Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Up to 40% of hospitalized patients treated with prednisolone will experience hyperglycemia.Current guidelines recommend management of acute hyperglycemia in hospitalized patients with basal-bolus insulin (BBI), but the optimum treatment strategy for patients with prednisolone-associated hyperglycemia has not been defined.WHAT THIS STUDY ADDS In hospitalized patients with prednisolone-associated hyperglycemia, use of an individualized BBI dose regimen was associated with improved glycemic control compared with standard weight-based BBI regimens.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY This study provides proof of concept that an individualized BBI regimen may improve glycemic control in patients with prednisolone-associated hyperglycemia; additional prospective studies evaluating the impact of individualized BBI on glycemic control compared with standard subcutaneous insulin regimens are needed.Introduction Glucocorticoids are a class of medications that are widely prescribed to hospitalized patients. 1 2 Glucocorticoids induce insulin resistance in skeletal muscle and the liver and acutely reduce insulin secretion.3 4 This predisposes to the development of hyperglycemia, which occurs in up to 40% of hospitalized patients prescribed glucocorticoids.5–8Inpatient hyperglycemia is associated with a range of adverse outcomes including increased risk of death, hospital-acquired infection, and length of hospital stay.9–11 Similarly, an increased risk of mortality, hospital-acquired infection, and cardiovascular events has been observed in glucocorticoid-associated hyperglycemia.12 Consequently, several international guidelines emphasize the importance of treating glucocorticoid-associated hyperglycemia in hospitalized patients.13–15 However, there is a lack of evidence to inform a standardized treatment approach.Prednisolone accounts for 55–80% of in-hospital glucocorticoid prescriptions.5 8 A morning dose of prednisolone causes a characteristic circadian pattern of glucose elevation, which occurs predominantly in the afternoon and early evening.16 However, in studies where patients were prescribed the same insulin dose calculated as a percentage of body weight, differences in the pharmacokinetic profile of an insulin regimen did not have a major effect on glycemic control.17 18 In these studies, there was wide variability in the response to a weight-based insulin dose in patients prescribed glucocorticoids, which may obscure the effect of insulin pharmacokinetics.17 18We recently reported a study investigating which clinical factors are associated with insulin requirements in patients with hyperglycemia prescribed prednisolone, with an aim to inform individualized insulin dosing.19 In this manuscript, we report additional data from members of that cohort investigating whether individualization of a subcutaneous glargine-based basal-bolus insulin (BBI) regimen improves glycemic control of prednisolone-associated hyperglycemia.Methods Study population and study design This was a cross-sectional study. The individualized treatment group included 23 adult inpatients with prednisolone-associated hyperglycemia prescribed an insulin infusion for 24 hours to assess their daily insulin requirements and who remained in hospital for a minimum of 24 hours on a subcutaneous BBI regimen. Details of the insulin infusion study have previously been reported. 19 In brief, all participants were admitted to the hospital with an acute inflammatory illness, prescribed a single morning dose of oral prednisolone 20 mg/day or greater for at least 3 days and had hyperglycemia (defined as two point-of-care (POC) capillary blood glucose levels >10.0 mmol/L (180 mg/dL) or one POC blood glucose level>15.0 mmol/L (270 mg/dL)) within the preceding 24 hours. Type 1 diabetes mellitus and patients who were nil by mouth were exclusion criteria in this study. The study protocol was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12618001211257).The standard treatment group comprised 24 adults from a previous observational trial evaluating glycemic control in hospitalized patients with prednisolone-associated hyperglycemia.20 This study was approved by the Southern Adelaide Clinical Human Research Ethics Committee, Adelaide (029.12); patient consent was not required. For both groups, participants were included in analyses if they had at least three BGLs and at least three insulin doses within a 24-hour period.Insulin