Article Text

Download PDFPDF

Mothers, places and small for gestational age births: a cohort study
  1. Jan Sundquist1,2,
  2. Kristina Sundquist1,
  3. Sven-Erik Johansson1,
  4. Xinjun Li1,
  5. Marilyn Winkleby2
  1. 1Center for Primary Health Care Research, Lund University, Lund, Sweden
  2. 2Stanford Prevention Research Center, Stanford University, Palo Alto, California, USA
  1. Correspondence to Dr Jan Sundquist, Center for Primary Health Care Research, Lund University/Region Skåne, CRC, hus 28, plan 11, ing 72, UMAS, 205 02 Malmö, Sweden; Jan.sundquist{at}


Objective This study examines whether neighbourhood deprivation increases the risk of giving birth to a small for gestational age (SGA) infant, after accounting for individual-level maternal socioeconomic characteristics.

Design An open cohort of women, aged 20–44 years, was followed from 1 January 1992 through 31 December 2004 for first singleton births. The women's residential addresses during the two consecutive years preceding the birth of their infants were geocoded and classified into three levels of neighbourhood deprivation. Gestational age was confirmed by ultrasound examinations. Multilevel logistic regression models were used in the statistical analysis.

Setting Sweden.

Results During the study period, women gave birth to 720 357 infants, of whom 20 487 (2.8%) were SGA. Age-adjusted incidence rates of SGA births increased with increasing level of neighbourhood deprivation. In the total population, 2.5% of births in the least deprived neighbourhoods and 3.5% of births in the most deprived neighbourhoods were SGA. A similar pattern of higher incidence with increasing level of neighbourhood-level deprivation was observed across all individual-level sociodemographic categories, including maternal age, marital status, family income, educational attainment, employment, mobility and urban/rural status. High neighbourhood-level deprivation remained significantly associated with SGA risk after adjusting for maternal sociodemographic characteristics (OR 1.28, 95% CI 1.22 to 1.34).

Conclusions This study is the largest to date of the influence of neighbourhood on SGA birth, with SGA confirmed by ultrasound examination. Results suggest that the characteristics of a mother's neighbourhood affect the risk of delivering an SGA infant independently of maternal sociodemographic characteristics.

View Full Text

Statistics from


Children who are born small for gestational age (SGA) have increased risks for a number of health consequences such as neurodevelopmental disabilities,1 psychiatric disorders2 and behavioural sequelae.3 Some maternal risk factors for giving birth to an SGA infant have been identified. Data from the US National Maternal and Infant Health Survey showed that both low educational and low occupational status were significantly associated with high risk of SGA births.4 Even in countries such as Sweden5 and Canada,6 which have smaller socioeconomic differences than the USA and universal access to prenatal care, there are higher rates of SGA births among women with lower socioeconomic status (SES) than among those with higher SES. Other maternal risk factors and maternal diseases that have been associated with SGA birth are smoking, substance abuse, malnutrition, anaemia, hypertension, chronic kidney disease, advanced diabetes, heart disease and infection.7,,9

What is already known on this topic

  • Neighbourhood-level deprivation is associated with small for gestational age (SGA) births, mostly defined according to mother's last menstrual period.

  • Maternal socioeconomic status (eg, low income, low educational attainment and/or low occupational status) is one of the strongest risk factors for SGA births and may often confound studies of the association between neighbourhood-level deprivation and SGA.

What this study adds

  • The Swedish population data allowed more precise determination of neighbourhood exposure and also adjustment of results for maternal age, marital status, family income, educational attainment, employment, mobility and urban/rural status.

  • Almost 95% of all small for gestational age (SGA) diagnoses were based on ultrasound examination, which is a gold standard to determine gestational age.

