Author + information
- Received October 13, 2003
- Revision received December 8, 2003
- Accepted December 15, 2003
- Published online May 5, 2004.
- P.Michael Ho, MD*,* (, )
- Frederick A Masoudi, MD, MSPH, FACC‡,
- Eric D Peterson, MD, MPH, FACC§,
- Gary K Grunwald, PhD∥,
- Anne E Sales, PhD¶,
- Karl E Hammermeister, MD† and
- John S Rumsfeld, MD, PhD, FACC†
- ↵*Reprint requests and correspondence:
Dr. P. Michael Ho, 1055 Clermont Street (111B), Denver, Colorado 80220, USA.
Objectives The goal of this study was to determine if cardiology subspecialty involvement improves the attainment of recommended low-density lipoprotein (LDL) cholesterol and blood pressure (BP) targets in coronary artery disease (CAD) patients.
Background The impact of physician specialty on secondary prevention measures for CAD in ambulatory care is unknown.
Methods This was a retrospective cohort study of 13,995 patients with CAD seen at eight ambulatory care Veteran Affairs facilities from 1998 to 2000. Patients with cardiology involvement were defined as those seen in cardiology clinic in addition to primary care. The main outcomes of interest were LDL cholesterol ≤100 mg/dl and BP ≤130/85 mm Hg. Multivariable hierarchical regression analyses were used to determine the independent association of cardiology involvement with improved LDL cholesterol and BP control.
Results Overall, 3,771 (27.0%) patients had cardiology involvement. A higher proportion of patients with cardiology involvement achieved LDL cholesterol (55.6% vs. 45.6%; p < 0.01) and BP (45.3% vs. 35.9%; p < 0.01) goals. In multivariable hierarchical regression analysis, cardiology involvement was independently associated with better LDL cholesterol (odds ratio [OR], 1.59; 95% confidence interval [CI], 1.40 to 1.82) and BP (OR, 1.52; 95% CI, 1.32 to 1.77) control. The benefit of cardiology involvement was consistent across a range of LDL and BP targets, in analysis of LDL and BP as continuous outcomes, and among subgroups of high-risk patients, including diabetic patients, the elderly, and those with prior revascularization.
Conclusions Cardiology involvement is associated with better LDL cholesterol and BP control among CAD patients. However, significant room for improvement in secondary prevention measures remains, irrespective of physician specialty.
In order to reduce the morbidity and mortality of the 12 million patients in the U.S. with coronary artery disease (CAD), aggressive secondary prevention measures, including low-density lipoprotein (LDL) cholesterol reduction and blood pressure (BP) control have been strongly recommended by multiple national guidelines (1–4). The American College of Cardiology/American Heart Association guidelines have established targets for treatment of LDL cholesterol at ≤100 mg/dl and BP at ≤130/85 mm Hg for patients with established CAD (4). Despite the strong evidence supporting these recommendations and the widespread dissemination of guidelines, achievement of these targets have been suboptimal (5–9).
Prior studies have found that some measures of quality of care and outcomes for patients with cardiovascular disease may differ by the specialty of the caring physician (10–16). Specifically, patients hospitalized with acute myocardial infarction (AMI) who were treated by cardiologists were more likely to receive thrombolytics, aspirin, and beta-blockers and to have lower mortality rates compared with those treated by generalist physicians. Additionally, Ayanian et al. (14)recently found that cardiology specialty of ambulatory care physicians was associated with decreased two-year mortality among elderly patients after a myocardial infarction. One possible explanation for this observed survival advantage with subspecialty care is that cardiologists are more aggressive in treating risk factors. These risk factors, including hyperlipidemia, hypertension, diabetes, and smoking are present in over 80% of CAD patients (17). To our knowledge, the association between cardiology specialty management and guideline adherence for secondary prevention measures has not been studied. Accordingly, the objective of this study was to determine if cardiology subspecialty involvement in ambulatory chronic care of CAD patients is associated with improved LDL cholesterol and BP control in a multi-center cohort of CAD patients.
All active patients with CAD enrolled in primary care clinics in any one of eight Pacific Northwest Veterans Affairs (VA) hospital facilities were included in the study. Patients with CAD were identified by any one or more of the following International Classification of Diseases-9th revision (ICD-9) diagnosis codes in the 24 months before the index date: 410.x, 411.x, 412, or 414.x. We searched for these diagnosis codes in the active problem list file of the VA computerized medical record, the outpatient care (OPC) file, and the patient treatment file (PTF) (6). The OPC file contains data on all outpatient visits, whereas the PTF file contains data on all in-patient encounters. These files are part of the Consumer Health Information & Performance Sets (CHIPS) Data Warehouse, which is a relational database mirroring the clinical information system residing at each VA Medical Center and outpatient facility (18).
