Abstract
Background:
Hepatocellular carcinoma (HCC) is a leading cause of cancer mortality worldwide. Liver resections and transplantations have increasingly become feasible options for potential cure. These complex surgeries are inherently associated with increased rates of readmission. In the meanwhile, hospital readmission rates are rapidly becoming an important quality of care metric. Therefore, it is very important to understand the effect of 30-day readmission on mortality and the factors associated with increased 30- and 90-day mortality rates.
Methods:
This is a retrospective cohort study utilizing data from the National Cancer Database. Patients included were 18 years or older who underwent liver resection or liver transplantation for HCC between 2003 and 2011. Our primary outcomes of interest were 30- and 90-day mortality rates. Our primary independent variable of interest was 30-day readmission.
Results:
16 658 patients underwent either a liver resection or transplantation for HCC between 2003 and 2011. For patients with liver transplantations, increased readmission rates were associated with lower risks of 30-day mortality (P = .012) but a trend toward higher 90-day mortality (P = .057). Patients who underwent liver resection for HCC also demonstrated increased readmission rates to be associated with lower risk of 30-day mortality (P = .014) but higher 90-day mortality (P ≤ .001).
Conclusion:
This is the only study to utilize a national database to investigate the association between readmission rates and mortality rates of both liver transplantations and resections for patients with HCC. We demonstrate 30-day readmission to show no increase in 30-day mortality, but rather higher 90-day mortality.
Keywords: live resection, liver transplant, hepatocellular carcinoma
Introduction
Hepatocellular carcinoma (HCC) continues to be a leading cause of cancer mortality, representing the second most common cause of cancer-related deaths worldwide.1 Surgical interventions, including liver transplantation and liver resection, are the only treatment options that offer patients a chance of cure.2,3 Liver transplantation is preferred for patients with HCC due to removal of the tumor and diseased liver. However, in light of the scarcity of donor organs, liver resection remains an important and potentially curative measure, with 5-year survival rates exceeding 70%.4,5 As such, liver resection for tumor clearance is now emerging as the first-line curative treatment for patients with preserved liver function.5 Due to careful patient selection, improved understanding of liver anatomy, surgical techniques, as well as advances in perioperative care, both surgical approaches can now be performed safely. Nevertheless, because both liver transplantation and liver resection are complex procedures, patients remain at risk of having subsequent, unplanned hospital readmissions.
Hospital readmission rates are rapidly becoming important, despite being a controversial quality of care metric. As complex surgeries like liver transplantation and hepatectomy are inherently associated with increased rates of readmission, it is important to know if these readmissions subsequently alter the mortality rates of these patients. The majority of previous studies have described readmission in single institutions, but national readmission data are lacking.6,7 Thus, in this study, we aimed to identify whether higher readmission rates are associated with increased mortality as well the factors associated with increased 30- and 90-day mortality rates after both liver transplantation and liver resection at the national level in patients with HCC.
Methods
Design and Data Sources
We conducted a retrospective cohort study using data from the National Cancer Database (NCDB). National Cancer Database was established by the American College of Surgeons and Commission on Cancer in 1989 and includes data from all Commission on Cancer-accredited hospitals in the United States and Puerto Rico. It is estimated to include approximately 70% of new cancer diagnoses and is comprised of more than 30 million records from 1500 hospitals. The database also includes census tract-level data from the US Census Bureau’s American Community Survey, which provides estimates of patient income, educational attainment, and urban/rural status.
Participants
We included all patients aged 18 years or older undergoing liver resection or liver transplantation for HCC between 2003 and 2011. We excluded any patients’ missing data for readmission, mortality, or key demographic variables.
Variables
Our primary outcomes of interest were 30- and 90-day mortality. Our primary independent variable of interest was 30-day readmission, defined as the rate of unplanned readmission to the reporting hospital within 30 days of hospital discharge after liver cancer-directed surgery.
We also included demographic, geographic, clinical, and hospital characteristics thought to impact readmission or mortality. Demographic data including age, sex, race/ethnicity, and type of insurance were collected at the patient level, while proxy measures of socioeconomic status were derived from the 2012 American Community Survey for each patient’s home ZIP code. These included ZIP code-level measures of median household income and educational attainment, measured as the proportion of patients in the ZIP code with less than a high school diploma. Patient’s urban/rural location was determined at the ZIP code-level from the 2012 American Community Survey and travel distance was measured as the Haversine distance in miles between the center of the patient’s ZIP code and the address of the hospital where they underwent surgery. Our analysis included cancer-specific clinical variables in addition to information about the patient’s surgical admission. Cancer stage was determined using American Joint Committee on Cancer (AJCC) clinical staging along with confirmation with biopsy samples or surgical pathology. Data from the surgical admission included postoperative inpatient length of stay and the Charlson-Deyo score, which is a commonly used measure of clinical comorbidities.8 We used the year of surgery to assess temporal trends in readmission.
