On the standardisation of the twinning rate.

In many studies the twinning rate, being strongly dependent on maternal age (and parity), has been standardised according to the maternal age distribution. The direct method requires very informative twinning data for the target population. The indirect method is used when the data for the target population is not sufficiently informative or when the target population is small. We have earlier introduced an alternative indirect technique for standardising the twinning rate. Our technique requires even less of the twinning data. Besides maternal age, parity is an influential factor, and should, if possible, be taken into account. In this study we present the traditional standardisation methods based on both maternal age and parity, we propose a new direct standardisation method and we develop our standardisation methods so that they take into account both maternal age and parity. We apply these standardisation methods to data from Finland, 1953-1964, from St. Petersburg, Russia, 1882-92, from Canada 1952-1967, and from Denmark, 1896-1967. These methods all give results very similar to those for the Finnish data, but the effect of parity is strongest with the direct methods. This may be due to the fact that, among extramarital maternities, parity has a strongly increasing effect on the twinning rate. This may be attributed to a higher reproduction capacity among unmarried mothers. Standardisations of the Canadian and the Danish data also give reliable results. With the St. Petersburg data, however, the different standardisations show notable discrepancies. These discrepancies are compared with Allen's findings.     

Temporal differences in the regional twinning rates in Sweden after 1750.

When twinning rates are studied, maternal age and parity should be considered. Data on parity are seldom available. We studied information about the mean parity, using the gross reproduction rate, the total fertility rate and the crude birth rate. These are strongly correlated with the mean parity. The crude birth rate is more readily available than the gross reproduction rate or the total fertility rate. Earlier studies have shown that it is difficult to model variations in the twinning rate with data for the macrolevel. In this study these findings are explained by theoretical analyses and illustrated by empirical data. Sweden, having the oldest continuous population statistics and high twinning rates, offers excellent possibilities for analyses of the twinning rate. We considered data for the counties of Gotland, Alvsborg and Stockholm and the city of Stockholm from 1749 to 1960 and for Sweden until 1996. For Alvsborg, the twinning rate was low for the whole period, showing no statistically significant decreasing trend. It is mainly about 11-14 per thousand, which is only 50-60% of the twinning rate in Gotland in the 18th century. In Gotland, in the county of Stockholm, in the city of Stockholm and in Sweden as a whole, the decreasing trends in the twinning rate were statistically significant. The decreasing twinning rates converge towards the low twinning rate of Alvsborg. After standardization of the twinning rate, the differences remained and the low rate in Alvsborg could not be explained by maternal age.     

Twinning rates in a rural area of Bangladesh

In this study we investigate the incidence of twin births over a period of 16 years in a rural area of Bangladesh using data from the Demographic Surveillance System of the International Centre for Diarrhoeal Disease Research. Over the study period twinning rates fluctuated between 7.8 and 11.2 per 1000 live births. The twinning rate was strongly correlated with maternal age; the rate for mothers over 35 years of age was about 3 times higher than for mothers younger than 20 years. The variation in twinning rate with maternal age is due to the variation in dizygotic twinning; the rate of monozygotic twinning is almost constant for all ages. Twinning rates were higher in the treatment area than in the comparison area after controlling for maternal age and parity. The rates were lower for monozygotic twinning and higher for dizygotic twinning in the treatment area than in the comparison area. Seasonality was observed for both twins and singletons, but the peak for twinning precedes that for singleton births by more than a month.     

Secular trends of twinning rate in Sweden.

The change in parity and maternal age in Sweden accounts for the decline in the twinning rate from the 19th century to the middle of the 20th century. The later decline in twinning rate cannot be explained by changes in maternal age and parity. Fecundity appears to be higher in mothers of twins than in other mothers.     

Should the statistical analyses of twinning-rate data be improved?

Every statistical model is based on explicitly or implicitly formulated assumptions. In this study we address new techniques of calculation of variances and confidence intervals, analyse some statistical methods applied to modelling twinning rates, and investigate whether the improvements give more reliable results. For an observed relative frequency, the commonly used variance formula holds exactly with the assumptions that the repetitions are independent and that the probability of success is constant. The probability of a twin maternity depends not only on genetic predisposition, but also on several demographic factors, particularly ethnicity, maternal age and parity. Therefore, the assumption of constancy is questionable. The effect of grouping on the analysis of regression models for twinning rates is also considered. Our results indicate that grouping influences the efficiency of the estimates but not the estimates themselves. Recently, confidence intervals for proportions of low-incidence events have been a target for revived interest and we present the new alternatives. These confidence intervals are slightly wider and their midpoints do not coincide with the maximum-likelihood estimate of the twinning rate, but their actual coverage is closer to the nominal one than the coverage of the traditional confidence interval. In general, our findings indicate that the traditional methods are mainly satisfactorily robust and give reliable results. However, we propose that new formulae for the confidence intervals should be used. Our results are applied to twin-maternity data from Finland and Denmark.     

