Author + information
- Received February 28, 1996
- Revision received August 3, 1996
- Accepted September 20, 1996
- Published online April 1, 1997.
- Marianne Laouri, PhDA,* (, )
- Richard L Kravitz, MD, MSPHB,
- William J French, MD, FACCC,
- Irene Yang, BAD,
- Jeffrey C Milliken, MD, FACCE,
- Lee Hilborne, MD, MPHF,
- Robin Wachsner, MD, FACCG and
- Robert H Brook, MD, ScDH
- ↵*Dr. Marianne Laouri, Quality Measurement and Research, PacifiCare Health Systems, 5995 Plaza Drive, MS 1330, Cypress, California 90630.
Objectives. Our main objective was to apply a new method to determine whether coronary revascularization procedures are underused, especially among African-Americans and uninsured patients.
Background. Although overuse of revascularization procedures has been studied, underuse as defined clinically has not been examined before.
Methods. The study was conducted at four public and two academically affiliated private hospitals in Los Angeles; 671 patients who underwent coronary angiography between June 1, 1990 and September 30, 1991 and met explicit clinical criteria for coronary revascularization were included. The main outcome measure was the proportion of patients undergoing an indicated procedure within 12 months (ascertained by medical record review and confirmed with a telephone survey). Adjusted relative odds of undergoing an indicated procedure for African-Americans and patients in public hospitals compared with whites and patients in private hospitals were calculated.
Results. Overall, 75% of patients underwent a revascularization procedure. Of 424 patients requiring bypass surgery, 107 angioplasty and 140 either bypass surgery or angioplasty, 59%, 66% and 75% underwent the procedure, respectively. African-Americans were less likely than whites to undergo operation (adjusted odds ratio [OR] 0.49, p < 0.05) and angioplasty (adjusted OR 0.20, p < 0.05). Patients in public hospitals were less likely than those in private hospitals to undergo angioplasty (adjusted OR 0.10, p < 0.005).
Conclusions. Underuse of coronary revascularization procedures is measurable and occurs to a significant degree even among insured patients attending private hospitals. Underuse is especially pronounced among African-Americans and patients attending public hospitals. Future cost-containment efforts must incorporate safeguards against underuse of necessary care.
(J Am Coll Cardiol 1997;29:891–7)
© 1997 by the American College of Cardiology
Variation in the use of medical and surgical procedures has been documented extensively. Recent studies focusing on access to cardiac services and use of coronary artery bypass graft surgery (CABG) and percutaneous transluminal coronary angioplasty (PTCA) have demonstrated that African-Americans, Latinos and poor and uninsured patients undergo fewer coronary artery revascularization procedures than white and insured patients ([1–7]). The difference in revascularization rates may be due to either overuse of these procedures among more privileged patients or underuse among those groups traditionally considered to have poor access to care.
This study examines possible underuse of coronary revascularization procedures and addresses two main questions: 1) Are coronary revascularization procedures underused? 2) Are there groups of patients in whom underuse is more common?
1.1 Definition of underuse and clinical necessity.
To measure underuse, we used the RAND/UCLA set of explicit clinical criteria for the necessity of coronary revascularization procedures ([8–10]). These criteria were developed by a nine-member multispecialty national panel of expert physicians and surgeons. Panel members represented internal medicine, invasive and noninvasive cardiology and cardiothoracic surgery. The panelists were selected through a national search of nominations from specialty societies and academic medical centers.
Necessitywas defined as follows for the panelists: 1) The procedure is appropriate (i.e., its benefits to the patient outweigh its risks); 2) it would be improper care not to provide this service to a patient; 3) there is a reasonable chance that this procedure will benefit the patient (i.e., the probability of benefit is not small); and 4) the magnitude of the benefit to the patient is not small ([8–10]). Panelists rated 2,004 clinically homogeneous patient scenarios (i.e., indications) first for the appropriateness and later for necessity of CABG and PTCA ([8–10]). The panel process produced 143 necessity indications for CABG, 36 for PTCA and 69 for any revascularization (invasive procedure recommended but exact type left to clinical discretion). Table 1contains examples of necessity indications for each procedure. The panelists’ ratings were based on a comprehensive review of the published reports and their clinical expertise.
