Abstract
BACKGROUND
Postoperative readmission, recently identified as a marker of hospital quality in the Affordable Care Act, is associated with increased morbidity, mortality and healthcare costs, yet data on readmission following lower extremity amputation is limited. We evaluated risk factors for readmission and post-discharge adverse events following lower extremity amputation in the ACS-NSQIP.
STUDY DESIGN
All patients undergoing transmetatarsal (TMA), below-knee (BKA) or above-knee amputation (AKA) in the 2011 – 2012 NSQIP were identified. Independent pre-discharge predictors of 30-day readmission were determined using multivariable logistic regression. Readmission indication and re-interventions, available in the 2012 NSQIP only, were also evaluated.
RESULTS
We identified 5,732 patients undergoing amputation (TMA: 12%; BKA: 51%; AKA: 37%). Readmission rate was 18%. Post-discharge mortality rate was 5% (TMA: 2%; BKA: 3%; AKA: 8%; p<.001). Overall complication rate was 43% (In-hospital: 32%; Post-discharge: 11%). Reoperation was for wound related complication or additional amputation in 79% of cases. Independent predictors of readmission included chronic nursing home residence (OR: 1.3; 95% CI: 1.0–1.7), non-elective surgery (OR: 1.4; 95% CI: 1.1–1.7), prior revascularization/amputation (OR: 1.4; 95% CI: 1.1–1.7), preoperative congestive heart failure (OR: 1.7; 95% CI: 1.2–2.4), and preoperative dialysis (OR: 1.5; 95% CI: 1.2–1.9). Guillotine amputation (OR: .6; 95%CI: .4–.9) and non-home discharge (OR: .7; 95%CI: .6–1.0) were protective of readmission. Wound related complications accounted for 49% of readmissions.
CONCLUSIONS
Post discharge morbidity, mortality and readmission are common following lower extremity amputation. Closer follow up of high risk patients, optimization of medical comorbidities and aggressive management of wound infection may play a role in decreasing readmission and post discharge adverse events.
INTRODUCTION
With internal reports from the Centers for Medicare & Medicaid Services estimating potentially preventable hospital readmission costs in the billions of dollars,1 hospital readmission has become a significant area of interest for both policy makers and clinicians. As we are now in the midst of the implementation of the Affordable Care Act, CMS is rolling out the Hospital Readmission Reduction Program. This program implements a payment algorithm by which hospital reimbursement will be partially based on risk-adjusted 30-day readmission rates.2 According to these recommendations, hospitals with high risk-adjusted rehospitalization rates will receive lower average per-case payments. This has led to increased interest in studying rehospitalization rates and contributing factors as indicated by a nearly three-fold increase (509 to 1,326) in “readmission” related scholarly articles on www.pubmed.gov since 2007.
Vascular surgery, in particular, has come to the forefront of readmission investigations related to a 2009 New England Journal of Medicine study whereby 24% of Medicare beneficiaries operated for peripheral vascular disease were readmitted; third highest of any diagnosis related group.3 Accordingly, recent studies have utilized data from individual institutions as well as regional and national sources to identify the incidence of and risk factors for readmission following either open or endovascular lower extremity revascularization.4–7 Thus, while these studies and others comparing reintervention rates following lower extremity revascularization have been common,8,9 lower extremity amputation (LEA) has not been the subject of similar research.
Despite our growing armamentarium for the treatment of lower extremity arterial disease and recent improvements in limb salvage rates,10 LEA is still frequently performed with an estimated two million Americans currently living with the loss of a limb.11 Two recent studies have demonstrated the significant commitment of long-term healthcare resources to patients following LEA with a vast majority of these patients undergoing multiple rehospitalizations over a period of months to years.12,13 Yet, despite these inquiries toward the long-term resource utilization of patients following LEA, perioperative (<=30-day) readmission, as targeted by CMS, has not been studied in detail. We now aim to use the American College of Surgeons National Surgery Quality Improvement Program (NSQIP) database to study the incidence of and risk factors for readmission in patients undergoing LEA.
METHODS
Data Source
We utilized data from the 2011 and 2012 NSQIP, a national, prospectively collected clinical database including over 300 institutions. Details regarding data collection and quality control have been previously described.14,15 In 2011, the NSQIP introduced a variable for readmission within 30-days of surgery to any hospital, including non-NSQIP hospitals, as determined by medical record review and direct patient contact. For this reason analysis was restricted to 2011 and 2012 only. The accuracy of NSQIP readmission data was compared to that of physician chart review and administrative data and found to be excellent.16 While the NSQIP began capture of 30-day readmission in 2011, indication for readmission only became available in the 2012 NSQIP. Thus, readmission indication is available for 2012 only. As this study contained only de-identified data without any protected health information, the study is not considered human research and therefore not subject to institutional review board approval.
Patients
Patients undergoing major LEA [above knee amputation (AKA), below knee amputation (BKA), trans-metatarsal amputation (TMA)] were identified via query of the 2011 and 2012 NSQIP Participant User Files using the following Current Procedural Terminology codes: 27590, 27591, 27592, 27596, 27880, 27881, 27882, 28805. Patients admitted to a NSQIP participating institution for an index trauma are specifically excluded from the database in the context of that admission. Baseline patient demographics, comorbidities, operative details and postoperative course were extracted from the database. Patients not at risk for readmission within 30 days of index amputation due to death during index admission or hospital stay greater than 30 days were excluded from analysis.
