A Cross-Sectional Analysis of Bronchodilator Prescribing in COPD and Cardiovascular Comorbidity

Background: Bronchodilators used to treat Chronic Obstructive Pulmonary disease have been associated with adverse cardiovascular events. Moreover, the high prevalence of cardiovascular diseases (CVD) in COPD (8-40%) requires an evaluation of treatment differences between comorbid groups in order to understand the appropriateness of drug prescribing. Objective: To determine whether COPD patients with concurrent CVD are less likely to be prescribed bronchodilators compared to those without CVD. Methods: A retrospective cross-sectional study was conducted using the 2010 National Ambulatory Medical Care Survey (NAMCS) to quantify the association between concurrent CVD morbidity and the probability of receiving a bronchodilator prescription. Visits included patients ≥40 years who had COPD diagnosis. Surveyweighted data were analyzed through descriptive analysis, univariate (unadjusted) and multivariate (adjusted) logistic regression models. Demographics, patient, physician and visit characteristics were assessed as covariates in the models. Results: Out of 11,627,061 ambulatory visits recorded by patients ≥40 years with COPD diagnosis, majority was male (57%), non-Hispanic White (80%) and used at least one bronchodilator (55.5%). We found that a significantly lower proportion of the COPD-CVD group (32.3%) was treated with bronchodilators versus 57.6% for the non-comorbid group. The effect of CVD on bronchodilator prescribing was modified by gender, beta-blockers and asthma. CVD patients who were female, not prescribed beta-blockers and not having concomitant asthma were 94% less likely to be prescribed bronchodilators compared the non-CVD females with the same characteristics. On the other hand, males not prescribed beta-blockers and not having concomitant asthma had 68% lower probability when CVD coexisted with COPD. Female CVD patients prescribed beta-blockers and not living with asthma had only a 20% chance of being a bronchodilator user than those without CVD. Cardiovascular disease did not affect the utilization of bronchodilators in males taking beta-blockers who either had or did not have asthma. Conclusion: Concurrent CVD diagnosis is a significant factor for reducing the probability of prescribing bronchodilators for COPD and our findings provide evidence of variations in bronchodilator prescribing for stratified groups of COPDCVD patients. Most patients with COPD and CVD are less likely to be prescribed bronchodilators, with the exception of males who were also prescribed beta-blockers. Thus, this study highlights a specific patient subgroup for whom the guidelines are less likely to be observed.


INTRODUCTION
Chronic Obstructive Pulmonary Disease (COPD) is defined as a "preventable and treatable disease characterized by persistent airflow limitation that is usually progressive and associated with an abnormal inflammatory response of the lung to noxious particles or gases. Exacerbations and comorbidities contribute to the overall severity in individual patients." 1 The disease is clinically diagnosed as one or more of 3 subtypes -chronic bronchitis, emphysema and chronic airway obstruction. 2 COPD is the fourth leading cause of chronic morbidity and mortality in the United States affecting 6.5% of adults (13.7 million diagnosed cases) and leading to 133,575 deaths. 2 In 2010, there were 10.3 million physician office visits, 1.5 million emergency department visits, and 699,000 hospital discharges for COPD. 2 Globally, COPD ranked fifth cause of years lived with disability (YLDs) in the 2010 global burden of disease study. 3 The World Health Organization predicts that COPD will become the third leading cause of death worldwide by 2030. 4 Inevitably, COPD is and will continue to be a major global health problem. 5 Bronchodilators are the mainstay for symptomatic treatment of COPD due to their ability to relieve bronchial obstruction and airflow limitation, improve emptying of the lungs and reduce hyperinflation during rest and exercise. 6,7 Three major classes are recommended as first-line agents -beta-(β) 2-agonists, anticholinergics (AC), and methylxanthines, used alone or in combination or as single or combination drugs. 7 Despite the benefits of bronchodilator agents in the treatment of COPD, patients may be at an increased risk of cardiovascular (CV) toxicity due to beta-adrenergic stimulation. Previous studies showing an association between bronchodilators and adverse cardiovascular events such as arrhythmias, myocardial ischemia, heart failure and sudden cardiac death, will be discussed in detail in the literature review section. [8][9][10][11][12][13] The potential cardiovascular risk of bronchodilators in COPD has important therapeutic implications for two main reasons: First of all, the prevalence of cardiovascular-related diseases (CVD) in COPD is estimated to range from 8% to as high as 40% 1, [14][15][16] making it the most frequent and most important disease group coexisting with COPD. Secondly, the exclusion of comorbid subjects from clinical trials often limits the use of trial data in providing practical treatment recommendations for the real world COPD patient population where multiple morbidities are highly prevalent.
We hypothesize that cardiac disorders play an independent role in the lesser probability of receiving a bronchodilator prescription for COPD. The objective of this study is to provide descriptive data on the classes of bronchodilator medications prescribed to ambulatory patients with diagnosed COPD in the United States and examine whether COPD patients with CVD are less likely to be prescribed bronchodilators compared to COPD patients without CVD.

