糖心vlog

Importance听 Reduced rates of cardiac catheterization, percutaneous coronary intervention (PCI), and coronary artery bypass graft (CABG) are an unintended consequence of public reporting of cardiogenic shock outcomes in New York.

Objectives听 To evaluate whether the referral rates for cardiac catheterization, PCI, or CABG have improved in New York since cardiogenic shock was excluded from public reporting in 2008 and compare them with corresponding rates in Michigan, New Jersey, and California.

Design, Setting, and Participants听 Patients with cardiogenic shock complicating acute myocardial infarction from 2002 to 2011 were identified using the National Inpatient Sample. Propensity score matching was used to assemble a cohort of patients with cardiogenic shock with similar baseline characteristics in New York and Michigan.

Main Outcomes and Measures听 Percutaneous coronary intervention (primary outcome), invasive management (cardiac catheterization, PCI, or CABG), revascularization (PCI or CABG), and CABG were evaluated with reference to 3 calendar year periods: 2002-2005 (time 1: cardiogenic shock included in publicly reported outcomes), 2006-2007 (time 2: cardiogenic shock excluded on a trial basis), and 2008 and thereafter (time 3: cardiogenic shock excluded permanently) in New York and compared with Michigan.

Results听 Among 2126 propensity score鈥搈atched patients representing 10鈥�795 (weighted) patients with myocardial infarction complicated by cardiogenic shock in New York and Michigan, 905 (42.6%) were women and mean (SE) age was 69.5 (0.3) years. A significantly higher proportion of the patients underwent PCI (time 1 vs 2 vs 3: 31.1% vs 39.8% vs 40.7% [OR, 1.50; 95% CI, 1.12-2.01; P鈥�=鈥�.005 for time 3 vs 1]), invasive management (time 1 vs 2 vs 3: 59.7% vs 70.9% vs 73.8% [OR, 1.84; 95% CI, 1.37-2.47; P鈥�<鈥�.001 for time 3 vs 1]), or revascularization (43.1% vs 55.9% vs 56.3% [OR, 1.66; 95% CI, 1.26-2.20; P鈥�<鈥�.001 for time 3 vs 1]) after the exclusion of cardiogenic shock from public reporting in New York. However, during the same periods, a greater proportion of patients underwent PCI (time 1 vs 2 vs 3: 41.2% vs 52.6% vs 57.8% [OR, 1.93; 95% CI, 1.45-2.56; P鈥�<鈥�.001 for time 3 vs 1]), invasive management (time 1 vs 2 vs 3: 64.4% vs 80.5% vs 78.6% [OR, 2.01; 95% CI, 1.47-2.74; P鈥�<鈥�.001 for time 3 vs 1]), or revascularization (51.2% vs 65.8% vs 68.0% [OR, 2.00; 95% CI, 1.50-2.66; P鈥�<鈥�.001 for times 3 vs 1]) in Michigan. Results were largely similar in several sensitivity analyses comparing New York with New Jersey or California.

Conclusions and Relevance听 Although the rates of PCI, invasive management, and revascularization have increased substantially after the exclusion of cardiogenic shock from public reporting in New York, these rates remain consistently lower than those observed in other states without public reporting.

Public reporting of operator-specific mortality for coronary artery bypass graft (CABG) surgery and percutaneous coronary intervention (PCI) in New York state began in the late 1980s with the creation of the New York State Cardiac Surgery Reporting System and the New York State Percutaneous Coronary Interventions Reporting System. Although the original objective of public reporting was to improve the quality of care and aid the public in selecting providers for cardiac interventions,¹ there is now a growing body of evidence suggesting unintended consequences of such public reporting.^2-4 In New York, 79% of interventional cardiologists claimed that public reporting had at times influenced their decision on whether to perform PCI.⁵

