- Email: [email protected]
Predictors of readmissions after head and neck cancer surgery: A national perspective
Oral Oncology 71 (2017) 106–112
Contents lists available at ScienceDirect
Oral Oncology journal homepage: www.elsevier.com/locate/oraloncology
Predictors of readmissions after head and neck cancer surgery: A national perspective Michelle M. Chen a, Ryan K. Orosco a, Jeremy P. Harris b, Julie B. Porter c, Eben L. Rosenthal a, Wendy Hara b, Vasu Divi a,⇑ a b c
Department of Otolaryngology-Head and Neck Surgery, Stanford University, 900 Blake Wilbur Drive, Third Floor, Stanford, CA 94305, USA Department of Radiation Oncology, Stanford University, 875 Blake Wilbur Drive, Rm CCG210 Clinic D, Stanford, CA 94305, USA Stanford Health Care, 875 Blake Wilbur Drive, Stanford, CA 94305, USA
a r t i c l e
i n f o
Article history: Received 14 February 2017 Received in revised form 14 June 2017 Accepted 15 June 2017
Keywords: Head and neck cancer Hospital readmission Patient readmission Patient discharge Patient care Laryngeal cancer Oropharyngeal cancer Oral cavity cancer Hypopharyngeal cancer
a b s t r a c t Objectives: Surgical oncology patients have multiple comorbidities and are at high risk of readmission. Prior studies are limited in their ability to capture readmissions outside of the index hospital that performed the surgery. Our goal is to evaluate risk factors for readmission for head and neck cancer patients on a national scale. Material and methods: A retrospective cohort study of head and neck cancer patients in the Nationwide Readmissions Database (2013). Our main outcome was 30-day readmission. Statistical analysis included 2-sided t tests, v2, and multivariate logistic regression analysis. Results: Within 30 days, 16.1% of 11,832 patients were readmitted and 20% of readmissions were at nonindex hospitals, costing $31 million. Hypopharyngeal cancer patients had the highest readmission rate (29.6%), followed by laryngeal (21.8%), oropharyngeal (18.2%), and oral cavity (11.6%) cancers (P < 0.001). Half of readmissions occurred within 10 days and were often associated with infections (27%) or wound complications (12%). Patients from lower household income areas were more likely to be readmitted (odds ratio [OR], 1.54; 95% confidence interval [CI], 1.16–2.05). Patients with valvular disease (OR, 2.07; 95% CI, 1.16–3.69), rheumatoid arthritis/collagen vascular disease (OR, 2.05; 95% CI, 1.27–3.31), liver disease (OR, 2.02, 95% CI, 1.37–2.99), and hypothyroidism (OR 1.30; 95% CI, 1.02–1.66) were at highest risk of readmission. Conclusion: The true rate of 30-day readmissions after head and neck cancer surgery is 16%, capturing non-index hospital readmissions which make up 20% of readmissions. Readmissions after head and neck cancer surgery are most commonly associated with infections and wound complications. Ó 2017 Elsevier Ltd. All rights reserved.
Introduction Hospital readmissions cost Medicare alone more than $17 billion annually in avoidable expenses, with over 17.5% of patients getting readmitted within 30-days of hospital discharge [1–3]. Only 10% of these Medicare readmissions are considered planned readmissions [1]. In order to encourage hospitals to reduce readmissions, the Affordable Care Act, which was passed in March 2010, created the Hospital Readmissions Reduction Program. Since October 2012, the Centers for Medicare and Medicaid Services (CMS) has levied a penalty on hospitals with readmissions in excess of the expected national rate. For fiscal year 2017, the penal-
⇑ Corresponding author. E-mail address: [email protected] (V. Divi). http://dx.doi.org/10.1016/j.oraloncology.2017.06.010 1368-8375/Ó 2017 Elsevier Ltd. All rights reserved.
ties are expected to total $528 million, which is $108 million more than the prior year [4]. The current program targets six medical conditions. Initially, it targeted acute myocardial infarctions, heart failure, and pneumonia. In 2015, total knee and hip surgery and chronic obstructive pulmonary disease were added; and in 2017, coronary artery bypass surgery was added [2,3]. Preliminary data since the introduction of the program has shown a reduction in the rates of avoidable readmissions, making the expansion of this program increasingly likely [2]. Oncology patients in particular have high rates of readmission following discharge, making them a possible target of the program. The concept of avoidable admissions in also present in the CMS Oncology Care Model, piloted on July 1, 2016, which seeks to capture the rates of emergency department visits and readmissions during chemotherapy.
