| Abstrakt: |
To the Editor.—The University Group Diabetes Program (UGDP) study has been a constant source of controversy since the original findings were published.1 While any set of data is always subject to more than one interpretation, we believe the methods of analysis used by Kilo et al (241:26, 1979) to dispute the most recent UGDP conclusions (240:37, 1978) are highly inappropriate, because the authors produced an extremely biased subsample of the original data. First, they excluded patients who died during the first year, changed medication (MC), or were considered dropouts (42% of the entire UGDP data base). Second, they found "a disproportionate number (14 of 17=82%) of insulin variable subjects, compared with placebo subjects (five of 17=29%)... at much higher risk of dying of cardiovascular causes at the start of the study,"2 where they defined "much higher risk" after scrutinizing the cardiovascular death data. Therefore, they further excluded all |