Etiology and Epidemiology of COPD and its Comorbidities
Cigarette smoking is the dominant and best-studied risk factor for COPD, though a substantial burden of disease occurs in nonsmokers. [17][18][19] Other etiologic factors such as advanced age, occupational exposure to dust, fumes and gases, indoor and outdoor air pollution, airway hyper-responsiveness, asthma, infections and genetic predisposition (α-antitrypsin deficiency) are also important. 1 Furthermore, it has been increasingly recognized that COPD is more than just a lung disease: it is a complex heterogeneous systemic disease, frequently associated with other chronic diseases including coronary heart disease, cardiovascular, respiratory tract, metabolic, hematological, musculoskeletal, gastrointestinal, renal diseases, psychiatric disorders and neoplasias. 14,15,[20][21][22][23] The prevalence of comorbidity in COPD has been studied using nationallyrepresentative data, National Health and Nutrition Examination Survey (NHANES) 1999-2008. 24 Using a ten-year analytic sample of 14,828 subjects (995 COPD cases) aged ≥45, subjects with self-reported physician diagnosis of COPD were more likely than subjects without physician-diagnosed COPD to have coexisting arthritis (54.6% vs. 36.9%), depression (20.6% vs. 12.5%), osteoporosis (16.9% vs. 8.5%), cancer (16.5% vs. 9.9%), coronary heart disease (12.7% vs. 6.1%), congestive heart failure (12.1% vs. 3.9%), and stroke (8.9% vs. 4.6%). 24 Vanfleteren and colleagues used a cluster analysis approach to assess the prevalence and severity of objectively identified chronic concomitant disorders in a cohort of 213 patients with moderate to very severe COPD. 25 The CIRO COmorbidity (CIROCO) study was an observational single-center study and almost all subjects (97.7%) had one or more comorbidities. More than half (53.5%) of the patients had at least four comorbidities.
Another study in a sample of COPD patients referred to an Australian outpatient hospital-based pulmonary rehabilitation program reported that 96% of patients had at least 1 other chronic conditions and 29% had 5 or more comorbidities. 26 Data were available on 70 patients and 64% of patients had multiple morbidities associated with cardiovascular disease.
While it is likely that the prevalence of comorbidities in people with COPD is influenced by common risk factors, such as smoking, advanced age, chronic systemic inflammation, physical inactivity and low socioeconomic status, studies have confirmed that concurrent conditions worsen health outcomes independent of preexisting risk factors. 26 Comorbidities in COPD lead to lower health-related quality of life, 27,28 more hospitalizations, severe disabilities 5 and higher rates of mortality 29,30 compared to those without comorbidities.

Burden of comorbid COPD and Cardiovascular disease (CVD)
Observational studies indicate that COPD is associated with a high prevalence of coronary artery disease, cardiac arrhythmias, cerebrovascular disease, peripheral vascular disease and congestive heart failure. [31][32][33] In addition, COPD patients with CVD pose significantly higher clinical and economic burden on patients and healthcare system than COPD patients without CVD.

