The focus of the 4 papers presented (one published, 3 manuscripts) is
assessment of the effect of genetic variants on mortality and
complications in diabetes patients, and has therefore required
accessing genotypical as well as phenotypical information from several
studies. The genotypical information acquisition seems well described
(however I have very limited knowledge in this field), whereas the
phenotypical (clinical) information is less consistently reported
between studies.

In general the epidemiological analyses appear weak and not well
explained in the papers, in particular the second paper (``Shared
genetic'') has flaws in the explanation of methods as well as
inconsistent event counts (12% extra deaths appear). 

Moreover the fourth paper has severe methodological omissions that
possibly invalidates the conclusions even though this question could
have been resolved quite easily (with unknown result, though). Also it
seems odd to make the last two papers (on the DOLCHE study) separate,
the famine analysis could easily have been included in the first paper
- it is basically a single extra variable (although pointing at
different loci).

In conclusion, I find the amount of work presented adequate, but I
find it difficult to recommend to the University of Lund to pass a PhD
with two major flawed manuscripts out of 4.