If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Division of Cardiothoracic Surgery, Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, TexMedical Scientist Training Program, Baylor College of Medicine, Houston, TexDepartment of Bioengineering, Rice University, Houston, Tex
Division of Cardiothoracic Surgery, Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, TexSection of Adult Cardiac Surgery, Department of Cardiovascular Surgery, Texas Heart Institute, Houston, Tex
Division of Cardiothoracic Surgery, Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, TexSection of Adult Cardiac Surgery, Department of Cardiovascular Surgery, Texas Heart Institute, Houston, Tex
Division of Cardiothoracic Surgery, Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, TexSection of Adult Cardiac Surgery, Department of Cardiovascular Surgery, Texas Heart Institute, Houston, Tex
Address for reprints: Ravi K. Ghanta, MD, Division of Cardiothoracic Surgery, Michael E. DeBakey Department of Surgery, Baylor College of Medicine, One Baylor Plaza, MC-390, Houston, TX 77030.
Division of Cardiothoracic Surgery, Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, TexSection of Adult Cardiac Surgery, Department of Cardiovascular Surgery, Texas Heart Institute, Houston, Tex
Safety-net hospitals (SNHs) provide essential services to predominantly underserved patients regardless of their ability to pay. We hypothesized that patients who underwent coronary artery bypass grafting (CABG) would have inferior observed outcomes at SNHs compared with non-SNHs but that matched cohorts would have comparable outcomes.
Methods
We queried the Nationwide Readmissions Database for patients who underwent isolated CABG from 2016 to 2018. We ranked hospitals by the percentage of all admissions in which the patient was uninsured or insured with Medicaid; hospitals in the top quartile were designated as SNHs. We used propensity-score matching to mitigate the effect of confounding factors and compare outcomes between SNHs and non-SNHs.
Results
A total of 525,179 patients underwent CABG, including 96,133 (18.3%) at SNHs, who had a greater burden of baseline comorbidities (median Elixhauser score 8 vs 7; P = .04) and more frequently required urgent surgery (57.1% vs 52.8%; P < .001). Observed in-hospital mortality (2.1% vs 1.8%; P = .004) and major morbidity, length of stay (9 vs 8 days; P < .001), cost ($46,999 vs $38,417; P < .001), and readmission rate at 30 (12.4% vs 11.3%) and 90 days (19.0% vs 17.7%) were greater at SNHs (both P < .001). After matching, none of these differences persisted except length of stay (9 vs 8 days) and cost ($46,977 vs $39,343) (both P < .001).
Conclusions
After matching, early outcomes after CABG were comparable at SNHs and non-SNHs. Improved discharge resources could reduce length of stay and curtail cost, improving the value of CABG at SNHs.
Safety-net hospitals perform coronary artery bypass grafting with comparable early outcomes to those of propensity-matched non-safety-net hospitals but with longer stays and greater costs.
Although previous studies have suggested that the quality of complex surgical care at safety-net hospitals lags behind nonsafety-net hospitals, we found that safety-net hospitals perform coronary artery bypass grafting with comparable propensity-score matched outcomes to non-safety-net hospitals, albeit with longer length of stay and greater costs.
Safety-net hospitals (SNHs) are defined as those that, by mission or mandate, provide care to a substantial share of vulnerable patients regardless of their ability to pay.
Nevertheless, SNHs have been the subject of scrutiny with respect to delivery of complex surgical care, including cardiac surgery, with several studies suggesting that the quality of surgical care at SNHs lags behind that at non-SNHs.
Given these disparities, along with a growing emphasis on health care equity, the quality of cardiac surgical care at SNHs versus non-SNHs is an area worthy of examination. In addition, identifying specific areas in which SNHs may fall short of non-SNHs can provide policymakers and cardiothoracic surgical societies with areas to target for improvement.
Coronary artery bypass grafting (CABG) is the most common cardiac operation performed in the United States.
Comparison of risk-standardized readmission rates of surgical patients at safety-net and non-safety-net hospitals using Agency for Healthcare Research and Quality and American Hospital Association data.
few have mitigated the effect of confounding variables by comparing matched patient cohorts. We hypothesized that patients who undergo CABG would have inferior outcomes and greater cost at SNHs than at non-SNHs but that matched cohorts drawn from a nationwide sample would have comparable outcomes and cost.
