Comparative Utilization and Temporal Trends in Cardiac Stress Testing in US Medicare Beneficiaries with and without Chronic Kidney Disease

Abstract

Objectives

We aimed to analyze temporal trends in cardiac stress testing in US Medicare beneficiaries, 2008–2012; types of stress testing; and comparative utilization related to presence and severity of chronic kidney disease (CKD).

Background

A long-held perception depicts patients with CKD as being treated less intensively for cardiovascular disease than non-renal patients. We wondered whether use of diagnostic testing for ischemic heart disease is affected by presence of CKD.

Methods

Using the 20% Medicare sample, we assembled yearly cohorts of Medicare beneficiaries (~4,500,000/year), 2008–2012. Beneficiaries aged 66 years or older undergoing a first cardiac stress test, with no prior history of coronary revascularization and no acute coronary syndrome within 60 days, were identified, as was type of stress test. We analyzed temporal trends and compared testing rates related to CKD stage vs. no CKD. A Poisson regression model estimated the likelihood of stress testing in 2012 by CKD stage, adjusted for demographic characteristics and comorbid conditions.

Results

Approximately 480,000 older patients (~29,000 with CKD) underwent stress tests in 2008, progressively declining to ~400,000 in 2012 (~38,000 with CKD). In 2008–2012, 78%–80% of all stress testing in non-CKD patients used nuclear imaging, as did 87%–88% in CKD patients. Rates of stress testing declined progressively for non-CKD and CKD patients from 2008 to 2012: 11.5 to 9.4/100 patient-years and 16.8 to 13.4/100 patient-years, respectively. The adjusted Poisson model, with non-CKD as the reference, showed increasing likelihood of stress testing with worsening CKD: Incidence rate ratios 1.01, stages 1–2 (P = ns); 1.05, stage 3, (P < 0.0001); 1.01, stage 4 (P = ns), 1.04, stage 5ND (P = ns), 1.15 stage 5D (P < 0.0001).

Conclusions

Overall rates of cardiac stress testing (over three-fourths using nuclear imaging) declined 2008–2012 in Medicare beneficiaries aged 66 years or older, but were consistently higher for CKD than for non-CKD patients; the effect of screening algorithms for transplant candidates was unknown. Our data refute under-utilization of cardiac stress testing in CKD patients.

Introduction

Historically, a long-held perception has depicted patients with chronic kidney disease (CKD), and especially those with dialysis-dependent end-stage renal disease (ESRD), as relatively under-treated (compared with patients without CKD) for cardiovascular disease in general, and for acute coronary syndrome in particular: use of evidence-based therapies was inversely related to CKD stage (despite higher mortality in patients with advanced CKD and acute myocardial infarction) (1–4). This approach was derided as “renalism” (5) or “therapeutic nihilism” (6) nearly two decades ago.

The “truism” of “renalism,” however, may actually be a misconception in the current era; as far back as 2011, the United States Renal Data System (USRDS) Annual Data Report commented on the “virtual sea change in clinical practice related to treatment of cardiovascular disease in ESRD patients,” particularly with respect to the rapid expansion of beta-blocker therapy, based on 2008 Medicare data (7).

In this context, we wondered whether use of diagnostic testing for ischemic heart disease is affected by the presence of CKD in the current era. The present study aimed to analyze temporal trends in use of cardiac stress testing in Medicare beneficiaries with respect to CKD status.

Methods

Using the 20% Medicare sample (2007–2012), a random 20% sample of all Medicare beneficiaries, we assembled yearly cohorts of Medicare beneficiaries with and without CKD who were alive, were aged 66 years or older, and had Medicare Parts A and B coverage on January 1 of each year from 2008 to 2012, with at least 12 months of continuous coverage preceding January 1 (baseline period). We excluded patients who underwent percutaneous coronary intervention, coronary artery bypass grafting, coronary angiography, or kidney transplant, or who participated in a health maintenance organization, anytime in the baseline period. We also excluded patients with unstable angina or acute myocardial infarction in the last 2 months of the baseline period, to mirror enrollment criteria for the International Study of Comparative Health Effectiveness with Medical and Invasive Approaches-Chronic Kidney Disease (ISCHEMIA-CKD) trial (ClinicalTrials.gov identifier NCT01985360). Follow-up started on January 1 of each year from 2008 to 2012 and ended at the earliest of stress test, death, disenrollment from Medicare coverage, the day before CKD/ESRD diagnosis for non-CKD patients, the day before kidney transplant for ESRD patients, or December 31 of that year. The research was approved by the institutional review board at Brigham & Women’s Hospital and by the Human Subjects Research Committee of the Hennepin County Medical Center/Hennepin Healthcare System, Inc.