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
Given the unique challenges faced by oncology patients, and head and neck cancer surgery patients in particular, there is a need to understand the risk factors for readmission so interventions can be effectively targeted and hospital benchmarks can be appropriately risk-adjusted. In addition, it is important for providers to understand the causes of these readmissions to help design programs that reduce avoidable readmissions. Prior studies on readmissions in otolaryngology are primarily institutional studies that are unable to capture readmissions that occur outside of a single institution. The purpose of our study was to evaluate the causes of and risk factors for readmission in head and neck cancer surgery patients utilizing a national database that covers readmissions at both the index hospital and outside facilities. This database also captures socioeconomic and comorbidity data to allow for evaluation of patient-level characteristics that influence readmission.
Materials and methods Our study was granted an exemption from our institutional review board. Patient data was drawn from the Healthcare Cost and Utilization Project’s Nationwide Readmissions Database (NRD) from January 1, 2013 to December 31, 2013. The NRD is a database of all-payer hospital inpatient stays from 21 states that accounts for 49.1% of all United States hospitalizations [5]. This stratified sample can be used to produce nationally weighted estimates. In 2013, the NRD covered 14 million discharges, resulting in a nationally weighted sample of 36 million discharges. All adult patients (age 18) with a primary diagnosis of head and neck cancer were chosen based on their International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes for oral cavity cancer (141.1–141.4, 141.8–141.9, 143.0–145.9), oropharyngeal cancer (141.0–141.6, 146.0–146.9), hyopharyngeal cancer (148.0–148.9), and laryngeal cancer (161.0–161.9). We limited our cohort to those who had surgery for the primary site based on their ICD-9-CM procedure codes. We also limited our index hospitalizations from January 1, 2013 to November 30, 2013, in order to ensure we were able to capture all 30-day readmissions. We excluded patients that died during the index hospitalization and any records that were combined transfer records that combined multiple hospitalizations into one. Demographic and socioeconomic status variables included age, sex, patient location, state residence, insurance status, household income, comorbidities, and number of chronic conditions. Age was grouped into four categories (18–49, 50–59, 60–69, and 70 years of age). Patient location was grouped into large metropolitan areas (populations 1 million), small metropolitan areas (populations from 50,000 to 999,999), and micropolitan and other areas. State residence was a binary characteristic that depended on whether the patient was a resident of the state of the index hospitalization. Insurance status was categorized as Medicare, Medicaid, private insurance, self-pay, and other (including no charge). Income was the median household income for the patient’s ZIP code and grouped into quartiles ($1–$37,999; $38,0 00–$47,999; $48,000–$63,999; and $64,000 or more). Comorbidity measures were identified by the Agency for Healthcare Research and Quality’s comorbidity software that identifies coexisting medication conditions that are likely present prior to hospital admission. The comorbidity measures included alcohol abuse, anemia (chronic blood loss and deficiency anemias), congestive heart failure, coagulopathy, depression, diabetes (complicated and uncomplicated), pulmonary disease (chronic lung disease and pulmonary circulation disorders), drug abuse, hypertension (complicated and uncomplicated), hypothyroidism, liver disease, fluid and electrolyte disorders, obesity, renal failure, weight loss, valvular heart disease, peripheral vascular disease, rheumatoid arthri-
107
