STRUCTURED ABSTRACT
Objective:
The influence of socioeconomic determinants of health on failure to rescue (FTR; mortality after a post-operative complication) after cardiac surgery is unknown. We hypothesized that increasing Distressed Communities Index (DCI), a comprehensive socioeconomic ranking by zip code, would be associated with higher FTR.
Methods:
Patients undergoing STS index operation in a regional collaborative (2011-2021) who developed a FTR complication were included. After excluding patients with missing zip code or STS predicted risk of mortality, patients were stratified by DCI scores (0-no distress, 100-severe distress) based on education level, poverty, unemployment, housing vacancies, median income, and business growth. The upper two quintiles of distress (DCI > 60) were compared to all other patients. Hierarchical logistic regression analyzed the association between DCI and FTR.
Results:
A total of 4,004 patients developed one or more of the defined complications across 17 centers. Of these, 582 (14.5%) experienced failure to rescue. High socioeconomic distress, (DCI > 60) was identified among 1272 patients (31.8%). Prior to adjustment, FTR occurred more frequently among those from socioeconomically distressed communities (DCI > 60; 16.9% vs. 13.4%, p = 0.004). After adjustment, residing in a socioeconomically distressed community was associated with 24% increased odds of FTR (OR 1.24 CI 1.003-1.54, p = 0.044).
Conclusions:
Increasing DCI, a measure of poor socioeconomic status, is associated with greater risk-adjusted likelihood of FTR following cardiac surgery. These findings highlight that current quality metrics do not account for socioeconomic status, and as such underrepresent procedural risk for these vulnerable patients.
Keywords: Failure to Rescue, Disparities, Outcomes
INTRODUCTION
In 1966, Dr. Martin Luther King asserted that, “Of all the forms of inequality, injustice in health care is the most shocking and inhumane.”1 In the years since King’s prescient attestation, a substantial body of research has explored the effect of socioeconomic status (SES) on health.2,3 In the field of cardiac surgery, evidence is accumulating that lower SES is associated with higher post-operative morbidity and mortality.4 Failure to rescue (FTR), defined as mortality after a postoperative complication, is a newly adopted Society of Thoracic Surgeons (STS) Adult Cardiac Surgery Database (ACSD) quality metric which promises to identify potentially preventable factors associated with mortality following cardiac surgery.
What influence that SES has on failure to rescue in cardiac surgery is unknown. Most studies have used limited proxies of SES, such as neighborhood income or insurance status, to estimate the strength of the relationship between SES and postoperative outcomes.5,6 It is therefore unclear whether demographic factors, insurance status, hospital safety net status, or broader measures of SES are driving differences in healthcare outcomes. Moreover, prior reports that have used more comprehensive measures of SES are limited in their inference by sample size and reliance on narrowly defined outcomes.7-9 To accurately delineate and disentangle the effects of socioeconomic determinants of health on clinical outcomes, use of a more comprehensive estimate of SES with a robust dataset is required.
One such database is Distressed Communities Index (DCI). Developed by the Economic Innovation Group (EIG), a bipartisan public policy organization “dedicated to forging a more dynamic and inclusive American economy”, the DCI aims to describe the socioeconomic health of communities in the United States.10 This estimate of social and economic position is calculated for each zip code and, is increasingly becoming appreciated as a valuable, direct measure of SES.11 Specifically, the DCI is calculated based on the values of seven metrics: percent of residents with a high school degree, housing vacancy rate, unemployment rate, poverty rate, median income ratio, change in employment, and change in business establishments. The index ranges from 0 to 100; a score of zero corresponds to a prosperous (or non-distressed community), whereas a score of 100 corresponds to a maximally distressed community.
The objective of this study was to assess the effect of SES, measured using the DCI, on failure to rescue (FTR) after cardiac surgery. We hypothesized that after adjusting for patient demographics, race, insurance, comorbidities, procedure type, year, and hospital random effect, high socioeconomic distress would still be a significant predictor of FTR.
METHODS
The Virginia Cardiac Services Quality Initiative includes 17 hospitals and surgical groups in Virginia. Virginia Cardiac Services Quality Initiative data include 99% of all adult cardiac surgeries in the region. Clinical data and cost methodology have been described previously12-13. STS data from individual centers are compiled in a central registry. This study was exempt from review by the University of Virginia’s Institutional Review Board due to the de-identified nature of the quality database (Protocol #23305, deemed exempt July 14th, 2021).
All patients undergoing a Society of Thoracic Surgeons (STS) index operation (CABG, AVR, MVR/r) between July 2011 and July 2021 were extracted from the Virginia Cardiac Services Quality Initiative database. These data were linked with the Economic Innovation Group’s (EIG) Distressed Communities Index (DCI) data file. Use of the EIG DCI in cardiac surgery has been described previously.4 Briefly, the DCI is an index used to rank the socioeconomic health of zip codes based on seven key indicators: percent of residents with a high school degree, housing vacancy rate, unemployment rate, poverty rate, median income ratio, change in employment, and change in business establishments. Patients were excluded if they did not experience an STS failure to rescue complication14, were missing Zip Code (required for linkage with the EIG DCI file), STS predicted risk of mortality, or underwent a non-index procedure (Figure 1).
Figure 1.
CONSORT Diagram
Standard Society of Thoracic Surgeons (STS) definitions were used for all variables.15 Operative mortality is defined as in-hospital mortality or death with in 30-day of discharge. Failure to rescue is defined as operative mortality after an STS-defined FTR complication (prolonged ventilation, post-operative renal failure requiring dialysis, reoperation, and stroke).