Patients with CKD were identified by International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes 585.X, requiring one inpatient or two outpatient/Part B claims at least 30 days apart in the baseline period. CKD stage was determined by the highest stage-specific code. Patients with ESRD were identified using an algorithm similar to the one used in the 2013 USRDS Annual Data Report (8). ESRD was defined by searching the Medicare Beneficiary Summary file for the variable “ESRD_FLG,” a yes/no switch for ESRD status. Revenue center outpatient codes were searched for evidence of outpatient dialysis on the basis of revenue codes for hemodialysis or peritoneal dialysis. Stress tests were identified via Current Procedural Terminology or ICD-9-CM procedure codes during the follow-up period. If two or more tests occurred on the same day, assignment of first test type followed a hierarchical approach starting with stress echocardiography, followed by stress nuclear imaging, stress magnetic resonance imaging, and stress electrocardiography. If none of these were found, we searched for non-invasive coronary computed tomography (CT) angiography (CCTA). If none of these were found, we searched for CT coronary calcium scan.

Statistical Methods

Patient characteristics were described for yearly cohorts of CKD and non-CKD patients and subgroups who received a first stress test. Frequency distributions of type of first stress test were examined. Unadjusted rates of first stress test (any type) were calculated using number of patients who received a stress test per 100 patient-years. Differences in trends in stress testing rates between the non-CKD and CKD cohorts and among CKD stages 1–5D were assessed using a generalized linear model with a negative binomial distribution. A Poisson regression model was used to estimate the likelihood of a stress test in 2012 for CKD patients overall and by CKD stage, compared with non-CKD patients, with adjustment for demographic characteristics and comorbid conditions. The validity of the Poisson model was visually inspected by plotting the observed against the predicted rates of a first stress test; the graph did not show deviation from the diagonal. To account for over-dispersion of the data, the covariance matrix of the parameter estimates was inflated by the dispersion parameter estimated by deviance statistics divided by its degrees of freedom.

Results

Table S1 summarizes the demographic characteristics of Medicare beneficiaries (approximately 4.5–4.6 million/cohort year) with and without CKD. Table 1 details the demographic characteristics of patients (approximately 480,000 in 2008, with progressive annual declines to about 400,000 in 2012) receiving stress tests from 2008 to 2012 by CKD status. CKD patients tended to be older, more were male, and proportions of black patients were higher compared with non-CKD patients. Table 2 shows the distribution of type of first stress test for each cohort year, related to CKD status. Stress nuclear imaging accounted for more than three-fourths of all stress testing, and for an even greater disproportion for CKD patients (87% in 2009–2012).

Table 1.

Demographic characteristics of patients receiving stress tests, by CKD status, 2008–2012

Characteristics	2008		2009		2010		2011		2012
	CKD Status		CKD Status		CKD Status		CKD Status		CKD Status
	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes
Total n	452,520	29,112	437,101	31,124	405,907	33,324	287,199	35,941	361,365	38,215
Age, yrs., %
66–74	57.1	46.1	57.6	46.0	58.0	46.0	58.6	45.5	58.8	45.4
75–84	37.3	45.2	36.8	44.9	36.2	44.6	35.5	44.4	35.2	44.4
≥ 85	5.5	8.7	5.7	9.1	5.8	9.4	6.0	10.0	6.0	10.2
Sex, %
Men	48.7	55.5	48.2	55.0	48.3	54.1	47.9	53.6	48.0	53.6
Women	51.3	44.6	51.8	45.0	51.7	45.9	52.1	46.4	52.0	46.4
Race/ethnicity, %
White	89.5	81.0	89.1	80.3	88.7	80.3	88.4	79.7	88.2	80.1
Black	5.9	13.4	6.2	13.5	6.4	13.4	6.5	13.8	6.6	13.5
Asian	1.5	1.8	1.6	2.1	1.7	2.0	1.7	2.2	1.7	2.1
Hispanic	1.4	2.1	1.4	2.2	1.5	2.3	1.5	2.3	1.5	2.1
Other/unknown	1.6	1.8	1.7	2.0	1.7	2.0	1.9	2.1	2.0	2.2
CKD stage, %
1–2		6.8		6.6		6.6		6.9		6.8
3		35.6		40.1		43.6		46.3		48.0
4		18.1		18.6		18.3		17.7		16.8
5ND		3.4		3.6		3.1		2.8		2.7
5D		10.8		10.2		9.5		9.4		8.5
Unknown		25.3		20.9		19.0		17.0		17.3

