Oral Presentation Australian Diabetes Society and the Australian Diabetes Educators Association Annual Scientific Meeting 2017

Identifying hospitalised patients at risk of adverse glycaemia: a risk stratification model (#8)

Mervyn Kyi 1 2 , Jane Reid 1 , Alex Gorelik 1 , Shanal Kumar 1 , Anna Galligan 1 , Lois M Rowan 1 , Alison J Nankervis 1 , Katie A Marley 1 , David M Russell 1 , Paul R Wraight 1 , Peter G Colman 1 , Spiros Fourlanos 1
  1. The Royal Melbourne Hospital, Parkville, VIC, Australia
  2. Department of Medicine, Royal Melbourne Hospital, The University of Melbourne, Melbourne, VIC, Australia

Background: In hospitalised patients, adverse glycaemia (both hypo- and hyperglycaemic extremes) should be avoided. We analysed a cohort of inpatients with diabetes and developed a risk-stratification model to predict those at risk of adverse glycaemia.

Methods: We recruited 643 consecutive inpatients with diabetes or new onset hyperglycaemia (random capillary blood glucose [BG] ≥11.1 mmol/L without known diabetes) with ≥2 day length of stay. Networked BG meters were used to collect capillary BG measures from the time of admission until discharge (or day 14 for long-stayers). Adverse glycaemia was defined as the occurrence of any capillary BG <4 or >15 mmol/L from day 2 onwards after admission.

Multivariable logistical regression analysis was used to investigate the association between adverse glycaemia and patient clinical factors (age, sex, Charlson index, admission creatinine, HbA1c, diabetes type & regimen), hospital treatment factors (surgery, glucocorticoid treatment, duration of hospital stay), and unstable day 1 BG (any BG <4, >15 or two BG >10 mmol/L). A split-sample approach was used for model construction and internal validation.

Results: Patient characteristics were: age 70±14 years; HbA1c: 7.6±1.7%; 33% insulin-treated. Adverse glycaemia occurred in 278 (43%) patients. Factors associated with adverse glycaemia were: Charlson index, HbA1c, duration of hospital stay, sulphonylurea or insulin treatment, and unstable day 1 BG (table). A risk-stratification model using these five factors had sensitivity 84%, specificity 60%, PPV 64%, NPV 82%. A second model using two practical factors easily available on admission (pre-admission insulin treatment or unstable day 1 BG) also predicted adverse glycaemia (sensitivity 83%, specificity 56%, PPV 56%, NPV 83%).

Conclusion: Factors associated with adverse glycaemia include pre-admission insulin or sulphonylurea treatment, unstable BG on day 1, higher HbA1c and greater comorbidities. These factors should be used to risk-stratify patients for concentration of specialist inpatient diabetes management efforts.

591bfa95967cd-Fourlanos_Table.JPG