Patients were stratified by Distressed Community Index. The most distressed patients (top two quintiles, DCI ≥ 60) were compared to all other patients. Median imputation was utilized for missingness (all missingness was less than 5%). Categorical variables are presented as counts (%) and continuous variables are presented as median (interquartile range) due to skewed distributions. Wilcoxon rank sum test was used for non-normal distributed continuous variables and the χ2 test for all categorical variables. LOESS analysis of the relationship between DCI and failure to rescue was employed to assess relationship linearity. Hierarchical logistic regression modeled the association between DCI and failure to rescue, adjusting for patient demographics, comorbid conditions, and operative characteristics. Hospital was included as a random effect to account for center-level differences. Variables utilized in the final multivariable model were selected based on clinical importance and statistical significance in univariate analyses. All statistical analyses were carried out using SAS Version 9.4 (SAS Institutive, Cary, NC) with a p-value less than 0.05 determining significance.
RESULTS
Baseline characteristics by DCI and FTR
A total of 4,004 patients who experienced a major complication were identified during the study period. Of these, 1,272 (31.8%) were in the highest two quintiles of DCI (DCI ≥ 60; Figure 2). Compared to patients with DCI < 60, patients with DCI ≥ 60 were younger (66 years vs 68 years, p = 0.004), more often of black race (26.5 vs. 12.7%, p < 0.001), and had a greater burden of peripheral arterial disease (19.8 vs. 16.8%, p = 0.019), hypertension (89.1 vs 85.1%, p < 0.001), end-stage renal disease (6.68 vs. 4.90%, p = 0.021), current tobacco use (23.7 vs. 20.6%, p = 0.011), oxygen-dependent lung disease (1.18 vs. 0.59%, p = 0.046), and previous percutaneous coronary intervention (PCI; 28.5 vs. 24.9%, p = 0.018; Table 1). STS Predicted risk of mortality was not significantly different between patients with DCI < 60 and those with DCI ≥ 60 (2.78 vs. 2.78%, p =0.627).
Figure 2 and Central Picture.
Mortality, major complications, and FTR by DCI quintile. Univariate logistic regression used for comparisons, with the lowest quintile of DCI as reference value.
Table 1.
Baseline characteristics by DCI
| Characteristic | High Socioeconomic Distress (DCI ≥ 60, n = 1272, 31.8%) |
Not Distressed (DCI < 60, n = 2732, 68.2%) |
p-value |
|---|---|---|---|
| Age | 66 (59-73) | 68 (60-75) | 0.004 |
| Year | 5.00 (3.00 – 8.00) | 5.00 (3.00 – 8.00) | 0.587 |
| Health Insurance | 0.475 | ||
| None | 856 (67.3%) | 1799 (65.8%) | |
| Private Insurance | 117 (9.2%) | 269 (9.85%) | |
| Medicaid | 40 (3.14%) | 62 (2.27%) | |
| Medicare | 248 (19.5%) | 569 (20.8%) | |
| Military | 9 (0.71%) | 23 (0.84%) | |
| Non-US Plan | 0 | 2 (0.07%) | |
| State-specific plan | 2 (0.15%) | 5 (0.18%) | |
| Other Government Insurance | 0 | 3 (0.19%) | |
| Any Health Insurance | 416 (32.7%) | 934 (34.2%) | 0.355 |
| Immunocompromised | 63 (4.95%) | 151 (5.53%) | 0.452 |
| Peripheral Arterial Disease | 252 (19.8%) | 458 (16.8%) | 0.019 |
| Hypertension | 1133 (89.1) | 2326 (85.1) | <0.001 |
| Diabetes | 626 (39.2%) | 1254 (45.9%) | 0.051 |
| Prior Stroke | 185 (14.5%) | 354 (13.0%) | 0.171 |
| Cerebrovascular Disease | 354 (27.8%) | 709 (25.9%) | 0.381 |
| Race | <0.001 | ||
| White | 903 (71.0%) | 2165 (79.3%) | |
| Black | 337 (26.49%) | 348 (12.7%) | |
| American Indian | 1 (0.08%) | 2 (0.07%) | |
| Asian | 14 (1.10%) | 113 (4.14%) | |
| Other | 17 (1.33%) | 104 (3.79%) | |
| Female gender | 478 (37.58%) | 943 (34.5%) | 0.059 |
| MELD Score | 8.15 (7.47-10.9) | 8.15 (7.31-10.9) | 0.192 |
| Liver Disease | 87 (6.84%) | 169 (6.19%) | 0.530 |
| Pre-operative Creatinine | 1.08 (0.87-1.40) | 1.00 (0.80-1.40) | 0.086 |
| Pre-operative Hemoglobin | 12.5 (11.0-13.9) | 12.6 (11.0-14.0) | 0.458 |
| Pre-operative WBC | 7.90 (6.40-10.1) | 8.10 (6.54-10.1) | 0.337 |
| Pre-operative Albumin | 3.80 (3.30-4.00) | 3.70 (3.30-4.00) | 0.579 |
| Body Surface Area | 2.84 (2.72 – 2.97) | 2.87 (2.71 -2.97) | 0.572 |