Patients and Methods
Study Population and Data Collection
The Nationwide Readmissions Database (NRD) is the largest publicly available all-payer inpatient database that provides linked admissions for a patient within a given calendar year. It is maintained by the Agency for Healthcare Research and Quality's Healthcare Cost and Utilization Project. The NRD provides demographic, clinical, and cost data from more than 90% of adult discharges from 28 states, accounting for approximately 60% of all patients and hospitals whose data are recorded in the American Hospital Association Annual Survey Database, and uses a complex survey design to allow for national estimates of patient outcomes.
We ensured that our data were coded consistently by abstracting admissions data between January 1, 2016, and December 31, 2018, during which time all data were coded according to the 10th Revision of the International Classification of Diseases and Related Health Problems (ICD-10). First, we used a combination of ICD-10-Clinical Modification and ICD-10-Procedure Coding System codes to identify all patients who underwent CABG during the study period. Next, we excluded patients who underwent concomitant percutaneous coronary intervention, valve repair or replacement, or open or endovascular aortic intervention to generate a cohort of patients who underwent isolated CABG (Table E1). Given that only deidentified data were used, this study was deemed exempt from institutional review board approval.
Study Definitions
To facilitate a direct comparison, and in line with previous studies, we defined safety-net burden as the percentage of all admissions with the patient's primary payer designated as uninsured or insured with Medicaid.
Comparison of risk-standardized readmission rates of surgical patients at safety-net and non-safety-net hospitals using Agency for Healthcare Research and Quality and American Hospital Association data.
In-hospital outcomes of ST-segment elevation myocardial infarction complicated with cardiogenic shock at safety-net hospitals in the United States (from the Nationwide Inpatient Sample).
We abstracted patient characteristics, including age, sex, payer, and median household income, directly from the NRD. Comorbidities were abstracted by using ICD-10-Clinical Modification codes. The Elixhauser Comorbidity Index,
a composite score of 30 chronic conditions, was used to quantify the burden of baseline comorbidities. Elixhauser score was calculated by using a Python (Python Software Foundation. Python Language Reference, version 3.7. http://www.python.org) implementation of Healthcare Cost and Utilization Project Software and Tools (hcuppy package, version 0.0.7).
Outcomes
Our primary outcome was in-hospital mortality, and our secondary outcomes were major morbidity (acute kidney injury, stroke, respiratory failure, infections), length of stay (LOS), cost, and 30- and 90-day readmission. Admissions during the month of December were excluded from 30-day readmission calculations, and admissions during the months of October through December were excluded from 90-day readmission calculations. The total hospitalization cost was calculated by multiplying the total hospital charge by cost-to-charge ratios provided by the NRD.
Statistical Analysis
Analyses were performed in R 4.1 (The R Project for Statistical Computing). In all patient-level analyses, we accounted for the complex survey design of the NRD, including clustering, stratification, and sample weighting. Data are reported as weighted national estimates, as is standard in studies using the NRD, unless otherwise noted. All hospital-level analyses are based on data directly abstracted from the NRD and were performed by traditional statistical methods, as these data are not weighted to provide national estimates. Of note, hospitals are given a unique identifier each year in the NRD and cannot be tracked across years; therefore, we stratified these data by year.
Categorical variables are presented as number (%), and continuous variables are presented as median (interquartile range). Univariate comparisons were performed with the χ2 test with Rao and Scott correction or Wilcoxon rank-sum test adjusted for complex survey design, as appropriate. National estimates were generated by using sampling weights assigned through the NRD.
A survey-adjusted multivariable logistic regression was performed on the complete cohort to determine which baseline patient characteristics were associated with in-hospital mortality. All significant variables in the univariate analysis, along with hospital safety-net status, were included in the initial model. All data were reduced to binary variables, and missing data cells were replaced by the mode of that column. We divided the data into a training set (80%) and a testing set (20%) with a partitioning tool. Logistic regressions were carried out on the training set, and predictions were made for the testing set, which were used to calculate the area under the curve (C-statistic) of the receiver operator characteristic curve. Next, we used the C-statistic to guide independent variable selection, and the variables in the final model were presented as adjusted odds ratio (OR) with 95% confidence interval (CI).
