There are limited data to support proposed increases to the minimum institutional mitral valve (MV) surgery volume required to begin a transcatheter mitral valve repair (TMVr) program. The current study examined the association between institutional MV procedure volumes and outcomes. All 2017 Medicare fee-for-service patients who received a TMVr or MV surgery procedure were included and analyzed separately. The exposure was institutional MV surgery volume: low (1 to 24), medium (25 to 39) or high (40+). Outcomes were in-hospital mortality and 1-year postdischarge mortality and cardiovascular rehospitalization. For MV surgery patients, in-hospital mortality rates were 6.4% at low-volume, 8.7% at medium-volume and 9.8% at high-volume facilities. Rates were significantly higher for low-volume [OR = 1.50, 95% CI (1.23 to 1.84)] and medium-volume [OR = 1.33, 95% CI (1.06 to 1.67)] compared with high-volume facilities. There was no statistically significant relationship between institutional MV surgery volume and in-hospital mortality for TMVr patients, either at low-volume [OR = 1.52, 95% CI (0.56, 4.13)] or medium-volume [OR = 1.58, 95% CI (0.82, 3.02)] facilities, compared with high-volume facilities. Across all volume categories, in-hospital mortality rates for TMVr patients were relatively low (2.3% on average). For both cohorts, the rates of 1-year mortality and cardiovascular rehospitalizations were not significantly higher at low- or medium-volume MV surgery facilities, as compared with high-volume. In conclusion, among Medicare patients, there was a relation between institutional MV surgery volume and in-hospital mortality for MV surgery patients, but not for TMVr patients.
The Federal Drug Administration’s recent approval expansion of transcatheter mitral valve repair (TMVr) to patients with secondary mitral regurgitation (MR) prompted a reconsideration of the national coverage determination (NCD) for reimbursement by the US Medicare program. A subsequent multisociety consensus statement recommended an increased threshold from ≥25 to ≥40 annual mitral valve (MV) surgeries before hospitals can establish a new TMVr program. However, there are limited data to justify the increase. Although a volume-outcome relationship has been shown for invasive MV surgical procedures, it is unknown whether the same relationship exists for MV surgical volume and outcomes of TMVr, a minimally invasive procedure with a favorable safety profile compared with open surgery and generally performed by a different cardiac specialty. The objective of this study was to update the analysis done by Barker et al that examined the association between institutional MV procedure volumes and outcomes in Medicare patients. First, we confirmed the level of association between institutional MV surgery volume and in-hospital mortality for both MV surgery and TMVr patients. Second, we examined the volume-outcome relationship for longer-term outcomes of mortality and cardiovascular rehospitalization up to 1 year postdischarge.
Methods
Patients enrolled in fee-for-service Medicare who received either a TMVr or MV surgery procedure in 2017 were included in the study. Data from January 1, 2017 through December 31, 2018 were derived from the 100% Medicare Limited Dataset Standard Analytic Files. The Medicare files contain detailed claims and beneficiary enrollment information including but not limited to diagnosis and procedure codes, facility ID, patient demographics, and death information. This study was exempt from institutional review board review under 45 CFR 46.101(b) as all data were de-identified and accessed in compliance with the Health Insurance Portability and Accountability Act.
The primary outcome was in-hospital mortality. Postdischarge outcomes of mortality and cardiovascular rehospitalization up to 1 year were also examined. Cardiovascular rehospitalization was defined as ICD-10 codes I00-I99 in the primary position. The main predictor was 2017 MV surgery volume for the procedure hospital. Hospitals were categorized into 3 groups based on the current requirements and those recommended by the societies to begin a TMVr program: low-volume (1 to 24 surgeries), medium-volume (25 to 39), and high-volume (40+). Hospital characteristics, including volumes, were obtained using the 2017 Definitive Healthcare Hospital & IDN Database. Definitive Healthcare uses a proprietary algorithm derived from the Centers for Medicare & Medicaid Services (CMS) Standard Analytical Files to estimate all-payor procedure volumes for US hospitals. Although only Medicare patients were examined for outcomes, hospital volume classifications were based on all-payer data since MV surgery is also performed on commercially-insured patients.
TMVr and MV surgery patients were analyzed as separate cohorts. Baseline patient and hospital characteristics were reported and tested for statistically significant differences at p <0.05 using chi-squared, Fisher’s exact test, or t-test, as appropriate. For outcomes modeling, in-hospital mortality was analyzed using generalized estimating equations models and accounted for clustering by institution. Odds ratios (OR), 95% confidence intervals (CI) and p values were presented as measures of association. Postdischarge outcomes were depicted using Kaplan-Meier curves and analyzed using a Cox proportional hazard model with a robust sandwich estimator to adjust for institutional clustering. Cumulative martingale residuals plots and Kolmogorov-type supremum tests were used to confirm the proportional hazards assumption. Hazard ratios (HR), 95% CIs and p values were presented as measures of association. All outcome models were risk-adjusted using a propensity score based on patient [age, sex, race, region, 31 comorbidity indicators from the Elixhauser comorbidity index (ECI) ] and hospital (bed size, teaching status, region) characteristics, as well as two-way interactions among the hospital characteristics. Propensity score adjustment was used to allow for the inclusion of many potential confounders without impeding model convergence. To address the potential for underpowered analysis due to low TMVr procedure volumes and outcome events, a sensitivity analysis was performed using TMVr procedures in both 2017 and 2018 to further evaluate the relationship between MV surgical volume and TMVr in-hospital mortality. Analytic data sets were created using the Instant Health Data platform from Panalgo and all statistical analyses were conducted using SAS software version 9.4 (SAS Institute, Cary, North Carolina).