Active patients were defined as being alive on October 1, 2000 (index date), and having at least one primary care clinic visit documented in the OPC file per year in each of the two previous years. A primary care visit was defined using the OPC file as any visit to one or more of the following clinics: General Internal Medicine, Women's Clinic, Primary Care/Medicine, or Geriatrics. Patients were defined as having cardiology involvement if they had at least one cardiology clinic visit anytime during the 24 months before the index date.
Clinical characteristics were defined by ICD-9 codes within either the OPC and/or PTF files within the 24 months before the index date: percutaneous coronary intervention (PCI), coronary artery bypass graft (CABG) surgery, diabetes, hypertension, hyperlipidemia, chronic obstructive pulmonary disease, peripheral vascular disease, cerebrovascular disease, congestive heart failure, depression, and renal disease.
Recent hospital admission for an acute coronary syndrome (ACS) and recent coronary revascularization procedure(s) (i.e., PCI and/or CABG) were defined as occurring within the 15 months before the index date. Hospital admission for ACS was defined by a primary discharge diagnosis of ICD-9 codes 410.x or 411.x in the PTF file. The number of outpatient clinic visits was aggregated over the 15 months before the index date.
Patients were defined as having a current medication prescription for the following classes of medications if they had a prescription written or renewed within the last 15 months before the index date: beta-blockers, angiotensin-converting enzyme (ACE) inhibitors or angiotensin receptor blockers, calcium channel blockers, diuretics, and 3-hydroxy-3-methylglutaryl coenzyme-A reductase inhibitors (statins).
Facility level data were obtained from the Veterans Integrated Service Network Support Service Center website, a health care information and technical support organization serving the needs of the networks (19). Data on volume of primary care patients and visits were obtained for each of the eight facilities included in the analysis.
The two primary outcomes of interest were achievement of: 1) LDL ≤100 mg/dl; and 2) BP ≤130/85 mm Hg, consistent with national guidelines as treatment targets for patients with underlying CAD (4). Any lipid level measured within 15 months before the index date was included, and the most current LDL level was used in the analyses. An LDL measurement was available for 83.6% of the patients. All patients had at least one BP measurement within the six months before the index date. Among the 68.8% of patients with two or more BP measurements, the BP readings were averaged.
We compared baseline patient characteristics and processes of care (e.g., number of clinic visits and prescribed medications) between patients with cardiology involvement and patients without cardiology involvement, using ttests for continuous variables and the chi-square tests for categorical variables. Next, we evaluated achievement of secondary prevention goals between patients with and without cardiology involvement using two different LDL cholesterol (i.e., LDL ≤100 mg/dl and ≤130 mg/dl) and BP (i.e., ≤130/85 mm Hg and ≤140/90 mm Hg) treatment targets. Then, we compared the mean LDL cholesterol, mean systolic BP, and mean diastolic BP between the two groups.
In order to determine if cardiology involvement was independently associated with improved LDL and BP control compared with no cardiology involvement, we first constructed a series of patient-level multivariable logistic regression models. Variables that were significantly different between the cardiology involvement and the no cardiology involvement groups (p ≤ 0.10) were candidates for the risk models. A baseline risk model was constructed consisting of patient demographics, medical history, number of days from the last LDL or BP measurement, recent ACS event or coronary revascularization procedures, and the number of clinic visits. These outpatient visits were also analyzed separately as primary care clinic visits only, cardiology clinic visits only, and primary care plus cardiology clinic visits. In addition, we fitted a separate logistic regression model to estimate the propensity that a patient would have a cardiology clinic visit and included the propensity score as an independent predictor variable in the baseline risk model (20). Then, the cardiology involvement variable was added to the risk model to determine its independent association with the outcome. Separate risk models were constructed for achievement of LDL ≤100 mg/dl and BP ≤130/85 mm Hg. To assess whether there was a “dose-response” association between the number of cardiology clinic visits and LDL or BP control, we fitted logistic regression models in which the number of cardiology clinic visits was categorized into quartiles.