| End-stage Renal Disease | 85 (6.68%) | 134 (4.90%) | 0.021 |
| Tobacco Use | 0.011 | ||
| None | 602 (47.3%) | 1426 (52.2%) | |
| Current | 301 (23.7%) | 562 (20.6%) | |
| Former | 369 (29.0%) | 744 (27.2%) | |
| Oxygen-dependent lung disease | 15 (1.18%) | 16 (0.59%) | 0.046 |
| Chronic Lung Disease | 486 (38.2%) | 1027 (37.6%) | 0.710 |
| Sleep Apnea | 227 (17.9%) | 472 (17.3%) | 0.659 |
| Pre-operative Arrhythmia, within 30 days | 172 (13.5%) | 459 (16.8%) | 0.003 |
| CHF | 609 (47.9%) | 1323 (48.4%) | 0.746 |
| Previous PCI | 362 (28.5%) | 681 (24.9%) | 0.018 |
| Previous CABG | 52 (4.09%) | 117 (4.28%) | 0.776 |
| Previous Valve Surgery | 60 (4.72%) | 142 (5.20%) | 0.518 |
| Prior MI | 707 (55.6%) | 1438 (52.6%) | 0.082 |
| Reoperative Surgery | 101 (7.94%) | 234 (8.57%) | 0.705 |
| Aortic Stenosis | 264 (20.8%) | 572 (20.9%) | 0.895 |
| Aortic Regurgitation | 0.709 | ||
| None | 806 (63.4%) | 1790 (65.5%) | |
| Trivial/Trace | 184 (14.5%) | 387 (14.2%) | |
| Mild | 179 (14.1%) | 353 (12.9%) | |
| Moderate | 60 (4.72%) | 120 (4.39%) | |
| Severe | 43 (3.37%) | 82 (3.01%) | |
| Mitral Stenosis | 65 (5.11%) | 136 (4.98%) | 0.859 |
| Mitral Regurgitation | 0.057 | ||
| None | 296 (23.3%) | 726 (26.6%) | |
| Trivial/Trace | 250 (19.6%) | 519 (19.0%) | |
| Mild | 336 (26.4%) | 724 (26.5%) | |
| Moderate | 197 (15.5%) | 345 (12.6%) | |
| Severe | 193 (15.2%) | 418 (15.3%) | |
| Tricuspid Stenosis | 3 (0.24%) | 7 (0.26%) | 0.904 |
| Tricuspid Regurgitation | 0.143 | ||
| None | 409 (32.1%) | 980 (35.9%) | |
| Trivial/Trace | 357 (28.1%) | 706 (25.8%) | |
| Mild | 341 (26.8%) | 714 (26.1%) | |
| Moderate | 128 (10.1%) | 271 (9.97%) | |
| Severe | 37 (2.90%) | 61 (2.23%) | |
| Pre-operative Ejection Fraction | 53.0 (40.0-60.0) | 55.0 (40.0 – 60.0) | 0.315 |
| Status | 0.0730 | ||
| Elective | 456 (35.8%) | 1000 (36.6%) | |
| Urgent | 710 (55.8%) | 1441 (52.7%) | |
| Emergent | 97 (7.64%) | 258 (9.46%) | |
| Emergent Salvage | 9 (0.76%) | 33 (1.24%) | |
| Intra-aortic balloon pump | 304 (23.9%) | 691 (25.3%) | 0.342 |
| Procedure Type | 0.358 | ||
| AV Replacement | 121 (9.51%) | 275 (10.1%) | |
| AV Replacement + CABG | 127 (9.98%) | 280 (10.25%) | |
| Isolated CABG | 757 (59.5%) | 1610 (58.9%) | |
| MV Repair | 47 (3.69%) | 132 (4.83%) | |
| MV Repair + CABG | 67 (5.27%) | 117 (4.28%) | |
| MV Replacement + CABG | 39 (3.07%) | 100 (3.66%) | |
| MV Replacement Only | 114 (8.98%) | 218 (7.98%) | |
| Cross clamp time | 81.0 (62.0-111) | 81.0 (64.0 – 108) | 0.998 |
| Cardiopulmonary Bypass Time | 118 (87.0 – 162) | 114 (89.0 – 155) | 0.333 |
| Intraoperative blood products, units | 0 (0-2) | 0 (0-2) | 0.02 |
| STS Predicted Risk of Mortality | 2.78 (1.29 – 6.14) | 2.78 (1.33 -6.24) | 0.627 |
Failure to rescue occurred in 582 patients (14.5%). Relative to patients without failure to rescue, patients who experienced failure to rescue were older (70 vs. 67 years, p < 0.001), more likely to have some form of health insurance (40.6% vs. 32.5%, p < 0.001), more often of female gender (42.1% vs. 34.4%, p < 0.001), had lower preoperative ejection fraction (53.0 vs. 55.0, p = 0.010), had a greater burden of immunodeficiency (8.25 vs. 4.85%, p < 0.001), peripheral arterial disease (25.6 vs. 16.4%, p < 0.001), cerebrovascular disease (30.2 vs. 25.9%, p = 0.010), liver disease (9.11 vs. 5.93%, p = 0.011), end-stage renal disease (9.11 vs. 4.85%, p < 0.001), congestive heart failure (CHF; 54.5 vs. 47.2%, p = 0.001), previous CABG (7.04 vs. 3.74%, p < 0.001), prior MI (59.6 vs. 52.5%, p < 0.001), mitral stenosis (8.42 vs. 4.44%, p < 0.001), were less likely to be elective status (30.9 vs. 37.3%, p < 0.001), more frequently required intra-aortic balloon pump (34.5 vs. 23.2%, p < 0.001), had longer cross clamp (83.0 vs. 80.0 minutes, p = 0.019) and cardiopulmonary bypass times (127 vs. 114 minutes, p < 0.001), required more intraoperative blood products (1.00 vs. 0.00 median units, p < 0.001) and had a higher STS predicted risk of mortality (5.10 vs. 2.58%, p < 0.001; Table 2). The components of DCI stratified by both DCI and FTR are included in Supplemental Table 1.
Table 2.
