Trends in cardiac surgery and percutaneous interventions in New York: a statewide registry analysis (2010–2019)

Abstract

Background

Contemporary practice in coronary and valve interventions continues to evolve with changing indications, technology, and systems of care. I characterized statewide procedure volumes and mortality for PCI, CABG, Valve ± CABG, and TAVR using a unified, internally consistent analytic framework.

Methods

I analyzed New York State public-use registry data (2010–2019). All PCI and Non-Emergency PCI are reported annually. Emergency PCI was derived annually at the hospital-year level as All PCI − Non-Emergency PCI; deaths and expected deaths were derived by subtraction and aggregated statewide. CABG is reported annually. Valve ± CABG and TAVR are provided as overlapping 3-year windows; I produced annualized values by averaging the per-year contribution from the two windows that include each year (single window at edges; TAVR available 2013–2019). Statewide Observed %, Expected %, O/E, and Risk-Adjusted % were computed as case-weighted aggregates. No hypothesis testing was performed.

Results

From 2010 to 2019, there were 1,005,980 PCI and 171,182 CABG procedures statewide. Overall PCI volume was stable (2010: 108,070; 2019: 108,552). Within PCI, Non-Emergency comprised 835,480 (83.1%) and Emergency 170,500 (16.9%); non-emergency PCI declined modestly (2010: 93,498 → 2019: 89,654), while emergency PCI increased (2010: 14,572 → 2019: 18,898). CABG volumes were broadly stable (2010: 18,842 → 2019: 17,876). Annualized Valve ± CABG volumes declined (≈ 14,822 in 2010 → 11,893 in 2019), whereas TAVR expanded after introduction (2013: 3,703 → 2019: 9,963). Pooled risk-adjusted mortality: All PCI 1.11%, Non-Emergency PCI 0.72%, Emergency PCI 3.07%, CABG 1.53%, Valve ± CABG 3.27% (2010–2019), and TAVR 2.93% (2013–2019). Across procedures and years, O/E ≈ 1.00, indicating good model calibration.

Conclusions

Between 2010 and 2019, statewide PCI volume was stable; non-emergency PCI declined modestly, emergency PCI rose modestly, CABG volumes were broadly stable, Valve ± CABG decreased, and TAVR increased substantially with improving mortality. Overall PCI mortality remained low and largely stable, consistent with higher-risk case mix over time. These contemporary benchmarks can inform quality improvement, capacity planning, and policy while highlighting the need for continued monitoring of high-risk PCI pathways and long-term TAVR durability in younger patients.

Introduction

Over the past decade, coronary revascularization practice has shifted markedly under evolving evidence and guidelines. Observational registry data show that total revascularization volumes declined after the mid-2000s, with CABG (Coronary Artery Bypass Grafting) use falling more steeply than PCI (Percutaneous Coronary Intervention) [1, 2]. For example, a Washington State registry found that annual CABG volume fell ≈ 22.6% from 2005 to 2017, whereas PCI volume fell 19.1% (2005–2014) and then stabilized or rose thereafter [2]. Consistently, U.S. Medicare analyses report a larger relative decline in CABG than PCI over a similar period [1]. These trends have coincided with an increasingly older, higher-risk PCI population: patients undergoing PCI in 2014–17 were more often > 80 years old and had higher rates of diabetes, renal disease, and heart failure than in 2005–09 [2]. Likewise, a nationwide cohort in Korea observed rising PCI volumes (∼32,000 to 52,000 annual cases from 2006 to 2015) and an increasing proportion of high-risk patients (with 1 year mortality rising from 15.5 to 19.4% in the highest-risk quartile) [3]. These findings mirror a shift toward conservative management of stable CAD: major trials (e.g. ISCHEMIA) showed no reduction in myocardial infarction or death with routine invasive strategy over optimal medical therapy [4]and current guidelines underscore that revascularization should be reserved for selected patients with symptomatic or prognostically significant disease [4, 5].