regimen Individualized treatment group Patients had been prescribed an intravenous insulin infusion comprising 50 units of Actrapid insulin (insulin aspart 100 IU/mL, Novo Nordisk A/S, Sydney, Australia) in 50 mL of 0.9% sodium chloride and wore a Freestyle Libre 1 flash glucose monitor (Abbott Laboratories, Chicago, Illinois, USA) for 24 hours. At the completion of a 24-hour insulin infusion, each patient’s total daily insulin dose (TDD) and average 24-hour flash glucose concentration were calculated. These were used to determine subcutaneous BBI doses for ongoing treatment. If the average 24-hour flash glucose concentration was >10.0 mmol/L, TDD was increased by 20%. For the average 24-hour flash glucose concentration between 7.5 and 10.0 mmol/L, TDD was unchanged. For the average 24-hour flash glucose between 4.5 and 7.5 mmol/L, TDD was reduced by 10%. If the participant experienced hypoglycemia during insulin infusion, defined as flash and POC glucose measurement <4.0 mmol/L with symptoms of hypoglycemia, TDD was reduced by 20%.The basal insulin dose was calculated by multiplying the insulin infusion dose between 00:00 and 06:00 hours by four. This was administered as a morning dose of insulin glargine (Optisulin 100 IU/mL, Sanofi Australia, Sydney, Australia). Prandial insulin doses were calculated by subtracting basal insulin dose from TDD. Prandial insulin doses of insulin aspart (Novorapid 100 IU/mL, Novo Nordisk A/S, Sydney, Australia) were divided between breakfast, lunch, and dinner in a 20%:40%:40% ratio; no further correctional insulin was administered.Flash glucose monitoring was ceased following the completion of the 24-hour intravenous insulin infusion as these data were unavailable in the historical control group. For this study extension, glucose monitoring was carried out in keeping with local hospital protocol as outlined below.Standard treatment group Basal insulin was insulin glargine (Optisulin 100 IU/mL, Sanofi Australia, Sydney, Australia) and prandial insulin was insulin aspart (Novorapid 100 IU/mL, Novo Nordisk A/S, Sydney, Australia). TDD was based on body weight using an institutional protocol (0.3 units/kg/day for diet-controlled diabetes, 0.4 units/kg/day for oral diabetes medications and continuation of their current daily dose in patients already treated with insulin). Basal insulin accounted for 50% of TDD, and the remaining 50% was divided into three equal portions administered with breakfast, lunch, and dinner. Correctional prandial insulin was administered as follows: 3 units for POC glucose 10.1–15.0 mmol/L and 6 units for POC glucose >15.0 mmol/L.Data collection Baseline demographic and biomedical data and insulin doses were recorded. POC glucose measurements were performed four times over 24 hours (fasted, prior to every meal and before bed (07:00, 12:00, 17:00 and 21:00)) in keeping with international guidelines and institutional protocol. 15 21Endpoints The primary endpoint was the mean 24-hour POC glucose achieved with each insulin regimen. We also compared mean POC glucose at each time point (07:00, 12:00, 17:00 and 21:00 hours), the proportion of POC glucose readings within target glucose range (defined as 4.0–10.0 mmol/L), prevalence of hypoglycemia, short-term (within 24-hours) glycemic variability by calculating the SD of the mean daily glucose concentration 22 and total daily, basal, and prandial insulin doses. Finally, we calculated the stress hyperglycemia ratio (SHR), a measure of relative hyperglycemia.23 SHR was calculated as: average 24-hour POC glucose/estimated average glucose over the previous 12 weeks. Estimated average glucose is derived from glycated hemoglobin (HbA1c), using the following formula: estimated average glucose=(1.59×HbA1c)–2.59.23 24Statistical analysis Continuous variables describing the baseline characteristics of the two groups are presented as mean±SD. The Shapiro-Wilk test was used to confirm the normality of the distribution. The initial analysis was for day 1 of the BBI insulin regimen, as all participants had sufficient data to be included in this analysis. Mean POC glucose over 24 hours, mean POC glucose at each time point (07:00, 12:00, 17:00, and 21:00), insulin doses, glycemic variability, and SHR were compared using unpaired t-tests and the proportion of POC glucose readings within target glucose range was compared using Fisher’s exact test.Patients who had 2 days of data were analyzed using a repeated measures linear mixed model, with insulin