Propensity Score–Matched Analysis
A propensity score–matched analysis was performed to compare outcomes between matched populations at SNHs versus non-SNHs. Cohorts were matched according to the following covariates: age, sex, insurance status, area of residence, household income quartile, elective admission status, hospital characteristics, and 28 comorbid conditions included in the composite Elixhauser comorbidities (Figure E1). Greedy nearest neighbor matching through the MatchIt package was used to create pairs in a 1:1 ratio between SNHs and non-SNHs with a caliper of 0.05 standard deviation of the logit. Of note, matching yielded a modest difference in weighted sample sizes between groups due to post hoc sample weighting calculations. Quality of matching was determined with balance diagnostics, including computing standardized mean differences and visually inspecting covariate distribution. An average standard mean difference of 0.10 was considered acceptable. After matching, univariate comparisons were performed with the tests described previously.
Results
Preoperative Characteristics
A total of 525,179 patients (75.5% male) underwent isolated CABG in the United States between 2016 and 2018 (96,133 [18.3%] at SNHs and 429,046 [81.7%] at non-SNHs) (Figure 1). There were several important baseline differences between the 2 cohorts. Patients at SNHs were younger (median age 65 vs 67 years) and more frequently in the lowest income quartile (32.5% vs 26.6%) than patients at non-SNHs (both P < .001). In addition, patients at SNHs had a greater burden of baseline comorbidities (median Elixhauser score 8 vs 7; P = .04), including several chronic conditions known to portend adverse outcomes after CABG: congestive heart failure (37.2% vs 34.3%), diabetes mellitus (52.2% vs 47.5%), and hypertension (89.7% vs 87.5%) (all P < .001). In contrast, patients at non-SNHs had greater rates of obesity (29.7% vs 28.1%; P = .04) and peripheral arterial disease (15.2% vs 14.1%; P = .003) than patients at SNHs. Tobacco use was similar between groups. Last, patients at SNHs more frequently required urgent surgery (57.1% vs 52.8%; P < .001) (Table 1).
Figure 1Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) diagram. CABG, Coronary artery bypass grafting; SNH, safety-net hospital.
Categorical variables are presented as number (%), and continuous variables are presented as median (interquartile range). SNHs, Safety-net hospitals; SMD, standardized mean difference.
An average of 783 hospitals performed CABG each year. In general, SNHs were larger, more likely to be in an urban location, and more likely to be designated as a teaching hospital, whereas non-SNHs were smaller and more likely to be in a rural location. Median annual CABG volume was 111 at SNHs and 158 at non-SNHs (P < .001) (Table 2).
Table 2Characteristics of SNH versus non-SNHs performing CABG, stratified by year
National estimates derived from weighted patient-level data.
145 (29, 335)
115 (34, 245)
.050
155 (38, 340)
110 (37, 257)
.01
174 (73, 348)
101 (31, 270)
< .001
Categorical variables are presented as number (%), and continuous variables are presented as median (interquartile range). SNH, Safety-net hospital; CABG, coronary artery bypass grafting.
∗ Observed data derived from unweighted hospital-level data.
† National estimates derived from weighted patient-level data.
Observed rates of in-hospital mortality (2.1% vs 1.8%; P = .004), acute kidney injury (21.6% vs 19.9%; P < .001), stroke (2.0% vs 1.7%; P < .001), pneumonia (6.2% vs 4.6%; P < .001), sepsis (2.7% vs 1.9%; P < .001), 30-day readmission (12.4% vs 11.3%; P < .001), and 90-day readmission (19.0% vs 17.7%; P < .001), the median LOS (9 vs 8 days; P < .001), and the median total cost ($46,999 vs $38,417; P < .001) were greater at SNHs than at non-SNHs. In addition, patients at SNHs were less likely to be discharged with home health care (35.1% vs 43.7%; P < .001) (Table 3).
Table 3Postoperative outcomes of patients who underwent coronary artery bypass grafting at SNH versus non-SNHs
Admissions in the months of October through December excluded. Categorical variables are presented as number (%) and continuous variables are presented as median (interquartile range).
13,476 (19.0%) (n = 70,859)
55,814 (17.7%) (n = 315,631)
<.001
13,427 (19.0%) (n = 70,688)
14,713 (18.6%) (n = 79,307)
.27
SNHs, Safety-net hospitals; LOS, length of stay; SNF, skilled nursing facility; LTACH, long-term acute care hospital.
∗ Admissions in the month of December excluded.
† Admissions in the months of October through December excluded. Categorical variables are presented as number (%) and continuous variables are presented as median (interquartile range).
