This population-level analysis of medical records from the Nationwide Readmissions Database evaluates risk factors, causes, and costs of hospital readmission after head and neck cancer surgery reconstruction.
Key Points
Question
What are the risk factors, causes, and hospital costs associated with 30-day readmission after flap reconstruction of head and neck oncologic defects?
Findings
In this retrospective population-level analysis of medical records in the Nationwide Readmissions Database, approximately 1 in 5 patients was readmitted within 30 days, most commonly for wound complications, and these readmissions represented about one-third the cost of the initial admission. Independent risk factors for readmission included socioeconomic status, complex ablative procedures, and patient comorbidities.
Meaning
Identifying the incidence, causes, and risk factors for readmission is an important step toward developing interventions to prevent or mitigate these costly occurrences.
Abstract
Importance
Thirty-day hospital readmissions have substantial direct costs and are increasingly used as a measure of quality care. However, data regarding the risk factors and reasons for readmissions in head and neck cancer surgery reconstruction are lacking.
Objective
To describe the rate, risk factors, and causes of 30-day readmission in patients with head and neck cancer following free or pedicled flap reconstruction.
Design, Setting, and Participants
This retrospective, population-based cohort study analyzed medical records from the Nationwide Readmissions Database of 9487 patients undergoing pedicled or free flap reconstruction of head and neck oncologic defects between January 1, 2010, and December 31, 2014. Data analysis was performed in October 2017.
Exposures
Pedicled or free flap reconstruction of an oncologic head and neck defect.
Main Outcomes and Measures
The primary outcome was 30-day all-cause readmissions. Secondary outcomes included risk factors, causes, and costs of readmission. Multivariate regression analyses were conducted to determine factors independently associated with 30-day readmissions.
Results
Among 9487 patients included in the study (6798 male; 71.7%), the median age was 63 years (interquartile range, 55-71 years), and the 30-day readmission rate was 19.4% (n = 1839), with a mean cost per readmission of $15 916 (standard error of the mean, $785). The most common indication for readmission was wound complication (26.5%, n = 487). On multivariate regression, significant risk factors for 30-day readmission were median household income in the lowest quartile (vs highest quartile: odds ratio [OR], 1.58; 95% CI, 1.18-2.11), congestive heart failure (OR, 1.68; 95% CI, 1.14-2.47), liver disease (OR, 2.02; 95% CI, 1.22- 3.33), total laryngectomy (OR, 1.40; 95% CI, 1.12-1.75), pharyngectomy (OR, 1.47; 95% CI, 1.08-2.01), blood transfusion (OR, 1.30; 95% CI, 1.04-1.64), discharge to home with home health care (vs routine: OR, 1.32; 95% CI, 1.04-1.67), and discharge to a nursing facility (vs routine: OR, 1.77; 95% CI, 1.30-2.40).
Conclusions and Relevance
Using the Nationwide Readmissions Database, we demonstrate that approximately 1 in 5 patients undergoing head and neck cancer surgery reconstruction is readmitted within 30 days of surgery. Readmissions are most commonly associated with wound complications. Socioeconomic status, complex ablative procedures, and patient comorbidities are independent risk factors for readmission. These findings may be useful to clinicians in developing perioperative interventions aimed to reduce hospital readmissions and improve quality of patient care.
Introduction
Preventing hospital readmissions is a growing priority for clinicians, hospitals, and policy makers. Avoidable readmissions are estimated to cost the Center for Medicare & Medicaid Services (CMS) up to $17 billion annually, with nearly 18% of patients requiring readmission within 30 days of discharge.1,2 In efforts to improve quality of care and reduce wasteful spending, CMS has instituted financial penalties for hospitals with excess 30-day readmissions after select medical hospitalizations and surgical procedures through the Hospital Readmissions Reduction Program (HRRP).3 Penalties were levied against nearly 80% of hospitals in 2017 with fines totaling $528 million, representing an increase of $108 million from the prior year.4 Although head and neck surgeries are not yet subject to such penalties, anticipated expansion in the scope of the program is likely to involve our specialty.
Previous research in otolaryngology has identified patients with head and neck cancer as a high-risk group for readmission, with rates ranging from 7.3% to 26.5%.5,6,7,8,9,10 Reconstructive surgery with pedicled or free flaps is often required for extensive oncologic defects, with the aim of facilitating optimal functional and aesthetic outcome. Although flap reconstruction adds operative complexity and alters the risk profile for complications after head and neck surgery,11 its impact on hospital readmissions has not been thoroughly explored. Postoperative risks specific to flap reconstruction, such as graft failure, donor site morbidity, and heightened wound care requirements may predispose these patients to rehospitalization. Furthermore, patients with oncologic defects requiring flap reconstruction tend to have worse preoperative medical status, longer postoperative hospital stays, and more complex supportive care needs, all of which can contribute to readmission. Existing studies on readmissions after these surgeries are sparse and have important methodological limitations including single-center design,12,13 shorter than 30-day postdischarge follow-up,14,15 and inability to capture the reasons for readmission.14,15 Additionally, no prior work, to our knowledge, has specifically described the financial impact of readmissions after these procedures. The purpose of this study, therefore, is to evaluate the particular risk factors, causes, and financial consequences associated with readmissions in patients undergoing flap reconstruction of head and neck oncologic defects using a population-level database that covers comorbidities, health care costs, and readmissions.
