“Renalism”: Inappropriately Low Rates of Coronary Angiography in Elderly Individuals with Renal Insufficiency : Journal of the American Society of Nephrology

Journal Logo

Advanced Search
Epidemiology and Outcomes

“Renalism”

Inappropriately Low Rates of Coronary Angiography in Elderly Individuals with Renal Insufficiency

Chertow, Glenn M.*; Normand, Sharon-Lise T.†,‡; McNeil, Barbara J.‡

Author Information

*Division of Nephrology, Departments of Medicine, Epidemiology and Biostatistics, University of California San Francisco, San Francisco, California; †Department of Health Care Policy, Harvard Medical School, Boston, Massachusetts; and ‡Department of Biostatistics, Harvard School of Public Health, Boston, Massachusetts

Correspondence to Dr. Glenn M. Chertow, Department of Medicine Research, University of California San Francisco, UCSF Laurel Heights Suite 430, 3333 California Street, San Francisco, CA 94118-1211. Phone: 415-476-2173; Fax: 415-476-9531; E-mail: [email protected]

Accepted June 02, 2004

Received March 04, 2002

Journal of the American Society of Nephrology 15(9):p 2462-2468, September 2004. | DOI: 10.1097/01.ASN.0000135969.33773.0B
  • Free

Abstract

Variation in the rates of major procedures such as coronary angiography has been well described (1,2). Initial studies by Wennberg et al. (3–5) highlighted regional variations in appendectomy and tonsillectomy rates, noting that variation was more likely when there was uncertainty in the optimal therapeutic approach and when several alternative treatment strategies were available. More recently, attention has been paid to inequities in procedure use by gender and race. Women and black individuals are less likely to undergo coronary angiography and bypass surgery (6–8) than men and white individuals. Black individuals are less likely to receive thrombolytic therapy after myocardial infarction (MI) (9), more likely to have amputation rather than revascularization surgery for peripheral vascular disease (10), and less likely to be referred for kidney transplantation (11,12). Older individuals may also receive less aggressive therapy for several conditions, prompting consideration of “ageism” (13,14). Although we cannot absolutely determine whether these inequities reflect underuse in women, black individuals, and the elderly or overuse in men, white individuals, and younger individuals, many have expressed concern that variation in procedure rates indicates bias that should be eliminated in the interest of optimizing medical care.

Fewer studies have examined how cardiac procedure rates vary by the presence or absence of coexisting medical conditions (15–17). In general, despite that higher risk patients tend to receive greater relative benefit from a variety of interventions in cardiovascular diseases, higher risk patients (including the elderly) tend to be treated more conservatively, perhaps because of their own or their physicians’ aversion to risk. The presence of one such condition, chronic kidney disease (CKD), influences the use of angiography as a result, in part, of the fear of inducing nephrotoxicity caused by radiocontrast media. We hypothesized that the presence of CKD in elderly patients would be associated with relative underuse of coronary angiography after MI (a bias we have termed “renalism”). Moreover, we hypothesized that a strategy that includes coronary angiography for patients with CKD would be associated with improved survival. Clinical decisions in the CKD population would then require balancing the potential benefit of an angiography-based strategy against the risks and consequences of radiocontrast-associated nephropathy combined with a more conservative medical approach.

Materials and Methods

Overview

Our approach involved creation of patient groups on the basis of the presence of CKD and the provision of coronary angiography. We then grouped CKD and non-CKD patients into strata on the basis of published criteria for the appropriateness of coronary angiography to evaluate whether the large observed differences in angiography rates between CKD and non-CKD groups could be explained by differences in clinical characteristics. Because CKD patients were not randomized to coronary angiography, we adopted a propensity score approach to adjust for observed confounders using data from medical records and administrative files. This approach provided us with the most accurate estimate of the relative effect of coronary angiography in the CKD population.