Patient-Level Analyses
To identify factors associated with hospital readmission, we used the Wilcoxon rank sum test and Chi-square or Fisher’s exact test to compare baseline characteristics for each outcome of interest. We used univariable logistic regression to calculate unadjusted odds ratios and 95% confidence intervals and included variables reaching a significance level of P < .20 in a multivariable logistic regression model. We assessed collinearity using variance inflation factors and evaluated goodness of fit using the Hosmer-Lemeshow test. We used similar analytic techniques to determine the effects of readmission on 30- and 90-day mortality. Two-tailed P-values below .05 were considered statistically significant. All analyses were conducted in Stata 13 (StataCorp, College Station, Texas).
Hospital-Level Analysis
To determine whether hospital-level readmission rates are associated with health outcomes, we aggregated patient data at the hospital level and calculated patient volume as well as mean age, Charlson-Deyo score, readmission rate, and 30- and 90-day mortality for each facility from 2003 to 2011. Hospitals with fewer than 10 procedures during the study period were excluded. We used data for the last 3 years of the study period to reduce the impact of temporal trends in readmission and mortality. After grouping the remaining hospitals into quartiles based on readmission rate, we then used Wilcoxon rank sum tests to determine whether hospital-level readmission rates were associated with hospital-level patient outcomes.
Results
We identified 16 658 patients who underwent surgical intervention for HCC between 2003 and 2011. Of these, 8023 patients underwent liver transplantation and 8635 patients underwent liver resection. For patients undergoing liver transplantation, 78.1% were men and the median age was 57 years (Table 1). There was no association between readmission rates and age, gender, race, type of insurance, education, income, urban/rural location of patient, or miles to hospital. Increased readmissions were associated with higher Charlson-Deyo scores, as well as longer hospital stays during initial hospitalization (Table 1). The major differences included an association between younger age, diagnosis at a higher cancer stage, and having nodes resected with increased readmission rates (Table 2). The extent of liver resection was not associated with increased readmission rates.
Table 1.
Characteristics of Patients Undergoing Liver Transplantation for Hepatocellular Carcinoma.
| All patients |
Not readmitted |
Readmitted |
||
|---|---|---|---|---|
| Characteristic | N = 8023 | N = 7254 | N = 769 | P-value |
| Age at diagnosis, median (IQR) | 57.0 (52.0, 62.0) | 57.0 (52.0, 62.0) | 56.0 (51.0, 61.0) | .087 |
| Sex, N(%) | .300 | |||
| Male | 6264 (78.1%) | 5675 (78.2%) | 589 (76.6%) | |
| Female | 1759 (21.9%) | 1579 (21.8%) | 180 (23.4%) | |
| Race, N(%) | .680 | |||
| White | 5623 (70.1%) | 5081 (70.0%) | 542 (70.5%) | |
| African American | 732 (9.1%) | 657 (9.1%) | 75 (9.8%) | |
| Asian/PI | 483 (6.0%) | 437 (6.0%) | 46 (6.0%) | |
| Hispanic | 926 (11.5%) | 848 (11.7%) | 78 (10.1%) | |
| Other/unknown | 259 (3.2%) | 231 (3.2%) | 28 (3.6%) | |
| Insurance type, N(%) | .890 | |||
| Private | 4469 (55.7%) | 4047 (55.8%) | 422 (54.9%) | |
| Medicaid | 884 (11.0%) | 792 (10.9%) | 92 (12.0%) | |
| Medicare | 2271 (28.3%) | 2057 (28.4%) | 214 (27.8%) | |