CKD, chronic kidney disease.

Table 2.

Distribution of type of first stress test by CKD status, 2008–2012

Type of Test, %	2008		2009		2010		2011		2012
	CKD Status		CKD Status		CKD Status		CKD Status		CKD Status
	No	Yes	No	Yes	No	Yes	No	Yes	No	Yes
Echo	11.6	7.4	12.4	8.0	11.9	7.5	12.2	7.2	12.6	7.4
Nuclear	80.3	88.2	80.0	87.3	79.1	87.3	78.5	87.4	77.7	87.1
MRI	0.04	*	0.04	0.04	0.05	*	0.5	*	0.06	0.03
ECG	8.0	4.5	7.7	4.7	8.0	4.7	8.3	5.0	8.7	5.0
CCTA					0.7	0.4	0.6	0.3	0.7	0.4
CT (calcium)					0.2	0.1	0.2	0.1	0.2	0.1

Note:

^*denotes suppressed data due to sample size less than 11.

CCTA, coronary computed tomography angiography; CKD, chronic kidney disease; CT, computed tomography; ECG, electrocardiogram; MRI, magnetic resonance imaging.

Figures 1A and 1B, respectively, show temporal trends in rates of stress testing by CKD status and CKD stage. Rates of stress testing declined progressively from 2008 to 2012 for non-CKD patients (11.5 to 9.4 per 100 patient-years) and CKD patients (16.8 to 13.4 per 100 patient-years). Rates of stress testing were higher for CKD patients (P < 0.0001), but temporal trends did not differ between the two groups (P = 0.93). Total stress testing rates declined significantly over time (P = 0.007). Figure 1B shows no significant difference in temporal trends by CKD stage (P = 0.98). Overall, rates of testing decreased significantly over time (P = 0.003). In 2012, there was a trend toward higher rates (per 100 patient-years) of stress testing with worse kidney function: non-CKD, 9.4; stage 1–2, 13.0; stage 3, 14.0; stage 4, 13.8, stage 5 non-dialysis (ND), 14.9; and stage 5 dialysis (D), 18.4 (stage 5D vs. stages 3–5ND, P = 0.0002; stage 3–5ND vs. stages 1–2, P = 0.06).

Figure 1. Temporal trends in rates of stress testing.

Panel A, CKD and non-CKD patients; panel B, CKD stages. CKD, chronic kidney disease.

In an unadjusted Poisson regression model, CKD patients were 43% more likely (P < 0.0001) to undergo stress testing than non-CKD patients (relative risk [RR] 1.43, 95% confidence interval [CI] 1.41–1.44). A similar model with categorized CKD stages showed a graded increased likelihood of stress testing related to CKD severity: no CKD, reference, RR 1.00; stages 1–2, 1.38, 1.33–1.44; stage 3, 1.49, 1.47–1.51; stage 4, 1.46, 1.43–1.50; stage 5ND 1.58, 1.48–1.68), stage 5D, 1.95, 1.89–2.02; stage unknown, 1.11, 1.09–1.14 (all stages P < 0.0001 vs. no CKD). When demographics were added to the model, we found that women were 21% less likely than men (P < 0.0001) and black patients 6% less likely than white patients (P < 0.0001) to undergo stress testing.