Cohort Definition

The Cooperative Cardiovascular Project (CCP) was a comprehensive study of treatment patterns and outcomes of acute MI in the elderly funded by the Centers for Medicare and Medicaid Services (CMS). The CCP included patients who were discharged with a diagnosis of acute MI (International Classification of Diseases, Ninth Revision code 410.xx, excluding a 2 in the fifth position) from >1600 hospitals located in California, Florida, Massachusetts, New York, Ohio, Pennsylvania, and Texas from February 1994 through July 1995. These states were selected because they were known to differ in frequency of cardiac procedure use, are large, and are geographically diverse. Trained nurses and medical records technicians abstracted hospital records, and data were entered directly into a computerized database management system. Abstracted data included information on hospital characteristics and patient demographics; medical history; cardiovascular symptoms and signs; radiographic, electrocardiographic, and echocardiographic findings; and selected laboratory studies, including admission serum creatinine.

The initial frame consisted of 85,743 patients. Patients with the following characteristics were excluded from the study sample: age <65 (n = 5547 [6.4%]) or >89 (n = 4169 [4.9%]); non-410 discharge diagnosis according to CMS claims data (n = 6) or a non-410 discharge diagnosis on a transfer admission (n = 6570 [7.7%]); discharged alive with a total length of stay of <4 d (n = 2381 [2.8%]); residence outside of the United States (n = 16); a reported date of death before admission date (n = 34); transfer admission outside the seven states (n = 342 [0.4%]); transfer more than once (n = 237 [0.3%]); CCP record with data missing for the second hospitalization (n = 3448 [4.0%]); and covered by plans other than fee-for-service (n = 5738 [6.7%]) or type of insurance was unknown (n = 14). Patients whose serum creatinine was ≥5 mg/dl or who were on dialysis (n = 1452 [1.7%]) or whose serum creatinine was missing (n = 3221 [3.8%]) at initial hospital admission also were excluded. These exclusions in sum left a cohort of 57,284 (some patients had more than one of the above exclusion criteria).

CKD was defined as a serum creatinine concentration ≥1.5 mg/dl and <5 mg/dl. Patients with serum creatinine concentrations <1.5 mg/dl were considered to have normal or near-normal renal function. Although we recognize the limitations of this approach, a serum creatinine concentration of 1.5 mg/dl has been used in previous studies that examined the risk of radiocontrast-associated nephropathy (18,19). We conducted companion analyses using an estimated glomerular filtration rate (GFR) <30 ml/min per 1.73 m2 as an alternative definition of CKD (20). We selected the <30 ml/min per 1.73 m2 cutoff (“stage 4 CKD” by National Kidney Foundation Kidney Disease Quality Initiative Clinical Practice Guidelines for CKD: Evaluation, Classification and Stratification) because with the advanced age of the CCP cohort, the majority of CCP patients would have been classified as having CKD if a more typical cutoff (<60 ml/min per 1.73 m2, or “stage 3 CKD”) had been selected.

Angiography and revascularization procedures were defined as having been completed at any time during the initial hospital admission after MI diagnosis. Mortality was defined as death within 1 yr of admission for the MI.

Appropriateness Criteria for Angiography

The RAND approach was used to generate ratings of appropriateness; the original RAND ratings were updated by this research team in October 1995 (16) and have been validated by practicing clinicians (16) and using empirical analyses (21). Among 92 clinical indications related to angiography during the initial hospitalization, the appropriateness score was determined using a combination of the following factors: duration of symptom onset (<6 h, 6 to 12 h, or >12 h); age (<75 or ≥75 yr); eligibility for and receipt of thrombolytic therapy; and the presence of complications, such as persistent or recurrent chest pain, stress-induced ischemia, and pulmonary edema (21). The appropriateness scores ranged from 1 (extremely inappropriate) to 5 (uncertain) to 9 (extremely appropriate). An indication was categorized as necessary (angiography is the best option available to the patient), appropriate but not necessary (the benefits of performing angiography exceed the risks), uncertain (the benefits and risks are approximately equal), or inappropriate (the risks outweigh the benefits). In this analysis, we considered cases that were either necessary or appropriate anytime during their admission (deemed “appropriate”) as having an indication for angiography to conservatively bias the analysis.