Meanwhile, transcatheter valve therapies have transformed management of aortic stenosis. Landmark randomized trials extended TAVR (Transcatheter Aortic Valve Replacement) from inoperable to intermediate and low-risk patients: for example, Leon et al. showed that at 2 years the composite of death or disabling stroke was similar for TAVR versus SAVR (Surgical Aortic Valve Replacement) in intermediate-risk patients [6]and Mack et al. found a significantly lower 1-year rate of death, stroke, or rehospitalization with transfemoral TAVR than surgery in low-risk patients (8.5% vs. 15.1%; HR 0.54) [7]. These data are reflected in practice: national registry reports document yearly growth in TAVR use, with TAVR volume surpassing surgical AVR by 2019 [8]. Indeed, population analyses suggest that TAVR’s arrival has driven a ≈ 60% increase in overall aortic-valve replacement volume among older adults [9]. Concomitantly, procedure outcomes have improved: for instance, the STS/ACC TVT registry showed 30-day TAVR mortality falling from 7.2% in 2011 to 2.5% in 2019 [8]. In sum, transcatheter technologies and guidelines favoring less-invasive AVR have markedly altered valvular intervention patterns, with many older patients now treated via catheter rather than open surgery.

Given these shifts, contemporary registry data are needed to quantify trends in procedure rates, patient risk profiles, and outcomes across revascularization and valvular interventions. In this study I analyze statewide U.S. registry data (2010–2019) to track volumes of PCI, CABG, TAVR, and surgical valve procedures (± CABG) over time, as well as changes in patient demographics and observed mortality. By comparing these real-world trends with major trial findings and guideline changes [2, 8]I aim to elucidate how evidence and technology have reshaped cardiovascular care delivery and outcomes in this period.

Methods

Study design and data source

I conducted a retrospective, population-based study using the New York State (NYS) Adult Cardiac Surgery and Percutaneous Coronary Interventions Registry, which mandatorily captures all cardiac surgical and coronary/structural catheter-based procedures performed in non-Federal hospitals statewide [10]. The registry is publicly released by the NYS Department of Health (DOH) with a data use policy statement [11]. The public use file include, for each hospital-year and procedure category, total cases, in-hospital deaths, and registry-derived expected and risk-adjusted mortality rates (All PCI, Emergency PCI, Non-Emergency PCI, isolated CABG, Valve surgery with or without CABG, and TAVR). Analyses were restricted to January 1, 2010 through December 31, 2019 (TAVR available 2013–2019 in annualized form; see below). The analysis was conducted independently using de-identified data released by the NYS Department of Health; institutional review board oversight was not required.

Data contains discharge-only outcomes and lacks readmission or long-term follow-up. Data file 2008–2019 was downloaded, concatenated, and analysed in accordance with STROBE guidelines for observational studies. Data from 2008 to 2009 included only emergency PCI and Valve or Valve/CABG procedures within the period window 2008–2010, so they were excluded from the study. Institutional review board oversight was not required because all data are de-identified and publicly available.

Case aggregation and data cleaning

I first collapsed hospital-level records to statewide totals by Procedure × Year. Facility identifiers and geographic columns were dropped prior to aggregation. For each stratum, I computed:

• Cases = Σ cases across hospitals.

• Deaths = Σ deaths.

• Observed mortality (%) = 100 × (Deaths ÷ Cases).

• Expected mortality (%) (statewide, case-weighted) = 100 × [Σ(Expected rate × Cases) ÷ Σ(Cases)]

• Risk-adjusted mortality (%) (statewide, case-weighted) = Σ(RA rate × Cases) ÷ Σ(Cases).

Emergency PCI (annual, derived by subtraction)

Because PCI partitions into All PCI = Non-Emergency PCI + Emergency PCI, I derived Emergency PCI annually at the hospital-year level:

• Cases_Emerg = Cases_All − Cases_Non-Em.

• Deaths_Emerg = Deaths_All − Deaths_Non-Em.

• ExpectedDeaths_Emerg = (ExpRate_All × Cases_All) − (ExpRate_Non-Em × Cases_Non-Em).

Statewide annual totals were then obtained by summing the hospital-year values. I did not subtract percentages. All rates were recomputed from aggregated counts.

The registry publishes Emergency PCI as overlapping 3-year windows. For comparability, I additionally produced annualized estimates by averaging the per-year contribution from the two windows that include a given year (single window at edges). These annualized Emergency PCI values closely matched the subtraction-based annual series: window roll-ups matched exactly on cases and deaths for non-overlapping windows, and year-specific observed/expected mortality differed only trivially, consistent with rounding and the smoothing inherent to 3-year windows. I therefore present the subtraction-derived annual series as primary and use the window-annualized series as a sensitivity check.

