Association of Physician Certification in Interventional Cardiology with In-Hospital Outcomes of Percutaneous Coronary Intervention

Abstract

Background

The value of American Board of Internal Medicine (ABIM) certification has been questioned. We evaluated the association of interventional cardiology (ICARD) certification with in-hospital outcomes of patients undergoing percutaneous coronary intervention (PCI) in 2010.

Methods and Results

We identified physicians who performed ≥10 PCIs in 2010 in the CathPCI Registry and determined ICARD status using ABIM data. We compared in-hospital outcomes of patients treated by certified and non-certified physicians using hierarchical multivariable models adjusted for differences in patient characteristics and PCI volume. Primary endpoints were all-cause in-hospital mortality and bleeding complications. Secondary endpoints included emergency coronary artery bypass grafting, vascular complications, and a composite of any adverse outcome. With 510,708 PCI procedures performed by 5,175 physicians, case mix and unadjusted outcomes were similar among certified and non-certified physicians. The adjusted risks of in-hospital mortality (OR 1.10, 95% CI 1.02-1.19) and emergency CABG (OR 1.32, 95% CI 1.12-1.56) were higher in the non-ICARD certified group, but the risks of bleeding, vascular complications, and the composite endpoint were not statistically significantly different between groups.

Conclusions

We did not observe a consistent association between ICARD certification and the outcomes of PCI procedures. Although there was a significantly higher risk of mortality and emergency CABG in patients treated by non-ICARD certified physicians, the risks of vascular complications and bleeding were similar. Our findings suggest that ICARD certification status alone is not a strong predictor of patient outcomes, and indicate a need to enhance the value of subspecialty certification.

Introduction

The American Board of Internal Medicine (ABIM) was founded nearly 80 years ago in an effort to establish uniform standards for internists¹. ABIM certification indicates “that internists have demonstrated – to their peers and to the public – that they have the clinical judgment, skills and attitudes essential for the delivery of excellent patient care.^”1 Over time, ABIM certification has evolved to include both recertification and, more recently, maintenance of certification requirements^2-3. Certification status plays a central role in many hospitals’ credentialing processes⁴, may influence patients’ choice of a physician⁵, and could be used to inform payment policies⁶. Although there is strong evidence for the internal validity of the testing process and the correlation of certification with other measures of physician quality such as program director’s ratings, the findings of studies examining the association of certification with patient outcomes have been mixed^7-13.

Board certification is of particular interest in the field of interventional cardiology (ICARD) where it is a relatively new development. Certification was first offered in 1999 in an effort to standardize a rapidly expanding field^14-15. Since then, there has been wide acceptance of the importance of ICARD certification. The current clinical competence statement for percutaneous coronary intervention (PCI) from the American College of Cardiology Foundation (ACCF), the American Heart Association (AHA), and the Society for Cardiovascular Angiography and Interventions strongly recommends that physicians performing PCI hold active ICARD ABIM certification¹⁶. Nevertheless, although ICARD certification has been in place for 16 years, the association between ICARD certification and the outcomes of patients undergoing PCI has not been evaluated. Critically assessing the certification process is essential, as physicians have challenged the certification process in both the lay and research press, citing time and financial pressures as well as a perceived lack of value of the process^17-19.

To address this gap in knowledge, we linked data regarding physician certification in ICARD from the ABIM database with patient outcomes data from the National Cardiovascular Data Registry’s (NCDR) CathPCI Registry. Specifically, we examined the association of ABIM physician certification in ICARD with patterns of care and outcomes of patients undergoing PCI. It is important to emphasize, however, that this study addresses certification status of the performing physician at the time of the procedure, and not maintenance of certification or recertification.

Methods

Data Sources

The CathPCI Registry was established as a mechanism to promote quality improvement efforts at hospitals performing diagnostic and interventional cardiac procedures²⁰. Hospitals voluntarily submit data to the registry using standardized definitions regarding patient demographics, medical history, risk factors, hospital presentation, initial cardiac status, procedural details, unique operator ID, medications, laboratory values and in-hospital outcomes of the PCI procedure. Data are entered by hospital personnel, and the data are only included in the analytic file if the hospital achieves >95% completeness of specified data elements. Currently, more than 1,600 sites submit data to the CathPCI Registry, which account for approximately 90% of catheterization labs performing PCI nationwide^21-22. Previous data quality reviews have supported the accuracy of the data submitted to the CathPCI Registry²³.

