Socioeconomic and cultural differentials in mortality in a late 19th century urban setting: A linked records study from Tartu, Estonia, 1897-1900

Mothers of twins had higher old-age survival than mothers of singletons in Estonian 19th-century birth cohorts

STUDY QUESTION

Do the mothers of twins and singletons differ regarding post-partum and old-age mortality?

SUMMARY ANSWER

Twin deliveries were associated with higher post-partum maternal mortality than singleton deliveries, but the lifetime post-partum mortality risk was similar for mothers of twins and singletons; survival of twinners was higher than survival of the mothers of singletons after the 67th lifespan percentile.

WHAT IS KNOWN ALREADY

Twinning is typically associated with higher post-partum maternal mortality. The evidence about whether twinning incurs long-term survival costs of reproduction or is a trait pertinent to long-lived women is scarce and contradictory.

STUDY DESIGN, SIZE, DURATION

The study is based on the data of the Estonian Family Register (operating from 1926 to 1943) and involves 5565 mothers of twins and 119 613 mothers of singletons born between 1850 and 1899. The subset for comparing maternal lifespans included 1703-1884 mothers of twins and 19 747-36 690 mothers of singletons.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Post-partum maternal mortality was analyzed in the whole sample (including mothers of a single child) by logistic regression. Most of the analyses were performed in samples where each mother of twins was matched against mothers of singletons based on parity (or number of deliveries), urban versus rural and inland versus coastal origin, whether their lifespan was known, date of birth and age at first birth. Lifespans were compared in linear mixed models. Quantile regression was used to analyze age-dependent variations in maternal mortality rates. All models were adjusted for relevant biodemographic covariates.

MAIN RESULTS AND THE ROLE OF CHANCE

The twinning rate in the whole sample was 4.4%. During the year after giving birth, maternal mortality for twin deliveries was 0.75% (17/2273) and 0.37% (449/122 750) for singleton deliveries (OR = 2.05, 95% CI = 1.21-3.23). However, the lifetime post-partum mortality risk for mothers of twins (0.51%; 28/5557) and singletons (0.37%; 438/119 466) did not differ significantly (OR = 1.38, 95% CI = 0.91-1.98). The life spans of the mothers of twins and singletons did not differ in matched samples. Past the 67th lifespan percentile, the odds of survival were significantly higher for mothers of twins than mothers of singletons, as indicated by non-overlapping 95% confidence intervals.

LIMITATIONS, REASONS FOR CAUTION

Relatively low number of individuals (22 802-28 335) with known age at death in matched datasets due to discontinuation of the register after 1943.

WIDER IMPLICATIONS OF THE FINDINGS

The finding that mothers of twins had higher odds of old-age survival than mothers of singletons is consistent with the contention that twinners represent a non-random subset of women whose robust phenotypic quality allows them to outlive the mothers of singletons in old age.

STUDY FUNDING/COMPETING INTEREST(S)

The study was funded by the Estonian Research Council grants PRG1137, PRG2248, and PSG669. The authors declare no competing interests.

TRIAL REGISTRATION NUMBER

N/A.

Re-examining the social gradient in health: A study of Dutch men, 1850-1984

Today, a social gradient in health is clearly visible. Individuals with higher socio-economic statuses tend to live longer lives, and are less likely to be disabled or chronically ill. However, there is debate over when the social gradient emerged: is it a constant across contexts, or a particular feature of certain societies? Often, social gradients are not found in historical contexts. This is perhaps because historical studies use mortality as their sole measure of health, which may not fully reflect the health statuses of the living. Using another health indicator may help to identify whether a social gradient in health was present in historical contexts. One alternative measure of health is body height, a barometer of population health. In this study, we accordingly examined socio-economic status's relationship to both adult mortality and body height. A sample of Dutch men (n=3396), born between 1850 and 1900, was used. Socio-economic status was measured with parental occupational class, and height was measured at age 20. Survival analyses (for mortality at age 20 or later) and linear regressions (for height at age 20) were performed. We found no clear gradient in occupational class's relationship to adult mortality. Regarding height, individuals from elite backgrounds were estimated to be 2.82 cm cm taller (95% CI: 1.41-4.24) than those from unskilled working backgrounds. While a gradient in height was present in earlier birth cohorts, it was not visible among men born between 1885 and 1900. These findings indicate that there was a social gradient in health in the late nineteenth and twentieth centuries, although the gradient perhaps changed based on the indicator and time period being examined. This may mean that the social gradient in health is more persistent over time than it appears when only examining the social gradient in mortality.

When Did the Health Gradient Emerge? Social Class and Adult Mortality in Southern Sweden, 1813-2015

Across today's developed world, there is a clear mortality gradient by socioeconomic status for all ages. It is often taken for granted that this gradient was as strong-or even stronger-in the past when social transfers were rudimentary and health care systems were less developed. Some studies based on cross-sectional data have supported this view, but others based on longitudinal data found that this was not the case. If there was no gradient in the past, when did it emerge? To answer this question, we examine social class differences in adult mortality for men and women in southern Sweden over a 200-year period, using unique individual-level register data. We find a systematic class gradient in adult mortality emerging at ages 30-59 only after 1950 for women and after 1970 for men, and in subsequent periods also observable for ages 60-89. Given that the mortality gradient emerged when Sweden transitioned into a modern welfare state with substantial social transfers and a universal health care system, this finding points to lifestyle and psychosocial factors as likely determinants.

Survival and sex composition of offspring: Individual-level responses in the quantum and tempo of childbearing during the demographic transition

Previous studies have documented varying fertility responses to childhood mortality and to the sex composition of the surviving offspring during the demographic transition. We contribute to this literature by applying a mixture cure model to reproductive histories of Estonian women born 1850-99. This model, unlike standard event history models, is capable of separating the effect of the covariates on the propensity of having another birth from their effect on its timing. Child fatalities, not having sons, and to a smaller extent, not having daughters, increased the propensity to have another child and decreased the interval to it. The response was stronger among later cohorts, but only with respect to parity progression. By contrast, the accelerated childbearing response diminished over time. Our findings suggest that behavioural responses in the quantum and tempo of childbearing can occur relatively independently.