Coronary Procedure Use

We calculated utilization rates for coronary angiography among all CKD and non-CKD patients as well as for patients who were judged to be appropriate for angiography. Differences in the odds of angiography among CKD and non-CKD patients, adjusting for the demographic, clinical, and hospital characteristics, were tested using the Breslow-Day χ2 test for homogeneity of odds ratios. The analytic sample for evaluation of appropriateness (n = 53,567) was slightly smaller than for other analyses because individuals who underwent angiography in a hospital before transfer (n = 574 [1.0%]) and those who died within 1 d (n = 3141 [5.5%]) were also excluded.

We also determined the use of percutaneous transluminal angioplasty (PTCA), coronary artery bypass grafting (CABG), and both invasive therapies during the hospital admission for CKD patients and non-CKD patients who received angiography. We repeated this calculation by considering only patients who were appropriate for angiography and received it.

Coronary Angiography, Revascularization, and Mortality in Appropriate CKD Patients

Because coronary angiography was not randomly assigned, we used two approaches to explore the effect of receiving angiography on 1-yr mortality for CKD patients who were judged appropriate for the procedure. We first estimated the adjusted odds of death using logistic regression. We also used a propensity score approach to estimate stratum-specific odds as well as an overall estimate.

Regression-Adjusted Estimates

We fit a multiple logistic regression model to the log-odds of mortality for appropriate CKD patients who underwent angiography compared with those who did not, adjusting for patient and hospital characteristics. The odds ratio (OR) and corresponding 95% confidence intervals (CI) were constructed on the basis of the estimated regression parameters. Model discrimination was assessed using the area under the receiver operating characteristic (ROC) curve (22). Calibration was assessed using the Hosmer-Lemeshow goodness-of-fit criterion (23).

Propensity Score Approach

We created a propensity score (24,25) for the likelihood of undergoing coronary angiography during hospitalization among CKD patients who were rated appropriate for the procedure. Multiple logistic regression with provision of angiography as the dependent variable was used in the development of the propensity score; it also incorporated patient and hospital characteristics. We included demographic characteristics (age, gender, race, state of residence), coexisting medical conditions (e.g., diabetes, congestive heart failure), physiologic derangements (e.g., conduction disturbances, heart rate, serum albumin), hospital characteristics of the initial admitting hospital (e.g., number of beds, presence of teaching affiliation, rural location, state), and interaction of selected factors in the model.

After fitting the model, we then ranked all appropriate CKD patients by their estimated propensity score and grouped patients within quintiles. We calculated the OR and 95% CI for 1-yr mortality, comparing within each quintile those who underwent coronary angiography and those who did not. A combined difference was estimated by averaging the OR across the quintiles (26) using weights proportional to the inverse of the variance of the estimates.

Revascularization Procedures

Because revascularization may influence mortality risk in CKD patients (and is dependent on angiography), we examined the relation between revascularization within the hospital admission and 1-yr mortality. We restricted this analysis to CKD patients who were considered appropriate and received angiography. We categorized patients into one of four mutually exclusive and exhaustive groups: those who underwent PTCA only, those who underwent CABG only, those who had both PTCA and CABG, and those who did not receive any revascularization procedure. We then performed a χ2 test to determine whether there was a significant association between mortality and revascularization.

For all analyses, two-tailed P < 0.001 was considered statistically significant. Statistical analyses were conducted using SAS (SAS Institute, Cary, NC).

Results

Characteristics of Study Groups

Baseline characteristics of the study population stratified by the presence of CKD and the use of coronary angiography are presented in Table 1. Using the serum creatinine ≥1.5 mg/dl definition, 15,093 (26.3%) were classified as CKD. Using the estimated GFR <30 ml/min per 1.73 m2 definition, 17,139 (30.0%) were classified as CKD, including 40.2 and 47.1% of patients aged 80 to 84 and 85 to 89 yr, respectively.

