Abstract
Objective:
Operative volume has been used as a marker of quality. Research from previous decades has suggested minimum open abdominal aortic aneurysm (AAA) repair volume requirements for surgeons of 9 to 13 open AAA repairs annually and for hospitals of 18 open AAA repairs annually to purportedly achieve acceptable results. Given concerns regarding the decreased frequency of open repairs in the endovascular era, we examined the association of surgeon and hospital volume with the 30- and 90-day mortality in the Vascular Quality Initiative (VQI) registry.
Methods:
Patients who had undergone elective open AAA repair from 2013 to 2018 were identified in the VQI registry. We performed a cross-sectional evaluation of the association between the average hospital and surgeon volume and 30-day postoperative mortality using a hierarchical Bayesian model. Cross-level interactions were permitted, and random surgeon- and hospital-level intercepts were used to account for clustering. The mortality results were adjusted by standardizing to the observed distribution of relevant covariates in the overall cohort. The outcomes were compared to the Society for Vascular Surgery guidelines recommended criteria of <5% perioperative mortality.
Results:
A total of 3078 patients had undergone elective open AAA repair by 520 surgeons at 128 hospitals. The 30- and 90-day risks of postoperative mortality were 4.1% (n = 126) and 5.4% (n = 166), respectively. The mean surgeon volume and hospital volume both correlated inversely with the 30-day mortality. Averaged across all patients and hospitals, we found a 96% probability that surgeons who performed an average of four or more repairs per year achieved <5% 30-day mortality. Substantial interplay was present between surgeon volume and hospital volume. For example, at lower volume hospitals performing an average of five repairs annually, <5% 30-day mortality would be expected 69% of the time for surgeons performing an average of three operations annually. In contrast, at higher volume hospitals performing an average of 40 repairs annually, a <5% 30-day mortality would be expected 96% of the time for surgeons performing an average of three operations annually. As hospital volume increased, a diminishing difference occurred in 30-day mortality between lower and higher volume surgeons. Likewise, as surgeon volume increased, a diminishing difference was found in 30-day mortality between the lower and higher volume hospitals.
Conclusions:
Surgeons and hospitals in the VQI registry achieved mortality outcomes of <5% (Society for Vascular Surgery guidelines), with an average surgeon volume that was substantially lower compared with previous reports. Furthermore, when considering the development of minimal surgeon volume guidelines, it is important to contextualize the outcomes within the hospital volumes.
Keywords: Abdominal aortic aneurysm, Hospital volume, Mortality, Outcomes, Surgeon volume
The relationship between surgical outcomes and institutional, as well as provider, case volume has been explored for nearly four decades.1,2 Accordingly, an inverse association appears to exist between mortality after open abdominal aortic aneurysm (AAA) repair and hospital volume,3–9 as well as surgeon volume.10,11 Recently, three healthcare systems voluntarily pledged to prohibit the performance of selected surgical procedures, including open AAA repair, by low-volume surgeons and low-volume hospitals within their networks. Initiatives of this nature could have policy and reimbursement implications for vascular surgeons and institutions in the future, given that payers have already implemented minimum volume restrictions for certain surgical procedures in other specialties, including bariatric and transplant surgery.12–14 Previous attempts at operationalizing regionalization have raised concerns about access to care, elimination of quality-based competition due to consolidation, and challenges to training the future surgical workforce, among others.15–17 A thoughtful, empirical approach to evaluating hypothetical operative volume thresholds and eventually implementing them is critical to avoid unintended consequences.
Details of the relationship between individual surgeon volume and outcomes after open AAA repair was initially explored in the late 1990s and early 2000s.10,11 Since then, the continued expansion of endovascular aneurysm repair and secular trends in perioperative management have led to a dramatic decline in the overall number of open AAA repairs performed and an increase in patient case complexity. More recent work describing contemporary surgeons’ open AAA repair volumes and their relationship with patient outcomes has relied on analyses of national administrative databases.18–23 Although these studies have provided valuable insights and have demonstrated an association between case volume and outcomes, such databases rely on billing codes and lack clinical detail. Furthermore, the most recent data time point previously investigated was 2008; however, since then, continued declines in open repair volume have been observed.5,6,24
We sought to leverage the Vascular Quality Initiative (VQI) database to provide surgeon and hospital volume-specific estimates of mortality after contemporary open AAA repair to inform future research and implementation efforts.25
METHODS
Study design.
The present study was conducted using observational data from the Society for Vascular Surgery (SVS) Patient Safety Organization VQI prospective longitudinal registry for open AAA repair. In the overall VQI registry, 100% of operations are contributed for a given member hospital, and 40%, 29%, and 31% of the participating institutions are community, teaching, and academic hospitals, respectively.26 The 5-year period from January 1, 2013 to April 30, 2018 was chosen as the most recent timeframe during which all 18 regional quality improvement groups had contributed cases to the VQI registry. Elective nonsymptomatic, nonruptured open AAA repairs were included. Patients with American Society of Anesthesiologists (ASA) class V (ie, moribund patients not expected to survive >24 hours without the operation; n = 6) were excluded. For patients who had had more than one open AAA repair recorded in the VQI, only the first operation was included. Hospitals with >15% of patients missing data on 90-day survival status (n = 72 hospitals) were excluded, for a total of 128 hospitals included. The resulting data have a multilevel structure, with patients (level 1; n = 3078) nested within surgeons (level 2; n = 520), nested within hospitals (level 3; n = 128). The Partners Human Research Committee institutional review board approved the study protocol.
Outcome, exposures of interest, and covariates.
