The main finding of this study was that the sweet drink given to children before two years of age was associated with adiposity at 24 years of age (Table 1). To our knowledge, this is the longest period over which such an association has been reported. There was, however, evidence that the consequences of the choice of drink relied on more than the drink itself.

In males, early exposure to cola, fruit squash (Table 2), or fizzy drinks (Supplementary Table S3), were at three years associated with a diet containing more energy, carbohydrates, fat, protein, and NME sugars, but less NSP. In contrast, early exposure to apple juice was not associated with energy intake at age three years, but there was a higher consumption of protein and NSP, and a lower consumption of fat and NME sugars. The pattern in females was similar. The choice of drink was only a part of a wider dietary pattern. Thus sugar in drinks should not be considered in isolation, as they may be associated with a healthy or less healthy diet (Tables 2 and 3). Another consideration is that FJ contains significant levels of vitamin C, folate, potassium, carotenoids, and flavonoids which are health promoting [19, 20].

The finding that dietary patterns are important was consistent with Northstone and Emmett [13], who found when diets at three, four, seven and nine years of age were compared, dietary patterns persisted over time. A ‘processed’ diet included the eating of chocolate, sweets and crisps and high fat, high sugar processed foods, but also the drinking of fizzy drinks. In contrast, fruit, vegetables, salads, pasta. and rice were part of a ‘health conscious’ diet that included fruit juice. Similarly in the present study, cola, other fizzy drinks, and fruit squash were associated with consuming burgers/sausages. pizza, fried foods, French fries, meat, chocolate, and sweets (Table 3, Supplementary Table S5). The pattern with apple juice was broadly opposite to SSB; it was less likely that burgers/sausages, fried food, and French fries were eaten (Table 3), although more fish, vegetables, and fruit were consumed.

These findings question an approach that concentrates on sugar per se, rather than a diet that happens to contain sources of sugar. Cola and apple juice both provide about 9–10 g of sugar per 100 ml, but in males cola was associated with greater adult adiposity, whereas apple juice was not (Table 1). In females, cola was also associated with higher adiposity, whereas apple juice was associated with lower adiposity (Table 1).

Such observations should not be a surprise as SSB consumption has been associated with greater adiposity when FJ consumption was not [5]. When FJ has been associated with higher body weight, 100% FJ has not been distinguished from exogenously sweetened juice [6]. The importance of this observation is illustrated by the present study, when fruit squash was found to be similar to SSB rather than FJ, in terms of macronutrient intake at three years (Supplementary Tables S4 and S6); dietary patterns at three years (Supplementary Table S5); mother’s weight and educational level; mother’s age at the child’s birth; partners BMI (Supplementary Table S7) and in females the BMI and total fat mass (Supplementary Table S2).

A reason to study early nutrition is that body fat increases after birth, and then declines only to subsequently increase from three to seven years. An earlier adiposity rebound has been associated with a greater risk of adult obesity [21, 22], such that if it occurs about three years of age rather than later, there is a greater risk of adult obesity [23], a phenomenon that occurs more often in girls [24, 25]. The question arises as whether the present findings were influenced by the age the drinks were given, as metabolic programming at critical stages in development predisposes to obesity throughout life [26]. If so, an insight into the influence of early feeding may prevent later weight gain, but attention needs to be directed to feeding in the first years of life, rather than the later ages that have often been studied in nutritional interventions.

However, the influence of diet needs to be placed in context, as although with a large sample size statistical significance was obtained, the regression equations accounted for a small amount of variance. In contrast, demographic variables had a far greater influence than the choice of drink given to the child (Table 5). Similarly, it has been reported that a child’s pattern of dietary intake is strongly influenced by socioeconomic characteristics. Factors such as a younger maternal age, and the early introduction of complementary feeding, are associated with an unhealthy dietary style [27]. Although nutritionists naturally emphasize diet, it should be assumed that a correlation between diet and adiposity reflects more than the calories provided, as the effects of demographic variables associated with diet (Table 5) had a greater influence than diet per se. A consideration of demographic variables will indicate to whom nutritional attention should be directed, and when.

A positive is that the findings add to the evidence that diet during early childhood is associated with risk of obesity in later life. As such, the attention of nutritionists should be directed to the early diet of young children, and the impact of a nutritional intervention at a younger rather than later age. There needs to be a greater emphasis on controlling energy intake in infancy and young childhood.