Glucocorticoid-associated blood glucose response and MS relapse recovery

Standard protocol approvals, registrations, and patient consents.

This retrospective chart review study was approved by the Institutional Review Board at the University of Virginia, and the study was executed in accordance with the Health Insurance Portability and Accountability Act of 1996.

We performed an administrative database search (2008–2015) to identify patients admitted to the inpatient service with a diagnostic code for MS between 2008 and June 2015 (MS International Classification of Diseases–9 [ICD-9] codes: 340, 341, 341.1, 341.2, 341.21, and 341.22). Patient and disease characteristics were abstracted from our electronic medical records (EMRs) using the following eligibility criteria: age 18–55 years (inclusive), confirmed diagnosis of MS, administration of IVMP 1,000 mg/d diluted in 0.9% normal saline (minimum 1 day exposure), serum BG values drawn prior to glucocorticoid treatment from a basic or comprehensive metabolic panel, availability of at least 1 nonfasting BG (NFBG) level (i.e., drawn between 09:00 and 24:00 hours) obtained within 18 hours after recorded glucocorticoid administration, no known diagnosis of diabetes, absence of glucose-regulating medications (e.g., metformin), and availability of demographic data in the EMR (e.g., age and sex).

Data collected from the EMR included age, sex, height, weight, glucocorticoid dose and number of days treated, time of glucocorticoid administration, and if sliding scale insulin was administered during hospitalization. Postglucocorticoid BG was defined as any BG drawn within 18 hours of glucocorticoid administration. (The half-life of MP has an estimated range of 18–24 hours.) Glucocorticoid-associated hyperglycemia is most prominent in the postprandial state, with relatively mild changes in fasting BG; thus, to provide a more sensitive assessment of glucocorticoid-associated BG changes, we classified BG values as either fasting or nonfasting. Since inpatient breakfast is routinely served at 8:00 am, any BG measured between 9:00 am and 24:00 am was considered to be an NFBG. Because of our inability to get reliable information on the frequency and dosing of sliding scale insulin administration, we chose to focus on the highest measured NFBG during glucocorticoid treatment (maximum NFBG [maxNFBG]). Given the very high doses and long duration of action of MP, we suspect that time since administration did not play a major role in differential glucose responses, at least within the 18-hour time frame of our study.

An Expanded Disability Status Scale (EDSS)-certified neurologist (MDG), who was masked to BG values, classified patients with MS as recovered (full, partial, or none) retrospectively using outpatient and inpatient medical records and examinations of visits occurring before, during, and after the acute relapse. Prerelapse and postrelapse examinations were taken from available records obtained within 6 months of relapse assessment. The recovery classification was based on the compendium of available event-related data collected from clinical notes before, during, and after relapse, including patient reports and treating physician impressions, neurologic examinations, and motor strength testing in available physical therapy records focused on the relapse-affected area (e.g., documented return to full power in an affected limb). When the available data indicated a return to prerelapse function, the patient was classified as recovered. When the data indicated persistent neurologic deficits, examinations were compared with prerelapse records to determine whether there was no or partial recovery. To provide further granularity, relapse-relevant EDSS Functional System Scores (FSS) before, during, and after the acute relapse were extracted from chart-based neurologic examinations and documented on the standard neurostatus scoring form (neurostatus.net/scoring/index.php). When needed (e.g., in cases of optic neuritis), additional ophthalmology notes for visual acuity were used for the visual FSS. Relapse severity was then calculated for prerelapse, intrarelapse, and postrelapse FSS domains (e.g., vision, brainstem, pyramidal, sensory, or cerebellar) based on the standard neurostatus scoring. Severity was calculated by subtracting the prerelapse FSS from the intrarelapse FSS for relapse-relevant domains; similarly, FSS-based recovery was calculated by subtracting the postrelapse FSS from the prerelapse FSS for relapse-relevant domains. For patients with more than one relapse-relevant FSS, an average was calculated for both severity and recovery of the FSS-based scores. Using these FSS obtained from through neurologic examinations available in the medical records, the EDSS scores were calculated based on these 8 FSS in the standard fashion.

We compared patient characteristics using 1-way analysis of variance, t tests, or χ² tests as appropriate. Logistic regression analysis was used to examine the association between the maxNFBG and relapse recovery. We estimated 1 unadjusted model to predict recovery using the highest NFBG alone and 2 other models that adjusted for relapse severity (mild, intermediate, and severe) and demographic variables such as age and sex. We additionally grouped patients into high and low maxNFBG using the median maxNFBG, thereafter using logistic regression to compare the probability of recovery between low and high maxNFBG. For all logistic regression models, we computed the area under the receiver operating characteristic curves as well as the Hosmer-Lemeshow goodness of fit statistic. We used SAS 9.2.6 (SAS Institute, Cary, NC) and Stata SE 14 (Statacorp, College Station, TX) for statistical analysis and graphing.