The ABIM database contains data on all physicians certified in internal medicine and each of its subspecialties. This includes demographic data, training dates, and certification status for internal medicine (IM), cardiovascular disease (CV), and ICARD.

Patient Population

A flow chart of patient and physician selection is shown in Figure 1. All patients who underwent a PCI procedure reported to the CathPCI Registry performed between January 1, 2010 and December 31, 2010, were eligible to be included in the analysis, which included 641,178 PCIs. We excluded PCIs that could not be linked to a physician and patients who were younger than eighteen years of age at the time of the procedure. We considered only the first PCI performed during a hospitalization. After these exclusions were applied, a total of 510, 708 PCIs were linked to 5,175 associated physicians.

Figure 1.

Flow chart of patient and physician selection.

Physician Population

All physicians who performed at least 10 PCI procedures in 2010 as captured in the CathPCI Registry were eligible to be included in the analysis. We identified physicians in the CathPCI Registry using their National Provider Identifier (NPI) number. NPIs are voluntarily submitted by participating hospitals. There were 6,314 physicians with an NPI number recorded who performed a PCI in 2010. However, physicians were listed in the ABIM database using both NPI and unique physician identification numbers (UPINs). Accordingly, we used a NPI/UPIN crosswalk supplemented by manual searches to identify all physicians within the ABIM database. We excluded physicians who could not be linked with the ABIM data, had missing certification information in the ABIM database, or performed fewer than 10 PCIs in 2010. The final analysis included 5,175 physicians.

Physician Certification

For the primary analysis, physicians were divided into two groups: “certified” and “not certified” based on their ABIM ICARD certification status on January 1, 2010. In secondary analyses, we further stratified physicians by the year that they completed their CV fellowship training. We used CV fellowship as a marker across physician groups as it is a prerequisite for ICARD training. Those physicians who performed PCI prior to the formalization of ICARD training will not have interventional fellowship training dates recorded, but will have CV fellowship training dates available. Physicians who performed PCIs prior to the formation of accredited fellowship training programs were eligible for ICARD certification using the ABIM’s “practice pathway,” which required certification in CV and either a total of 500 lifetime PCIs or 150 PCIs in the previous two years. The first group included physicians who held ICARD certification in 2010 and had finished CV fellowship prior to 1999 (the first year of the ABIM ICARD certification; n = 2,200); the second group included physicians who held ICARD certification in 2010 and finished CV fellowship in 1999 or later (n = 1,466); the third group included physicians who had never held ICARD certification and who finished CV fellowship prior to 1999 (n = 1,044); the fourth group included physicians who never held ICARD certification and who finished CV fellowship in 1999 or later (n = 149); the fifth group included physicians who had previously held ICARD certification but whose certification had lapsed (n = 316).

Outcomes

The CathPCI Registry captures information about complications occurring during or following the PCI procedure up until hospital discharge²². Our two pre-specified primary endpoints were all-cause in-hospital mortality and bleeding complications. Bleeding complications consisted of suspected or confirmed bleeding events within 72 hours of the PCI associated with any of the following: a hemoglobin drop of ≥3 g/dl, transfusion of red blood cells, and/or an intervention at the bleeding site to reverse of stop the bleeding. Pre-specified secondary endpoints included need for emergency coronary artery bypass grafting (CABG), vascular complications requiring therapy (any access site occlusion, peripheral embolization, dissection, pseudoaneurysm, or AV fistula requiring intervention), and a composite endpoint of any outcome (death, bleeding, vascular complications, and emergency revascularization). These outcomes represent clinically significant, well-defined outcomes that have frequently been used in analyses of the CathPCI Registry.

Discharge Medications

In order to examine the association of ICARD certification with processes of care, we compared patterns of discharge medications- aspirin, statin therapy, and thienopyridines. We excluded patients with a contraindication to the medication. Furthermore, we excluded patients who did not undergo stent implantation from our assessment of thienopyridines.