tis/collagen vascular disease, other neurological disorders, and psychoses. Tobacco use was a binary characteristic created for any patient with an ICD-9-CM diagnosis code of tobacco use. Clinical variables for index hospitalization included tumor site, use of a flap during the index surgery, hospital length of stay, discharge location, costs, and primary diagnosis on readmission. Tumor site was classified as oral cavity, oropharynx, hypopharynx, and larynx. Discharge location was grouped into home, home with home health, discharge to skilled nursing facility (SNF) or other facilities, and leaving against medical advice (AMA). Total costs were determined by total charges and hospital-specific cost to charge ratios. Primary diagnosis on readmission was based on ICD-9-CM diagnosis codes and grouped into twelve categories (infection, other medical, head and neck cancer, other head and neck, wound complication, hematoma/hemorrhage, tracheostomy complication, electrolyte/nutrition, gastrostomy complication, pulmonary embolism/deep venous thromboembolism/stroke, chemotherapy/radiotherapy, or pain). ‘‘Other medical” encompassed diagnoses such as cardiovascular (heart failure, atrial fibrillation, chest pain, cardiac arrest, coronary artery disease), pulmonary (dyspnea), psychologic (anxiety, depression), renal, gastrointestinal, and genitourinary disorders. ‘‘Other head and neck” included diagnosis codes referring to edema in the head and neck, laryngeal stenosis, vocal fold paralysis, foreign bodies in the head and neck, and other diseases of the head and neck not classified elsewhere. Infections of tracheostomy and gastrostomy sites were grouped under the tracheostomy and gastrostomy complication categories rather than under infection. Hospital level variables included hospital size, ownership, location, and teaching status. Hospital size is grouped into small, medium, and large based on bedside and hospital location with larger cut-offs for hospitals in urban locations. Hospital ownership was categorized into government, private non-profit, and private investor-owned hospitals. Hospital location was classified as large metropolitan area (1 million residents), small metropolitan area (<1 million residents), and other (includes micropolitan and non-urban areas). Teaching status was divided into metropolitan non-teaching hospitals, metropolitan teaching hospitals, and non-metropolitan hospitals. Our primary outcome of interest was 30-day readmission defined as any hospital admission that occurred within 30 days of discharge from the index hospitalization. All statistical analysis was performed using STATA/SE software (version 14.2; StataCorp, College Station, TX, USA). Bivariate analysis using chi-square and t tests were used to analyze our categorical and continuous variables, respectively. Multivariate logistic regression analysis was used to identify factors associated with 30-day readmission. Odds ratios (OR) and 95% confidence intervals (95% CIs) were calculated for the strength of association. All tests were 2-sided, and a P value < 0.05 was considered to be statistically significant.
Results We identified a weighted total of 11,832 patients who had primary surgery for oral cavity, oropharyngeal, hypopharyngeal, and laryngeal cancer with a 30-day readmissions rate of 16.1%. Over half of these patients were readmitted through the Emergency Department and 20% were readmitted to a different hospital (Table 1). Just over half (51%) of initial 30-day readmissions occurred within 10 days of hospital discharge (see Fig. 1). Other than a diagnosis of head and neck cancer, the most common primary diagnosis on 30-day readmission was an infectious etiology (23.5%) (Fig. 2). In the first 10 days after hospital discharge, nearly half of readmissions were infectious (27.2%), wound-related (11.9%), and
108
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
Table 1 30-day readmissions for head and neck cancer surgery patients, national weighted sample. Totala
Readmissionsa
Emergency department readmissionsa
Mean proportion of readmissions at index hospitalb
Site
N
N
% of Total
N
% of Readmissions
% of Readmissions (SE)
$ (SE)
Oral Cavity Oropharynx Hypopharynx Larynx Total
6183 2090 452 3108 11832
717 380 134 678 1908
11.6 18.2 29.6 21.8 16.1
417 128 67 362 974
58.2 33.7 50.3 53.4 51.1
73.1 90.1 78.9 82.8 80.3
13,587 10,609 15,795 18,445 14,895
(3.5) (3.0) (6.6) (2.3) (2.2)
Mean cost per readmissionb
(1097) (834) (2031) (1815) (926)
SE, standard error. a National weighted totals. b National weighted means with linearized standard errors.
Fig. 1. Over half of head and neck cancer surgery patients are readmitted within the first 10 days after discharge.