and day of treatment as the two variables included in the analysis. Only participants who remained in the hospital were included in the 2-day analysis. Statistical analysis was performed using GraphPad Prism V.9 (GraphPad Software, San Diego, California, USA) and IBM SPSS Statistics (V.27.0). A p value <0.05 was considered statistically significant.Results Participant characteristics There were no significant differences in sex, weight, body mass index, reason for admission to hospital, proportion of participants with known T2DM, HbA1c, or diabetes treatment between the groups ( table 1). Most participants had a preexisting diagnosis of T2DM (standard treatment 96% vs individualized treatment 83%). Of participants with known T2DM, 31% in the individualized group and 58% in the standard group were treated with glucose-lowering medications prior to hospitalization (table 1). This study was conducted prior to the approval of long-acting glucagon-like peptide 1 receptor antagonist use in Australia. There were no participants with known hemoglobinopathies. Participants in the individualized group were younger (70.4±10.7 vs 76.9±11.1 years, p=0.03) and had higher estimated glomerular filtration rate (eGFR) (72±22 vs 44±24 mL/min/1.73 m2) than those in the standard group (table 1).Table 1Baseline characteristics of 23 study participants with prednisolone-associated hyperglycemia treated with an individualized basal-bolus insulin regimen versus 24 historical controls with prednisolone-associated hyperglycemia treated with standard institutional weight-based basal-bolus insulin protocolStandardIndividualizedP valueNumber of participants2423Age (years)76.9±11.170.4±10.70.03Male (n, %)15 (63)17 (74)0.53Weight (kg)83.6±19.189.3±24.70.46BMI (kg/m2)30.0±5.430.3±8.30.99eGFR (mL/min/1.73 m2)44±2472±22<0.01Admission diagnosis (n, %) Respiratory21 (88)21 (91)0.99 Rheumatological3 (12)0 Other02 (8)Known diabetes (n, %)23 (96)19 (83)0.99HbA1c (%)8.1±1.08.6±2.30.64 (mmol/mol)65±1271±14Diabetes treatment (n, %) Diet only1 (4)4 (17)0.11 Orals only9 (38)12 (52) Orals and insulins14 (58)7 (31)Values represent mean±SD.BMI, body mass index; eGFR, estimated glomerular filtration rate; HbA1c, glycated hemoglobin.Day 1 analysis 23 patients were prescribed individualized BBI and 24 patients prescribed standard insulin were included in the day 1 analysis. The number of BGLs and insulin doses was 3.9±0.3 and 4.0±0.0 per day, respectively, in the individualized group and 3.9±0.3 and 4.0±0.0 per day, respectively, in the standard group. The mean prednisolone dose was similar in the two groups ( table 2).Table 2Daily prednisolone, insulin dose, and point-of-care (POC) glucose concentrations in participants with prednisolone-associated hyperglycemia on the first day of treatment with an individualized basal-bolus insulin regimen versus historical controls with prednisolone-associated hyperglycemia treated with a standard institutional weight-based basal-bolus insulin protocolStandardIndividualizedP valueNumber of participants (n)2423Prednisolone dose (mg)30.8±8.830.8±8.50.99Total daily insulin (units/kg)0.70±0.310.78±0.460.36Total daily insulin (units)56±2769±490.67Total daily insulin dose range (units)26–14920–182Insulin dose (units) Basal25±1338±380.13 Breakfast10±77±40.02 Lunch11±812±80.79 Dinner11±1412±80.71Total prandial insulin (units)33±1631±200.60POC glucose (mmol/L) 07:009.3±4.18.0±2.40.17 12:0011.0±4.411.3±4.90.93 17:0013.8±5.212.1±4.90.24 21:0014.1±4.611.6±4.90.07Values represent mean±SD.The TDD expressed as an absolute dose (units/day) and as per kg body weight (units/kg) were 11% and 23% higher in the individualized group; these differences were not statistically significant (table 2). Basal insulin accounted for 55% of TDD in the individualized group and 44% of TDD in the standard group; again, this difference was not statistically significant (table 2). The range of starting basal insulin dose ranged from 0 to 132 units/day and 10–64 units/day in the individualized and standard groups, respectively. There was minimal difference in the amount of prandial insulin prescribed to the two groups, with less insulin prescribed at breakfast to the individualized group. The SD and dose range for TDD were far wider in the individualized group, reflecting greater variability in insulin dosing (table 2).Mean 24-hour POC glucose was 1.2 mmol/L lower in the individualized group; this difference did not reach statistical significance (p=0.07, figure 1A). In both groups, mean 