Appropriate Use Criteria

In a secondary analysis, we examined differences in PCI procedural appropriateness as characterized by the 2012 appropriate use criteria for coronary revascularization (AUC)²³. Procedures were characterized as either appropriate, of uncertain appropriateness, inappropriate, or unmappable^24-25. Because prior analyses have shown that almost all acute PCIs were classified as appropriate²⁶, we excluded patients undergoing PCI in the setting of an acute coronary syndrome from AUC analysis.

Analysis

We summarized physician and patient characteristics including physician demographics, certification attributes, and baseline patient characteristics. To assess the relationship between certification status and outcomes, we used methods appropriate for clustered data^27-28. We used χ² tests, adjusted for clustering by physician, to compare patient characteristics and outcomes across physician groups²⁸. Then, to account for physician and patient characteristics, we estimated for each outcome a hierarchical generalized linear model with a logistic link function and a random intercept across physicians²⁸. Each model was adjusted for physician PCI volume, years since their initial certification in cardiovascular medicine (as a proxy for years of experience), the PCI volume of the hospital where the physician performed most of his or her PCIs, as well as patient characteristics. Physician and hospital PCI volume were calculated using procedures performed in 2010. Physician volume was included as a dichotomous variable in the model (less than 50 procedures, and 50 or greater procedures). To account for the fact that some physicians perform procedures at more than one hospital, we performed automated and manual abstraction of CathPCI data to identify operators who performed cases at more than one hospital submitting data to the CathPCI Registry. We included the hospital at which the provider performed the majority of his or her procedures. We also included in each model patient risk factors previously shown to be significantly associated with our outcomes of interest (including death, bleeding, emergency CABG, and vascular complications). For each model we reported the odds ratio (OR) and 95% confidence interval (CI) for the effect of certification status (or, for the secondary outcomes, certification group) on the outcome. To improve the interpretation of multilevel model results we used recycled predictions to express effects as absolute differences in outcome rates; this method uses the model results to predict the outcome for all patients as if they were in each exposure group, and summarizes the differences²⁹. In a secondary analysis, we included pairwise interaction terms between physician annual volume, hospital volume category, and ICARD status in the model. All analyses were performed using Stata 13.1 (StataCorp 2013, College Station TX). Analyses were approved by the Yale University School of Medicine’s Human Investigation Committee.

Results

Among the 6,314 eligible physicians, 717 were excluded because they performed fewer than 10 PCIs in 2010, 4 were excluded because they were enrolled in a fellowship program in 2010, 3 were excluded who received ICARD certification after 2010, and 415 had missing certification information. The remaining 5,175 interventional cardiologists performed a total of 510,708 PCI procedures between January 1, 2010 and December 31, 2010. The characteristics of the patients are summarized in Table 1 based on certification status of the performing physician. Overall, patient characteristics were similar across the certification status groups. Of note, the ICARD certified group had a slightly higher proportion of patients with heart failure (11.8% vs 11.1%), ST-segment elevation myocardial infarction (16.3% vs 15.5%), and acute coronary syndrome (70.2% vs 68.0%). The non-ICARD certified group had a higher proportion of patients with prior PCI (41.7% vs 40.3%), and patients undergoing elective PCI (48.9% vs 44.2%). Physician characteristics are shown in Table 2. Only 161 (3.2%) were female, and the mean age was 49.9 years. The mean age of noncertified physicians was 54.8 years compared with 48.0 in the certified group (P<0.001). A total of 3,666 (70.8%) held ICARD certification on January 1, 2010. ICARD certified physicians performed the majority of procedures (399,153 procedures; 78.2%). Among patients for whom information about practice type was available, a higher percentage of ICARD certified physicians practiced in an academic setting (7.8% vs 0.9%). On average, ICARD certified physicians performed more PCI in 2010 than non-ICARD certified physicians (111.8 PCI vs 75.8 PCI), and a higher proportion performed at least 50 PCI (77.6% vs 55.6, P<0.001).

Table 1.