Recent AAA guideline updates from the SVS have recommended that elective open AAA repair should only be performed at hospitals with a documented perioperative mortality of <5%.27 Thus, the primary outcome was 30-day postoperative all-cause mortality, as assessed by linkage with the Social Security Death Index.26 In addition to the 30-day point, we examined mortality at 90 days after AAA repair because the perioperative window of elevated risk has been shown to persist for this duration.22 The primary exposures of interest were the average surgeon annual volume and hospital annual elective open AAA repair volume. Surgeon and hospital volume were averaged over the time active in the VQI to determine the average annual volumes during the study period.
Patient-level covariates that were possible confounders on the relationship between operative volume and postoperative mortality included age, sex, race (African American vs non-African American), ASA physical status classification (I-II vs III vs IV), previous congestive heart failure, chronic obstructive pulmonary disease, chronic kidney disease, preoperative smoking status (current vs former or never), body mass index (kg/m2), diabetes, maximal AAA diameter (mm), previous aortic surgery, and year of surgery. Removing all patients with one or more missing covariate value (4% of total sample) in a complete case approach yielded a final analytical sample of 2958 patients for the outcome of 30-day mortality and 2939 for the outcome of 90-day mortality.
Statistical analysis.
We conducted a cross-sectional Bayesian regression analysis to describe the association between the average surgeon volume, average hospital volume, and postoperative mortality. Clustering by surgeon and hospital was accounted for by allowing for varying intercepts at these levels. The analysis also accounted for surgeons who had performed operations at more than one of the included VQI hospitals. Mortality was estimated across the average surgeon and hospital volume values (and allowing for their cross-level interaction), standardized to the observed distribution of covariates of the study population under each joint level of surgeon and hospital volume, using a binomial model adjusted for these covariates. The regression analysis included a flexible functional form for both annual surgeon and hospital volume using restricted cubic splines with three knots and permitted interaction on mortality by their first-order terms. Model parameters and central 95% percentile credible intervals (CIs) were estimated using Markov chain Monte Carlo estimators. Improper flat priors were used for fixed parameters and weakly informative priors were used for the intercept and variances. Additionally, to assess between-surgeon and between-hospital variability, median odds ratios (MORs) were calculated for hospital-level and surgeon-level variance, which is a measure directly comparable to fixed effect ORs.28 For example, an MOR of 1 at the hospital level indicates no variation between hospitals. In contrast, a large MOR at the hospital-level indicates substantial between-hospital variability. The data were truncated at the 99th percentile of surgeon (ie, 11 repairs annually) and hospital volume (ie, 60 repairs annually). All analyses were performed in R, version 3.5.2 (R Foundation for Statistical Computing, Vienna, Austria), and included the use of the package brms for multilevel analysis.29,30
RESULTS
The overall attributes of the 3078 individuals who had undergone initial elective open AAA repair and characteristics stratified by 30-day survival status are summarized in Table I. The cohort was 5% (n = 148) African American and 26% (n = 800) female. The median age at surgery was 70 years (interquartile range [IQR], 65–75 years). A few patients were healthy or had had mild systemic disease (ASA class I or II; n = 130; 4%). The median AAA diameter was 57 mm (IQR, 52–65 mm).
Table I.
Baseline patient characteristics
| Characteristic | Overall (n = 3078) | Died within 30 days (n = 126) | Survived at 30 days (n = 2952) |
|---|---|---|---|
| Age category, years | |||
| ≤60 | 329 (11) | 3 (2) | 326 (11) |
| 60–69 | 1171 (38) | 27 (21) | 1144 (39) |
| 70–79 | 1248 (41) | 63 (50) | 1185 (40) |
| ≥80 | 330 (11) | 33 (26) | 297 (10) |
| Female (vs male) | 800 (26) | 48 (38) | 752 (25) |
| African American (vs non-African American) | 148 (5) | 9 (7) | 139 (5) |
| BMI, kg/m2 | 27.1 [24.0–30.8] | 26.1 [22.5–30.8] | 27.1 [24.0–30.8] |
| ASA class | |||
| I or II | 130 (4) | 3 (2) | 127 (4) |
| III | 1870 (61) | 72 (57) | 1798 (61) |
| IV | 1078 (35) | 51 (40) | 1027 (35) |
| Private (vs nonprivate) insurance | 1198 (39) | 44 (35) | 1154 (39) |
| Hypertension | 2584 (84) | 118 (94) | 2466 (84) |
| Diabetes | 547 (18) | 34 (27) | 513 (17) |
| CAD | 753 (24) | 42 (33) | 711 (24) |
| CHF | 248 (8) | 25 (20) | 223 (8) |
| COPD | 984 (32) | 54 (43) | 930 (32) |
| CKD | 1022 (34) | 59 (48) | 963 (33) |
| Current smoking | 1178 (38) | 50 (40) | 1128 (38) |
| Previous aortic surgery | 353 (11) | 24 (19) | 329 (11) |
| Maximum AAA diameter, mm | 57.0 [52.0–65.0] | 60.0 [54.8–70.0] | 57.0 [52.0–65.0] |
| Surgeon volume, elective AAA repairs/y | 2.5 [1.1–4.6] | 1.6 [0.9–2.6] | 2.5 [1.1–4.6] |
| Hospital volume, elective AAA repairs/y | 9.4 [4.4–20.3] | 6.5 [3.2–18.5] | 9.4 [4.6–20.3] |
AAA, Abdominal aortic aneurysm; ASA, American Society of Anesthesiologists (physical status classification); BMI, body mass index; CAD, coronary artery disease; CHF, congestive heart failure; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease. Data presented as number (%) or median [interquartile range].