T1-24
Table 1:
Baseline characteristics by serum creatinine and angiography statusa

As expected, younger patients, men, and white individuals were more likely to undergo coronary angiography than older patients, women, and black individuals, respectively. Within both the non-CKD and CKD groups, patients with coexisting diseases (e.g., congestive heart failure, stroke, low ejection fraction, shock in hospital) were less likely to undergo coronary angiography. An exception to this rule was the more frequent use of coronary angiography among patients with recurrent ischemic chest pain. One-year mortality was markedly increased (52.6% versus 26.4%; P < 0.0001) among patients with CKD compared with those with normal or near-normal renal function.

Coronary Procedure Use in CKD Patients

Coronary Angiography.

Fewer patients with CKD than without CKD underwent angiography during their hospital stay (25.2% versus 46.8%; P < 0.0001). Angiography rates were lower for CKD patients with diabetes (n = 5851) than for those without diabetes (n = 9242; 23.6% versus 26.3%; P < 0.001). Adjusting for other predictors of angiography, the OR for angiography for CKD patients was 0.47 (95% CI, 0.40 to 0.52). Of the 13,348 patients who had CKD and for whom sufficient data were available from which to calculate an appropriateness score, 6794 (50.9%) were considered “appropriate” (i.e., appropriate or necessary) during at least one phase of hospitalization, 2014 (29.6%) of whom actually had the procedure. In the non-CKD population, 20,283 (50.4%) were considered appropriate, 10,056 (49.6%) of whom actually had the procedure. Results were similar when CKD was defined using the estimated GFR cutoff. Patients with estimated GFR <30 ml/min per 1.73 m2 were significantly less likely to undergo angiography (23.3% versus 48.7%; P < 0.0001). Among the 7783 patients who had estimated GFR <30 ml/min per 1.73 m2 and were considered appropriate for angiography, only 2184 (28.1%) underwent the procedure.

Revascularization.

Of the 3805 CKD patients who underwent coronary angiography, 1195 (31.4%) underwent PTCA, 786 (20.7%) underwent CABG, and 99 (2.6%) received both invasive therapies (54.7% total). Compared with non-CKD patients, a smaller fraction of patients who underwent angiography eventually received revascularization (54.7% versus 62.0%; P < 0.0001).

Coronary Angiography and Mortality in CKD Patients: Regression Approach

Among all CKD patients who underwent coronary angiography, unadjusted 1-yr mortality was 30.2% compared with 60.2% among those who did not undergo angiography. Adjusting for other factors associated with 1-yr mortality, coronary angiography was associated with a significant reduction in the risk of death (adjusted OR, 0.54; 95% CI, 0.49 to 0.60). When the sample was restricted to the 6794 CKD patients who were considered appropriate for angiography, the adjusted OR was 0.58 (95% CI, 0.50 to 0.67). The logistic regression model above had good discrimination (area under the ROC curve, 0.76) and was well calibrated (Hosmer-Lemeshow χ2 = 12.73, P = 0.12). The corresponding adjusted OR for estimated GFR <30 ml/min per 1.73 m2 was 0.59 (95% CI, 0.51 to 0.67).

Propensity Score Approach in Appropriate CKD Patients

Table 2 shows the 1-yr mortality results stratified by quintiles of propensity score. In all quintiles except the first (likelihood of angiography, <6%), there was a significant decrease in the odds of death in patients who had CKD and were considered appropriate and received angiography. The combined propensity score-adjusted odds of death were similar to those obtained with logistic regression alone (0.62; 95% CI. 0.54 to 0.70). The model fit to determine the propensity score had very good discrimination (area under the ROC curve of 0.85) and was well calibrated (Hosmer-Lemeshow χ2 = 8.25, P = 0.41). As above, results were similar when the alternative definition for CKD was used (Table 2).