Next, to account for the clustering of patients within providers and facilities, hierarchical regression models were constructed using NLMIXED in SAS (Version 8.02, SAS Institute, Cary, North Carolina). The cardiology involvement variable was entered into the model, adjusting for patient-level risk factors (i.e., all patient-level variables associated with either of the outcomes in the baseline risk models) and clustering of patients within providers and facilities to determine its independent association with LDL and BP control. Subsequent models also adjusted for other facility characteristics such as tertiary versus primary care facility and primary care clinic volume as defined by both the number of clinic visits and the number of patients enrolled.
We performed sensitivity analyses to determine if the observed relationship between cardiology involvement and improved LDL and BP control was significant when different definitions for the outcome variables were applied. Using similar multivariable regression modeling techniques, we redefined the outcomes for LDL cholesterol and BP as ≤130 mg/dl and ≤140/90 mm Hg, respectively, and evaluated the independent association between cardiology involvement and achievement of these new targets. Next, we analyzed LDL cholesterol, systolic BP, and diastolic BP as continuous outcome variables in multivariable linear regression models, adjusting for patient co-morbidities and processes of care. Finally, to determine if the benefit of cardiology involvement was consistent across different high-risk subgroups, we evaluated this relationship in older patients (age ≥65 years), diabetics, and those with prior coronary revascularization (i.e., PCI or CABG). All analyses were performed using SAS (Version 8.02, SAS Institute). The study was approved by the Colorado Multiple Institutional Review Board.
Table 1compares the baseline characteristics of patients with and without cardiology involvement. Of 13,955 active CAD patients in the Pacific Northwest VA facilities, 3,771 (27.0%) patients had at least one cardiology clinic visit and 2,177 (15.6%) had at least two or more visits. Patients with cardiology involvement were younger, but a greater proportion had cardiac co-morbidities including hypertension, diabetes, hyperlipidemia, prior myocardial infarction, and prior coronary revascularization. Cardiology involvement patients also had a higher burden of non-cardiac co-morbidities including chronic obstructive pulmonary disease and depression. A greater proportion of cardiology involvement patients experienced a recent ACS event or coronary revascularization procedure, and they also had more overall clinic visits.
In unadjusted analyses, patients with cardiology involvement were more likely to achieve LDL cholesterol and BP targets as recommended by national guidelines (Fig. 1). Furthermore, patients with cardiology involvement had lower mean systolic BPs (134.2 ± 17.4 mm Hg vs. 138.9 ± 18.6 mm Hg; p <0.01), lower mean diastolic BPs (71.2 ± 10.1 mm Hg vs. 74.1 ± 10.7 mm Hg; p <0.01), and lower mean LDL cholesterol (99.8 ± 32.0 mg/dl vs. 107.8 ± 34.7 mg/dl; p <0.01) compared with patients without cardiology involvement. This observed benefit with cardiology involvement remained consistent among subgroups of older patients, diabetics, and those with prior coronary revascularization (Fig. 2). Patients with cardiology involvement were also more likely to be prescribed beta-blockers, ACE inhibitors, or angiotensin receptor blockers and diuretics (Fig. 3). Furthermore, there was a dose-response relationship between the number of cardiology clinic visits and the odds of having LDL or BP treated to guideline recommendations.
In multivariable hierarchical regression models, cardiology involvement was independently associated with better LDL control (odds ratio [OR], 1.59; 95% confidence interval [CI], 1.40 to 1.82) and BP control (OR, 1.52; 95% CI, 1.32 to 1.77) after adjusting for patient demographics, medical history, number of clinic visits, recent ACS or coronary revascularization procedure, number of days from last LDL or BP measurement, propensity to have a cardiology visit, and clustering of patients within providers and facilities. The magnitude of benefit for cardiology involvement remained unchanged in subsequent risk models that also adjusted for tertiary care facility and primary care clinic volume and when less stringent guideline targets were used to define the outcomes (i.e., ≤130 mg/dl for LDL cholesterol and ≤140/90 mm Hg for BP).
When the outcomes were analyzed as continuous variables in multivariable linear regression analyses, cardiology involvement was associated with lower systolic BP (−2.70 ± 0.38 mm Hg; p <0. 01), lower diastolic BP (−1.97 ± 0.23 mm Hg; p <0. 01), and lower LDL cholesterol (−5.82 ± 0.74 mg/dl; p <0.01) even after adjustment for patient characteristics and processes of care.
The primary goal of this study was to determine if cardiology subspecialty involvement in ambulatory care improves secondary prevention guideline concordance, specifically LDL cholesterol and BP control, in CAD patients. We found that cardiology involvement (i.e., patients seen in cardiology in addition to primary care) was independently associated with improved LDL and BP control. This observed benefit with cardiology involvement remained significant when different targets for LDL and BP were used, when LDL and BP were analyzed as continuous outcome variables, and among subgroups of high-risk patients.