Methods
Data Source
We performed a retrospective cohort study using data from the Nationwide Readmissions Database (NRD) from 2010 to 2014. The NRD is a database of all-payer hospital inpatient stays developed as part of the Healthcare Cost and Utilization Project (HCUP) by the Agency for Healthcare Research and Quality (AHRQ).16 Data in the NRD are constructed as a compilation from individual state inpatient databases, which from January 2010 to December 2014 included data from 22 states accounting for nearly 50% of all US hospitalizations. Sample weights provided by the database allow investigators to produce estimates representative of 100% of national discharges. The database includes data typical of a hospital discharge record, including predefined variables for demographics, comorbidities, hospital characteristics, and admission characteristics. The NRD also contains verified patient-linkage numbers that can track patients across hospitals within a state for that year. The patient-linkage numbers do not track patients across years. Further details regarding the NRD is available through HCUP.16 Because the database uses publicly available information with no personal identifiers, full review by the University of California, Los Angeles institutional review board was not required.
Study Cohort
We identified hospital discharges for adult patients (≥18 years of age) who underwent an ablative procedure for a diagnosis of a malignant oropharyngeal, laryngeal, hypopharyngeal, or oral cavity neoplasm from 2010 to 2014 using International Classification of Disease, 9th revision (ICD-9) codes (eTable in the Supplement). We further selected those who underwent subsequent pedicled or free flap reconstruction of the head and neck oncologic defect (eTable in the Supplement).17 This yielded an initial sample of 12 270 patients. We excluded patients who died during the index admission (n = 203; in-hospital mortality, 1.6%). We further excluded patients who were discharged in December (n = 807) to ensure 30-day follow-up for each index hospitalization and patients with out-of-state residence (n = 1773) to avoid potential loss of follow-up. Thus, the study cohort included 9487 patients.
Study Variables
Patient characteristics included age, sex, primary payer, median household income, relevant comorbidities, and procedures performed at the index admission. Comorbidities were drawn from the Elixhauser comorbidity measures for administrative data.18 In addition to the Elixhauser comorbidities, we further evaluated the impact of tobacco use (305.1, V15.82) and prior radiation exposure (V15.3).
Admission characteristics included admission source, admission type (elective vs nonelective), discharge destination, and prolonged length of hospital stay (defined as longer than the median length of stay in our cohort). Hospital characteristics included hospital bed number, teaching status, ownership, and safety-net hospital status. Safety-net burden was defined as the percentage of treated patients per hospital with Medicaid or uninsured payer status.19 Safety-net hospitals were defined as those hospitals in the highest quartile of safety-net burden.
Outcomes Measures
The primary outcome of interest was 30-day all-cause readmissions. We defined a readmission as the CMS defines it, namely a readmission to the same (index) or different (nonindex) hospital within 30 days of discharge from the index hospitalization. Hospital transfers were not considered readmissions. For patients who had multiple readmissions within 30 days, only the first readmission was included. The primary cause of readmission was determined by searching the primary ICD-9 diagnostic codes for each readmission (eTable in the Supplement). Secondary outcomes of interest included patient-, admission-, and hospital-level risk factors for readmission as well as cost of readmission.
Statistical Analysis
We compared index admission characteristics of readmitted and nonreadmitted patients using χ2 tests for categorical variables and independent t tests for continuous variables. We generated national estimates using survey weights from the NRD. Inpatient costs were converted from NRD charges using the hospital-specific cost-to-charge ratios provided by the NRD, and adjusted for inflation to 2014 dollars using the medical component of the consumer price index.20
Univariate analysis was performed to identify associations between patient-, admission-, and hospital-level factors and the risk of readmission. To assess the independent contribution of each variable to the risk of readmission, variables significantly associated with readmission on univariate analysis (P < .05) were included in the multivariate model. Statistical tests were 2 sided. Statistical significance was indicated by P < .05. Statistical analyses were performed using Stata 14 (StataCorp LLC) in October 2017.
Results
Study Population
A total of 9487 eligible patients were identified who underwent flap reconstruction for a head and neck oncologic defect (Table 1 and Table 2). The median age was 63 years (interquartile range [IQR], 55-71 years), and the majority of patients were male (71.7%; n = 6798). Most operations were performed at large, teaching hospitals (74.6%; n = 7077). In univariate analysis, compared with patients who were not readmitted within 30 days, those readmitted were more likely to have Medicaid or Medicare insurance status, lower household income, advanced comorbidity, length of stay of 10 days or longer, and nonroutine discharge. Readmitted patients also more commonly had undergone total laryngectomy, pharyngectomy, blood transfusion, and percutaneous gastrostomy placement. No statistically significant differences were noted in patient age, history of prior radiation therapy, or any of the hospital characteristics.
Table 1. Demographics and Comorbidities Among Patients Undergoing Flap Reconstruction for Head and Neck Oncologic Defects by 30-Day Readmission Status.
