Over the last two decades it has become increasingly acknowledged that events occurring in early life have a lasting effect on an individuals’ health [1]. The diet of infants and toddlers has short and long-term implications for health and development [2,3,4]. In the first two years of life, an infant’s diet changes from milk as the sole food source, to foods and beverages that reflect the family diet. This transition represents the most rapid change in diet over the life course and is the developmental period when dietary preferences and habits are first established [5]. It is important to be able to characterize early life dietary patterns and understand the composition of such patterns in order to examine how early diet influences later outcomes.

Principal component analysis (PCA) is a technique for characterizing the whole diet. PCA is based on correlations between foods and is used to identify foods that are frequently consumed together. Individuals are assigned a score on each dietary pattern and higher scores indicate a greater intake of the foods associated with that pattern.

A number of studies have characterized dietary patterns in infancy and toddlerhood [6,7,8,9,10,11]. For example, using dietary questionnaires at 6 and 15 months of age, we identified four dietary patterns among a United Kingdom birth cohort [7]. A traditional stylepattern characterized by home-prepared meats, vegetables and desserts; aready-prepared baby foods pattern; and a discretionary pattern characterized by foods such as biscuits, sweets and crisps were identified at both ages. At 6 months, the fourth pattern was characterized predominantly by breastfeeding and at 15 months, by contemporary-style foods including herbs, legumes, nuts, raw fruit and vegetables. Similar early dietary patterns have been reported in other studies, including in a more contemporary UK birth cohort [8] and other countries including Norway [6], the Netherlands [11] and Japan [9].

While dietary patterns provide a useful summary measure of diet, it is important to understand the energy and nutrient intake underlying the pattern. However this is rarely examined, as we have been able to identify only two studies involving young children that report associations between nutrients and dietary patterns [11,12]. In 14-month-old toddlers from the Netherlands (n = 2420), weak correlations were observed between health-conscious or Western-like dietary patterns and energy and macronutrient intakes. The health-conscious pattern was positively correlated with energy, protein and polysaccharide intake, and was inversely correlated with saturated fat intake. The Western-like pattern was positively correlated with energy and saturated fat intake and inversely correlated with protein and polysaccharide intake. Shin et al. found that the total energy intake of five-year-old Korean children was associated with higher scores on each of the three patterns they identified (Korean healthy; animal foods; sweets patterns, n = 1441). Energy-adjusted intake of protein, fiber, iron, calcium, vitamin A, carotene and vitamin C was associated with higher scores on the Korean-healthy dietary pattern. In contrast, scores on the sweets pattern was inversely associated with nutrient intakes.

Understanding the energy and nutrient intakes that underpin a dietary pattern provides a measure of the patterns’ validity and can improve our understanding of the possible pathways through which diet might influence health and development. Therefore the aim of this study is to determine whether dietary patterns obtained from questionnaires at 6 and 15 months of age are associated with nutrient intakes assessed by diet record at 8 and 18 months, respectively, using data from the Avon Longitudinal Study of Parents and Children (ALSPAC).

ALSPAC is an ongoing longitudinal birth cohort study that is designed to investigate the determinants of health and development [13]. All pregnant women residing in the former county of Avon, Southwest England who were expected to deliver between 1st April 1991 and 31st December 1992 were invited to participate in ALSPAC. Information about the ALSPAC cohort and study questionnaires are published on the ALSPAC website. The study sample is considered to be broadly representative of the population although ethnic minorities, single parents and unmarried couples are slightly underrepresented compared with the 1991 National Census of women residing in Avon. Ethical approval for the study was obtained from the ALSPAC Ethics and Law Committee and the Local Research Ethics Committees. The core ALSPAC sample consists of 14,541 pregnancies and of these pregnancies, 13,988 infants survived to 1 year of age. Children in Focus (CIF) is a subsample of 1432 children born in the last 6 months of the cohort recruitment (July to December 1992) that were invited for regular clinical assessments from birth. The study described in this paper involves the 10% CIF subsample of children, whose nutrient intakes were collected at 8 and/or 18 months of age using an un-weighed 3-day diet diary [14,15], and for whom dietary patterns were extracted from dietary questionnaires at 6 and/or 15 months of age, respectively [7].

Dietary pattern scores at 6 months and nutrient intakes at 8 months were available from n = 725 participants. Dietary pattern scores at 15 months and nutrient intakes at 18 months were available from n = 535 participants. The characteristics of the participants included in the analyses at 6–8 months and at 15–18 months of age were similar (Table 1). Compared with the core ALSPAC cohort, participants included in the present analyses were born to mothers who were slightly older, more likely to be primiparous, have a higher level of education, from social classes II and III, and were more likely to have never smoked or quit smoking. Although the magnitude of these differences is quite small, the statistical analysis was significant at the p < 0.01 level due to the large sample size. The proportion of males to females and singletons to twins did not differ from the core cohort.

The present study shows that the dietary patterns of infant and toddlers exhibit variation in their underlying energy and nutrient composition. Each pattern was associated with a different profile of nutrients. We have previously shown that the intake of foods vary across quartiles of pattern scores at 6 and 15 months of age [7]. The trends in nutrient intakes across dietary pattern scores presented here further demonstrate that the patterns obtained using questionnaires describe dietary differences not only in food intakes but also in nutrient profile.