Cardiogenic shock complicating myocardial infarction (MI) is a leading cause of in-hospital mortality; an invasive strategy of emergent revascularization reduces this risk and is a class I indication by national and international guidelines.^6-9 However, despite improved outcomes with emergent revascularization, in-hospital mortality remains high鈥攎ore than 30%¹⁰鈥攖hereby creating a disincentive in public health systems in which revascularization outcomes are publicly reported. Data² suggest that patients with cardiogenic shock are less likely to undergo coronary angiography, PCI, and CABG in New York state compared with other states (eg, Michigan). Although the data on this unintended consequence have been reported² previously, it is not known whether selective exclusion from reporting would have a positive effect on the use of guideline-recommended care of these patients. In New York, cardiogenic shock was excluded from public reporting on a trial basis in 2006-2007 and permanently since 2008. The objective of the present study was to evaluate whether the referral rates for cardiac catheterization, PCI, or CABG for cardiogenic shock have improved with this measure in New York and compare those with corresponding rates in Michigan, New Jersey, and California (nonreporting states).

Box Section Ref ID

Patients with cardiogenic shock complicating acute MI between 2002 and 2011 in New York, Michigan, New Jersey, and California were included in this analysis. The comparator state was chosen based on prior reported data comparing New York with the state (Michigan), geographic proximity (New Jersey), and availability of a 100% patient sample file (California). The patients were identified using the National Inpatient Sample (NIS),¹¹ which is the largest publicly available all-payer database that contains discharge-level data yielding national estimates of hospital inpatient stays. The database contains discharge data from more than 1000 acute care and nonfederal hospitals each year, which approximates a 20% stratified sample of US community hospitals with approximately 8 million unweighted hospital stays each year. Because the NIS is an administrative data set and is deidentified, no informed consent was needed and the study was exempt from review by the New York University School of Medicine institutional review board.

Patients were included if they had a principal diagnosis of cardiogenic shock (International Classification of Diseases, Ninth Revision, code 785.5x) with a secondary diagnosis of acute MI or a principal diagnosis of acute MI with a secondary diagnosis of cardiogenic shock. Only patients admitted to hospitals in New York, Michigan, New Jersey, or California were included in this analysis. Patients were excluded if they were admitted to a non-PCI center (defined as those with data on <5 PCIs performed for acute MI during the respective year of the study).

Operator-level public reporting of outcomes data primarily pertains to revascularization (PCI and CABG and not diagnostic catheterization). In addition, there has been a significant shift in revascularization for acute MI with a significant increase in PCI and a decline in CABG during the past decade. Therefore, our primary outcome was the proportion of patients with cardiogenic shock who underwent PCI before and after cardiogenic shock was excluded from publicly reported data in New York. We also evaluated invasive management (defined as any cardiac catheterization, PCI, or CABG), revascularization (PCI or CABG), or CABG, as well as in-hospital mortality during the periods and between the states.

The primary analysis cohort compared New York with Michigan using data from NIS. Several sensitivity analysis cohorts were compared as discussed below. Outcomes were compared across times and between states. The periods evaluated were 2002-2005 (cardiogenic shock included in publicly reported outcomes in New York), 2006-2007 (cardiogenic shock excluded on a trial basis in New York), and from 2008 onward (cardiogenic shock excluded permanently in New York).

Since the difference in management between the 2 states compared may be related to baseline differences between patients admitted, propensity score matching was used to assemble a cohort of patients with cardiogenic shock with similar baseline characteristics. The propensity score is the conditional probability of having a particular exposure given a set of baseline-measured covariates.^12,13 Propensity score matching was performed using a nonparsimonious logistic regression model¹⁴ with New York vs the comparator state as the dependent variable and the variables listed in the Table with year of admission as the independent variable. Matching was performed using a 1:1 protocol without replacement (greedy matching algorithm) with a caliper width equal to 0.2 of the SD of the logit of the propensity score. The C statistic for the model was 0.72 for the New York vs Michigan comparison. Absolute standardized differences were estimated for all the baseline covariates before and after matching to assess prematch imbalance and postmatch balance.¹⁵ Absolute standardized differences of less than 10% for a given covariate indicate a relatively small imbalance.¹⁵ All further analyses were performed in the matched cohort. Paired comparisons were performed using the McNemar test for binary variables and a 2-tailed, paired t test for continuous variables.

