Summary

Variable rate intravenous insulin infusions (VRIII) are frequently used to manage blood glucose in inpatients. They are often used inappropriately and are associated with significant harm to patients.

The Joint British Diabetes Society (JBDS) published guidance in 2014 that a new balanced fluid was recommended as a substrate for patients on VRIII, given the significant risk of hypoglycaemia and electrolyte imbalance in these patients. This fluid was not widely available and was expensive compared to the usual fluids. No data had been published to show its superiority. Following publication of the guideline, a practice review found that the fluid was poorly managed across the Trust, resulting in high levels of hypokalaemia. A quality improvement project (QIP) was set up at Queen Elizabeth Hospital Birmingham (QEHB) to improve the outcomes of patients on VRIII and to allow a consistent standard of care in keeping with JBDS guidelines. The Trust’s medicines management group would be advised on the efficacy of the new fluid. The QIP led to a significant improvement in clinical outcomes in patients on VRIII, brought cost savings (over £600,000 per year) and provided the first evidence of the cost-effectiveness of the fluid along with VRIII in the NHS.

Innovation

At the time of publication of the JBDS guideline on VRIII, there was no published evidence to suggest the economic benefits of this expensive fluid. The main innovation was the readiness to introduce it to demonstrate efficacy where there was no previous evidence. The results are a consequence of many changes that helped to bridge the care gap. Staff training raised awareness of the guideline and the importance of adhering to it. Examples include the design of an online e-learning module for junior doctors and review of staff education at ward level by the ‘Diabetes Back to the Floor’ programme. The national guideline was adapted and made available electronically on the Trust’s website. The VRIII quick guide is a simple, one-page document with links to different important areas. An electronic insulin prescription was introduced, with preset hourly insulin rates to minimise variation so that the prescriber can simply choose the ‘Actrapid Inf Standard scale’. Also, when insulin is prescribed on the system, a link appears with the QEHB logo to point towards the quick guide to help people prescribe insulins appropriately with VRIII. The project only worked because of support from the multidisciplinary team; doctors from foundation year to consultant, pharmacists, diabetic nurse specialists, members of the hospital education team and IT technicians involved in the electronic prescribing systems. After an initial pilot in theatres and the medical assessment unit, the stocks of fluids were rolled-out to different clinical areas. Team-working and a robust QI methodology successfully demonstrated the efficacy of the introduction of a new fluid with VRIII to improve patient care, solving the nationwide issue of adverse events, such as electrolyte imbalance and hypoglycaemia in patients on VRIII. The hope is to take the QIP nationally. The team’s next challenge is to develop a computerised insulin infusion protocol with this substrate fluid to further enhance patient safety for the NHS.

Results

Measurable improvements in JBDS guideline-orientated care quality outcomes for diabetic inpatients treated with VRIII were demonstrated following intervention. Appropriate fluid choice (0.9%/0.45% NaCl with 5% glucose and 0.15%/0.3% KCl IV) increased from 23% at baseline to 95% at the final cycle. There were significant reductions in episode incidence of hyponatraemia from 28.6% to 8.3% and hypoglycaemia from 24.2% to 8.3%, as well as an absolute (but non-significant) reduction in episode incidence of hypokalaemia from 6.6% to 2.8%. Similarly, there were absolute increases in the proportion of patients with basal insulin continued and other diabetic medications stopped when on VRIII, while an increase in the average frequency of CBG measurement did not meet the significance level but did meet JBDS recommendations of >1/hour. National Diabetes Inpatient Audit (NaDIA) data showed a reduction in use of VRIII by 80% (16.8% to 9.9%) between 2016 and 2017, with inappropriate use falling from 5.7% to 0%.

Dissemination and Sustainability

The evidence suggests the combination of education and use of this recommended but ‘expensive’ fluid provides both clinical benefit for the diabetic inpatient on VRIII and potential economic benefit for the NHS. This work has been presented at the Association of British Clinical Diabetologists (ABCD) Spring 2018 meeting (winning Best Research Travel Grant) and is to be presented at the RCP Innovation in Medicine 2018 conference. This should help facilitate QEHB’s proposal to set up a national VRIII QIP in diabetic inpatients. A baseline survey questionnaire will be circulated to hospitals across the country to find out more on their current use of VRIII and thereafter initiate a multicentre QIP.

Method

Data was collected retrospectively using electronic records (PICS system), accessed via health informatics. All adults who had been on a VRIII between September and November 2016 (ten weeks) were identified as a baseline set and compared with cycles of two weeks of data in January, February, March, September, October and November 2017. All patients on a VRIII for longer than one hour on a medical or surgical ward were included. ITU patients were excluded. Between each cycle of data, interventions were introduced. Data was collected on: fluid use with VRIII and addition of potassium; daily monitoring of Us+Es; number of episodes of electrolyte abnormality;

Potassium <3.5mmol/l or Sodium <135mmol/l; the number of episodes of hypoglycaemia (glucose <4mmol/l); the frequency of CBG measurements; whether long-acting insulin continued during VRIII, and whether all other insulin/diabetic medications paused while on VRIII. The data was analysed and plotted on run charts and chi square tests were used to look at the statistical significancebetween the first (baseline data) and the last data collection in November 2017.