T2-24
Table 2:
Propensity score-adjusted 1-year mortality in appropriate CKD patients by coronary angiographya

Revascularization and Mortality

One-year mortality rates for CKD patients who were considered appropriate and underwent angiography were 23.4, 25.1, 14.3, and 29.8% for those who ultimately underwent CABG, PTCA, PTCA + CABG, and no revascularization, respectively (P = 0.01). In contrast, among the 4751 CKD patients who were considered appropriate but did not undergo angiography (and did not undergo revascularization), 1-yr mortality was 47.4%.

Discussion

The management of coronary heart disease has been the subject of numerous health services research inquiries over the past two decades, largely because the conditions are relatively common and there is wide variation in practice patterns. Several groups have previously studied factors that influence the provision of coronary angiography after acute MI. Gender and race inequities have been the focus of much of this research, with grave concerns expressed by many that personal biases (and/or other barriers to care) are responsible for the relative underutilization of coronary angiography among women and individuals of minority backgrounds. Additional work has explored regional variation, including one analysis published by Guadagnoli et al. (27) showing that angiography rates were 50% higher in Texas than in New York, with no evidence at that time of a significant symptomatic benefit or improvement in long-term survival. The question remains whether there is over- or underutilization (or both) of cardiac procedures as a result of this variation.

In this study, we aimed to explore a different bias in the provision of care—one presumably based on fear of an immediate complication (radiocontrast-associated nephrotoxicity) and virtually restricted to the population of interest (people with CKD). In contrast to most other causes of acute renal failure, the timing of radiocontrast-associated nephrotoxicity is predictable, and the complication is preventable by avoidance of radiocontrast exposure. Typically, there is a decrease in GFR, often with evidence of renal hypoperfusion (i.e., low urinary sodium, low fractional excretion of sodium, high specific gravity) (28). The course of radiocontrast-associated nephrotoxicity is usually self-limited, with the peak serum creatinine concentration 3 to 5 d after exposure.

Several studies have identified predictors for development of radiocontrast-associated nephrotoxicity (18,28–31). The common risk factors in all studies are baseline CKD and diabetes. The dose of radiocontrast, congestive heart failure, proteinuria, hyponatremia (as a proxy for the severity of heart failure), and multiple myeloma has also been suggested to increase risk. Overall, the absolute risk is low, and progression to ESRD is rare. In most studies, individuals without diabetes and with baseline serum creatinine concentrations <1.5 mg/dl experience <0.5% incidence of radiocontrast-associated nephrotoxicity. Only patients with diabetes and CKD experience a >2% incidence of significant (yet still reversible) renal dysfunction.

In this study, we showed that patients with CKD underwent coronary angiography at nearly one half the rate of individuals with normal or near-normal renal function, even though the proportion of patients who were deemed appropriate by published criteria was roughly the same. Coronary angiography was associated with a significant reduction in the odds of death. The propensity score-adjusted OR in CKD patients who were considered appropriate for angiography was 0.62 (95% CI, 0.54 to 0.70). Results obtained using propensity scores (OR, 0.62) and logistic regression (OR, 0.58) were similar, although the propensity score analysis indicated that the size of the benefit differed across the quintiles. Given that the propensity score approach requires fewer assumptions and tends to balance differences between treated and untreated groups, we prefer these results to those of the logistic regression model.

There are several important limitations to this study. First, propensity scores can adjust only for the associations among observed covariates and the chosen treatment or strategy (coronary angiography in this case). Other unobserved covariates could influence the likelihood of treatment, and there is no guarantee that the correlation among observed and unobserved covariates is sufficiently high to account adequately for this deficiency. Second, the population was restricted to Medicare beneficiaries. Therefore, these results cannot be extrapolated to the universe of patients who have CKD and are younger than 65 yr. Third, the use of serum creatinine to distinguish CKD from non-CKD patients is relatively crude. Use of a direct measure of kidney function, such as creatinine clearance or iothalamate GFR, to define CKD might have been superior, especially among the elderly population, in whom the Modification of Diet in Renal Disease (MDRD) equation has not been validated. However, the serum creatinine concentration is the single laboratory value used by most clinicians to define CKD, and an analysis based on estimated GFR yielded qualitatively similar results. Finally, we were unable to evaluate the incidence (or severity) of radiocontrast-associated nephrotoxicity in patients who underwent angiography to confirm complication rates reported in other cohorts.