Outcomes
Our primary outcome measure was overall readmission to any hospital within 30-days of index amputation. Readmissions for 2012 were further categorized as planned versus unplanned and related to the index procedure or unrelated to the index procedure. Planned readmissions were determined by whether the readmission was planned at the time of the index amputation. Readmissions were deemed related to the primary procedure if considered related by the NSQIP reviewer. Data regarding time to readmission, multiple readmissions, readmission indication and re-interventions occurring on readmission were also noted for 2012 when these parameters were introduced to the NSQIP database. Secondary outcome measures included overall 30-day morbidity and mortality stratified as either pre or post discharge.
Measures/Terms
While definitions for all NSQIP terms may be found in the NSQIP user guide, this study also utilized newly created terms defined here. Any wound complication refers to a composite variable inclusive of any NSQIP defined surgical site infection (SSI) including superficial SSI, deep SSI, organ-space SSI and dehiscence.
Statistical Analysis
All analyses were conducted using IBM SPSS Statistics version 21.0.0 for Macintosh (IBM Corp., Armonk, NY). Categorical variables were analyzed using the chi-square or Fisher’s exact test where appropriate. Continuous variables were compared using two-tailed independent samples t-test or ANOVA. Cases missing data for any given parameter were eliminated from consideration for the purposes of bivariate analysis. Multiple logistic regression was performed to determine independent predictors of readmission. All variables with a p-value less than .10 on bivariate analysis were included in the model. Models were then constructed via two methods; (A) one in which any cases missing data for a candidate predictor were excluded and (B) one in which any cases missing data had the missing parameter set to the reference group for that parameter. Method A uses only cases with complete data for candidate predictors at the cost of limited sample size. Method B maximizes sample size but produces conservative estimates. Clinical judgment was used to eliminate redundant variables such as emergent case status which was largely represented by elective case status, a more informative variable in this context. Similarly, ASA class was eliminated as it serves as a proxy for other comorbidities represented in the model. Length of stay was eliminated from consideration related to the inherent bias associated with length of stay as collected by the NSQIP. As the NSQIP follow up period is 30 days from the date of surgery, increased length of stay decreases the time at risk for readmission. Backward stepwise elimination was used to determine final independent predictors with variables eliminated for p-value greater than .05. Model discrimination was assessed using c-statistics with a c-statistic of 1.0 denoting perfect predictive power and a c-statistic of .5 denoting prediction equivalent to random chance. Hosmer-Lemeshow test was used to assess model calibration. Spearman’s rho was used to assess the correlation of predicted readmission probability between the two model methods. Throughout all analyses, statistical significance was determined by a criterion of p < .05.
RESULTS
Demographics/Clinical Details
Overall, 6,571 patients underwent LEA in the 2011–2012 NSQIP; 2,486 AKA (38%), 3,310 BKA (50%), 775 TMA (12%). Exclusions were made for patients not at risk for 30-day readmission at the time of discharge [death on index admission, N = 298 (5%) or in hospital at 30 days, N = 269 (4%)] and those for whom readmission data was unavailable, N = 272 (4%). The remaining 5,732 patients formed the basis for our study. Comorbidities are outlined in Table I, which demonstrates an increasing burden of comorbid illness with proximal amputation. Operative and post-operative course are discussed with readmission analysis. Discharge was to home in 27% of cases with the remainder discharged to skilled or unskilled facilities though discharge to home varied by amputation level (Home discharge: TMA: 54%, BKA: 23%, AKA: 23%; p<0.001).
Table 1.
Demographic Characteristics and Comorbidities of LEA Patients Undergoing Readmission Analysis in the 2011–2012 NSQIP
| ALL N (%) |
TMA N (%) |
BKA N (%) |
AKA N (%) |
p-value | |
|---|---|---|---|---|---|
|
| |||||
| N | 5,732 (100) | 700 (12) | 2,909 (51) | 2,123 (37) | - |
|
| |||||
| Age, years; mean (SD) | 66.9 (13.6) | 63.6 (13.2) | 64.5 (13.3) | 70.9 (13.2) | <.001 |
|
| |||||
| Male | 3,650 (64) | 494 (71) | 1,952 (67) | 1,204 (57) | <.001 |
|
| |||||
| Race | .001 | ||||
| - White | 3,562 (65) | 410 (62) | 1,849 (67) | 1,303 (64) | |
| - Black | 1,535 (28) | 186 (28) | 735 (27) | 614 (30) | |
| - Asian | 89 (2) | 11 (2) | 36 (1) | 42 (2) | |
| - Native American | 31 (1) | 5 (1) | 20 (1) | 6 (0) | |
| - Native Hawaiian/Pacific Islander | 33 (1) | 6 (1) | 20 (1) | 7 (0) | |
|
| |||||
| Chronic Nursing Home Residence | 1,035 (18) | 61 (9) | 376 (13) | 598 (28) | <.001 |
|
| |||||
| Dependent Functional Status | 2,296 (40) | 173 (25) | 988 (34) | 1,135 (54) | <.001 |
|
| |||||
| Diabetes Mellitus | 3,663 (64) | 526 (75) | 2,054 (71) | 1,083 (51) | <.001 |
|
| |||||
| Smoker within Last Year | 1,535 (27) | 163 (23) | 789 (27) | 583 (28) | .081 |
|
| |||||
| Chronic Obstructive Pulmonary Disease | 657 (12) | 49 (7) | 305 (11) | 303 (14) | <.001 |
|
| |||||
| Congestive Heart Failure within 30 Days | 371 (7) | 39 (6) | 194 (7) | 138 (7) | .569 |
|
| |||||
| History Myocardial Infarction Last 6 Months | 107 (4) | 12 (3) | 59 (4) | 36 (4) | .414 |
|
| |||||
| History Prior Percutaneous Coronary Intervention | 461 (17) | 68 (17) | 245 (18) | 148 (15) | .102 |
|
| |||||
| History Prior Cardiac Surgery | 599 (21) | 76 (19) | 312 (23) | 211 (21) | .229 |
|
| |||||
| History Revascularization or Amputation | 1,691 (61) | 249 (62) | 837 (61) | 605 (60) | .604 |
|
| |||||
| Rest Pain/Gangrene | 1,676 (60) | 232 (58) | 837 (61) | 607 (60) | .576 |
|
| |||||
| Preoperative Open Wound | 3,971 (69) | 512 (73) | 2,073 (71) | 1,386 (65) | <.001 |
|
| |||||
| Dialysis Dependence | 1,062 (19) | 126 (18) | 609 (21) | 327 (15) | <.001 |
|
| |||||
| Preoperative Sepsis | <.001 | ||||
| - SIRS | 713 (12) | 43 (6) | 352 (12) | 318 (15) | |
| - Sepsis | 731 (13) | 407 (14) | 407 (14) | 213 (10) | |
| - Septic Shock | 111 (2) | 62 (2) | 62 (2) | 42 (2) | |
|
| |||||
| ASA ≥ 4 | 2,314 (40) | 187 (27) | 1,117 (38) | 1,010 (48) | <.001 |
|
| |||||
| BMI ≥ 30 kg/m2 | 1,750 (32) | 259 (38) | 1,013 (36) | 478 (23) | <.001 |
All percents reflect valid denominator given missing data.
