Reader Comments

Post a new comment on this article

Diagnosis Time Lag.

Posted by RaviR on 28 Feb 2013 at 20:43 GMT

If I understand the approach correctly, the data are for the 11 yr duration. (I came to this site via NYT.)

Given that the symptoms, and thus thus the diagnosis, of diabetes can take several years to show up, is there a data-set (or methodology) that can run the model/analysis with a time lag. This might even provide a stronger case for causation, if the current conclusions are true.

No competing interests declared.

RE: Diagnosis Time Lag.

sanjaybasu replied to RaviR on 28 Feb 2013 at 22:27 GMT

Some time lag were incorporated as the length of the available data allowed, as described in the SI text.

No competing interests declared.