Reduction of cardiovascular medication when guidelines change: personalized prediction of who will be able to stop successfully

Background Patients whose indication for the use of antihypertensive and/or lipid-lowering drugs changes, may want to stop their medication. We aimed to develop a decision rule for successfully stopping preventive cardiovascular medication, thus providing the physician with individualised information and enhancing decision making concerning deprescription. Methods We re-analyzed data from the intervention group of our own previously published Evaluating Cessation of STatins and Antihypertensive Treatment In primary Care (ECSTATIC) study, a controlled trial in primary care in which we assessed the (cost-) effectiveness and safety of an attempt to deprescribe antihypertensive and/or lipidlowering drugs in a population with low cardiovascular disease risk. Potential determinants of successful deprescription were found in literature and expert opinion. We assessed demographic factors, physical examination measures, laboratory results, and information from questionnaires. Potential determinants showing a univariable association with a P<0.2 were tested in a multivariable prediction model with generalised estimating equations in SPSS version 23. We used cross-validation for internal validation of the model. Results Among those in the intervention group (N=492) 135 patients successfully stopped medication (27%). We found a systolic blood pressure (SBP) ≤140, using preventive cardiovascular medication ≤10 years, using either an antihypertensive or a lipidlowering drug, and using ≤1 class of antihypertensive drugs to predict successful stopping independently. Discrimination and calibration were reasonable, with an area under the curve of 0.70 (95% CI 0.65 to 0.75), reduced to 0.65 in cross-validation (95% CI 0.60 to 0.71). The decision rule derived from our model showed that the probability of successfully stopping medication was 45% if all four predictors were positive. Conclusion The highest probability of successful stopping (redundant) preventive cardiovascular medication is approximately 50% for patients who show all four factors when the decision is taken. If one of these factors is absent, probability is substantially lower. This information will help GPs to inform their patients and to improve decision making during deprescribing consultations. ABSTRACT Background Patients whose indication for the use of antihypertensive and/or lipid-lowering drugs changes, may want to stop their medication. We aimed to develop a decision rule for successfully stopping preventive cardiovascular medication, thus providing the physician with individualised information and enhancing decision making concerning deprescription. Methods We re-analyzed data from the intervention group of our own previously published Evaluating Cessation of STatins and Antihypertensive Treatment In primary Care (ECSTATIC) study, a controlled trial in primary care in which we assessed the (cost-) effectiveness and safety of an attempt to deprescribe antihypertensive and/or lipidlowering drugs in a population with low cardiovascular disease risk. Potential determinants of successful deprescription were found in literature and expert opinion. We assessed demographic factors, physical examination measures, laboratory results, and information from questionnaires. Potential determinants showing a univariable association with a P<0.2 were tested in a multivariable prediction model with generalised estimating equations in SPSS version 23. We used cross-validation for internal validation of the model. Results Among those in the intervention group (N=492) 135 patients successfully stopped medication (27%). We found a systolic blood pressure (SBP) ≤140, using preventive cardiovascular medication ≤10 years, using either an antihypertensive or a lipidlowering drug, and using ≤1 class of antihypertensive drugs to predict successful stopping independently. Discrimination and calibration were reasonable, with an area under the curve of 0.70 (95% CI 0.65 to 0.75), reduced to 0.65 in cross-validation (95% CI 0.60 to 0.71). The decision rule derived from our model showed that the probability of successfully stopping medication was 45% if all four predictors were positive. Conclusion The highest probability of successful stopping (redundant) preventive cardiovascular medication is approximately 50% for patients who show all four factors when the decision is taken. If one of these factors is absent, probability is substantially lower. This information will help GPs to inform their patients and to improve decision making during deprescribing consultations.