Polypill and Risk of Vascular Morbidity and Mortality in CAD
Polypill and Risk of Vascular Morbidity and Mortality in CAD
Patients originated from the Second Manifestations of ARTerial disease (SMART) study, an ongoing prospective cohort at the University Medical Center Utrecht. Patients were newly referred by general practitioners or medical specialists from other hospitals in the region. Patients with terminal malignant disease, those not independent in daily activities, and/or those with insufficient knowledge of the Dutch language to understand the patient information were excluded. Patients with recently diagnosed clinically manifest CAD between January 1996 and February 2010 in the SMART study were included. Coronary artery disease was defined as myocardial infarction (MI), angina pectoris, elective percutaneous coronary intervention, or coronary artery bypass grafting.
The SMART study was approved by the local medical ethics committee and conducted in accordance with the guidelines of the Declaration of Helsinki. All patients gave written informed consent. The rationale and design of the SMART study have been described in detail elsewhere.
Participating patients underwent a diagnostic screening protocol for atherosclerotic disease and vascular risk factors. This screening protocol was done within the first weeks after referral. Participants completed questionnaires on cardiovascular history, risk factors, and current medication use. Physical examination consisted of measurements of height and weight. Body mass index (BMI) was calculated as weight in kilogram divided by the square of height in meters. Blood pressure was measured by sphygmomanometry at the right and left upper arm and repeated on the side with the highest values. The mean of the last 2 obtained measurements was used in the analyses. Blood samples were collected after overnight fasting. Low-density lipoprotein cholesterol was calculated using the Friedewald formula.
At baseline, brand name, generic name, frequency, and dosage were registered of current medication. Data from the self-reported use of pharmacologic agents, verified by the (electronic) medical records, have been recoded into drug classes. If the pharmacologic treatment comprised multiple pharmacologic agents combined into a single tablet/capsule, the treatment was categorized in the corresponding drug classes according to the active pharmacologic agents. Combination treatment with aspirin, a statin, and ≥2 BP-lowering agents was defined as the use of ≥2 BP-lowering agents from different treatment categories (β-blocker, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker, calcium channel blocker, and diuretic).
An indication for aspirin was defined as no antiplatelet or anticoagulant agent(s) and a history of established vascular disease. For statins, this was defined as no lipid-lowering agent(s) and an LDL cholesterol >2.5 mmol/L, and for BP-lowering agents, this was defined as no BP-lowering agent(s) and a systolic BP >140 mm Hg.
During follow-up, patients were asked to fill out a questionnaire biannually to report newly diagnosed diseases and hospital admissions. Outcomes of interest for this study were first (subsequent) occurrence of MI, ischemic cerebrovascular accident (iCVA), vascular mortality, a composite of these events (composite vascular end point), and all-cause mortality (Table I). When a possible event of interest was reported, patients' medical records and documentation were retrieved from their treating specialist or general practitioner. For those who died, specific cause of death was retrieved to differentiate vascular death from nonvascular causes of death. All events were adjudicated by 3 members of the SMART study end point committee comprising specialists from different departments. End point adjudication was performed without knowledge of the research question. Follow-up duration (in years) was defined as the period between study enrollment and first vascular event or death from any cause, date of loss to follow-up, or the preselected date of March 1, 2010. Of the 2,706 participants, 2% were lost to follow-up because of migration or discontinuation of the study.
Because complete case analysis leads to loss of statistical power and possibly biased results, missing values of key covariates were reduced by single imputation with regression methods (systolic BP [n = 4; 0.1%] and diastolic BP [n = 4; 0.2%], BMI [n = 1; 0.1%], triglycerides [n = 20; 0.7%], total cholesterol [n = 9; 0.3%], high-density lipoprotein [HDL] cholesterol [n = 24; 0.9%], and LDL cholesterol [n = 25; 0.9%]). Continuous variables are presented as mean and SD in a normal distribution and as median and interquartile range (IQR) in a skewed normal distribution. Categorical variables are presented as a percentage of the total (n [%]).