| Variable | Patients, No. (%) | P Value | OR (95% CI)a | ||
|---|---|---|---|---|---|
| Total (N = 9487) | Without Readmission (n = 7648) | With Readmission (n = 1839) | |||
| Age, y | .27 | ||||
| <45 | 506 (5.3) | 437 (5.7) | 69 (3.7) | 1 [Reference] | |
| 45-59 | 3423 (36.1) | 2775 (36.3) | 648 (35.2) | 1.49 (0.89-2.50) | |
| 60-74 | 3943 (41.6) | 3143 (41.1) | 800 (43.5) | 1.62 (0.97-2.71) | |
| ≥75 | 1615 (17.0) | 1293 (16.9) | 322 (17.5) | 1.59 (0.93-2.72) | |
| Male | 6798 (71.7) | 5440 (71.1) | 1358 (73.9) | .21 | 1.15 (0.93-1.42) |
| Payer | .004 | ||||
| Private | 2724 (28.8) | 2312 (30.3) | 412 (22.4) | 1 [Reference] | |
| Medicare | 4456 (47.1) | 3496 (45.8) | 960 (52.4) | 1.54 (1.22-1.95) | |
| Medicaid | 1727 (18.3) | 1357 (17.8) | 370 (20.1) | 1.53 (1.15-2.03) | |
| Other | 555 (5.9) | 461 (6.1) | 93 (5.0) | 1.13 (0.73-1.74) | |
| Median household income quartile for zip code | .005 | ||||
| 76th-100th Percentile | 1881 (20.1) | 1601 (21.3) | 280 (15.5) | 1 [Reference] | |
| 51st-75th Percentile | 2159 (23.1) | 1746 (23.2) | 413 (22.8) | 1.35 (1.00-1.81) | |
| 26th-50th Percentile | 2565 (27.5) | 2075 (27.6) | 490 (27.1) | 1.35 (1.01-1.80) | |
| ≤25th Percentile | 2735 (29.3) | 2110 (28.0) | 625 (34.6) | 1.69 (1.27-2.25) | |
| APRDRG severity scale | .001 | ||||
| Minor loss of function | 624 (6.6) | 560 (7.3) | 64 (3.5) | 1 [Reference] | |
| Moderate loss of function | 3046 (32.1) | 2477 (32.4) | 569 (31.0) | 1.99 (1.13-3.51) | |
| Major loss of function | 4520 (47.6) | 3639 (47.6) | 881 (47.9) | 2.10 (1.20-3.65) | |
| Extreme loss of function | 1297 (13.7) | 973 (12.7) | 324 (17.6) | 2.89 (1.63-5.14) | |
| Anemia | 1477 (15.6) | 1181 (15.4) | 296 (16.1) | .68 | 1.05 (0.82-1.34) |
| CHF | 420 (4.4) | 288 (3.8) | 132 (7.2) | <.001 | 1.98 (1.35-2.89) |
| Chronic pulmonary disease | 2149 (22.7) | 1620 (21.2) | 529 (28.7) | <.001 | 1.50 (1.20-1.88) |
| Coagulopathy | 308 (3.2) | 233 (3.0) | 75 (4.1) | .27 | 1.35 (0.79-2.31) |
| Diabetes, uncomplicated | 1294 (13.6) | 1011 (13.2) | 283 (15.4) | .18 | 1.20 (0.92-1.55) |
| Diabetes, chronic complications | 154 (1.6) | 112 (1.5) | 42 (2.3) | .09 | 1.55 (0.93-2.60) |
| Hypertension | 4857 (51.2) | 3890 (50.9) | 967 (52.6) | .48 | 1.07 (0.88-1.30) |
| Liver disease | 310 (3.3) | 206 (2.7) | 104 (5.7) | .002 | 2.16 (1.33-3.51) |
| Fluid and electrolyte disorders | 2676 (28.2) | 2067 (27.0) | 609 (33.1) | .005 | 1.34 (1.09-1.64) |
| Obesity | 433 (4.6) | 353 (4.6) | 80 (4.3) | .78 | 0.94 (0.60-1.46) |
| Peripheral vascular disorders | 483 (5.1) | 355 (4.6) | 127 (6.9) | .02 | 1.53 (1.07-2.19) |
| Pulmonary circulation disorders | 162 (1.7) | 133 (1.7) | 29 (1.6) | .80 | 0.93 (0.51-1.69) |
| Renal failure | 369 (3.9) | 276 (3.6) | 93 (5.1) | .12 | 1.43 (0.90-2.26) |
| Weight loss | 2051 (21.6) | 1583 (20.7) | 468 (25.4) | .02 | 1.31 (1.05-1.63) |
| Prior radiation | 1309 (13.8) | 1021 (13.3) | 288 (15.7) | .18 | 1.20 (0.92-1.58) |
| Tobacco use | 4757 (50.1) | 3885 (50.8) | 871 (47.4) | .17 | 0.87 (0.72-1.06) |
Abbreviations: APRDRG, All Patient Refined Diagnosis-Related Group; CHF, congestive heart failure; OR, odds ratio.
Odds ratios and confidence intervals in columns 6 and 7 are generated by univariate logistic regression, whereas descriptive statistics in columns 2-4 are generated by cross-tabulation. The results of these different analyses are presented in the same table to avoid excessive duplication of tables.
Table 2. Procedure, Admission, and Hospital Characteristics Among Patients Undergoing Flap Reconstruction for Head and Neck Oncologic Defects by 30-Day Readmission Status.