In summary, elderly patients with CKD undergo coronary angiography much less frequently than do patients with normal or near-normal renal function, despite being considered equivalently appropriate for coronary angiography by conventional criteria. Invasive strategies of care including coronary angiography and subsequent revascularization procedures are associated with a significant decrease in the odds of death in this population, even after adjusting for confounding and selection bias. Sicker patients may undergo coronary angiography less frequently for a variety of reasons, including their risk for nephrotoxicity. Nonetheless, this modest risk does not outweigh the potential benefit of diagnostic coronary angiography in patients who have CKD and are otherwise considered appropriate candidates for the procedure. Although efforts should continue to limit over- and underuse of coronary angiography, these data suggest that the relative usage in individuals with CKD is low and probably not justified.

This work was supported in part by Agency for Healthcare Research and Quality Grant #HS06503.

We are grateful to Margaret Volya, MS, and Fung-Yea Huang, PhD, for programming assistance.

1. Shekelle PG, Kahan JP, Bernstein SJ, Leape LL, Kamberg CJ, Park RE: The reproducibility of a method to identify the overuse and underuse of medical procedures. N Engl J Med 338: 1888–1895, 1998
2. Kahan JP, Bernstein SJ, Leape LL, Hilborne LH, Park RE, Parker L, Kamberg CJ, Brook RH: Measuring the necessity of medical procedures. Med Care 32: 357–365, 1994
    3. Wennberg JE, Blowers L, Parker R, Gittelsohn AM: Changes in tonsillectomy rates associated with feedback and review. Pediatrics 59: 821–826, 1977
    4. Wennberg J, Gittelsohn A: Variations in medical care among small areas. Sci Am 246: 120–134, 1982
      5. Wennberg JE, Freeman JL, Shelton RM, Bubolz TA: Hospital use and mortality among Medicare beneficiaries in Boston and New Haven. N Engl J Med 321: 1168–1173, 1989
        6. Ayanian JZ, Epstein AM: Differences in the use of procedures between women and men hospitalized for coronary heart disease. N Engl J Med 325: 221–225, 1991
        7. Ayanian JZ, Udvarhelyi IS, Gatsonis CA, Pashos CL, Epstein AM: Racial differences in the use of revascularization procedures after coronary angiography. JAMA 269: 2642–2646, 1993
          8. Peterson ED, Wright SM, Daley J, Thibault GE. Racial variation in cardiac procedure use and survival following acute myocardial infarction in the Department of Veterans Affairs. JAMA 271: 1175–1180, 1994
            9. Canto JG, Allison JJ, Kiefe CI, Fincher C, Farmer R, Sekar P, Person S, Weissman NW: Relation of race and sex to the use of reperfusion therapy in Medicare beneficiaries with acute myocardial infarction. N Engl J Med 342: 1094–1100, 2000
            10. Guadagnoli E, Ayanian JZ, Gibbons G, McNeil BJ, LoGerfo FW: The influence of race on the use of surgical procedures for treatment of peripheral vascular disease of the lower extremities. Arch Surg 130: 381–386, 1995
            11. Kasiske BL, Neylan JF 3rd, Riggio RR, Danovitch GM, Kahana L, Alexander SR, White MG: The effect of race on access and outcome in transplantation. N Engl J Med 324: 302–307, 1991
            12. Ayanian JZ, Cleary PD, Weissman JS, Epstein AM: The effect of patients’ preferences on racial differences in access to renal transplantation. N Engl J Med 341: 1661–1669, 1999
              13. Krumholz HM, Pasternak RC, Weinstein MC, Friesinger GC, Ridker PM, Tosteson AN, Goldman L: Cost effectiveness of thrombolytic therapy with streptokinase in elderly patients with suspected acute myocardial infarction. N Engl J Med 327: 7–13, 1992
              14. Bowling A: Ageism in cardiology. BMJ 319: 1353–1355, 1999