Previous studies have examined the effects of socioenvironmental variables on SGA births. A study from Quebec found that women who lived in low-income neighbourhoods had a higher risk of giving birth to an SGA infant than women who lived in affluent neighbourhoods.10 Studies from the USA found that neighbourhood economic conditions and neighbourhood deprivation were associated with SGA births.11 12

The current study of the Swedish population makes a novel contribution to the literature for several reasons. Unlike earlier studies, the data on SGA births had exceptionally high validity as they were based on ultrasound examination, which is used in approximately 95% of all pregnancies in Sweden to calculate expected date of delivery.13 SGA was chosen as the outcome because it includes growth restriction which could be related to chronic maternal stress in deprived neighbourhoods. Furthermore, it was possible to determine neighbourhood exposure more precisely than in previous studies, as data were available on the neighbourhood where each pregnant woman resided during the two consecutive years preceding the birth of her infant. In addition, it was possible to adjust the results for a comprehensive set of individual-level sociodemographic factors.

This study had the following three aims: (1) to ascertain whether exposure to neighbourhood-level deprivation in the two consecutive years prior to childbirth predicted risk of delivering an SGA infant, (2) to determine whether the relationship between neighbourhood deprivation and risk of SGA remained significant after adjusting for individual-level maternal sociodemographic factors, and (3) to examine possible cross-level interactions between individual-level sociodemographic factors and neighbourhood-level deprivation to determine whether neighbourhood-level deprivation had a differential effect on risk of SGA births across the subgroups of women (effect modification).


The study population consisted of an open cohort of all women in Sweden aged 20–44 of years, who were followed from 1 January 1992 through 31 December 2004 for first singleton births. Data were retrieved from a national research database located at the Center for Primary Health Care Research at Lund University. The database contains nationwide information on maternal and infant factors at both the individual and neighbourhood level, including comprehensive demographic and socioeconomic data. The information in the database comes from several Swedish national registries. The registers used in the present study were the Medical Birth Register, the Hospital Discharge Register and the Population Register.

The Swedish nationwide population and healthcare registers are exceptionally complete and have very high validity.14 There was no loss to follow-up, as each woman was tracked with a personal identification number that was replaced by a serial number to provide anonymity.

Definition of SGA

A combination of gestational age (based on ultrasound or date of last menstrual period) and birth weight were used to determine SGA. All pregnant women in Sweden are offered free antenatal care. At the first antenatal visit, usually at 10–12 gestational weeks, information on the date of the last menstrual period is obtained. In addition, between 16 and 18 gestational weeks, almost all women (95% or more13) have an ultrasound examination to date their pregnancy. Infants more than 2 SD below the mean for gestational age (ie, the 2.5th percentile) are defined as SGA.15 This corresponds to a birth weight more than 24% (approximately 850 g) lower than expected for a full-term infant.

Neighbourhood-level deprivation

The home addresses of all Swedish women have been geocoded to small geographical units that have boundaries defined by homogeneous types of buildings. These neighbourhood areas, called small area market statistics, or SAMS, have an average of 1000 people each and were created by Statistics Sweden, the statistics bureau owned by the Swedish Government. SAMS were used as proxies for neighbourhoods, as in previous research.16 17 SAMS with fewer than 50 people aged 25–64 (the most socioeconomically active) were excluded (n=1024 SAMS), as were women whose addresses could not be geocoded to a neighbourhood area (n=37 143 women, 0.5% of the sample). The final sample included 8505 SAMS.

In the analyses, neighbourhood of residence was determined for two consecutive years prior to childbirth to ensure that the correct neighbourhood of exposure was selected for all women.18 19 Neighbourhood of residence is calculated annually from the National Land Survey of Sweden's register.