Morbidity/Mortality
Thirty-day morbidity and mortality stratified by amputation level and pre/post-discharge occurrence are displayed in Table II. For those patients that died within 30 days of surgery but after hospital discharge (N = 266), mean days from discharge until death was 10.9 (SD ± 7.0). As seen in table II, the incidence of NSQIP defined post-discharge adverse events varied by amputation level. Mortality increased with proximal amputation, as did several infectious complications including sepsis, pneumonia and urinary tract infection. The proportion of patients with multiple NSQIP defined complications also varied by amputation level, both pre and post discharge.
Table 2.
Overall, Pre and Post Discharge* Adverse Events by Amputation Level in Patients Undergoing Readmission Analysis in the 2011–2012 NSQIP
| TMA N (%) |
BKA N (%) |
AKA N (%) |
p-Value | |
|---|---|---|---|---|
|
| ||||
| Length of Stay, days; Mean (SD) | 8.2 (6.6) | 7.5 (5.0) | 6.7 (4.8) | .001 |
|
| ||||
| Readmission | 119 (17) | 527 (18) | 369 (17) | .694 |
|
| ||||
| Any Complication | 270 (39) | 1,271 (44) | 905 (43) | .049 |
| - Pre Discharge | 170 (24) | 952 (33) | 699 (33) | <.001 |
| - Post Discharge | 65 (9) | 320 (11) | 227 (11) | .419 |
|
| ||||
| Any Wound Complication | 84 (12) | 233 (8) | 142 (7) | <.001 |
| - Pre Discharge | 13 (2) | 29 (1) | 22 (2) | .137 |
| - Post Discharge | 31 (4) | 163 (6) | 87 (4) | .042 |
|
| ||||
| Post Discharge Mortality | 11 (2) | 87 (3) | 168 (8) | <.001 |
|
| ||||
| Myocardial Infarction | 5 (1) | 45 (2) | 28 (1) | .228 |
| - Pre Discharge | 4 (1) | 32 (1) | 15 (1) | .215 |
| - Post Discharge | 1 (0) | 13 (0) | 13 (1) | .280 |
|
| ||||
| New Dialysis Dependence | 10 (1) | 34 (1) | 15 (1) | .148 |
| - Pre Discharge | 9 (1) | 29 (1) | 12 (1) | .121 |
| - Post Discharge | 1 (0) | 5 (0) | 3 (0) | .959 |
|
| ||||
| Sepsis | 31 (4) | 142 (5) | 155 (7) | <.001 |
| - Pre Discharge | 20 (3) | 82 (3) | 80 (4) | .145 |
| - Post Discharge | 11 (2) | 60 (2) | 75 (4) | .001 |
|
| ||||
| Pneumonia | 13 (2) | 82 (3) | 78 (4) | .034 |
| - Pre Discharge | 8 (1) | 38 (1) | 35 (2) | .484 |
| - Post Discharge | 3 (0) | 25 (1) | 29 (1) | .055 |
|
| ||||
| Urinary Tract Infection | 8 (1) | 108 (4) | 107 (5) | <.001 |
| - Pre Discharge | 3 (0) | 38 (1) | 45 (2) | .003 |
| - Post Discharge | 2 (0) | 23 (1) | 29 (1) | .018 |
|
| ||||
| Any Return to Operating Room** | 110 (16) | 324 (11) | 118 (6) | <.001 |
| - Unplanned Pre Discharge Reoperation*** | 28 (9) | 62 (4) | 25 (2) | <.001 |
| - Unplanned Post Discharge Reoperation*** | 21 (7) | 76 (5) | 33 (3) | .010 |
| - Any Reoperation Guillotine Only | n/a | 63 (16) | 5 (5) | .003 |
| - Any Reoperation Non- Guillotine Only | n/a | 261 (10) | 113 (6) | <.001 |
|
| ||||
| Multiple Complications | 55 (8) | 304 (11) | 264 (12) | .002 |
| - Pre Discharge | 23 (3) | 147 (5) | 124 (6) | .028 |
| - Post Discharge | 8 (1) | 64 (2) | 73 (3) | .001 |
Pre and post discharge distinction not applied when exact date of event unknown
2011 and 2012 Patients
2012 Patients Only
Morbidity and mortality are presented stratified by readmission status and pre/post discharge occurrence in Table III. Temporal distribution of complications varied according to specific complication. SSI’s were over four times more likely to occur post-discharge than pre-discharge (5% vs. 1%) though all other NSQIP documented complications occurred more frequently pre-discharge. Specifically, individual post-discharge wound complications were documented as occurring at a mean of 12–15 days (SD ± 6) after discharge from the acute care setting.
