Norwegian children consume less fruit and vegetables (FV) 1 and more added sugar 12 and saturated fat 1, than recommended. A number of intervention studies have demonstrated that it is possible to increase children's FV intake, even though effects have been small and the long-term persistence of such changes are unknown 3. The more effective studies have been comprehensive multi-component interventions targeting several determinants. However, very few studies have been able to evaluate the effect of the separate components, and little is known about what mediate the effect observed. Studies of single intervention components, targeting specific strong determinants of intake, are largely missing. Recently, an extensive review of determinants of children's and adolescents' fruit and vegetable intake identified availability and accessibility among the strongest determinants 4, a finding also supported by Norwegian studies 56. Thus, an intervention increasing the accessibility of fruit and vegetable should in theory be effective, as is also predicted based on ecological models emphasising the important influence on health-related behaviours exercised by the environment 7.

Most elementary school children in Norway bring their own lunch (usually sandwiches) to school. Few elementary schools have canteens, and traditionally FV have not been available at school. To increase the pupils' intake of FV, a School Fruit Programme is now offered to all Norwegian elementary and junior high schools 8. Pupils who subscribe receive a piece of fruit or a carrot each school day, usually at lunch time. The cost for the pupils/parents is NOK 2.50 per school day (approximately EUR 0.30). The programme is subsidised by the Norwegian Government by NOK 1.00 per pupil per school day.

A major problem with the current programme is the low participation rate by schools and by pupils at participating schools. Only 12% of the total Norwegian elementary school population (grades 1–10) did subscribe in the spring semester 2006, and the effect on overall FV intake is therefore limited. It is also noticeable that subscribing pupils tend to report healthier behaviours to start with than do non-subscribing pupils 89. Thus, the programme might contribute to increase the inequality seen in FV intake in this age group 9.

This paid programme increases the availability of fruits and vegetables at participating schools. However, it does not increase the accessibility for the individual pupil unless the parents subscribe to the programme on behalf of their children. If the programme was offered for free, it would have increased the accessibility of fruits and vegetables for all children. Recent evaluations of this programme when it was offered at no-cost for the pupils, have shown good effects in increasing all children's FV intake, even among those that usually eat little FV 8. A sustained elevated FV intake was furthermore found one year after the programme was provided for free 10. This effect could in part be explained by higher subscription rates in the standard paid Norwegian School Fruit Programme the subsequent year 10. The no-cost subscription also resulted in decreased consumption of unhealthy snacks, i.e. snacks high in added sugar and/or saturated fat, for pupils of parents without higher education 8.

A recent cost-effectiveness analysis concluded that a School Fruit Programme provided for free to all 10 years of elementary and junior high school would be cost-effective if it resulted in a sustained (lifelong) increase in mean FV intake of 2.5 grams (conservative estimate) per day 11.

The aim of the present paper is to evaluate the effect of the Norwegian School Fruit Programme, provided at no-cost to the pupils, three years after it was provided for free, for both fruit and vegetable intake and consumption of unhealthy snacks.

The third order interaction term between sex, time and intervention was not significant in any of the models for the four response variables (0.13 ≤ p ≤ 0.90). Thus, the intervention effect was similar for boys and girls (on the log_e-scale, back transformed estimates differ slightly). All analyses included the interaction between time and intervention which represents the intervention effect defined above. This interaction term was highly significant for FV at school, FV all day and usual FV intake (p < 0.001), but not for soda/candy/chips (p = 0.18). In addition, we found a significant interaction between time and sex in the models for FV at school and soda/candy/chips (p-values 0.03 and 0.002, respectively), and a significant interaction between sex and intervention in the models for FV all day and usual FV intake (p-values 0.04 and 0.008, respectively).

The pupils in the Free fruit group significantly increased their FV intake from baseline to May 2002, compared to the control group (Table 1). This short-term effect has previously been described in Bere et al. (2005). Sustained significant effects on FV intake three years after the end of the intervention were also observed (all p-values < 0.001). For boys the estimated change in FV intake from baseline to May 05, compared to the control group, were 0.13 portions/day for FV at school, 0.38 portion/day for FV all day and 1.6 times/week for usual FV intake. Corresponding estimates for girls were 0.15 portions/day, 0.44 portion/day and 1.6 times/week, respectively. No significant intervention effect was observed for soda/candy/chips.