Safety of Bronchodilators in COPD
Bronchodilators are the mainstay for symptomatic treatment of COPD due to their ability to relieve bronchial obstruction and airflow limitation, improve emptying of the lungs and reduce hyperinflation during rest and exercise. 6,7 Bronchodilator treatments include beta-(β) 2-agonists, anticholinergics (AC), and methylxanthines, used alone or in combination. They are recommended as first-line agents useful to prevent or reduce symptoms as-needed or on a regular basis. 7 Short-acting bronchodilators are typically used as rescue medication for immediate symptom relief, routinely in the early disease stages or in COPD patients with a low symptom burden whereas long-acting bronchodilators are used as regular maintenance therapy and are generally more effective and convenient. 7 Combinations of bronchodilators may improve efficacy and reduce risk of adverse effects rather than increasing the dose of a single agent. 7 Inhaled bronchodilators are preferred over oral preparations based on favorable efficacy and side effect profile. 1 Despite the benefits of bronchodilator agents in the treatment of COPD, patients may be at an increased risk of cardiovascular (CV) toxicity due to beta-adrenergic stimulation. Evidence on the safety of bronchodilator therapy in COPD is mixed.
There have been reports on the relative safety of bronchodilator use in COPD 38 whereas some researchers have found clinically significant higher risks in the incidence of cardiovascular events in patients on selected bronchodilator agents than control groups. 8,11,12,39,40 In 2003 were dispensed within 60 days prior to the hospitalization. While it is likely that bronchodilator use was simply a marker of disease severity and this accounted for their association with cardiovascular disease, the authors note that ICS was not associated with these outcomes. The findings on specific drug interactions were that each case the receipt of cardiac drugs decreased the likelihood of hospitalizations associated with bronchodilator. For example, bronchodilators increased the risk of stroke in patients who were not taking cardiac drugs but did not increase the risk of arrhythmia in the presence of anti-arrhythmic agents.
In 2012, another nested case-control study involving 6018 subjects with COPD from the Saskatchewan Cohort Study 9 found that the rate of arrhythmia was elevated with the new use of ipratropium, a short-acting anticholinergic agent (RR 2.4; 95% CI, 1.4-4.0) and of long-acting β-agonists (RR, 4.5; 95% CI, 1. 4-14.4). The researchers did not find an elevated risk with short-acting beta-2-agonists or methylxanthines and proceeded to ascertain the results in a larger patient pool. For the reassessment, the larger Quebec cohort of 76,661 subjects was analyzed. 10  There was no influence of β2 agonist duration of use, dose, or type on outcome in fully adjusted models.
The possible role of the mode of administration of bronchodilators in observed differences in treatment effects has also been explored. Available formulations include metered dose inhalers (MDI), dry powder inhalers (DPI) and soft mist inhalers (SMI).
Using a retrospective study, Verhamme and colleagues 12  The publications on safety of bronchodilators reveal the likelihood of different study designs to bias and confounding. Smoking status, non-respiratory drugs with known arrhythmogenic properties taken concomitantly may bias the study results. The adverse cardiovascular outcomes may reflect COPD severity or confounding by indication rather than an adverse effect of bronchodilator therapy. It is also possible that bronchodilators impose a larger relative risk in people with undiagnosed cardiovascular disease than in people with known disorders since bronchodilators did not increase the risk of arrhythmia in the presence of anti-arrhythmic agents.

Utilization of Bronchodilators
Although, long-term decline in lung function is irreversible in COPD, bronchodilators are recommended to reduce airflow obstruction, reduce the frequency of exacerbations, improve health status and decrease mortality. 1 Therapy usually includes a short-acting bronchodilator or a long-acting bronchodilator (either a Beta-2agonist or an anticholinergic drug or a methylxanthine) or combination products for adequate clinical response and tolerance of side effects. 1 A case-control study was conducted on the use of bronchodilators and arrhythmias in COPD in 2012. 9 Out of 469 cases of arrhythmias, the proportion taking at least one short-acting β-agonist was 37.9%, long-acting β-agonists 1.9%, Women had a consistently higher proportion of visits with COPD medication prescriptions than visits by men. The frequency of COPD medications prescribed to women decreased by 7% (from 73% to 67%) during the study period (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004).
The average number of COPD drug mentions per visit decreased for both women and men and trends were significantly different between men and women (P < 0.001).
Prescription of methylxanthines decreased in visits by women and men, whereas prescription of anticholinergics increased continuously in the recent years.
Certainly, the high prevalence of COPD and cardiovascular disease warrant further study of prescribing patterns in COPD patients with cardiac conditions. Thus, the purpose of the present study is to determine whether COPD patients in ambulatory care in the United States have a lower likelihood of being prescribed bronchodilators if they have concurrent CVD compared to those without CVD.

