To Correct or Not to Correct: Lost in Inpatient Translation

The era of intensive glycemic management in the hospital began following the landmark Leuven trial in the surgical intensive care unit (1), followed by other studies demonstrating mixed benefits in critical and noncritical care settings and in specialized populations (2–5). While questions remain about how aggressive glycemic targets can safely be, the delicate balance between intensive glycemic management and subsequent hypoglycemia and other deleterious outcomes remains elusive to many health systems (6). While the Normoglycemia in Intensive Care Evaluation–Survival Using Glucose Algorithm Regulation (NICE-SUGAR) trial questioned the benefit gained from “intensive” treatment, implementation differences across hospitals and health systems including system-level barriers likely affected rates of hyperglycemia, hypoglycemia, and outcome attainment (3,7). Understanding this balance in the real world is still lacking. Insulin safety and preventing hypoglycemia are still major concerns in the inpatient setting, as insulin remains a high alert medication that poses a greater risk of causing significant harm if there is an error in type, amount, or timing of the dose (8). Use of simplified regimens and glycemic targets to maintain euglycemia with the potential of reducing workflow burden and potential for errors should be prioritized.

To this end, Vellanki et al. (9) completed a noninferiority study evaluating the use of intensive (correction for blood glucose >140 mg/dL), and nonintensive (correction for blood glucose >260 mg/dL) supplemental sliding scale insulin (SSI) before meals and bedtime in patients with type 2 diabetes (T2D). The study was a randomized controlled trial (RCT) of a hard-to-treat group of patients: those with uncontrolled T2D (mean ± SD HbA_1c 9.1 ± 2.3), with obesity (BMI 34.4 ± 9.8), with long T2D duration (12 ± 9 years), and on high home insulin doses (61% on insulin, 70 ± 50 units/day). Study investigators were successful in achieving a similar mean daily glucose in both groups (intensive N = 108, 172 ± 38 mg/dL, vs. nonintensive, N = 107, 173 ± 43 mg/dL) with significance for noninferiority (P = 0.001). Of note, while significantly fewer subjects received correction insulin between the groups (34% vs. 91% for intensive and nonintensive, respectively; P < 0.0001), there were no differences in correction insulin doses between groups who received SSI. These results, presented in this issue of Diabetes Care, support a potential paradigm-shifting approach to starting and continuing SSI at less aggressive targets that can have widespread implications for diabetes management.

Noninferiority trials are generally used when there is an accepted standard where placebo control may be unethical and/or when the new treatment is unlikely to outperform known standards of care. In this case, the premise of less supplemental insulin at a higher threshold, in a group with diabetes so difficult to control, is radical. The design and premise of the scientific question are to be commended, since simplification of regimens without forsaking glycemic outcomes can have far-reaching implications. Vellanki et al. should also be applauded for applying rigorous research methods to question the standard of one of the most commonplace tools in the armamentarium of intensive management: the supplemental or correction scale. The research team was highly successful in recruitment, through informed consent, of a largely underrepresented group in clinical trials (N = 81% Black) while safely attaining the outcomes of interest, demonstrating noninferiority.

The study’s strengths also highlight its limitations, especially in the hospital setting. This is a single-system gold standard setting, where individual patients were randomized and seasoned study investigators ordered insulin and maintained treatment fidelity. In addition, in evaluating the hospital protocol used, it was not just the supplemental scale that was driving glycemic outcomes but also likely the underlying basal-bolus strategy used (10). While glucose targets were used in initiation of treatment (glucose >140 vs. >260 mg/dL), a highly prescribed set of protocols dictated both supplemental (based on sensitive/usual/resistant scales) and basal-bolus protocols (daily insulin adjustments ± 10% or 20%). There were no significant differences in mean daily glucose or prandial, daily, or supplemental insulin. The average doses of supplemental insulin given were nominal (7 ± 4, 8 ± 4 units/day) in comparison with the total daily insulin needed (total estimated mean total daily dose 44 ± 37 units/day).

The assumption of the premise, that this simplified regimen with a high target threshold could improve adherence to and safety of the proposed protocols, needs to be examined in a real-world context. From a safety perspective, not only targets and dosing are of importance but also the systems in place for implementation and continued use. Errors can range from incorrect prescriber orders to pharmacy dispensing errors to nurse administration mishaps, in addition to dosing errors. The era of computerized prescriber order entry using structured insulin order sets and “right” patient-insulin-dose-time barcoding has revolutionized the ability to prevent insulin errors (11,12). Optimal use of electronic health records can potentially provide just-in-time dosing guidance, autocalculations of insulin doses, and best practice advisories in an effort to prevent such errors. The Institute for Safe Medication Practices along with the American Diabetes Association Standards of Medical Care in Diabetes recommend validated structured insulin order sets to prevent insulin errors and therapeutic inertia (8,13). Structured insulin orders can ease calculation burden and reduce therapeutic inertia, especially when the majority of items such as controlled carbohydrate meal plans, point-of-care glucose monitoring, HbA_1c, hypoglycemia treatment, and prandial, correction, and basal insulin are preselected. It is noted that many hospitals report that they do not have all the necessary tools (measurement, order sets) at their disposal (14). In this study, it is not clear what system-level interventions were present, albeit the systems in place may be equally as important as the threshold and the target.

Vellanki et al. have clearly ushered in a new era of correctional scale using a two-tier threshold of above and below 260 mg/dL. But some questions remain: Will glycemic targets be reached outside of this hospital system? How much “treatment burden” is there and in a real-world setting, will this truly reduce staff burden and maintain optimal glycemic outcomes at all hospitals? Rigorously evaluating these questions through health services research methods and implementation science within institutions with varying resources and in diverse settings will be necessary.

Learning from our past, we should be cautious in adopting a “less” intensive supplemental correction scale strategy, as was done with intensive glycemic targets post-Leuven in 2001. While RCTs are considered the gold standard for clinical guidelines (15), in the hospital setting, where treatment fidelity may rely on a multimodal set of factors (institutional and EHR technology support, nurse staffing, and workflows), we may need to expand our methodology (clustered randomization, interrupted time series designs) to understand truly optimal care delivery for our sickest patients (16). And while a change to less aggressive SSI may not exacerbate inequities at the individual level, system inequities at the institutional or health systems level may occur if appropriate basal-bolus and supplemental protocols and support for their implementation are not in place.

Questions will always remain regarding how institutions can effectively translate RCT results into delivery of high-quality and equitable care. Here it will hinge on investment, along with multipronged system-level approaches and innovative redesigns, with the prerequisite of EHR optimization, interdisciplinary team management, and supporting policies. Perhaps with the advent of this study, in the height of our current day technology and systems alignment, the holy grail of simplified regimens with optimal glycemic outcomes for all patients in all hospitals may not be so far from reach.