| Not insured | 219 (2.7%) | 197 (2.7%) | 22 (2.9%) | |
| Other/unknown | 180 (2.2%) | 161 (2.2%) | 19 (2.5%) | |
| Income ($USD), N(%) | .680 | |||
| <$38 000 | 1435 (17.9%) | 1291 (17.8%) | 144 (18.7%) | |
| $38 000-$47 999 | 1911 (23.8%) | 1741 (24.0%) | 170 (22.1%) | |
| $48 000-$62 999 | 2213 (27.6%) | 1999 (27.6%) | 214 (27.8%) | |
| >$63k | 2464 (30.7%) | 2223 (30.6%) | 241 (31.3%) | |
| No HS diploma, N(%) | .490 | |||
| 21%+ | 1633 (20.4%) | 1493 (20.6%) | 140 (18.2%) | |
| 13-20.9% | 2146 (26.7%) | 1936 (26.7%) | 210 (27.3%) | |
| 7-12.9% | 2491 (31.0%) | 2245 (30.9%) | 246 (32.0%) | |
| <7% | 1753 (21.8%) | 1580 (21.8%) | 173 (22.5%) | |
| Patient urban/rural location, N(%) | .500 | |||
| Metro areas | 6873 (85.7%) | 6211 (85.6%) | 662 (86.1%) | |
| Metro-adjacent | 721 (9.0%) | 656 (9.0%) | 65 (8.5%) | |
| Not metro-adjacent | 315 (3.9%) | 288 (4.0%) | 27 (3.5%) | |
| Rural | 114 (1.4%) | 99 (1.4%) | 15 (2.0%) | |
| Miles to hospital, median (IQR) | 33.5 (12.0, 98.8) | 33.7 (12.0, 98.7) | 32.3 (12.1, 101.2) | .680 |
| Tumor size, N(%) | .180 | |||
| <2 cm | 2229 (27.8%) | 1992 (27.5%) | 237 (30.8%) | |
| 2-5 cm | 4869 (60.7%) | 4424 (61.0%) | 445 (57.9%) | |
| >5 cm | 677 (8.4%) | 617 (8.5%) | 60 (7.8%) | |
| Unknown | 248 (3.1%) | 221 (3.0%) | 27 (3.5%) | |
| AJCC cancer stage, N(%) | .191 | |||
| Stage I | 3629 (45.2%) | 3282 (45.2%) | 347 (45.1%) | |
| Stage II | 3061 (38.2%) | 2776 (38.3%) | 285 (37.1%) | |
| Stage III | 397 (4.9%) | 366 (5.0%) | 31 (4.0%) | |
| Stage IV | 37 (.5%) | 35 (.5%) | 2 (.3%) | |
| Unknown | 899 (11.2%) | 795 (11.0%) | 104 (13.5%) | |
| Charlson-Deyo score, N(%) | <.001 | |||
| 0 | 2772 (34.6%) | 2555 (35.2%) | 217 (28.2%) | |
| 1 | 2428 (30.3%) | 2186 (30.1%) | 242 (31.5%) | |
| 2 | 2823 (35.2%) | 2513 (34.6%) | 310 (40.3%) | |
| Hospital type, N(%) | .059 | |||
| Community cancer programa | 1173 (14.6%) | 1082 (14.9%) | 91 (11.8%) | |
| Academic/research program | 6850 (85.4%) | 6172 (85.1%) | 678 (88.2%) | |
| Facility location, N(%) | <.001 | |||
| New England | 281 (3.5%) | 256 (3.5%) | 25 (3.3%) | |
| Middle Atlantic | 1026 (12.8%) | 883 (12.2%) | 143 (18.6%) | |
| South Atlantic | 1588 (19.8%) | 1466 (20.2%) | 122 (15.9%) | |
| East North Central | 1208 (15.1%) | 1078 (14.9%) | 130 (16.9%) | |
| East South Central | 613 (7.6%) | 557 (7.7%) | 56 (7.3%) | |
| West North Central | 692 (8.6%) | 597 (8.2%) | 95 (12.4%) | |
| West South Central | 1139 (14.2%) | 1067 (14.7%) | 72 (9.4%) | |
| Mountain | 502 (6.3%) | 447 (6.2%) | 55 (7.2%) | |
| Pacific | 974 (12.1%) | 903 (12.4%) | 71 (9.2%) | |
| Nodes resected, N(%) | 2646 (33.0%) | 2376 (32.8%) | 270 (35.1%) | .190 |
| Postoperative length of stay (days), median (IQR) | 8 (6, 13) | 9 (6, 15) | <.0001 |
Abbreviations: AJCC, American Joint Committee on Cancer; HCC, hepatocellular carcinoma.
This category includes community programs and comprehensive community programs.
Table 2.
Characteristics of Patients Undergoing Liver Resection by Readmission Status.
| All patients |
Not readmitted |
readmitted |
||
|---|---|---|---|---|
| Characteristic | N = 8635 | N = 8132 | N = 503 | P-value |
| Age in years, median (IQR) | 64.0 (55.0, 73.0) | 64.0 (55.0, 73.0) | 62.0 (53.0, 72.0) | .020 |
| Sex, N (%) | ||||
| Male | 5929 (68.7) | 5569 (68.5) | 360 (71.6) | .15 |
| Female | 2706 (31.3) | 2563 (31.5) | 143 (28.4) | |
| Race, N(%) | ||||
| White | 5525 (64.0) | 5210 (64.1) | 315 (62.6) | .54 |
