Birthweight reflects intrauterine growth and wellbeing and is recognised globally as an indicator of infant, child health and increasingly adult health [1]. A low birthweight (< 2500 g) has previously been associated with increased infant mortality and morbidity [2, 3], poorer education outcomes and developmental delay in childhood [4, 5] and an increased risk of adult disease [6] via a range of programming mechanisms. It has been suggested that one such mechanism is the potential for reduced immune function [7] and an inverse association between birthweight and infection related morbidity has been identified in children up to the age of 14 [8], and there is some evidence of longer term effects of birthweight on the immune system where antibody response to vaccination in teenagers and adults is lower in those who were small at birth [9, 10]. Worldwide mean birthweight is lower and the incidence of low birthweight higher among South Asian origin populations when compared to White US/European populations [11, 12]. In South Asia, this is in part thought to reflect environmental exposure to poverty and poor nutrition. However, babies born in high income countries such as the UK, to mothers of South Asian origin are considerably lighter (around 200-300 g) than babies born to White British mothers and this difference does not appear to reduce over subsequent generations of UK South Asians [13, 14]. This has led to the suggestion that differences may not be expressions of growth restriction but rather are genetically or culturally programmed [15–17], and in the absence of environmental risk factors, may not necessarily lead to increased mortality or morbidity i.e. some South Asian infants may be small and healthy rather than small and at risk. Whether it can be proven that smaller size is a normal phenomenon within South Asians or not, the important question is whether the risks of greater mortality or morbidity arising from smaller birth size still apply. Customised growth charts have been developed that take into account ethnicity and other maternal factors [15] but when these have been evaluated in terms of neonatal risk, there is no strong evidence that these better predict the risk of adverse outcomes than population based charts [18]. Whether this is also the case beyond the neonatal period, to our knowledge remains unclear.

Our aim in this paper was to examine the association between being born small and early childhood morbidity estimated using use of health services information for White British and Pakistani origin children aged 0–3 participating in the Born in Bradford (BiB) cohort study. We chose GP prescribing as a marker of morbidity and selected the three most common prescribing categories: analgesics, antibiotics and bronchodilators, where analgesics was the most common category followed by antibiotics and third, bronchodilators. We also considered the number of GP consultations and emergency and elective hospital episodes as further morbidity indicators. We defined being born small in two ways using a cut-off of being born weighing less than 2500 g and also using customised birthweight charts.

Table 3 and Fig. 2 a – d show the adjusted IRRs for Pakistani children relative to White British children for each marker of morbidity. Pakistani children generally had a higher rate of episodes for all outcomes compared to White British children whether they were normal weight or categorised as small at birth by either method. They had 48–55% more GP appointments depending on the birthweight category, compared to White British children. The IRR for Pakistani children relative to White British children for analgesic prescriptions ranged from 1.76 (95% CI 1.37, 2.25) to 2.31 (95% CI 2.19, 2.45) across the categories of normal and small birthweight and antibiotic prescriptions ranged from 13 to 38% higher among Pakistani origin children. Compared to White British children, bronchodilator prescriptions were slightly more common among Pakistani children categorised as being small at birth (by either method) compared to White British children, although there was no strong statistical evidence for this difference. The incidence of Hospital Emergency episodes was greater among Pakistani children and especially where children were born small although again, these results were not statistically significant which in this case, mostly reflects the small number of emergency episodes overall. The incidence of Hospital Elective episodes was markedly higher among Pakistani origin children compared to White British children where they were classified as normal weight using either the 2500 g cut-off or the customised charts (IRR 1.65 (95% CI 1.21–2.25) and 1.49 (95% CI 1.09, 2.04) respectively). However, among children born small the IRR was 0.61 (95% CI 0.26, 1.44) for children categorised as low birthweight (< 2500 g) and 0.84 (95% CI 0.29, 2.47) for those defined small using the customised charts.