Data Source
The 2010 National Ambulatory Medical Care Survey (NAMCS) 43  Therefore, we consider the dataset suitable for this analysis. The 2010 NAMCS data were used because they were the most recently available.  Figure 1 depicts the inclusion and exclusion criteria.
Further stratification of COPD sub-groups was done by presence or absence of concurrent diagnosis of cardiovascular conditions of interest. Based on previous studies, the following cardiovascular diseases were selected: myocardial infarction, angina, coronary artery disease, cardiac arrhythmias (including tachycardia, atrial fibrillation, and cardiac arrest), and congestive heart failure. The ICD-9 diagnosis codes are presented in Table 2.

Variables Specification
The dependent variable is defined as the prescription of a bronchodilator at a visit (Rx =1 or 0). New or continued bronchodilators recorded in any of the 32 therapeutic class fields were identified using the therapeutic class codes extracted from NCHS's Ambulatory Care Drug Database.  Figure 3) and implies it can be coded as either a continuous or categorical variable. Age was modeled as a continuous variable for the purpose of this study.

Descriptive Statistics and Modeling Strategy
Descriptive statistics, univariate and multivariate techniques were used. All analyses were performed on weighted data as recommended by the NCHS. The patient visit weighting, as provided, uses the most recently available census data to provide a stratified representation of the nation's patient population. All statistical analyses were done using SAS version 9.3 and the results report weighted information.
For descriptive analysis, the means (with standard error) and proportions of demographic, clinical and medication utilization characteristics for the study population were calculated. Rates of drug visits were computed for the sub-categories of bronchodilator medications. Bivariate associations between the dependent variable (prescription of one or more bronchodilators) and each independent variable were separately estimated using univariate logistic regression to obtain unadjusted Odds Ratios and 95% confidence intervals. This will enable inferences to be deduced about the true population parameter lies between the confidence limits with 95% certainty.
Confounding and interaction were assessed given that they may bias the relationship between CVD morbidity and treatment with bronchodilators. Two multivariate logistic regression (LR) models were developed -one containing interaction terms plus confounding variables and the other without interaction (confounders only). First of all, interaction was tested by applying the hierarchy principle to combinations of variables. This implies that if certain interaction terms are significant, all lower order components of such terms cannot be deleted and will remain in any further models considered. Two-way interaction analysis was performed for the following pairs of variables: CVD and gender, CVD and beta-blockers, CVD and asthma. Here, the Odds Ratios of prescribing a bronchodilator for COPD in a CVD patient is described in terms of the modifying variable, either gender, betablockers or asthma. The multivariate logistic regression (LR) model without interaction terms was developed.
Both multivariate models included a number of confounders -Age, gender, race, insurance coverage, asthma, concomitant antiarrhythmic and beta-blocker use, and physician specialty. Therefore, the multiple regression analysis presented adjusted models and allowed us to quantify the true association between concurrent COPD-CVD morbidity and the probability of receiving a bronchodilator prescription, while controlling for the effects of other explanatory variables.

Model fit
The next step in the model building procedure was to verify whether or not the model fitted the data well. Hosmer and Lemeshow goodness-of-fit test and a likelihood ratio test were conducted on the two multivariate logistic models -full model with interaction and reduced model without interaction. A final model was selected based on the adequacy of the fit.
In interpreting the results of the likelihood of prescribing a bronchodilator for COPD visit of a patient with cardiovascular comorbidity, the Adjusted Odds Ratios derived from the selected multivariate logistic regression model were used. The 95% confidence intervals were also calculated for each Odds Ratio.

Study Sample Characteristics
An estimated 13,799,072 visits of patients above 40 years had a diagnosis of COPD occurred in 2010 in the US. This is approximately 1.4% of 1billion annual office visits. Missing or unknown data were removed leaving 11,627,061 visits in the analytic sample ( Table 4).