Baseline characteristics by FTR
| Characteristic | Failure to Rescue (14.5%, n = 582) |
No Failure to Rescue (85.46%, n = 3422) |
p-value |
|---|---|---|---|
| Age | 70.0 (63.0-77.0) | 67.0 (59.0-74.0) | <0.001 |
| Year | 6.0 (4.0-9.0) | 5.0 (3.0-8.0) | <0.001 |
| Health Insurance | <0.001 | ||
| None | 346 (59.5%) | 2309 (67.5%) | |
| Private Insurance | 43 (7.39%) | 343 (10.0%) | |
| Medicaid | 20 (3.44%) | 82 (2.40%) | |
| Medicare | 8 (1.37%) | 24 (0.70%) | |
| Military | 1 (0.17%) | 1 ( 0.03%) | |
| Non-US Plan | 1 (0.17%) | 6 (0.18%) | |
| State-specific plan | 0 | 0 | |
| Other Government Insurance | 0 | 3 (0.09%) | |
| Any Health Insurance | 236 (40.6%) | 1114 (32.5%) | <0.001 |
| Immunocompromised | 48 (8.25%) | 166 (4.85%) | <0.001 |
| Peripheral Arterial Disease | 149 (25.6%) | 561 (16.4%) | <0.001 |
| Hypertension | 513 (88.1%) | 2946 (86.1%) | 0.181 |
| Diabetes | 280 (48.1%) | 1600 (46.8%) | 0.545 |
| Prior Stroke | 89 (15.3%) | 450 (13.1) | 0.162 |
| Cerebrovascular Disease | 176 (30.2%) | 887 (25.9%) | 0.010 |
| Race | 0.797 | ||
| White | 453 (77.8%) | 2615 (76.42%) | |
| Black | 91 (15.6%) | 594 (17.4%) | |
| American Indian | 0 | 3 (0.09%) | |
| Asian | 20 (3.44%) | 107 (3.13%) | |
| Other | 18 (3.09%) | 103 (3.01%) | |
| Female gender | 245 (42.1%) | 1176 (34.4%) | <0.001 |
| High Socioeconomic Distress | 215 (16.9%) | 367 (13.4%) | 0.004 |
| MELD Score | 9.23 (7.47-13.4) | 7.67 (7.31-10.7) | <0.001 |
| Liver Disease | 53 (9.11%) | 203 (5.93%) | 0.011 |
| Pre-operative Creatinine | 1.1 (0.90-1.54) | 1.0 (0.80-1.30) | <0.001 |
| Pre-operative Hemoglobin | 12.0 (10.3-13.4) | 12.7 (11.0-14.0) | <0.001 |
| Pre-operative WBC | 8.50 (6.55-11.1) | 8.00 (6.50-9.98) | <0.001 |
| Pre-operative Albumin | 3.60 (3.10-3.90) | 3.80 (3.30-4.00) | <0.001 |
| Body Surface Area | 2.83 (2.67-2.97) | 2.87 (2.72-2.98) | <0.001 |
| End-stage Renal Disease | 53 (9.11%) | 166 (4.85%) | <0.001 |
| Tobacco Use | 0.097 | ||
| None | 292 (50.2%) | 1736 (50.7%) | |
| Current | 110 (18.9%) | 753 (22.0%) | |
| Former | 180 (30.9%) | 933 (27.3%) | |
| Oxygen-dependent lung disease | 8 (1.37%) | 23 (0.67%) | 0.074 |
| Chronic Lung Disease | 239 (41.1%) | 1274 (37.2%) | 0.078 |
| Sleep Apnea | 88 (15.1%) | 611 (17.9%) | 0.108 |
| Pre-operative Arrhythmia, within 30 days | 118 (20.3%) | 513 (15.0%) | <0.001 |
| CHF | 317 (54.5%) | 1615 (47.2%) | 0.001 |
| Previous PCI | 168 (29.9%) | 875 (25.6%) | 0.094 |
| Previous CABG | 41 (7.04%) | 128 (3.74%) | <0.001 |
| Previous Valve Surgery | 36 (6.19%) | 166 (4.85%) | 0.174 |
| Prior MI | 347 (59.6%) | 1798 (52.5%) | <0.001 |
| Reoperative Surgery | 59 (10.1%) | 276 (8.07%) | 0.095 |
| Aortic Stenosis | 133 (22.9%) | 703 (20.5%) | 0.205 |
| Aortic Regurgitation | 0.963 | ||
| None | 384 (66.0%) | 2212 (64.6%) | |
| Trivial/Trace | 82 (14.1%) | 489 (14.3%) | |
| Mild | 75 (12.9%) | 457 (13.4%) | |
| Moderate | 25 (4.30%) | 155 (4.53%) | |
| Severe | 16 (2.75%) | 109 (3.19%) | |
| Mitral Stenosis | 49 (8.42%) | 152 (4.44%) | <0.001 |
| Mitral Regurgitation | 0.001 | ||
| None | 119 (20.5%) | 903 (26.4%) | |
| Trivial/Trace | 104 (17.9%) | 665 (19.4%) | |
| Mild | 156 (26.8%) | 903 (26.4%) | |
| Moderate | 103 (17.7%) | 439 (12.8%) | |
| Severe | 100 (17.2%) | 511 (14.9%) | |
| Tricuspid Stenosis | 1 (0.17%) | 9 (0.26%) | 0.684 |
| Tricuspid Regurgitation | <0.001 | ||
| None | 168 (28.9%) | 1221 (35.7%) | |
| Trivial/Trace | 147 (25.3%) | 916 (26.8%) | |
| Mild | 171 (29.4%) | 884 (25.8%) | |
| Moderate | 70 (12.0%) | 329 (9.61%) | |
| Severe | 26 (4.47%) | 72 (2.10%) | |
| Pre-operative Ejection Fraction | 53.0 (37.0-60.0) | 55 (40.0-60.0) | 0.010 |
| Status | <0.001 | ||
| Elective | 180 (30.9%) | 1276 (37.3%) | |
| Urgent | 321 (55.2%) | 1830 (53.5%) | |
| Emergent | 67 (11.5%) | 288 (8.42%) | |
| Emergent Salvage | 14 (2.41%) | 28 (0.82%) | |
| Intra-aortic balloon pump | 201 (34.5%) | 794 (23.2%) | <0.001 |
| Procedure Type | 0.119 | ||
| AV Replacement | 46 (7.90%) | 350 (10.2%) | |
| AV Replacement + CABG | 63 (10.8%) | 344 (10.0%) | |
| Isolated CABG | 328 (56.4%) | 2039 (59.6%) | |
| MV Repair | 29 (4.98%) | 150 (4.38%) | |
| MV Repair + CABG | 33 (5.67%) | 151 (4.41%) | |
| MV Replacement + CABG | 27 (4.64%) | 112 (3.27%) | |
| MV Replacement Only | 56 (9.62%) | 276 (8.07%) | |
| Cross clamp time | 83.0 (65.0-119) | 80.0 (63.0-107) | 0.019 |
| Cardiopulmonary Bypass Time | 127 (93.0-180.0) | 114 (88.0-153.0) | <0.001 |
| Intraoperative blood products, units | 1.00 (0-3) | 0 (0-2) | <0.001 |
| STS Predicted Risk of Mortality | 5.10 (2.18-12.4) | 2.58 (1.25-5.44) | <0.001 |
Unadjusted outcomes by DCI
Postoperative outcomes stratified by DCI are presented in Table 3. Failure to rescue occurred more frequently among patients with DCI ≥ 60, relative to those with DCI < 60 (16.9 vs. 13.4%, p = 0.003). Higher FTR among the DCI ≥ 60 was primarily driven by a significantly higher rate of prolonged ventilation (79.9 vs. 76.7%, p =0.017) and a qualitatively higher rate of renal failure requiring dialysis (16.8 vs. 14.6%, p = 0.080). Significant differences by DCI were also present in non-FTR complication rates. Specifically, deep sternal wound infection was more common among DCI ≥ 60 patients (1.26 vs. 0.48%, p = 0.007), as was cardiac arrest (15.3 vs. 11.7%, p = 0.001). The relationship between DCI and FTR as assessed by LOESS analysis was mostly linear (Supplemental Figure 1).