Valve or valve/cabg and TAVR (annualization from overlapping 3-year windows)

The registry reports Valve ± CABG and TAVR as overlapping 3-year windows. To obtain annual series for 2010–2019 (TAVR 2013–2019), I used the prespecified averaging rule:

• For interior year y, take the two windows that include it—(y − 2…y) and (y…y + 2)—divide each window’s Cases, Deaths, and Expected deaths by 3, then average the two per-year contributions.

• At edges with one available window (e.g., early TAVR years), use the single contributing window divided by 3.

• From the resulting annual totals I recomputed Observed %, Expected %, and O/E.

This procedure yields annualized estimates and is exact for linear trends within windows; pooled totals in Table 1 were subsequently computed by summing annualized values over 2010–2019 (TAVR 2013–2019).

Table 1.

Overall statewide aggregated procedure summary between 2008–2019

Procedure	Period used in pooled total	Cases	Deaths	Observed %	Expected %	O/E	Risk-Adjusted %
All PCI	2010–2019 (annual)	1,005,980	11,224	1.12	1.12	1	1.11
Non-Emergency PCI	2010–2019 (annual)	835,480	6,022	0.72	0.72	1	0.72
Emergency PCI	2010–2019 (annual)	170,500	5,202	3.05	3.06	1	3.07
CABG	2010–2019 (annual)	171,182	2,614	1.53	1.53	1	1.53
Valve or Valve/CABG	2010–2019 (annual)	138,953	4,485	3.23	3.23	1	3.27
TAVR	2013–2019 (annual)	47,527	1,365	2.87	2.87	1	2.93

Pooled summaries (Table 1) and internal checks

Pooled, case-weighted statewide rates for 2010–2019 were computed for All PCI, Non-Emergency PCI, Emergency PCI (derived annually), CABG, and Valve ± CABG; TAVR pooled 2013–2019. As a consistency check, All PCI = Non-Emergency PCI + Emergency PCI was verified for each year (cases and deaths), and for every annual stratum Observed %, Expected %, and O/E were recomputed from totals.

Outcome definitions

The registry defines in-hospital mortality as death during the index admission (or ≤ 30 days if still hospitalized). Expected mortality is produced by NYS DOH logistic-regression models using demographic, comorbidity, hemodynamic, and anatomical predictors. Risk-adjusted mortality at the centre level is defined as (Observed/Expected) × the statewide mean mortality for that procedure-year; statewide RA% is the case-weighted mean of centre-level RA rates [12].

For Emergency PCI, hospital-level O/E was computed from derived Deaths_Emerg and ExpectedDeaths_Emerg; if a statewide Emergency RA% was needed, I used the manuscript’s definition (centre-level O/E × the statewide Emergency mean in that year), then case-weighted across centres to obtain the statewide RA%.

Statistical analysis

Analyses were performed in Python (pandas for data wrangling; matplotlib for graphics). Descriptive statistics are presented as counts, proportions, or case-weighted means. Because the dataset enumerates the statewide treated population, no hypothesis tests or confidence intervals were required for primary endpoints; temporal patterns were assessed graphically. Sensitivity checks confirmed that pooled PCI identities held annually after Emergency derivation and that Valve ± CABG/TAVR pooled results were robust to the annualization step (no overlapping windows were ever summed).

Results

Procedural activity

From 2010 to 2019, the registry recorded 1,005,980 PCI and 171,182 CABG procedures statewide, alongside substantial volumes of Valve or Valve/CABG and TAVR (annualized from overlapping 3-year windows per Methods). Overall PCI volume was stable (2010: 108,070; 2019: 108,552). Within PCI, non-emergency cases comprised 83.1% (835,480/1,005,980) and emergency cases 16.9% (170,500/1,005,980). Emergency PCI increased over time (2010: 14,572; 2019: 18,898), while non-emergency PCI was broadly flat to slightly lower (2010: 93,498; 2019: 89,654). CABG volumes were relatively steady (2010: 18,842; 2019: 17,876). Annualized Valve or Valve/CABG counts declined (≈ 14,822 in 2010 to 11,893 in 2019), whereas TAVR expanded following introduction (2013: 3,703 to 2019: 9,963) (See Table 2 for PCI; Table 3 for CABG, Valve ± CABG, and TAVR; Fig. 1).