Univariate Logistic Regression
The univariate logistic regression analysis performed indicated that most of the independent variables were significantly associated with the dependent variable (

Multivariate Logistic Regression
Two multivariate logistic regression (LR) models were developed -one containing interaction terms, exposure and other independent variables and the other without interaction (exposure plus independent variables only).
In the interaction model, the effects were significant between CVD and each of three possible modifying variables tested -gender, beta blockers (BB) and asthma since all p-values < 0.05. All three modifiers and their product terms with CVD were retained in the model according to the hierarchy principle.
The reduced model without interaction terms contained the exposure variable (CVD), and potential confounding variables yielded the results in Table 8. A likelihood ratio test was conducted based on the null hypothesis of 'no interaction' versus 'interaction' as alternative hypothesis. The test statistic was 134.4, p-value < 0.0001 produced evidence that the 'interaction' model is a significantly better fitting model.
Therefore, the interaction model was selected as the best model and this final choice guided the interpretation of the parameter estimates and odds ratios. Table 7 displays the Odds Ratio for the modifying effect of gender, asthma and β-blockers on the association between CVD and bronchodilator prescribing in COPD.
For effect modification by gender, CVD had a greater effect on bronchodilator prescribing compared to the non-CVD group in females with neither asthma nor betablocker medication. OR=0.06, 95%CI 0.04-0.09 implies a 94% reduction in the probability of receiving bronchodilator. In males with no asthma and no beta-blocker utilization, the effect of CVD was 68% since OR=0.32, 95%CI 0.25-0.40. The highest difference in the odds of bronchodilator use between CVD and non-CVD groups was observed in females who had asthma and did not use beta-blockers (OR=0.03, 95%CI 0.02-0.06). Thus, this patient group had the least chance of taking bronchodilators.
CVD was not associated with bronchodilator prescribing in male beta-blocker users, regardless of asthma status (for male, BB users with asthma OR=0.62 95%CI 0.33-1.16; for male BB users without asthma OR=1.11 95%CI 0.82 -1.51). Therefore, cardiovascular disease did not affect the outcome in this category of patients. On the contrary, CVD female beta-blocker users without asthma had 79% less probability of receiving a bronchodilator than similar patients without CVD.
The effect of cardiovascular disease was also modified by concomitant asthma to different degrees. CVD asthma patients who were males and not taking BB were only 17% as likely to be prescribed bronchodilators compared to those without CVD, that is, 83% less chance of receiving the drug. The effect of CVD was stronger for female asthma patients taking BBs with 89% lower likelihood, and strongest for female asthma patients who did not use BB (97% lower likelihood).

CHAPTER 5 DISCUSSION
Our study revealed that β-agonists were the most frequently prescribed class of bronchodilators and were prescribed at 56.4% of COPD visits. This fact supports previous studies. 44 We conducted this cross-sectional study to determine whether prescribing rates of bronchodilators in COPD patients were different in cardiovascular comorbidities and if so, to what extent patient characteristics modified the prescribing pattern in ambulatory care in the United States.
We found that most COPD patients, who had been diagnosed with cardiovascular disease, were prescribed bronchodilators less frequently than were those who had COPD without cardiovascular diagnosis, but that this relationship was modified by patient characteristics. This pattern was more evident for females, whereas males had less pronounced differences between CVD and non-CVD groups. Our finding was consistent with literature on treatment differences in bronchodilator utilization for comorbid groups. 45 There were significant interactions between cardiovascular disease and gender, βblocker and asthma, shown in Table 7. The presence of cardiovascular diagnosis significantly decreased the likelihood of bronchodilator prescription by 97% for females who were not taking beta-blockers but had asthma (OR=0.03, 95% CI 0.02-0.06). Males with similar characteristics had 83% reduced odds of taking bronchodilators. This was not entirely suspected as a previous cross-sectional study by Suh et al based on 10 years data from the NAMCS and hospital files (NHAMCS) 42 found that women had a consistently higher proportion of visits with COPD medication prescriptions than visits by men.
No significant difference in bronchodilator utilization between CVD and non-CVD groups was observed with the use of beta-blockers in males who were either asthma (OR=0.62 95% CI 0. 33-1.16) or non-asthma patients (OR=1.11 95%CI 0.82-1.51). A possible explanation may be that the concurrent use of beta-blockers for cardiac conditions and bronchodilators (which are majorly beta-agonists) in patients with multiple morbidities, leads to opposing pharmacological activity due to interaction between these medications. Other studies have found that the effects of β-agonists may be diminished by β-blockers and vice-versa. 46 In the Cooperative Cardiovascular Project 47 , β-blocker use was not associated with lower mortality among patients receiving concurrent beta-agonists. Subjects enrolled in the Veterans Administration's ACQUIP trial who received both beta-blockers and beta-agonists had no increase in risk of acute coronary syndromes. 48 However, we did not observe this nullifying effect among females with comparable characteristics. The effect of CVD on bronchodilator prescribing in females taking beta-blockers with coexisting asthma was about 79% less than those without CVD (OR=0.21 95%CI 0.14 -0.30). These results point to the fact that the effect of CVD on prescribing differs between males and females who concurrently use beta-blockers and bronchodilators (mainly beta-agonists) and suggest more conservative prescribing practices for females than for males.