Demographic analysis of the variation in the rates of multiple maternities in Sweden since 1751

To elucidate the causes and mechanisms of twinning and higher multifetal maternities, we have taken advantage of the statistical sources of Sweden, where continuous statistics for the whole population are the oldest available. We found strong secular and regional fluctuations. The rates of multiple maternities were the highest during the last three decades of the 18th century, when the twinning rate was more than 17 per 1,000, the triplet rate was more than 3 per 10,000, and the quadruplet rate was almost 7 per 1 million maternities. During 1849-1873 the twinning rate in Sweden was 14.2 per 1,000, but this rate showed great regional differences, being 18.0 per 1,000 on the island of Gotland and 12.6 per 1,000 in the county of Alvsborg. During this period the twinning rate in the countryside in the county of Stockholm was 20.4, but in the city of Stockholm it was only 14.1 per 1,000. In Sweden after the 1930s there was a marked decrease in the twinning rate, which by the 1960s had fallen to only about half of what it had been two centuries earlier. The corresponding reductions for triplet and quadruplet rates were about 75%. The aim of this paper was to study the temporal and regional variations in multiple maternities in Sweden from 1751 to 1960 based on demographic and some socioeconomic data for the counties. We confirmed our earlier studies that maternal age and parity cannot satisfactorily explain the secular and regional differences in the twinning rates. In contrast to studies in France (1901-1968), we found no unequivocal association between the twinning rates and the crude birth rates. The correlation coefficients between the twinning rate and the crude birth rate showed statistically significant regional and temporal variations. After eliminating the temporal trends, regional differences in the correlation coefficients remained. The twinning rates for the counties seem to converge toward a common low level, 10-12 per 1,000. The observed convergence toward relatively similar levels may be caused by the increased matrimonial migration distances and decreased endogamy of the citizens as a consequence of better communications. The increased urbanization and industrialization that started in the last decades of the 19th century broke up the old static agrarian isolates and caused Sweden, within 2-3 generations, to develop from a poor nation to one of the most prosperous in the world. A more urban and affluent lifestyle, a better diet, and increased stress and sedentary occupations may have reduced the physical capacity of mothers to carry gestations with multiple embryos or fetuses to completion.     

The epidemiological characteristics of twinning in Northwest India

The data on twin births from four different places in Northwest India were analyzed to study epidemiological characteristics of twinning. Gujjars manifested the highest incidence of twinning (30.2 per 1000 births). The twinning rate at Amritsar for the year 1987 was 19.20. At Bathinda and Jammu, the average twinning rate over the period 1984–93 were 10.70 and 11.40 respectively. Stillbirth rate among twins was much higher in Gujjars as compared to other three samples. Regional trends of twinning rate showed the highest incidence in the state of Uttar Pradesh, while the lowest in Tamilnadu. There were no significant differences between states for MZ twinning rate, while differences in the DZ twinning rate were significant in some instances. Maternal age and parity specific twinning rate showed the highest incidence at parity 4 and in the maternal age group 30–34 years. There was evidence of both seasonality and secular trends in twin births. The highest incidence of twin births over the period 1984–93 was noted in the summer season followed by rainy season, while the lowest in autumn. The average incidence of twinning in the Jammu region decreased from 13.76 during the years 1984–89 to 9.07 during the year 1990–93.     

Maternal smoking and twinning.

