Monte Carlo simulation of uncertainty to identify barriers to optimizing blood pressure control

OBJECTIVES: To assess the impact of variable drug response and measurement error on SBP control.

METHODS: We simulated a treat-to-target strategy for populations with different pretreatment SBP, whereby medications were added sequentially until measured SBP (mSBP) less than 140 mmHg. Monte Carlo simulations determined variability of both drug response (drugeff ± σdrug; 10 ± 5 mmHg base case) and measurement error (σmeas; 10 mmHg base case) of true SBP (tSBP). The primary outcome measure was the proportion of individuals who achieved target less than 140 mmHg.

RESULTS: Decision-making based on mSBP resulted in 35.0% of individuals with initial tSBP 150 mmHg being either inappropriately given, or inappropriately denied a second drug. When the simulation was run for multiple drug titrations, measurement error limited tSBP control for all populations tested. A strategy of drug titration based on a second measurement for individuals at risk of incorrect decisions (mSBP 120-150 mmHg; σmeas 15 mmHg) reduced the proportion above target from 40.1 to 30.0% when initial tSBP 160 mmHg. When the measurement variability for the second reading was reduced below that usually seen in clinical practice (σmeas 5 mmHg), the proportion above target decreased further to 17.4%.

CONCLUSION: In this simulation, measurement error had the greatest impact on the proportion of individuals achieving their SBP target. Efforts to reduce this error through repeated measures, alternative measurement techniques or changing thresholds, are promising strategies to reduce cardiovascular morbidity and mortality and should be investigated in clinical trials. Here we have shown that Monte Carlo simulations are a useful technique to investigate the influence of uncertainty for different hypertension management strategies.