| Variable | Patients, No. (%) | P Value | OR (95% CI)a | ||
|---|---|---|---|---|---|
| Total (N = 9487) | Without Readmission (n = 7648) | With Readmission (n = 1839) | |||
| Ablative Procedures | |||||
| Oral cavity excision | 2680 (28.2) | 2215 (29.0) | 465 (25.3) | .09 | 0.83 (0.66-1.03) |
| Partial glossectomy | 2258 (23.8) | 1861 (24.3) | 397 (21.6) | .19 | 0.86 (0.67-1.08) |
| Total glossectomy | 611 (6.4) | 499 (6.5) | 112 (6.1) | .66 | 0.93 (0.66-1.30) |
| Oropharynx excision | 421 (4.4) | 342 (4.5) | 79 (4.3) | .88 | 0.96 (0.57-1.63) |
| Tonsillectomy | 362 (3.8) | 289 (3.8) | 73 (3.9) | .87 | 1.04 (0.61-1.79) |
| Pharyngectomy | 1086 (11.4) | 827 (10.8) | 259 (14.1) | .05 | 1.35 (1.00-1.82) |
| Mandibulectomy | 2610 (27.5) | 2090 (27.3) | 520 (28.3) | .67 | 1.04 (0.85-1.30) |
| Maxillectomy | 95 (1.0) | 79 (1.0) | 16 (1.0) | .66 | 0.84 (0.37-1.86) |
| Partial laryngectomy | 129 (1.4) | 103 (1.3) | 26 (1.4) | .87 | 1.06 (0.51-2.21) |
| Total laryngectomy | 2640 (27.8) | 1967 (25.7) | 673 (36.6) | <.001 | 1.67 (1.35-2.05) |
| Esophagectomy | 90 (1.0) | 75 (1.0) | 17 (0.9) | .90 | 0.94 (0.37-2.38) |
| Hypopharynx excision | 238 (2.5) | 176 (2.3) | 62 (3.4) | .23 | 1.49 (0.77-2.86) |
| Concurrent Procedures | |||||
| Neck dissection | 5650 (59.6) | 4615 (60.3) | 1035 (56.3) | .10 | 0.85 (0.69-1.03) |
| Blood transfusion | 1904 (20.1) | 1434 (18.7) | 470 (25.6) | <.001 | 1.49 (1.19-1.86) |
| Central venous catheter placement | 401 (4.2) | 302 (4.0) | 98 (5.3) | .13 | 1.37 (0.90-2.08) |
| Percutaneous gastrostomy | 2277 (24.0) | 1788 (23.4) | 488 (26.6) | .12 | 1.19 (0.95-1.47) |
| Invasive mechanical ventilation for <96 h postoperatively | 804 (8.5) | 630 (8.2) | 174 (9.5) | .34 | 1.17 (0.85-1.60) |
| Admission Characteristics | |||||
| Admit from ED | 554 (5.8) | 428 (5.6) | 125 (6.8) | .24 | 1.23 (0.86-1.76) |
| Nonelective case | 1235 (13.0) | 969 (12.7) | 266 (14.5) | .22 | 1.17 (0.91-1.50) |
| Prolonged length of stay (≥10 d) | 4842 (51.0) | 3774 (49.3) | 1068 (58.1) | <.001 | 1.42 (1.17-1.73) |
| Discharge destination | <.001 | ||||
| Routine | 3044 (32.1) | 2615 (34.2) | 429 (23.3) | 1 [Reference] | |
| Home with home health care | 4479 (47.2) | 3584 (46.9) | 895 (48.7) | 1.52 (1.21-1.91) | |
| Nursing facility | 1802 (19.0) | 1314 (17.2) | 488 (26.5) | 2.26 (1.72-2.97) | |
| Short-term hospital | 140 (1.5) | 117 (1.5) | 24 (1.3) | 1.23 (0.63-2.39) | |
| Hospital Characteristics | |||||
| Bed number | .75 | ||||
| Small | 679 (7.2) | 544 (7.1) | 134 (7.3) | 1 [Reference] | |
| Medium | 1373 (14.5) | 1089 (14.2) | 284 (15.4) | 1.06 (0.70-1.59) | |
| Large | 7435 (78.4) | 6015 (78.6) | 1420 (77.3) | 0.96 (0.69-1.33) | |
| Teaching status | .11 | ||||
| Metropolitan nonteaching | 444 (4.7) | 364 (4.8) | 80 (4.4) | 1 [Reference] | |
| Metropolitan teaching | 8960 (94.4) | 7205 (94.2) | 1754 (95.4) | 1.10 (0.76-1.60) | |
| Nonmetropolitanb | Censored | Censored | Censored | Censored | |
| Ownership | .08 | ||||
| Government, nonfederal | 2336 (24.6) | 1928 (25.2) | 408 (22.2) | 1 [Reference] | |
| Private, nonprofit | 6670 (70.3) | 5371 (70.2) | 1299 (70.6) | 1.14 (0.92-1.41) | |
| Private, invest-own | 482 (5.1) | 350 (4.6) | 132 (7.2) | 1.77 (1.02-3.08) | |
| Safety-net hospital | 2270 (23.9) | 1798 (23.5) | 472 (25.7) | .33 | 1.12 (0.86-1.43) |
Abbreviations: ED, emergency department; OR, odds ratio.
Odds ratios and confidence intervals in columns 6 and 7 are generated by univariate logistic regression, whereas descriptive statistics in columns 2-4 are generated by cross-tabulation. The results of these different analyses are presented in the same table to avoid excessive duplication of tables.
Censored due to fewer than the Healthcare Cost and Utilization Project minimum of 11 cases.
Incidence and Risk Factors for Readmission
Among 9487 included patients, 1839 (19.4%) were readmitted within 30 days. Readmissions to a hospital other than the initial treating facility accounted for 26.3% (n = 484) of all readmissions. After controlling for other covariates, factors significantly associated with increased odds of readmission were median household income in the lowest quartile (vs highest quartile: odds ratio [OR], 1.58; 95% CI, 1.18-2.11), congestive heart failure (OR, 1.68; 95% CI, 1.14-2.47), liver disease (OR, 2.02; 95% CI, 1.22- 3.33), total laryngectomy (OR, 1.40; 95% CI, 1.12-1.75), pharyngectomy (OR, 1.47; 95% CI, 1.08-2.01), blood transfusion (OR, 1.30; 95% CI, 1.04-1.64), discharge to home with home health care (vs routine: OR, 1.32; 95% CI, 1.04-1.67), and discharge to a nursing facility (vs routine: OR, 1.77; 95% CI, 1.30-2.40) (Table 3).