24-hour POC glucose was above 10.0 mmol/L, but fasting POC glucose was within the target range, and mean POC glucose at other time points was above target range (table 2). POC glucose at 3 of 4 time points was lower in the individualized group, but these differences were not statistically significant (table 2).Figure 1Day 1 mean capillary point-of-care glucose concentration (A), proportion of point-of-care capillary glucose within glucose range over day 1 (B), and SD of glucose on day 1 (C) in participants with prednisolone-associated hyperglycemia treated with an individualized basal-bolus insulin regimen versus historical controls with prednisolone-associated hyperglycemia treated with a standard institutional weight-based basal-bolus insulin protocol.On day 1, there were no episodes of hypoglycemia in the individualized group versus two episodes in the standard group. The proportion of POC glucose concentration within the target range (4–10 mmol/L) was higher in the individualized group (52.0±4.8 vs 37.0±4.5, p=0.0007, figure 1B). Intraday glycemic variability as assessed by the SD of mean glucose was lower in the individualized group (3.1±0.3 vs 4.0±0.3, p=0.04, figure 1C). There was no significant difference in SHR between the individualized and standard groups on day 1 (1.02±0.22 vs 1.17±0.40, p=0.13).2-day analysis 11 patients prescribed individualized insulin and 19 patients prescribed standard insulin were included in a 2-day analysis. The remainder of participants was discharged from the hospital by day 2. The number of BGLs and insulin doses was 3.9±0.3 and 4.0±0.2 per day, respectively, in the individualized group and 3.9±0.3 and 4.0±0.2 per day, respectively, in the standard group. There was no significant difference in prednisolone dose between treatment groups on either day (day 1, see table 2; day 2, 28.4±8.3 vs 27.9±8.3 mg, p=0.73).The SD for glucose increased from day 1 to day 2 across both treatment groups (figure 2A). Mean glucose and SHR were 5–29% higher in the standard group across day 1 and 2; these differences were not statistically significant (figure 2B,C). There was no significant interaction between time and treatment group in any analysis (data not shown). There were two further episodes of hypoglycemia over the 2-day analysis in the standard group.Figure 2Day 1 and 2 mean SD of glucose (A), capillary point-of-care glucose concentration (B), and stress hyperglycemia ratio (C) in participants with prednisolone-associated hyperglycemia treated with an individualized basal-bolus insulin regimen versus historical controls with prednisolone-associated hyperglycemia treated with a standard institutional weight-based basal-bolus insulin protocol.Discussion This analysis evaluated the performance of an individualized BBI regimen for the treatment of prednisolone-associated hyperglycemia in hospitalized patients. Compared with patients treated with a standard weight-based BBI regimen, use of an individualized insulin regimen was associated with a mean 24-hour POC glucose on day 1 that was 1.2 mmol/L, but this difference did not reach statistical significance. There was a significantly higher proportion of glucose measurements within the target glucose range and reduced glycemic variability on day 1 of individualized insulin. These results provide proof of concept that individualized BBI treatment of prednisolone-associated hyperglycemia in hospitalized patients may improve glycemic control.Much of the literature on treatment of prednisolone-associated hyperglycemia focuses on insulin pharmacokinetics; in particular, the concurrent administration of isophane insulin and prednisolone, as the profile of isophane approximates the pattern of prednisolone-induced hyperglycemia.16 25 When extra isophane is coadministered with prednisolone in addition to usual treatment, glycemic control improves.26 27 However, when comparable doses of glargine-based and isophane-based insulin regimens are studied, mean daily glucose levels are similar.17 18 28 29 This suggests that determination of an appropriate insulin dose is more important than insulin pharmacokinetics to optimize glycemic control in patients prescribed prednisolone.In this study, the mean individualized daily insulin dose prescribed per kg body weight was 23% higher and much more variable than in the standard treatment. Individualized insulin dosing was associated with a lower mean POC glucose that approached statistical significance, but mean blood glucose remained above