Things We Do For No Reason: Sliding-Scale Insulin as Monotherapy for Glycemic Control in Hospitalized Patients

SSI administration does not attempt to replicate normal pancreatic physiology, which involves basal insulin secretion to impair hepatic gluconeogenesis and meal-associated insulin spikes to promote uptake into glucose-avid tissues. SSI is a reactive strategy, not a proactive one, and perhaps unsurprisingly, to our knowledge, it has never been shown to prevent hyperglycemia in hospitalized patients, an impression corroborated by a systematic review of the topic between 1964 and 2003.⁷ More recently, one multicenter trial analyzed the effect of adding SSI to oral antihyperglycemic medications in hospitalized diabetics and found no differences in rates of hyperglycemia.⁸ Another study found that 84% of administered SSI doses failed to correct hyperglycemia.⁹

However, does adding basal insulin to SSI raise a patient’s risk of hypoglycemia? When basal insulin is dosed carefully, the answer appears to be no. In a trial in which diabetic long-term care residents who were receiving SSI at baseline were randomized to either continued SSI or basal-bolus insulin, the investigators found that the basal-bolus group experienced significantly lower average blood glucose levels without an increase in adverse glycemic events.¹⁰ Perhaps the most significant milestone to date, however, was the RABBIT 2 multicenter trial, published in 2007, that randomized hospitalized, insulin-naïve diabetics to either a weight-based regimen of basal and prandial insulin or SSI only.¹¹ Rates of hypoglycemia and length of stay did not differ between the groups, and 66% of patients receiving basal-prandial insulin achieved their glycemic control target as opposed to just 38% of patients in the SSI-only group. The SSI group also required more total insulin. A weight-based, basal-bolus strategy was later proven to be similarly effective, without causing severe hypoglycemia, for patients undergoing surgery who could not maintain consistent oral alimentation.¹² Basal-bolus insulin was associated with fewer surgical complications, and it produced a cost savings of $751 per day as determined by a post hoc comparative effectiveness study.¹³

Prolonged use of SSI as monotherapy may be not only ineffective but also harmful. Clearly, the absence of basal insulin will harm type 1 diabetics, who need basal insulin to prevent diabetic ketoacidosis. However, even for type 2 diabetics and nondiabetics, hyperglycemia has been established as a marker for adverse outcomes among hospitalized patients,¹⁴ and SSI monotherapy has been associated with a three-fold higher risk of hyperglycemia compared with the use of a sliding scale plus other forms of insulin.¹⁵ At least one other study has also linked this practice with a significantly increased length of stay compared with patients who were receiving insulin proactively.¹⁶ We believe that the potential for harm is difficult to disregard, especially because safer alternatives are available. Ultimately, it can be stated that in hospitalized patients with persistent hyperglycemia who require insulin, SSI alone should not be the preferred treatment choice regardless of whether the patient carries a known diagnosis of diabetes mellitus or has used insulin previously.

As discussed above, there is no known clinical scenario in which SSI as monotherapy has been proven to be effective; however, the use of SSI as monotherapy as a short-term approach has not been well studied. Hospitalized patients who are at risk for adverse glycemic events should be monitored with periodic finger-stick blood glucose draws per guidelines, and in the first 24 hours, it may be reasonable to withhold basal insulin for insulin-naive patients, particularly if the medication reconciliation or other key components of the history are in doubt, or if there are risk factors for hypoglycemia such as a history of bariatric surgery. The amount of insulin received in the first 24 hours of such monitoring may inform subsequent insulin dosing, but this method of “dose finding” has not been validated in the literature.