Underusewas defined as failure to undergo a necessary revascularization procedure. Using this definition, underuse does not necessarily reflect poor quality care because some patients may have refused procedures that were offered to them.
1.2 Selection of hospitals.
The study was conducted at six teaching hospitals: four public and two large private nonprofit hospitals in Los Angeles. The study hospitals represented geographically diverse parts of Los Angeles; all had cardiology fellowship programs; four had cardiac surgery facilities, and two did not. The number of beds ranged from 400 to 1,400, and the number of coronary angiographies performed in 1 year ranged from 125 to 1,930.
The four public hospitals were selected because, traditionally, public hospitals in California have been the main source of medical care for the state’s uninsured residents ([11, 12]), and we wanted to investigate whether underuse of CABG and PTCA is more prevalent among poor and uninsured patients than among insured patients in private hospitals. The two private teaching hospitals were included to represent a practice standard in hospitals that serve predominantly insured, middle-class patients.
1.3 Sampling of patients and data collection.
To determine whether a procedure was underused, patients with previous medical need for that procedure must be identified first; then it can be determined whether they indeed underwent the necessary procedure. Because all patients who undergo coronary revascularization must first undergo coronary angiography, we identified eligible patients from among ∼9,100 patients who had undergone coronary angiography at a study hospital between January 1, 1990 and September 30, 1991. We reviewed patient logs kept at each hospital’s cardiac catheterization laboratory and randomly sampled 4,226 cardiac catheterization reports. The 4,226 reports were screened to identify patients with coronary artery disease. Coronary artery diseasewas defined as a 50% left main coronary artery lesion or a 70% lesion in at least one other coronary artery. At this stage, we excluded patients without coronary artery disease, patients with history of CABG or heart transplantation and patients referred for open heart surgery for reasons other than coronary artery disease. We also excluded patients with single-vessel coronary artery disease and no symptoms (because such patients, by definition, did not meet revascularization necessity criteria). Trained data collectors abstracted the medical records of the remaining patients to determine whether they met a necessity indication, and if so, whether they underwent CABG or PTCA or were offered revascularization but refused.
Data abstracted from medical records included clinical presentation at the time of the diagnostic catheterization, intensity of medical therapy (i.e., 0, 1 or >1 type of antianginal medications), angiographic findings, left ventricular ejection fraction, presence of other comorbid conditions (e.g., diabetes, hypertension, kidney failure), results of noninvasive diagnostic tests for cardiac ischemia, patient age, gender, ethnicity and performance of a revascularization procedure or evidence that the patient refused revascularization. Surgical risk was calculated on the basis of the presence of comorbidities and age, using the modified Parsonnet Scale ([8, 9, 13]). Abstraction forms were reviewed by physicians to decide whether each case met a necessity indication. We identified 671 patients who needed a necessary revascularization procedure. We also reviewed CABG and PTCA logs at study hospitals to confirm whether each patient underwent revascularization.
To collect additional demographic information and to determine whether patients whose medical records did not indicate that they had undergone revascularization, we attempted to contact by telephone all patients not known to be dead. Interviews were conducted between 14 and 38 months after the coronary angiogram. We followed a protocol that involved using directory assistance and calling nonrespondents at least 15 times over several weeks.
1.4 Confirmation of necessity among candidates for CABG.
Because preliminary analysis demonstrated a high rate of underuse even at the private hospitals, we performed a more detailed clinical review of the 108 patients who needed CABG but received only medical therapy. Two interventional cardiologists reviewed the primary clinical data on the 108 CABG candidates to decide whether according to their clinical judgment these patients had clinical need for CABG. At least one cardiologist rated each of the cases as having a clinical need for CABG; and they agreed about the vast majority of cases (78%).