A summary index was calculated to characterise neighbourhood-level deprivation. The neighbourhood index was based on information on women and men aged 20–64 who lived in the neighbourhood because people in this age group are the most socioeconomically active, that is, as population group they have a stronger impact on the socioeconomic structure of the neighbourhood than children, younger women and men and/or retirees. The neighbourhood index was based on four items: low educational status (<10 years of formal education), low income (income from all sources, including that from interest and dividends, defined as less than 50% of individual median income), unemployment (not employed, excluding full-time students, those completing compulsory military service and early retirees) and receipt of social welfare. The index was categorised into the following three groups (higher scores reflect more deprived neighbourhoods): low neighbourhood deprivation (more than 1 SD below the mean), moderate neighbourhood deprivation (within 1 SD of the mean) and high neighbourhood deprivation (more than 1 SD above the mean).20 Low deprivation neighbourhoods represent the most affluent neighbourhoods.

Individual-level sociodemographic variables

All individual variables were measured for each woman for the year during childbirth. For example, if the first singleton birth took place during 2001, the individual variables were measured on 1 January 2001.

Maternal age: Divided into 5-year age groups as follows: 20–24, 25–29, 30–34, 35–39 and 40–44 years.

Marital status: Data were collected from the medical birth record and grouped as: (1) married/cohabitating or (2) never married, widowed or divorced.

Family income: Calculated using the annual family income divided by the number of people in the family, that is, individual family income per capita. The income parameter also took into consideration the ages of people in the family and used a weighted system whereby small children were given lower weights than adolescents and adults. The calculation procedure was performed as follows. The sum of all family members' incomes was multiplied by the individual's consumption weight divided by the family members' total consumption weight. The final variable was calculated as empirical quartiles from the distribution.

Educational attainment: Divided into completion of compulsory school or less (≤9 years), practical high school or some theoretical high school (10–11 years) or completion of theoretical high school and/or college/university (≥12 years).

Employment: (1) Employed or (2) unemployed. Unemployed also included students and homemakers.

Mobility: Women were classified as having ‘not moved’ or ‘moved’ to another neighbourhood with the same or a different level of deprivation during the 2 years prior to childbirth. This procedure created five categories:

  • (1) not moved

  • (2) consistently low/moderate deprivation (women who moved from one low deprivation neighbourhood to another low deprivation neighbourhood or women who moved from one moderate deprivation neighbourhood to another moderate deprivation neighbourhood)

  • (3) consistently high deprivation (women who moved from one high deprivation neighbourhood to another high deprivation neighbourhood)

  • (4) downward mobility (women who moved to a neighbourhood with a higher level of deprivation, that is, low to moderate, moderate to high, low to high) and

  • (5) upward mobility (women who moved to a neighbourhood with a lower level of deprivation, that is, moderate to low, high to moderate, high to low).

Urban/rural status: Mothers were classified as living in a large city, middle-sized town, or a small town or rural area. This variable was included because urban/rural status may be associated with access to antenatal care. Large cities included cities with a population of more than 200 000, that is, the three largest cities in Sweden (Stockholm, Gothenburg and Malmö). Middle-sized towns included towns with a population of more than 90 000, excluding the three largest cities in Sweden. Small towns included towns with a population of more than 27 000 and less than 90 000, and rural areas were areas with a smaller population than population of small towns. This classification yielded three equally sized groups.

Statistical analysis

Age-standardised incidence proportions were calculated by direct age standardisation. The entire Swedish population of women aged 20–44 years in 1992 was used as the standard population. The incidence rate of SGA births during the study period was calculated as the total number of SGA births in Sweden from 1 January 1992 through 31 December 2004 among women who gave birth during the study period divided by the number of women on 1 January 1992. Incidence rates were calculated for the total population and for each subgroup after each woman's neighbourhood of residence for the 2 years prior to childbirth had been determined.

Multilevel (hierarchical) logistic regression models with incidence proportions were estimated. The analyses were performed using MLwiN, v 2.02. First, a null model was calculated to determine the variance among neighbourhoods. Then a neighbourhood model was calculated that included only neighbourhood-level deprivation to determine the crude risk of SGA birth by level of neighbourhood deprivation (aim 1). Next, a full model was calculated that included neighbourhood-level deprivation and the seven individual-level maternal sociodemographic variables, which were added simultaneously to the model (aim 2). Finally, a full model tested the cross-level interactions between the individual-level sociodemographic variables and neighbourhood-level deprivation to determine if the effects of neighbourhood-level deprivation on SGA incidence differed across the sociodemographic variables, that is, effect modification (aim 3).

