Our pragmatic trial demonstrated that, in adults with newly diagnosed T2DM, MNT with an advice to restrict fruit intake resulted in a decreased fruit intake while MNT with an advice to eat more fruit resulted in an increased fruit intake. However this difference in fruit intake did not significantly affect glycemic control, body weight or waist circumference.
To our knowledge, this is the first randomized intervention study examining the effects of dietary advice to restrict fruit intake on glycemic control in T2DM. Most intervention studies with fruit have investigated fruit as a part of a whole diet, fruit mixed with vegetables or only one type of fruit and often as a single meal study e.g. glycemic index studies. Very few intervention studies have tested a variety of fruit over several weeks and none have investigated long-term glycemic control in T2DM subjects.
In our study we found that restriction of fruit intake does not significantly affect HbA1c. In a study by Rodriguez et al. 15 obese women randomized to either a low-fruit or a high-fruit diet for 8 weeks a difference between the groups of 550 kJ from fructose was obtained . This is around twice the difference obtained in the present study. The study reported no differences in HOMA, glucose or insulin levels between the groups. In another study by Madero et al. 131 obese subjects were randomized to a low-fructose diet or a moderate-natural- fructose diet for 6 weeks . The groups’ intake of fruit corresponded to approximately 250 and 2200 kJ, respectively in low-fructose and moderate-natural-fructose which corresponds to a difference between the groups about three times as large as in the present study. Significant reductions in HOMA and glucose values were seen within the moderate-natural-fructose group, but no difference between the groups was seen. Cross-sectional studies have shown that fruit intake is not associated [12–15] or inversely associated  with HbA1c or other parameters reflecting glycemic control. Further, cohort studies addressing the impact of fruit intake on the incidence of T2DM have shown either no association or an inverse association [22, 23]. The evidence, including our present study, therefore suggests that a high fruit intake does not have a negative impact on glycemic control.
We found a tendency towards reduced body weight and waist circumference in the group that ingested most fruit 0.9 (CI 95%; −0.4 to 2.2) kg and 1.2 (CI 95%; −0.5 to 3.0) cm respectively. This corroborates with a few intervention studies. The study by Rodriguez et al. in which the high-fruit group had a significant reduction in waist circumference compared to low-fruit group (5.5 vs. 2.4 cm; p=0.048) . Weight loss was similar in the two groups (6.1 vs. 6.4 kg; p=0.78). In another intervention study 49 obese women were randomized to add either three apples, three pears or three oat cookies to their usual diet for 10 weeks . The total energy and fiber content of the supplements were matched. The two groups with fruit supplements lost significantly more body weight than the group with oat cookies (−0.9 vs −0.8 vs 0.2 kg). In a third study by Madero et al. the moderate-natural-fructose group reduced body weight more than the low-fructose group (4.1 vs 2.9 kg; p=0.02) . A recent review study concluded that in most studies a higher fruit intake has a beneficial effect on body weight and that no studies have found a negative effect .
In spite of a difference in fruit intake of about two pieces daily between the groups we did not find any effect on HbA1c, body weight or waist circumference. The most likely explanation is that fruit is eaten as a part of a daily diet and therefore when changing the fruit intake it will lead to other changes in the diet. We did not measure total energy intake, but weight and physical activity were similar between the groups and therefore energy intake must have been more or less the same in both groups too. When changing the fruit intake other changes in the diet most likely occur and this would explain that there was no difference in HbA1c, body weight and waist circumference despite the significant difference in fruit intake.
Our study has several strengths. It is the first randomized controlled study investigating the relevant scientific question: does fruit intake matter in relation to glycemic control in T2DM subjects? We chose to do this in a “real life” setting. Thus, almost all subjects were fully compliant and there were no drop-outs.
However, our study also has some weaknesses. First, it can be argued that a greater difference in fruit intake between the high and low fruit groups would have resulted in a significant effect, positive or negative. However, we consider a difference of about two pieces of fruit as clinically relevant and we think it reflects a “real life” situation. We admit that testing whether an even higher fruit intake may impact significantly the glycemic control would be interesting, but this was not the intention in this pragmatic trial. Secondly, we did not control (and had no pre-trial intention to do so) the intake of medication. A difference in baseline use of OADs could bias the results. However, adjustments did not significantly change the results (Table 3). Therefore we do not believe it has biased the results. Thirdly, fruit intake and physical activity were self-reported and therefore could have been subject to under or over-reporting. Measurement of biomarkers of fruit intake, e.g. plasma vitamin C and plasma carotenoids would have strengthened the study.