Table 3. Multivariate Analysis of Factors Associated With 30-Day Readmission.
| Variable | OR (95% CI) | P Value |
|---|---|---|
| Payer | ||
| Private | 1 [Reference] | NA |
| Medicare | 1.23 (0.96-1.57) | .09 |
| Medicaid | 1.26 (0.93-1.71) | .13 |
| Other | 1.15 (0.52-2.52) | .72 |
| Median household income quartile for zip code | ||
| 76th-100th percentile | 1 [Reference] | NA |
| 51st-75th percentile | 1.35 (1.01-1.82) | .04 |
| 26th-50th percentile | 1.27 (0.95-1.71) | .10 |
| ≤25th percentile | 1.58 (1.18-2.11) | .002 |
| APRDRG severity scale | ||
| Minor loss of function | 1 [Reference] | NA |
| Moderate loss of function | 1.43 (0.81-2.53) | .22 |
| Major loss of function | 1.33 (0.76-2.34) | .31 |
| Extreme loss of function | 1.45 (0.77-2.69) | .24 |
| Congestive heart failure | 1.68 (1.14-2.47) | .009 |
| Chronic pulmonary disease | 1.25 (0.99-1.57) | .06 |
| Liver disease | 2.02 (1.22-3.33) | .006 |
| Fluid and electrolyte disorders | 1.04 (0.83-1.30) | .70 |
| Peripheral vascular disorders | 1.23 (0.85-1.80) | .28 |
| Weight loss | 1.06 (0.82-1.36) | .67 |
| Total laryngectomy | 1.40 (1.12-1.75) | .002 |
| Pharyngectomy | 1.47 (1.08-2.01) | .02 |
| Blood transfusion | 1.30 (1.04-1.64) | .02 |
| Percutaneous gastrostomy | 1.02 (0.80-1.29) | .85 |
| Discharge destination | ||
| Routine | 1 [Reference] | NA |
| Home with home health care | 1.32 (1.04-1.67) | .02 |
| Nursing facility | 1.77 (1.30-2.40) | <.001 |
| Short-term hospital | 0.86 (0.44-1.68) | .68 |
Abbreviations: APRDRG, All Patient Refined Diagnosis-Related Group; NA, not applicable; OR, odds ratio.
Timing, Causes, and Costs of Readmission
The distribution of time to first readmission is shown in Figure 1. Median time to readmission was 9 days (IQR, 4-19 days). Median length of stay for readmissions was 5 days (IQR, 3-9 days) compared with 10 days (IQR, 7-17 days) for the index admission. The rate of mortality for the readmission hospitalization was 3.9%.
Figure 1. Days to First Readmission.
Readmissions are scattered throughout the postoperative period, although 44% occur during the first week, and 64% occur during the first 2 weeks.
Causes of 30-day readmission are shown in Figure 2A. Wound complications were the most common reason for readmission, accounting for 26.5% (n = 487) of readmissions (surgical site infection [SSI], 13.6% [n = 250]; fistula, 6.6% [n = 121]; dehiscence, 6.3% [n = 116]). Other medical diagnoses were the second most common cause of readmissions (15.7%; n = 289). Sepsis/septicemia resulted in 7.3% of readmissions (n = 134) and head and neck cancer diagnoses accounted for 5.7% (n = 105) of readmissions. Pneumonia and tracheostomy complications were identified as the primary cause of readmission in 5.5% (n = 101) and 4.3% (n = 79) of patients, respectively. Among early readmissions (≤14 days after discharge), wound complications remained the most common reason for readmission (31.8%; n = 374), although for late readmissions (15-30 days after discharge), other medical diagnoses were more common (18.6%; n = 122) (Figure 2A).
Figure 2. Readmission Reasons and Costs.
A, For all 30-day readmissions, wound complications (26.5%, n = 487) were the most common readmission diagnoses. For early readmissions (≤14 days after discharge), wound complications remained the most common diagnoses (31.8%, n = 374), but for late readmissions (15-30 days after discharge), other medical diagnoses were the leading causes (18.6%, n = 122), suggesting that a shorter time window (eg, 14 days postdischarge) may better capture related readmissions. B, The mean cost per readmission varied widely by the primary diagnosis at readmission, with the highest costs associated with graft complications and respiratory failure.
The total aggregate costs of all 30-day readmissions were $29.27 million. The mean cost per readmission was $15 916, and median cost was $10 029 (IQR, $5273-$19 029). The mean cost of the index admission was $48 088, and median cost was $39 866 (IQR, $25 840-$60 454). Cost of the first 30-day readmission accounted for 22.9% of the total cost of the episode of care (index admission + readmission) among readmitted patients. There was no significant difference in cost between a readmission to the index hospital or to a different hospital. When stratified by reason for readmission, graft complications ($35 749; standard error of the mean [SEM], $11 019) and respiratory failure ($30 606; SEM, $4561) led to the costliest readmissions, while tracheostomy complications ($11 697; SEM, $1830) and electrolyte/nutrition problems ($9070; SEM, $1769) were the least costly (Figure 2B).
Discussion
Rate of Readmission
Prior studies in head and neck surgery have quantified 30-day readmissions after different operations including general otolaryngologic procedures (7.3%-9.5%),21,22,23 multiple-subsite head and neck cancer operations (7.3%-16.1%),5,6,7 and total laryngectomy (13.9%-26.5%).8,9,10 However, there is very limited representation of head and neck reconstructive surgery within the readmission literature. The overall rate in the present study (19.4%) was higher than 2 previous national level analyses of head and neck free flap surgery by Carniol et al (9.6%)15 and Garg et al (8.8%),14 both using the American College of Surgeons’ National Quality Improvement Program (ACS-NSQIP) database. However, the ACS-NSQIP only captures readmissions that occur within 30 days of surgery, rather than the 30-day postdischarge metric used by the CMS. With average length of stays in these studies exceeding 9 days, this limitation is likely to underestimate the rate of 30-day readmissions. Two single-institution studies of head and neck cancer reconstruction, at Brigham and Women’s Hospital12 and Massachusetts Eye and Ear Infirmary,13 have reported rates of 11.6% and 19.8%, respectively. Single-center readmission rates must be interpreted with caution given the inability to capture readmissions to outside institutions, which have represented as many as one-third of readmissions in prior studies.24 Furthermore, case complexity, patient factors, and thresholds for readmitting patients after postdischarge complications likely vary across institutions; thus, the generalizability of these studies may be limited. To our knowledge, the present study uses the largest nationwide data set of head and neck flap reconstruction readmissions to index and nonindex hospitals within 30 days of discharge; it provides the most reliable existing baseline value for rehospitalizations after these operations.