Our multivariable model of the complete cohort identified several factors associated with in-hospital mortality: congestive heart failure (OR, 3.22; 95% CI, 2.96-3.49), chronic kidney disease (OR, 2.00; 95% CI, 1.85-2.16), age >65 years (OR, 1.99; 95% CI, 1.84-2.14), need for urgent surgery (OR, 1.69; 95% CI, 1.56-1.84), female sex (OR, 1.60; 95% CI, 1.49-1.72), peripheral arterial disease (OR, 1.55; 95% CI, 1.43-1.68), lowest household income quartile (OR, 1.17; 95% CI, 1.08-1.27), and Medicaid/uninsured (OR, 1.17; 95% CI, 1.03-1.34). Of note, hospital safety-net status was not significantly associated with in-hospital mortality (OR, 1.09; 95% CI, 0.98-1.20) (Table 4).
Table 4Multivariable logistic regression model for in-hospital mortality
Variable
Adjusted OR (95% CI)
P value
Congestive heart failure
3.22 (2.96-3.49)
<.001
Chronic kidney disease
2.00 (1.85-2.16)
<.001
Age >65 y
1.99 (1.84-2.14)
<.001
Non-elective surgery
1.69 (1.56-1.84)
<.001
Female sex
1.60 (1.49-1.72)
<.001
Peripheral arterial disease
1.55 (1.43-1.68)
<.001
Lowest household income quartile
1.17 (1.08-1.27)
<.001
Uninsured or insured with Medicaid
1.17 (1.03-1.34)
.02
Safety-net hospital
1.09 (0.98-1.20)
.11
Obesity
0.80 (0.73-0.87)
<.001
Diabetes mellitus
0.80 (0.74-0.86)
<.001
Hypertension
0.56 (0.51-0.62)
<.001
C-statistic = 0.754. Categorical variables are presented as number (%), and continuous variables are presented as median (interquartile range). OR, Odds ratio; CI, confidence interval.
After propensity-score matching, 204,053 patients were included (95,911 at SNHs and 108,142 at non-SNHs). Comparable rates of in-hospital mortality (2.1% vs 1.9%; P = .11), acute kidney injury (21.6% vs 21.3%; P = .54), stroke (2.0% vs 2.0%; P = .90), respiratory failure (20.2% vs 20.3%; P = .85), 30-day readmission (12.3% vs 12.0%; P = .33), and 90-day readmission (19.0% vs 18.7%; P = .52) were observed at SNHs and non-SNHs; however, the rates of pneumonia (6.2% vs 5.0%) and sepsis (2.7% vs 2.0%) were greater at SNHs than at non-SNHs (both P < .001). In addition, the longer median LOS (9 vs 8 days) and greater median total cost ($46,977 vs $39,343) observed at SNHs persisted after matching (both P < .001). After matching, patients at SNHs also remained less likely to be discharged with home health care (35.0% vs 45.8%; P < .001) (Table 3).
In a subanalysis of Medicaid-insured and -uninsured patients, we found that differences in preoperative characteristics between groups were generally similar to those observed in the original cohort (Table E2). Of note, SNHs had more Medicaid patients (81.3% vs 72.9%) whereas non-SNHs had more uninsured/self-pay patients (27.1% vs 18.7%) (both P < .001). Observed and propensity-score matched in-hospital mortality, major morbidity, and 30- and 90-day readmission for this cohort of patients were comparable at SNHs and non-SNHs. Again, after matching, these patients had greater total cost ($50,110 vs $40,318) and were less likely to be discharged with home health care (32.8% vs 39.7%) at SNHs than at non-SNHs (both P < .001) (Table E3).
Discussion
In this retrospective cohort study of patients who underwent isolated CABG in the NRD, as expected, patients at SNHs were more medically and socioeconomically complex and more often required urgent surgery than those at non-SNHs, which is in accordance with previous studies.
Consistent with our hypothesis, observed outcomes were generally inferior at SNHs compared with non-SNHs; however, after matching, we found that most outcomes after CABG were comparable at SNHs and non-SNHs. Interestingly, patients who underwent CABG at SNHs had marginally longer LOS and greater total cost, even after propensity-score matching. Of note, previous studies have identified postoperative complications as a primary driver of variation in cost after CABG.
Although there were comparable rates of most postoperative complications, patients at SNHs had greater rates of pneumonia and sepsis; pneumonia is often cited as a source of prolonged LOS and greater total cost after cardiac surgery.
Pneumonia after cardiac surgery: experience of the National Institutes of Health/Canadian Institutes of Health Research Cardiothoracic Surgical Trials Network.