tracheostomy (5.6%) or gastrostomy tube-associated (3.0%). For patients readmitted after 10 days, the most common diagnoses on readmission were other medical diagnoses (19.9%) or a head and neck cancer diagnosis (28.6%). For admissions occurring from January to November 2013, readmissions accounted for $31.1 million in total cost. The mean cost per readmission was $14,895 (SEM, $926), which was 44.9% of the average cost of the index hospitalization. The readmission cost was highest for laryngeal cancer patients (mean cost, $18,445) and lowest for oropharyngeal cancer patients (mean cost, $10,609). There was no difference in cost between readmission at the initial hospital and at a non-index hospital (P = 0.43). Patients with laryngeal and hypopharyngeal primaries had the highest rate of readmission at 21.8% and 29.6%, respectively; while patients with oropharyngeal cancer (18.2%) and oral cavity cancer (11.6%) had significantly lower readmission rates (P < 0.001). There was no difference in likelihood of readmission across age groups (P = 0.06), but Medicare (18.0%) and Medicaid patients (18.5%) were more likely to be readmitted than patients with private insurance (13.7%) or who self-paid for their care (10.9%) (P < 0.001) (Table 2). Compared with patients discharged home, patients dis-
charged to skilled nursing facilities and other facilities were more likely to be readmitted within 30 days (13.3% vs 22.3%, P < 0.001). Readmitted patients had longer index hospitalizations and more chronic conditions. On multivariate analysis, tumor site, a flap during the index procedure, lower median household incomes, and teaching hospital status were independently associated with increased odds of readmission (Table 3). Of the 21 comorbidities analyzed, valvular heart disease (odds ratio [OR], 2.07; 95% confidence interval [CI], 1.16–3.69), rheumatoid arthritis or collagen vascular disease (OR, 2.05; 95% CI, 1.27–3.31), liver disease (OR, 2.02, 95% CI, 1.37–2.99), and hypothyroidism (OR 1.30; 95% CI, 1.02–1.66) were independent risk factors for readmission. Hospital size, ownership, and location were not associated with readmission. The only hospital level factor associated with readmission was teaching status. Relative to non-teaching hospitals, metropolitan teaching hospitals were associated with increased odds of readmission (OR, 1.50; 95% CI, 1.04–2.16). We analyzed a subset of patients with only one 30-day readmission in order to investigate factors associated with index hospital readmission. Larger index hospitals (medium hospital: OR, 3.20; 95% CI, 1.23–8.32; large hospital: OR, 3.72; 95% CI, 1.75–7.92) were
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
109
Fig. 2. For all 30-day readmissions, infection (23.5%) was the most common primary diagnosis on readmission. For early readmissions (10 days after discharge), infection continued to be the most common diagnosis on readmission (27.2%), but for late readmissions (11–30 days after discharge), head and neck cancer (28.6%) and other medical diagnoses (19.9%) were more common. PE, pulmonary embolism; DVT, deep venous thromboembolism.
more likely to have patients return upon readmission than small index hospitals. Patients readmitted to non-index hospitals were more likely to be readmitted from the emergency department (OR, 0.21; 95% CI, 0.12–0.35). There was no difference in index hospital readmission between patients discharged home and those who were discharged to skilled nursing facilities. Discussion This study utilizes a national inpatient discharge database to provide national weighted estimates of readmissions after head and neck cancer surgery. We demonstrate that there is a 16.1% national 30-day readmission rate and a fifth of the readmissions were not at the index hospital. The total cost of readmissions for head and neck cancer surgery patients over 11 months was $31 million. Surgical site, flaps, areas with lower median household incomes, and hospital type were associated with 30-day readmission. Over half of all initial readmissions occurred within 10 days of discharge and early readmissions had a higher rate of infections and wound complications than late readmissions. Prior studies on otolaryngologic readmissions have been primarily single institution studies, state-specific studies, or focused on otolaryngologic procedures in aggregate. While readmissions after otolaryngologic procedures in aggregate have been shown to range from 3.1 to 9.5% [6–11], cancer patients represent a complex subset of patients who tend to be more frequently readmitted. Evaluating 155 total laryngectomy patients treated at one institution, Graboyes et al. found a 26.5% 30-day readmission rate [12]. Chaudhary et al. examined 1518 elderly oropharyngeal and laryngeal cancer Medicare patients and 14.1% were readmitted within 30 days of hospital discharge [13]. Similarly, our study reports a 16.1% overall readmission rate with the highest readmission rates for patients with laryngeal (21.8%) and hypopharyngeal (29.6%) cancer. Head and neck cancer patients are complex and have a higher rate of unavoidable readmissions than other otolaryngology patients. Readmissions are particularly important in this patient population since they can delay initiation of adjuvant therapy and thus potentially lead to decreased survival [14].
