Socioeconomic conditions have been shown to influence the risk of many types of ill health in Sweden. This is manifested as increased mortality and in-hospital care due to accidents and diseases such as cardiovascular disease and lung cancer in lower socioeconomic position (SEP) groups [1, 2], while there are smaller differences in out-of-hospital medical care consumption [3]. However, studies on social differences in medical care consumption have been criticized for neglecting or excluding the use of health services by children and young people, despite evidence that people in younger age groups form a significant proportion of all healthcare users [4]. The social patterning of medical care consumption seen in children does not necessarily follow the pattern seen among adults, due to differences in disease patterns and factors such as costs and availability. In Sweden, medical care visits are free for children under the age of 12 years, and the drugs prescribed are subsidized with taxes. Furthermore, there is a well-developed preventive care organization offering regular visits until the child is 5-6 years old, controlling his or her physical and psychological development and administering vaccinations according to a general base program. These consultations are free of charge and optional.

It is widely acknowledged that the social position of the family is closely related to the health risks that small children are exposed to, and so the environment in which children grow up is closely associated with their health [1, 5]. The family's socioeconomic situation affects the child's health from the very beginning, through the mother's living conditions and health habits during her pregnancy. Parental smoking, short duration of breast feeding and low social support are more common in lower socioeconomic groups, and have all been shown to be associated with increased morbidity in infancy [1, 6–12], which in turn might lead to increased medical care consumption. However, to date there are only a few studies on socioeconomic differences in medical care consumption among small children [13, 14]. To the best of our knowledge, there is no study that has investigated the contribution of differences in parental behavioral factors and social support to socioeconomic differences in medical care consumption. The aim of the present study was to investigate potential differences in parental preventive behavior, risk behavior, social support, and the use of medical care among small children in Malmö, Sweden, by maternal educational level. We also wanted to examine whether potential differences in child medical care consumption by maternal educational level could be explained by differences in parental behavior and social support.

Among mothers without breastfeeding problems 83% exclusively breastfed their child for at least 4 months, while the corresponding proportion among mothers with such problems was 42%. Many of the factors associated with a shorter period of exclusive breastfeeding were also more often associated with breastfeeding problems, such as presence of secondhand tobacco smoke at 0-4 weeks, low birth weight, and having been admitted to neonatal care ward after delivery (data not shown). However, there were no differences in the frequency of reported breastfeeding problems between those of low and higher educational level (data not shown).

Although, there was no difference in the frequency of breastfeeding problems according to mother's educational level, the types of problems were different in mothers with different educational levels and the various types of breastfeeding problems also showed clear differences in terms of the effect on length of breastfeeding. Insufficient breast milk was associated with the lowest proportion of exclusive breastfeeding ≥4 months (Figure 1). Various types of breastfeeding problems also differed by maternal educational level (Figure 2).

Figure 3 shows the association between exposure to unfavorable parental behavioral factors and child medical care consumption. Increased exposure to unfavorable parental behavioral factors (maternal smoking during pregnancy, secondhand tobacco smoke during 0-4 weeks of age, and duration of exclusive breastfeeding <4 months) was associated with increased odds of child medical care consumption (i.e., doctor's visits and in-hospital care). The odds were doubled when exposed to all three risk factors (Figure 3).

The results showed that maternal educational level is closely related to the health risks that small children are exposed to. Children with less-educated mothers were exposed to more health risks, fewer health promoting factors, worse social support, and had a higher medical care consumption than children whose mothers had higher levels of education. After adjustment for parental behavioral factors (maternal smoking during pregnancy, secondhand tobacco smoke during the child's first four weeks of life, and exclusive breastfeeding <4 months), the excess odds of doctor's visits and in-hospital care among children with less-educated mothers were reduced and became statistically non-significant.

It is not easy to make international comparisons in medical care consumption due to differences in factors such as availability and fees. A study from Great Britain [18] showed that children and young people living under materially deprived conditions were more likely to be admitted to hospital. However, two other studies, one from Sweden and one from Great Britain, found no socioeconomic differences in the use of health services by children [4, 19]. According to another Swedish study (published in 2000) among children aged 1-5 years, less-educated parents were less likely to have paid a visit to a physician when their child had an infection and less likely to have consumed antibiotics, compared to more highly-educated parents [20]. These results are in contrast to the findings in one of our previous studies, published in 2009, where we showed that the odds of small children's use of antibiotics increased as parental educational level decreased [21]. Some of these differences in results might perhaps be related to differences in day care and reduced costs of medical care visits for children. Low educational groups are likely to be more sensitive to the costs of medical care.