Three analyses were performed. The first of these was the proportion of PCI (primary outcome), invasive management, revascularization, and CABG over the 3 periods within a state and between states. Next, the adjusted change in proportion was determined before and after exclusion of cardiogenic shock from public reporting in each of the states. For this analysis, multivariable logistic regression models were created to evaluate the change in management and in-hospital mortality for the 3 periods using time 1 as reference and after controlling for baseline clinical variables for each state. Third, to compare the rate of change (slope) in management over time between New York and each of the comparator states, we calculated a test for interaction using a logistic regression model with both the linear and nonlinear terms for year and the interaction terms with state and adjusted for baseline covariates. An interaction value of P鈥�<鈥�.05 was used to denote a significant interaction between the rate of change in management over time in New York and that in the comparator state. However, a nonsignificant test for interaction should be interpreted with caution since it may merely mean that the 2 curves are parallel to each other and could be separated by a constant percentage. All analyses were performed with SAS, version 9.3 (SAS Institute Inc). A 2-sided value of P鈥�<鈥�.05 was used to denote statistical significance.

Several sensitivity analyses were performed to evaluate the robustness of the results. New York was compared with New Jersey (to assess whether the results observed for New York vs Michigan were by chance alone). New York was compared with Michigan and with New Jersey after also matching for hospital characteristics between the states to ensure that the results observed were not due to differences in hospital characteristics between the states. Validation of the results of the above analyses was performed using a 100% patient sample file of the Statewide Planning and Research Cooperative System (SPARCS) (New York)¹⁶ and the California inpatient file to evaluate whether the results observed were similar to the 20% sample used in NIS. Finally, the results observed using the 100% patient sample of New York and California were compared with those using data for New York and California from NIS (20% sampling).

From 2002 to 2011, a total of 2817 patients (unweighted) representing 14鈥�370 patients (weighted) received a diagnosis of cardiogenic shock complicating acute MI in New York. In the same timeframe, 1587 patients representing 7911 patients received a diagnosis of cardiogenic shock complicating acute MI in Michigan. Propensity score matching identified 1063 patients in New York (representing 5446 patients) and 1063 patients in Michigan (representing 5349 patients) with similar propensity scores. After matching, there were no considerable differences between the baseline characteristics of patients in New York and Michigan (Table). In the matched cohort, there were 336 (representing 1675 patients), 233 (representing 1225 patients), and 494 (representing 2546 patients) patients in times 1, 2, and 3, respectively, in New York and 318 (representing 1597 patients), 229 (representing 1192 patients), and 516 (representing 2559 patients) patients, respectively, in Michigan.

The baseline characteristics before and after propensity score matching for the sensitivity analysis cohort of New York vs Michigan matched on hospital characteristics (2288 patients representing 11鈥�555 patients), New York vs New Jersey (2722 patients representing 13鈥�844 patients), New York vs New Jersey matched on hospital characteristics (2082 patients representing 10鈥�570 patients), New York vs California using 100% patient sample files (22鈥�460 patients), and New York vs California using data from NIS (3948 patients representing 19鈥�683 patients) are outlined in eTables 1-5 in the Supplement.

The proportion of patients with cardiogenic shock who underwent PCI increased significantly after exclusion of cardiogenic shock from public reporting in New York (time 1 vs 2 vs 3: 31.1% vs 39.8% vs 40.7% [OR, 1.50; 95% CI, 1.12-2.01; P鈥�=鈥�.005 for time 3 vs 1]) (Figure 1A). There was a 19% (time 3 vs 1: adjusted relative risk [RR], 1.19; 95% CI, 0.96-1.45 [P鈥�=鈥�.11]) increase in PCI after cardiogenic shock was excluded from public reporting in New York. A similar increase in the proportion of patients who underwent PCI was also seen in Michigan (time 1 vs 2 vs 3: 41.2% vs 52.6% vs 57.8% [OR, 1.93; 95% CI, 1.45-2.56; P鈥�<鈥�.001 for time 3 vs 1]) (Figure 1A). In fact, PCI increased by 40% (adjusted RR, 1.40; 95% CI, 1.18-1.60 [P鈥�<鈥�.001 for time 3 vs time 1]) in Michigan. At each of the 3 times, the proportion of patients who underwent PCI was significantly lower in New York than in Michigan (Figure 1A). The test for interaction for the rate of change in PCI over time between the 2 states was statistically significant (P鈥�=鈥�.02 for interaction) (eFigure 1A in the Supplement) such that the change was greater in Michigan than in New York.