| African American | 1159 (13.4) | 1094 (13.5) | 65 (12.9) | |
| Asian/PI | 1099 (12.7) | 1037 (12.8) | 62 (12.3) | |
| Hispanic | 567 (6.6) | 527 (6.5) | 40 (8.0) | |
| Other/unknown | 285 (3.3) | 264 (3.2) | 21 (4.2) | |
| Insurance type, N(%) | ||||
| Private | 3469 (40.2) | 3256 (40.0) | 213 (42.3) | .093 |
| Medicaid | 746 (8.6) | 693 (8.5) | 53 (10.5) | |
| Medicare | 3807 (44.1) | 3600 (44.3) | 207 (41.2) | |
| Not insured | 368 (4.3) | 345 (4.2) | 23 (4.6) | |
| Other/unknown | 245 (2.8) | 238 (2.9) | 7 (1.4) | |
| Income ($USD), N(%) | ||||
| <$38 000 | 1710 (19.8) | 1610 (19.8) | 100 (19.9) | .92 |
| $38 000-$47 999 | 2044 (23.7) | 1931 (23.7) | 113 (22.5) | |
| $48 000-$62 999 | 2323 (26.9) | 2183 (26.8) | 140 (27.8) | |
| >$63k | 2558 (29.6) | 2408 (29.6) | 150 (29.8) | |
| No HS diploma, N(%) | ||||
| 21%+ | 1842 (21.3) | 1741 (21.4) | 101 (20.1) | .19 |
| 13-20.9% | 2330 (27.0) | 2175 (26.7) | 155 (30.8) | |
| 7-12.9% | 2606 (30.2) | 2455 (30.2) | 151 (30.0) | |
| <7% | 1857 (21.5) | 1761 (21.7) | 96 (19.1) | |
| Patient urban/rural location, N(%) | ||||
| Metro areas | 7440 (86.2) | 6999 (86.1) | 441 (87.7) | .57 |
| Metro-adjacent | 755 (8.7) | 717 (8.8) | 38 (7.6) | |
| Not Metro-adjacent | 314 (3.6) | 299 (3.7) | 15 (3.0) | |
| Rural | 126 (1.5) | 117 (1.4) | 9 (1.8) | |
| Miles to hospital, median (IQR) | 14.8 (5.8, 47.3) | 15.0 (5.8, 47.5) | 13.2 (6.2, 41.2) | .19 |
| Tumor size, N(%) | ||||
| <2 cm | 657 (7.6) | 605 (7.4) | 52 (10.3) | .11 |
| 2-5 cm | 2979 (34.5) | 2812 (34.6) | 167 (33.2) | |
| >5 cm | 4655 (53.9) | 4388 (54.0) | 267 (53.1) | |
| Unknown | 344 (4.0) | 327 (4.0) | 17 (3.4) | |
| Postoperative length of stay (days), median (IQR) | 6.0 (4.0, 9.0) | 7.0 (5.0, 11.0) | <.001 | |
| AJCC cancer stage, N(%) | ||||
| Stage I | 3553 (41.1) | 3378 (41.5) | 175 (34.8) | .025 |
| Stage II | 2058 (23.8) | 1922 (23.6) | 136 (27.0) | |
| Stage III | 1819 (21.1) | 1695 (20.8) | 124 (24.7) | |
| Stage IV | 302 (3.5) | 283 (3.5) | 19 (3.8) | |
| Unknown | 903 (10.5) | 854 (10.5) | 49 (9.7) | |
| Charlson-Deyo score, N(%) | ||||
| 0 | 4668 (54.1) | 4434 (54.5) | 234 (46.5) | .002 |
| 1 | 2522 (29.2) | 2352 (28.9) | 170 (33.8) | |
| 2 | 1445 (16.7) | 1346 (16.6) | 99 (19.7) | |
| Hospital type, N(%) | ||||
| Community | 2854 (33.1) | 2710 (32.3) | 144 (28.6) | .087 |
| Academic | 5781 (66.9) | 5422 (67.7) | 359 (71.4) | |
| Facility location, N(%) | ||||
| New England | 408 (4.7) | 384 (4.7) | 24 (4.8) | <.001 |
| Middle Atlantic | 1509 (17.5) | 1380 (17.0) | 129 (25.6) | |
| South Atlantic | 1671 (19.4) | 1569 (19.3) | 102 (20.3) | |
| East North Central | 1391 (16.1) | 1330 (16.4) | 61 (12.1) | |
| East South Central | 588 (6.8) | 555 (6.8) | 33 (6.6) | |
| West North Central | 647 (7.5) | 614 (7.6) | 33 (6.6) | |
| West South Central | 861 (10.0) | 808 (9.9) | 53 (10.5) | |
| Mountain | 330 (3.8) | 303 (3.7) | 27 (5.4) | |
| Pacific | 1230 (14.2) | 1189 (14.6) | 41 (8.2) | |
| Nodes resected, N(%) | 1585 (18.4) | 6657 (81.9) | 393 (78.1) | .036 |
| Care transitions, N(%) | ||||
| Yes | 5204 (60.3) | 4885 (60.1) | 319 (63.4) | .14 |
| No | 3431 (39.7) | 3247 (39.9) | 184 (36.6) | |
| Surgery type | ||||
| Wedge/segmental | 4630 (53.6) | 4379 (53.8) | 251 (49.9) | .36 |
| Lobectomy | 2470 (28.6) | 2318 (28.5) | 152 (30.2) | |
| Extended lobectomy | 613 (7.1) | 572 (7.0) | 41 (8.2) | |
| Unknown | 922 (10.7) | 863 (10.6) | 59 (11.7) |
Abbreviations: AJCC, American Joint Committee on Cancer.