Results
In 2017, a total of 16,610 Medicare patients underwent MV surgery, with 14.3% of surgeries performed at low-volume facilities, 10.6% at medium-volume facilities, and 75.2% at high-volume facilities ( Table 1 ). The majority of MV surgery patients were male (52.8%), white (85.9%), and 70.1 years on average. A total of 3,435 Medicare patients underwent TMVr (i.e., 6.9% at low-volume, 11.0% at medium-volume, 82.1% at high-volume facilities). The majority were white (88.8%) and the average age was 79.7 years. TMVr patients had a greater number of baseline comorbidities than MV surgery patients (mean ECI score = 7.2 vs 5.4, respectively). In both cohorts, there were statistically significant baseline differences in patient age, race, region and select ECI categories ( Table 2 ) across hospital volume categories. There were also significant differences in hospital characteristics where patients at high-volume facilities were more likely to be treated at teaching hospitals with larger bed sizes in the Northeast or South Atlantic regions.
| Patient and hospital characteristics | MV surgery patients | TMVr patients | |||||
|---|---|---|---|---|---|---|---|
| MV surgery institutional volume | |||||||
| Low (n = 2,369) | Medium (n = 1,758) | High (n = 12,483) | Low (n = 237) | Medium (n = 379) | High (n = 2,819) | ||
| Age (years) a , c , d | 69.7 ± 8.9 | 70.2 ± 9.0 | 70.1 ± 8.9 | 77.6 ± 9.4 | 78.7 ± 8.1 | 80.0 ± 8.7 | |
| Women | 1123 (47.4%) | 829 (47.2%) | 5893 (47.2%) | 128 (54.0%) | 177 (46.7%) | 1352 (48.0%) | |
| Race ⁎ , † , ‡ , § | |||||||
| White | 2004 (84.6%) | 1540 (87.6%) | 10730 (86.0%) | 205 (86.5%) | 329 (86.8%) | 2515 (89.2%) | |
| Black | 202 (8.5%) | 119 (6.8%) | 956 (7.7%) | 12 (5.1%) | 22 (5.8%) | 165 (5.9%) | |
| Hispanic | 40 (1.7%) | 20 (1.1%) | 134 (1.1%) | – | 11 (2.9%) | 23 (0.8%) | |
| Asian | 41 (1.7%) | 15 (0.9%) | 184 (1.5%) | – | – | 43 (1.5%) | |
| Unknown | 82 (3.5%) | 64 (3.6%) | 479 (3.8%) | – | – | 73 (2.6%) | |
| Patient Region ⁎ , † , ‡ , § | |||||||
| Midwest | 766 (32.3%) | 379 (21.6%) | 3040 (24.4%) | 66 (27.8%) | 78 (20.6%) | 557 (19.8%) | |
| Northeast | 279 (11.8%) | 338 (19.2%) | 2771 (22.2%) | 33 (13.9%) | 56 (14.8%) | 570 (20.2%) | |
| South | 831 (35.1%) | 678 (38.6%) | 4610 (36.9%) | 95 (40.1%) | 135 (35.6%) | 1038 (36.8%) | |
| West | 493 (20.8%) | 363 (20.6%) | 2062 (16.5%) | 43 (18.1%) | 110 (29.0%) | 654 (23.2%) | |
| Elixhauser score c | 5.3 ± 3.7 | 5.4 ± 3.6 | 5.4 ± 3.7 | 7.8 ± 4.1 | 7.4 ± 3.8 | 7.2 ± 3.9 | |
| Hospital Region ⁎ , † , ‡ , § | |||||||
| Northeast | 268 (11.3%) | 337 (19.2%) | 2797 (22.4%) | 33 (13.9%) | 57 (15.0%) | 573 (20.3%) | |
| South Atlantic | 368 (15.5%) | 336 (19.1%) | 2479 (19.9%) | 49 (20.7%) | 82 (21.6%) | 626 (22.2%) | |
| East North Central | 555 (23.4%) | 245 (13.9%) | 2044 (16.4%) | 65 (27.4%) | 37 (9.8%) | 298 (10.6%) | |
| East South Central | 154 (6.5%) | 137 (7.8%) | 896 (7.2%) | – | 19 (5.0%) | 183 (6.5%) | |
| West North Central | 207 (8.7%) | 128 (7.3%) | 1224 (9.8%) | – | 36 (9.5%) | 268 (9.5%) | |