Table 3 shows the Poisson model adjusted for demographics and comorbid medical conditions. In this fully adjusted model, the independent likelihood of stress testing by CKD stage and sex is attenuated (with the addition of comorbid medical conditions). Compared with no CKD, the likelihood of stress testing ranged from a 1% increase (P = NS) for stages 1–2 to a 15% increase for stage 5D (P < 0.0001). Unknown CKD stage was 20% less likely (P < 0.0001) to be associated with stress testing after adjustment for comorbid conditions. Women were 11% less likely than men (P < 0.0001) and black patients 6% less likely than white patients (P < 0.0001) to undergo stress testing in the fully adjusted model. The strongest negative predictor was age: patients aged 85 years or older were 65% less likely than patients ages 66–74 years (P < 0.0001) to undergo stress testing. Not surprisingly, a history of coronary artery disease (vs. none) was strongly associated with stress testing (RR 2.61, 95% CI 2.58–2.63). In contrast, congestive heart failure and stroke/transient ischemic attack were negatively associated with stress testing (19% and 23% decreased likelihood, respectively, P < 0.0001).

Table 3.

Adjusted Poisson regression model for incidence of stress testing

Parameter	RR (95% CI)	P
CKD stage
No CKD	1.00 (ref)
1–2	1.01 (0.97–1.06)	0.575
3	1.05 (1.03–1.07)	< 0.0001
4	1.01 (0.98–1.04)	0.722
5ND	1.04 (0.97–1.13)	0.255
5D	1.15 (1.10–1.20)	< 0.0001
Unknown	0.80 (0.78–0.83)	< 0.0001
Age, years
66–74	1.00 (ref)
75–84	0.86 (0.85–0.87)	< 0.0001
≥ 85	0.35 (0.34–0.35)	< 0.0001
Sex
Men	1.00 (ref)
Women	0.89 (0.88–0.89)	< 0.0001
Race
White	1.00 (ref)
Black	0.94 (0.93–0.96)	< 0.0001
Other	0.97 (0.95–0.99)	0.0002
CAD
No	1.00 (ref)
Yes	2.61 (2.58–2.63)	< 0.0001
CHF
No	1.00 (ref)
Yes	0.81 (0.80–0.83)	< 0.0001
Diabetes mellitus
No	1.00 (ref)
Yes	1.14 (1.13–1.15)	< 0.0001
CVA/TIA
No	1.00 (ref)
Yes	0.77 (0.75–0.78)	< 0.0001
PVD
No	1.00 (ref)
Yes	1.01 (1.00–1.03)	0.03
Other cardiac disease
No	1.00 (ref)
Yes	1.22 (1.20–1.24)	< 0.0001
COPD
No	1.00 (ref)
Yes	1.05 (1.04–1.06)	<.0001
Dysrhythmia
No	1.00 (ref)
Yes	1.23 (1.21–1.24)	< 0.0001
Hypertension
No	1.00 (ref)
Yes	1.44 (1.43–1.46)	< 0.0001

CAD, coronary artery disease; CHF, congestive heart disease; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; CVA/TIA, cerebrovascular accident/transient ischemic attack; PVD, peripheral vascular disease.

Discussion

Few published data report on the relative utilization of cardiac stress testing in CKD patients. In 2012, about 2,000,000 Medicare beneficiaries (aged 66 years or older), underwent at least one cardiac stress test; approximately 190,000 had CKD. Based on our analysis, we draw several important conclusions. First, our data do not suggest that cardiac stress testing is under-utilized in CKD patients. To the contrary, rates of stress testing were higher in CKD patients than in non-CKD patients from 2008 to 2012; even after adjustment for demographics and comorbid medical conditions, we found a graded increase in the likelihood of stress testing related to CKD stage. Second, we found a progressive decline in the overall rate of cardiac stress testing from 2008 to 2012, and the magnitude was not related to CKD, as it was similar for non-CKD and CKD patients. Third, over 75% of all stress testing in the study population involved nuclear imaging, and despite potential concerns regarding the accuracy of stress nuclear scintigraphy versus stress echocardiography in patients with advanced CKD (9), stress nuclear imaging was used more in CKD patients (87% in 2012). Fourth, we found relative underuse of cardiac stress testing related to sex and race; even after adjustment for demographics and comorbid conditions, women and black patients were respectively 11% and 6% less likely to undergo a cardiac stress test.