Table 3.
Post-operative Outcomes by DCI
| Outcomes | High Socioeconomic Distress (DCI ≥ 60, n = 1272, 31.8%) |
Not Distressed (DCI < 60, n = 2732, 68.2%) |
p-value |
|---|---|---|---|
| Failure to Rescue (Major morbidity AND operative mortality)* | 215 (16.9%) | 367 (13.4%) | 0.003 |
| Prolonged Ventilation | 1017 (79.9%) | 2092 (76.7%) | 0.017 |
| Permanent Stroke | 142 (11.2%) | 306 (11.2%) | 0.972 |
| Reoperation | 313 (24.6%) | 681 (24.9%) | 0.827 |
| Renal failure requiring dialysis | 213 (16.8%) | 399 (14.6%) | 0.080 |
| Readmission | 157 (13.2%) | 289 (10.6%) | 0.098 |
| Deep Sternal Wound Infection | 16 (1.26%) | 13 (0.48%) | 0.007 |
| Any surgical site infection | 37 (2.91%) | 66 (2.42%) | 0.359 |
| Atrial fibrillation | 470 (37.0%) | 1027 (37.6%) | 0.700 |
| Cardiac arrest | 195 (15.3%) | 319 (11.7%) | 0.001 |
| Pneumonia | 176 (13.8%) | 434 (15.9%) | 0.093 |
| DVT/PE | 73 (5.74%) | 163 (5.97%) | 0.776 |
| Readmission to ICU | 157 (12.3%) | 392 (14.4%) | 0.180 |
| Total ICU Hours | 169 (95.5 – 317) | 159 (93.7 – 307) | 0.020 |
| Total Length of Stay, Days | 13.0 (8.0 – 21.0) | 12.0 (8.0 – 20.0) | 0.150 |
| Discharge Home | 498 (39.2 %) | 1202 (44.0%) | <0.001 |
Major morbidity defined as any of the following complications: Prolonged ventilation, New-onset dialysis requirement, reoperation, or stroke.
Multivariable Model of FTR
After risk-adjustment, DCI ≥ 60 was independently predictive of FTR (OR 1.24, CI 1.01 −1.54, p = 0.044; Figures 3 and 4). Full model output is included in Table 4. This effect was qualitatively unchanged when DCI was modeled in its continuous form (OR 1.005 CI 1.001-1.009, p = 0.018). The c-statistic of the final multivariable model was 0.73.
Figure 3.
Risk-adjusted association of socioeconomic determinants of health with FTR. a Black race compared to white race, female gender compared to male, private insurance compared to no insurance. Multivariable model includes adjustment for patient demographics, race, insurance, comorbidities, procedure type, year and hospital random effect.
Figure 4.
Graphical Abstract
Table 4.
Multivariable logistic regression of the effect of DCI on FTR
| Characteristic | Odds Ratio | 95% CI | p-value |
|---|---|---|---|
| High Socioeconomic Distress | 1.24 | 1.01-1.54 | 0.044 |
| Age, years | 1.04 | 1.02-1.05 | <0.001 |
| Year | 1.08 | 1.02-1.15 | 0.010 |
| Health Insurance | 0.484 | ||
| None | 0.83 | 0.512-1.35 | |
| Private Insurance | 1.49 | 0.786-2.84 | |
| Medicaid | 1.04 | 0.706-1.52 | |
| Medicare | 1.88 | 0.739-4.80 | |
| Military | 4.64 | 0.269-80.1 | |
| Non-US Plan | 1.21 | 0.123-12.0 | |
| State-specific plan | 0.008 | <0.001->999 | |
| Other Government Insurance | 0.784 | 0.458-1.34 | |
| Immunocompromised | 1.36 | 0.935-1.97 | 0.108 |
| Peripheral Arterial Disease | 1.48 | 1.17-1.88 | 0.001 |
| Hypertension | 0.988 | 0.728-1.34 | 0.937 |
| Diabetes | 0.919 | 0.749-1.13 | 0.420 |
| Prior Stroke | 1.15 | 0.806-1.63 | 0.447 |
| Cerebrovascular Disease | 0.927 | 0.697-1.23 | 0.225 |
| Race | 0.972 | ||
| White | Ref | Ref | |
| Black | 0.923 | 0.694-1.23 | |
| American Indian | 0.007 | <0.001->999 | |
| Asian | 1.03 | 0.582-1.82 | |
| Other | 0.930 | 0.524-1.65 | |
| Female gender | 1.25 | 0.954-1.63 | 0.106 |
| MELD Score | 1.07 | 1.04-1.11 | <0.001 |
| Liver Disease | 1.71 | 1.20-2.45 | 0.012 |
| Pre-operative Creatinine, mg/dL | 0.904 | 0.817-1.00 | 0.048 |
| Pre-operative Hemoglobin, g/dL | 0.985 | 0.931-1.04 | 0.600 |
| Pre-operative WBC, cells per cubic mm | 1.035 | 1.01-1.06 | 0.001 |
| Pre-operative Albumin, g/dL | 0.796 | 0.655-0.968 | 0.022 |
| Body Surface Area, m2 | 0.630 | 0.328-1.21 | 0.166 |
| End-stage Renal Disease | 1.41 | 0.819-2.41 | 0.217 |
| Tobacco Use | 0.722 | ||
| None | Ref | Ref | |
| Current | 0.890 | 0.670-1.18 | |
| Former | 0.965 | 0.760-1.22 | |
| Oxygen-dependent lung disease | 1.23 | 0.502-3.00 | 0.654 |
| Chronic Lung Disease | 1.00 | 0.815-1.24 | 0.974 |
| Sleep Apnea | 0.754 | 0.572-0.995 | 0.046 |
| Pre-operative Arrhythmia, within 30 days | 1.10 | 0.84-1.43 | 0.311 |
| CHF | 0.955 | 0.757-1.20 | 0.695 |
| Previous PCI | 1.11 | 0.885-1.39 | 0.347 |
| Previous CABG | 2.39 | 1.21-4.70 | 0.012 |
| Previous Valve Surgery | 0.850 | 0.467-1.54 | 0.594 |
| Prior MI | 1.03 | 0.806-1.31 | 0.823 |
| Reoperative Surgery | 0.607 | 0.313-1.18 | 0.139 |
| Aortic Stenosis | 1.43 | 0.982-2.09 | 0.062 |
| Aortic Regurgitation | 0.122 | ||
| None | Ref | Ref | |
| Trivial/Trace | 0.781 | 0.585-1.04 | |
| Mild | 0.718 | 0.527-0.979 | |
| Moderate | 0.640 | 0.387-1.06 | |
| Severe | 0.761 | 0.377-1.53 | |
| Mitral Stenosis | 2.06 | 1.33-3.19 | 0.001 |
| Mitral Regurgitation | 0.895 | ||
| None | Ref | Ref | |
| Trivial/Trace | 1.02 | 0.733-1.41 | |
| Mild | 0.953 | 0.699-1.30 | |
| Moderate | 1.05 | 0.721-1.53 | |
| Severe | 0.853 | 0.508-1.43 | |
| Tricuspid Stenosis | 0.282 | 0.029-2.73 | 0.274 |