The 5-year mean annual operative volume for surgeons ranged from <1 to 18 repairs (Fig) and for hospitals ranged from <1 to 62 (Fig). The patient characteristics were generally similar across the levels of surgeon volume and hospital volume (Tables II and III, respectively). Operative duration, estimated blood loss, clamp placement, and approach did not have clear trends across the surgeon volume categories. Likewise, no clear trends were found across the hospital volume categories. The overall risk of mortality by 30 and 90 days postoperatively was 4.1% (n = 126) and 5.4% (n = 166) in the overall cohort, respectively. Comparing the reported outcomes among the included and excluded hospitals, the 30-day mortality (4.1% vs 4.4%), mean ± standard deviation length of stay (8.7 ± 7.0 days vs 9.8 ± 8.7 days), and mean ± standard deviation operative time (255 ± 105 minutes vs 257 ± 107 minutes) were similar.
Fig.
Histograms of surgeons and hospitals stratified by volume.
Table II.
Baseline characteristics and outcomes stratified by annual surgeon volume
| Variable | Annual surgeon volume, cases/y | ||||
|---|---|---|---|---|---|
| <2 (n = 1342) | 2–5 (n = 1069) | 5–8 (n = 337) | ≥8 (n = 330) | ||
| Age group, years | |||||
| ≤60 | 162 (12) | 113 (11) | 23 (7) | 31 (9) | |
| 60–69 | 531 (40) | 397 (37) | 120 (36) | 123 (37) | |
| 70–79 | 520 (39) | 443 (41) | 148 (44) | 137 (42) | |
| ≥80 | 129 (10) | 116 (11) | 46 (14) | 39 (12) | |
| Female (vs male) | 350 (26) | 269 (25) | 91 (27) | 90 (27) | |
| African American (vs non-African American) | 71 (5) | 53 (5) | 13 (4) | 11 (3) | |
| BMI, kg/m2 | 26.9 [23.8–30.4] | 26.9 [23.9–30.6] | 27.4 [24.3–31.6] | 28.2 [24.2–31.6] | |
| ASA class | |||||
| I or II | 72 (5) | 41 (4) | 9 (3) | 8 (2) | |
| III | 810 (60) | 701 (66) | 187 (55) | 172 (52) | |
| IV | 460 (34) | 327 (31) | 141 (42) | 150 (45) | |
| Private (vs nonprivate) insurance | 559 (42) | 402 (38) | 130 (39) | 107 (33) | |
| Hypertension | 1134 (85) | 896 (84) | 282 (84) | 272 (82) | |
| Diabetes | 242 (18) | 189 (18) | 52 (15) | 64 (19) | |
| CAD | 337 (25) | 274 (26) | 71 (21) | 71 (22) | |
| CHF | 126 (9) | 82 (8) | 25 (7) | 15 (5) | |
| COPD | 421 (31) | 345 (32) | 107 (32) | 111 (34) | |
| CKD | 448 (34) | 346 (33) | 124 (37) | 104 (32) | |
| Current smoking | 538 (40) | 416 (39) | 97 (29) | 127 (38) | |
| Previous aortic surgery | 145 (11) | 127 (12) | 50 (15) | 31 (9) | |
| Maximum AAA diameter, mm | 57.0 [52.0–65.0] | 57.0 [52.0–65.0] | 57.0 [53.0–65.0] | 55.0 [52.0–60.0] | |
| Hospital volume of elective AAA, repairs/y | 4.6 [2.8–9.4] | 10.1 [6.7–20.3] | 21.9 [7.9–61.9] | 19.2 [18.5–61.9] | |
| Retroperitoneal exposure | 271 (20) | 224 (21) | 118 (35) | 75 (23) | |
| Proximal clamp placement | |||||
| Infrarenal | 713 (54) | 612 (58) | 171 (51) | 177 (55) | |
| Above 1 renal artery | 218 (17) | 129 (12) | 51 (15) | 52 (16) | |
| Above both renal arteries | 291 (22) | 247 (23) | 93 (28) | 77 (24) | |
| Supraceliac | 98 (7) | 69 (7) | 22 (7) | 15 (5) | |
| EBL, mL | 1200.0 [750.0–2112.8] | 1200.0 [723.8–2300.0] | 1800.0 [1000.0–3000.0] | 1325.0 [831.2–2200.0] | |
| Operative time, minutes | 238.5 [180.0–307.0] | 237.0 [182.8–301.0] | 258.0 [197.0–314.0] | 219.0 [182.0–280.0] | |
| 30-Day mortality | 83 (6) | 30 (3) | 10 (3) | 3 (1) | |
| 90-Day mortality | 105 (8) | 44 (4) | 14 (4) | 3 (1) | |
AAA, Abdominal aortic aneurysm; ASA, American Society of Anesthesiologists (physical status classification); BMI, body mass index; CAD, coronary artery disease; CHF, congestive heart failure; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; EBL, estimated blood loss. Reported as number (%) or median [interquartile range].
Table III.