Ethics approval

This study was approved by the Ethics Committee at Lund University.


Of the total female population, 21.9%, 59.5% and 18.6% lived in low, moderate and high deprivation neighbourhoods, respectively. During the follow-up period from 1 January 1992 through 31 December 2004, there were 720 357 first singleton births among the women; 20 487 (2.8%) of these births were SGA (table 1). Age-adjusted incidence rates of SGA births increased from 2.5% of births in neighbourhoods with low deprivation to 2.8% of births in neighbourhoods with moderate deprivation and 3.5% of births in neighbourhoods with high deprivation. A similar pattern of higher incidence with each increasing level of neighbourhood-level deprivation was observed across all seven individual-level sociodemographic categories.

Table 1

Distribution of the total population of women aged 20–44 years (N=720 357), number of small for gestational age (SGA) births and age-standardised incidence rates of SGA births (%) by neighbourhood-level deprivation

The OR of giving birth to an SGA infant for women living in a high versus low deprivation neighbourhood was 1.38 (95% CI 1.32 to 1.44) in the crude neighbourhood-level model (table 2). Neighbourhood-level deprivation remained significantly associated with SGA risk after adjusting for the seven individual-level maternal sociodemographic variables (OR 1.28, 95% CI 1.22 to 1.34).

Table 2

OR and 95% CI for small for gestational age births: results of multi-level logistic regression models

Women with the highest risk of giving birth to an SGA infant were older; never married, widowed or divorced; had low family incomes; and/or the lowest educational attainments. All categories showed a gradient effect across level of neighbourhood deprivation. Three of the four categories of women who moved had higher SGA risks than women who did not move.

The test for cross-level interactions between the individual-level sociodemographic variables and neighbourhood-level deprivation on risk of SGA births showed no meaningful cross-level interactions or effect modification.

The between-neighbourhood variance (ie, the random intercept) was over 1.96 times the SE in all models, indicating that there were significant differences in SGA incidence between neighbourhoods after accounting for the neighbourhood-level variable and the individual-level variables. The neighbourhood-level variable explained 54% of the between-neighbourhood variance in the null model (see table 2). After inclusion of the individual-level variables, the explained variance was 62%.


We found that living in a deprived neighbourhood increased the odds of giving birth to an SGA infant by 28%. This is a modest increase in risk, but one with important public health consequences in deprived neighbourhoods. It is noteworthy that these effects were found in a country with a comparatively strong system of universal health and social welfare benefits.

Our finding that neighbourhood deprivation exerts an independent effect on risk of SGA birth is consistent with the findings of a small but growing number of studies that have provided evidence of an association between neighbourhood-level socioeconomic factors and adverse birth outcomes, including low birth weight21,,23 and preterm birth.24 Low birth weight (an easy-to-measure indicator of the health of newborn infants that is routinely measured and documented on birth certificates in most countries) is a more commonly used outcome that includes both SGA as well as preterm births. However, low birth weight has been criticised as lacking a causal link to health outcomes.25 The causal link may be lacking because low birth weight is a broad diagnosis that includes both SGA births and preterm births, two aetiologically different outcomes.25 26 Few previous researchers conducting neighbourhood-level studies have had data that enable them to use SGA births as a specific outcome variable. One study on neighbourhood deprivation and risk of SGA birth from Quebec, Canada, found an increased risk of SGA birth of 1.18 among women who lived in the most deprived quartile of neighbourhoods after adjustment for age, mother's education and marital status.10 Our findings confirm the findings of the Canadian study in a larger, nationwide sample of women. Both studies have been conducted in countries with universal healthcare systems and a similar standard of living; it is reasonable to hypothesise that neighbourhood effects may be larger in countries that do not have universal healthcare, such as the USA.