Table 2.

Aggregated PCI Procedure-Year summary 2010–2019

Procedure	Year of Hospital Discharge	Number of Cases	Number of Deaths	Observed Mortality Rate (aggregate)	Expected Mortality Rate (weighted)	Risk-Adjusted Mortality Rate (weighted)
All PCI	2010	108,070	908	0.84	0.84	0.84
All PCI	2011	100,474	974	0.97	0.97	0.97
All PCI	2012	94,090	942	1	1	0.99
All PCI	2013	95,930	1,096	1.14	1.14	1.14
All PCI	2014	94,392	1,110	1.18	1.18	1.19
All PCI	2015	98,070	1,116	1.14	1.14	1.13
All PCI	2016	100,674	1,246	1.24	1.24	1.25
All PCI	2017	102,858	1,162	1.13	1.13	1.13
All PCI	2018	102,870	1,416	1.38	1.38	1.36
All PCI	2019	108,552	1,254	1.16	1.16	1.15
Non-Emergency	2010	93,498	478	0.51	0.51	0.51
Non-Emergency	2011	85,308	558	0.65	0.65	0.65
Non-Emergency	2012	77,524	498	0.64	0.64	0.64
Non-Emergency	2013	78,782	582	0.74	0.74	0.74
Non-Emergency	2014	77,316	574	0.74	0.74	0.75
Non-Emergency	2015	80,824	600	0.74	0.74	0.75
Non-Emergency	2016	83,344	696	0.84	0.84	0.85
Non-Emergency	2017	84,770	662	0.78	0.78	0.78
Non-Emergency	2018	84,460	738	0.87	0.87	0.87
Non-Emergency	2019	89,654	636	0.71	0.71	0.7
Emergency	2010	14,572	430	2.95	2.95	3.02
Emergency	2011	15,166	416	2.74	2.76	2.74
Emergency	2012	16,566	444	2.68	2.68	2.66
Emergency	2013	17,148	514	3	2.99	3.06
Emergency	2014	17,076	536	3.14	3.16	3.14
Emergency	2015	17,246	516	2.99	3	3
Emergency	2016	17,330	550	3.17	3.17	3.21
Emergency	2017	18,088	500	2.76	2.77	2.77
Emergency	2018	18,410	678	3.68	3.7	3.69
Emergency	2019	18,898	618	3.27	3.28	3.32

Table 3.

Aggregated cardiac surgery Procedure-Year summary 2010–2019

Procedure	Year of Hospital Discharge	Number of Cases	Number of Deaths	Observed Mortality Rate (aggregate)	Expected Mortality Rate (weighted)	Risk-Adjusted Mortality Rate (weighted)
CABG	2010	18,842	298	1.58	1.58	1.6
CABG	2011	17,254	214	1.24	1.24	1.26
CABG	2012	16,284	238	1.46	1.46	1.46
CABG	2013	16,336	300	1.84	1.84	1.85
CABG	2014	15,884	194	1.22	1.22	1.24
CABG	2015	16,712	260	1.56	1.56	1.54
CABG	2016	17,386	290	1.67	1.67	1.66
CABG	2017	17,564	278	1.58	1.58	1.6
CABG	2018	17,044	252	1.48	1.48	1.5
CABG	2019	17,876	290	1.62	1.62	1.61
Valve or Valve/CABG	2010	14,798	617	4.17	4.17	4.18
Valve or Valve/CABG	2011	14,820	569	3.84	3.84	3.87
Valve or Valve/CABG	2012	14,825	513	3.46	3.46	3.5
Valve or Valve/CABG	2013	14,781	479	3.24	3.24	3.28
Valve or Valve/CABG	2014	14,615	460	3.15	3.15	3.21
Valve or Valve/CABG	2015	14,165	427	3.02	3.02	3.07
Valve or Valve/CABG	2016	13,558	402	2.97	2.97	3.02
Valve or Valve/CABG	2017	12,736	356	2.8	2.8	2.84
Valve or Valve/CABG	2018	12,762	357	2.8	2.8	2.87
Valve or Valve/CABG	2019	11,893	305	2.57	2.57	2.63
TAVR	2013	3,703	176	4.75	4.75	4.78
TAVR	2014	5,116	184	3.6	3.6	3.63
TAVR	2015	5,239	182	3.47	3.47	3.77
TAVR	2016	6,751	194	2.87	2.87	3.02
TAVR	2017	8,369	205	2.45	2.44	2.49
TAVR	2018	8,386	203	2.42	2.42	2.4
TAVR	2019	9,963	221	2.22	2.22	2.21

