- Top of page
Objective To estimate the intergenerational effects of preterm birth and reduced intrauterine growth.
Design Population-based cohort study.
Settings Mother–first-born offspring pairs recorded in the Swedish Medical Birth Registry.
Population Children born before 2001 to 38 720 women born in 1973–75.
Methods The relationships between the mother's and the child's birth characteristics were estimated using logistic regression analysis. Adjustments were made for smoking habits, body mass index (BMI), and current and childhood socio-economic conditions. Analyses were performed on all mother–offspring pairs and on the pairs for which information on neither of the included background variables was missing (n= 24 520).
Main outcome measures Preterm birth (<37 weeks of gestation) and small for gestational age (SGA) (<−2 SD of the Swedish standard).
Results Mothers who themselves had been born preterm were not significantly more likely to deliver their own children preterm, compared with those who had been born at term (adjusted OR 1.24, 95% CI 0.95–1.62). Also, preterm birth in the mothers did not influence the occurrence of SGA in the children. However, the odds ratio for giving birth to SGA and preterm children, respectively, was higher among SGA mothers (OR 2.68, 95% CI 2.11–3.41 and OR 1.30, 95% CI 1.05–1.61). Mothers whose intrauterine growth was moderately reduced but who did not meet the criterion of being born SGA were also at higher risk of giving birth to both preterm and SGA children, respectively.
Conclusions The present study showed evidence of intergenerational effects of reduced intrauterine growth even when socio-economic factors as well as BMI and smoking were adjusted for. There was, however, no consistent intergenerational effect of preterm birth.
- Top of page
Reduced intrauterine growth and preterm birth seem to be related to subsequent adverse outcomes, such as altered physical health in adult life,1 poorer school performances,2–5 and, to some extent, impaired reproduction in women.2,5–7 Researchers have also speculated about a connection between reduced intrauterine growth as well as preterm birth in the mother and/or father and different adverse pregnancy outcomes in the offspring, such as increased infant mortality,8,9 and preterm birth as well as reduced intrauterine growth.10–20 Previous research has shown an intergenerational effect in birthweight,10–13,17,18 and in some of the studies, adjustments were made for gestational age. Other studies have used a slightly different approach when evaluating the intergenerational effect of birthweight for gestational age by dichotomising the birthweight for gestational age into small for gestational age (SGA), typically defined by external standards, and appropriate weight for gestational age (AGA).14–16,20 These studies also suggest that an effect might be present, but larger studies are needed to confirm these findings. The intergenerational effect of preterm birth has also been investigated and although these studies taken together provide evidence of an intergenerational effect, the conclusions drawn in these studies are not always in agreement.15–17,19,20 In a few studies, preterm birth in the mother has been related to SGA in the offspring and vice versa,15,16,20 but the relationships were not well established.
Socio-economic factors such as parental education, maternal age when giving birth, and marital status play a part in determining the occurrence of preterm birth and SGA in the offspring2,8,21,22 as do the mother's body mass index (BMI) and smoking habits. Preterm birth and SGA may have consequences related to the socio-economic situation as well as the BMI, and smoking habits in adult life.2,3,5,11,15,16 There is growing evidence that the intergenerational effects of preterm birth and SGA (as well as birthweight and gestational length) persist even after adjustments are made for current socio-economic status of the mother and/or father.10–13,16,19,23,24 Previous research has also shown an association between maternal social and environmental factors during the childhood of the mother and birthweight of her offspring.25 Therefore, it is of value to try to account not only for the mother's current socio-economic factors but also for the socio-economic situation during her own childhood, when investigating the intergenerational effects of preterm birth and SGA. In addition, it is important to account for the mother's BMI and smoking habits when investigating the intergenerational effects, as these variables otherwise could confound the results. Studies have indicated that the mother's smoking habits or weight/BMI during pregnancy, or her socio-economic situation during childhood, are not likely to explain the intergenerational effects found.12,14–16,23,24 Although some previous studies have been able to simultaneously take account of the above-mentioned variables,23,24 the number of observations in these studies is relatively small, which affects the power of the study particularly when studying the more pathological groups (i.e. preterm and SGA).
Swedish population-based registries offer an opportunity to study the intergenerational effects of preterm birth and SGA in a large sample. Through the registries used, it is possible to get access to socio-economic characteristics of the mothers and the mothers’ parents (also referred to as grandparents), as well as information on the mothers’ smoking habits and BMI at the time of pregnancy. Thus, the primary aim of this study was to estimate the intergenerational effects of preterm birth and reduced intrauterine growth (i.e. SGA) and to evaluate the impact of current and childhood socio-economic factors, as well as the mothers’ BMI and smoking habits during pregnancy, on these effects.