There are a number of general mechanisms through which level of neighbourhood deprivation may influence risk of SGA births, including unfavourable health-related behaviours,16 27 28 neighbourhood social disintegration (ie, criminality, high mobility or unemployment),29 low social capital17 30 31 or neighbourhood stress mediated by factors that can influence immunological and/or hormonal stress reactions.32,,34 For example, it has been suggested that crime lies in the direct pathway between the neighbourhood social environment and health,35 36 and a consistent association between neighbourhood social deprivation and crime has been found in previous studies.35 Socially deprived neighbourhoods in the USA are often associated with both criminal violence and residential instability.36 It is possible that women are particularly vulnerable to stressors such as crime during pregnancy.37 The results of a 2008 Canadian study are consistent with this hypothesis. The Canadian researchers found that women's perceptions of neighbourhood security were associated with SGA births.38

Neighbourhood-level stressors may also influence risk factors for SGA birth. A recent study examined the effects of neighbourhood environment on birth weight and used sequential models to separate the effects of health-related behaviours from the effects of sociodemographic factors. That study, which included all infants born to 726 women, found that smoking, drug use and delayed prenatal care explained 30% of the neighbourhood effect, whereas stress explained 12%.28

The present study has several limitations. First, we did not have information on maternal risk factors such as smoking,39 drinking and/or drug use that may be related to both neighbourhood deprivation and risk of SGA births. Individual-level lifestyle factors such as smoking and drinking are difficult to assess in an entire population, which was the study population in the present study. Second, it is possible that residual confounding exists because SES cannot be measured entirely by family income, educational attainment and employment status. Third, the sample sizes were too small to allow for an additional analysis of women from specific non-Swedish countries, some of whom may have higher risks of SGA birth. This could be a limitation if ethnicity is one of the mechanisms underlying the neighbourhood effect on SGA births. Fourth, the variable for marital status underestimated the proportion of women who were cohabiting with a partner. However, we kept this variable in the analysis because there was an association between marital status and risk of SGA birth. In addition, any bias would rather result in an underestimation of the risk estimates rather than an overestimation.

The limitations of the study are countered by its strengths, which include: (1) the ability to analyse data on a large national cohort of women over a period of 9 years (20 487 SGA births), (2) the prospective design, (3) the completeness of data (eg, only 0.3% of the data on infant birth weights, 0.1% of the data on gestational age and 1% of the data on maternal education and family income were missing), (4) the use of small, well-defined neighbourhoods with an average of 1000 people each, (5) the high validity of SGA diagnoses based on ultrasound examinations that are used to confirm 95% of all pregnancies in Sweden13 and (6) the ability to adjust for a set of individual-level maternal sociodemographic factors, including family income. This last strength is particularly important, as family income is a major confounder that can affect an individual's choice of neighbourhood. Another strength is the possibility of generalisation of our results to other populations (external validity), particularly to populations in industrialised societies. Finally, the ability to measure exposure to neighbourhoods in the 2 years preceding the birth of the infant is a major strength. Nearly 25% of all pregnant women move during their pregnancy,40 which makes it difficult in most studies to estimate neighbourhood exposure in the period immediately preceding and during pregnancy.40

This nationwide, prospective study showed that after accounting for individual-level sociodemographic factors, neighbourhood-level deprivation increased the risk of SGA birth in first-born singletons. These findings are valuable for healthcare professionals who work in neighbourhoods with varying levels of deprivation.


The authors thank Kimberly Kane, scientific editor, for useful comments on the text.


View Abstract


  • Funding This work was supported by a grant to JS from the National Institute of Child Health and Human Development (NICHD) (1R01HD052848-01) and grants to JS and KS from the Swedish Research Council (2008-3110 and 2008-2638) and the Swedish Council for Working Life and Social Research (2006-0386, 2007-1754 and 2007-1962).

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Ethics Committee at Lund University.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.