GP appointments	White British n(%)	Pakistani n(%)	Incidence Rate Ratio Pakistani/ White British
Not low birthweight	3913 (95.0%)	4319 (91.3%)	1.49 (1.44–1.55)
Low birthweight	206 (5.0%)	412 (8.7%)	1.48 (1.27–1.73)
Not SGA	3658 (93.8%)	3899 (86.4%)	1.49 (1.43–1.55)
SGA	241 (6.2%)	613 (13.6%)	1.55 (1.36–1.76)
Analgesic prescriptions
Not low birthweight	3913 (95.0%)	4319 (91.3%)	2.31 (2.19–2.45)
Low birthweight	206 (5.0%)	412 (8.7%)	1.76 (1.37–2.25)
Not SGA	3658 (93.8%)	3899 (86.4%)	2.30 (2.17–2.44)
SGA	241 (6.2%)	613 (13.6%)	1.92 (1.54–2.39)
Antibiotic prescriptions
Not low birthweight	3913 (95.0%)	4319 (91.3%)	1.38 (1.30–1.46)
Low birthweight	206 (5.0%)	412 (8.7%)	1.13 (0.88–1.46)
Not SGA	3658 (93.8%)	3899 (86.4%)	1.38 (1.30–1.48)
SGA	241 (6.2%)	613 (13.6%)	1.26 (1.00–1.59)
Bronchodilator prescriptions
Not low birthweight	3913 (95.0%)	4319 (91.3%)	0.95 (0.84–1.09)
Low birthweight	206 (5.0%)	412 (8.7%)	1.02 (0.58–1.77)
Not SGA	3658 (93.8%)	3899 (86.4%)	0.96 (0.84–1.10)
SGA	241 (6.2%)	613 (13.6%)	1.10 (0.65–1.87)
Hospital Emergency
Not low birthweight	3913 (95.0%)	4319 (91.3%)	1.24 (1.10–1.40)
Low birthweight	206 (5.0%)	412 (8.7%)	1.42 (0.87–2.31)
Not SGA	3658 (93.8%)	3899 (86.4%)	1.20 (1.06–1.36)
SGA	241 (6.2%)	613 (13.6%)	1.46 (0.92–2.32)
Hospital Elective
Not low birthweight	3913 (95.0%)	4319 (91.3%)	1.65 (1.21–2.25)
Low birthweight	206 (5.0%)	412 (8.7%)	0.61 (0.26–1.44)
Not SGA	3658 (93.8%)	3899 (86.4%)	1.49 (1.09–2.04)
SGA	241 (6.2%)	613 (13.6%)	0.84 (0.29–2.47)

^aLow birthweight models adjusted for maternal parity, infant sex, gestational age, maternal age, social economic factors (maternal education, housing tenure, means- tested benefits) and smoking; SGA models adjusted for maternal age, social economic factors (maternal education, housing tenure, means- tested benefits) and smoking

In all analyses, results were generally similar whether small at birth was defined using the cut-off for low birthweight of a birthweight less than 2500 g or defined using the customised SGA charts. Adjustment for social economic variables did not markedly alter the results (see Additional file 1: Tables S1 and S2).

To our knowledge, this is the first time that detailed research information has been linked to primary and secondary care outcome data to examine the association between being born small and early childhood morbidity. Consistent with previous results using this cohort [13, 22] and other UK studies [12, 23], Pakistani children had a lower mean birthweight and were more likely to have a low birthweight (< 2500 g) compared to White British children. Being born small has previously been associated with an increased risk of adverse neonatal outcomes [24] and there is some association with adult morbidity [6, 25] although this is possibly modified by adult risk factors, for example adult BMI [26]. There is however, a notable lack of evidence to identify whether similar associations are present in childhood. We used GP consultations, prescription data for the three most common prescriptions, and hospital episode information as markers of morbidity in a cohort of children all born and growing up in the same UK city. We found that in both ethnic groups, children who were born small, regardless of how that was categorised (i.e. either low birthweight or SGA), had a higher incidence of most markers of morbidity from birth to age 3, compared to children of normal weight (i.e. not low birthweight or SGA). This suggests that the association of being born small and poorer health outcomes identified in the neonatal period [24] may persist into early childhood and underlines how prevention of low birthweight remains important to the development of public health interventions.