1.5 Data analysis.
The primary outcome of interest was performance of the necessary revascularization procedure. We gave credit for performance of necessary CABG or PTCA up to 12 months after coronary angiography. In the bivariate analysis, we compared the proportion of patients in different subgroups who underwent clinically necessary revascularization procedures; results were considered significant if the 95% confidence interval for the difference between groups did not include 0. The independent associations between gender, ethnicity and hospital system and receipt of necessary CABG or PTCA were assessed in three separate logistic regression models evaluating 1) receipt of CABG, given that CABG was necessary; 2) receipt of PTCA, given that PTCA was necessary; and 3) receipt of any revascularization, given that revascularization (type unspecified by panel) was necessary. All three multivariate models evaluated the effect of gender, ethnicity and hospital system (private vs. public) while controlling for age, clinical presentation, angiographic findings and ejection fraction, but the exact specifications differed due to differences in sample size and distribution of covariates for each of three groups of patients. For example, the model predicting receipt of CABG included a dummy variable for left main coronary artery disease, but this variable was omitted from the model predicting PTCA because left main disease was not found among patients meeting criteria for PTCA.
The first model assessed receipt of necessary CABG among 424 patients who met necessity indications for CABG. Independent variables in this model included Canadian Cardiovascular Society class III/IV chronic stable angina, unstable angina, recent myocardial infarction, left main disease, left ventricular ejection fraction ≥50%, age 60 to 69 years, age 70 to 79 years, age ≥80 years, female gender, African-American, Latino, Asian/other and coronary angiography performed at a public hospital.
The second model assessed receipt of necessary PTCA among 107 patients who met necessity indications for PTCA. The independent variables in this model included unstable angina, recent myocardial infarction, left ventricular ejection fraction ≥50%, age 60 to 69 years, age 70 to 79 years, female gender, African-American, Latino and coronary angiography performed at a public hospital.
The third model included 140 patients who met necessity indications for any revascularization. Receipt of a revascularization procedure was the outcome variable for this logistic regression model. Independent variables included unstable angina, recent myocardial infarction, left ventricular ejection fraction ≥50%, age 60 to 69 years, age 70 to 79 years, female gender, African-American, Latino and coronary angiography performed at a public hospital.
2.1 Patient characteristics.
The mean age of patients who met necessity indications for CABG or PTCA was 60 years: 32% were female, 55% white, 21% Latino and 12% African-American (Table 2). Of 601 patients not known to be dead, 374 completed a telephone interview (net response rate 62%). At the time of coronary angiography, ∼36% had chronic stable angina, 40% had unstable angina, and 23% had an acute myocardial infarction within 21 days of angiography. Two-thirds of the patients had left main or triple-vessel coronary artery disease (Table 2). Of the 671 patients, 424 met necessity indications for CABG, 107 for PTCA and 140 for any revascularization (type unspecified).
2.2 Receipt of necessary CABG or PTCA.
2.2.1 Unadjusted results.
Overall, 506 patients (75%) underwent either CABG or PTCA within 1 year; of these, 85% underwent the procedure preferred by the expert panel, whereas 15% underwent an alternative procedure (i.e., PTCA when CABG was indicated or CABG when PTCA was indicated). Patients needing an unspecified revascularization procedure (n = 140) were almost evenly split between CABG (39%) and PTCA (40%), and 21% were treated medically (Table 3). After exclusion of 16 patients who refused operation (10 identified through medical record review, 6 through the interview), the overall revascularization receipt rate increased from 75% to 77% (data not in table).
On bivariate analysis, among all three groups of patients, no significant associations were seen between receipt of necessary CABG or PTCA and gender, ethnicity or clinical presentation at the time of angiography (Table 4). Patients with two-vessel disease were significantly less likely than those with left main disease to undergo necessary CABG (Table 4). Left ventricular ejection fraction <50% was significantly associated with receipt of an invasive procedure among patients needing any revascularization (type unspecified). Patients who underwent coronary angiography at a public hospital were less likely than those treated at a private hospital to undergo necessary PTCA (p < 0.05) (Table 4). Rates of receipt of necessary CABG were similar at private and public hospitals (60% vs. 58%) (Table 4).