- Top of page
Since 1973, all births in Sweden are registered in the Swedish Medical Birth Registry (MBR),26 which contains information on birth outcomes as well as certain maternal characteristics such as previous reproductive history, and family situation and age when giving birth. In the 1980s and 1990s, additional information was given, for example, the smoking habits and weight of the mothers.27 The births and deaths reported to the MBR are validated annually against the Total Population Register (TPR)28 through individual record linkage. The TPR also contains information on deaths, marital status, migration, and country of origin for Swedish residents born abroad. Other registries used were the Causes of Death Register29 in which information on all deceased persons is recorded, and the Multi-generation Register,30 which makes it possible to identify the parents and grandparents of the children registered in the MBR and the TPR. Information on educational levels was included by using the Education Register31 and the 1970 Population and Housing Census.32
A total of 155 494 female births were registered in Sweden during the years 1973, 1974, and 1975 according to both the MBR and the TPR. Of the women who were alive and still living in Sweden at 13 years of age (n= 150 425), 1029 were excluded due to missing values on birthweight and/or gestational length, as were 126 women with extremely high birthweights compared with the length of gestation (the procedure of exclusion is explained in a previous study2). The remaining 149 270 women were then individually linked to the maternal personal identification numbers for births occurring in the MBR before the year of 2001 (the first birth occurred in 1987). Thus, at the time of the study, the maximum maternal age was 27 years. A total of 40 152 mother–first-born offspring pairs were identified, of which 262 pairs were excluded due to missing values on the child's birthweight and/or gestational length and an additional five due to the children having extremely high birthweights compared with the length of their gestation. Missing values on the first-born child's birthweight and gestational age were more common among teenage mothers and among mothers with lower educational levels. Furthermore, mother–offspring pairs where either the mother or the child was the result of a multiple birth (n= 548 and 356, respectively) were excluded, as were pairs for whom the grandmother and/or grandfather could not be identified in the registries (n= 273). Thus, 38 720 mother–offspring pairs were available for analysis.
Preterm birth was defined as birth before 37 completed weeks of gestation; SGA was defined according to the Swedish standard (i.e. as a birthweight at least 2 SD below the mean weight for the gestational length).33 The impact of the mother's gestational length, as well as her birthweight for gestational length, on the occurrence of preterm birth and SGA in the child was also evaluated in more detail. In these analyses, the gestational length of the mother was split into four categories: ≤31, 32–36, 37–41, and ≥42 weeks of gestation. The mothers’ birthweight for gestational length was expressed in standardised SDs and then split into six categories: <−2, −2 to −1.01, −1 to −0.01, 0 to +1, +1.01 to +2, and >+2 SDs. The standardised SDs were computed by using the formula presented by Marsal et al.33 A similar categorisation of the birthweight for gestational age was also performed on the children.
Data on the mothers’ age, smoking habits, weight, height, and cohabitation with the infant's father were collected by the midwives at the first antenatal visit, and then recorded in the MBR. Smoking was coded as ‘yes’ or ‘no’, as was the cohabitation variable. The mother's age when giving birth was split into three categories: ≤19, 20–24, and ≥25 years. The weight and height of the mothers were used to calculate the BMI, which was divided into four categories: <20, 20–24.9, 25–29.9, and ≥30. Information on the educational levels and marital status of the mothers at 25 years of age as well as information on the socio-economic situations of the mothers’ parents in 1970 (i.e. close to the time of the birth of the mothers) were collected and included in the analyses. The educational levels were coded according to the Swedish educational system: elementary school (9 or 10 years of education), high school (11–13 years), and graduate and postgraduate education (14 years or more). The marital status of the mothers was coded as ‘married’ or ‘unmarried’, and the marital status of the grandmothers as ‘married’, ‘unmarried’, and ‘divorced or widowed’. The grandmothers’ parity at the time of giving birth to the mothers in the study was split into two categories: no previous children and one or more previous children; and their age was split into four categories: ≤19, 20–26, 27–33, and ≥34 years. Information on the grandparents’ country of origin was coded to indicate if at least one of the grandparents had been born outside the Nordic countries or if both grandparents had been born in the Nordic countries.