In order to investigate a possible association between maternal smoking during pregnancy and twinning, information on 1,096,330 single births and 12,342 twin births in 1983-95 was obtained from the Swedish Medical Birth Registry (MBR). All odds ratios (OR) were estimated after stratification for year of birth and maternal age, parity, and educational level. Smoking women, compared with non-smoking women, were at increased risk of having dizygotic (DZ) twins, but the risk increase was only evident among multiparas. A strong association between previous involuntary childlessness and dizygotic (DZ) twinning (especially in primiparas) was found. The strongest association between maternal smoking and DZ twinning was found among multiparas without any history of involuntary childlessness (OR: 1.35, 95%CI:1.22-1.49), whereas among women who had experienced involuntary childlessness, the opposite was seen (OR: 0.82, 95%CI:0.66-1.00, no difference between parity strata). Weinberg's differential method was used to estimate the number of monozygotic (MZ) twins, and a method of estimating stratified ORs among mothers of MZ twins was presented. No association was found between MZ twinning and maternal smoking (OR: 0.96, 95%CI:0.86-1.07), and no confounding by parity or previous involuntary childlessness was indicated. Several non-causal explanations to the positive association between DZ twinning and maternal smoking among multiparas were discussed, but homogeneity over strata indicated that maternal smoking may be a true risk factor for double ovulation.     

Recent secular trends in dizygotic twinning rates in Europe

The author analyzes trends of declining maternal-age-specific dizygotic twinning rates in Europe for the period 1960-80. Using data from country publications on vital statistics, rates were available for most European countries; no useful data was, however, available for Albania, Romania, and Turkey. Dizygotic twinning rates declined for most countries tries through the 1970s, yet ceased their decline in the 1970s following initial drops in the 1960s were England and Wales, Scotland, Sweden, Norway, Denmark, Switzerland, Hungary, and Czechoslovakia. In the absence of concrete reasons for the observed twinning rates declines and subsequent decline cessation in several countries, the author hypothesizes potential causes. It is speculated that declining sperm quality over the period is related to the declines in European dizygotic twinning rates. Specifically, environmental pollution may have been at the root of deteriorating sperm quality, which in turn fueled the observed declines in dizygotic twinning rates. Cessation of the rate declines for selected countries in the 1970s may be related to the removal of certain polluting environmental agents from those countries. Forming 2 geographic European subregions, countries experiencing rate decline cessation or increase may have common elements to their environmental policy which foster the improvement of sperm quality. Examination of regulations regarding certain agents in these countries within a few years of 1970 may help identify agents responsible for dizygotic twinning rate trends in most European countries. The author also suggests that rate decline cessation may have stemmed from an increase in hormonally-induced twinning, counterbalancing the decline in natural twinning over the period.     

Congenital malformations and variations in reproductive performance in the ferret: effects of maternal age, color and parity

Demographic data of ferrets from a commercial breeding colony were analyzed for the effects of maternal age, parity and strain on reproductive performance and the frequency of gross congenital abnormalities observed at parturition. Litter size (mean +/- SEM) was found to be greatest for young, primiparous females (10.3 +/- 0.2) and decreased with advancing maternal age and parity to a cohort mean of 8.1 +/- 0.1 for third parity females 16 months of age. Age, parity or strain had no effect on 24-hour neonatal mortality (7%) or mortality from birth to weaning (20%) and an examination of the causes for death suggested that these rates can be reduced. The malformation rate from two cohorts of females whelping at different times of the year was low (less than 1.0%) and not significantly different. A higher frequency of malformed offspring was detected in females of low previous parity (0-2) than in those with three or more. Based on data obtained in this survey, the ferret would seem a valuable alternative, nonrodent species for teratologic investigations using currently recommended protocols.     

Secular trends in twinning rates

Based on national data from civil birth registration systems, this paper provides an overview of trends in the twinning birth rates in fifteen developed countries. Patterns and differentials in such rates across populations are described, and trends over time are given. The evolution of the twinning rate in industrialized countries over the last century follows broadly the same pattern. One exception is the period around World War I, with a peak in the twinning rate being observed during the war, or just after it, although this was not registered in all countries. Since the mid-1970s twinning rates have increased in many developed countries in response to a growing use of fertility-stimulating treatments such as in vitro fertilization.     

Frequency of twinning in southwest Nigeria

BACKGROUND:

In the human species, twin is a type of multiple birth in which the mother gives birth to two offspring from the same pregnancy. The occurrence and frequency of twinning, however, varies across human populations. The maternal age, socio-environmental factors, increase in the use of contraceptives, the race of human population, increase in the spontaneous abortion rate, and seasonal variations are among the factors that could influence twinning rate. Information on twinning rates in southwest Nigeria is limited.

AIMS:

This study presents information on the frequency of twinning, as well as its analysis by maternal age, in four urban settings in southwest Nigeria. This is with the aim of extending current knowledge on the frequency of twinning in southwest Nigeria and contributing to the demographic studies in the country.