Thirty-Day Mortality
Mortality data were available for 7171 patients who underwent liver transplantation to treat HCC. Of these, 183 (2.6%) died within 30 days of surgery. In our adjusted analysis, readmission was significantly associated with a lower risk of 30-day mortality (P = .012, Table 3). Patients with Medicaid or Medicare insurance or Stage IV cancer had a significantly increased risk of 30-day mortality. Patients with Stage II cancer and those with longer postoperative inpatient hospital stays were significantly less likely to die within 30 days of surgery (Table 3).
Table 3.
Adjusted Analyses of Factors Associated with Mortality After Liver Transplantation for Hepatocellular Carcinoma.
| 30-day mortality |
90-day mortalityb |
|||
|---|---|---|---|---|
| N = 7171 |
N = 6988 |
|||
| Characteristic | OR (95% CI) | P-valuea | OR (95% CI) | P-valuea |
| Age at diagnosis (years) | — | — | 1.041 (1.012-1.072) | .006 |
| Race | — | — | — | — |
| White | — | — | 1.00 (Reference) | |
| African American | — | — | .874 (.442-1.728) | .699 |
| Asian/PI | — | — | .940 (.444-1.991) | .871 |
| Hispanic | — | — | .395 (.169-.922) | .027 |
| Other/unknown | — | — | 2.158 (1.032-4.516) | .041 |
| Miles to hospital | .999 (.997-1.000) | .066 | — | — |
| Insurance type | <.001 | <.001 | ||
| Private | 1.00 (Reference) | 1.00 (Reference) | ||
| Medicaid | 2.272 (1.440-3.585) | <.001 | 1.350 (.736-2.477) | .332 |
| Medicare | 1.834 (1.27-2.634) | .001 | .716 (.440-.168) | .181 |
| Not insured | 1.035 (.320-3.350) | .954 | 2.397 (1.060-5.420) | .036 |
| Other/unknown | 3.666 (1.754-7.664) | .001 | 4.367 (2.115-9.016) | <.001 |
| AJCC cancer stage | .005 | — | — | |
| Stage I | 1.00 (Reference) | — | — | |
| Stage II | .658 (.456-.950) | .025 | — | — |
| Stage III | 1.455 (.792-2.676) | .227 | — | — |
| Stage IV | 4.090 (1.173-14.265) | .027 | — | — |
| Unknown | 1.023 (.618-1.695) | .928 | — | — |
| Academic hospital (vs. Communityc) | 1.703 (.991-2.928) | .054 | .506 (.320-.801) | .004 |
| Care transitions | .669 (.454-.987) | .043 | — | — |
| Length of stay (days) | .879 (.847-.912) | <.001 | 1.029 (1.023-1.035) | <.001 |
| Readmission | .347 (.152-.791) | .012 | 1.627 (.987-2.682) | .057 |
Abbreviations: AJCC, American Joint Committee on Cancer, CI, confidence interval, HCC, hepatocellular carcinoma; OR, odds ratio.
P-values calculated using multivariable logistic regression.
Analysis includes patients who were alive for at least 30 days after surgery.
Includes comprehensive community cancer programs.
As for patients who underwent liver resection for their HCC, 535 (6.2%) died within 30 days of surgery. Of the 503 patients who were readmitted, 19 (3.8%) died within 30 days of surgery. By comparison, 516 (6.3%) of the 8132 patients who were not readmitted within 30 days experienced 30-day mortality. In a model adjusted for known confounders, patients who were readmitted were 45% less likely to die within 30 days (P = .014, Table 4). Older age, advanced tumor stage, and increased Charlson-Deyo scores were significantly associated with increased risks of 30-day mortality in the adjusted model (Table 4). Having no insurance or Medicare/Medicaid insurance was also associated with increased risk of 30-day morality (Table 4). Meanwhile, higher income, treatment at an academic center, and care transitions were negatively associated with 30-day mortality (Table 4).
Table 4.
Multivariable Analysis of Factors Associated with Mortality After Liver Surgery for Hepatocellular Carcinoma.
