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

IF t~=T

TotalCosts(h,t) = STCosts(h,t)+sum(CumLTCost(t,:);

TotalQALYs(h,t) = STQALYs(h,t)+sum(CumLTQaly(t,:);

ELSE

TotalCosts(h,t) = STCosts(h,t)+sum(CumLTCost(t:end,:);

TotalQALYs(h,t) = STQALYs(h,t)+sum(CumLTQaly(t:end,:);

END

END

END % End of loop h.

END % End of loop t.
% Calculate the total net benefit.

TNB = sum(TotalQALYs(:,:))*WTP-sum(TotalCosts(:,:));

Figure ‎6.. Pseudo-code of the function ‘EvModel’ included in the hypertension SDDP model
Given the treatment regimen used for h at t and the mean SBP with the variation before treatment, the function ‘SBPmodelling’ provided the treatment success rate, the mean SBP and SD after treatment (separately for the controlled and the uncontrolled), the probability of CVD, HF, DM and AE for a group of patients who were in h at t.

An individual Monte Carlo modelling method was used for the function ‘SBPmodelling’. Assuming there are 1,000 individual entrants in this cohort, the simulation randomly generated 1,000 samples of baseline SBPs (i.e., SBP_Before) and the SBP lowering effects (i.e., SBP_Rdt) using a lognormal distribution. The SBPs after treatment (i.e., SBP_After) were calculated based on the baseline SBP, age-dependent SBP change and the SBP lowering effect. For each SBP result after treatment, the 10-year CVD risk (i.e., TenYearCVD) was calculated using QRISK2, which considers a set of variables including gender, age, treatment history, DM, BMI, ethnicity, family history and other comorbidities. The 10-year CVD risk was adjusted to the three month basis (i.e., ThreeMonthCVD).

For the AEs, a beta distribution was used to generate a probability of AEs including DM. Given the probability of the patients who have an AE (i.e., pAE), 1,000 samples of AE occurrence were randomly generated by assigning 1 for the patients having an AE and 0 for the patients having no AE (i.e., Sampled_AE).

If a set of the sampled SBP, CVD and AE results met the criteria of treatment success, it was assumed that the treatment was successful for the individual patient. For the rest it was assumed that the treatment failed. The proportions of treatment success/failure, CVD, HF, DM and AE were the output of the function ‘SBPmodelling’ and used to calculate the transition probabilities to Success, Failure, Death and the long-term CVD model in the next period. The mean SBP and SD of the entrants were used to extrapolate forward their long-term costs and health outcomes.

% Given the treatment regimen, the mean SBP and the variation of the patients who are in h at t, the function ‘SBPmodelling’ provides the treatment success rate, the SBP and SD for the controlled patients and the uncontrolled patients, the probability of CVD, HF, DM and AE.