We found that each 30-day readmission costs $14,895 on average for all-payers, nearly half of the cost of the initial surgical hospitalization. This corroborates prior data by Chaudhary et al. on Medicare patients that demonstrated oropharyngeal and laryngeal cancer patients who were readmitted had mean costs that were $15,123 greater than those who were not readmitted [13]. Readmissions are a small, but significant proportion of healthcare costs for head and neck cancer surgery. Over half of patients were readmitted within 10 days of discharge. Of these, 27% had a primary diagnosis of infection and 12% had a diagnosis of a wound complication. This supports findings by Bur et al. who demonstrated that causes of unplanned readmissions after head and neck surgery were mostly infectious, including wound infection or breakdown (24%) and pneumonia (6.4%). However this study was limited because it looked at readmissions starting from 30 days of the surgical date instead of the discharge date, which may underestimate the actual 30-day readmissions for complex head and neck patients with prolonged hospital stays [11]. These admissions may present a unique opportunity for quality improvement. More aggressive assessment of potential infections and wound complications prior to discharge may help identify patients who require further treatment and close follow-up. This may be done with a combination of improved caregiver and patient education on early signs of infection or wound complications, post-discharge phone calls with targeted questions, and close follow-up in outpatient clinics. In our cohort, while infection was the most common diagnosis for early readmissions (10 days after discharge), other medical diagnoses and head and neck cancer diagnoses were more common for readmissions after 10 days. This suggests that perhaps a shorter time frame for readmissions may better capture related and avoidable readmissions. The 30-day readmission metric was initially chosen as a quality metric for both medical and surgical patients since readmissions after 30-days were less likely to be due to the acute care received in the hospital. It has been shown previously that half of all post-discharge complications in otolaryngology occur soon after discharge [15]. Readmissions that
110
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
Table 2 Baseline characteristics for head and neck cancer surgery patients by 30-day readmission status. 30-day readmissiona No N (%)
P value
177 518 633 579
0.06
Age
18–49 50–59 60–69 70+
1250 2909 3032 2733
Sex
Male Female
6768 (83.0) 3156 (85.8)
1386 (17.0) 522 (14.2)
0.03
Flap
No Yes
7654 (84.6) 2270 (81.6)
1395 (15.4) 513 (18.4)
0.02
Discharge Location
Home Home Health SNF/Other AMA
5724 (86.7) 3153 (81.3) 1029 (77.7) 19
877 (13.3) 725 (18.7) 295 (22.3) 10
<0.001
Insurance
Medicare Medicaid Private Self-Pay Other
4366 (82.0) 1191 (81.5) 3669 (86.3) 281 (89.1) 405 (87.0)
959 (18.0) 270 (18.5) 584 (13.7) 34 (10.9) 61 (13.0)
<0.001
Patient Location
Large metropolitan area Small metropolitan area Micropolitan/Other
4615 (83.8) 3100 (84.6) 2185 (83.0)
892 (16.2) 566 (15.4) 448 (17.0)
0.73
Patient Residence in State of Index Hospitalization
Nonresident Resident $1–$37,999 $38,000–$47,999 $48,000–$63,999 $64,000
1274 8651 2423 2695 2414 2235
216 (14.5) 1691 (16.4) 550 (18.5) 571 (17.5) 449 (15.7) 312 (12.2)
0.39
Length of stay, days (SEM)b Chronic conditions (SEM)b
Hospital length of stay Number of chronic conditions
7.3 (0.2) 5.1 (0.1)
8.9 (0.5) 5.8 (0.1)
<0.001 <0.001
Hospital size
Small Medium Large
647 (84.0) 1433 (84.2) 7844 (83.8)
123 (16.0) 269 (15.8) 1517 (16.2)
0.96
Hospital ownership
Government, nonfederal Private, not-profit Private, invest-own
1766 (82.9) 7522 (84.0) 636 (84.7)
363 (17.1) 1429 (16.0) 115 (15.3)
0.77
Hospital Location
Large metropolitan area Small metropolitan area Other
6084 (83.7) 3638 (84.0) 203 (86.7)
1186 (16.3) 691 (16.0) 31 (13.3)
0.81
Teaching status of Hospital
Metropolitan non-teaching Metropolitan teaching Non-metropolitan hospital
978 (87.9) 8743 (83.4) 203 (86.7)
134 (12.1) 1743 (16.6) 31 (13.3)
0.06
Median Household Income
(87.6) (84.9) (82.7) (82.5)
Yes N (%)
(85.5) (83.6) (81.5) (82.5) (84.3) (87.8)
(12.4) (15.1) (17.3) (17.5)
0.002
SNF, skilled nursing facility; AMA, against medical advice; SEM, standard error of mean. a Weighted national sample. b Weighted national means.