Fig. 1.

Annual case volumes (2010–2019). Overall PCI is stable; the emergency PCI share rises modestly as non-emergency PCI plateaus/slightly declines; CABG is broadly stable; valve interventions shift toward rapid TAVR growth with a concurrent decline in valve ± CABG

Percutaneous coronary intervention (PCI)

All PCI pooled risk-adjusted mortality was 1.11% over 2010–2019. Non-emergency PCI pooled risk-adjusted mortality was 0.72%. Emergency PCI was derived annually at the hospital-year level as All PCI − Non-Emergency PCI; deaths and expected deaths were derived by subtraction and aggregated statewide. Emergency PCI rose from 14,572 cases (430 deaths) in 2010 to 18,898 (618 deaths) in 2019; pooled risk-adjusted mortality was 3.07% (Tables 1 and 2; Figs. 2 and 3).

Fig. 2.

Annual risk-adjusted mortality (2010–2019; TAVR from 2013). Valve ± CABG and TAVR show sustained declines; non-emergency PCI remains low; CABG varies within a narrow band; emergency PCI is higher with a modest mid-decade uptick followed by a plateau

Fig. 3.

Pooled risk-adjusted mortality by procedure (2010–2019; TAVR from 2013). A clear gradient is seen: non-emergency PCI is lowest, all PCI remains low, CABG is intermediate, TAVR is higher, and valve ± CABG and emergency PCI are highest

Coronary artery bypass grafting (CABG)

CABG activity was broadly stable through the decade. Pooled risk-adjusted mortality was 1.53% over 2010–2019. Observed and expected mortality percentages closely tracked one another year by year (Tables 1 and 3; Figs. 2 and 3).

Valve surgery and transcatheter therapy

Valve or Valve/CABG annual values (from overlapping 3-year windows) showed decreasing volumes and improving mortality; pooled risk-adjusted mortality over 2010–2019 was 3.27%. TAVR volumes increased substantially from 2013 onward, with pooled risk-adjusted mortality 2.93% (2013–2019). Year-by-year risk-adjusted mortality declined for both series (e.g., Valve ± CABG from ~ 4.2% in 2010 to ~ 2.6% in 2019; TAVR from ~ 4.8% in 2013 to ~ 2.2% in 2019) (Tables 1 and 3; Figs. 2 and 3).

Risk-adjusted mortality (pooled and annual)

Figure 3 summarizes pooled, case-weighted risk-adjusted mortality: Non-Emergency PCI 0.72%, All PCI 1.11%, CABG 1.53%, TAVR 2.93%, Emergency PCI 3.07%, Valve or Valve/CABG 3.27%. Annual trajectories in Fig. 2 show stable or improving patterns across modalities, with the most pronounced declines for TAVR and Valve ± CABG.

Calibration (observed-to-expected)

Across procedures and years, O/E ≈ 1.00, indicating close agreement between observed and expected mortality. This pattern held for All PCI, Non-Emergency PCI, Emergency PCI (derived), CABG, Valve ± CABG, and TAVR (Table 1; Fig. 2). No hypothesis testing was performed.

Internal consistency checks

• For every year, All PCI = Non-Emergency PCI + Emergency PCI for both cases and deaths (Table 2).

• Annual Valve ± CABG and TAVR series were produced from overlapping windows as prespecified; pooled summaries in Table 1 use unobjectionable, non-overlapping accounting to prevent double counting.

• No imputation or forward-filling was used; gaps reflect registry availability (e.g., TAVR begins 2013).