Sodium intake tended to increase while many other nutrients tended to decrease across tertile of discretionary pattern scores. These results suggest that higher scores on the discretionary pattern reflect a poorer nutritional profile and a less nutrient dense diet. In contrast, the ready-prepared baby foods pattern was positively associated with key micronutrients such as iron and calcium, yet negatively associated with sodium intake. However scores on the ready-prepared baby foods pattern was also negatively associated with NSP intake at 6–8 months. These results suggest a generally positive nutritional profile with some room for improvement. The favorable nutrient composition of the ready-prepared baby foods pattern is pertinent given than the UK infant feeding guidelines specifically promote the use of home-made rather than ready-prepared baby foods [29].

The associations between the home-made traditional dietary patterns with energy and nutrient intake were more difficult to interpret and showed little consistency between 6–8 and 15–18 months of age. At 6–8 months, higher scores on the home-made traditional pattern were positively associated with energy, protein, NSP and sodium intake, and negatively associated with calcium and iron. However at 15–18 months of age, the association between the home-made traditional pattern and protein intake was consistent with the data at 6–8 months, but there was no association with carbohydrate, calcium, sodium and iron intake. These results suggest the traditional patterns at different ages reflect a mixed nutrient profile, making it difficult to hypothesize how the traditionalpatterns might influence health and development. There was limited variation in fat intake across the dietary patterns at either time point, which may limit their use in predicting outcomes such as blood lipids.

At 6–8 months of age, the breastfeeding pattern was negatively associated with calcium and iron intake. It is important to note that breast milk intake was not assessed directly, instead mothers reported the duration of each breastfeed and an algorithm was applied to the rate of transfer of milk to the infant [30]. Furthermore, the nutrient content of breast milk does not allow for the bioavailability of the nutrient, which is often enhanced in breast milk. Therefore, the associations between scores on the breastfeeding pattern and nutrients may be less reliable than for other patterns. Nevertheless, median nutrient intakes exceeded the UK reference values [31] suggesting that the predominantly breastfeeding pattern reflects an age-appropriate nutritional profile. Finally, the home-made contemporary pattern at 15–18 months of age was negatively associated with intakes of saturated fat, and positively with NSP, potassium and iron. With respect to the macronutrient intakes, higher scores on the home-made contemporary pattern tended to reflect a healthier diet.

The present study is one of the few that have examined the association between dietary patterns and nutrient intakes in a pediatric sample [32]. Consistent with previous literature [11,12], we found that patterns characterized by higher intakes of healthier foods had more positive nutrient profiles than patterns characterized by higher intake of discretionary (nutrient-poor) foods. Even though the present study was conducted in a subset of a large cohort, the number of children involved is substantial (n = 535–725) and to the best of our knowledge, this is the only study to have examined associations between nutrient intakes obtained from diet diaries and dietary patterns in the first year of life.

The present findings should be interpreted within the context of the study strengths and limitations. The ALSPAC cohort is one of the most well-characterized birth cohorts in the world and is considered broadly representative of the population living in the local area during the early 1990’s when the study commenced. Although the dietary data was collected prospectively, the questionnaires were not validated in the study population. While the questionnaire was adapted to the age of the infant, all studies that involve dietary questionnaires have the potential for recall bias. The fact that the dietary patterns were not extracted at the same age as the nutrient data may increase the measurement error. We have assumed that the patterns at 6 and 15 months of age would be similar to those at 8 and 18 months, respectively. This is because new foods are usually introduced to an established repertoire of foods and therefore we may expect to see minor or gradual changes over time, rather than a major change to a different style of diet. Furthermore, the effect of measurement error is likely to result in an underestimation or null association between dietary patterns and nutrient intakes rather than an over-estimation. This may have also contributed to some of the inconsistencies in the associations across each age. By contrast, a strength of this analysis is that associations are observed using two independent measures of diet (i.e., dietary patterns derived from questionnaires and nutrient intakes from diet diaries), because the results may be less influenced by the disadvantages or reliance on one particular method of dietary assessment.

An advantage of using PCA to summarize the whole diet is that the dietary patterns reflect what individuals actually eat. However as dietary patterns are based on actual consumption patterns rather than an established diet-disease relationship, patterns may or may not be associated with health outcomes [33]. The present article highlights the complexity of the associations between dietary patterns and nutrient intakes, as pattern scores may be associated with nutrients that are frequently linked to higher risk of disease (e.g., sodium, saturated fat) and also to protective factors (e.g., iron, NSP). Our analysis demonstrates that an understanding of the nutrient intakes underpinning a dietary pattern can provide important and in insightful information that may be helpful for interpreting the association between dietary patterns and health.

Early life dietary patterns vary in their underlying energy and nutrient composition. By gaining insight into the energy and nutrient intakes underpinning each pattern, our overall understanding of dietary patterns is improved. While pattern food loadings suggest that the discretionary, ready-prepared baby foods, and home-made traditional patterns are similar at 6 and 15 months of age, the nutrient analysis show that these patterns exhibit some differences in their associations with nutrient intakes.