A minimal selection bias might not be totally excluded in the
group of uncomplicated diabetes type1, since some of them
have been treated only by their family doctor and different
centres might use heterogeneous documentations.
The macrovascular events are evaluated retrospectively
and not prospectively, which might also cause differences.
HbA1c levels were not measured centrally. In order to
reduce inter laboratory variations of HbA1c values, they were
standardised according to DCCT, using the multiple of the
mean method.