| 30-day mortality |
90-day mortalityb |
|||
|---|---|---|---|---|
| N = 8635 |
N = 8100 |
|||
| Characteristic | Odds ratio (95% CI) | P-valuea | Odds ratio (95% CI) | P-valuea |
| Age at diagnosis (years) | 1.024 (1.014-1.034) | <.001 | 1.028 (1.017-1.039) | <.001 |
| Sex | ||||
| Male | 1.00 (Reference) | 1.00 (Reference) | ||
| Female | .824 (.673-1.007) | .059 | .720 (.569-.911) | .006 |
| Race/ethnicity | ||||
| White | — | — | 1.00 (Reference) | |
| African American | — | — | 1.321 (.981-1.778) | .066 |
| Asian/PI | — | — | .849 (.600-1.202) | .357 |
| Hispanic | — | — | 1.200 (.814-1.768) | .357 |
| Other/unknown | — | — | 1.026 (.560-1.879) | .933 |
| Miles to hospital | — | — | .999 (.998-1.000) | .095 |
| Income ($USD) | ||||
| <$38 000 | 1.00 (Reference) | — | — | |
| $38 000-$47 999 | .744 (.580-.953) | .019 | — | — |
| $48 000-$62 999 | .705 (.551-.901) | .005 | — | — |
| >$63k | .512 (.393-.667) | <.001 | — | — |
| Insurance type | ||||
| Private | 1.00 (Reference) | 1.00 (Reference) | ||
| Medicaid | 2.304 (1.678-3.163) | <.001 | 1.561 (1.079-2.259) | .018 |
| Medicare | 1.302 (1.016-1.669) | .037 | .965 (.740-1.258) | .793 |
| Not insured | 2.807 (1.912-4.121) | <.001 | 1.182 (.688-2.028) | .545 |
| Other/unknown | 1.907 (1.152-3.156) | .012 | 1.060 (.552-2.037) | .861 |
| Tumor size | ||||
| <2 cm | 1.00 (Reference) | 1.00 (Reference) | ||
| 2-5 cm | 1.262 (.821-1.939) | .289 | 1.380 (.838-2.273) | .206 |
| >5 cm | 1.247 (.813-1.912) | .312 | 1.225 (.747-2.008) | .421 |
| Unknown | 4.183 (2.536-6.900) | <.001 | 2.899 (1.575-5.338) | .001 |
| AJCC cancer stage | ||||
| Stage I | 1.00 (Reference) | 1.00 (Reference) | ||
| Stage II | 1.563 (1.223-1.996) | <.001 | 1.198 (.887-1.618) | .238 |
| Stage III | 1.910 (1.477-2.470) | <.001 | 2.827 (2.133-3.746) | <.001 |
| Stage IV | 2.101 (1.347-3.277) | .001 | 3.904 (2.504-6.088) | <.001 |
| Unknown | 1.881 (1.400-2.526) | <.001 | 1.531 (1.057-2.217) | .024 |
| Charlson-Deyo score | ||||
| 0 | 1.00 (Reference | 1.00 (Reference) | ||
| 1 | 1.237 (1.003-1.525) | .046 | 1.086 (.855-1.379) | .497 |
| 2 | 1.647 (1.304-2.079) | <.001 | 1.361 (1.038-1.783) | .026 |
| Hospital Type | ||||
| Communityc | 1.00 (Reference) | — | — | |
| Academic | .801 (.663-.966) | .020 | — | — |
| Care Transitions | ||||
| No | 1.00 (Reference) | — | — | |
| Yes | .708 (.583-.859) | <.001 | — | — |
| Surgery Type | ||||
| Wedge/segmental | — | — | 1.00 (Reference) | |
| Lobectomy | — | — | 1.438 (1.134-1.824) | .003 |
| Extended lobectomy | — | — | 1.261 (.847-1.878) | .254 |
| Unknown | — | — | 1.333 (.955-1.859) | .091 |
| Readmission | ||||
| Yes | .552 (.344-.887) | .014 | 2.186 (1.580-3.026) | <.001 |
Abbreviations: AJCC, American Joint Committee on Cancer, CI, confidence interval.
Odds ratios calculated using multivariable logistic regression.
Includes patients who were alive for at least 30 days after surgery: American Joint Committee on Cancer.
Includes comprehensive community cancer centers.
Ninety-Day Mortality
Of the 6988 patients who were alive at 30 days after liver transplantation, 127 (1.8%) died within 90 days of surgical discharge. In our adjusted analyses, there was a trend toward higher 90-day mortality among patients who were readmitted, but this was not statistically significant (P = .057, Table 3). Older age, other/unknown race, other or no insurance, and longer postoperative inpatient hospital stays were positively associated with 90-day mortality, while Hispanic race/ethnicity and treatment at an academic hospital were associated with reduced risks of 90-day mortality (Table 3). Unlike the lower 30-day mortality rates associated with increased length of stay, the 90-day mortality rates were actually higher (P = <.001, Table 3).
Of the 8100 patients who lived for 30 or more days after liver resection for HCC, 404 (5.0%) died within 90 days of surgery. Of the 484 patients who were readmitted, 48 (9%) experienced 90-day mortality, compared to 356 (4.7%) of the 7616 patients who were not readmitted. In a model adjusted for known confounders, the risk of 90-day mortality was 2.19 times as high in patients who were readmitted as in those who were not readmitted (P < .001, Table 3). In addition, factors including older age, Medicaid insurance, advanced AJCC stage, Charlson-Deyo scores of 2 or higher, and lobectomy (compared to wedge/segmental resection) were associated with significantly higher risks of 90-day mortality in the adjusted model (Table 4). Female sex was significantly associated with a lower risk of 90-day mortality.
Hospital-Level Readmission
To analyze the impact of readmission rates on overall outcomes by hospital, we included cases between 2009 and 2011 completed at hospitals that performed at least ten liver transplants or liver resections/year. There were 73 hospitals that performed more than 10 liver transplants and 88 hospitals that performed ten or more liver resections between 2009 and 2011. The median hospital transplant volume was 28 cases per year (IQR: 16.0-48.0 Table 5) and 18 cases per year (IQR: 13-30.5), respectively. There was no significant difference in case volume based on hospital readmission quartile (Table 5).