Baseline patient characteristics and outcomes, stratified by annual hospital volume
| Variable | Hospital surgeon volume, cases/y | |||
|---|---|---|---|---|
| <5 (n = 918) | 5–10 (n = 801) | 10–20 (n = 526) | ≥20 (n = 833) | |
| Age group, years | ||||
| ≤60 | 113 (12) | 80 (10) | 52 (10) | 84 (10) |
| 60–69 | 348 (38) | 311 (39) | 186 (35) | 326 (39) |
| 70–79 | 360 (39) | 326 (41) | 214 (41) | 348 (42) |
| ≥80 | 97 (11) | 84 (10) | 74 (14) | 75 (9) |
| Female (vs male) | 236 (26) | 229 (29) | 138 (26) | 197 (24) |
| African American (vs non-African American) | 40 (4) | 58 (7) | 19 (4) | 31 (4) |
| BMI, kg/m2 | 26.9 [24.0–30.6] | 26.8 [23.4–30.3] | 26.7 [23.5–30.4] | 27.7 [24.7–31.6] |
| ASA class | ||||
| I or II | 48 (5) | 29 (4) | 28 (5) | 25 (3) |
| III | 572 (62) | 511 (64) | 345 (66) | 442 (53) |
| IV | 298 (32) | 261 (33) | 153 (29) | 366 (44) |
| Private (vs nonprivate) insurance | 357 (39) | 319 (40) | 252 (48) | 270 (32) |
| Hypertension | 773 (84) | 682 (85) | 446 (85) | 683 (82) |
| Diabetes | 168 (18) | 146 (18) | 92 (17) | 141 (17) |
| CAD | 223 (24) | 200 (25) | 126 (24) | 204 (24) |
| CHF | 80 (9) | 68 (8) | 41 (8) | 59 (7) |
| COPD | 276 (30) | 239 (30) | 187 (36) | 282 (34) |
| CKD | 312 (35) | 248 (31) | 190 (37) | 272 (33) |
| Current smoking | 361 (39) | 311 (39) | 228 (43) | 278 (33) |
| Previous aortic surgery | 99 (11) | 91 (11) | 60 (11) | 103 (12) |
| Maximum AAA diameter, mm | 57.0 [52.0–65.0] | 57.0 [52.0–65.0] | 56.0 [52.0–64.0] | 57.0 [53.0–64.0] |
| Surgeon volume, elective AAA repairs/y | 1.0 [0.6–1.6] | 2.3 [1.2–3.5] | 3.7 [1.9–8.8] | 4.9 [3.4–7.8] |
| Retroperitoneal exposure | 233 (26) | 162 (20) | 70 (13) | 223 (27) |
| Proximal clamp placement | ||||
| Infrarenal | 506 (56) | 453 (57) | 299 (59) | 415 (50) |
| Above 1 renal artery | 150 (17) | 104 (13) | 67 (13) | 129 (16) |
| Above both renal arteries | 170 (19) | 187 (24) | 98 (19) | 253 (30) |
| Supraceliac | 72 (8) | 50 (6) | 47 (9) | 35 (4) |
| EBL, mL | 1200.0 [750.0–2000.0] | 1200.0 [700.0–2200.0] | 1100.0 [750.0–2000.0] | 1700.0 [1000.0–3000.0] |
| Operative time, minutes | 236.0 [180.0–306.0] | 212.0 [166.0–270.5] | 211.5 [173.0–270.0] | 277.0 [225.0–340.0] |
| 30-Day mortality | 54 (6) | 33 (4) | 21 (4) | 18 (2) |
| 90-Day mortality | 70 (8) | 44 (6) | 28 (5) | 24 (3) |
AAA, Abdominal aortic aneurysm; ASA, American Society of Anesthesiologists (physical status classification); BMI, body mass index; CAD, coronary artery disease; CHF, congestive heart failure; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; EBL, estimated blood loss. Reported as number (%) or median [interquartile range].
One way to evaluate the impact on mortality of surgeon volume, relative to the hospital volume, is to compare their MORs, which is a measure of their variances. Unadjusted MORs of 2.64 (95% CI, 1.80–3.89) for surgeons and 1.32 (95% CI, 1.01–1.91) for hospitals indicated that the between-surgeon differences accounted for approximately double the variability in mortality compared with the between-hospital volume differences. After adjusting for surgeon and hospital volumes to evaluate the extent to which the operative volume contributed to the between-surgeon and between-hospital variability, the adjusted MORs were 2.24 (95% CI, 1.47–3.33) for surgeons and 1.23 (95% CI, 1.01–1.72) for hospitals.
The adjusted risk of 30- and 90-day mortality expected across surgeon volumes (adjusting for hospital volume) and the adjusted risk of 30- and 90-day mortality expected across hospital volumes (adjusting for surgeon volume) are reported in Tables IV and V. Surgeon volume and hospital volume were both inversely related to 30-day mortality. The median surgeon volume was 2.5 (IQR, 1.1–4.6) and 1.6 (IQR, 0.9–2.6) for the patients who survived and did not survive 30 days, respectively. The median hospital volume was 9.4 (IQR, 4.6–20.3) and 6.5 (IQR, 3.2–18.5) for the patients who survived and did not survive 30 days, respectively (Table I). Averaged across all hospitals, a 96% probability was found that surgeons who performed four or more repairs annually would achieve <5% 30-day mortality, in accordance with SVS recommendations. A 91% probability was found that hospitals that performed $10 repairs annually would achieve <5% 30-day adjusted mortality. Of the 520 surgeons, 489 (94%) had averaged fewer than four open elective AAA repairs annually (Fig). Surgeons who performed fewer than four repairs annually had treated 71% of patients (n = 2197; Supplementary Fig, online only). Of the 128 hospitals, 116 (91%) averaged <10 repairs annually (Fig). Hospitals that performed <10 repairs annually had treated 56% (n = 1719) of patients (Supplementary Fig, online only). A total of 2335 patients (76%) had been treated by surgeons who performed fewer than four cases annually and/or at hospitals that had performed <10 annually.
Table IV.
30-Day and 90-day mortality stratified by surgeon volumea
| No. of surgeon cases/y | Mean 30-day mortality | 95% CI | Probable mortality <5% | Mean 90-day mortality | 95% CI | Probable mortality <5% |
|---|---|---|---|---|---|---|
| 1 | 5.2 | 3.7–7.0 | 0.43 | 6.9 | 5.2–8.8 | 0.01 |
| 2 | 4.7 | 3.5–6.2 | 0.68 | 6.2 | 4.8–7.7 | 0.05 |
| 3 | 4.1 | 2.8–5.7 | 0.87 | 5.3 | 3.8–7.1 | 0.35 |
| 4 | 3.5 | 2.2–5.1 | 0.97 | 4.4 | 3.0–6.2 | 0.76 |
| 5 | 2.8 | 1.7–4.4 | 0.99 | 3.5 | 2.2–5.1 | 0.97 |
| 6 | 2.3 | 1.1–3.8 | >0.99 | 2.8 | 1.6–4.3 | >0.99 |
| 7 | 1.8 | 0.7–3.4 | >0.99 | 2.1 | 1.0–3.7 | >0.99 |
| 8 | 1.4 | 0.4–3.2 | >0.99 | 1.6 | 0.6–3.3 | >0.99 |
| 9 | 1.2 | 0.2–3.1 | >0.99 | 1.3 | 0.4–3.0 | >0.99 |
| 10 | 1.0 | 0.1–3.1 | >0.99 | 1.0 | 0.2–2.8 | >0.99 |
CI, Credible interval.