Reasons for Readmission
In the present cohort, wound complications were the most common reason for 30-day readmissions, which supports prior studies of major ablative6 and reconstructive surgery12 of the upper aerodigestive tract. The risk of surgical site complications after head and neck reconstruction is thought to be higher than for other anatomic sites because of factors including the formation of communication between the upper aerodigestive tract and the neck, contamination from salivary and respiratory secretions, and the relative inaccessibility of certain head and neck sites such as the larynx and pharynx to local wound care.12,25 Antibiotic prophylaxis and postoperative wound care practices vary widely among reconstructive surgeons.26 Therefore, this may represent a potential area for quality improvement. More aggressive predischarge evaluation of surgical sites may detect brewing wound infections or identify at-risk patients who may benefit from closer follow-up. Late-onset infections (>14 days after surgery) represent a considerable portion of flap SSIs (35%-39%)27,28 and likely pose a greater threat for readmission than early infections, which may be detected before discharge. While early-onset flap SSIs are generally thought to develop from the patient’s own oral or skin flora introduced to the wound during surgery, late-onset flap SSIs may be caused by bacteria acquired through secondary contamination of the intraoral wound postoperatively. Given that oral flora change quickly during hospitalization,29,30 late-onset SSIs due to bacteria acquired several days after surgery may not be prevented by perioperative antibiotics. Longer courses of perioperative antibiotics, which one might consider as a potential solution, have not effectively prevented infection and may actually increase them.31,32 An additional prophylactic pulse of antibiotics (eg, <24 hours) given about 7 days postoperatively targeting pathogens in oral flora cultures 4 to 5 days postoperatively has been suggested as a means to further decrease late-onset SSIs.28
In the present study, wound complications were the most common reason for early readmissions (≤14 days after discharge), but other medical diagnoses were more common for late readmissions (15-30 days after discharge). This suggests that potentially a shorter postdischarge interval may be more effective for capturing related and preventable readmissions. The 30-day time frame for readmissions was initially chosen by the CMS to capture both medical and surgical readmissions; however, given the nature of surgery, complications directly related to the index hospitalization tend to occur early in the postoperative period. Courtwright et al33 studied lung transplantation patients and found that 78% of readmissions classified as preventable occurred with the first week. A shorter time interval for hospital readmissions (eg, 14 days) may be a better metric of an institution’s quality of care for surgical hospitalizations.
Risk Factors for Readmission
Analysis of patient-level comorbidities demonstrated that congestive heart failure and liver disease were significantly associated with readmission. Patients with these conditions often have multiple risk factors for poor surgical wound healing including chronic tissue ischemia, malnutrition, and cachexia, which may increase the risk for readmission.34 Preoperative clinic appointments designed to optimize a patient’s medical status may be effective at reducing readmissions related to underlying medical conditions. Dziegielewski et al7 found an 8-fold reduction in head and neck surgery readmissions for patients who attended a presurgical clinic. Kamal et al35 observed a 3-fold reduction in intensive-care unit admission and mortality from high-risk orthopedic surgery after implementing a multidisciplinary preoperative assessment with an anesthesiologist, surgeon, and nurse practitioner. Such encounters may involve medication adjustment, additional radiologic or laboratory examinations, or further patient education. Nurse-led preoperative assessments have also been shown to decrease patient anxiety, a risk factor that has been associated with unplanned readmissions in prior research.33
Lower median household income was a significant risk factor for 30-day readmission. Socioeconomically disadvantaged patients tend to present with more advanced disease,36 have greater rates of nonadherence to medication regimens and lifestyle restrictions,37 and may have difficulty affording care supplies or accessing other resources needed to manage postoperative problems. Increased attention to discharge planning and coordination of postdischarge care is warranted in this particularly high-risk demographic.
Patients who underwent ablative surgeries that included total laryngectomy or pharyngectomy prior to reconstruction had increased odds of readmission. These operations transgress anatomic structures essential to speech, breathing, and swallowing. Resultant physiologic changes can be life-altering, often requiring the placement of a tracheostomy or gastrostomy tube, and impose a substantial burden on patients and caretakers in the early postoperative period. Graboyes et al10 found that most emergency department visits after total laryngectomy were related to the laryngectomy stoma or tracheoesophageal puncture, and that emergency department visits predicted a 5-fold increased risk of readmission. Flap reconstruction introduces further challenges to the airway management of these patients because obstruction via oropharyngeal edema, swelling of the flap, or hemorrhage can be life-threatening and require urgent action. While flap failure and wound complications have received the most attention in the microvascular literature (and justifiably so), the need for increased supportive care has also represented a substantial portion (15%-20%)12,13 of reasons for readmission in prior case series, including the present study, and deserve further investigation. Greater emphasis on stomal care, potentially through formal predischarge evaluations, may be beneficial at reducing these hospital revisits. Studies comparing the efficacy of different strategies for patient and family education are lacking and, when conducted, may provide the foundation for standards of care to guide clinicians and caregivers.
Limitations
The current study is not without limitations. The use of administrative data relies on ICD-9 coding practices, which may vary among clinicians and administrative staff or contain coding errors. However, the ICD-9 coding system has been frequently used and validated to identify the diagnoses and procedures defined in our study.17,21,38 The primary diagnosis codes of a readmission may not always correlate with the true reason for readmission. We were unable to reliably separate planned and unplanned readmissions. Additionally, the ICD-9 code for flap reconstruction does not differentiate between microvascular and pedicled flap reconstruction. Finally, the NRD does not contain tumor-specific information, which prevented us from adjusting for cancer stage or pathologic characteristics in our analysis.
