Interestingly, there is data to suggest that the thiazide-induced hyperglycemia effect is a potassium effect: a 2006 review of >50 trials in which thiazides were compared to other blood pressure-lowering medications or placebo, it was found that for every 1 mEq/L decrease in K, there is a 10 mg/dL increase in glucose. Furthermore, normalization of serum potassium in patients on thiazides (with potassium supplements) will lower the serum glucose.
From a molecular level, this is perhaps explained by the fact that the pancreatic b-cell secretion of insulin is regulated in large part by ATP-sensitive K channels in the beta-cell membrane. Alterations in the extracellular potassium concentration to which these cells are exposed could potentially lead to decreased insulin secretion.
If this mechanism is correct, it would imply that thiazide-induced hyperglycemia occurs by a different mechanism (decreased insulin secretion) than the standard type 2 diabetes (peripheral insulin resistance), and furthermore that thiazide-induced hyperglycemia is reversible with normalization of potassium levels. Thus, it is probably not necessary to remove thiazides from the list of useful first-line agents for treatment of hypertension. That being said, as a budding nephrologist I am always looking for a good reason to prescribe an ACE-inhibitor or ARB, and therefore I personally tend to add one of the RAAS blockers first given the population of clinic patients that I see.