| Tricuspid Regurgitation | 0.219 | ||
| None | Ref | Ref | |
| Trivial/Trace | 1.12 | 0.843-1.49 | |
| Mild | 1.22 | 0.920-1.61 | |
| Moderate | 1.01 | 0.696-1.48 | |
| Severe | 1.83 | 1.02-3.30 | |
| Pre-operative Ejection Fraction, % | 0.995 | 0.987-1.00 | 0.212 |
| Status | 0.400 | ||
| Elective | Ref | Ref | |
| Urgent | 0.895 | 0.689-1.16 | |
| Emergent | 0.947 | 0.634-1.41 | |
| Emergent Salvage | 1.62 | 0.738-3.57 | |
| Intra-aortic balloon pump | 1.65 | 1.30-2.10 | <0.001 |
| Procedure Type | 0.212 | ||
| AV Replacement | 0.784 | 0.458-1.34 | |
| AV Replacement + CABG | 0.660 | 0.405-1.076 | |
| Isolated CABG | Ref | Ref | |
| MV Repair | 1.34 | 0.694-2.60 | |
| MV Repair + CABG | 0.789 | 0.462-1.35 | |
| MV Replacement + CABG | 0.779 | 0.444-1.37 | |
| MV Replacement Only | 0.947 | 0.634-1.41 | |
| Cardiopulmonary Bypass Time | 1.01 | 1.01-1.01 | <0.001 |
| Intraoperative blood product transfusion, units | 1.04 | 1.01-1.07 | 0.019 |
DISCUSSION
In this regional retrospective cohort study low SES, as defined by the Distressed Communities Index, was significantly associated with risk-adjusted failure to rescue among patients undergoing STS index procedures. This observation appears to be driven primarily by higher rates of prolonged ventilation and renal failure requiring dialysis among low SES patients. Similar to prior reports, patients who are members of the most distressed two quintiles of communities (DCI ≥ 60) experience significantly higher rates of housing vacancy, poverty, unemployment, business closure, and lower rates of educational achievement, relative to patients in less distressed communities (DCI < 60). Patients who experienced FTR were members of communities with higher rates of vacancy, poverty, unemployment, business closure, and lower rates of educational achievement, relative to those who did not experience FTR.
Failure to rescue is an established quality metric by which to measure a system’s response to postoperative complications, and has been adopted by the Agency for Healthcare Quality and Research (AHRQ) as a patient safety indicator.16 Reddy et al. were among the first to apply this metric in cardiac surgery, demonstrating that variation in center-level observed-to-expected mortality may be explained in part by their observed-to-expected FTR ratio.17 Edwards and coauthors built upon this with their national study of the STS ACSD, which validated the relationship between increasing center-level mortality and increasing FTR rates.18 Most recently, Likosky and colleagues studied FTR using a ‘collaborative of collaboratives’ consisting of 90 hospitals across the United States, and identified significant interhospital variation in mortality rates which were driven primarily by failure to rescue rather than overall complication rates.19 In response to this growing body of evidence, the STS recently announced that FTR would be incorporated into the ACSD as a risk-adjusted metric to further assist in benchmarking and quality improvement.14 While this new risk-adjusted metric is the product of extremely sophisticated statistical technique, it importantly does not incorporate any direct measures of socioeconomic status. However, the decision to include Area Deprivation Index, an index similar to the DCI which reports socioeconomic health of census tracks using extended 9-digit ZIP code, in future releases of the STS ACSD is encouraging and will greatly enhance efforts tailored towards enhancing equity in cardiothoracic surgery. 20,21
Stratification of our cohort by DCI suggests how the socioeconomic condition of a patient’s community may contribute to disparities in outcomes. Patients identified as members of DCI ≥ 60 communities more frequently had a history of prior PCI (28.5 vs. 24.9%, p =0.018), end-stage renal disease (6.68 vs. 4.90%, p = 0.021) and hypertension (89.1 vs. 85.1%, p < 0.001). These typically sub-acute to chronic disease processes may be a reflection of high DCI communities’ relative lack of access to primary care physicians, as well as other local inputs, which do not promote health-positive choices. However, patients in the highest two quintiles of DCI were not significantly more likely to require a preoperative intraaortic balloon pump (23.9% vs. 25.3%, p = 0.342), undergo a non-elective procedure (35.8% vs. 36.6%, p = 0.073), and had statistically equivalent STS predicted risk of mortality (2.78 vs. 2.78%, p = 0.627), relative to patients from the bottom three quintiles of DCI. These findings suggest that among patients who develop major postoperative complications, preoperative presentation and comorbidities should not be used as a ‘catch-all’ by which to explain the entirety of their complicated postoperative course. Notably, patients with DCI ≥ 60 were significantly more often of non-white race than those with DCI < 60 (29.0 vs. 20.7, p < 0.001).