Conclusions
In the first national database analysis to our knowledge to capture 30-day readmissions to primary and outside hospitals after flap reconstruction for head and neck oncologic defects, we demonstrate that approximately 1 in 5 patients is readmitted within 30 days. Readmissions are most commonly associated with wound complications and occur early after discharge. Significant risk factors for readmission include lower socioeconomic status, complex ablative procedures, and certain patient comorbidities. These findings may be useful to clinicians in developing preoperative initiatives to improve medical optimization of high-risk patients and postoperative protocols aimed to reduce hospital readmissions and improve quality of patient care.
eTable. ICD-9 Procedure and Diagnosis Codes
References
- 1.Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med. 2009;360(14):1418-1428. doi: 10.1056/NEJMsa0803563 [DOI] [PubMed] [Google Scholar]
- 2.Zuckerman RB, Sheingold SH, Orav EJ, Ruhter J, Epstein AM. Readmissions, observation, and the hospital readmissions reduction program. N Engl J Med. 2016;374(16):1543-1551. doi: 10.1056/NEJMsa1513024 [DOI] [PubMed] [Google Scholar]
- 3.Centers for Medicare and Medicaid Services Readmissions reduction program. https://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/AcuteInpatientPPS/Readmissions-Reduction-Program.html. Accessed November 1, 2017.
- 4.Rau J. Medicare’s readmission penalties hit new high. 2016; https://khn.org/news/more-than-half-of-hospitals-to-be-penalized-for-excess-readmissions/. Accessed November 1, 2017.
- 5.Chaudhary H, Stewart CM, Webster K, et al. . Readmission following primary surgery for larynx and oropharynx cancer in the elderly. Laryngoscope. 2017;127(3):631-641. doi: 10.1002/lary.26311 [DOI] [PubMed] [Google Scholar]
- 6.Chen MM, Orosco RK, Harris JP, et al. . Predictors of readmissions after head and neck cancer surgery: a national perspective. Oral Oncol. 2017;71:106-112. doi: 10.1016/j.oraloncology.2017.06.010 [DOI] [PubMed] [Google Scholar]
- 7.Dziegielewski PT, Boyce B, Manning A, et al. . Predictors and costs of readmissions at an academic head and neck surgery service. Head Neck. 2016;38(suppl 1):E502-E510. doi: 10.1002/hed.24030 [DOI] [PubMed] [Google Scholar]
- 8.Ferrandino R, Garneau J, Roof S, et al. . The national landscape of unplanned 30-day readmissions after total laryngectomy. Laryngoscope. 2018;128(8):1842-1850. doi: 10.1002/lary.27012 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Goepfert RP, Hutcheson KA, Lewin JS, et al. . Complications, hospital length of stay, and readmission after total laryngectomy. Cancer. 2017;123(10):1760-1767. doi: 10.1002/cncr.30483 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Graboyes EM, Yang Z, Kallogjeri D, Diaz JA, Nussenbaum B. Patients undergoing total laryngectomy: an at-risk population for 30-day unplanned readmission. JAMA Otolaryngol Head Neck Surg. 2014;140(12):1157-1165. doi: 10.1001/jamaoto.2014.1705 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Tan BK, Por YC, Chen HC. Complications of head and neck reconstruction and their treatment. Semin Plast Surg. 2010;24(3):288-298. doi: 10.1055/s-0030-1263070 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Offodile AC II, Pathak A, Wenger J, Orgill DP, Guo L. Prevalence and patient-level risk factors for 30-day readmissions following free tissue transfer for head and neck cancer. JAMA Otolaryngol Head Neck Surg. 2015;141(9):783-789. doi: 10.1001/jamaoto.2015.1323 [DOI] [PubMed] [Google Scholar]
- 13.Osborn HA, Rathi VK, Tjoa T, et al. . Risk factors for thirty-day readmission following flap reconstruction of oncologic defects of the head and neck. Laryngoscope. 2018;128(2):343-349. doi: 10.1002/lary.26726 [DOI] [PubMed] [Google Scholar]
- 14.Garg RK, Wieland AM, Hartig GK, Poore SO. Risk factors for unplanned readmission following head and neck microvascular reconstruction: Results from the National Surgical Quality Improvement Program, 2011-2014. Microsurgery. 2017;37(6):502-508. doi: 10.1002/micr.30116 [DOI] [PubMed] [Google Scholar]
- 15.Carniol ET, Marchiano E, Brady JS, et al. . Head and neck microvascular free flap reconstruction: An analysis of unplanned readmissions. Laryngoscope. 2017;127(2):325-330. doi: 10.1002/lary.26039 [DOI] [PubMed] [Google Scholar]
- 16.Healthcare Cost and Utilization Project (HCUP) Overview of the Nationwide Readmissions Database (NRD). 2015. https://www.hcup-us.ahrq.gov/nrdoverview.jsp. Accessed November 1, 2017.
- 17.Gooi Z, Gourin CG, Boahene KD, Byrne PJ, Richmon JD. Temporal trends in head and neck cancer surgery reconstruction. Head Neck. 2015;37(10):1509-1517. doi: 10.1002/hed.23786 [DOI] [PubMed] [Google Scholar]
- 18.Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care. 1998;36(1):8-27. doi: 10.1097/00005650-199801000-00004 [DOI] [PubMed] [Google Scholar]
- 19.Genther DJ, Gourin CG. The effect of hospital safety-net burden status on short-term outcomes and cost of care after head and neck cancer surgery. Arch Otolaryngol Head Neck Surg. 2012;138(11):1015-1022. doi: 10.1001/jamaoto.2013.611 [DOI] [PubMed] [Google Scholar]
- 20.Bureau of Labor Statistics Consumer price index. https://www.bls.gov/cpi/. Accessed November 8, 2017.