Patient Characteristics by Physician Certification

Patient Characteristic	ICARD* certified, n (%)	Non-ICARD certified n (%)	p-value
Total PCI	399,153	111,555
Age (years, mean (SD))	64.7 (12.1)	64.8 (12.0)	0.048
Female	129,966 (32.6)	36,752 (32.9)	0.114
White	354,014 (88.7)	99,008 (88.8)	0.965
Black	31,483 (7.9)	8,686 (7.8)	0.965
Prior heart failure	47,160 (11.8)	12,332 (11.1)	0.002
Prior valve surgery	5,723 (1.4)	1,569 (1.4)	0.585
Prior CV† disease	48,999 (12.3)	13,227 (11.9)	0.030
Prior PCI‡	160,969 (40.3)	46,526 (41.7)	0.001
Hypertension	327,368 (82.0)	91,047 (81.6)	0.174
Current hemodialysis	9,347 (2.3)	2,331 (2.1)	0.002
Chronic lung disease	60,097 (15.1)	17,257 (15.5)	0.132
Diabetes	142,938 (35.8)	40,136 (36.0)	0.182
STEMI§	65,128 (16.3)	17,260 (15.5)	0.026
NSTEMI∥	185,517 (46.5)	48,536 (43.5)	<0.001
Acute coronary syndrome	280,277 (70.2)	175,849 (68.0)	0.001
Cardiogenic shock	7,628 (1.9)	1,845 (1.7)	0.001
NYHA# class I/II	16,167 (4.1)	3,656 (3.3)	<0.001
NYHA class III/IV	23,678 (5.9)	5,187 (4.6)	<0.001
Pre-procedure TIMI flow			<0.001
TIMI** -1	65,157 (16.3)	17,678 (15.8)
TIMI-2	31,872 (8.0)	12,101 (10.8)
TIMI-3	72,883 (18.3)	23,676 (21.2)
TIMI-4	227,190 (56.9)	57,426 (51.5)
Pre Procedure LVEF††			0.636
< 30	16,518 (4.1)	4,229 (3.8)
31 – 45	47,215 (11.8)	12,693 (11.4)
> 45	228,944 (57.4)	65,048 (58.3)
Missing	106476 (36.7)	29,686 (26.5)
PCI Status			<0.001
Elective	176,542 (44.2)	54,516 (48.9)
Urgent	151,753 (38.0)	38,207 (34.2)
Emergent	69,330 (17.4)	18,446 (16.5)
Salvage	1,292 (0.3)	342 (0.3)
Discharge prescriptions
Aspirin	375,937 (94.2)	103,527 (92.8)	<0.001
Statin	343,705 (86.1)	93,091 (83.4)	<0.001
Thienopyridine	351,137 (88.0)	96,597 (86.6)	<0.001

^*interventional cardiology;

^†cardiovascular disease;

^‡percutaneous coronary intervention;

^§ST segment elevation myocardial infarction;

^∥non-ST segment elevation myocardial infarction;

^#New York Heart Association;

^**thrombolysis in myocardial infarction;

^††left ventricular ejection fraction.

Table 2.

Physician Characteristics (n = 5,175)

	Non-ICARD* certified	ICARD certified	Total	P-value
	n (%)	n (%)	n (%)
Total # of Physicians	1509 (29.2)	3666 (70.8)	5175 (100.0)
Gender				0.373
Male	1466 (97.2)	3544 (96.7)	5010 (96.8)
Female	43 (2.8)	122 (3.3)	165 (3.2)
Type of Practice				<0.001
Teaching	14 (0.9)	287 (7.8)	301 (5.8)
Solo or Group	455 (30.2)	1946 (53.1)	2401 (46.4)
HMO/Other	29 (1.9)	177 (4.8)	206 (4.0)
Hospital Practice	99 (6.6)	604 (16.5)	703 (13.6)
Missing	912 (60.4)	652 (17.8)	1564 (30.2)
Volume category				<0.001
<50	670 (44.4)	822 (22.4)	1492 (28.8)
50+	839 (55.6)	2844 (77.6)	3683 (71.2)
PCI volume				<0.001
Mean (sd)	75.8 (75.8)	111.8 (85.3)	101.3 (84.3)
Certification Status
Yes			3,666 (70.8)
Yes, CV < 1999			2,200 (42.5)
Yes, CV ≥ 1999			1,466 (28.3)
No, CV < 1999			1,044 (20.2)
No, CV ≥ 1999			149 (2.9)
Lapsed			316 (6.1)
Years since CV† fellowship‡				<0.001
0-5yrs	52 (3.4)	555 (15.1)	607 (11.7)
6-10yrs	81 (5.4)	762 (20.8)	843 (16.3)
11-15yrs	185 (12.2)	730 (19.9)	915 (17.7)
16-20yrs	284 (18.8)	578 (15.8)	862 (16.7)
21-30yrs	659 (43.7)	897 (24.5)	1556 (30.1)
30+yrs	247 (16.4)	144 (3.9)	391 (7.6)
Hospital PCI Volume				0.016
1-199	129 (8.5)	283 (7.7)	412 (8.0)
200-399	286 (19.0)	588 (16.0)	874 (16.9)
400+	1094 (72.5)	2795 (76.2)	3889 (75.1)
Age (yrs) on 1/1/2010				<0.001
mean (sd)	54.8 (7.3)	48.0 (8.0)	49.9 (8.4)