Table 3.
Overall, Pre and Post Discharge* Adverse Events By Readmission Status in LEA Patients who Underwent Readmission Analysis in the 2011–2012 NSQIP
| Overall N (%) |
Readmitted N (%) |
Non- Readmitted N (%) |
p- Value | |
|---|---|---|---|---|
|
| ||||
| N | 5,732 (100) | 1,015 (18) | 4,717 (82) | - |
|
| ||||
| Length of Postoperative Stay, days; Mean (SD) | 7.3 (5.2) | 6.8 (4.4) | 7.4 (5.4) | <.001 |
|
| ||||
| Any NSQIP Complication | 2,446 (43) | 724 (71) | 1,722 (37) | <.001 |
| - Pre Discharge | 1,821 (32) | 340 (34) | 1,481 (31) | .194 |
| - Post Discharge | 612 (11) | 440 (43) | 172 (4) | <.001 |
|
| ||||
| Any Wound Complication | 459 (8) | 248 (24) | 211 (5) | <.001 |
| - Pre Discharge | 64 (1) | 11 (1) | 53 (1) | 1.000 |
| - Post Discharge | 281 (5) | 180 (18) | 101 (2) | <.001 |
|
| ||||
| Post Discharge Mortality | 266 (5) | 96 (10) | 170 (4) | <.001 |
|
| ||||
| Myocardial Infarction | 78 (1) | 33 (3) | 45 (1) | <.001 |
| - Pre Discharge | 51 (1) | 9 (1) | 42 (1) | 1.000 |
| - Post Discharge | 27 (1) | 24 (2) | 3 (0) | <.001 |
|
| ||||
| New Dialysis Dependence | 59 (1) | 24 (2) | 35 (1) | <.001 |
| - Pre Discharge | 50 (1) | 15 (2) | 35 (1) | .038 |
| - Post Discharge | 9 (0) | 9 (1) | 0 (0) | <.001 |
|
| ||||
| Sepsis | 328 (6) | 162 (16) | 166 (4) | <.001 |
| - Pre Discharge | 182 (3) | 33 (3) | 149 (3) | .844 |
| - Post Discharge | 146 (3) | 129 (13) | 17 (0) | <.001 |
|
| ||||
| Pneumonia | 173 (3) | 78 (8) | 95 (2) | <.001 |
| - Pre Discharge | 81 (1) | 18 (2) | 63 (1) | .304 |
| - Post Discharge | 57 (1) | 51 (5) | 6 (0) | <.001 |
|
| ||||
| Unplanned Reintubation | 120 (2) | 59 (6) | 61 (1) | <.001 |
| - Pre Discharge | 75 (1) | 22 (2) | 53 (1) | .014 |
| - Post Discharge | 45 (1) | 37 (4) | 8 (0) | <.001 |
|
| ||||
| Urinary Tract Infection | 223 (4) | 89 (9) | 134 (3) | <.001 |
| - Pre Discharge | 86 (2) | 20 (2) | 66 (1) | .199 |
| - Post Discharge | 54 (1) | 34 (3) | 20 (0) | <.001 |
|
| ||||
| Any Return to Operating Room** | 570 (18) | 310 (48) | 260 (10) | <.001 |
| - Unplanned Pre Discharge Reoperation*** | 115 (4) | 22 (5) | 93 (4) | .367 |
| - Unplanned Post Discharge Reoperation*** | 130 (5) | 112 (24) | 18 (1) | <.001 |
| - Unplanned Pre Discharge Reoperation (Non-Guillotine Only)*** | 67 (3) | 14 (4) | 53 (3) | .413 |
Pre and post discharge distinction not applied when exact date of event unknown
2011 and 2012 Patients
2012 Patients Only
Bivariate Analysis – Readmission
Overall readmission rate was 18%. Mean days from discharge to readmission were 11 (SD ± 7). Three percent of readmitted patients (N = 16/465; 2012 only) were readmitted a second time. Factors associated with readmission on bivariate analysis were primarily patient related (Table IV). Operative details are listed in Table V with elective surgery and guillotine amputation showing a protective effect from readmission. Further investigation showed that guillotine amputation was associated with a decreased incidence of post-discharge wound infection (2% vs. 5%; p=.002). New dialysis dependence and unplanned reintubation on index admission were the lone pre-discharge complications to predict readmission (Table III). Given the expectation that many guillotine amputations may have a planned return to the OR for wound revision, we evaluated the risk of readmission following reoperation independently for guillotine and non-guillotine amputation patients (Table III).
Table 4.
