SiMSK Model Description (Page in development)

This table and list summarise the attendance data used in the baseline model. From our NNUH dataset we have over 80,000 data lines describing attendances of 10,000 patients over a three year period. Not all of these variables are essential, but they help paint a fuller picture of the system.

• Attendance day (starting at 1 for first day in data period)
• Clinic Code (Code for intra departmental clinics)
• Clinic lead (clinician, nurse etc)
• Appointment Priority - urgent or routine
• Appointment Category (New, FU, FUT, NFU)
• Booking System (Full Booking, No Choice Offered, Partial Booking)
• Referral source(Cons to Cons, GP/DT Referrals, Not Recorded or All Other Referrals)
• Referral Status (Ongoing or Discharged)
• Referral Discharge Reason(Moved Away, No Longer Required, Patient Died, Self Discharge, Treatment Completed, Ongoing )
• Unique Patient Key (anonymised)

The model can upload data from different sources and analyses them together for combined use. The SiMSK model uses patient data from the BSR-Biologics Register (BSRBR), the Norfolk Arthritis Register (NOAR) and the HQIP Rheumatology audit. This provides us with data on patients with Rheumatoid Arthritis being treated with both standard DMARD and Biologic treatments over a long period of time.

The model analyses the data to produce probabilistic disease progression pathways, broken down by clinically relevant variables: gender, DAS state last appointment, time since last appointment, smoking status and age group. The model can easily be adjusted to input a different set of disease progression variables. A similar technique is used in the model to simulate Global Health and HAQ scores.

The model is flexible enough to model disease progression based on limited data. For example to use limited trials data to model what the effect of implementing these drugs in a real world setting might be. It can even model completely hypothetical data, this is useful for cases where we do not have sufficient real world data to base these predictions on. For example on prescribing biologic drugs to patients with mild Rheumatoid Arthritis. We can then estimate these variables and explore the effects on the system, noting that it is based on our assumptions.