Implementation of a dedicated hospital subcutaneous insulin prescription chart: effect on glycaemic control

Limited Documentation and Treatment Quality of Glycemic Inpatient Care in Relation to Structural Deficits of Heterogeneous Insulin Charts at a Large University Hospital

OBJECTIVES

Insulin charts represent a key component in the inpatient glycemic management process. The aim was to evaluate the quality of structure, documentation, and treatment of diabetic inpatient care to design a new standardized insulin chart for a large university hospital setting.

METHODS

Historically grown blank insulin charts in use at 39 general wards were collected and evaluated for quality structure features. Documentation and treatment quality were evaluated in a consecutive snapshot audit of filled-in charts. The primary end point was the percentage of charts with any medication error.

RESULTS

Overall, 20 different blank insulin charts with variable designs and significant structural deficits were identified. A medication error occurred in 55% of the 102 audited filled-in insulin charts, consisting of prescription and management errors in 48% and 16%, respectively. Charts of insulin-treated patients had more medication errors relative to patients treated with oral medication (P < 0.01). Chart design did support neither clinical authorization of individual insulin prescription (10%), nor insulin administration confirmed by nurses' signature (25%), nor treatment of hypoglycemia (0%), which resulted in a reduced documentation and treatment quality in clinical practice 7%, 30%, 25%, respectively.

CONCLUSIONS

A multitude of charts with variable design characteristics and structural deficits were in use across the inpatient wards. More than half of the inpatients had a chart displaying a medication error. Lack of structure quality features of the charts had an impact on documentation and treatment quality. Based on identified deficits and international standards, a new insulin chart was developed to overcome these quality hurdles.

A human factors approach to subcutaneous insulin chart design improves user-performance: An experimental study

Insulin is a high-risk medicine that has been implicated in serious adverse events for hospital inpatients, including medication-error related deaths. Most insulin errors occur during administration, and "wrong dose" is the most common type. A paper-based subcutaneous insulin chart (the "NSIC") was developed for the Australian Commission on Safety and Quality in Health Care, using a range of human factors methods, with the aim of reducing the opportunity for errors. The present lab-based study empirically assessed whether the NSIC's human factors design translates into improved user-performance in the determination of insulin doses, compared with a pre-existing chart. Forty-one experienced nurses and 48 novice chart-users completed 60 experimental trials (30 per chart), in which they determined doses to administer to patients. Both groups determined insulin doses faster, and made fewer dose errors, when using the NSIC. These results support the utility of the usability heuristics employed in developing the chart.

Glucometric benchmarking in an Australian hospital enabled by networked glucose meter technology

OBJECTIVE

To assess glucometric outcomes and to estimate the incidence of hypo- and hyperglycaemia among non-critical care inpatients in a major Australian hospital.

DESIGN, SETTING AND PARTICIPANTS

A prospective 10-week observational study (7 March - 22 May 2016) of consecutive inpatients with diabetes or newly detected hyperglycaemia admitted to eight medical and surgical wards at the Royal Melbourne Hospital. Point-of-care blood glucose (BG) data were collected with networked glucose meters.

MAIN OUTCOME MEASURES

Glycaemic control, as assessed with three glucometric models (by population, by patient, by patient-day); incidence of adverse glycaemic days (AGDs; patient-days with BG levels below 4 mmol/L or above 15 mmol/L).

RESULTS

During the study period, there were 465 consecutive admissions of 441 patients with diabetes or newly detected hyperglycaemia, and 9817 BG measurements over 2953 patient-days. The mean patient-day BG level was 9.5 mmol/L (SD, 3.3 mmol/L). The incidence of hyperglycaemia was higher than for a United States hospital benchmark (patient-days with mean BG level above 10 mmol/L, 37% v 32), and that of hypoglycaemia lower (proportion of patient-days with mean BG level below 3.9 mmol/L, 4.1% v 6.1%). There were 260 (95% CI, 245-277) AGDs per 1000 patient-days; the incidence was higher in medical than surgical ward patients (290 [CI, 270-310] v 206 [CI, 181-230] per 1000 patient-days). 604 AGDs (79%) were linked with 116 patients (25%). Episodes of hyperglycaemia (BG above 15 mmol/L) were more frequent before lunch, dinner, and bedtime; 94 of 187 episodes of hypoglycaemia (BG below 4 mmol/L) occurred between 11 pm and 8 am.

DISCUSSION

Glucometric analysis supported by networked glucose meter technology provides detailed inpatient data that could enable local benchmarking for promoting safe diabetes care in Australian hospitals.

The high incidence of steroid-induced hyperglycaemia in hospital

AIM

The objective of this study was to systematically evaluate the incidence of steroid-induced hyperglycaemia in a tertiary referral hospital. We conducted a glucometric audit of a prospective protocol where glucose monitoring was routinely performed on patients treated with high-dose steroids.

METHODS

The protocol specified routine fingerprick glucose monitoring for patients commencing high-dose steroid therapy (prednisone 25mg/day, dexamethasone 4 mg/day, hydrocortisone 100mg/day, or more) for a minimum of 48 h. The medical records and charts of these patients were audited after a 6 month period.

RESULTS

There were 80 non-diabetic patients treated with high-dose steroids and 862 blood glucose (BG) readings were recorded. The mean BG was ≥8 mmol/L in 38 (48%) patients and ≥10 mmol/L in 11 (14%) subjects. Sixty-nine (86%) subjects had at least one BG ≥8 mmol/L, and 56 (70%) subjects had at least one BG ≥10 mmol/L. Among those with hyperglycaemia, it had developed within 48 h in 94% of subjects. When prednisone was administered as a once daily morning dose, glucose levels peaked in the afternoon, and would return to baseline by the next morning.

CONCLUSIONS

We conclude that steroid-induced hyperglycaemia is common in hospital. Patients should be monitored for hyperglycaemia upon commencement of high-dose steroid therapy and treatment commenced as indicated.