occur closer to hospital discharge are also more likely to be classified as avoidable readmissions [16]. Chin et al. found that overall hospital level variation in readmissions was highest on the first day after discharge and decreased until it reached its lowest point at seven days after discharge [17]. A shorter time frame for capturing readmissions, such as 7–10 days, could be a better representative of a hospital’s quality of surgical care. Evaluation of detailed patient-level comorbidities showed that valvular heart disease, rheumatoid arthritis or collagen vascular disease, liver disease, and hypothyroidism were independently associated with readmission. Given that most readmissions are for infectious and wound healing issues, these comorbidities become particularly important. Rheumatoid arthritis and collagen vascular disease are markers of long-term steroid use and chronic immunosuppression. This is consistent with a previous study which found long-term steroid use (OR, 7.78; 95% CI, 1.50–41.83) was associated with readmission in laryngectomy patients [12]. Hypothyroidism and liver disease can contribute to difficulties in wound healing [18,19]. There is a paucity of knowledge about
how to optimize post-operative care for these immunosuppressed and coagulopathic patients, who might be more susceptible to infection and wound breakdown. Our finding that diabetes and hypertension were not associated with readmission suggests that using an aggregate comorbidity index may not be as accurate in risk-adjusting patients. This is in contrast to prior studies in the otolaryngology literature, which have been limited to the Medicare population and have shown that increases in the aggregate Charlson comorbidity score is associated with readmission [13]. Insurance status was not associated with readmission, but patients in areas with lower median household incomes were more likely to be readmitted. This is an important factor to consider when risk-adjusting for a hospital’s patient population, as failure to do so would unfairly punish hospitals in low-income areas which typically have fewer resources. Carey et al. investigated the impact of HRRP on safety-net hospitals with a high proportion of low-income patients and found that safety-net hospitals were able to reduce their readmissions, but not to the extent of nonsafety-net hospitals that started with similar readmission rates
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112 Table 3 Multivariate logistic regression for risk factors associated with 30-day readmission. 30 Day Readmission OR (95% CI)
P Value
Site Oral Cavity Oropharynx Hypopharynx Larynx
1 [Reference] 2.17 (1.32–3.57) 2.76 (1.94–3.92) 2.07 (1.55–2.76)
0.002 <0.001 <0.001
Age group <50 50–59 60–69 >=70
1 [Reference] 1.04 (0.73–1.50) 1.14 (0.77–1.70) 1.21 (0.77–1.91)
0.83 0.51 0.40
Sex Male Female
1 [Reference] 0.92 (0.73–1.16)
0.47
Flap No Yes
1 [Reference] 1.32 (1.07–1.63)
0.009
Discharge location Home Home Health SNF/Other AMA
1 [Reference] 1.17 (0.90–1.51) 1.26 (0.91–1.73) 3.55 (0.85– 14.78)
0.23 0.16 0.08
Insurance Medicare Medicaid Private Insurance Self-Pay Other
1 [Reference] 1.04 (0.74–1.45) 0.91 (0.67–1.24) 0.60 (0.31–1.13) 0.77 (0.50–1.19)
Median Household Income (Quartiles) $64,000 $48,000–$63,999 $38,000–$47,999 $1–$37,999 Length of Stay Number of chronic conditions
1 [Reference] 1.34 (1.01–1.78) 1.50 (1.18–1.90) 1.54 (1.16–2.05) 1.00 (0.99–1.02) 1.01 (0.96–1.06)
0.04 0.001 0.003 0.55 0.70
Teaching status Metropolitan non-teaching Metropolitan teaching Non-metropolitan hospital
1 [Reference] 1.50 (1.04–2.16) 0.94 (0.48–1.84)
0.03 0.85
1.12 (0.78–1.63) 1.37 (0.88–2.14) 1.27 (0.97–1.66) 1.10 (0.88–1.39) 1.30 (1.02–1.66) 2.02 (1.37–2.99) 0.97 (0.78–1.21) 1.16 (0.88–1.53) 2.07 (1.16–3.69) 1.28 (0.86–1.90) 2.05 (1.27–3.31)
0.53 0.17 0.08 0.40 0.03 <0.001 0.78 0.28 0.01 0.22 0.003
Comorbidities Anemia Coagulopathy Diabetes Pulmonary Disease Hypothyroidism Liver Disease Fluid/Electrolyte Disorders Weight Loss Valvular Disease Peripheral Vascular Disease Rheumatoid Arthritis/Collagen Vascular Disease
0.83 0.55 0.11 0.24
111
non-index hospital readmission and these were associated with a 2.1-fold increased risk of in-hospital mortality [22]. Among general surgery patients, 25.0% of readmissions are to non-index hospitals and these admissions are associated with 48% higher odds of mortality than patients admitted to the index hospital [23]. Care fragmentation does not lead to higher costs, but it can potentially lead to increased mortality, delay in initiation of adjuvant therapy, and redundancy in care. Limitations to our study include coding and misclassification errors that are inherent to database research. In particular, we relied on this database to capture comorbidities, which required accurate diagnosis and coding by the hospital. We cannot validate that the coding of the primary diagnosis was the critical reason for the readmission. We cannot definitively exclude all oropharyngeal cancer patients who received a biopsy rather than definitive surgery based on procedure codes, however, we believe