Adjusted for sex, race, age, body mass index, maximum abdominal aortic aneurysm diameter, American Society of Anesthesiologists physical status classification, previous congestive heart failure, previous smoking, chronic kidney disease, previous aortic surgery, year of surgery, and hospital volume.
Table V.
30-Day and 90-day mortality stratified by hospital volumea
| No. of hospital cases/y | Mean 30-day mortality | 95% CI | Probable mortality <5% | Mean 90-day mortality | 95% CI | Probable mortality <5% |
|---|---|---|---|---|---|---|
| 1 | 4.4 | 2.8–6.7 | 0.74 | 5.6 | 3.7–7.9 | 0.31 |
| 2 | 4.4 | 2.9–6.5 | 0.77 | 5.5 | 3.9–7.7 | 0.30 |
| 5 | 4.3 | 3.1–5.7 | 0.85 | 5.5 | 4.2–7.0 | 0.26 |
| 10 | 4.2 | 3.1–5.4 | 0.91 | 5.4 | 4.2–6.8 | 0.25 |
| 20 | 3.9 | 2.3–6.1 | 0.87 | 5.2 | 3.3–7.6 | 0.45 |
| 40 | 2.7 | 1.2–4.9 | 0.97 | 3.8 | 1.9–6.5 | 0.85 |
CI, Credible interval.
Adjusted for sex, race, age, body mass index, maximum abdominal aortic aneurysm diameter, American Society of Anesthesiologists physical status classification, previous congestive heart failure, previous smoking, chronic kidney disease, previous aortic surgery, year of surgery, and surgeon volume.
Next, we fit a multilevel regression model to examine the contextual effect (ie, interaction) between the surgeon and hospital volumes. The model of 30-day mortality is presented in Supplementary Table I (online only), and the model of 90-day mortality is presented in Supplementary Table II (online only). Substantial interplay was found between surgeon volume and hospital volume. Supplementary Table III (online only) presents the adjusted risk of 30-day mortality associated with AAA repair under different combinations of surgeon and hospital volumes, incorporating the contextual effect of hospital volume on surgeon volume (eg, a low-volume surgeon operating at a higher volume hospital might have different outcomes than the same low-volume surgeon operating at a lower volume hospital). At lower volume hospitals performing two repairs annually, <5% 30-day mortality would be expected 64% of the time for surgeons performing three operations annually. In contrast, at higher volume hospitals performing 40 repairs annually, a <5% 30-day mortality would be expected 96% of the time by surgeons performing three operations annually (Supplementary Table III, online only). As the hospital volume increased, a diminishing difference was found in the 30-day mortality between lower and higher volume surgeons (Supplementary Table III, online only). Likewise, as the surgeon volume increased, a diminishing difference was found in 30-day mortality between the lower and higher volume hospitals (Supplementary Table III, online only). Similar findings for 90-day mortality are reported in Supplementary Table IV (online only).
DISCUSSION
We explored the relationships between the average surgeon volume, hospital volume, and postoperative mortality after elective open AAA repair in the contemporary era of endovascular aneurysm repair. We found a 4.1% and 5.4% risk of mortality at 30 and 90 postoperative days, respectively. The variation in mortality between surgeons (regardless of volume) was approximately double compared with the variation in mortality between hospitals. Although hospitals and surgeons across volume categories treated patients with similar baseline patient characteristics and aneurysm size, mortality was greater among lower volume surgeons and hospitals compared with those with higher volume.
On average, higher volume surgeons achieved the SVS guideline-recommended #5% mortality, regardless of hospital volume. In contrast, lower volume surgeons only achieved such results at higher volume hospitals. Lower volume surgeons at higher volume hospitals had only slightly higher than expected mortality compared with higher volume surgeons at these hospitals. In contrast, lower volume surgeons at lower volume hospitals had substantially greater mortality rates compared with higher volume surgeons at those hospitals. The support systems available at higher volume hospitals might mediate more favorable outcomes for lower volume surgeons compared with their outcomes in lower volume hospitals. Such support systems could include experienced surgical assistance, anesthesia care, and postoperative critical care, among other factors. In contrast, higher volume surgeons achieved generally favorable results, regardless of the hospital volume, likewise indicating the importance of surgical experience.
Most previous reports of the relationship between outcomes and surgeon volume of open AAA repair were based on data from the late 1990s. Since then, the dramatic shift to endovascular repair has altered the number of open AAA repairs performed annually. Even the most recently reported studies included data w10 years olddsince then significant changes have continued to occur in the open AAA repair volume.19,24 Interest has been increasing in the development of minimum volume thresholds for aortic procedures, with high profile voluntary pledges at some institutions.13 Nonprofit healthcare quality and safety organizations, such as the Leapfrog Group, have historically advocated volume criteria.16 Volume thresholds set for complex aortic repair differ from previously reported critical thresholds for AAA repair, and the leaders at some large healthcare systems have expressed reservations with accepting this “Volume Pledge” without further evaluation of the specific threshold values laid out.4,31–33
Reservations about the use of minimum volume thresholds are not unfounded, given that existing studies have not explicitly attempted to target causal estimands. The present study is an initial effort to describe the relationships between surgeon and hospital operative volume and postoperative mortality in broad strokes. For these effects to be interpreted causally, several unreasonably strong assumptions would be required, including that the surgeon and hospital volumes would be constant over time (given that the exposure is the average surgeon and hospital volume during the study period), surgeon- and hospital-specific postoperative mortality would be constant over time (given the random effects model), that the patients had selected the surgeons and hospitals for intervention (instead of surgeons increasing or decreasing operative volume longitudinally), that no violations of positivity had occurred, and that no confounders were present in the relationship between outcome and the specific treatment surgeon (ie, which particular surgeon and hospital were selected among those with the operative volumes of interest).34 Despite this, the present analysis has provided a useful description of the relationship between surgeon volume, hospital volume, and postoperative mortality, which will help inform future research. Future research should be sure to delineate whether the target of intervention is patients who are selecting surgeons (and hospitals) or surgeons who are increasing or decreasing their operative volume longitudinally. Without doing so and satisfying the necessary underlying assumptions to obtain causal estimates, we caution against using such research to determine minimum volume thresholds.