In part, the differences in LOS and cost may also be attributable to differences in discharge disposition between groups. We found that patients at SNHs were less likely to be discharged with home health care than patients at non-SNHs. Given that outcomes were generally similar between groups, this finding is more likely to reflect disparities in socioeconomic factors and access to skilled care, rather than a difference in patient condition. Consequently, patients at SNHs may have required additional time in the hospital to facilitate safe discharge. Given the substantial overall cost of CABG, targeted strategies to reduce LOS are warranted for SNHs to facilitate cost containment. In addition, readmission rates were comparable at SNHs and non-SNHs in our matched analysis; this is encouraging, given that previous studies have highlighted readmission rates as a metric of concern at SNHs.
Comparison of risk-standardized readmission rates of surgical patients at safety-net and non-safety-net hospitals using Agency for Healthcare Research and Quality and American Hospital Association data.
Comparison of risk-standardized readmission rates of surgical patients at safety-net and non-safety-net hospitals using Agency for Healthcare Research and Quality and American Hospital Association data.
that examined the effect of hospital safety-net burden on the outcomes of 304,080 patients who underwent CABG in the State Inpatient Database, an initial association between safety-net burden and observed in-hospital mortality did not persist after adjustment for baseline patient and hospital characteristics by multivariable regression. Our study extends these findings by providing a matched comparison and more rigorously isolating the effect of hospital safety-net status on outcomes after CABG. Although additional studies have shown comparable early outcomes after cardiac surgery at SNHs versus non-SNHs after adjusting for patient and hospital characteristics,
It is worth noting, however, that failure to rescue may be particularly sensitive to the disproportionate financial pressures and limited hospital resources at SNHs.
In addition, patient- and hospital-level socioeconomic factors are inherently interconnected, rendering studies of these factors complex. Patients at SNHs, many of whom are uninsured, may be more likely to forgo necessary care and screening
are associated with greater morbidity and mortality after cardiac surgery, and they probably exert a synergistic effect on access to cardiac surgical care.
Additional studies are needed to elucidate the complex relationship between patient- and hospital-level socioeconomic factors, their relative effects on outcomes after cardiac surgery, and where the greatest return on investment may lie for policymakers and other stakeholders aiming to improve the quality of cardiac surgical care for vulnerable populations. In general, SNHs are especially vulnerable due to their public funding, low operating margins, and relatively high rate of uncompensated care.
One potential strategy that has been proposed to reduce the rate of uncompensated care and financial strain at SNHs is Medicare/Medicaid expansion, with efforts currently underway.
In designing this study, the primary aims were 2-fold: first, to provide a snapshot of outcomes of CABG at SNHs in reference to non-SNHs, and second, to identify specific areas in which SNHs may fall short of non-SNHs, to provide policymakers and cardiac surgical societies areas to target for improvement. After propensity-score matching to mitigate the effect of confounding baseline covariates, we found that the rates of in-hospital mortality, major morbidity, and readmission at SNHs were comparable with those at non-SNHs. Given the practical barriers to comparing hospital safety-net status in a randomized clinical trial, this propensity-score matched study provides important insight into the controversy as to whether SNHs can provide routine cardiac surgical care with comparable outcomes to those of non-SNHs. In addition, the NRD provides a nationwide sample of all-payer admissions, giving the results of our study strong external validity and generalizability. Although the United States Department of Health and Human Services launched a large, federal action plan aimed at improving health care equity in 2011,
U.S. Department of Health and Human Services HHS action plan to reduce racial and ethnic health disparities: a nation free of disparities in health and health care.
and renewed initiatives to mitigate disparities are warranted. Ultimately, the results of our study are encouraging because they show that patients can obtain similar outcomes whether they undergo CABG at an SNH or a non-SNH. Although beyond the direct scope of the present study, it was also encouraging to find that these hospitals were more likely to be teaching hospitals. In turn, these results support continued funding of cardiac surgical care at SNHs, along with further investigation regarding potential cost-containment strategies such as enhanced care coordination and disposition planning to reduce LOS and minimize unplanned readmissions at SNHs.