^*interventional cardiology;

^†cardiovascular disease;

^‡years since end date of general cardiovascular disease fellowship.

The proportions of patients discharged on aspirin, thienopyridines, and statins following the PCI procedure are also shown in Table 1. There were relatively small but statistically significant differences such that patients treated by certified physicians were more likely to be discharged on each medication. The proportions of appropriate procedures by certification status are shown in Figure 2. A higher proportion of PCIs performed by non-ICARD certified physicians were not mappable compared with PCIs performed by ICARD certified physicians (42.9% vs 39.9%, p=0.002). Among procedures that were mappable, higher proportions of PCI performed by ICARD certified physicians were classified as inappropriate (13.1% vs 11.8%, p=0.002) and appropriate (24.7% vs 23.1%, p=0.038). The proportions of procedures classified as uncertain were similar (22.2% vs 22.3%, p=0.934).

Figure 2.

Procedural Appropriateness of Non-acute PCI Stratified by Physician Certification.

In the secondary analysis that stratified physicians based on when they finished CV training, Table 2 shows that 20.2% of physicians had never been ICARD certified and finished CV fellowship prior to 1999. These physicians performed 14.2% (72,566 procedures) of all PCIs included in the analysis. A smaller proportion of PCIs (11,849 procedures; 2.3%) were performed by the 148 physicians (2.9%) who had never been ICARD certified and had finished CV fellowship in 1999 or later. Only 309 physicians (6.1%) had a lapsed certification during 2010, and these physicians performed 27,140 procedures (5.3% of the total number of procedures included in the analysis). There were modest differences in patient characteristics and discharge medications across these five physician categories.

Outcomes

Overall, the crude outcomes of patients treated by ICARD certified physicians were almost identical to those of non-ICARD certified physicians (Table 3). However, in multivariable analyses, the risks of both mortality and emergency CABG were statistically significantly higher for patients treated by non-ICARD board certified physicians (Table 4). As estimated by recycled predictions, this corresponded to absolute increases in mortality and emergency CABG of 0.08% and 0.03% respectively. In contrast, there were no significant differences in the adjusted risks of bleeding, vascular complications, and the composite endpoint of any adverse outcome. In a secondary analysis including interaction terms between physician annual volume category, hospital volume category, and ICARD status, only emergency CABG remained statistically significant.

Table 3.

Crude Outcomes of Primary and Secondary Analyses

Outcome	Non-ICARD* n, (%)	ICARD n, (%)	ICARD, CV†<1999 n (%)	ICARD, CV≥1999 n (%)	Non-ICARD, CV<1999 n (%)	Non-ICARD, CV≥1999 n (%)	Lapsed n (%)
	111,555 (21.8)	399,153 (78.2)	242,452 (47.5)	156,701 (30.7)	72,566 (14.2)	11,849 (2.3)	27,140 (5.3)
In-hospital death	1,509 (1.4)	5,436 (1.4)	3,245 (1.3)	2,191 (1.4)	1,016 (1.4)	173 (1.5)	320 (1.2)
Bleeding	1,878 (1.7)	7,116 (1.8)	4,321 (1.8)	2,795 (1.8)	1,232 (1.7)	190 (1.6)	456 (1.7)
Vascular complication	475 (0.4)	1,851 (0.5)	1,084 (0.4)	767 (0.5)	296 (0.4)	57 (0.5)	122 (0.4)
Emergency CABG‡	255 (0.2)	707 (0.2)	405 (0.2)	302 (0.2)	174 (0.2)	32 (0.3)	49 (0.2)
Composite§	3,740 (3.4)	13,659 (3.4)	8,156 (3.4)	5,503 (3.5)	2,473 (3.4)	408 (3.4)	859 (3.2)

^*interventional cardiology;

^†cardiovascular disease fellowship;

^‡coronary artery bypass graft;

^§composite endpoint of any adverse event (including in-hospital death, bleeding, vascular complication, and emergency CABG).

