Sequential drug decision problems in long-term medical conditions: a case Study of Primary Hypertension Eunju Kim ba, ma, msc

4.4.1The set of time

A cohort of hypertensive patients moves through the model experiencing hypertension-related events at discrete time periods until either death or the patient reaches an age of 100-years. At each time period, the decision maker (i.e., clinician) makes a decision on the choice of drug(s) (e.g., continue with the same drug(s), replace with a new drug or add a new drug into the current treatment regimen) considering the patient’s current blood pressure level, medical history and the risk of the CVD-related health states. In practice, the time to revisit the clinician after treatment initiation and to decide upon treatment success and tolerability is fairly short. If the time interval is too short, however, there is a potential risk of carry-over effect and maximal antihypertensive efficacy is unlikely to have been reached. From a methodological viewpoint, the total drug switching period would be too short to consider the long-term cost and effectiveness of sequential treatment policies. Therefore this study assumes that this decision-making cycle is three months.

It is practically impossible to consider all possible drug switches until 100 years old because of the huge number of possible disease pathways and treatment sequences: thus, this study reduces the size of the problem by restricting the maximum number of drug switches. The NICE clinical guidelines on primary hypertension recommend a 4-step treatment algorithm (i.e., three switches are allowed after the initial drug is prescribed at step 1) including two or three drug combinations^[63]. For the untreated hypertension after step 3 (i.e., blood pressure is still not controlled after three types of treatment), the NICE guidelines recommend to consider further Ds-based therapy, adding an AB or BB or seeking specialist advice because it may be associated with secondary hypertension. Depending on the cause of high blood pressure, treatments after step 4 may involve pharmacological treatment, surgery or even renal dialysis, which are beyond the scope of this study. Furthermore, the proportion of hypertensive patients who fail to control blood pressure after step 4 is small in practice^{[8, 63]}. As such, the hypertension SDDP model assumes that the maximum number of drug switches is three, and drug switching beyond step 4 is not included in this analysis.

Markov models are suitable to modelling a long-term disease, which involves an ongoing risk over time^[74]. Where a maximum number of drug switches are set, however, the hypertension SDDP model can be structured as a successive decision tree with an add-on Markov model. In this case, drug switching is only considered in the successive decision tree; and the add-on Markov model can be used to calculate the long-term impact of sequential drug use on cost and effectiveness. Therefore, the total follow-up period in the successive decision tree is one year (i.e., three months x four time periods, or T=(t₁,t₂,t₃,t₄)), whereas the overall cost-effectiveness of the sequential treatment strategy is modelled long-term (i.e., until death or 100 years old). This thesis refers to the successive decision tree as the short-term drug switching model and the add-on Markov model as the long-term CVD model in the following sections 5.4 and 5.5 for description of their structures.