Training to estimate blood glucose and to form associations with
initial hunger
Mario Ciampolini  and Riccardo Bianchi

Nutrition & Metabolism 2006, 3:42     doi:10.1186/1743-7075-3-42
http://www.nutritionandmetabolism.com/content/3/1/42
Published   8 December 2006
Abstract (provisional)

The complete article is available as a provisional PDF. The fully
formatted PDF and HTML versions are in production.

Background: The will to eat is a decision associated with conditioned
responses and with unconditioned body sensations that reflect changes
in metabolic biomarkers. Here, we investigate whether this decision can
be delayed until blood glucose is allowed to fall to low levels, when
presumably feeding behavior is mostly unconditioned. Following such an
eating pattern might avoid some of the metabolic risk factors that are
associated with high glycemia.

Results: In this 7-week study, patients were trained to estimate their
blood glucose at meal times by associating feelings of hunger with
glycemic levels determined by standard blood glucose monitors and to
eat only when glycemia was < 85 mg/dL. At the end of the 7-week
training period, estimated and measured glycemic values were found to
be linearly correlated in the trained group (r = 0.82; p = 0.0001) but
not in the control (untrained) group (r = 0.10; p = 0.40). Fewer
subjects in the trained group were hungry than those in the control
group (p = 0.001). The 18 hungry subjects of the trained group had
significantly lower glucose levels (80.1 +/- 6.3 mg/dL) than the 42
hungry control subjects (89.2 +/- 10.2 mg/dL; p = 0.01). Moreover, the
trained hungry subjects estimated their glycemia (78.1 +/- 6.7 mg/dL;
estimation error: 3.2 +/- 2.4% of the measured glycemia) more
accurately than the control hungry subjects (75.9 +/- 9.8 mg/dL;
estimation error: 16.7 +/- 11.0%; p = 0.0001). Also the estimation
error of the entire trained group (4.7 +/- 3.6%) was significantly
lower than that of the control group (17.1 +/- 11.5%; p = 0.0001). A
value of glycemia at initial feelings of hunger was provisionally
identified as 87 mg/dL. Below this level, estimation showed lower error
in both trained (p = 0.04) and control subjects (p = 0.001).

Conclusion: Subjects could be trained to accurately estimate their
blood glucose and to recognize their sensations of initial hunger at
low glucose concentrations. These results suggest that it is possible
to make a behavioral distinction between unconditioned and conditioned
hunger, and to achieve a cognitive will to eat by training.