MATERIALS AND METHODS:

Data on single births and twin births from January 1995 to December 2004 were collected from the Oyo State General Hospital (OSGH), Wesley Guild Hospital (WGH), Obafemi Awolowo University Teaching Hospital (OAUTH), and Ekiti State Specialist Hospital (ESSH) in Ogbomoso, Ilesa, Ile-Ife, and Ado-Ekiti respectively. These were analyzed by year and maternal age groups of 15-19, 20-24, 25-29, 30-34, 35-39, 40-44, and 45-49 years according to the standard method.

RESULTS:

A frequency of twin births of 46.5 per 1000 deliveries and 46.2 per 1000 deliveries was recorded for Ilesa and Ile-Ife respectively. The frequency recorded for Ogbomoso and Ado-Ekiti was 38.5 and 22.1 per 1000 deliveries respectively. The overall average frequency of 40.2 per 1000 deliveries for the four hospitals ranks among the highest recorded rates of twin births in the world. The maternal age group of 25-29 years had the highest occurrence of twin births, while the lowest was recorded in the 45-49 years age group.

CONCLUSION:

This analysis reveals high incidence of twinning in the studied areas and supports previous assertion that the southwestern part of Nigeria has the highest twinning rate in the country and in the whole world. It is our opinion that diet, maternal history of twinning, and some socio-environmental factors may have influenced the results.     

Twinning and heteropaternity in chimpanzees (Pan troglodytes)

Unlike monozygotic (MZ) twins, dizygotic (DZ) twins develop from separate ova. The resulting twins can have different sires if the fertilizing sperm comes from different males. Routine paternity testing of a pair of same-sexed chimpanzee twins born to a female housed with two males indicated that the twins were sired by two different males. DNA typing of 22 short-tandem repeat (STR) loci demonstrated that these twins were not MZ twins but heteropaternal DZ twins. Reproductive data from 1926-2002 at five domestic chimpanzee colonies, including 52 twins and two triplets in 1,865 maternities, were used to estimate total twinning rates and the MZ and DZ components. The average chimpanzee MZ twinning rate (0.43%) equaled the average human MZ rate (0.48%). However, the chimpanzee DZ twinning rate (2.36%) was over twice the human average, and higher than all but the fertility-enhanced human populations of Nigeria. Similarly high twinning rates among African chimpanzees indicated that these estimates were not artifacts of captivity. Log-linear analyses of maternal and paternal effects on recurrent twinning indicated that females who twinned previously had recurrence risks five times greater than average, while evidence for a paternal twinning effect was weak. Chimpanzee twinning rates appear to be elevated relative to corresponding estimated human rates, making twinning and possibly heteropaternity more important features of chimpanzee reproductive biology than previously recognized.     

Determinants of the sex ratio at birth: review of recent literature

The fact that more boys are born than girls (104-107 boys for every 100 girls) has been known since 1662. Factors determining the sex ratio at birth rate are of 2 kinds: factors determining the primary sex ratio, i.e., sex ratio at conception, and factors determining the survival of the embryo in utero. Y-bearing and X-bearing sperm may have different motility or different survival time. The age of the ovum at fertilization and the chemical balance of the female genital tract have an effect on sex ratio at conception. High levels of circulating gonadotropins may imply a lower sex ratio at birth as well as a higher rate of dizygotic twinning. Male conception also appears to be higher early and late in the menstrual cycle. The fact that women exposed to higher coital rates conceive earlier in the menstrual cycle may account for the greater number of boys born during wars. Prenatal male mortality is reportedly highest between gestational months 3-5, lower between months 6-8, and higher again st term. Also, immunological interaction between mother and embryo may account for some sex selective spontaneous abortions. 3 sociodemographic determinants of sex ratio at birth are thought to be maternal age, paternal age, and birth order. Higher prenatal male mortality may be correlated with socioeconomic conditions, since higher socioeconomic status lowers prenatal mortality in general. The effects of parental age, birth order, and parity are less clear. Race is also a factor, since the sex ratio at birth for blacks is lower (102-104) than for whites (106). 14 univariate and 19 multivariate studies of effects of maternal age, paternal age, parity, birth order, race, and socioeconomic status on sex ratio at birth, with sample sizes in the millions from various countries have been analyzed. More boys are born to younger parents, and lower order births have a higher proportion of males than do higher order births. In the multivariate analyses, when the effects of paternal and and birth order are controlled for, the effect of maternal age weakens, and the effect of paternal age appears to be stronger. The effect of birth order remains but is very small, and the effect of race persists independent of any effect of other variables. Maternal age, parity, and birth order are positively correlated with proportion of male stillbirths. The results of the multivariate analyses show all of the effects to be very small, but that maternal age has no effect on sex ratio at birth; paternal age and birth order have a negative effect, and the racial effect persists independent of any other effect. The racial effect is clearly biologically determined at conception because blacks have higher levels of circulating gonadotropin and therefore a higher probability of conceiving girls. Parents in higher socioeconomic classes are more likely to have sons, but the effect is largely due to the excess male mortality during most of the gestational period.     