The temporal trend finding of declining rates of (predominantly nuclear) stress tests and numerical dominance of nuclear imaging (reflecting clinicians’ preferences for pharmacologic testing in elderly patients) are concordant with prior studies. McNulty et al (10) reported a decline in nuclear myocardial perfusion imaging in northern California after 2006, and Levin et al (using Medicare Part B data) reported peak rates of nuclear myocardial perfusion imaging in 2006, followed by declines in 2009 through 2010 (11). The rate of decline accelerated after 2009; by 2013, it was slightly lower than in 2001 (12). In contrast, rates of stress echocardiography changed little from 2001 to 2010 (12.5/1000 beneficiaries in 2010), but declined through 2013 (to 10.8/1000 beneficiaries, versus 61.9/1000 beneficiaries for nuclear myocardial perfusion imaging). The utilization rates of CCTA in fee-for-service Medicare beneficiaries was low in 2012 (1.1/1000 beneficiaries) (12). Andrus and Welch (13) reported on Medicare services provided by cardiologists in the US from 1999–2008. For 2008, they reported that nearly all (95%) imaging stress tests (nuclear or echocardiographic) involved nuclear imaging.

Our study has several limitations. Ascertainment of test utilization is based on Medicare claims data, and we may have underestimated test rates (particularly CCTA and CT calcium scans) due to non-reimbursable claims. We excluded patients with recent (within 60 days) acute coronary syndrome. Possibly, the differential rate of stress testing related to CKD status might differ with recent acute coronary syndrome. Our analysis focused on the first stress test. Clearly, the total number of stress tests would be higher if repeat testing for the same patient were included. Based on our findings related to CKD stage, we believe that caution should be exercised in interpreting the data for unspecified CKD stage, as ambiguity regarding CKD severity is likely. Also, the reported distribution of 7% for CKD stages 1/2 may be lower than expected clinically, likely reflecting coding practices. Our data were based on Medicare claims for patients aged 66 years or older; the relative use of cardiac stress testing might be different in younger patients, or in elderly patient with Medicare Advantage coverage (also not included in this study). Our regression model does not include all potentially relevant clinical factors (such as symptoms, laboratory data (including cholesterol measurements), or electrocardiogram abnormalities); inclusion of additional variables could have potentially altered our findings. Our analysis focused on 2008 through 2012; we are agnostic to temporal trends occurring before or after the study period. Finally, our study does not separately account for the potential impact on rates of cardiac stress testing driven by screening algorithms for renal transplant candidates. This could inflate the numbers of patients with advanced CKD undergoing stress tests. Unfortunately, it is not feasible to assess transplant listing or tests done specifically for pretransplant evaluation in Medicare data.

Based on our data, we conclude that cardiac stress testing is not under-utilized in CKD patients (compared with non-CKD patients) in the current era. At least with regard to cardiac stress testing, our data do not support the concept of “renalism” or “nihilism” in the non-invasive assessment of ischemic heart disease in CKD patients.

Perspectives.

Competency in Medical Knowledge

Patients with chronic kidney disease are at increased risk for cardiovascular disease. Few data describe the utilization of cardiac stress testing in patients with chronic kidney disease. We found a progressive decrease in rates of cardiac stress testing in patients both with and without chronic kidney disease from 2008 to 2012. The absolute rate of stress testing was consistently higher for chronic kidney disease patients over the study period; over three-fourths received a stress nuclear imaging test.

Translational Outlook

Based on our data, cardiac stress testing is not under-utilized in patients with chronic kidney disease, refuting a long-held perception that these patients are treated less intensively than non-renal patients for cardiovascular disease. Further, despite potential concerns regarding the accuracy of stress nuclear scintigraphy versus stress echocardiography in patients with advanced chronic kidney disease, stress nuclear imaging was used in more chronic kidney disease patients than non-renal patients.

Abbreviations

CKD

chronic kidney disease

CT

computed tomography

CCTA

coronary computed tomography angiography

ESRD

end-stage renal disease

ICD-9-CM

International Classification of Diseases, Ninth Revision, Clinical Modification

USRDS

United States Renal Data System