that the majority of our patients are receiving definitive surgery since they were all admitted post-operatively and the average length of stay was 5.1 days (SEM, 0.2 days). We cannot separate unplanned from planned readmissions. The database does not contain any cancerspecific variables therefore we cannot determine or adjust for clinical staging or pathologic characteristics. Teaching hospital status may be associated with readmission due to having more complex cases and salvage cases, however, the NRD does not include reliable data on history of prior head and neck chemoradiation. We do not have mortality data to know if lack of readmission was due to patient death. Conclusion This is the first national study that includes both readmissions at index and non-index hospitals to capture the true rate of readmission after head and neck cancer surgery in the United States. Readmissions are most commonly associated with infections and wound complications and occur soon after hospital discharge. Quality improvement initiatives targeting avoidable readmissions should concentrate on early detection and prevention of infection and wound complications. The use of 30-day all-cause unplanned readmission rates as a global metric for hospital performance should be reconsidered as a shorter time period (10 days) may better capture the majority of avoidable readmissions for surgical patients. Further research is needed to define what is considered an avoidable readmission for head and neck cancer surgery patients. Meetings A portion of this data was presented at the American Head and Neck Society 2017 Annual Meeting, San Diego, CA. Financial disclosures/conflicts of interest None declared.
[20]. Herrin et al. demonstrated that 58% of the national variation in hospital readmissions rates is explained by the county in which the hospital is located [21]. Risk-adjustment for socioeconomic status in readmission reduction programs can help account for this variability or perhaps safety-net hospitals should be exempted from readmission reduction program penalties. This is the first study to capture a true national 30-day readmission rate that encompasses both index and non-index hospital readmissions. Nearly 20% of readmitted head and neck cancer surgery patients have non-index hospital readmissions. Costs were similar between index and non-index readmissions. In head and neck cancer patients treated in California, there is a 37.4% rate of
References [1] Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009;360:1418–28. [2] Zuckerman RB, Sheingold SH, Orav EJ, Ruhter J, Epstein AM. Readmissions, observation, and the hospital readmissions reduction program. N Engl J Med 2016;374:1543–51. http://dx.doi.org/10.1056/NEJMsa1513024. [3] American Hospital Association. Rethinking the Hospital Readmssions Reduction Program. TrendWatch; 2015. [4] Rau J. Medicare’s Readmission Penalties Hit New High. Kais Health News; 2016.
112
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
[6] Graboyes EM, Liou T-N, Kallogjeri D, Nussenbaum B, Diaz JA. Risk factors for unplanned hospital readmission in otolaryngology patients. Otolaryngol – Head Neck Surg 2013;149:562–71. http://dx.doi.org/10.1177/ 0194599813500023. [7] Dziegielewski PT, Boyce B, Manning A, Agrawal A, Old M, Ozer E, et al. Predictors and costs of readmissions at an academic head and neck surgery service. Head Neck 2016;38:E502–10. http://dx.doi.org/10.1002/hed.24030. [8] Danino JF, Taylor T, Metcalfe CW, Muzaffar SJ, Sinha A. Readmission rates and financial penalties after ear, nose and throat surgery: how can we improve? Br J Hosp Med Lond Engl 2005;2015(76):655–7. http://dx.doi.org/10.12968/ hmed.2015.76.11.655. [9] Han S, Smith TS, Gunnar W. Descriptive analysis of 30-day readmission after inpatient surgery discharge in the veterans health administration. JAMA Surg 2014;149:1162. http://dx.doi.org/10.1001/jamasurg.2014.1706. [10] Rajasekaran K, Revenaugh P, Benninger M, Burkey B, Sindwani R. Development of a quality care plan to reduce otolaryngologic readmissions early lessons from the Cleveland clinic. Otolaryngol – Head Neck Surg 2015;153:629–35. http://dx.doi.org/10.1177/0194599815570025. [11] Bur AM, Brant JA, Mulvey CL, Nicolli EA, Brody RM, Fischer JP, et al. Association of clinical risk factors and postoperative complications with unplanned hospital readmission after head and neck cancer surgery. JAMA Otolaryngol Neck Surg 2016. http://dx.doi.org/10.1001/jamaoto.2016.2807. [12] 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 Neck Surg 2014;140:1157. http://dx.doi.org/ 10.1001/jamaoto.2014.1705. [13] Chaudhary H, Stewart CM, Webster K, Herbert RJ, Frick KD, Eisele DW, et al. Readmission following primary surgery for larynx and oropharynx cancer in the elderly. The Laryngoscope; 2016, n/a-n/a. http://dx.doi.org/10.1002/lary. 26311. [14] Tribius S, Donner J, Pazdyka H, Münscher A, Gröbe A, Petersen C, et al. Survival and overall treatment time after postoperative radio(chemo)therapy in patients with head and neck cancer. Head Neck 2016;38:1058–65. http://dx. doi.org/10.1002/hed.24407.