The average hospital and surgeon mortality were lower at each average volume level in the VQI dataset compared with previous national claims and clinical databases.19,20 The reported 30-day mortality after elective open AAA repair in the VQI sample extending through 2018 was only 4.1%. This was in line with the SVS and Centers for Medicare and Medicaid Services in-hospital mortality benchmarks, lower than the 30-day mortality reported from other large national databases (5.3%), and comparable to even the highest volume quantiles in previous studies.19,22,27,35 It is unclear whether the secular trends that have occurred since these earlier reports could account for this discrepancy or whether they are a byproduct of care delivery by presumably more quality-conscious hospitals and their affiliating surgeons, given their voluntary participation in VQI.
Contextualizing the effect of a surgeon’s volume was found to be important. For example, high-volume surgeons at low-volume hospitals and low-volume surgeons at high-volume hospitals both had acceptable 30-day mortality results. The associations between surgeon volume, hospital volume, and mortality were even more pronounced when considering the 90-day outcomes. Thus, it is clear that ignoring hospital context (data presented in Tables IV and V compared with data presented in Supplementary Table III, online only) is an oversimplification. Inaccurately assuming no relationship between surgeon volume and hospital volume would likewise lead to policy interventions with unintended consequences.
We built on previous work on other national databases, which compared the importance of surgeon and hospital.20 Recent work studying the contemporary volumeeoutcome relationship for open AAA repair has often lacked an analysis of either surgeon or center volume and, therefore, could not consider their interplay.36,37 Although surgeon selection played a generally greater role than hospital selection in our analysis, both surgeon volume and hospital volume had important associations with mortality. We emphasize the critical role of context (ie, the hospital in which an operation is being performed) when considering minimum surgeon volume thresholds. Patient safety organizations have advocated for minimum annual volume thresholds for open AAA repair of 10 cases per hospital and 7 cases per surgeon.38 Recently updated SVS guidelines have recommended open AAA repair only occur at hospitals with ≥10 open aortic procedures of any type annually.27 However, only a small portion of even the highest volume surgeons and hospitals would be able to satisfy these volume requirements.
Given these findings, overly simplified thresholds might unduly restrict care without substantially improving quality or patient safety. Setting arbitrary surgeon volume criteria could prevent lower volume surgeons from performing open AAA repair despite their favorable results when operating in high-volume hospitals. Similarly, setting arbitrary hospital volume criteria could prevent higher volume surgeons from performing open AAA repair in low-volume hospitals despite their favorable outcomes.
Traditional mortality rates have been evaluated at the 30th postoperative day. Schermerhorn et al22 have demonstrated that postoperative mortality rates do not level off until 90 days after surgery. The data presented in Tables IV and V demonstrated greater mortality rates at 90 days after surgery. The association of surgeon and hospital volume with mortality observed at 30 days persisted at 90 days. Whether 90-day mortality rates should be given more emphasis is an important question but was beyond the focus of our analysis.
Our findings had important limitations. First, the VQI dataset involves voluntary center participation and captures self-reported data and, thus, could be subject to a lack of generalizability and misclassification, respectively. As stated in the Methods section, six patients were classified as ASA class V, although their repairs had been categorized as “elective,” potentially suggesting the presence of measurement error or extenuating circumstances. These patients were excluded from analysis, given this concern. However, the VQI data are audited and represent patients treated by a group of surgeons and hospitals invested in quality improvement. As such, the volume thresholds that would achieve similar mortality might differ in non-VQI hospitals and for non-VQI surgeons. Second, given the relatively short 5-year inclusion window of the present study, the operative volume was determined concurrently with the outcome as the average annual volume. As such, each surgeon and each hospital were assigned a single volume for the duration of the study period, an oversimplification that might have introduced a measurement error. As a result of this cross-sectional design, the results are intended to be descriptive. We found relatively very few higher volume surgeons at low- or medium-volume hospitals, given that hospital volume was defined by the aggregate volume of its individual surgeons, such that the mortality estimates for higher volume surgeons at lower volume hospitals might have been subjected to extrapolation. The volume estimates additionally did not include other major aortic operations and did not account for the possibility that some surgeons operate outside of the VQI network, which would have led to an underestimation of surgeon operative volume. Previous cumulative surgeon and hospital experience were also not considered, and their impact on outcomes is unknown, nor is the impact of senior surgeon guidance of lower volume, less-experienced recent graduates. The impact of the latter would be likely to erode over time as senior surgeons training and practicing in the pre-endovascular era retire. It is possible that different parameterizations of the exposure, such as cumulative case experience during a surgeon’s career, should be considered when delineating volume thresholds. Furthermore, if interest lies in the effect on postoperative mortality of changing a surgeon’s (or hospital’s) volume, time-varying confounding should be considered. Third, substantial loss to follow-up had occurred at certain hospitals, which were excluded from the analysis, as previously stated. The present analysis assumed that missingness was not systematically associated with volume or postoperative mortality. This could have affected external validity.