The relationship between combination therapy and cardiovascular events and all-cause mortality was assessed using Cox proportional hazards (Cox PH) regression models to calculate hazards ratios (HRs) with a 95% CI, after confirming that the conditions of proportionality were met. Different models were used to adjust for potential confounders. Model II was adjusted for age and gender, and model III was adjusted for age and gender and, additionally, BMI, smoking, pack-years of smoking, and presence of concomitant vascular disease (cerebrovascular disease [CVD], peripheral arterial occlusive disease [PAOD], and abdominal aortic aneurysm [AAA]). In model IV, total cholesterol, HDL cholesterol, and systolic BP were added.
Furthermore, the propensity score (PS) was calculated. The use of PSs is a method for dealing with confounding by indication. Correcting observational findings using a PS makes it possible to get similar baseline variables between treated and untreated subjects conditional on the PS. The PS is defined as an individual's probability of being treated with the intervention of interest, given the complete set of all information about that individual. Thus, the use of a probability that a subject would have been treated allows adjustment of the estimated treatment effect, creating "quasi-randomized" trial and reducing confounding by indication. The PS provides a single metric that summarizes all the information from explanatory variables. The PS was calculated using a logistic regression model, saving predicted probabilities in the PS, which included the following variables: age, gender, BMI, smoking, pack-years of smoking, presence of concomitant vascular disease (CVD, PAOD, AAA), total cholesterol, HDL cholesterol, and systolic BP. Balance of the covariates was checked for the covariates for the strata of the PS. More important, to check balance conditional on the covariates, the multivariate method was used, with the treatment as the dependent variable in a logistic regression model and the PS and covariates as independent variables. If balance is achieved, the PS should be the only significant explanatory variable for status of treatment. The other covariates should not add significant information. Balance on the covariates was achieved because the odds ratios of the covariates were near 1.0, and none reached a predefined significance level of P < .10. The PS was used as a continuous covariate in Cox PH regression models, replacing all covariates in the final PS model. The Cox PH was based on 2 independent variables: the indicator variable denoting treatment assignment and the estimated PS.
No extramural funding was used to support this work. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the manuscript, and its final contents.
Patients and Methods
Study Design and Population
Patients originated from the Second Manifestations of ARTerial disease (SMART) study, an ongoing prospective cohort at the University Medical Center Utrecht. Patients were newly referred by general practitioners or medical specialists from other hospitals in the region. Patients with terminal malignant disease, those not independent in daily activities, and/or those with insufficient knowledge of the Dutch language to understand the patient information were excluded. Patients with recently diagnosed clinically manifest CAD between January 1996 and February 2010 in the SMART study were included. Coronary artery disease was defined as myocardial infarction (MI), angina pectoris, elective percutaneous coronary intervention, or coronary artery bypass grafting.
The SMART study was approved by the local medical ethics committee and conducted in accordance with the guidelines of the Declaration of Helsinki. All patients gave written informed consent. The rationale and design of the SMART study have been described in detail elsewhere.
Baseline Examination
Participating patients underwent a diagnostic screening protocol for atherosclerotic disease and vascular risk factors. This screening protocol was done within the first weeks after referral. Participants completed questionnaires on cardiovascular history, risk factors, and current medication use. Physical examination consisted of measurements of height and weight. Body mass index (BMI) was calculated as weight in kilogram divided by the square of height in meters. Blood pressure was measured by sphygmomanometry at the right and left upper arm and repeated on the side with the highest values. The mean of the last 2 obtained measurements was used in the analyses. Blood samples were collected after overnight fasting. Low-density lipoprotein cholesterol was calculated using the Friedewald formula.
Combination Therapy
At baseline, brand name, generic name, frequency, and dosage were registered of current medication. Data from the self-reported use of pharmacologic agents, verified by the (electronic) medical records, have been recoded into drug classes. If the pharmacologic treatment comprised multiple pharmacologic agents combined into a single tablet/capsule, the treatment was categorized in the corresponding drug classes according to the active pharmacologic agents. Combination treatment with aspirin, a statin, and ≥2 BP-lowering agents was defined as the use of ≥2 BP-lowering agents from different treatment categories (β-blocker, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker, calcium channel blocker, and diuretic).