the recommended blood glucose target range.In the individualized group, we first calculated a basal insulin dose based on each patient’s overnight insulin requirements during an insulin infusion. This was administered as insulin glargine, as over 80% of patients in the individualized group had type 2 diabetes and commonly need long-acting insulin to optimize glucose control. In our study, there was wide variability in glargine dose (range 0–132 units/day) in the individualized group. As prednisolone predominantly increases glucose concentration during the afternoon and evening, some patients in this group required little or no insulin overnight on the insulin infusion to maintain glucose in the target range. In contrast, in other patients, the overnight insulin requirements were high, likely reflecting the severity of underlying insulin resistance associated with type 2 diabetes.8 20 The average glargine dose of 0.4±0.3 units/kg accounted for 55% of the TDD and was approximately 50% higher than the basal insulin prescribed in the standard group. This resulted in a mean fasting POC glucose of 8 mmol/L with no fasting hypoglycemia, suggesting the amount of glargine prescribed to the individualized group was appropriate.The total daily prandial insulin dose for the individualized group was also derived from the 24-hour insulin infusion and then split into a 20:40:40 ratio for breakfast, lunch, and dinner to counter the circadian pattern of prednisolone-associated hyperglycemia.16 Across day 1, the average POC glucose concentration in the individualized group remained above the target glucose range at 12:00, 17:00, and 21:00 hours and was not significantly different from the standard group. The lack of benefit is not surprising as our calculation of insulin doses from the insulin requirements on a 24-hour insulin infusion did not result in higher doses of prandial insulin to the individualized group. Our findings are consistent with previous studies using various BBI regimens to treat glucocorticoid-induced hyperglycemia in which postprandial hyperglycemia persisted despite increasing short-acting insulin doses.18 26 27 29 Our findings demonstrate that, despite calculating insulin doses from an insulin infusion, which is a surrogate marker of insulin sensitivity, we underestimated the insulin doses needed to treat postprandial hyperglycemia in patients prescribed prednisolone and highlight the importance of considering higher prandial insulin doses in future studies.Increased time in the target glucose range is associated with reduced hypoglycemia and 30-day mortality in hospitalized patients.30 Previous studies assessing prednisolone-associated hyperglycemia management reported that different BBI regimens did not significantly affect the proportion of time in the target glucose range, which varied between 19.2% and 66%.17 18 27 29 However, in this study, individualized insulin was associated with a greater proportion of POC glucose levels in the target range. There were no episodes of hypoglycemia in the individualized group, whereas four episodes of hypoglycemia occurred in the standard group. Previous studies have reported both lower18 and higher27 frequency of hypoglycemia with glargine-based BBI regimens. Our study highlights the potential of dosing individualization to reduce the frequency of adverse events such as hypoglycemia and to improve the proportion of glucose measurements within the target range.Increased glycemic variability and relative hyperglycemia are additional measures of glycemia that are associated with poor outcomes in hospitalized patients.31 In the individualized group, the SD of mean glucose concentration, a measure of glycemic variability, was lower on day 1. Further, the mean SHR in the individualized group was 1.0, which is associated with the lowest rates of mortality in hospitalized patients.31 These positive findings in other glycemic variables further suggest that individualized treatment of prednisolone-associated hyperglycemia may be of benefit.There were fewer participants in the 2-day analysis, which limits the conclusions that can be drawn. However, there was an increase in glycemic variability across time with no positive change in mean POC glucose or SHR. These results are concordant with previous studies demonstrating that prednisolone-associated hyperglycemia does not generally improve over time, despite increasing insulin and reducing prednisolone doses.17 18 27 29This study provides proof of