Bivariate Comparison of Preoperative Characteristics for Readmitted and Non-Readmitted LEA Patients in the 2011–2012 NSQIP
| Overall N (%) |
Readmitted N (%) |
Non- Readmitted N (%) |
p-Value | |
|---|---|---|---|---|
|
| ||||
| N | 5,732 (100) | 1,015 (18) | 4,717 (82) | - |
|
| ||||
| Age, years; mean (SD) | 67.7 (13.6) | 67.4 (13.8) | 66.7 (13.6) | .135 |
|
| ||||
| Male | 3,650 (64) | 634 (63) | 3,016 (64) | .368 |
|
| ||||
| Race | .007 | |||
| - White | 3,562 (65) | 637 (65) | 2,925 (65) | |
| - Black | 1,535 (28) | 299 (30) | 1,236 (28) | |
| - Asian | 89 (2) | 11 (1) | 78 (2) | |
| - Native American | 31 (1) | 6 (1) | 25 (1) | |
| - Native Hawaiian/Pacific Islander | 33 (1) | 5 (1) | 28 (1) | |
|
| ||||
| Chronic Nursing Home Residence | 1,035 (18) | 219 (22) | 816 (17) | .001 |
|
| ||||
| Dependent Functional Status | 2,296 (40) | 458 (46) | 1,838 (39) | <.001 |
|
| ||||
| Diabetes Mellitus | 3,663 (64) | 666 (66) | 2,997 (64) | .221 |
|
| ||||
| Smoker within Last Year | 1,535 (27) | 260 (26) | 1,275 (27) | .369 |
|
| ||||
| Chronic Obstructive Pulmonary Disease | 657 (12) | 138 (14) | 519 (11) | .022 |
|
| ||||
| Congestive Heart Failure within 30 Days | 371 (7) | 92 (9) | 279 (6) | <.001 |
|
| ||||
| History Myocardial Infarction Last 6 Months | 107 (4) | 18 (4) | 89 (4) | .898 |
|
| ||||
| History Prior Percutaneous Coronary Intervention | 461 (17) | 92 (19) | 369 (16) | .183 |
|
| ||||
| History Prior Cardiac Surgery | 599 (21) | 120 (24) | 479 (21) | .117 |
|
| ||||
| History Revascularization or Amputation | 1,691 (61) | 332 (67) | 1,359 (59) | .002 |
|
| ||||
| Rest Pain/Gangrene | 1,676 (60) | 329 (66) | 1,347 (59) | .002 |
|
| ||||
| Preoperative Open Wound | 3,971 (69) | 709 (70) | 3,262 (69) | .680 |
|
| ||||
| Dialysis Dependence | 1,062 (19) | 256 (25) | 806 (17) | <.001 |
|
| ||||
| Preoperative Sepsis | .158 | |||
| - SIRS | 713 (12) | 147 (15) | 566 (12) | |
| - Sepsis | 731 (13) | 131 (13) | 600 (13) | |
| - Septic Shock | 111 (2) | 21 (2) | 90 (2) | |
|
| ||||
| ASA ≥ 4 | 2,314 (40) | 475 (47) | 1,839 (39) | <.001 |
|
| ||||
| BMI ≥ 30 kg/m2 | 1,750 (32) | 304 (31) | 1,446 (32) | .452 |
All percents reflect valid denominator given missing data.
Table 5.
Operative Details for LEA Patients Who Underwent Readmission Analysis in the 2011–2012 NSQIP
| Overall N (%) |
Readmitted N (%) |
Not Readmitted N (%) |
p-Value | |
|---|---|---|---|---|
|
| ||||
| N | 5,732 (100) | 1,015 (18) | 4,717 (82) | - |
|
| ||||
| OR Time, minutes; Mean (SD) | 66 (40) | 66 (35) | 66 (42) | .616 |
|
| ||||
| Amputation Level | .694 | |||
| - TMA | 700 (12) | 119 (12) | 581 (12) | |
| - BKA | 2,909 (51) | 527 (52) | 2,382 (51) | |
| - AKA | 2,123 (37) | 369 (36) | 1,754 (37) | |
|
| ||||
| Guillotine Amputation | 495 (10) | 58 (7) | 437 (11) | <.001 |
|
| ||||
| Emergency Surgery | 716 (13) | 110 (11) | 606 (13) | .084 |
|
| ||||
| Elective Surgery | 2,044 (36) | 318 (32) | 1,726 (37) | .002 |
Multivariable Analysis – Readmission
Independent predictors of readmission as determined through the model techniques outlined in the methods section are shown in Figure 1. Predictors were generally patient related though also included non-elective surgery, non-home discharge and post-operative new dialysis dependence. Non-home discharge was associated with a higher baseline comorbidity burden (Home discharge ASA4: 28%, Non-home discharge ASA4: 45%; p<.001) and increased incidence of pre-discharge complications (Home discharge: 23%, Non-home discharge: 35%; p<.001). The model depicted in Figure 1A, excluding any patient with missing data, included a sample size of 2,712 of which 480 were readmitted. The model in Figure 1B included all patients. The readmission rate of 18% (N = 480/2,712) for the patients in the exclusionary model did not differ from the overall readmission rate (p = 1.000), also 18%, suggesting that these patients were a representative sample with respect to readmission. Model discrimination is as shown by the c-statistics demonstrated in the figure. The Hosmer-Lemeshow test for each model was non-significant. For those patients included in both models (N = 2,712), the spearman’s rho correlating predicted probability of readmission was .836.
Figure 1.
(A) Independent Predictors of Readmission for LEA Patients in the 2011–2012 NSQIP – Missing Variables Excluded; (B.) Independent Predictors of Readmission for LEA Patients in the 2011–2012 NSQIP – Missing Variables Set to Reference Group
Reoperation Details – 2012 Only
One hundred fifteen patients (N = 115/2,874; 4%) in the 2012 NSQIP underwent 128 pre-discharge reoperations on their index admission. Approximately 70% of these patients returned to the operating room for either additional amputation (N = 54/115; 47%) or wound related procedures (N = 28/115; 24%) including debridement, incision and drainage or revision. An additional 11% underwent either open or endovascular revascularization (N = 13/115). One hundred thirty patients in the 2012 NSQIP (N = 130/2,874; 5%) underwent 150 reoperations after discharge. One in four readmitted patients (N = 112/465) had a post-discharge return to the operating room. Similar to pre-discharge reoperation, wound related procedures (N = 50/130; 38%) and additional amputations (N = 61/130; 47%) accounted for the majority of post-discharge reoperations. Revascularization was rare in the post-discharge setting (N = 3/150; 2%).