                15. Selby JV, Fireman BH, Lundstrom RJ, Swain BE, Truman AF, Wong CC, Froelicher ES, Barron HV, Hlatky MA: Variation among hospitals in coronary-angiography practices and outcomes after myocardial infarction in a large health maintenance organization. N Engl J Med 335: 1888–1896, 1996
                16. Ayanian JZ, Landrum MB, Normand SL, Guadagnoli E, McNeil BJ: Rating the appropriateness of coronary angiography—Do practicing physicians agree with an expert panel and with each other? N Engl J Med 338: 1896–1904, 1998
                17. Kilaru PK, Kelly RF, Calvin JE, Parrillo JE: Utilization of coronary angiography and revascularization after acute myocardial infarction in men and women risk stratified by the American College of Cardiology/American Heart Association guidelines. J Am Coll Cardiol 35: 974–979, 2000
                  18. Barrett BJ, Parfrey PS, Vavasour HM, O’Dea F, Kent G, Stone E: A comparison of nonionic, low-osmolality radiocontrast agents with ionic, high-osmolality agents during cardiac catheterization. N Engl J Med 326: 431–436, 1992
                  19. Erley CM, Duda SH, Rehfuss D, Scholtes B, Bock J, Muller C, Osswald H, Risler T: Prevention of radiocontrast-media-induced nephropathy in patients with pre-existing renal insufficiency by hydration in combination with the adenosine antagonist theophylline. Nephrol Dial Transplant 14: 1146–1149, 1999
                    20. National Kidney Foundation. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification and stratification. Am J Kidney Dis 39: S1–S85, 2002
                    21. Normand SL, Landrum MB, Guadagnoli E, Ayanian JZ, Ryan TJ, Cleary PD, McNeil BJ: Validating recommendations for coronary angiography following acute myocardial infarction in the elderly: A matched analysis using propensity scores. J Clin Epidemiol 54: 387–398, 2001
                    22. Hanley JA, McNeil BJ: The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143: 29–36, 1982
                    23. Hosmer DW, Lemeshow S. Applied Logistic Regression, New York, John Wiley & Sons, 1989
                    24. Rosenbaum PR, Rubin DB: Reducing bias in observational studies using subclassification on the propensity score. J Am Stat Assoc 79: 516–524, 1984
                    25. Rubin DB: Estimating causal effects from large data sets using propensity scores. Ann Intern Med 127 [Suppl]: 757–763, 1997
                      26. Petersen LA, Normand SL, Daley J, McNeil BJ: Comorbidity, severity, and outcome of patients with acute myocardial infarction in the Veterans Administration compared with Medicare. N Engl J Med 343: 1934–1941, 2000
                      27. Guadagnoli E, Hauptman PJ, Ayanian JZ, Pashos CL, McNeil BJ, Cleary PD: Variation in the use of cardiac procedures after myocardial infarction. N Engl J Med 333: 573–578, 1995
                      28. Barrett BJ: Contrast nephrotoxicity. J Am Soc Nephrol 5: 125–137, 1994
                      29. Taliercio CP, Vlietstra RE, Fisher LD, Burnett JC: Risks for renal dysfunction with cardiac angiography. Ann Intern Med 104: 501–504, 1986
                        30. Parfrey PS, Griffiths SM, Barrett BJ, Paul MD, Genge M, Withers J, Farid N, McManamon PJ: Contrast material-induced renal failure in patients with diabetes mellitus, renal insufficiency, or both. A prospective controlled study. N Engl J Med 320: 143–149, 1989
                          31. Lautin EM, Freeman NJ, Schoenfeld AH, Bakal CW, Haramati N, Friedman AC, Lautin JL, Braha S, Kadish EG, Sprayregen S, et al.: Radiocontrast-associated renal dysfunction: incidence and risk factors. AJR Am J Roentgenol 157: 49–58, 1991
                            Copyright © 2004 The Authors. Published by Wolters Kluwer Health, Inc. All rights reserved.