Discussion

In this statewide registry analysis (2010–2019), I observed a two-phase pattern in PCI volumes. From 2010 to 2014, all PCI declined by ~ 13% (≈ 108k→≈94k), driven by a ~ 16–17% reduction in non-emergency (elective) PCI (≈ 92k→≈77k), while Emergency PCI rose modestly (≈ 15k→≈17k), increasing its share of all PCI from ~ 14% to ~ 18%. From 2014 to 2019, volumes rebounded: all PCI returned to ≈ 109k and non-emergency PCI to ≈ 90k, while Emergency PCI continued a mild increase to ≈ 19k. This pattern aligns with national reports—e.g., Almarzooq et al. noted a ~ 10% decline in overall PCI from 2010 to 2017 largely attributable to a ~ 34% fall in elective PCI [13, 14]. These shifts are plausibly related to evolving guidance and neutral trial results in stable ischemic heart disease (e.g., appropriate-use criteria; ISCHEMIA), which favored conservative management [15].

In risk-adjusted mortality after PCI, our data showed little change or slight worsening over time – again aligning with external reports. National data 2003–2016 found risk-adjusted PCI mortality essentially flat, with only minor rises in certain subgroups (STEMI: 4.9%→5.3%; UA/SIHD: 0.8%→1.0%)[16]. Likewise, state data from Washington showed increased risk-adjusted PCI mortality over time [2]. Possible drivers include higher-risk patients (e.g. older, more STEMI/NSTEMI, shock) now undergoing PCI. Indeed, Kataruka et al. reported more STEMI and comorbid presentations in the later era [2]which mirrors our registry’s rise in STEMI/urgent case burden. Thus, even as devices and techniques improved, the higher acuity of PCI cases appears to have offset mortality gains, consistent with US Medicare observations [14, 16]. In short, our findings of stable or slightly higher PCI mortality rates align with multi-database analyses [2, 16].

By contrast, CABG volumes were broadly stable. This observation is congruent with several reports from both the U.S. and abroad. For example, U.S. Medicare saw > 18,000 fewer CABGs in 2008–2012 [14], Washington state reported a 22.6% decline from 2005 to 2017 [1], and UK centers documented ongoing volume decreases into 2016 [17]. In our state, CABG case numbers were broadly stable over the decade. The relative stability of CABG likely reflects multiple factors: improved medical therapy, selective use of PCI for less complex disease, and the aging population with varying surgical risk.

Importantly, risk-adjusted mortality after CABG remained low and stable over the study period. Prior studies found that surgical outcomes steadily improved. U.S. inpatients showed significant mortality reductions for CABG between 2003 and 2016 (e.g. MI case mortality 5.6%→3.4%; non-MI 2.8%→1.7%)¹⁶. Similarly, Washington’s CABG mortality fell over 2005–2017 [2], and the UK’s reported isolated-CABG mortality reached ~ 1.0% (elective ~ 0.6%) by 2016 [17]. Our statewide values were low and stable, consistent with improved surgical techniques, perioperative care, and patient selection. Thus, stable CABG volume in our data was accompanied by sustained low mortality – concordant with registry reports from STS, EACTS, and others.

In transcatheter aortic valve replacement (TAVR), our registry showed explosive growth. TAVR was virtually non-existent pre-2011, then accelerated annually after FDA approval. National reports indicate TAVR volume surpassed surgical AVR by 2019 [8], and I observed rising TAVR adoption in our registry, especially after expanded indications for intermediate- and low-risk patients in the mid-2010s. Outcomes improved dramatically: early TAVR 30-day mortality was ~ 6–7%, but by 2018–2019 it fell to ~ 2–3%[8, 18]. Our state’s TAVR mortality declines track those in large registries. The STS/ACC TVT Registry showed 30-day mortality drop from 6.7% (2012) to 2.4% (2018) and 1-year mortality from 19.9–10.1%[18]. Similarly, an international cohort saw 30-day TAVR mortality shrink from 7.2 to 2.5% over 2011–2019 [8]. I observed the same trend: later patients had far lower TAVR mortality, reflecting improved devices, experience, and expanding use in lower-risk subsets. By 2019–22, U.S. registry reports indicate these gains leveled off (30-day ≈ 2.2%)[18], suggesting a new baseline. Our findings also align with key trials: pivotal randomized trials (e.g. PARTNER and SURTAVI) [7, 19] consistently showed TAVR’s safety and noninferiority (even superiority) compared to SAVR in high- and intermediate-risk patients. Recent low-risk trials (PARTNER 3, NEJM 2019) [7] extended TAVR to younger patients, further boosting volumes. In short, the trend toward more TAVR and improving outcomes in our state registry mirrors national/international experience [8, 18].