2.3 Adjusted results.
Controlling for clinical covariates using logistic regression, we found that left main disease was strongly associated with receipt of necessary CABG (adjusted odds ratio [OR] 3.84, 95% confidence interval [CI], 2.13 to 6.92, p < 0.0001) (Table 5). No differences were observed between men and women or between patients who underwent coronary angiography at a private or public hospital. However, African-Americans were less likely than whites to undergo necessary CABG (adjusted OR 0.49, 95% CI, 0.23 to 0.99, p < 0.05).
With respect to PTCA, we found that clinical presentation with unstable angina or a recent myocardial infarction and ejection fraction ≥50% were strongly associated with receipt of necessary PTCA (Table 5). Patients 70 to 79 years old were less likely than those <60 years old to undergo necessary PTCA. No differences were found between men and women. African-American patients were less likely than whites to undergo necessary PTCA (adjusted OR 0.20, 95% CI 0.06 to 0.72, p < 0.05). Patients who underwent coronary angiography at a public hospital were less likely than those at private hospitals to undergo necessary PTCA (adjusted OR 0.10, 95% CI 0.02 to 0.44, p < 0.005).
In the logistic regression model examining the effects of patient gender, ethnicity and hospital system on receipt of CABG or PTCA among patients who met necessity indications for any revascularization, no statistically significant differences were seen, except that patients with a left ventricular ejection fraction ≥50% were significantly more likely to undergo a necessary procedure than those with impaired left ventricular function.
Although methods to assess the appropriateness of care have been used to identify and measure overuse of coronary artery revascularization, to our knowledge, the present study is the first to demonstrate that these methods can be used to measure underuse of this costly but potentially beneficial pair of procedures. In our study of 671 patients who met explicit indications for the necessity of coronary artery revascularization, approximately one-fourth were treated medically, and more than one-third either did not undergo a revascularization procedure or underwent one that was presumably suboptimal. The validity of the results depends on the soundness and applicability of the 248 necessity indications and the integrity of the process used to develop them. As described elsewhere ([8–10]), the indications are the product of the careful deliberations of nine experts, guided by a comprehensive review of published reports. They were judged reasonable by cardiologists at each of the six participating hospitals, and external evidence for their validity was demonstrated in a related study that concluded that failure to undergo necessary revascularization was correlated with an excess risk of death ().
Although the overuse of CABG ([15, 16]) and PTCA () has been demonstrated in previous studies, our finding that necessary revascularization procedures are not always provided means that clinically meaningful underuse and overuse of the same procedure may occur simultaneously. Underuse of revascularization procedures was not confined to the poor and uninsured. Clinically significant underuse was observed among insured patients at the two private hospitals. This finding suggests that elimination of financial barriers to care will not avert underuse of CABG and PTCA without attention to factors influencing clinical decision making and physician–patient communication.
The provision of necessary revascularization procedures was strongly driven by clinical factors. As expected, clinical severity of illness was the most important influence on whether revascularization was performed. For example, left main coronary artery disease was the strongest factor associated with receipt of necessary CABG. Similarly, clinical presentation with an acute syndrome (unstable angina or chest pain after an acute myocardial infarction) was strongly associated with receipt of necessary PTCA.
Nevertheless, nonclinical factors were also associated with underuse of CABG and PTCA. Patients in public hospitals were less likely to undergo PTCA (but not CABG) than those in private hospitals. African-Americans were less likely than whites to undergo both necessary CABG and PTCA. This latter finding not only corroborates results of previous studies ([1, 2, 4, 5, 7]) but shows that CABG and PTCA are underused among African-Americans relative to whites even after controlling for important clinical factors, including objective anatomic data. Refusal of recommended procedures, at least as documented in the medical record or reported by telephone survey responders, did not account for the racial disparities. More research is needed to understand and correct these apparent inequities, which may reflect real but unmeasured differences in patient preferences, bias in physician judgment or cultural barriers to physician–patient communication.