Readmission Indication – 2012 Only
In the 2012 NSQIP, 465 patients were readmitted of which 451 readmissions (97%) were unplanned. Of the unplanned readmissions, half (N = 227/451) were considered related to the index amputation. TMA patients had the highest proportion of related readmissions (65%) as compared to BKA (49%) and AKA (49%) [p = .219]. The readmission indications given by the NSQIP for the unplanned, related readmissions (N = 227) are presented in Figure 2. The non-NSQIP defined wound related indications include wound infections not meeting NSQIP diagnostic criteria for infection as well as other non-infectious wound complications such as non-healing wound and hematoma. Including both NSQIP defined and non-NSQIP defined wound complications, these accounted for half of all readmissions (49%). Those patients readmitted with peripheral vascular disease complications include indications such as lower extremity ulcer, gangrene or cellulitis. Readmission indication varied by amputation level. The proportion of readmitted patients readmitted for non-wound related infectious indication increased with proximal amputation (TMA: 14%, BKA: 20%, AKA: 35%; p = .020). Though not reaching statistical significance, a greater proportion of patients were readmitted for wound related indication with distal amputation (TMA: 62%, BKA: 54%, AKA: 44%; p = .197). Patients readmitted for wound complications were more likely to undergo additional amputation than were patients readmitted for other indications (33% vs. 6%; p<.001). Peripheral vascular disease readmissions decreased with proximal amputation (TMA: 62%, BKA: 54%, AKA: 44%; p = .222). Readmission indication for the unplanned, unrelated readmissions are unavailable in the NSQIP database.
Figure 2.
Unplanned Related Readmission Indication Among LEA Patients in the 2012 NSQIP Cohort
DISCUSSION
This study represents the first investigation of a national, multi-center, prospective clinical database to evaluate risk factors for readmission following LEA. Readmission following LEA is extremely common with nearly one in five patients undergoing LEA in the 2011–2012 NSQIP readmitted. Independent risk factors for readmission in the NSQIP cohort were primarily patient related though non-elective surgery, non-home discharge and post-operative dialysis dependence also predicted readmission (Figure 1). Both methods of logistic regression model construction produced similar results. Readmissions were unplanned in the vast majority of cases (97%) though approximately half of unplanned readmissions were deemed unrelated to the index amputation. Of patients with unplanned, related readmissions, half were related to wound complications. Highlighting the importance of wound management in these patients, reoperations, both pre and post discharge, were related to wound issues in approximately a third of patients and included additional amputation in approximately half. Finally, readmitted patients represent an extremely high-risk population as they demonstrated a mortality rate two and a half times that of non-readmitted patients (10% vs. 4%).
Whereas prior studies on readmission following LEA have focused on long-term resource utilization, our study is the first to evaluate the incidence of perioperative (30-day) readmission and its risk factors. Feinglass and colleagues12 reviewed the cumulative risk of readmission for patients undergoing LEA in the VA NSQIP system in the early 1990s showing that over 70% of patients were readmitted over a median follow up period of 32 months. Similarly, Henry et al13 reported on long-term resource utilization following LEA at two tertiary care centers yet, this study also reported a 30-day readmission rate of 20.1% (N = 73/364). This rate of readmission is similar to the rate found in the NSQIP cohort (19%) showing that the issue of readmission after LEA is relevant to tertiary referral centers as well as others.
Given the chronic and largely non-modifiable nature of preoperative nursing home residence, CHF, chronic obstructive pulmonary disease and dialysis dependence, quality improvement measures focused on these patient-related readmission risk factors are likely limited. While we may hope to medically optimize these patients preoperatively via diuresis and/or medication management, substantive improvement in their baseline comorbidity profile is often not feasible in the preoperative period. Non-home discharge, also a predictor of readmission, was associated with a higher burden of preoperative comorbid illness and pre-discharge complications. Robust primary care management of these complex patients with early referral to vascular surgeons may limit the proportion of amputations performed under emergent circumstances, which could also play a role in mitigating readmissions by allowing for optimization.
With few patient-related risk factors for readmission conducive to preoperative modification, we look to the indication for readmission and interventions performed upon readmission for insight toward how to address this issue. Approximately half of patients were readmitted for wound related complications and one quarter of readmitted patients returned to the operating room; most for wound related procedures or additional amputations. In fact, patients readmitted for wound complications were seven times more likely to undergo additional amputation than were patients readmitted for other reasons. This, in conjunction with a trend toward increasing wound complications with distal amputation, may suggest that surgeons attempting to leave their patients with maximum limb function via a limited amputation could be susceptible to higher readmission rates. As the wound complications and the need for reoperation likely relate to a number of patient and operative factors, these present several opportunities for quality improvement. Avoidance of reoperation was the focus of a 2013 study by O’Brien and colleagues17 that evaluated risk factors for early failure of lower extremity amputation as indicated by a return to the OR on index admission. Using the 2005–2010 NSQIP, O’Brien et al showed active tobacco use (OR: 1.18, 95%CI: 1.00–1.38) and intraoperative surgical trainee participation (OR: 1.37, 95%CI: 1.20–1.57) to increase reoperation risk while locoregional anesthesia (OR: .75; 95%CI: .63–.89) and OR time greater than 40 minutes were found to be protective.