- 21.Graboyes EM, Kallogjeri D, Saeed MJ, Olsen MA, Nussenbaum B. 30-day hospital readmission following otolaryngology surgery: Analysis of a state inpatient database. Laryngoscope. 2017;127(2):337-345. doi: 10.1002/lary.25997 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Graboyes EM, Liou TN, Kallogjeri D, Nussenbaum B, Diaz JA. Risk factors for unplanned hospital readmission in otolaryngology patients. Otolaryngol Head Neck Surg. 2013;149(4):562-571. doi: 10.1177/0194599813500023 [DOI] [PubMed] [Google Scholar]
- 23.Han S, Smith TS, Gunnar W. Descriptive analysis of 30-day readmission after inpatient surgery discharge in the Veterans Health Administration. JAMA Surg. 2014;149(11):1162-1168. doi: 10.1001/jamasurg.2014.1706 [DOI] [PubMed] [Google Scholar]
- 24.Graboyes EM, Kallogjeri D, Saeed MJ, Olsen MA, Nussenbaum B. Postoperative care fragmentation and thirty-day unplanned readmissions after head and neck cancer surgery. Laryngoscope. 2017;127(4):868-874. doi: 10.1002/lary.26301 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Karakida K, Aoki T, Ota Y, et al. . Analysis of risk factors for surgical-site infections in 276 oral cancer surgeries with microvascular free-flap reconstructions at a single university hospital. J Infect Chemother. 2010;16(5):334-339. doi: 10.1007/s10156-010-0108-y [DOI] [PubMed] [Google Scholar]
- 26.Reiffel AJ, Kamdar MR, Kadouch DJ, Rohde CH, Spector JA. Perioperative antibiotics in the setting of microvascular free tissue transfer: current practices. J Reconstr Microsurg. 2010;26(6):401-407. doi: 10.1055/s-0030-1249606 [DOI] [PubMed] [Google Scholar]
- 27.Yarlagadda BB, Deschler DG, Rich DL, et al. . Head and neck free flap surgical site infections in the era of the Surgical Care Improvement Project. Head Neck. 2016;38(suppl 1):E392-E398. doi: 10.1002/hed.24005 [DOI] [PubMed] [Google Scholar]
- 28.Durand ML, Yarlagadda BB, Rich DL, et al. . The time course and microbiology of surgical site infections after head and neck free flap surgery. Laryngoscope. 2015;125(5):1084-1089. doi: 10.1002/lary.25038 [DOI] [PubMed] [Google Scholar]
- 29.Yoshioka H, Fujita K, Maruyama S, Oka R. Changes in aerobic pharyngeal flora related to antibiotic use and the emergence of gram-negative bacilli. Clin Pediatr (Phila). 1982;21(8):460-462. doi: 10.1177/000992288202100802 [DOI] [PubMed] [Google Scholar]
- 30.Johanson WG, Pierce AK, Sanford JP. Changing pharyngeal bacterial flora of hospitalized patients: emergence of gram-negative bacilli. N Engl J Med. 1969;281(21):1137-1140. doi: 10.1056/NEJM196911202812101 [DOI] [PubMed] [Google Scholar]
- 31.Lotfi CJ, Cavalcanti RdeC, Costa e Silva AM, et al. . Risk factors for surgical-site infections in head and neck cancer surgery. Otolaryngol Head Neck Surg. 2008;138(1):74-80. doi: 10.1016/j.otohns.2007.09.018 [DOI] [PubMed] [Google Scholar]
- 32.Avery CM, Ameerally P, Castling B, Swann RA. Infection of surgical wounds in the maxillofacial region and free flap donor sites with methicillin-resistant Staphylococcus aureus. Br J Oral Maxillofac Surg. 2006;44(3):217-221. doi: 10.1016/j.bjoms.2005.05.020 [DOI] [PubMed] [Google Scholar]
- 33.Courtwright AM, Zaleski D, Gardo L, et al. . Causes, preventability, and cost of unplanned rehospitalizations within 30 days of discharge after lung transplantation. Transplantation. 2018;102(5):838-844. doi: 10.1097/TP.0000000000002101 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Rhou YJ, Henshaw FR, McGill MJ, Twigg SM. Congestive heart failure presence predicts delayed healing of foot ulcers in diabetes: an audit from a multidisciplinary high-risk foot clinic. J Diabetes Complications. 2015;29(4):556-562. doi: 10.1016/j.jdiacomp.2015.02.009 [DOI] [PubMed] [Google Scholar]
- 35.Kamal T, Conway RM, Littlejohn I, Ricketts D. The role of a multidisciplinary pre-assessment clinic in reducing mortality after complex orthopaedic surgery. Ann R Coll Surg Engl. 2011;93(2):149-151. doi: 10.1308/003588411X561026 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Chen AY, Schrag NM, Halpern MT, Ward EM. The impact of health insurance status on stage at diagnosis of oropharyngeal cancer. Cancer. 2007;110(2):395-402. doi: 10.1002/cncr.22788 [DOI] [PubMed] [Google Scholar]
- 37.Ultee KHJ, Tjeertes EKM, Bastos Gonçalves F, et al. . The relation between household income and surgical outcome in the Dutch setting of equal access to and provision of healthcare. PLoS One. 2018;13(1):e0191464. doi: 10.1371/journal.pone.0191464 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Gourin CG, Frick KD. National trends in oropharyngeal cancer surgery and the effect of surgeon and hospital volume on short-term outcomes and cost of care. Laryngoscope. 2012;122(3):543-551. doi: 10.1002/lary.22447 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable. ICD-9 Procedure and Diagnosis Codes