| West South Central | 330 (13.9%) | 207 (11.8%) | 1022 (8.2%) | 41 (17.3%) | 33 (8.7%) | 205 (7.3%) | |
| Mountain | 205 (8.7%) | 150 (8.5%) | 613 (4.9%) | 26 (11.0%) | 50 (13.2%) | 201 (7.1%) | |
| Pacific | 282 (11.9%) | 218 (12.4%) | 1408 (11.3%) | 16 (6.8%) | 65 (17.2%) | 465 (16.5%) | |
| Bed size ⁎ , † | 300 ± 149 | 370 ± 179 | 593 ± 312 | 323 ± 183 | 470 ± 179 | 583 ± 289 | |
| Teaching hospital ⁎ , † , ‡ | 1303 (55.0%) | 1248 (71.0%) | 10511 (84.2%) | 172 (72.6%) | 320 (84.4%) | 2451 (86.9%) | |
⁎ Significant difference between low-volume and high-volume for MV surgery patients
† Significant difference between medium-volume and high-volume for MV surgery patients
‡ Significant difference between low-volume and high-volume for TMVr patients
§ Significant difference between medium-volume and high-volume for TMVr patients
| Patient and hospital characteristics | MV surgery patients | TMVr patients | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MV surgery institutional volume | ||||||||||||
| Low (n = 2,369) | Medium (n = 1,758) | High (n = 12,483) | Low (n = 237) | Medium (n = 379) | High (n = 2,819) | |||||||
| AIDS a | – | – | 44 (0.4%) | 0 | – | – | ||||||
| Alcohol abuse b | 64 (2.7%) | 64 (3.6%) | 266 (2.1%) | – | – | 41 (1.5%) | ||||||
| Anemia | 294 (12.4%) | 205 (11.7%) | 1527 (12.2%) | 40 (16.9%) | 67 (17.7%) | 445 (15.8%) | ||||||
| Arrhythmia a | 1114 (47.0%) | 873 (49.7%) | 6502 (52.1%) | 167 (70.5%) | 269 (71.0%) | 1962 (69.6%) | ||||||
| CHF | 1098 (46.3%) | 803 (45.7) | 5739 (46.0%) | 177 (74.7%) | 282 (74.4%) | 2088 (74.1%) | ||||||
| Chronic pulm dis a | 675 (28.5%) | 462 (26.3) | 3086 (24.7%) | 93 (39.2%) | 132 (34.8%) | 949 (33.7%) | ||||||
| Coagulopathy | 191 (8.1%) | 154 (8.8) | 1047 (8.4%) | 35 (14.8%) | 48 (12.7%) | 365 (12.9%) | ||||||
| Depression | 302 (12.7%) | 227 (12.9) | 1534 (12.3%) | 40 (16.9%) | 60 (15.8%) | 367 (13.0%) | ||||||
| Diabetes (Comp) c | 385 (16.3%) | 275 (15.6) | 1865 (14.9%) | 58 (24.5%) | 82 (21.6%) | 538 (19.1%) | ||||||
| Diabetes (Uncomp) c | 574 (24.2%) | 387 (22.0) | 2655 (21.3%) | 77 (32.5%) | 107 (28.2%) | 675 (23.9%) | ||||||
| Drug abuse b | 58 (2.4%) | 51 (2.9) | 257 (2.1%) | – | – | 35 (1.2%) | ||||||
| Fluid & electrolyte | 559 (23.6%) | 366 (20.8) | 2827 (22.6%) | 80 (33.8%) | 139 (36.7%) | 970 (34.4%) | ||||||
| HTN (Comp) | 848 (35.8%) | 610 (34.7) | 4429 (35.5%) | 144 (60.8%) | 221 (58.3%) | 1666 (59.1%) | ||||||
| HTN (Uncomp) b | 1470 (62.1%) | 1136 (64.6) | 7689 (61.6%) | 174 (73.4%) | 269 (71.0%) | 1920 (68.1%) | ||||||
| Hypothyroidism c | 385 (16.3%) | 268 (15.2) | 2113 (16.9%) | 42 (17.7%) | 102 (26.9%) | 677 (24.0%) | ||||||
| Iron anemias a | 37 (1.6%) | 54 (3.1) | 306 (2.5%) | – | 17 (4.5) | 103 (3.7%) | ||||||
| Liver disease | 119 (5.0%) | 106 (6.0) | 671 (5.4%) | – | 23 (6.1) | 132 (4.7%) | ||||||
| Lymphoma | 24 (1.0%) | 25 (1.4) | 146 (1.2%) | – | – | 50 (1.8%) | ||||||
| Metastatic cancer | 15 (0.6%) | – | 98 (0.8%) | – | – | 32 (1.1%) | ||||||
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