Based on epidemiologic studies, even small decreases in LDL and BP levels translate into significant reductions in morbidity and mortality, with a linear relationship between serum cholesterol or BP and subsequent cardiovascular events (21–25). Law et al. (21)estimated that a 10% decrease in the serum cholesterol concentration was associated with a reduction in cardiovascular events of at least 19% and as high as 54%, depending on the patient population. Van den Hoogen et al. (24)estimated that the risk of death due to CAD was 1.17 per 10 mm Hg increase in systolic BP and 1.13 per 5 mm Hg increase in diastolic BP. The lower levels of LDL cholesterol (−8.0 mg/dl or 7% lower), systolic BP (−4.7 mm Hg), and diastolic BP (−2.9 mm Hg) associated with cardiology involvement in this study would, thus, translate into significant reductions in morbidity and mortality if extrapolated on a population-wide basis.
Recent studies suggest that hyperlipidemia and hypertension are prevalent conditions among patients with CAD. Khot et al. (17)demonstrated that 80% to 90% of patients with CAD have at least one of the four conventional cardiac risk factors (i.e., hypertension, hyperlipidemia, smoking, and diabetes) with hyperlipidemia and hypertension being the two most common risk factors. Furthermore, Greenland et al. (26)noted that at least one of these risk factors was present in 87% to 100% of patients who subsequently have a fatal cardiovascular event. These studies emphasize the urgent need to identify and aggressively treat these risk factors, especially among patients with preexisting CAD who have a five-fold to sevenfold increased risk of recurrent myocardial infarction compared with the general population (27).
Prior studies evaluating the association between physician specialty and outcomes of patient's with CAD have focused mainly on the hospital setting. Among patients with AMI, cardiologists were more likely to prescribe guideline-indicated therapies, such as thrombolytics, aspirin, and beta-blockers and to refer patients for revascularization (10–13). In addition, treatment by a cardiologist during hospitalization for AMI has been associated with lower in-hospital and one-year mortality (28,14). Although AMI patients are an important component of the CAD population, the majority of CAD patients receive care in the ambulatory care setting, where less is known about the association between physician specialty and outcomes of care.
This study suggests that cardiology subspecialty involvement can have an impact on chronic ambulatory care in addition to hospital care. A prior single-centered study found improved guideline compliance with LDL goals among patients with cardiology involvement (29). In addition, ambulatory visits to cardiologists have previously been associated with greater use of cardiac procedures and decreased mortality after myocardial infarction (14). Although the greater use of cardiac procedures and revascularization may explain, in part, the mortality difference, other factors such as LDL and BP control are as important in reducing recurrent cardiac events. In this study, CAD patients managed with cardiology involvement were more likely to have their LDL cholesterol and BP treated to levels recommended by national guidelines despite having more cardiac and non-cardiac co-morbidities. Furthermore, we found that patients with cardiology involvement were more likely to receive beta-blockers and ACE inhibitors, therapies that have demonstrated survival benefits for CAD patients (30,31).
Our findings suggest that a collaborative approach between cardiologists and primary care providers may improve outcomes for patients with CAD. Other studies evaluating a collaborative approach involving cardiologists have also demonstrated better guideline concordance and outcomes for conditions such as heart failure and dyslipidemia (32–34). For example, Ryan et al. (33)demonstrated better dyslipidemia care through a multidisciplinary effort lead by a cardiologist with assistance from an advanced practice nurse. Patients underwent an intensive assessment and treatment program with frequently scheduled visits, suggesting that intensity of therapy may be an important component for improving risk factor control. Future studies should attempt to identify the potential mechanisms whereby specialty care results in improved outcomes. Such factors could be studied prospectively to determine if implementation results in improved morbidity, mortality, and quality of life outcomes for patients with a wide spectrum of cardiovascular diseases.
There are several potential explanations for the higher proportion of cardiology involvement for patients achieving guideline recommendations for LDL cholesterol and BP. Cardiologists may have more up-to-date and detailed knowledge of the evidence supporting certain therapies. In a survey of cardiologists, internists, and family practitioners about four treatments demonstrated to have improved survival after AMI, cardiologists believed more strongly in the results and were more likely to prescribe thrombolytics, aspirin, and beta-blockers (35).