Table 5.
Characteristics of Hospitals Performing Liver Transplantation for Hepatocellular carcinoma.
| Characteristic | All hospitals N = 73 |
Readmission quartile |
P-valuea | |||
|---|---|---|---|---|---|---|
| 1 |
2 |
3 |
4 |
|||
| N = 19 | N = 19 | N = 17 | N = 18 | |||
| Patient age, mean (SD) | 57.5 (1.5) | 57.5 (1.7) | 57.3 (1.3) | 57.8 (1.3) | 57.4 (1.6) | .730 |
| Transplant volume, median (IQR) | 28.0 (16.0, 48.0) | 19.0 (14.0, 45.0) | 40.0 (28.0, 55.0) | 41.0 (21.0, 48.0) | 24.0 (13.0, 41.0) | .074 |
| Charlson-Deyo score, median (IQR) | 1.13 (.88, 1.31) | 1.2 (.8, 1.3) | 1.1 (.9, 1.3) | 1.1 (.8, 1.2) | 1.2 (1.0, 1.4) | .710 |
| Facility type, N(%) | ||||||
| Communityb | 12 (16.4) | 5 (26.3) | 1 (5.3) | 4 (23.5) | 2 (11.1) | .250 |
| Academic | 61 (83.6) | 14 (73.7) | 18 (94.7) | 13 (76.5) | 16 (88.9) | |
P-values calculated using Wilcoxon rank sum and Chi-square tests
Includes comprehensive community cancer programs.
Similarly, we identified 98 hospitals that performed more than 10 liver resections between 2009 and 2011. Of these, 78 (80%) were academic, the median procedure volume was 6 cases per year, and the average patient was 62.2 years old. In the patients who were readmitted, the majority of cases performed were wedge or segmental resections (49.9%), followed by lobectomies (30.2%) (Table 2). The median hospital-level readmission rate was 3.1% (0-8.3%). The median 30-day mortality rate was 3.7% (IQR: 0-7.7%), while the median 30-day survival rate was .8% (0-6.7%).
Discussion
Despite substantial improvements in surgical care over the past years, hepatic resection and transplantation are still associated with a substantial risk of mortality and high postoperative readmission rates. In this study, of those who underwent liver transplantations, increased readmission rates were associated with higher Charlson-Deyo scores, longer hospital stays, and facility location (Table 1). For those who underwent liver resections, the results showed that younger age, diagnosis of cancer at a higher stage, and having nodes resected were also associated with increased readmission rates (Table 2). Patients with HCC undergoing resection or transplant represent a unique population, subject to higher medical and surgical morbidities than those encountered in the general surgery population.9,10 Many of these patients have chronic liver disease, ascites, and protein-calorie malnutrition. Therefore, it was of no surprise that Charlson comorbidity scores were significantly higher in patients from both groups that had a 30-day readmission. This was in accordance with previous studies reporting that patients’ overall status is a significant predictor of readmission after liver resection or transplantation.5,6
Many have argued against the use of readmission rates as the measure of quality has not been firmly established to correlate with known quality metrics such as mortality or volume. In the adjusted analysis, both patients who were readmitted after undergoing a liver transplantation or resections had a lower risk of 30-day mortality. On the other hand, patients who underwent a liver transplantation and were readmitted had a trend toward higher 90-day mortality, despite it not being significant (P = .057, Table 3), and those who underwent a liver resection demonstrated increased 90-day mortality (P < .001). Having Medicaid, Medicare, or advanced stage cancer was significantly associated with increased risk of 30-day mortality in both groups. Meanwhile, higher 90-day mortality rates in the liver transplantation cohort were associated with being of an “other ethnicity,” not having insurance or having “other insurance,” receiving treatment at an academic hospital, and increased LOS. Patients who were readmitted after a liver resection demonstrated 90-day mortality rates to be associated with increased age, female ethnicity, Medicaid insurance, unknown tumor size, stage III or IV HCC, Charlson-Deyo score of 2, and last, undergoing a lobectomy. Higher 90-day mortality after lower 30-day mortalities associated with readmission suggests that patient salvage was attempted, but it was not successful in the long run.
Furthermore, we compared low readmission rate centers with high readmission rate facilities in regard to type, volume, and patient characteristics. Numerous previous studies have demonstrated improved surgical outcomes in higher volume centers for both procedures.11,12 We did not find a significant independent difference in center volume with 30-day readmissions after either resection or transplantation. This is consistent with prior publications, which have failed to show a beneficial impact of hospital volume on 30-day readmission rates.13-15 These results continue to build upon the argument against the use of 30-day readmission as a quality measure and subsequent metric for penalization. Therefore, caution should be exercised when utilizing 30-day readmissions as a marker of quality because the rates do not correlate with hospital-level mortality and volume and appear to be mainly driven by patients’ overall heath and comorbidities. We do believe there is still reason to further study the effect of 30-day readmission on longer term mortality as we do show higher 90-day mortality rates.
