Hi Sid,
Thanks for your thoughts. Unfortunately (for us insulin users!) I don't think that is the most likely interpretation of the study results, for the following reasons: 1). people in all comparison groups were 'incident cases' (that is, the authors used a criterion that to be included people had to be recently diagnosed and diagnosed for the first time), so in theory at least they won't have been people who had failed to comply with any therapy for years and gone through an escalation of medication, for whatever reason insulin for the insulin group was chosen as pretty much a first line therapy (we could query why). 2). Anyone with prior health issues of relevance to the study was excluded (so any health problem evaluated developed subsequent to the therapy of choice) 3) an extensive list of baseline characteristics was controlled for in multivariate analyses (meaning that these differences therefore did not impact on outcome (characteristics were: age, gender, HbA1C, blood pressure, smoking history, cholesterol, weight, duration of diabetes, LVD, creatinine levels, experience of antihypertensives, antiplatelets, aspirin, Charlston co-mobidity status and GP contacts in prior year. The outcomes therefore are given with all of these risk factors 'equal'. So, sadly, the clear implication made by the authors is that it is insulin as such that causes the reported higher rates of adverse events.
We could critiscise the study for being retrospective, but given the very large sample sizes, I am not sure we would see much difference with a prospective study.
I don't want to give people (including myself!) sleepless nights, but I think we need to at least assume for the moment that maybe insulin is dangerous and all try to cut down on it whilst keeping our BGs as good as possible - it is a balancing act for those who 'have' to have it, maybe 'better insulins can be developed if we press for that.