In May 2002, 100% and 13% of the intervention and control pupils, respectively, subscribed to the School Fruit programme (Table 2). The corresponding numbers were 16% and 1% in May 2005. When subscription to the School Fruit Programme was included in the analyses the effect of the intervention decreased some, and became less significant for all measures, but the effect was still significant for all FV measures (Table 1). We found a significant interaction between individual subscription and intervention for usual FV intake (p = 0.04), but not for FV all day (p = 0.20) and FV at school (p = 0.92).

The present results show that free school fruit had a positive effect on Norwegian school children's FV intake, also three years after the fruit was provided for free. Such long term effect of a FV intervention for children has, to our knowledge, not previously been reported. Previously we have reported that a free school fruit programme, in addition to increasing pupil's FV intake, also reduce the consumption of unhealthy snacks among pupil's of parents without higher education 8. This effect is not sustained three years after the end of the intervention.

The long term effects were estimated to be 0.38 and 0.44 portions/day for FV all day (boys and girls, respectively). A portion in the present study is about 80 grams 12. The effect of the free fruit intervention three years after the fruit was provided for free can therefore be estimated to be about 30–35 grams/day. This is a considerable larger increase than what is concluded to be needed for the Norwegian School Fruit Programme, offered for free for all 10 years, to be cost effective (2.5 grams as a conservative estimate) – if the effect is sustained throughout life 11. The observed significant long-term effects in May 2005 decreased some when adjusting for individual subscriptions to the school fruit programme, but the effects were still significant. The subscription rates in May 2005 were clearly higher in the Free fruit group than in the control group, and this then accounted for some of the difference between the groups. Available FV (e.g. such as a paid subscription programme) therefore promotes a sustained long term effect. Longer term follow-up surveys are, however, needed to state whether lifelong FV habits have been established in order to state the cost-effectiveness of the programme.

The long-term effect seen in the present paper contradicts with findings reported from the British National School Fruit Scheme where short-term but no long-term effect was seen 14. That study had some clear limitations as it was cross-sectional (no baseline data), had a low response rate (51%) and schools were not randomly assigned to condition 14. In addition, no paid programme (or something equivalent) was offered to the pupils after the intervention. Thus, the setting was different, and it is therefore difficult to compare the results with those from the present study. Different versions of free school fruit initiatives have also been tried out, or are planned, in several other countries; e.g. in Belgium, Canada, Denmark, Ireland, the Netherlands, New Zealand, Scotland and USA (Robert Pederson, the Danish Cancer Society, personal communication), but effect evaluations have not yet been published in scientific journals.

There are some limitations of the present study. The schools were randomly drawn from two different counties, but the schools receiving free fruit were for practical reasons randomly drawn from one of the two counties only. Some of the pupils (6th graders at nine schools in both counties, but no 7th graders) did also receive a FV educational programme during the intervention year (2001/2002). This educational programme showed, however, no effect in increasing FV intake 1015, and all pupils in the FVMM cohort are therefore included in the present study to increase the statistical power. We also included the variable for the educational programme as a covariate in the model. However, the educational programme was highly correlated with class level and we found that using only class level as covariate gave a better fit (using the Akaike information criterion). We also performed analyses restricted to the 7th grade pupils, and the same conclusions apply although with this reduced sample we see a weaker significance for the FV variables (p-values were 0.02, 0.02, 0.04 for FV at school, FV all day and usual FV intake, respectively).

The results show long-term effects of a free school fruit programme. The effects are in part mediated through higher participation in the national (paid) school fruit programme the years following the intervention.

The author(s) declare that they have no competing interests.

KIK conceived the study and is together with EB responsible for the design of the study. MBV, ØS and EB worked with the statistical models. ØS ran the final analyses. All authors contributed to the interpretation of the data. EB drafted the article, and MBV, ØS and KIK revised it critically. All authors have approved the final version of the manuscript.