The differences in child medical care consumption were reduced after adjustment for parental behavioral factors, i.e., maternal smoking during pregnancy, early exposure to secondhand tobacco smoke, and duration of exclusive breastfeeding <4 months. These factors may theoretically be regarded as intermediate factors in the same causal chain and were shown to be associated both with maternal educational level and medical care consumption. There may be different mechanisms underlying the association between such behavioral factors and medical care consumption. Earlier studies have shown breastfeeding to be protective and associated with decreased morbidity in infancy with a lowered incidence of infectious diseases [8, 22], asthma [9], overweight [23], and some childhood cancers [10], as well as a better cognitive development [11]. Maternal smoking during pregnancy has been shown to increase the odds for having a child with cleft lip/palate [24], decreased lung function [1], colic pain [25], and meningococcal disease during the first years of life [26]. Maternal smoking and environmental smoking have also been shown to be associated with wheeze in infancy [7]. A recent review showed that secondhand tobacco smoke was associated with both the onset and severity of asthma, as well as respiratory and middle ear infections among children [27]. The latter effect has been shown to be of particular importance during the first two years of life [1].

Earlier studies have shown that parental behavior is influenced by material resources and knowledge of favorable life styles [1]. Similarly, in the present study, children with lower educated mothers also more often lacked the protective advantages of breastfeeding. The most common problem with breastfeeding among less-educated mothers was insufficient breast milk, which in turn was the breastfeeding problem associated with the lowest proportion of exclusive breastfeeding ≥4 months. Knowledge about the type and occurrence of the various problems related to duration of breastfeeding is important in order to allow the construction of effective breastfeeding promotion programs.

Low emotional and practical support have previously been shown to be associated with health care seeking [28, 29]. For example, a recent American study showed that children whose parents experienced low emotional support had higher odds for having paid visits to primary care with the child [28]. Poor maternal psychosocial functioning may influence the child's psychosocial and physical health both directly, by inhibiting the mother's ability to care for the child, and also indirectly through her limited personal resources to help her child by turning to a physician for assistance [29].

Certain methodological issues need to be addressed. First, about two thirds (65%) of the parents who received the study questionnaire responded to it. An earlier study showed only small differences between participants and the general population with regard to the parental educational level, but with an underrepresentation of children whose parents were born outside Sweden [17]. Such an underrepresentation might affect absolute descriptive measures such as prevalences and proportions. However, when studying associations, it is more important that the relation between exposure and disease is similar for those who participate in the study, and those who are theoretically eligible for the study, including those who do not participate, rather than strict criteria of representativity [30]. The likelihood of such selection bias in a given study is often assessed by subjective reasoning. Second, self-reported data on parental behavioral factors is a potential source of bias. A recent review found a tendency to underestimate self-reported smoking, especially when smoking is seen as undesirable, such as during pregnancy [31]. Such an underestimation would lead to a dilution of the true associations. However, there are studies that have validated questionnaire information on presence of maternal smoking through blood samples measuring cotinine and thiocyanate in cord serum, reporting good agreement between high and low levels of biomarkers and daily and non-smoking mothers [32]. Self-reported length of breastfeeding is another potential source of bias. Previous studies on the validity of self-reported breast-feeding have been made [33, 34]. A review on the validity and reliability of maternal recall of breastfeeding practice concluded that maternal recall is a valid and reliable estimate of breastfeeding duration, especially when the duration of breastfeeding is recalled after a shorter period (≤ 3 years). However, validity and reliability of maternal recall for the age at introduction of food and fluids other than breast milk was less satisfactory [34]. Thus, there might be some recall error with regard to self-reported length of breastfeeding in the present study. Self-reported visits to medical care might be another potential source of bias. An American study, of self-reported information about children's hospitalizations and emergency visits, showed that parents had good recall information over the first 3 years of a child's life [35]. Although some degree of recall error cannot be ruled out, it would, if anything, most likely be of a nonsystematic nature, thus attenuating the true association between maternal educational level and medical care consumption. Third, we use maternal educational level as a measure of socioeconomic position (SEP). There are also other well-known measures of SEP such as parental occupational level and income that could be used, but, unfortunately, we do not have information on these other measures of SEP.

In conclusion, the results show that children with less-educated mothers were exposed to more health risks, fewer health-promoting factors, worse social support and had a higher medical care consumption than children with higher-educated mothers. Furthermore, the differences in child medical care consumption by maternal educational level emphasize the notion that children's health seems to be influenced by the characteristics of the families into which they are born. Thus, improving children's health calls for policies that target the parent's health-related behaviors and social support.

The authors declare that they have no competing interests.