Sensitivity analyses comparing New York with Michigan after matching for hospital characteristics (Figure 1B) and comparing New York with New Jersey before (Figure 2A) and after (Figure 2B) matching for hospital characteristics yielded largely similar results with a significantly lower proportion of patients with cardiogenic shock who underwent PCI in New York than in the comparator state at each documented time. Moreover, analyses using the 100% patient sample files of New York (SPARCS) and California (inpatient data files) yielded largely similar results (Figure 3A). In addition, comparison between New York and California using data from NIS produced results similar to those obtained with the 100% patient sample files (Figure 3B).

The proportion of patients with cardiogenic shock who underwent invasive management increased significantly after the exclusion of cardiogenic shock from public reporting in New York (time 1 vs 2 vs 3: 59.7% vs 70.9% vs 73.8% [OR, 1.84; 95% CI, 1.37-2.47; P鈥�<鈥�.001 for time 3 vs 1]) (eFigure 2A in the Supplement). In a multivariable model adjusted for baseline characteristics, compared with time 1, there was an 18% (adjusted RR,鈥�1.18; 95% CI, 1.05-1.29 [P鈥�=鈥�.01]) increase in invasive management after cardiogenic shock was excluded from public reporting (time 3) in New York. A similar increase in the proportion of patients who underwent invasive management was also seen in Michigan (time 1 vs 2 vs 3:鈥�64.4% vs 80.5% vs 78.6% [OR, 2.01; 95% CI, 1.47-2.74; P鈥�<鈥�.001 for time 3 vs 1]) (eFigure 2A in the Supplement). Invasive management increased by 23% (adjusted RR, 1.23; 95% CI, 1.12-1.32 [P鈥�<鈥�.001 for time 3 vs time 1]) in Michigan. At each of the 3 times, the proportion of patients who underwent invasive management was numerically lower in New York than in Michigan (eFigure 2A in the Supplement). The test for interaction for the rate of change in invasive management over time between the 2 states was not statistically significant (P鈥�=鈥�.27 for interaction) (eFigure 1B in the Supplement). However, the model indicated that, overall, there was a 12% increase in invasive management in Michigan compared with New York (adjusted RR, 1.12; 95% CI, 1.05-1.17 [P鈥�<鈥�.001]). Sensitivity analyses with pairwise comparison of New Jersey and California with New York all yielded largely similar results (eFigures 3A, 4A, 5A, 6A, and 7A in the Supplement).

The proportion of patients with cardiogenic shock who underwent revascularization increased significantly after the exclusion of cardiogenic shock from public reporting in New York (time 1 vs 2 vs 3: 43.1% vs 55.9% vs 56.3% [OR, 1.66; 95% CI, 1.26-2.20; P鈥�<鈥�.001 for time 3 vs 1]) (eFigure 2B in the Supplement). There was a 23% (time 3 vs 1: adjusted RR, 1.23; 95% CI 1.05-1.41 [P鈥�=鈥�.01]) increase in revascularization after cardiogenic shock was excluded from public reporting. A similar increase in the proportion of patients who underwent revascularization was also seen in Michigan (time 1 vs 2 vs 3:鈥�51.2% vs 65.8% vs 68.0% [OR, 2.00; 95% CI, 1.50-2.66; P鈥�<鈥�.001 for time 3 vs 1]) (eFigure 2B in the Supplement). Revascularization increased by 35% (adjusted RR, 1.35; 95% CI, 1.19-1.49 [P鈥�<鈥�.001 for time 3 vs time 1]) in Michigan. At each of the 3 times, the proportion of patients who underwent revascularization was significantly lower in New York than in Michigan (eFigure 2B in the Supplement). The test for interaction for the rate of change in revascularization over time between the 2 states was not statistically significant (P鈥�=鈥�.08) (eFigure 1C in the Supplement). However, the model indicated that, overall, there was a 25% increase in revascularization in Michigan compared with New York (adjusted RR, 1.25; 95% CI, 1.16-1.34 [P鈥�<鈥�.001]). Sensitivity analyses with pairwise comparison of New Jersey and California with New York all yielded largely similar results (eFigures 3B, 4B, 5B, 6B, and 7B in the Supplement).