We found that compared to White British children, Pakistani children had a higher incidence of all morbidity markers with the exception of bronchodilator prescriptions and Hospital Elective episodes. This greater health service use among Pakistani origin children is consistent with other studies that report higher rates of GP consultations [27], and hospital emergency admissions [28, 29] among ethnic minority groups. It has previously been suggested that these differences might in part be explained by social and economic differences between groups, however our results did not differ substantially with or without adjustment for social and economic markers (Additional file 1: Tables S1 and S2). In contrast to previous evidence of higher rates of antibiotic prescriptions among UK white populations across all age groups [30], we found antibiotic prescriptions were more common for Pakistani children compared to White British children across all birthweight categories but especially among normal weight children (defined by either method) where the rate of prescriptions was up to 38% higher. Given the current concerns around antibiotic use [31], these population differences require further investigation and may be a potential area for future targeted interventions to protect the population’s health. Pakistani children also had on average, a greater number of prescriptions for analgesics, which were the most common prescription category in both ethnic groups and may reflect current UK guidance for optimising analgesia for the treatment of childhood infections [32]. We found that bronchodilator prescriptions were slightly more common in White British children despite previous findings from this cohort that identified a higher proportion of Pakistani children as being diagnosed with asthma compared to White British children (13.4% compared to 8.9%) [33]. Our results are consistent with other studies that have identified differences in specialty asthma care and higher rates of related emergency admissions among Pakistani populations [29, 34] and also with the wider possibility that minority ethnic groups may receive less preventive health care which leads to greater use of emergency care [34]. Our finding that emergency episodes were more common among Pakistani children across all birthweight categories seems to support this. Similarly, previous studies have identified that South Asians are more likely to consult their GP but are less likely to be referred to secondary care [27, 35]. Here, our results are partly consistent with this in that we found that Pakistani children had a higher rate of GP consultations in all birthweight categories but that the rate of elective episodes was only lower for those children defined as small at birth. In normal birthweight children, the rate of elective episodes was on average higher than that for White British children. GP decision making is likely to be independent of birthweight therefore rather than being suggestive of ethnic differences in elective referral, fewer elective episodes among Pakistani children defined as small might suggest that some of these children are born small and healthy as opposed to small and at risk thus needing less elective care. However, the rate of elective episodes was lower for those Pakistani children born small even when customised growth charts were applied, our other markers of morbidity do not support the possibility of less morbidity in these children and our estimates are based on a small number of elective episodes and the confidence limits for the ethnic difference suggest some uncertainty.

We used two methods to identify children who were born small. First, we used a cut-off of birthweight less than 2500 g as although this is a crude measure [36], it is well recognised and established as an indicator of health [37]. Second, we used GROW customised birthweight charts which adjust for ethnicity and maternal characteristics (height, BMI, age and parity). In our study population, we found that when using the cut-off of birthweight < 2500 g, 8.7% of Pakistani children and 5% of White British were defined as low birthweight. When we applied the customised charts, more children in both ethnic groups (13.6 and 6.2% respectively) were defined as SGA compared to the number who were categorised as low birthweight (< 2500 g) but this was especially marked among Pakistani children. Despite being intuitively appealing, customised charts have not improved the prediction of growth restriction [18] or adverse neonatal outcomes [38, 39] compared with population standard charts. Here, whilst there are some differences in outcomes between low birthweight and SGA definitions, we have found no robust evidence that customised charts better predict early childhood morbidity than a crude cut-off of being born weighing less than 2500 g.

The key strengths of this study are our linkage of research data with routine primary and secondary care data which has allowed us to examine the effects of being born small on child health beyond the perinatal period, our detailed ethnicity information and the ability to adjust for a range of covariables. A limitation of our study is that our outcomes may not accurately reflect morbidity for a number of reasons. First, the data are dependent on the accuracy and quality of coding. Second, for hospital episodes the sample numbers are small in some categories and this is evident in the confidence intervals for these outcomes. Third, GP consultations include routine appointments (for example immunisations) that may not be indicative of illness however, we expect this to not differ markedly between the two ethnic groups. Over 99% of analgesic prescriptions were paracetamol or paracetamol based and it is possible that some of these prescriptions may have been associated with routine immunization but we were unable to examine this with the data available, however if this is the case we do not expect it would differ markedly between the two groups’. Likewise, Hospital Emergency episodes will include accident related episodes that do not necessarily reflect morbidity, it is possible that these may differ between the two groups but we are not able to examine this further with the data we have available. Children born prematurely might have a greater risk of respiratory illness and wheeze and as such experience greater morbidity in early childhood [40], however prematurity did not differ markedly between the two ethnic groups and gestational age was accounted for in all models whether low birthweight was defined as less than 2500 g or using customised growth charts. In addition, we were only able to examine Pakistani origin children due to the small number of other South Asian groups in the BiB cohort. This means that our results may not be generalisable to other South Asian groups.

These results suggest that being categorised as small at birth is associated with increased morbidity estimated using health service use information, in early childhood in both White British and Pakistani origin UK children. This combined with evidence that birthweight is inversely associated with neonatal mortality, educational achievement and adult disease risk [2–6], highlights the importance of birthweight to health throughout the life-course and that the development of interventions to reduce low birthweight, remains a public health priority. Overall, Pakistani children access primary and secondary health services more frequently and are more commonly prescribed analgesics and antibiotics than White British children irrespective of whether they are born small or how this is defined. This has implications for health service planning in areas with large South Asian populations and suggests a need for a better understanding of ethnic differences in health service use. Despite the marked difference in the criteria used to define low birthweight and SGA, we found our results did not differ substantially using either method which supports the suggestion that customised charts do not necessarily better predict outcomes.