For valvular surgery with or without CABG (“Valve ± CABG”), our registry trends reflected the interplay of surgical valve and TAVR volumes. As TAVR took over most isolated aortic valve replacements, surgical AVR (often combined with CABG) declined or stabilized. Overall AVR (TAVR + SAVR) increased ~ 60% nationally (2012–2019) [9]but that growth was entirely due to TAVR; SAVR cases fell. In our data, SAVR + CABG procedures trended downward, matching U.S. Medicare trends [9]while TAVR + PCI and valve-in-valve cases rose. Risk in valve operations has generally improved. STS reports show in-hospital mortality ~ 2–4% for isolated SAVR, and lower for isolated mitral repairs; combined procedures carry higher risk. Consistent with external registries, combined AVR + CABG carries higher mortality than isolated cases. This is consistent with registry analyses: a UK cohort (NACSA) found that each additional bypass graft during AVR raised mortality [20]and Korean data showed 1-year mortality 13.2% for AVR + CABG vs. 9.5% for CABG alone [21]. Our risk-adjusted death rates for valve ± CABG fell modestly over time, reflecting both improved surgical care and patient selection (e.g. hybrid approaches for multivessel disease).

Internationally, our trends align with many reports. In Europe and Asia, elective PCI also fell as in the U.S., while STEMI PCI was stable or rising. For example, Chinese PCI registries show total PCI tripled from 2010 to 2018, but in-hospital PCI mortality remained low (~ 0.3%) [22], echoing our low elective death rates. CABG declines have been widespread: the UK saw > 50% drop in isolated CABG over 15 years with mortality falling to ~ 1.0%[17], echoing STS data of improving CABG outcomes. Moreover, emerging international analyses of PCI/CABG largely confirm that in regional registry-type comparisons, CABG is associated with better long-term survival compared to PCI in most regions of the world without evidence for higher periprocedural mortality [23]. TAVR growth and outcome improvements have been global: Australian data show TAVR cases growing by ~ 30–40% annually since 2008 [24], with corresponding declines in surgical AVR [9]. The “all-AVR” survival improvements I observed match those seen in Medicare beneficiaries (1-year AVR mortality from ~ 12–9%)[9].

Study limitations

This analysis is subject to several inherent constraints. First, it is a retrospective observational study that relies on administrative registry data; although statewide capture is robust, coding errors and missing fields (e.g., frailty indices, left-ventricular ejection fraction, STS risk scores) may lead to misclassification and residual confounding. Second, the public NYS registry suppresses patient-level demographic and presentation fields (age, sex, STEMI/NSTEMI, cardiogenic shock, detailed comorbidity codes). Consequently, I cannot show how those characteristics evolved over time, nor can I isolate contribution to the patterns in PCI or CABG mortality. Weighted expected-mortality trends partially mitigate this gap, but unmeasured shifts in patient complexity may still influence our findings. Third, risk adjustment employed registry-standard models, which—while well calibrated overall—do not include contemporary metrics such as Society for Cardiovascular Angiography and Interventions (SCAI) shock stages or STS PROM for valve-in-valve procedures; unmeasured clinical variables could bias comparisons across eras. Fourth, follow-up was confined to in-hospital outcomes; late complications (e.g., structural valve deterioration, repeat revascularization, unplanned readmissions) were not available, constraining survival inferences beyond discharge. Fifth, because the dataset ends 31 Dec 2019, the analysis does not capture subsequent practice changes. Sixth, results derive from a single U.S. state with high procedural volumes; external generalizability to lower-volume or rural settings should be made cautiously. Finally, Valve or Valve/CABG and TAVR annual values were annualized from overlapping 3-year windows, and Emergency PCI was derived annually by subtraction (All − Non-Emergency); although window-annualized and subtraction-based series were closely concordant in checks, these steps introduce modest estimation error.