There was no change in the proportion of patients with cardiogenic shock who underwent CABG in New York after exclusion of cardiogenic shock from public reporting (time 1 vs 2 vs 3: 14.2% vs 18.3% vs 17.8% [OR, 1.28; 95% CI, 0.87-1.88; P鈥�=鈥�.16 for time 3 vs 1]) or in Michigan (time 1 vs 2 vs 3: 13.4% vs 16.7% vs 12.9% [OR, 0.95; 95% CI, 0.63-1.44; P鈥�=鈥�.83 for time 3 vs 1]) (eFigure 2C in the Supplement). The test for interaction for the rate of change in CABG over time between the 2 states was not statistically significant (P鈥�=鈥�.10) (eFigure 1D in the Supplement). However, the model indicated that overall there was a 17% decrease in CABG in Michigan compared with New York (adjusted RR,鈥�0.83; 95% CI, 0.66-1.02 [P鈥�=鈥�.07]). Sensitivity analyses with pairwise comparison of New Jersey and California with New York all yielded largely similar results (eFigures 3C, 4C, 5C, 6C, and 7C in the Supplement).

For the overall cohort, there was a lower mortality rate with time for both New York (time 1 vs 2 vs 3: 45.3% vs 39.1% vs 31.8% [OR, 0.57; 95% CI, 0.43-0.76; P鈥�=鈥�.001 for time 3 vs 1]) and Michigan (time 1 vs 2 vs 3: 41.2% vs 36.4% vs 33.4% [OR, 0.71; 95% CI, 0.54-0.95; P鈥�=鈥�.02 for time 3 vs 1]), with no significant difference between the states at each time (time 1, P鈥�=鈥�.29; time 2, P鈥�=鈥�.55; and time 3, P鈥�=鈥�.59). There was no change in in-hospital mortality of patients with cardiogenic shock who underwent invasive management (eFigure 2D in the Supplement), revascularization (eFigure 2E in the Supplement), PCI, or CABG in New York after exclusion of cardiogenic shock from public reporting. The results were similar in Michigan (eFigure 2D-E in the Supplement). In the group managed conservatively, in-hospital mortality was uniformly high across the 3 periods in both New York and Michigan (eFigure 2F in the Supplement). Sensitivity analyses comparing New York with New Jersey yielded largely similar results (eFigure 8A-C in the Supplement).

The proportion of patients with cardiogenic shock who underwent right heart catheterization increased significantly after the exclusion of cardiogenic shock from public reporting in New York (time 1 vs 2 vs 3: 9.0% vs 13.8% vs 17.1% [OR, 2.06; 95% CI, 1.32-3.21; P鈥�=鈥�.001 for time 3 vs 1]). However, this proportion decreased with time in MI (time 1 vs 2 vs 3: 21.9% vs 13.2% vs 13.3% [OR, 0.54; 95% CI, 0.37-0.78; P鈥�=鈥�.002 for time 3 vs 1]).

Among 2126 patients representing 10鈥�795 patients with MI complicated by cardiogenic shock in New York and Michigan, 905 (42.6%) were women and mean (SE) age was 69.5 (0.3) years. A significantly higher proportion of patients underwent PCI, invasive management, or revascularization after the exclusion of cardiogenic shock from public reporting in New York. However, during the same periods, there was a greater proportion of patients with cardiogenic shock who underwent these procedures in Michigan. The results were largely similar in several sensitivity analyses, including 5 pairwise comparisons of New York with 3 different states.