Information on the mothers’ smoking habits, cohabitation status, and BMI was missing in 5.4, 6.0, and 16.1% of the 38 720 mothers, respectively. As data on maternal weight were not available in the MBR during the years 1990–91,27 missing values on BMI were more common among teenage mothers. The same held true for the cohabitation variable, even though information was recorded in the MBR during the whole study period. Missing values on smoking habits were more common among mothers aged 25 years or more when giving birth to their first child. Information on the educational level of the grandmothers was missing in 14.8% of the mother–child pairs and the corresponding percentage among the grandfathers was 8.1. It has previously been shown that the missing category on these variables constitutes a greater proportion of individuals who were young when giving birth as well as individuals who immigrated later than 1970 or were of non-Nordic origin.2,34 Missing values on other socio-economic variables of the mother and her parents were found in less than 1% of the population, for each variable.
Due to the relatively high percentages of missing values on the mothers’ smoking habits, cohabitation status, and BMI, as well as the grandparents’ educational levels, analyses were performed both on the original study population (n= 38 720) and on the mother–offspring pairs for which information on neither of the included variables was missing (n= 24 520).
The relationships between the mothers’ birth characteristics (i.e. preterm birth and SGA) and socio-economic factors as well as BMI and smoking habits in adult life (see 1Table 1) were estimated using χ2 tests. We also performed multiple logistic regression analyses in which the above-mentioned relations were adjusted for socio-economic characteristics of the mother's parents (i.e. the grandparents’ educational levels, grandmother's marital status, parity, and age when giving birth to the mothers of study, as well as the grandparents’ country of origin). In these additional analyses (see first paragraph of Results), the following categories of the dependent variables were treated as reference categories: ‘age <25 years’, ‘not smoking’, ‘living with the child's father’, ‘BMI ≥20’, ‘less than 14 years of education’, and ‘not married’. Multiple logistic regression analysis was also used when estimating the effects of maternal smoking habits, cohabitation status, age, BMI, as well as the effects of the socio-economic characteristics of the mother's parents on preterm birth and SGA in the offspring. The odds ratios were adjusted for all variables included in the models, and the variables were treated as categorical.
Table 1. Preterm birth and SGA among the mothers in relation to subsequent socio-economic factors as well as BMI and smoking habits*
| ||Mother born preterm**||Mother born SGA**|
| ||Yes (n= 807)||No (n= 23 713)||P***||Yes (n= 1283)||No (n= 23 237)||P***|
|Age† (years)|| ||0.116|| ||0.842|
|≤19||6.8||6.2|| ||6.6||6.2|| |
|20–24||59.1||56.1|| ||56.0||56.2|| |
|≥25||34.1||37.6|| ||37.3||37.5|| |
|Smoking habits†|| ||0.111|| ||0.001|
|Not smoking||79.3||81.5|| ||78.1||81.6|| |
|Smoking||20.7||18.5|| ||21.9||18.4|| |
|Cohabitation status†|| ||0.001|| ||0.124|
|Living with the child's father||88.7||92.0|| ||90.7||91.9|| |
|Other family situation||11.3||8.0|| ||9.3||8.1|| |
|BMI†|| ||0.831|| ||<0.001|
|<20||13.6||14.0|| ||18.5||13.7|| |
|20–24.9||54.8||55.9|| ||54.4||55.9|| |
|25–29.9||23.0||21.8|| ||20.3||21.9|| |
|≥30||8.6||8.4|| ||6.9||8.4|| |
|Educational level‡ (years)|| ||<0.001|| ||<0.001|
|9–10||18.3||14.7|| ||17.1||14.7|| |
|11–13||68.5||67.5|| ||69.6||67.4|| |
|≥14||13.1||17.8|| ||13.1||17.9|| |
|Marital status‡|| ||0.693|| ||0.015|
|Married||22.3||22.9|| ||20.1||23.0|| |
|Not married||77.7||77.1|| ||79.9||77.0|| |
In addition, logistic regression analysis was used to estimate the intergenerational effects of preterm birth and SGA. Analyses were performed both on the original study population (n= 38 720) and on the mother–offspring pairs for which information on neither of the included variables was missing (n= 24 520). Crude odds ratios and 95% confidence intervals were calculated for each analysis, and, for the 24 520 mother–offspring pairs with complete information on the background variables, adjusted odds ratios were also calculated. Adjustments were made for the mothers’ smoking habits, cohabitation status, age, and BMI, as well as for the socio-economic characteristics of the mothers’ parents. Since the mothers’ educational level and marital status were measured at 25 years of age, we performed subgroup analyses where only the mothers aged ≥25 years when giving birth were included, and we then evaluated the effect of these two additional variables on the investigated intergenerational effects.