CONCLUSIONS
We performed a cross-sectional investigation of a contemporary, national cohort of patients who had undergone open AAA repair in the VQI. We found short-term mortality was lower compared with previous estimates. Hospital volume and surgeon volume were both associated with patient outcomes, and a contextual effect was found of hospital volume on the effect of surgeon volume. The patients had an average of <5% mortality after elective open AAA repair with hospitals and surgeons at average volume thresholds substantially lower than those currently recommended by patient safety and provider groups. However, in practice, a substantial proportion of patients will undergo open AAA repair by lower volume surgeons at lower volume centers, conditions that might not meet the SVS clinical practice guideline target mortality. Data from further evaluation of national and regional trends in open repair volume and variation should guide incorporation of minimum volume thresholds to ensure that regionalization of care and credentialing are effectively implemented in a manner supported by the evidence, ensuring vascular patients’ access to care, and maintaining an adequate training environment for the future vascular surgery workforce.
Supplementary Material
ARTICLE HIGHLIGHTS.
Type of Research: Cross-sectional analysis of prospectively collected, multicenter Vascular Quality Initiative national registry data
Key Findings: The 30- and 90-day risks of mortality after open abdominal aortic aneurysm repair were 4.1% and 5.4%, respectively. Averaged across all patients and hospitals, a 96% probability was found that surgeons who averaged four or more repairs annually achieved 30-day mortality of <5%. A substantial contextual effect by hospital volume on the relationship between surgeon volume and postoperative mortality was found.
Take Home Message: Surgeons and hospitals in the Vascular Quality Initiative registry achieved mortality outcomes <5% (Society for Vascular Surgery guidelines) with a surgeon volume that was substantiallylower compared with previous reports.
Acknowledgments
The present research was supported by the National Institute on Aging, National Institutes of Health (award F32 AG064831-01 to A.L.M.).
Footnotes
Author conflict of interest: none.
Presented at the Forty-sixth Annual Meeting of the New England Society for Vascular Surgery, Providence, RI, September 13–15, 2019; recipient of the Deterling Award.
Additional material for this article may be found online at www.jvascsurg.org.
The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a conflict of interest.
REFERENCES
- 1.Luft HS, Bunker JP, Enthoven AC. Should operations be regionalized? The empirical relation between surgical volume and mortality. N Engl J Med 1979;301:1364–9. [DOI] [PubMed] [Google Scholar]
- 2.Hughes RG, Hunt SS, Luft HS. Effects of surgeon volume and hospital volume on quality of care in hospitals. Med Care 1987;25:489–503. [DOI] [PubMed] [Google Scholar]
- 3.Birkmeyer JD, Dimick JB, Staiger DO. Operative mortality and procedure volume as predictors of subsequent hospital performance. Ann Surg 2006;243:411–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Finks JF, Osborne NH, Birkmeyer JD. Trends in hospital volume and operative mortality for high-risk surgery. N Engl J Med 2011;364: 2128–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Holt PJ, Poloniecki JD, Gerrard D, Loftus IM, Thompson MM. Meta-analysis and systematic review of the relationship between volume and outcome in abdominal aortic aneurysm surgery. Br J Surg 2007;94:395–403. [DOI] [PubMed] [Google Scholar]
- 6.Holt PJ, Poloniecki JD, Loftus IM, Michaels JA, Thompson MM. Epidemiological study of the relationship between volume and outcome after abdominal aortic aneurysm surgery in the UK from 2000 to 2005. Br J Surg 2007;94:441–8. [DOI] [PubMed] [Google Scholar]
- 7.Troëng T Volume versus outcome when treating abdominal aortic aneurysm electivelydis there evidence to centralise? Scand J Surg 2008;97:154–9. [DOI] [PubMed] [Google Scholar]
- 8.Landon BE, O’Malley AJ, Giles K, Cotterill P, Schermerhorn ML. Volume-outcome relationships and abdominal aortic aneurysm repair. Circulation 2010;122:1290–7. [DOI] [PubMed] [Google Scholar]
- 9.Waits SA, Sheetz KH, Campbell DA, Ghaferi AA, Englesbe MJ, Eliason JL, et al. Failure to rescue and mortality following repair of abdominal aortic aneurysm. J Vasc Surg 2014;59:909–14. [DOI] [PubMed] [Google Scholar]
- 10.Birkmeyer JD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. Surgeon volume and operative mortality in the United States. N Engl J Med 2003;349:2117–27. [DOI] [PubMed] [Google Scholar]
- 11.Pearce WH, Parker MA, Feinglass J, Ujiki M, Manheim LM. The importance of surgeon volume and training in outcomes for vascular surgical procedures. J Vasc Surg 1999;29:768–76. [DOI] [PubMed] [Google Scholar]
- 12.Sternberg S Hospitals move to limit low-volume surgeries. US News & World Report. May 19, 2015. Available at: http://www.usnews.com/news/articles/2015/05/19/hospitals-move-to-limit-low-volume-surgeries. Accessed January 31, 2017. [Google Scholar]
- 13.Urbach DR. Pledging to eliminate low-volume surgery. N Engl J Med 2015;373:1388–90. [DOI] [PubMed] [Google Scholar]
- 14.Jha AK. Back to the future: volume as a quality metric. JAMA 2015;314: 214–5. [DOI] [PubMed] [Google Scholar]
