Contribution of socioeconomic status to the risk of small for gestational age infants – a population-based study of 1,390,165 singleton live births in Finland

Data were extracted from the Finnish Medical Birth Register (MBR), which is a compilation of the clinical records of all of the obstetric care units in Finland. The MBR is currently maintained by the National Institute for Health and Welfare (THL), which authorized the use of sensitive health register data in scientific research for the years 1987–2010, as required by national data protection legislation.

Permission to use the confidential register data in this study was granted on February 16^th, 2012 by the National Institute for Health and Welfare (THL) in Finland. (Reference number 1749/5.05.00/2011). Informed consent of the registered individuals was not needed since only anonymized data from national health registers were used, and the individuals were not contacted.

SGA was defined according to the International Societies of Pediatric Endocrinology and the Growth Hormone Research Society as a birth weight more than 2 SDs below the sex- and gestational age-specific reference mean [15]. The Finnish population-based reference for birth size was used to classify the newborn infants as SGA (Unpublished Finnish data from THL). Gestational age was estimated based on parturient’s last menstrual period, unless there was a discrepancy with the first- or second-trimester ultrasonography measurements of more than seven or 14 days, respectively, in which case the latter measurements were used. Information on gestational age and birthweight was missing in 8,754 (0.6%) and 1,304 (0.1%) cases, respectively, and therefore those cases were excluded from the analyses. Information on in-vitro fertilization (IVF) was only available from 1991 onwards. However, only 170 IVF children were born in Finland before 1991. IVF also included intracellular sperm injection (ICSI) and frozen embryo transfers (FET). Maternal anemia was defined as haemoglobin levels below 100 g/L. Maternal smoking was self-reported and recorded as non-smoking or smoking during pregnancy. After 1991, information was also available about the number of women who quitted smoking during the first trimester or continued smoking throughout the pregnancy. Information on maternal height and weight was available from 2004 onwards, and thus multivariate analyses of the body mass index (BMI) were only performed for the years 2004–2010. BMI was calculated by dividing body weight in kilograms by the squared height in meters (kg/m²). Information on socioeconomic status (SES) based on mother’s occupation at time of birth was gathered in the register from October 1990, and thus used in the analyses for the years 1991–2010. SES was recorded as either upper white-collar workers such as teachers and physicians, lower white-collar workers such as nurses and secretaries, blue-collar workers such as cookers and cleaners or “others”, as described in detail elsewhere [10]. The category “others” included entrepreneurs, housewives, students, retired, unemployed and unclassifiable women. Information on SES was missing in 151,694 (13.2%) cases, and hence they were analyzed as a separate group in the multivariate analyses. Parturients’ place of residence was grouped by the twenty hospital districts in Finland. Ethnic variation is Finnish population was minor during the study period [16]. Marital status was recorded as married or cohabiting, or single. The study period from 1987–2010 was divided into five time periods (1987–1991, 1992–1996, 1997–2001, 2002–2006, 2007–2010) to examine the variation in SGA incidence over time.

Information on major congenital anomalies was obtained from the Finnish Register of Congenital Malformations established in 1963 and currently maintained by THL. The two data sources were linked together using encrypted unique personal identification numbers. Apart from number of deliveries, miscarriages and prior terminations, the data were dichotomous or categorical.

Statistical differences in frequencies (categorical and dichotomous variables) between women with SGA infants and the reference population were evaluated by Pearson’s chi-square test. Differences between continuous variables were evaluated by Student’s t- and Mann–Whitney U-tests as appropriate. Multivariate logistic regression analyses were used to model the risk factors for SGA in comparison with controls in 1991–2010 due to availability of information on SES and smoking status. Odds ratios (OR) with 95% confidence intervals (CI) were calculated. Possible confounding variables were selected based on univariate analyses (p < 0.05). Differences were deemed to be significant if p < 0.05. Furthermore, to quantify the contribution of SGA risk factors (smoking, parturient’s place of residence grouped by hospital district, amniocentesis, placenta previa, placental abruption, congenital anomaly and prior cesarean section (CS)) to socioeconomic disparities, we evaluated the percentage reduction in the SES – SGA association in sequential logistic regression models. Each covariate was added separately to a partially adjusted model (Model 3: adjusted for smoking, SES, maternal age, and parity), and the percentage reduction in the odds ratio was calculated. The formula used was (OR Model 3 – OR Model A/B/C/D) / (OR Model 3 – 1) [17]. The data were analyzed using SPSS for Windows 19.0, Chicago, IL.

The aim of the present study was to identify risk factors for SGA among singleton live births and quantify their contribution to socioeconomic disparities in SGA. Out of the total population of singleton live births (n = 1,390,165) in Finland during the study period (1987–2010), 3.1% (n = 42,702) were classified as SGA, which corresponds well to a Swedish study that used the same definition of SGA and reported a 2.0% SGA incidence for the period 1999 to 2010 [13]. In the fully adjusted models, we detected a 11 − 24% higher SGA incidence in the lowest SES groups in comparison with the highest SES group, and an up to 1.5-fold difference in SGA incidence among the twenty hospital districts. The main finding of the present study was that smoking alone made the largest contribution, explaining about 40 − 50% of the excess risk of SGA among the lower SES groups. In contrast, the combined contribution of amniocentesis, placental abruption, placenta previa, congenital anomaly and prior CS was only around 4%.