The impact of the mother's gestational length and the mother's birthweight for gestational length (expressed in SDs) on the occurrence of preterm birth and SGA in the child was also evaluated in more detail. These models were compared with the previously described models (i.e. in which the mother's gestational length and birthweight for gestational length were dichotomised) using the −2ln(likelihood) of the models. In this way, it was possible to evaluate if the more detailed categorisations of the mother's gestational length and birthweight for gestational length, respectively, added more information in explaining the occurrence of preterm birth and SGA in the offspring. In addition, the effect of the mother's birthweight for gestational length on reduced intrauterine growth in the children was also evaluated in more detail by performing a series of three multiple logistic regression analyses. In the first analysis, all mother–offspring pairs were included; in the next analysis, we excluded children who were SGA (i.e. whose birthweights for gestational age were <−2 SD) and examined the children whose birthweights for gestational length were between −2 and −1.01 SD. Finally, we excluded all children below −1 SD and evaluated the effect of the mother's birthweight for gestational length on the children being born −1 to −0.01 SD.
In order to evaluate the final models, two-way interactions (also referred to as effect modifiers) between the mothers’ birth characteristics and the other background variables of the mothers were created and included in the models. The interaction terms were determined by means of forward stepwise regression, and the significance level of the interaction terms was set to P≤ 0.01.
This study was approved by the Human Research Ethics Committee, Faculty of Health Sciences, Linköping University.
- Top of page
Unlike most previous studies, we had the opportunity to account not only for the mother's current socio-economic factors but also for the socio-economic situation during her own childhood as well as her smoking habits and BMI during early pregnancy when investigating the intergenerational effects of preterm birth and reduced intrauterine growth (i.e. SGA). The present study is also one of the largest studies to date, which enables us to make more precise estimates of the intergenerational effects investigated than has previously been possible. In the present study, the estimated risk of preterm birth in the offspring was about 30% higher among Swedish mothers who themselves were born preterm, compared with mothers born at term. This effect was no longer significant once those with incomplete data were removed from the analyses and the odds ratio was adjusted for socio-economic factors. Increased risk, although not always statistically significant, has been suggested in several studies.16,17,19,20 Klebanoff et al.15 reported a reduced risk (OR = 0.53) in a smaller Swedish study, but, due to the wide confidence interval, their risk estimate did not differ from the risk estimate and corresponding confidence interval demonstrated in the present study. Furthermore, we found no significant increases in risk for SGA in the child among mothers who were born preterm, which is in line with previously reported results.15,16,20 The results of our study showed an intergenerational effect of SGA, but also that SGA in the mother had a positive effect on the occurrence of preterm birth in the offspring, which is in agreement with previously reported results.12–14,20 Klebanoff et al.15 reported a significantly higher unadjusted odds ratio for giving birth to preterm children in mothers who themselves were born SGA (OR 2.96, 95% CI 1.47–5.94). However, the odds ratio in their study became markedly lower when adjustments were made for maternal weight and smoking habits as well as marital status, and the adjusted analysis did not differ from the one presented in our study. In addition, we found evidence that the mothers whose intrauterine growth was reduced but who did not meet the criterion of being born SGA also were at higher risk of giving birth to preterm and SGA children, but also to children whose intrauterine growth was reduced but who did not meet the criterion of being born SGA.
There seems to be a relationship between preterm birth and SGA among the mothers and subsequent socio-economic situation in adult life. As expected from previous results,8,21 the mothers’ pre-pregnancy BMI seemed to be related to preterm birth in the offspring and there was also some evidence that underweight mothers were more likely to give birth to SGA children. Mothers who were themselves born SGA also seem to be more likely to be smokers in adult life, compared with mothers who were born AGA, and they also seem to have a lower BMI. These associations are also apparent in other studies.2,5,11,15,16 Mothers who were smokers have previously been shown to be more likely to give birth to preterm or growth-restricted children.8,21,22 We found that the mother's smoking habits were associated with SGA but not preterm birth in the offspring. The mother's educational level at 25 years of age was found to be related to SGA in the offspring, which has also been shown in previous studies,2,8 whereas the mother's marital status and cohabitation status seemed to be of minor importance in affecting the outcomes studied, although these associations were found in earlier research.2,21 A possible reason for these discrepancies might be that our population was smaller than in some of the previous studies;2,8,21 for instance, in the study by Clausson et al.8 the population consisted of nearly 100 000 women. Other explanations might be that the mothers in the present study were relatively young (i.e. 25–27 years at the end of follow up), compared with the mothers in other studies, or the fact that we were able to adjust the relationships studied for the socio-economic characteristics during the childhood of the mothers. Earlier research has shown an association between maternal socio-economic situation during childhood and offspring birthweight, independent of the mother's current socio-economic position.25 In the present study, preterm birth (but not SGA) in the offspring was related to the mothers’ socio-economic characteristics during childhood, even after adjustments were made for the mother's current socio-economic situation, smoking habits, and BMI.