Table 4.

Multivariable Outcomes of Primary and Secondary Analyses

Outcome	ICARD* Certified	Non-ICARD (OR, 95% CI)§	P-value	ICARD, CV†<1999∥	ICARD, CV≥1999 (OR, 95% CI)	Non-ICARD, CV<1999 (OR, 95% CI)	Non-ICARD, CV≥1999 (OR, 95% CI)	Lapsed (OR, 95% CI)
In-hospital death	ref	1.10 (1.02, 1.19)	0.018	ref	0.94 (0.78, 1.14)	1.12 (1.02, 1.24)	1.17 (0.90, 1.52)	0.91 (0.79, 1.06)
Bleeding	ref	0.96 (0.89, 1.04)	0.322	ref	0.88 (0.73, 1.07)	0.94 (0.86, 1.04)	0.77 (0.59, 1.02)	0.96 (0.85, 1.13)
Vascular complication	ref	0.98 (0.87, 1.11)	0.784	ref	0.90 (0.68, 1.18)	0.93 (0.80, 1.08)	0.87 (0.59, 1.29)	1.07 (0.86, 1.33)
Emergency CABG‡	ref	1.32 (1.12, 1.56)	0.001	ref	1.32 (0.91, 1.91)	1.29 (1.06, 1.57)	1.86 (1.10, 3.09)	0.97 (0.70, 1.34)
Composite endpoint	ref	1.03 (0.98, 1.09)	0.276	ref	0.96 (0.84, 1.09)	1.03 (0.96, 1.10)	0.97 (0.80, 1.17)	0.97 (0.87, 1.07)

^*interventional cardiology;

^†cardiovascular disease fellowship;

^‡coronary artery bypass graft;

^§reference group is ICARD certified group in the primary analysis;

^∥reference group is ICARD < 1999 certified group in the secondary analysis.

Findings were similar when we classified physicians into the five categories (Table 3). In multivariable analyses, the risk of death was significantly higher in patients treated by non-ICARD certified physicians who finished CV fellowship prior to 1999, and the risk of emergency CABG was significantly higher for patients treated by both non-ICARD certified groups (Table 4). Again there were no significant differences in the adjusted risks of bleeding, vascular complications, and the composite endpoint of any adverse outcome.

Discussion

In the United States in 2010, interventional cardiologists who did not hold ICARD certification performed a substantial proportion of PCI procedures. This included both physicians who had never been ICARD certified and those who had allowed their certification to lapse. We found that the care and outcomes of patients undergoing PCI procedures were generally similar regardless of the ICARD certification status of the performing physician. After adjusting for patient characteristics and PCI volume, the risks of mortality and emergency CABG were statistically higher among non-ICARD certified physicians compared with ICARD certified physicians. However, the overall event rates were low, and the clinical significance of these differences may be modest.

The requirements for obtaining certification in ICARD are rigorous. Given this rigor, it is notable that the majority of practicing interventional cardiologists have obtained ABIM ICARD certification. This is particularly striking among physicians who completed their CV training since ICARD certification was introduced in 1999. In this group, certification has become the norm, and only 2.9% were not certified in 2010. The reasons why physicians choose to become certified are probably multifactorial, but surveys of physicians suggest that both professional pride and hospital credentialing policies play major roles³⁰. As noted, ABIM certification has been symbolic of professionalism in medicine, and has also been seen as indicative of a commitment to excellence and continued learning on the part of the physician.

In the present study, we did not observe a consistent association between ICARD certification and patient outcomes. Although patients treated by non-ICARD certified physicians were at statistically significantly higher risk of both mortality and emergency CABG, we did not see a similar association for the clinical endpoints of bleeding, vascular complications, and a composite of any adverse outcome. Furthermore, the clinical significance of the mortality and CABG findings is subject to interpretation. The adjusted absolute increase in mortality was 0.08%, which corresponds to 1 additional death for every 1250 patients. Similarly there was 1 additional emergency bypass surgery for every 3333 patients treated by non-certified physicians, when compared with certified physicians. Finally, a smaller proportion of non-acute PCI procedures performed by non-ICARD certified physicians were classified as inappropriate compared with procedures performed by ICARD certified physicians. Collectively, our findings suggest that ICARD certification status alone is not a strong predictor of the quality of care and in-hospital outcomes.