Population based study of rates of multiple pregnancies in Denmark, 1980-94.

OBJECTIVE: To study trends in multiple pregnancies not explained by changes in maternal age and parity patterns. DESIGN: Trends in population based figures for multiple pregnancies in Denmark studied from complete national records on parity history and vital status. POPULATION: 497,979 Danish women and 803,019 pregnancies, 1980-94. MAIN OUTCOME MEASURES: National rates of multiple pregnancies, infant mortality, and stillbirths controlled for maternal age and parity. Special emphasis on primiparous women > or = 30 years of age, who are most likely to undergo fertility treatment. RESULTS: The national incidence of multiple pregnancies increased 1.7-fold during 1980-94, the increase primarily in 1989-94 and almost exclusively in primiparous women aged > or = 30 years, for whom the adjusted population based twinning rate increased 2.7-fold and the triplet rate 9.1-fold. During 1989-94, the adjusted yearly increase in multiple pregnancies for these women was 19% (95% confidence interval 16% to 21%) and in dizygotic twin pregnancies 25% (21% to 28%). The proportion of multiple births among infant deaths in primiparous women > or = 30 years increased from 11.5% to 26.9% during the study period. The total infant mortality, however, did not increase for these women because of a simultaneous significant decrease in infant mortality among singletons. CONCLUSIONS: A relatively small group of women has drastically changed the overall national rates of multiple pregnancies. The introduction of new treatments to enhance fertility has probably caused these changes and has also affected the otherwise decreasing trend in infant mortality. Consequently, the resources, both economical and otherwise, associated with these treatments go well beyond those invested in specific fertility enhancing treatments.     

Social class and human twinning

In a comparative study to examine the effect of social factors on human twinning, data on sociodemographic and other factors were collected from parents of all twins born alive in Denmark in 1984 or 1985 and from a random sample of parents of singleton infants born in the same years. A postal questionnaire was used. The twins were classified as monozygotic (MZ) or dizygotic (DZ) twins by the similarity method. A trend was found in DZ-twinning, with significantly fewer DZ-twins born in the lower social classes, but not in MZ-twinning. All results were controlled for maternal age and parity.     

Dizygotic to monozygotic twinning ratio at The Royal Women's Hospital, Melbourne 1947-1997, compared with Australian national twinning incidence.

The incidence of dizygotic (Dz) twinning can be used as an index of natural human fertility. A retrospective study was done at The Royal Women's Hospital, Australia, to see whether the dizygotic to monozygotic (Mz) twinning ratio from one hospital can accurately reflect the national incidence of Dz twinning. The yearly twinning incidence from 1947-1997 was expressed as a Dz:Mz ratio, standardised for maternal age and plotted against previously published national statistics. The proportion of mothers born in Asia (of both singleton and multiples) between 1983-1997 was analysed to see whether different racial mixes might influence twinning trends. There were 5275 twins born of known sex and maternal age between 1947-1997. The age-standardised Dz:Mz ratio increased non-significantly from 1.39 in 1947 to 2.29 in 1953 (P = 0.08), underwent a significant decline to 0.73 in 1977, then remained stable until 1997 (P>0.05). The same trends were also apparent when the data was pooled into 2-year groups with the increase from 1947/48-1953/54 becoming highly significant (P<0.009). These trends observed in the hospital-based data were in close agreement with those found in the national statistics, with the exception of a rise in 1977-1982 only reflected in the Australia-wide data. In 1993, 2.6% of mothers were born in Asian countries; by 1997, this had risen to 10.6%. We found that the Dz:Mz ratio from one hospital closely reflects national twinning trends. Prospective studies must account for race, and would need around 200-300 twin pairs per year to minimise fluctuations of the ratio.     