As with any retrospective cohort study, our analysis has some limitations. The use of large databases such as the NCDB limits the granularity of detail for each patient. National Cancer Database collects limited clinical data; we are not able to identify the reason for readmission or predischarge complications which are likely to be important factors in readmissions. Hospital costs, patient charges, and reimbursement figures were not reported which, therefore, precluded the performance of a direct cost analysis of the impact of readmission. Despite these limitations, the data presented are derived from a large comprehensive cancer database and highlight important aspects associated with readmission after curative surgery for HCC.
In summary, this study provides the largest and most comprehensive analysis of readmission rates after liver surgery for HCC in the United States and thus offers providers, hospitals, and policy makers with an understanding of the readmission patterns in this specific population of patients. It is prudent to study whether readmissions are associated with patient outcomes most notably, mortality. We identified several patient-level factors that are predictive of readmission such as severity of underlying disease, cancer stage, and age. However, the evidence on the relationship between hospital factors, surgical quality metrics, and 30-day readmission was less robust. This raises the question whether the assessing 30-day mortality after 30-day readmission is prudent and subsequently raises serious caution against the use of this in penalizing health care systems. On the other hand, we do believe there is to be a need to further study the effect of 30-day readmission on longer term mortality as well as the culture to shift toward the use of 90-day mortality rates to better assess outcomes. We recommend future directions to include prospective collection of data across multiple centers, with broader data collection and inclusion criteria to allow for better identification of risk factors and the development of effective interventions.
Acknowledgment
The University of Pittsburgh holds a Physician-Scientist Institutional Award from the Burroughs Wellcome Fund.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
References
- 1.Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer incidence and mortality worldwide: Sources, methods and major patterns in GLOBOCAN 2012. Int J Canc. 2015;136:E359–E386. [DOI] [PubMed] [Google Scholar]
- 2.Nathan H, Segev DL, Mayo SC, et al. National trends in surgical procedures for hepatocellular carcinoma: 1998-2008. Cancer. 2012;118:1838–1844. [DOI] [PubMed] [Google Scholar]
- 3.Vivarelli M, Montalti R, Risaliti A. Multimodal treatment of hepatocellular carcinoma on cirrhosis: An update. World J Gastroenterol. 2013;19:7316–7326. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kitisin K, Packiam V, Steel J, et al. Presentation and outcomes of hepatocellular carcinoma patients at a western centre. HPB. 2011;13:712–722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Fan ST, Mau Lo C, Poon RTP, et al. Continuous improvement of survival outcomes of resection of hepatocellular carcinoma: A 20-year experience. Ann Surg. 2011;253:745–758. [DOI] [PubMed] [Google Scholar]
- 6.Pereira AA, Bhattacharya R, Carithers R, Reyes J, Perkins J. Clinical factors predicting readmission after orthotopic liver transplantation. Liver Transpl. 2012;18:1037–1045. [DOI] [PubMed] [Google Scholar]
- 7.Fan ST, Poon RTP, Yeung C, et al. Outcome after partial hepatectomy for hepatocellular cancer within the Milan criteria. Br J Surg. 2011;98:1292–1300. [DOI] [PubMed] [Google Scholar]
- 8.Deyo R, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45(6):613–619. doi: 10.1016/0895-4356(92)90133-8 [DOI] [PubMed] [Google Scholar]
- 9.Schneider EB, Hyder O, Wolfgang CL, et al. Patient readmission and mortality after surgery for hepato-pancreatobiliary malignancies. J Am Coll Surg. 2012;215:607–615. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Barbas AS, Turley RS, Mallipeddi MK, et al. Examining reoperation and readmission after hepatic surgery. J Am Coll Surg. 2013;216:915–923. [DOI] [PubMed] [Google Scholar]
- 11.Reames BN, Ghaferi AA, Birkmeyer JD, Dimick JB. Hospital volume and operative mortality in the modern era. Ann Surg. 2014;260:244–251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Birkmeyer JD, Siewers AE, Finlayson EV, et al. Hospital volume and surgical mortality in the United States. N Engl J Med. 2002;346:1128–1137. [DOI] [PubMed] [Google Scholar]
- 13.Makela KT, Hakkinen U, Peltola M, Linna M, Kroger H, Remes V. The effect of hospital volume on length of stay, re-admissions, and complications of total hip arthroplasty. Acta Orthop. 2011;82:20–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Eskander RN, Chang J, Ziogas A, Anton-Culver H, Bristow RE. Evaluation of 30-day hospital readmission after surgery for advanced-stage ovarian cancer in a medicare population. J Clin Oncol. 2014;32:4113–4119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Goodney PP, Stukel TA, Lucas FL, Finlayson EV, Birkmeyer JD. Hospital volume, length of stay, and readmission rates in high-risk surgeryy. Ann Surg. 2003;238:161–167. [DOI] [PMC free article] [PubMed] [Google Scholar]