We did not observe differences in the receipt of necessary CABG or PTCA between men and women. The absence of detectable gender differences is consistent with results of previous research by Tobin et al. (), Maynard et al. () and Krumholtz et al. () showing that the gender differences commonly observed in studies of noninvasive cardiac procedures were not observed once the coronary anatomy was known, and revascularization rates did not differ between men and women.
For the 85% of Americans with health insurance and their employers, health care costs are of paramount concern (). As cost-containment pressures escalate, it is important that safeguards be established to promote the provision of highly beneficial care. The present study shows that even in settings where most patients are insured and relatively well off, useful procedures may not be provided. The methods described herein represent one approach to the measurement and prevention of underuse of health care services. If criteria for necessity of revascularization were incorporated into standard teaching and placed on hospital computers, patients needing revascularization would be identified more readily. Clinicians could be supported and encouraged to offer necessary procedures to their patients or, alternatively, to provide explicit clinical justification for not doing so in the medical record. In this way physicians and hospitals could monitor their own practice, and outside agencies could assure the public that cost-cutting measures were not a threat to health.
3.1 Study limitations.
The major limitations of this study include the manner of patient selection and limited generalizability to other procedures and geographic regions. Because the sample was drawn from patients who had already undergone coronary angiography, underuse of CABG and PTCA may be higher among persons who encounter difficulties accessing the health care system or obtaining necessary preliminary diagnostic tests. One might expect such difficulties to be most pronounced for Medicaid patients and for uninsured patients using the public hospital system. Evidence for this hypothesis is provided in a related study that showed lower use of necessary coronary angiography in three public hospitals compared with a private university hospital (). In contrast, the results of the current study may overestimate the extent of underuse of CABG and PTCA for Los Angeles County as a whole. Examination of a larger and more diverse sample of nonpublic hospitals might have revealed a lower overall level of underuse. Furthermore, although nonclinical factors, such as organizational characteristics of the hospitals, and the physicians’ training and specialty, may be important for understanding underuse as a function of organizational decision making, such information was not collected in this study. A larger number of hospitals would have allowed for a hospital level analysis of how decisions about revascularization are made for clinically homogenous patient populations.
In recent years, a large portion of health services research has been devoted to cost containment and the elimination of overuse of medical services. The results of the present study demonstrate the feasibility of a method for identifying and measuring underuse of medical services, as well as a clear demonstration that underuse occurs. Given this knowledge, together with the well documented problems in access to care experienced by large groups of the population, further investigation of underuse of medical care is needed to identify vulnerable patients groups and types of care most susceptible to underuse. As we reform our health care system and strive to improve access and quality of care while containing costs, it is important to bear in mind the health and economic implications of underuse and to build in safeguards against both inappropriate overuse and underuse of medical care.
☆ This study was supported by Grant 91-8976 from the Kaiser Family Foundation, Menlo Park, California. It was presented in part at the 43rd Annual Scientific Session of the American College of Cardiology, Atlanta, Georgia, March 1994.
- coronary artery bypass graft surgery
- confidence interval
- odds ratio
- percutaneous transluminal coronary angioplasty
- Received February 28, 1996.
- Revision received August 3, 1996.
- Accepted September 20, 1996.
- The American College of Cardiology
- Valdez RB,
- Dallek G
- Leape LL,
- Hilborne LH,
- Kahan JP,
- et al.
- Hilborne LH,
- Leape LL,
- Kahan JP,
- Park RE,
- Kamberg CJ,
- Brook RH
- California Legislative Assembly. AB799, S2012. 1982.
- Parsonnet V,
- Dean D,
- Bernstein A
- Krumholz HM,
- Douglas PS,
- Lauer MS,
- Pasternak RC
- Blendon RJ,
- Marttila J,
- Benson JM,
- Shelter MC,
- Connolly FJ,
- Kiley T
- ↵Laouri M, Kravitz RL, Bernstein SJ, et al. Underuse of coronary angiography: application of a clinical method. Int J Quality Health Care. In press.