Monitoring population productivity in the saiga antelope

Effective conservation requires a good understanding of factors causing variation in population growth rate. We here analyse the relationship between female age and fecundity in the saiga antelope Saiga tatarica tatarica , a critically endangered ungulate of the Eurasian steppes and semideserts, at both individual and population levels. Annual variation in age structure and twinning rates was investigated using long-term datasets, sampling a total of 3308 females in four populations over more than 40 years. Further, a new non-invasive method is presented, estimating twinning rates from both calves and placentas encountered during calving aggregation transects. At an individual level, the most parsimonious model for twinning rates included three age classes (1, 2 and ≥3 years); however, the model with only two classes (1 and ≥2 years) was competitive and particularly useful for monitoring because these two age classes can reliably be determined by direct observation in the field. Among yearlings, 77.4% were fecund and 11.7% twinned, whereas among older females 94.6% were fecund and 72.6% twinned. At a population level, annual variation in age structure (proportion ≥2 years) correlated well with annual variation in twinning rate except in the north-west Pre-Caspian population. Our results suggest that the recent poaching-driven collapse in saiga numbers has potentially resulted in reductions in fecundity, which will have an impact on population growth rate. Our results highlight the potential for monitoring of twinning rate using non-invasive calving aggregation transects as a cost-effective additional tool to population counts for monitoring the status of this critically endangered species. These monitoring methods are also potentially transferable to other ungulate species.     

Association between the rates of multiple maternities.

We analyze the association between the rates of multiple maternities. Correlation analysis is suitable if there are at least monotonic relationships between the variables. A decreasing tendency can be observed in the rates of multiple maternities in Sweden and in its 25 counties for the period 1751-1960 and the effect of external variables can be assumed to be monotonic. After 1960, the rates of multiple maternities show marked increases, mainly caused by ovulation stimulants and the introduction of other artificial reproduction techniques. We transformed the triplet and quadruplet rates according to Hellin's law in order to obtain rates on the same scale as the twinning rate. The time trends for the multiple maternities in Sweden as a whole were quite similar. The regional differences in the twinning rates could not be satisfactorily explained by maternal age. Alternative attempts to eliminate the effect of other time-dependent factors were to study partial correlation coefficients when the time was kept fixed and the correlation coefficients based on cross-sectional regional data. Both the ordinary and the partial correlation coefficients showed strong regional variation. Cross-sectional analyses gave correlation coefficients similar to the partial correlation coefficients for Sweden as a whole. The variations in the correlation coefficients between the twinning and the triplet rates seem to be caused by other time-dependent factors and the effects of these factors show strong regional variation. After elimination of such factors, the correlation between the twinning and the triplet rates is moderate.     

Influence of maternal age, parity and social class on perinatal mortality in Scotland: 1960-82

This study examines secular changes in the influence of maternal age, parity and social class on perinatal mortality in Scotland. Using cross-sectional national data on all Scottish legitimate births the effects of these factors are estimated on the risk of stillbirths, neonatal and perinatal deaths, and the extent to which the current pattern of relative risks in the early 1980s has changed over the past 2 decades is investigated. Social class is used as a crude measure of relative as opposed to absolute differences in socioeconomic conditions which may influence reproductive outcomes. The effects of age, parity and social class are estimated using logistic models. The most parsimonious model adequately describing the data is provided by a main effects model without interactions. Despite changes in reproductive behavior, improved access to maternity services and more effective perinatal care, the influence of maternal age and social class on perinatal mortality remained unchanged between 1960 and 1982. Although the absolute risks of stillbirths and neonatal deaths declined in all maternal age groups, this improvement was not accompained by a significant change in the relative risks traditionally associated with age. Despite no significant changes in the traditional J-shaped association between parity and stillbirths, cross-sectional analysis shows that in the early 1980s the risk of both neonatal and perinatal deaths decreased as parity increased. This finding is consistent with the pattern of risks observed in longitudinal studies and retrospective surveys of reproductive histories. In view of the stability of age, parity and social class effects on the risk of perinatal mortality, little if any of the overall decrease in Scottish stillbirth and neonatal death rates can be attributed to a significant narrowing of relative risks. The results suggest that the attributable risk of high maternal age or low social class on perinatal mortality is negligible. Future improvements in perinatal mortality are thus likely to result from a continuation of the uniform decrease in perinatal mortality for women of all ages, parities and social classes and not from a diminishing of differences in relative risks which are now virtually identical for a large and growing % of women in Scotland.