No information can be obtained about the sex ratio of spontaneous abortions. Northern Ireland does not have an equivalent to the Abortion Act 1967 in Great Britain so that comprehensive data on induced abortions in Northern Ireland women are not available. A steadily increasing number of Northern Ireland women obtain abortioneach year in Great Britain; in 1978 the number aborted in England and Wales was 13019, but the sex of the aborted foetuses is not known. A corresponding fall in the sex ratio of births in 1978 did not occur in England and Wales10, where the ratio was 1060, nor in Scotland"1, where the ratio was 1058, neither ratio differing greatly from that expected. In the Isle of Man12 the ratio was 1122.
In the Republic of Ireland on the other hand the sex ratio in 1978 was 1038, the lowest recorded in the past 20 years, and the ratio in 1979 was back to normal at 1060. The ratio in 1978 in the two cities, Dublin and Cork, and in the provincial areas of the Republic is variable, but the three counties of the province of Ulster within the Republic, Monaghan, Cavan, and Donegal, show the very low ratio of 1005 (Table).

DISCUSSION
The sex ratio of births shows considerable variability from place to place and from year to year when small areas are considered, but the main parts of the British Isles this century appear to have had very similar mean ratios, about 1060 males per 1000 females. The occurence in one region of the British Isles with over 26,000 births per annum of a ratio well below the lower 99.9 per cent confidence limit gives rise to speculation as to a possible cause. If the cause lay in some temporary change in the environment, such as an air-, water-, or foodborne contaminant or a drug having a selective effect on male spermatozoa, male conceptions or male embryos, this should in principle be discoverable and could suggest an approach to control of the sex of infants and livestock. Colleagues in agricultural and veterinary science with whom we discussed the matter had not noticed any disturbance in the sex ratio of animals, but they thought that, in the absence of records for animals comparable to human births registration, all but the grossest changes in sex ratio would have remained undetected. We have been unable to think of a likely explanation for the 1978 ratio and have found no associated conditions worth investigating. The fact that the low ratio was confined to a single calendar year does not make the matter easier to understand, but made us look for some form of artefact as a possible cause. This we did not find, and the low ratio is confirmed by the birth notification in Northern Ireland which are made quite independently of the registrations of birth. The departure from the expected sex ratio in 1978 will be noticed when this cohort of Northern Ireland children reaches school age and a number of boys sufficient to fill several schools are missing mainly from the east of the province.
We have here a phenomenon affecting the whole of Ireland, but the north more than the south, and not affecting Great Britain. The finding that the three Ulster counties outside Northern Ireland had a ratio even lower than Northern Ireland's 1007 is intrigueing and makes us reluctant to accept this as merely an extreme example of random variation.
SUMMARY Northern Ireland has a mean sex ratio of live births of 1063 males per 1,000 females. In 1978 the very low ratio 1007 was recorded. This is significantly different from the predicted value at the 0.1 per cent level. A low ratio occurred also in the Republic of Ireland but not in Great Britain.