There are several potential explanations for this finding. First, PCI is a much safer and more reliable procedure today than in its early development. As prior studies have demonstrated, the introduction of coronary stents greatly improved procedural success and reduced the need for emergency CABG^31-32. Similarly, increasing use of radial access and novel anticoagulant and antithrombotic strategies have likely reduced the risks of periprocedural bleeding and myocardial infarction^33-34. In addition, there is increasing recognition that the outcomes of patients undergoing PCI are less attributable to an individual provider and more reflective of the totality of care delivered by the team treating the patient both during and after the PCI procedure³⁵. Collectively, these factors may have had the effect of leveling the playing field, making it possible for interventional cardiologists with different training, knowledge, and technical expertise to achieve comparable results.

Second, the group of non-ICARD certified physicians is heterogeneous, consisting of physicians who did and did not complete an accredited fellowship in ICARD. In our analyses, we did not identify a significant interaction between physician training year and certification status, but we observed that the risks of mortality and emergency CABG were highest among physicians without ICARD certification who completed their CV training after 1999. Further research is warranted to determine whether the differences between ICARD certified and non-ICARD certified physicians will become more apparent over time.

Third, in spite of the rigor of the ABIM’s certification requirements, it is possible that the qualities and abilities currently captured by the certification process may not be the same as those needed to discriminate between physicians who perform PCI. For example, the certification exam can be used to assess an individual’s knowledge, but it cannot test for many of the qualities that are associated with highly skilled proceduralists, including the ability to make good decisions under extreme stress, superior manual dexterity, and quickly recognizing and effectively treating procedural complications. One promising approach to better assess this skillset could be the use of simulators, which could be folded into fellowship training and perhaps incorporated into the certification process³⁶. Alternatively, there may be a role for direct observation of procedural skills by interventionalists specifically trained in observation and assessment of procedural skills³⁷.

Our findings must be interpreted in the context of several limitations. Our analysis examined only in-hospital outcomes, and it is possible that differences in the procedural quality may only become evident over longer periods of time or when considering other PCI-relevant outcomes including quality of life. We only considered patients who had a PCI, and could not assess whether certification status was associated with differences in the overall management of patients with acute and stable coronary artery disease. Nevertheless, the characteristics and measures of procedural appropriateness were similar across the two groups. The clinical endpoints that we examined are relatively infrequent in occurrence, and therefore may have made it difficult to find variation across operators. In addition, our study reflects the outcomes of patients undergoing PCI in 2010, and it is possible that the underlying associations between ICARD certification status and outcomes may be different in more recent data. However, studies using more recent CathPCI data show similar rates of complications³⁸.

The ABIM identified 6,172 physicians who held ICARD certification in 2010, but we were unable to identify certification status for 16.2% (997) of these physicians in the linked dataset. This likely includes physicians who were practicing at non-NCDR hospitals and physicians for whom the NPI number was not available (as this was not a mandatory field on the NCDR data reporting sheet). Nevertheless, the case mix of physicians who could and could not be matched were comparable, suggesting that our findings are generalizable. Our study was also limited to hospitals that participate in the NCDR CathPCI Registry. The primary goal of the NCDR is to drive quality improvement through measurement and feedback and substantial research has demonstrated the power and importance of feedback for professional development³⁹. It is possible that physician participation in NCDR-related activities may mitigate the association of certification with PCI outcomes. As noted, we assessed only ICARD certification at the time of the procedure, and did not evaluate whether or not participation or completion of the maintenance of certification program impacts PCI outcomes. Finally, the development of ICARD certification was a relatively recent event, and most if not all non-ICARD physicians participated in a training program even if it was not formally recognized. As such, our findings may not be generalizable to other specialties.

In conclusion, we found that the outcomes of patients undergoing PCI were excellent and varied modestly depending on the certification status of the performing physician. Our findings suggest there is an opportunity to enhance the value of subspecialty certification.