Our findings should be interpreted in the context of the inherent limitations associated with retrospective studies and administrative databases. Although the NRD provides a large sample size capable of powering a robust propensity-score matched analysis, administrative data can be incomplete, and they rely on accurate coding. Consequently, patients with limited access to or use of routine health care services before surgery may be “undercoded” with respect to comorbidities. In addition, we attempted to minimize the influence of baseline covariates by using established comorbidity codes; however, the influence of residual confounding factors cannot be fully excluded. Several potential unmeasured confounders influencing the findings of this study include race and ethnicity, time from diagnosis to surgery, preoperative risk modification, frailty, left ventricular ejection fraction, lesion complexity (ie, SYNTAX score), and variation in surgical technique (eg, on- vs off-pump, number of grafts, arterial vs venous grafts). Additional commonly used risk scores, including the Society of Thoracic Surgeons Postoperative Risk of Mortality and European System for Cardiac Operative Risk Evaluation II, were not available within the NRD, precluding an analysis of observed versus expected outcomes between groups. The Society of Thoracic Surgeons Adult Cardiac Surgery Database captures several of the aforementioned data points, which are unavailable in the NRD, and could therefore provide an important extension of the present study. Nevertheless, we expect that if differences in the aforementioned factors did exist, this would bias the results against our hypothesis. In addition, we were not able to account for differences in postoperative resources or failure to rescue, key factors that have been identified as potential areas for improvement at SNHs,
although again, we would expect that these differences would bias the results against our hypothesis. Other important limitations of the NRD are the lack of granularity with respect to intensive care unit versus overall hospital LOS and the availability of intermediate care or “step-down” units, both of which can influence discharge planning and the total cost of admission. Cost was also calculated indirectly by using cost-to-charge ratios provided by the NRD. Last, there is no consensus definition of “safety-net hospital,” and although we used a definition previously described in the literature to facilitate a 1:1 propensity-score matched analysis, the definition varies across studies in this space.
After propensity-score matching, we found that early outcomes after isolated CABG were comparable at SNHs and non-SNHs (Figure 2). Initiatives to reduce LOS through improved discharge planning and resources could curtail cost and improve the value of CABG for vulnerable patients at SNHs.
Figure 2In the present study, we found that early outcomes after coronary artery bypass grafting (CABG) were similar at safety-net hospitals (SNHs) compared with nonsafety-net hospitals (non-SNH) after propensity-score matching.
The Journal policy requires editors and reviewers to disclose conflicts of interest and to decline handling or reviewing manuscripts for which they may have a conflict of interest. The editors and reviewers of this article have no conflicts of interest.
The authors thank Scott A. Weldon, MA, CMI, FMI, for creating the illustrations and Stephen N. Palmer, PhD, ELS, for editorial support.
Categorical variables are presented as number (%), and continuous variables are presented as median (interquartile range). SNH, Safety-net hospital. SMD, standardized mean difference.
Excluding admissions in the months of October through December. Categorical variables are presented as number (%), and continuous variables are presented as median (interquartile range).
2408 (21.0%) (n = 11,443)
5319 (20.2%) (n = 26,353)
.22
2359 (20.9%) (n = 11,271)
2907 (21.5%) (n = 13,508)
.46
SNH, Safety-net hospital; LOS, length of stay; SNF, skilled nursing facility; LTACH, long-term acute care hospital.
∗ Excluding admissions in the month of December.
† Excluding admissions in the months of October through December. Categorical variables are presented as number (%), and continuous variables are presented as median (interquartile range).
Figure E1Assessment of covariate balance before and after propensity-score matching. A, Love plot of standardized mean differences before and after matching. B, Density plot of propensity scores before and after matching. SNH, Safety-net hospital.
Figure E1Assessment of covariate balance before and after propensity-score matching. A, Love plot of standardized mean differences before and after matching. B, Density plot of propensity scores before and after matching. SNH, Safety-net hospital.
Comparison of risk-standardized readmission rates of surgical patients at safety-net and non-safety-net hospitals using Agency for Healthcare Research and Quality and American Hospital Association data.
In-hospital outcomes of ST-segment elevation myocardial infarction complicated with cardiogenic shock at safety-net hospitals in the United States (from the Nationwide Inpatient Sample).
Pneumonia after cardiac surgery: experience of the National Institutes of Health/Canadian Institutes of Health Research Cardiothoracic Surgical Trials Network.
00017-7&id=fx1.jpg){kind=link}
00017-7&id=fx2.jpg){kind=link}
00017-7&id=gr1.jpg){kind=link}
00017-7&id=gr2.jpg){kind=link}
00017-7&id=fx3a.jpg){kind=link}
00017-7&id=fx3b.jpg){kind=link}