- 15.Chhabra KR, Dimick JB. Hospital networks and value-based payment: fertile ground for regionalizing high-risk surgery. JAMA 2015;314:1335–6. [DOI] [PubMed] [Google Scholar]
- 16.Christian CK, Gustafson ML, Betensky RA, Daley J, Zinner MJ. The Leapfrog volume criteria may fall short in identifying high-quality surgical centers. Ann Surg 2003;238:447–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Livingston EH, Burchell I. Reduced access to care resulting from centers of excellence initiatives in bariatric surgery. Arch Surg 2010;145:993–7. [DOI] [PubMed] [Google Scholar]
- 18.Dua A, Koprowski S, Upchurch G, Lee CJ, Desai SS. Progressive shortfall in open aneurysm experience for vascular surgery trainees with the impact of fenestrated and branched endovascular technology. J Vasc Surg 2017;65:257–61. [DOI] [PubMed] [Google Scholar]
- 19.Zettervall SL, Schermerhorn ML, Soden PA, McCallum JC, Shean KE, Deery SE, et al. The effect of surgeon and hospital volume on mortality after open and endovascular repair of abdominal aortic aneurysms. J Vasc Surg 2017;65:626–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.McPhee JT, Robinson WP III, Eslami MH, Arous EJ, Messina LM, Schanzer A. Surgeon case volume, not institution case volume, is the primary determinant of in-hospital mortality after elective open abdominal aortic aneurysm repair. J Vasc Surg 2011;53:591–9.e2. [DOI] [PubMed] [Google Scholar]
- 21.Marlow NE, Barraclough B, Collier NA, Dickinson IC, Fawcett J, Graham JC, et al. Effect of hospital and surgeon volume on patient outcomes following treatment of abdominal aortic aneurysms: a systematic review. Eur J Vasc Endovasc Surg 2010;40:572–9. [DOI] [PubMed] [Google Scholar]
- 22.Schermerhorn ML, Giles KA, Sachs T, Bensley RP, O’Malley J, Cotterill P, et al. Defining perioperative mortality after open and endovascular aortic aneurysm repair in the US Medicare population. J Am Coll Surg 2011;212:349–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Zettervall SL, Soden PA, Buck DB, Cronenwett JL, Goodney PP, Eslami MH, et al. ; Society for Vascular Surgery Vascular Quality Initiative. Significant regional variation exists in morbidity and mortality after repair of abdominal aortic aneurysm. J Vasc Surg 2017;65: 1305–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Suckow BD, Goodney PP, Columbo JA, Kang R, Stone DH, Sedrakyan A, et al. National trends in open surgical, endovascular, and branched-fenestrated endovascular aortic aneurysm repair in Medicare patients. J Vasc Surg 2018;67:1690–7.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Cronenwett JL, Kraiss LW, Cambria RP. The Society for Vascular Surgery Vascular Quality Initiative. J Vasc Surg 2012;55:1529–37. [DOI] [PubMed] [Google Scholar]
- 26.Desai SS, Kaji AH, Upchurch G Jr. Practical guide to surgical data sets: Society for Vascular Surgery Vascular Quality Initiative (SVS VQI). JAMA Surg 2018;153:957–8. [DOI] [PubMed] [Google Scholar]
- 27.Chaikof EL, Dalman RL, Eskandari MK, Jackson BM, Lee WA, Mansour MA, et al. The Society for Vascular Surgery practice guidelines on the care of patients with an abdominal aortic aneurysm. J Vasc Surg 2018;67:2–77.e2. [DOI] [PubMed] [Google Scholar]
- 28.Larsen K, Merlo J. Appropriate assessment of neighborhood effects on individual health: integrating random and fixed effects in multilevel logistic regression. Am J Epidemiol 2005;161:81–8. [DOI] [PubMed] [Google Scholar]
- 29.Bürkner PC. brms: an R package for Bayesian multilevel models using Stan. J Stat Softw 2017;80:1–28. [Google Scholar]
- 30.Bürkner PC. Advanced Bayesian multilevel modeling with the R package brms. The R J 2018;10:395–411. [Google Scholar]
- 31.Hannan EL, Radzyner M, Rubin D, Dougherty J, Brennan MF. The influence of hospital and surgeon volume on in-hospital mortality for colectomy, gastrectomy, and lung lobectomy in patients with cancer. Surgery 2002;131:6–15. [DOI] [PubMed] [Google Scholar]
- 32.Hu JC, Gold KF, Pashos CL, Mehta SS, Litwin MS. Role of surgeon volume in radical prostatectomy outcomes. J Clin Oncol 2003;21: 401–5. [DOI] [PubMed] [Google Scholar]
- 33.Morgan TM, Barocas DA, Keegan KA, Cookson MS, Chang SS, Ni S, et al. Volume outcomes of cystectomydis it the surgeon or the setting? J Urol 2012;188:2139–44. [DOI] [PubMed] [Google Scholar]
- 34.Wanis KN, Madenci AL, Dokus MK, et al. The meaning of confounding adjustment in the presence of multiple versions of treatment: an application to organ transplantation. Eur J Epidemiol 2019;34:225–33. [DOI] [PubMed] [Google Scholar]
- 35.Centers for Medicare & Medicaid Services. Physician Quality Reporting System: About PQRS. Available at: http://www.cms.gov/Medicare/Quality-Initiatives-Patient-Assessment-Instruments/PQRS/Payment-Adjustment-Information.html. Accessed November 4, 2018.
- 36.O’Donnell TFX, Boitano LT, Deery SE, Lancaster RT, Siracuse JJ, Schermerhorn ML, et al. Hospital volume matters: the volume-outcome relationship in open juxtarenal AAA repair. Ann Surg 2020;271:184–90. [DOI] [PubMed] [Google Scholar]
- 37.Esce A, Medhekar A, Fleming F, Glocker R, Ellis J, Raman K, et al. Increasing surgeon volume correlates with patient survival following open abdominal aortic aneurysm repair. J Vasc Surg 2019;70:762–7. [DOI] [PubMed] [Google Scholar]
- 38.Leapfrog Survey Content: Surgical Volume. Available at: http://www.leapfroggroup.org/ratings-reports/surgical-volume. Accessed November 4, 2018.
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.