Web Resource Last Updated: 31-05-2023

This short video contains a description of how the analytics and risk prediction tool works

 

Individual Patient Predictions

When viewing individual patient records, on the home page, left column, you will see a section which supports individual risk predictions for key diabetes complications. You can set this to make predictions over 1,2, 5 or 10 year time periods.

Current Predictions include:

  • Renal (ESRF)- Probability of renal failure defined as creatinine >250 umol/l not ascribable to any acute intercurrent illness (ICD-9 codes 250.3 and >=585 and <=586) and death due to renal failure (ICD-9 codes >=580 and <=593.9)
  • Amputation- Probability of first amputation of a digit or limb (ICD-9 code >=5.845 and <=5.848, or 250.6) or a fatal peripheral vascular event (ICD-9 code 997.2, 997.6, 250.6 or 440.2)
  • Blindness- Probability of blindness in one eye defined as a visual acuity Snellen 6/60 or ETDRS logMAR 1.0 or worse for any reason, persisting for 3 months (ICD-9 codes >=369 and <=369.9).
  • Mortality- 5-year, all-cause, mortality for individuals with type-2 diabetes using a model with demographic, clinical, and renal variables.
  • MI (Myocardial Infarction)- Probability (in type 2 diabetes) of first MI, defined as acute myocardial infarction (ICD-9 code 410) or sudden death (ICD-9 code >=798 and <=798.9)

 

These predictions have been developed using information from the published literature, and from independent analysis of large volumes of diabetes population data. Models have been developed and/or validated using data from large diabetes populations in Scotland and England. Predictions will be based on data available within the system. Missing or inaccurate data could lead to false predictions. Predictions should be used as a guide only, and clinical discretion should always be used when making any treatment decisions.

Population level Predictions

Within the analytics platform, the “future risk” section calculates risk at population level using the same models as described above. Populations are then divided into low, medium and high risk (coloured green, amber and red)

This is currently based on a threshold of

  • Low: <5%
  • Medium: 5-10%
  • High: >10%

 

By clicking on any section you can create a list of patient within that risk category, and export the list to enable further targeted intervention.

 

 

 

For further information, please see terms and conditions for more information about use of this platform.

 

 

Leave a review