Utilizing the risk factors for reoperation elucidated by O’Brien and colleagues, tangible steps to decrease the need for reoperation, and consequently readmission, are evident. A recent meta-analysis on the benefit of preoperative smoking cessation found a 40% decrease in postoperative complications in patients who discontinued tobacco use preoperatively.18 Routine use of proactive smoking cessation strategies such as in-clinic enrollment in smoking cessation programs during preoperative consultation may improve our patients’ success in this regard.19,20 Intraoperatively, the use of locoregional anesthesia when possible avoids the physiologic stress of general anesthesia and may play a role in preventing certain systemic postoperative complications. Finally, while careful supervision of surgical trainees and a deliberate technical approach may also aid in the avoidance of reoperation and rehospitalization, it is difficult to draw conclusions regarding the impact of trainee participation on postoperative outcomes using the NSQIP database given the difficulty in eliminating hidden confounders related to the selection of intraoperative assistants.
In further evaluating indication for readmission and the role it may play in readmission avoidance, it must be noted that, likely related to baseline differences in health status for the patients undergoing these procedures, readmission indication was not be similarly distributed across amputation level given. Patients receiving an AKA as compared to a TMA in the NSQIP cohort were nearly two times more likely to be ASA class 4 or greater (Table I). Accordingly, AKA patients were more likely to be readmitted for non-wound related infectious complications compared to BKA or TMA patients such as pneumonia, UTI and sepsis. This is in contrast to readmissions for wound complications or complications of peripheral vascular disease that showed trends toward an increase in incidence with distal amputation.
While the particular care needs of LEA patients may vary by amputation level, a preponderance of post-discharge wound complications were seen for amputations at all levels. Henry and colleagues reported similar results at their institutions with 52.8% of 30-day readmissions taking place for amputation and peripheral arterial disease related issues. Interestingly, guillotine amputations were seen to have decreased readmission rates relative to non-guillotine amputations which may be attributable to a 2.5-fold increase in SSI for non-guillotine amputations. However, guillotine amputations, whether later revised in the OR or allowed to heal secondarily with a negative pressure dressing, often dramatically delay patients’ return to function and as such, this approach must be weighed carefully. A 2011 report by Hasanadka et al21 reviewed the NSQIP for risk factors of surgical site infection to find elevated preoperative INR and smoking status to be the lone modifiable risk factors for wound infection. These findings further emphasize the critical importance of preoperative smoking cessation while also highlighting the necessity for excellent intra-operative hemostasis to avoid hematoma in this population already at high risk for SSI.
Any discussion of readmission necessarily involves assessment of the appropriateness of hospital length of stay as this is thought to provide a counterbalance to the incidence of readmission. In the NSQIP cohort, increased length of stay was correlated with a decreased risk of readmission (Table III). However, given that readmissions occurred at a mean of 11 days (SD ± 7) after discharge and NSQIP follow up ends at 30 days post-surgery, a longer length of stay shortens the post-discharge follow up for these patients thus resulting in a decreased likelihood of readmission. For this reason, length of stay was excluded from our multivariable model. Of note, Medicare readmission penalties are assessed for readmission within 30 days of hospital discharge while the NSQIP only follows patients for 30 days after their index operation.2,15 Thus, this data set is unable to provide accurate estimates regarding the effect of hospital stay duration on readmission risk. However, as noted in the Results section, post-discharge wound infections were diagnosed, on average, over a week after leaving the hospital and post-discharge mortality occurred approximately ten days after discharge. While this certainly argues for close postoperative follow up, it is unclear whether marginal increases in length of stay would provide a benefit with respect to the development of these complications.
The findings of this study must be interpreted in the context of the study design. The NSQIP database provides data de-identified at the patient, institutional and regional level that precludes the investigation of institutional and regional variation in readmission. Future studies using alternate data sources may provide valuable insight through the identification of high performing centers and their best practices. The database also does not include information such as insurance type or median zip code income to assess the impact of socioeconomic factors on readmission. However, as a prospectively collected, multi-center, clinical database whose methodology includes direct patient contact within 30 days, the NSQIP database has the particular strength of capturing readmissions to both the operating institution and others. While the NSQIP database did collect readmission indication in 2012, it did so only for readmissions deemed related to the index procedure; approximately half of unplanned readmissions. As the relationship of certain systemic complications to the index procedure is highly subjective, future iterations of the NSQIP may consider including readmission indication for all readmitted patients.
CONCLUSIONS
This study represents the first and largest report from a multi-center clinical database on the incidence of and risk factors for readmission after LEA. With one in five patients rehospitalized after LEA, we have shown readmission to be common and given its association with a two-fold increase in mortality, we have also shown readmission to be incredibly costly to our patients. Risk factors for readmission include patient characteristics as well as the occurrence of reoperation on index admission. While indications for readmission require further study, we have shown that a majority of readmissions are related to wound complications. Strategies to mitigate postoperative readmission should place particular attention on preoperative optimization of comorbid illness with an emphasis on smoking cessation and intraoperative factors to prevent both wound complications and reoperation.