Cardiogenic shock complicating MI is the leading cause of in-hospital mortality.^17,18 Early revascularization in such patients results in a significant reduction in cardiovascular events.⁸ Despite this intervention, the in-hospital mortality rate remains high, even in patients who undergo revascularization.¹⁰ Public reporting of outcomes may serve as a disincentive for operators to choose revascularization in patients with a high likelihood of adverse outcomes. This limitation is especially important in the era of decreasing operator volumes for PCI and CABG in the United States since only a few deaths can result in worse outcome data for the operator. Although many public reporting systems use risk adjustment models to account for case mix and complexity, such models cannot completely account for many unmeasured confounders. Furthermore, with cardiogenic shock, risk adjustment is complicated by the degree of shock (eg, level of hypotension), its duration, and its trajectory. Moreover, other measures of the severity of hypoperfusion, including end organ damage, may be a more important indicator of outcomes than blood pressure. It is therefore not surprising that, given a mortality rate of 1 in 3 patients with cardiogenic shock even after PCI, prior reports^2-4,19 have shown 鈥渃herry picking鈥� of these cases in health care systems in which there is public reporting of procedural outcomes.

As a result of this unintended consequence of public reporting, an initiative was undertaken in New York whereby patients with cardiogenic shock who underwent revascularization were excluded from public reporting on a trial basis from 2006 to 2007 and permanently since 2008. However, whether the risk-averse behavior is mitigated by this measure is not known. Our study highlights the impact of excluding cardiogenic shock from public reporting. Although analyses of only New York data suggest substantial improvement in PCI, invasive management, and revascularization after exclusion of cardiogenic shock from public reporting in New York, comparison with other nonreporting states suggests that these improvements are lower in New York. The data suggest that gaps in the rates of these interventions persist between New York and other states. The improved proportion of PCI in Michigan, New Jersey, and California is likely due to several factors, including increased understanding of the importance of invasive management and revascularization due to greater acceptance of the results of the Should We Emergently Revascularize Occluded Coronaries for Cardiogenic Shock (SHOCK) trial⁶ with time, and American College of Cardiology/American Heart Association⁸ guideline recommendations that strongly favor revascularization in patients with cardiogenic shock.

Although it is likely that a proportion of patients with cardiogenic shock may not undergo revascularization owing to comorbidities (eg, advanced cancer) and patient preference, one of the more likely reasons for the observed low rates of PCI in New York could still be continued reluctance to perform interventions on high-risk cases.¹⁹ Although the improvement seen in New York could be attributable to the exclusion of cardiogenic shock from public reporting, the comparatively lower rates may be the result of the strict definition of cardiogenic shock in New York that qualifies for exemption from public reporting. The definition of cardiogenic shock in New York is 鈥淓pisode of systolic blood pressure <80 mm Hg and/or cardiac index <2.0 L/min/m² determined to be secondary to cardiac dysfunction despite the use of parenteral inotropic or vasopressor agents or mechanical support (eg, IABP [intra-aortic balloon pump], extracorporeal circulation, VADs [ventricular assist devices]). If the patient has an IABP, the augmented or non-augmented systolic blood pressure <80 mm Hg can be used as supporting documentation to code refractory shock. If the patient is VAD dependent, then refractory shock can be coded. For these purposes ECMO [extracorporeal membrane oxygenation] is treated like a VAD. Use of Impella [Abiomed] is treated like a VAD when there is evidence prior to insertion that the hemodynamic criteria above are met. When coding refractory shock, be careful of timing. It needs to be just prior to the start of the PCI. All elements of the definition must be clearly documented to have occurred prior to the guide-wire leaving the catheter.鈥�^20[p38]

Thus, if the above criteria are not met or, more likely, are not documented given the emergent nature of these procedures, New York denies the hospital鈥檚 request to categorize a patient as having cardiogenic shock. In New York, refractory shock is excluded from public reporting. A previously published report²¹ has shown an increase in the absolute number of PCIs for refractory shock in New York since the exclusion of cardiogenic shock from public reporting, thereby indicating a positive impact of this measure. Despite the relative difference between states in the use of invasive management, an overall improvement in survival over the periods examined was documented in patients with cardiogenic shock in all of the states compared that paralleled the increase in the use of revascularization.