concept that an individualized BBI regimen may improve some components of glycemic control in patients with prednisolone-associated hyperglycemia and highlights the importance of adequate prandial insulin dosing in these patients. The individualized group in this study had been prescribed an insulin infusion for 24 hours, and this was used to determine each patient’s daily insulin requirements. We do not necessarily propose that an initial insulin infusion should be standard practice for these patients. Rather, we undertook this analysis to investigate the potential benefit that could arise from individualized insulin dosing for patients with prednisolone-associated hyperglycemia. We have recently reported additional clinical factors to body weight, namely sex, HbA1c, diabetes status, and diabetes treatment prior to hospital admission, that could be used to individualize insulin dose calculations in patients with prednisolone-associated hyperglycemia.19 Taken together, study findings emphasize first the wide variability in basal insulin dose required in this patient group, and the importance of adequate prandial hyperglycemia treatment.We acknowledge the limitations of the analysis. First and foremost, the use of a historical control group undergoing standard insulin treatment who were older and had a lower eGFR than the individualized group may have impacted glucose levels. Second, there was a difference in how the prandial insulin dose was apportioned and supplemental insulin was only prescribed to the control group. However, the historical control group consisted of patients from the same study site, on the insulin protocol in current clinical use, that is, our hospitals’ current standard of care. Moreover, the difference in prescription of supplemental insulin will result in underestimation and not overestimation of the benefit of individualized insulin. All patients completed the insulin infusion between 09:00 and 16:00 hours, and given the short elimination half-life of Actrapid insulin when administered intravenously, this is unlikely to have affected glucose control the next day, although this is a possibility.32 The subject group comprised patients both with and without known diabetes. While in the future, these patient groups may be managed differently, in hospitalized patients the association between hyperglycemia and mortality is greatest in patients without known diabetes, and current evidence suggests hyperglycemia should be treated regardless of underlying diabetes status.9 31 Additional limitations include a short duration of follow-up and inability to evaluate daily carbohydrate intake; however, the latter is in keeping with standard daily clinical care. Mean prednisolone dose in our study was 30.8 mg/day as this study only enrolled patients prescribed >20 mg/day prednisolone, and this dose is higher than reported in other studies of prednisolone use in hospitalized patients.8 Finally, the sample size was small, particularly for the 2-day analysis. However, the clinical decisions to discharge participants were made by the participant’s treating physician, based on clinical care needs and reflecting routine clinical practice in teaching hospitals.In conclusion, in this proof-of-concept study, we report that individualizing a subcutaneous BBI regimen for treatment of prednisolone-associated hyperglycemia in hospitalized patients is associated with positive effects on glycemic control. Improved glycemic control was characterized by a modest reduction in mean POC glucose, an increased proportion of blood glucose measurements within the target range, and reduced short-term glycemic variability. In our individualized treatment group, we used insulin glargine as the basal insulin and with a starting dose ranging from 0 to 132 units/day, and this was associated with a fasting glucose within the recommended range for hospitalized patients. The prandial insulin doses prescribed in this study were inadequate to control glucose levels at other times of the day. Given the limitations of the current study, including the use of historical controls and moderate sample size, this data should be viewed as an exploratory, hypothesis-generating analysis. The analysis does support the need for a prospective randomized study comparing the effect of standard weight-based versus individualized insulin dosing on glycemic control in hospitalized patients with prednisolone-associated hyperglycemia.SP110.1136/bmjdrc-2025-004963.supp1Supplementary data