Strengths and Limitations
One of the strengths of the current study is the dataset. The NAMCS provides a national perspective on drug utilization in ambulatory medical care. We included patients with COPD diagnosis, race and ethnicity data, Bronchodilator prescriptions were assumed to be directly indicated for COPD treatment which may not be accurate. The NAMCS data has a comprehensive list of medications and diagnosis codes documented for each patient encounter but since there is no clear match between the prescribed medications and ICD-9 diagnosis codes, we could not verify the specific diagnosis for which bronchodilators are being prescribed.
Future improvements to the survey design to establish a link between these disease condition and medications prescribed would be helpful to researchers. The dataset did not contain specific information that would indicate that patients may have a contraindication or past drug-related adverse event, thereby preventing the use of one or more bronchodilator medication.
Another limitation is the type of drug formulation. 17.4% of the prescribed products per visit were fixed combinations of different classes of bronchodilators or a bronchodilator in addition to an inhaled corticosteroid. However, we categorized and assessed them like any other bronchodilator in the study because evidence suggests inhaled corticosteroid present less cardiovascular risk unlike oral corticosteroids that may cause sodium and fluid retention leading to elevated blood pressure and adverse cardiovascular effects. 49,50 Combination regimens are substantially utilized in COPD due to the increased effectiveness, reduced burden of medication use and increased adherence that can potentially achieved with this agents. Therefore, all bronchodilatorcontaining drugs whether single agents or in combination with corticosteroids were included in the definition of the outcome variable.
We considered including disease severity measures as covariates in this study but were unable to because the cross-sectional data did not contain clinical outcomes or follow-up information. Prescribing practices often differ from provider to provider and it may be worth investigating predictors of bronchodilator prescribing among physicians. Future research should attempt documenting the physician characteristics influencing bronchodilator prescribing for COPD patients. The outcome of this type of analysis can provide inputs for shaping integrated disease management programs for concurrent COPD and CVD conditions.

CHAPTER 6 CONCLUSION
Despite considerable research interest in the potential risk of cardiac events associated with use of bronchodilators in COPD patients, few studies have analyzed the real-world prescribing rates of this highly utilized COPD therapy in cardiovascular disease. This study presents a unique national perspective to understand the effect of CVD on prescribing practices of bronchodilators in the United States.
Concurrent CVD diagnosis is a significant factor for reducing the probability of prescribing bronchodilators for COPD and the odds of prescribing were further decreased due to differences in gender, asthma and beta-blocker use. CVD patients who were female, not taking beta-blockers and not having concomitant asthma were 94% less likely to be taking bronchodilators compared the non-CVD females with same characteristics. On the other hand, males not taking beta-blockers and not having concomitant asthma had 68% lower probability when CVD coexisted with COPD. Female CVD patients taking beta-blockers and not living with asthma had only 20% chance of being a bronchodilator user than those without CVD. It is noteworthy that the general pattern observed in these national data is consistent with treatment guidelines that advise using bronchodilators with caution in the comorbid COPD-CVD patient population, however cardiovascular disease did not affect the utilization of bronchodilators in males taking beta-blockers.
Our findings provide evidence that most patients with COPD and CVD are less likely to be prescribed bronchodilators, with the exception of males who were also prescribed beta-blockers. Thus this study highlights a specific patient subgroup for whom the guidelines are less likely to be observed.    * Mean age and standard error in years ** CVD includes myocardial infarction (0.05%), angina (0.3%), coronary artery disease (5.15%), cardiac arrhythmias 2.0% (including tachycardia, atrial fibrillation), congestive heart failure (0.7%) ┼ Weighted values represent the sample adjusted to represent national visit characteristics   Female, Asthma, β-blocker 1.00 reference * Effect modifiers are gender, asthma and β-blockers ** Effect measure modification was calculated by the difference between the odds in the CVD group and the non-CVD comparison group, on a percent-scale