Our manuscript adds to the growing body of work focused on the association of SES with surgical outcomes. In a national study of the STS Adult Cardiac Surgery Database, Mehaffey and coauthors reported that the DCI remained significantly associated with mortality (OR 1.12, p < 0.001) and composite morbidity and mortality (OR 1.03, p = 0.002) even after adjustment for the STS predicted risks for those respective outcomes.8 Here, membership in communities with DCI ≥ 60 was associated with 24% increased odds of failure to rescue after developing an STS major complication. Mehaffey and colleagues work, and the findings of this analysis, suggest that efforts towards quality improvement in cardiothoracic surgery must address socioeconomic distress if their full benefits are to be realized by all patients.
While experience to-date with interventions focused on improving outcomes among surgical of low SES are limited, successful examples are found in other fields of medicine. The Alliance to Reduce Disparities in Diabetes was a five-year initiative across five centers in the United States to improve patient outcomes among populations which are both underserved and have significantly higher rates of diabetes than the general population.22,23 Participants in the program, who were members of historically underserved communities (i.e., African American, Hispanic, Female), received enhanced patient education and care coordination. Physicians and clinic staff at the participating centers also underwent cultural competency training and behavior change education. Ultimately, participants in the intervention arm experienced significantly greater decreases in Hemoglobin A1c and blood pressure than participants in the control arm. The inclusion of additional patient- and physician-facing resources to enhance the care of patients of lower SES communities is likely a critical component of any successful intervention, given the potential for implicit bias.
We acknowledge the limitations of the present study. First, and as with all retrospective analyses, our results are vulnerable to unmeasured confounding. These findings reflect rigorous risk adjustment. Additionally, our results may not be generalizable outside of our regional collaborative. However, the VCSQI encompasses a demographically and socioeconomically diverse population. Third, communities’ DCI may have changed since this index was calculated. Nevertheless, our findings provide important insight into the relationship between the socioeconomic fitness of a patient’s community and their outcomes following cardiac surgery. Fourth, DCI utilizes five-digit zip code. The discriminatory power of the DCI and other indices like it would be strengthened by the inclusion of calculations based on extended 9-digit zip code; however, extended zip code is not reliably present in the STS ACSD. Fourth, FTR is somewhat narrowly defined by the STS and does not include all complications following cardiac surgery from which a patient may be ‘rescued’ from mortality (ex., cardiac arrest, pulmonary embolism, sepsis). However, we employed the STS definition of FTR in this analysis as it is this definition which will be included in the STS ACSD and used for calculation of a patient’s predicted risk of FTR. As such, it is likely that STS-defined FTR will be used extensively for quality improvement and research in cardiac surgery. Our intent is for these results to encourage investigators and trialists to include a measure of SES, such as the DCI or area deprivation index, in future work using STS-defined FTR so as to avoid unmeasured confounding by SES.
CONCLUSION
Low socioeconomic status is associated with poor health outcomes. In this regional analysis of patients undergoing STS index procedures, increasing socioeconomic distress (low SES) is significantly associated with higher FTR. This finding suggests that socioeconomic determinants of health should be accounted for in risk prediction model, and in efforts to mitigate failure to rescue, which may allow for improved outcomes.
Supplementary Material
Central Picture Legend.
Failure to Rescue Associated with Increasing Socioeconomic Distress
Central Message.
Low socioeconomic status, assessed by the Distressed Communities Index, is associated with higher failure to rescue for patients undergoing STS index procedures.
Perspective Statement.
Patients undergoing STS index procedures have higher rates of failure to rescue, defined as death after a major complication, if they live in a low socioeconomic zip code region, as assessed by the Distressed Communities Index. Socioeconomic determinants of health should be accounted for in risk prediction model, and in efforts to mitigate failure to rescue, which may allow for improved outcomes.
ACKNOWLEDGEMENTS
Research reported in this publication/presentation/work was supported in part by the National Heart, Lung, and Blood Institute (grant T32 HL007849-21A1), as well as by a grant under Award Number 2UM HL088925.The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Glossary of Abbreviations
- FTR
Failure to Rescue
- DCI
Distressed Communities Index
- STS
Society of Thoracic Surgeons
- SES
Socioeconomic Status
- EIG
Economic Innovation Group
- CABG
Coronary artery bypass grafting
- AVR
Aortic Valve Replacement
- MVR
Mitral Valve Replacement
- MVr
Mitral Valve Repair
- PCI
Percutaneous coronary intervention
- CHF
Congestive heart failure
- VCSQI
Virginia Cardiac Services Quality Initiative
Biographies
Footnotes
Conflict of Interest Statement and Separate Funding Statement: The authors had full control of the design of the study, methods used, results, data analysis and production of the written manuscript.