An indication for aspirin was defined as no antiplatelet or anticoagulant agent(s) and a history of established vascular disease. For statins, this was defined as no lipid-lowering agent(s) and an LDL cholesterol >2.5 mmol/L, and for BP-lowering agents, this was defined as no BP-lowering agent(s) and a systolic BP >140 mm Hg.
Follow-up
During follow-up, patients were asked to fill out a questionnaire biannually to report newly diagnosed diseases and hospital admissions. Outcomes of interest for this study were first (subsequent) occurrence of MI, ischemic cerebrovascular accident (iCVA), vascular mortality, a composite of these events (composite vascular end point), and all-cause mortality (Table I). When a possible event of interest was reported, patients' medical records and documentation were retrieved from their treating specialist or general practitioner. For those who died, specific cause of death was retrieved to differentiate vascular death from nonvascular causes of death. All events were adjudicated by 3 members of the SMART study end point committee comprising specialists from different departments. End point adjudication was performed without knowledge of the research question. Follow-up duration (in years) was defined as the period between study enrollment and first vascular event or death from any cause, date of loss to follow-up, or the preselected date of March 1, 2010. Of the 2,706 participants, 2% were lost to follow-up because of migration or discontinuation of the study.
Data Analysis
Because complete case analysis leads to loss of statistical power and possibly biased results, missing values of key covariates were reduced by single imputation with regression methods (systolic BP [n = 4; 0.1%] and diastolic BP [n = 4; 0.2%], BMI [n = 1; 0.1%], triglycerides [n = 20; 0.7%], total cholesterol [n = 9; 0.3%], high-density lipoprotein [HDL] cholesterol [n = 24; 0.9%], and LDL cholesterol [n = 25; 0.9%]). Continuous variables are presented as mean and SD in a normal distribution and as median and interquartile range (IQR) in a skewed normal distribution. Categorical variables are presented as a percentage of the total (n [%]).
The relationship between combination therapy and cardiovascular events and all-cause mortality was assessed using Cox proportional hazards (Cox PH) regression models to calculate hazards ratios (HRs) with a 95% CI, after confirming that the conditions of proportionality were met. Different models were used to adjust for potential confounders. Model II was adjusted for age and gender, and model III was adjusted for age and gender and, additionally, BMI, smoking, pack-years of smoking, and presence of concomitant vascular disease (cerebrovascular disease [CVD], peripheral arterial occlusive disease [PAOD], and abdominal aortic aneurysm [AAA]). In model IV, total cholesterol, HDL cholesterol, and systolic BP were added.
Furthermore, the propensity score (PS) was calculated. The use of PSs is a method for dealing with confounding by indication. Correcting observational findings using a PS makes it possible to get similar baseline variables between treated and untreated subjects conditional on the PS. The PS is defined as an individual's probability of being treated with the intervention of interest, given the complete set of all information about that individual. Thus, the use of a probability that a subject would have been treated allows adjustment of the estimated treatment effect, creating "quasi-randomized" trial and reducing confounding by indication. The PS provides a single metric that summarizes all the information from explanatory variables. The PS was calculated using a logistic regression model, saving predicted probabilities in the PS, which included the following variables: age, gender, BMI, smoking, pack-years of smoking, presence of concomitant vascular disease (CVD, PAOD, AAA), total cholesterol, HDL cholesterol, and systolic BP. Balance of the covariates was checked for the covariates for the strata of the PS. More important, to check balance conditional on the covariates, the multivariate method was used, with the treatment as the dependent variable in a logistic regression model and the PS and covariates as independent variables. If balance is achieved, the PS should be the only significant explanatory variable for status of treatment. The other covariates should not add significant information. Balance on the covariates was achieved because the odds ratios of the covariates were near 1.0, and none reached a predefined significance level of P < .10. The PS was used as a continuous covariate in Cox PH regression models, replacing all covariates in the final PS model. The Cox PH was based on 2 independent variables: the indicator variable denoting treatment assignment and the estimated PS.
No extramural funding was used to support this work. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the manuscript, and its final contents.
Source...