Footnotes
Presented as a poster at the 2013 Vascular Annual Meeting of the Society for Vascular Surgery (May 2013, San Francisco, CA).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- 1.Medicare Payment Advisory Commission (U.S.) Report to the Congress: promoting greater efficiency in Medicare. Washington, DC: Med PAC; 2007. [Google Scholar]
- 2.United States., United States. Congress. House. Office of the Legislative Counsel, United States. Congress. House. Committee on Ways and Means., United States. Congress. House. Committee on Energy and Commerce., United States. Congress. House. Committee on Education and Labor. Compilation of Patient Protection and Affordable Care Act: as amended through November 1, 2010 including Patient Protection and Affordable Care Act health-related portions of the Health Care and Education Reconciliation Act of 2010. Washington: U.S. Government Printing Office; 2010. [Google Scholar]
- 3.Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med. 2009;360:1418–28. doi: 10.1056/NEJMsa0803563. [DOI] [PubMed] [Google Scholar]
- 4.Vogel TR, Kruse RL. Risk factors for readmission after lower extremity procedures for peripheral artery disease. J Vasc Surg. 2013;58:90–7. e1–4. doi: 10.1016/j.jvs.2012.12.031. [DOI] [PubMed] [Google Scholar]
- 5.McPhee JT, Barshes NR, Ho KJ, Madenci A, Ozaki CK, Nguyen LL, et al. Predictive factors of 30-day unplanned readmission after lower extremity bypass. J Vasc Surg. 2013;57:955–62. doi: 10.1016/j.jvs.2012.09.077. [DOI] [PubMed] [Google Scholar]
- 6.McPhee JT, Nguyen LL, Ho KJ, Ozaki CK, Conte MS, Belkin M. Risk prediction of 30-day readmission after infrainguinal bypass for critical limb ischemia. J Vasc Surg. 2013;57:1481–8. doi: 10.1016/j.jvs.2012.11.074. [DOI] [PubMed] [Google Scholar]
- 7.Gupta PK, Fernandes-Taylor S, Ramanan B, Engelbert TL, Kent KC. Unplanned readmissions after vascular surgery. J Vasc Surg. 2013 doi: 10.1016/j.jvs.2013.09.002. [DOI] [PubMed] [Google Scholar]
- 8.Lo RC, Darling J, Bensley RP, Giles KA, Dahlberg SE, Hamdan AD, et al. Outcomes following infrapopliteal angioplasty for critical limb ischemia. J Vasc Surg. 2013;57:1455–63. doi: 10.1016/j.jvs.2012.10.109. discussion 63–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Vogel TR, Dombrovskiy VY, Carson JL, Graham AM. In-hospital and 30-day outcomes after tibioperoneal interventions in the US Medicare population with critical limb ischemia. J Vasc Surg. 2011;54:109–15. doi: 10.1016/j.jvs.2010.12.055. [DOI] [PubMed] [Google Scholar]
- 10.Jones WS, Patel MR, Dai D, Subherwal S, Stafford J, Calhoun S, et al. Temporal trends and geographic variation of lower-extremity amputation in patients with peripheral artery disease: results from U.S. Medicare 2000–2008. J Am Coll Cardiol. 2012;60:2230–6. doi: 10.1016/j.jacc.2012.08.983. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Ziegler-Graham K, MacKenzie EJ, Ephraim PL, Travison TG, Brookmeyer R. Estimating the prevalence of limb loss in the United States: 2005 to 2050. Archives of physical medicine and rehabilitation. 2008;89:422–9. doi: 10.1016/j.apmr.2007.11.005. [DOI] [PubMed] [Google Scholar]
- 12.Feinglass J, Pearce WH, Martin GJ, Gibbs J, Cowper D, Sorensen M, et al. Postoperative and late survival outcomes after major amputation: findings from the Department of Veterans Affairs National Surgical Quality Improvement Program. Surgery. 2001;130:21–9. doi: 10.1067/msy.2001.115359. [DOI] [PubMed] [Google Scholar]
- 13.Henry AJ, Hevelone ND, Hawkins AT, Watkins MT, Belkin M, Nguyen LL. Factors predicting resource utilization and survival after major amputation. J Vasc Surg. 2013;57:784–90. doi: 10.1016/j.jvs.2012.09.035. [DOI] [PubMed] [Google Scholar]
- 14.ACS NSQIP User Guide for the 2011 Participant Use Data File. 2012 at http://site.acsnsqip.org/participant-use-data-file/
- 15.ACS NSQIP User Guide for the 2012 Participant Use Data File. 2013 at http://site.acsnsqip.org/participant-use-data-file/
- 16.Sellers MM, Merkow RP, Halverson A, Hinami K, Kelz RR, Bentrem DJ, et al. Validation of new readmission data in the american college of surgeons National Surgical Quality Improvement Program. J Am Coll Surg. 2013;216:420–7. doi: 10.1016/j.jamcollsurg.2012.11.013. [DOI] [PubMed] [Google Scholar]
- 17.O’Brien PJ, Cox MW, Shortell CK, Scarborough JE. Risk factors for early failure of surgical amputations: an analysis of 8,878 isolated lower extremity amputation procedures. J Am Coll Surg. 2013;216:836–42. doi: 10.1016/j.jamcollsurg.2012.12.041. discussion 42–4. [DOI] [PubMed] [Google Scholar]
- 18.Mills E, Eyawo O, Lockhart I, Kelly S, Wu P, Ebbert JO. Smoking cessation reduces postoperative complications: a systematic review and meta-analysis. The American journal of medicine. 2011;124:144–54. e8. doi: 10.1016/j.amjmed.2010.09.013. [DOI] [PubMed] [Google Scholar]
- 19.Rigotti NA, Pasternak RC. Cigarette smoking and coronary heart disease: risks and management. Cardiology clinics. 1996;14:51–68. doi: 10.1016/s0733-8651(05)70260-5. [DOI] [PubMed] [Google Scholar]
- 20.Stead LF, Buitrago D, Preciado N, Sanchez G, Hartmann-Boyce J, Lancaster T. Physician advice for smoking cessation. Cochrane Database Syst Rev. 2013;5:CD000165. doi: 10.1002/14651858.CD000165.pub4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hasanadka R, McLafferty RB, Moore CJ, Hood DB, Ramsey DE, Hodgson KJ. Predictors of wound complications following major amputation for critical limb ischemia. J Vasc Surg. 2011;54:1374–82. doi: 10.1016/j.jvs.2011.04.048. [DOI] [PubMed] [Google Scholar]