Meeting: Oral Presentation, AATS Annual Meeting, May 15th, 2022, Boston, MA
Institutional Review Board (IRB) Approval: This study was exempt from review by the University of Virginia’s Institutional Review Board due to the de-identified nature of the quality database (Protocol #23305).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- 1.Munro D. The 50th anniversary of Dr. King’s healthcare quote. Jersey City, NJ: Forbes, 2016. Mar. 25. [Google Scholar]
- 2.Reames BN, Birkmeyer NJ, Dimick JB, Ghaferi AA. Socioeconomic disparities in mortality after cancer surgery: failure to rescue. JAMA Surg. 2014. May;149(5):475–481. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Metcalfe D, Castillo-Angeles M, Olufajo OA, et al. Failure to rescue and disparities in emergency general surgery. J Surg Res. 2018. Nov;231:62–68. [DOI] [PubMed] [Google Scholar]
- 4.Charles EJ, Mehaffey JH, Hawkins RB, et al. Socioeconomic Distressed Communities Index Predicts Risk Adjusted Mortality After Cardiac Surgery. Ann Thorac Surg. 2019. Jun;107(6):1706–1712. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hoyler MM, Feng TR, Ma X, et al. Insurance Status and Socioeconomic Factors Affect [DOI] [PubMed] [Google Scholar]
- 6.Early Mortality After Cardiac Valve Surgery. J Cardiothorac Vasc Anesth. 2020. Dec;34(12):3234–3242. [DOI] [PubMed] [Google Scholar]
- 7.Anderson BR, Fieldston ES, Newburger JW, et al. Disparities in Outcomes and Resource Use After Hospitalization for Cardiac Surgery by Neighborhood Income. Pediatrics. 2018. Mar;141(3). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mehaffey JH, Hawkins RB, Charles EJ, et al. Distressed communities are associated with worse outcomes after coronary artery bypass surgery. J Thorac Cardiovasc Surg. 2020. Aug;160(2):425–432. [DOI] [PubMed] [Google Scholar]
- 9.Bilfinger T, Nemesure A, Pyo R, et al. Distressed Communities Index in Patients Undergoing Transcatheter Aortic Valve Implantation in an Affluent County in New York. J Interv Cardiol. 2021 2021. Aug. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kesler P. About Us. Economic Innovation Group; 2022. https://eig.org/about-us/ (accessed June 8, 2022). [Google Scholar]
- 11.Introduction to the Distressed Communities Index [Internet]. Washington, D.C.: Economic Innovation Group. Available from: https://eig.org/dci. [Google Scholar]
- 12.Osnabrugge RLJ, Speir AM, Head SJ, Fonner CE, Fonner E Jr, Ailawadi G, et al. Costs for surgical aortic valve replacement according to preoperative risk categories. Ann Thorac Surg 2013;96:500–6. [DOI] [PubMed] [Google Scholar]
- 13.Hawkins RB, Downs EA, Johnston LE, Mehaffey JH, Fonner CE, Ghanta RK, et al. Impact of Transcatheter Technology on Surgical Aortic Valve Replacement Volume, Outcomes, and Cost. Ann Thorac Surg 2017;103:1815–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kurlansky PA, O’Brien SM, Vassileva CM, Lobdell KW, Edwards FH, Jacobs JP, et al. Failure to Rescue: A New Society of Thoracic Surgeons Quality Metric for Cardiac Surgery. Ann Thorac Surg 2021. 10.1016/j.athoracsur.2021.06.025. [DOI] [PubMed] [Google Scholar]
- 15.https://www.sts.org/registries-research-center/sts-national-database/adult-cardiac-surgery-database/data-collection. Accessed 03/01/2022.
- 16.https://psnet.ahrq.gov/primer/failure-rescue. Accessed 03/01/2022.
- 17.Reddy HG, Shih T, Englesbe MJ, Shannon FL, Theurer PF, Herbert MA, et al. Analyzing “failure to rescue”: is this an opportunity for outcome improvement in cardiac surgery? Ann Thorac Surg 2013;95:1976–81; discussion 1981. 10.1016/j.athoracsur.2013.03.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Edwards FH, Ferraris VA, Kurlansky PA, Lobdell KW, He X, O’Brien SM, et al. Failure to Rescue Rates After Coronary Artery Bypass Grafting: An Analysis From The Society of Thoracic Surgeons Adult Cardiac Surgery Database. Ann Thorac Surg 2016;102:458–64. 10.1016/j.athoracsur.2016.04.051 [DOI] [PubMed] [Google Scholar]
- 19.Likosky DS, Strobel RJ, Wu X, Kramer RS, Hamman BL, Brevig JK, et al. Interhospital failure to rescue after coronary artery bypass grafting. J Thorac Cardiovasc Surg 2021. 10.1016/j.jtcvs.2021.01.064. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Shahian DM, Badhwar V, O’Brien SM, et al. Social Risk Factors in Society of Thoracic Surgeons Risk Models. Part 1: Concepts, Indicator Variables, and Controversies. Ann Thorac Surg 2022;113:1703–17. 10.1016/j.athoracsur.2021.11.067. [DOI] [PubMed] [Google Scholar]
- 21.Shahian DM, Badhwar V, O’Brien SM, et al. Social Risk Factors in Society of Thoracic Surgeons Risk Models. Part 2: Empirical Studies in Cardiac Surgery; Risk Model Recommendations. Ann Thorac Surg 2022;113:1718–29. 10.1016/j.athoracsur.2021.11.069. [DOI] [PubMed] [Google Scholar]
- 22.Lewis MA, Bann CM, Karns SA, Hobbs CL, Holt S, Brenner J, et al. Cross-site evaluation of the Alliance to Reduce Disparities in Diabetes: clinical and patient-reported outcomes. Health Promot Pract 2014;15:92S – 102S. 10.1177/1524839914545168. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Lewis MA, Williams PA, Fitzgerald TM, Heminger CL, Hobbs CL, Moultrie RR, et al. Improving the implementation of diabetes self-management: findings from the Alliance to Reduce Disparities in Diabetes. Health Promot Pract 2014;15:83S – 91S. 10.1177/1524839914541277. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.