Benefits and Risk
Modelling benefit-risk, comparative-harm and effectiveness of contraindication
An EMR dataset defines the patient population (real patients)
​
The entire EMR dataset is used to give granular estimates of the rates of the events which are assigned to each patient
For each patient time to events are generated using Discrete Event simulation.
The rate estimates (λ) are varied over multiple simulations (quantifying uncertainty)
Time to a single event
Cohort simulation
Applying relative risks (Ω) creates drug exposure scenarios
​
​
The distribution of the results (number of events) is described probabilistically to answer “what is the chance that the new treatment is better than the standard of care?”
-
Example* of comparative harm between latanoprost and timolol maleate to treat glaucoma.
-
THIN database used to create the cohort and to provide risk estimates for outcomes.
-
The results of 1000 simulations are plotted in terms of net differences between therapies (timolol-latanoprost) for death due to COPD and due to hear failure (CHF).
-
Uncertainty in the risk and relative risk estimates gives the cloud of points; 78% are in the "northeast" quadrant implying we are 78% certain that there will be more deaths from both causes due to timolol compared to latanoprost in light of evidence and uncertainty
​
*only for methodological demonstration