Although some of the background characteristics studied, such as the mother's smoking habits, were significant confounding variables, they had notably little impact on the intergenerational effects. This could imply that the factors adjusted for in this study did not capture the whole social panorama. For instance, we were not able to investigate the effect of behavioural factors such as maternal use of drugs, which is known to increase the occurrence of preterm birth and SGA in the offspring.22 However, there is growing evidence that the intergenerational effects of preterm birth and SGA (as well as birthweight and gestational length) persist even after adjustments are made for current socio-economic status of the mother and/or father.10–13,16,19,23,24 Some studies have also indicated that the mother's smoking habits or weight/BMI during pregnancy, or her socio-economic situation during childhood, are not likely to explain the intergenerational effects found.12,14–16,23,24 Thus, it seems unlikely that the intergenerational associations of preterm birth and SGA are entirely explained by socio-economic characteristics or BMI and smoking habits. It has been suggested that the intergenerational effects of birthweight and gestational age might be explained by a genetic mechanism.17,18,35 Another explanation could be that the mothers were exposed to some unknown factor before they were born, a factor that caused them to be SGA or preterm at birth and also predisposed their offspring to be born SGA or preterm.
Although the overall information on gestational length and birthweight recorded in the MBR has been evaluated and found to be reliable,26,27 evaluations of the registry have shown evidence of measurement errors in both extremes of the distributions of birthweight and gestational length, respectively.26,36 In the present study, we have made efforts to exclude misreported and/or wrongly entered values from the dataset, and also chose to categorise the data in larger categories instead of basing the analyses on individual values, which decreases the influence of possible incorrect values on the results. However, there may be other sources of error that affect the information regarding preterm birth and SGA. The results show that across the two generations studied, the prevalence of preterm birth has increased from 3.7% in the mothers to 6.4% in the children, while the prevalence of SGA decreased from 5.5 to 2.8%. Since the middle of the 1980s, the gestational length has been measured by means of ultrasound examinations predominantly made at 16–18 weeks of gestation.27 As the mothers of this study were born in 1973–75, their gestational age was estimated by using last normal menstrual period dating, which has been evaluated to produce, on average, two to three days longer estimates than the more precise ultrasound examination.37 Some researchers believe that the difference in estimates between the two methods is involved in explaining the recent trend of increased prevalence of preterm births.37 If correct, this means that in the present study, there are some mothers who ought to have been classified as preterm but were wrongly classified as full-term (i.e. ‘false negatives’). As a result, the analyses involving the mothers’ gestational length may be slightly conservative. An explanation to the decline in the prevalence of SGA could be that the current Swedish SGA standard is based on intrauterine measures of children born in the middle of the 1990s. As the mean birthweight for children born in 1973–75 was 50–60 g lower compared with children born in 1996,38 this implies that there may be some ‘false positives’ among the SGA mothers. However, although adjusting for this increase in birthweight decreased the prevalence of SGA among the mothers to 4.2%, the intergenerational effect of SGA did not substantially change. Thus, we believe that this increase in birthweight is of minor importance to the conclusions drawn from this study.
These differences in prevalence could also theoretically be due to the fact that not all mothers were first-born or due to biases arising from missing values. However, neither the results of the analyses in which only the first-born-mother–offspring pairs were included nor those of analyses that included all eligible mother–offspring pairs differed substantially from the results on the pairs with complete data on all background variables. Missing values on the child's gestational length and/or birthweight were more common among teenage mothers and mothers with lower educational levels. This could also have biased the results. However, we believe that this possible bias would have had little effect on the results as the amount of missing values was small (less than 1%). As some of the previous research in this area has indicated that teenage mothers and mothers with low education tend to give birth to children with lower birthweights and length of gestation,2,21,22 our results could be slightly more conservative than if there were no missing values on birthweight and gestational length. During the period of study, there have been some other changes in obstetric practice in Sweden that affect the information in the MBR, and these may have affected the results of this study. For example, caesarean sections, including caesarean sections for fetal indication at an early gestational age, have become more widely used in the 1990s compared with the 1970s; 5.3–7.0% of the children born in 1973–75 were delivered by means of caesarean section, compared with 11.3–14.8% in 1986–2000.38