Although the objective of public reporting is to encourage improvement in quality of care, findings regarding the results have been variable. The response from the public has been mediocre at best since publicly reported outcomes data for PCI and CABG are difficult to understand.²² Moreover, a survey published in New York²³ suggests that most clinicians believe that the publicly reported information did not change their referral patterns to surgeons. Despite this, the general perception among health care professionals is that public reporting influences their decisions on providing care for their patients.⁵ However, other analyses have shown a positive impact of public reporting. In an analysis from the National Cardiovascular Data Registry CathPCI registry, Boyden et al¹⁹ found that fewer high-risk patients underwent PCI in New York than in Michigan. However, they also observed lower mortality and fewer adverse events in New York for the overall PCI cohort. Whether further loosening of the criteria for shock in New York would improve access to care is yet to be determined. In addition, the process of care may be a better measure of quality than outcomes given the challenges to adequately risk adjust in this group of patients with cardiogenic shock. Moreover, process of care measures are less likely to be subject to upcoding.

The study was based on an administrative database; hence, there is a potential for miscoding. However, there is no reason to believe that there would be differential coding across calendar years. The improvement in the proportion of patients who underwent PCI over time could also be the result of improved coding of cardiogenic shock in the group that underwent PCI (to qualify for exemption from public reporting of outcomes). The improvement could also be due to potential upcoding for cardiogenic shock in high-risk patient subgroups (without cardiogenic shock) that underwent PCI given the knowledge on the providers (in New York) that these patients would be excluded in the publicly reported data. The study did not demonstrate a mortality difference among patients with invasive management between New York and Michigan. However, given the small sample size, the study is likely underpowered to detect a difference in mortality. The sample size of patients with cardiogenic shock was small, and we were therefore not able to evaluate whether there are differences in revascularization over time based on hospital size, as well as type or volume of cases. Although we evaluated the trends in its use, right heart catheterization is subject to under coding in administrative databases.

Finally, Healthcare Cost and Utilization Project discourages state-to-state comparison since the NIS sampling scheme is not stratified by state and hospitals in a given stratum are unlikely to be representative of those in a given state. However, the above limitation is applicable to state-to-state comparisons in which there is a need to preserve the sampling design, such as comparison of absolute rates (eg, absolute rates of cardiogenic shock in New York and Michigan) since the rates can vary based on the sampling design. Although random sampling from each state can eliminate certain kinds of bias, it does not ensure that the patient and hospital characteristics between the 2 states will be similar. Our analysis used a propensity score鈥搈atched sample, which is a more robust way to reduce state-to-state variability in patient- and hospital-related characteristics and evaluated the relative proportion of patients with cardiogenic shock who underwent PCI, invasive management, or revascularization within this sample. The study therefore loses the sampling design of NIS since many patients are excluded during the propensity score鈥搈atching process. The consistency of the results across 6 pairwise comparisons with the use of 3 states point to the robustness of the analyses. Moreover, sensitivity analyses matching for hospital characteristics as well as year of admission yielded largely similar results. Finally, analyses using 100% patient sample files of SPARCS and California inpatient data largely yielded similar results. Moreover, we were able to demonstrate that the results obtained using the 100% sample files are largely similar to those of the New York vs California comparison from NIS, thereby showing that the comparison of New York with New Jersey and New York with Michigan from NIS are also valid for the reasons outlined above.

We observed a significantly higher proportion of patients with cardiogenic shock as a complication of acute MI who underwent PCI, invasive management, or revascularization after the exclusion of cardiogenic shock from public reporting in New York. However, these rates remained lower than those in nonreporting states, such as Michigan, New Jersey, and California, with a persistent gap between the states over time.

Corresponding Author: Sripal Bangalore, MD, MHA, New York University School of Medicine, New York, NY 10016 (sripalbangalore@gmail.com).

Accepted for Publication: March 14, 2016.

Published Online: July 27, 2016. doi:10.1001/jamacardio.2016.0785

Author Contributions: Dr Bangalore and Ms Guo had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Bangalore, Hochman.

Acquisition, analysis, or interpretation of data: Bangalore, Guo, Xu, Blecker, Gupta, Feit.

Drafting of the manuscript: Bangalore.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Guo, Xu, Gupta.

Administrative, technical, or material support: Blecker, Gupta.

Study supervision: Bangalore, Hochman.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Blecker was supported by the Agency for Healthcare Research and Quality grant K08HS23683. No other disclosures were reported.