The most important strength of our study was that the data derived from the national, mandatory MBR covered the entire population, and therefore offered a comprehensive view of the risk factors for SGA. The MBR content is extensive, and thus allows the use of numerous birth characteristics, perinatal outcomes and socioeconomic factors as exposures. Further, to best of our knowledge, the present study population is one of the largest studied to date. On the other hand, this kind of register information might include errors and missing values because the data are produced mainly for administrative and statistical purposes rather than for research. However, the MBR has been shown to have excellent data coverage and quality [18, 19]. A possible limitation of the study was that the gestational age was not estimated by ultrasonography for all women, which may have masked or indicated certain trends in the SGA rate across the time periods (because data for the earlier periods were more reliant on the last menstrual period without correction). Further, a possible limitation was that SES was determined by parturients’ occupation during pregnancy and we had no information on maternal educational or family income. In general, however, both of these variables are known to be related in our country to the occupation which in turn is a register-based, available indicator for studies on socioeconomic health differences in the perinatal period [10, 20]. Spouses’ information was not available in the MBR due to the confidentiality of such information by the Finnish law.

Our results are in line with previous results and confirm that the etiology of SGA is multifactorial and substantially affected by SES [9]. After adjustment for background information, the risk of SGA was 11 − 24% higher in the lowest SES group than in the highest SES group. The risks of SGA and other adverse pregnancy outcomes have been shown to be affected by SES in multifarious ways. For example, exposure to harmful substances may result from occupational, residential and lifestyle factors that are socially patterned [9]. Our results revealed an up to 1.5-fold difference in SGA risk due to the parturient’s place of residence (after adjustment for background information). In the hospital districts with high SGA risk (districts 5, 6 and 16), the gross domestic product (GDP) per capita and unemployment rate have been reported to be among the lowest and highest in Finland, respectively (Local Finland, http://​www.​localfinland.​fi/​en/​Pages/​default.​aspx). The main finding of the present study was that smoking alone made the largest contribution, explaining about 40 − 50% of the excess risk of SGA among the lowest SES groups. This implies that smoking during pregnancy resulted in 0.2 extra cases of SGA per 1,000 births in lower white-collar workers compared to 5.3 extra cases per 1,000 births in blue-collar workers. Among the total population, smoking during the first trimester and throughout the pregnancy resulted in 624 and 5850, respectively, additional cases of SGA during the study period. The apparently large contribution of smoking to socioeconomic disparities in SGA is in line with the results of a previous study that utilized data from the Finnish MBR for the years 1991 to 1999 [10]. The strong relation between smoking and birth weight is not a novel and unexpected finding, and the relationship has even shown to be dose-dependent [12]. However, our data did not contain information on the number of cigarettes smoked per day, and hence we did not examine the dose dependence. Nevertheless, the Finnish MBR has been shown to cover daily smoking habits during pregnancy relatively well [21]. Our results showed that women who quitted smoking during the first trimester of pregnancy had only a 7% higher risk of SGA to that of non-smokers, suggesting that interventional measures to reduce the socioeconomic gradient of SGA might prove effective. It could be argued that the inclusion of women (13.2%, n = 151,694) with missing information on SES may have biased the results. However, the incidence of SGA (3.4%) and proportions of women who gave up smoking (3.4%) or were smokers (14.1%) among the women with missing SES information were similar to those of the general population, showing that the results were unlikely to be biased by the missing information.

Our results also confirmed associations between SGA and several reproductive risk factors, such as nulliparity [6, 7], advanced maternal age [6], BMI ≤24.9 [5, 22], placental abruption [6, 23], and congenital anomalies [24]. Furthermore, we found an association between amniocentesis or prior CS and SGA. Amniocentesis has previously been shown to be associated with an increased risk of pregnancy loss [25]. However, in our study, amniocentesis was probably a non-causal indicator of a high-risk pregnancy. Vaginal birth after prior CS (VBAC) has also been shown to be associated with other adverse maternal and neonatal outcomes [26, 27]. Our results suggested that the combined contribution of amniocentesis, placenta previa, placental abruption, congenital anomaly and prior CS to the excess SGA risk among the lowest SES groups was up to 3.6%. However, we were not able to investigate all possible confounding factors found in previous studies, e.g., maternal chronic diseases, such as gestational hypertension [6] and preeclampsia [6] due to limitations in the availability and quality of data used. Furthermore, our results showed that the risk of SGA was inversely associated with gestational age, in particular, the incidence of SGA decreased by 41 − 52% after gestational week 40. This indicates that women with SGA were diagnosed and treated before term.