Readers hoping for clear protocols about glucose control in nonintensive care hospitalized patients with diabetes mellitus may note conflicting recommendations in this issue of American Family Physician. Nau and colleagues recommend stopping home oral diabetes medications, switching to basal (long-acting) insulin protocols, and aiming for tighter glucose control,¹ whereas Sawin and Shaughnessy recommend continuing home medications and avoiding tight glucose control regimens.² Which recommendations are correct?

These two articles do not conflict as much as it seems about whether to continue oral diabetes medications. Both articles emphasize the inferiority of sliding-scale insulin to support their recommendations. We see this all of the time: treatment A is better than placebo; treatment B is better than placebo; but which treatment is better, A or B? One of the reasons experts disagree is because there are a lack of head-to-head clinical trials directly comparing treatments. In hospitalized patients with diabetes, basal insulin regimens are superior to sliding-scale insulin, and continuing patients' home regimens on admission is at least equivalent to sliding-scale insulin. Randomized controlled trials comparing oral home regimens with basal regimens would clarify which option is truly better. In the study supporting home regimens, one half of the patients' home medications (which were continued in the hospital) included long-acting basal insulin.³ Therefore, this study also could be used to advocate basal insulin over sliding-scale insulin.

Both articles address specific risks of continuing oral medications in the hospital.^1,2 Although they differ on what to do, they both advocate clinical judgment to make this decision.

The second apparent conflict concerns tight glycemic control. Sawin and Shaughnessy recommend against routine intravenous insulin drips to achieve tight glycemic control (i.e., 81 to 108 mg per dL [4.50 to 5.99 mmol per L]) in patients in intensive care, specifically in the management of myocardial infarction.² To be clear, the subcutaneous basal insulin protocols for patients who are not in intensive care presented by Nau and colleagues follow American Diabetes Association's guideline targets of 140 to 180 mg per dL (7.77 to 9.99 mmol per L).¹ Thus, there is no overt conflict. However, Nau and colleagues recommend that hospital teams establish their own target glucose range¹; therefore, one should be familiar with local protocols and avoid overly tight parameters.

Inpatient costs aside, it should be noted that basal insulin analogues (glargine [Lantus], detemir [Levemir]) in the outpatient setting are much more expensive (up to $643,000 per quality-adjusted life-year), with little to no proven benefit over isophane insulin (NPH).⁴ Be wary of protocols favoring analogues, especially if patients are discharged on these medications. Both articles support good communication with patients' primary care physicians upon discharge.^1,2

Overall, these two articles reflect different perspectives about evidence. A primary care versus specialty care undertone may arise when the evidence is unclear. Poor in-hospital glycemic control (hyperglycemic and hypoglycemic) is associated with increased mortality, poor patient outcomes, and longer hospital stays. However, association does not prove cause. Poor glycemic control may be a marker of disease severity. Associations of hypoglycemia with bad outcomes and lack of patient-oriented outcomes for tight glycemic control argue for a less aggressive approach that some primary care physicians may find appealing. Strong, independent associations of hyperglycemia with bad outcomes and disease-oriented evidence (e.g., oxidative stress, endothelial dysfunction, markers of inflammation) support more aggressive treatments that other physicians may prefer. Until randomized controlled trials assessing patient-oriented outcomes provide clarity, experts may legitimately disagree about which associations are markers or causes.

Apart from straddling these fault lines in the evidence, each article supports similar management of nonintensive care inpatients with diabetes—use clinical judgment about continuing oral medications; when stopping oral medications, use basal insulin protocols (avoiding sliding-scale insulin and intravenous insulin drips); and select glucose targets no lower than 140 to 180 mg per dL.