In addition, cardiologists have a narrower focus in their clinical practice and can defer the care of non-cardiovascular conditions to primary care providers. In our cohort, patients had an average of three co-morbidities in addition to CAD and were prescribed an average of 11 medications. Primary care providers are, thus, faced with the management of a larger number of conditions, where they have to weigh the risks and benefits of prescribing additional medications or adjusting current medications with the potential for side effects that occur with polypharmacy or higher doses of medications (36). Furthermore, primary care physicians often experience “tyranny of the urgent,” where the acute symptoms and concerns of patients crowd out the less urgent need to bring chronic illness (e.g., hypertension and hyperlipidemia) under optimal management (37).
Although a larger proportion of patients with cardiology involvement met secondary prevention guideline recommendations, overall rates of LDL cholesterol and BP control were low regardless of specialty involvement in this study, 56.8% and 38.5%, respectively. Prior studies evaluating treatment of LDL cholesterol in secondary prevention have also noted low rates of guideline concordance (5,6). Although studies on hypertension treatment patterns have not focused specifically on CAD patients who have stricter treatment goals, these studies have also found similarly low rates of guideline adherence (7–9). Together, these studies highlight an urgent need to improve secondary prevention practice patterns because patients with preexisting CAD have a higher risk for recurrent cardiac events.
There are several potential limitations to this study. Although the database included complete laboratory, pharmacy records, and visit histories from all Pacific Northwest VA facilities, information from outside the VA system were not available. However, given an average of eight clinic visits per patient during the study period, it is likely that we have captured a significant proportion of all health care utilization by these patients. Also, we defined the measurement period for LDL cholesterol as the 15 months before the index date, but patients could have been referred to cardiology after this time period, resulting in misclassification bias. This misclassification, however, would tend to bias the results towards the null. Next, we did not have reliable data on medication doses. One possible explanation for the finding of the study is that cardiology involvement patients were prescribed higher doses of medications. For example, although a similar proportion of patients with and without cardiology involvement received statin drugs, a higher proportion of cardiology involvement patients achieved LDL ≤100 mg/dl. Future studies should evaluate whether intensity of drug therapy is a principal mechanism for the differential outcomes.
Next, this was an observational study with significant differences between patients with and without cardiology involvement. The cardiology involvement patients were sicker at baseline with a heavier burden of co-morbidities. However, this disproportionate disease burden would bias the results against cardiology involvement patients, and, if anything, we may have underestimated the magnitude of association between cardiology involvement and improved secondary prevention measures. Finally, although LDL cholesterol and BP levels are surrogate outcomes, they are strongly linked to cardiovascular morbidity and mortality. Large population-based studies have demonstrated significant declines in cardiovascular events with even small reductions in cholesterol or BP (21,24).
In conclusion, we found that cardiology involvement in ambulatory care is associated with improved lipid and hypertension control among CAD patients. However, significant opportunity for improvement in the treatment of secondary prevention measures remains for CAD patients, irrespective of physician specialty. A collaborative approach between cardiologists and primary care providers may improve outcomes for CAD patients. Future investigations should address specific mechanisms, such as intensity of therapy that may explain the differences in outcomes noted in this study.
☆ Dr. Ho is supported by a NIH NRSA award. Dr. Masoudi is supported by a NIH Research Career Award K08-AG01011. Dr. Rumsfeld is supported by a VA HSR Advanced Research Career Development Award.
- angiotensin-converting enzyme
- acute coronary syndrome
- acute myocardial infarction
- blood pressure
- coronary artery bypass graft
- coronary artery disease
- confidence interval
- International Classification of Diseases- 9th revision
- low-density lipoprotein
- outpatient care file
- odds ratio
- percutaneous coronary intervention
- patient treatment file
- Veterans Affairs
- Received October 13, 2003.
- Revision received December 8, 2003.
- Accepted December 15, 2003.
- American College of Cardiology Foundation
- American Heart Association
- Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults
- Gibbons R.J,
- Chatterjee K,
- Daley J,
- et al.
- Schreiber T.L,
- Elkhatib A,
- Grines C.L,
- O'Neill W.W
- Ansari M,
- Alexander M,
- Tutar A,
- Bello D,
- Massie B.M
- ↵VISN 20 Data Warehouse. Available at: http://vaww.visn20.med.va.gov/visn20/CHIPS/index.html. Accessed on March 19, 2004.
- ↵VISN Support Service Center. Available at: http://klfmenu.med.va.gov-.url. Accessed on March 19, 2004.
- Law M.R,
- Walkd N.J,
- Thompson S.G
- National Cholesterol Education Program
- Casale P.N,
- Jones J.L,
- Wolf F.E,
- Pei Y,
- Eby L.M