Future perspectives

Ongoing evolution in device design and patient selection is likely to reshape these trends further. Long-term durability data for low-risk TAVR cohorts will be pivotal in determining the balance between trans-catheter and surgical aortic valve strategies. The integration of intravascular imaging, physiologic lesion assessment, and robotic assistance may improve PCI outcomes, particularly for complex multivessel and chronic total occlusion cases. Artificial-intelligence–driven risk-prediction models that incorporate frailty, social determinants, and real-time hemodynamics could refine patient selection and benchmarking. Prospective linkage of procedural registries with longitudinal claims and electronic-health-record datasets will enable evaluation of late clinical and economic outcomes, while pragmatic trials comparing hybrid revascularization, complete revascularization, and definitive medical therapy remain a priority.

Health-Policy implications

Several actionable insights emerge for policymakers and system planners:

1. Regionalization and Volume Thresholds: Broadly stable CABG volumes juxtaposed with stable excellent outcomes support concentrating complex surgery in high-volume centers while maintaining minimum activity thresholds to safeguard proficiency.

2. Heart-Team Reimbursement Models: The sustained mortality gap between emergency PCI and other categories underscores the value of multidisciplinary Heart-Team decision-making; bundled or team-based reimbursement could incentivize collaborative case selection and peri-procedural care pathways.

3. Program Accreditation and Expansion: Explosive TAVR growth, coupled with improving safety, justifies expanding trans-catheter programs—provided rigorous accreditation standards (operator experience, surgical backup, structural-heart coordinators) remain in place to avoid volume dilution.

4. Quality Reporting and Transparency: Observed-to-expected ratios near unity validate current risk-adjusted public-reporting frameworks; continued investment in real-time dashboards and mandatory data submission can drive iterative quality improvement, particularly in high-risk PCI and combined valve–CABG surgery.

5. Resource Allocation and Workforce Planning: Rising emergency PCI and TAVR volumes necessitate 24/7 catheterization-laboratory staffing, structural-heart teams, and postoperative intensive-care capacity; strategic funding must anticipate these needs while supporting surgical training pipelines.

6. Equity and Access: Disparities in procedural access—especially for rural, minority, and low-income populations—should be monitored through geospatial mapping of registry data, with targeted outreach and tele-cardiology networks to mitigate inequities.

Collectively, these measures will help health systems capitalize on technological advances while safeguarding patient safety, fiscal sustainability, and equitable access to contemporary cardiovascular care.

Conclusions

Between 2010 and 2019, overall PCI volume was stable, non-emergency (elective) PCI declined modestly, emergency PCI rose modestly, and TAVR expanded substantially while surgical valve (Valve ± CABG) volumes decreased; outcomes for both surgical valve procedures and TAVR improved over time. In contrast, overall PCI mortality remained low and largely stable, consistent with a shift toward higher-risk case mix. CABG volumes were broadly stable across the decade with low, stable mortality. These findings underscore the need for quality-focused pathways in high-risk PCI, vigilant long-term monitoring of TAVR durability in younger patients, and policies that support high-performing centers while ensuring equitable access to contemporary cardiac therapies.

Authors’ contributions

R.M.E. conceived and designed the study, developed and executed the search strategy, carried out data extraction and statistical analyses, interpreted the results, drafted the manuscript, critically revised it for important intellectual content, and approved the final version for submission. R.M.E. agrees to be accountable for all aspects of the work, ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Funding

This research received no external funding.

Data availability

Data is available at New York State Department of Health. *Adult Cardiac Surgery and PCI Reporting System Public Use File (2008-2019)* Albany (NY): NYSDOH; 2024.10 Cardiac Surgery and Percutaneous Coronary Intervention data portal [Internet]. Albany (NY): NYSDOH; 2025 [cited 2025 May 1]. Available by clicking export at the top right of the page: https://health.data.ny.gov/Health/Cardiac-Surgery-and-Percutaneous-Coronary-Interven/jtip-2ccj/about_data.

Declarations

Ethics approval and consent to participate

Institutional review board oversight was not required because all data are de-identified and publicly available.

Consent for publication

The registry is publicly released by the NYS Department of Health (DOH) with a data use policy statement [11].

Competing interests

The authors declare no competing interests.

Clinical trial number

not applicable.