[15] Chen MM, Roman SA, Sosa JA, Judson BL. Postdischarge complications predict reoperation and mortality after otolaryngologic surgery. Otolaryngol – Head Neck Surg 2013:0194599813505078. http://dx.doi.org/10.1177/ 0194599813505078. [16] van Walraven C, Jennings A, Taljaard M, Dhalla I, English S, Mulpuru S, et al. Incidence of potentially avoidable urgent readmissions and their relation to all-cause urgent readmissions. Can Med Assoc J 2011;183:E1067–72. http:// dx.doi.org/10.1503/cmaj.110400. [17] Chin DL, Bang H, Manickam RN, Romano PS. Rethinking thirty-day hospital readmissions: shorter intervals might be better indicators of quality of care. Health Aff (Millwood) 2016;35:1867–75. http://dx.doi.org/10.1377/ hlthaff.2016.0205. [18] Zimmermann E, Ribas-Filho JM, Malafaia O, Ribas CAPM, Nassif PAN, Stieven Filho E, et al. Tracheal suture in rats with hypothyroidism: wound healing study. Acta Cir Bras 2009;24:282–9. [19] Karalis M, Pavlidis TE, Psarras K, Ballas K, Zaraboukas T, Rafailidis S, et al. Effect of experimentally induced liver cirrhosis on wound healing of the postextraction tooth socket in rats. Eur Surg Res 2007;40:190–6. http://dx.doi.org/ 10.1159/000110860. [20] Carey K, Lin M-Y. Hospital readmissions reduction program: safety-net hospitals show improvement, modifications to penalty formula still needed. Health Aff (Millwood) 2016;35:1918–23. http://dx.doi.org/10.1377/ hlthaff.2016.0537. [21] Herrin J, St. Andre J, Kenward K, Joshi MS, Audet A-MJ, Hines SC. Community factors and hospital readmission rates. Health Serv Res 2015;50:20–39. http:// dx.doi.org/10.1111/1475-6773.12177. [22] Graboyes EM, Kallogjeri D, Saeed MJ, Olsen MA, Nussenbaum B. Postoperative care fragmentation and thirty-day unplanned readmissions after head and neck cancer surgery. The Laryngoscope; 2016:n/a-n/a. http://dx.doi.org/10. 1002/lary.26301. [23] Tsai TC, Orav EJ, Jha AK. Care fragmentation in the postdischarge period: surgical readmissions, distance of travel, and postoperative mortality. JAMA Surg 2015;150:59. http://dx.doi.org/10.1001/jamasurg.2014.2071.
Contents lists available at ScienceDirect
Oral Oncology journal homepage: www.elsevier.com/locate/oraloncology
Predictors of readmissions after head and neck cancer surgery: A national perspective Michelle M. Chen a, Ryan K. Orosco a, Jeremy P. Harris b, Julie B. Porter c, Eben L. Rosenthal a, Wendy Hara b, Vasu Divi a,⇑ a b c
Department of Otolaryngology-Head and Neck Surgery, Stanford University, 900 Blake Wilbur Drive, Third Floor, Stanford, CA 94305, USA Department of Radiation Oncology, Stanford University, 875 Blake Wilbur Drive, Rm CCG210 Clinic D, Stanford, CA 94305, USA Stanford Health Care, 875 Blake Wilbur Drive, Stanford, CA 94305, USA
a r t i c l e
i n f o
Article history: Received 14 February 2017 Received in revised form 14 June 2017 Accepted 15 June 2017
Keywords: Head and neck cancer Hospital readmission Patient readmission Patient discharge Patient care Laryngeal cancer Oropharyngeal cancer Oral cavity cancer Hypopharyngeal cancer
a b s t r a c t Objectives: Surgical oncology patients have multiple comorbidities and are at high risk of readmission. Prior studies are limited in their ability to capture readmissions outside of the index hospital that performed the surgery. Our goal is to evaluate risk factors for readmission for head and neck cancer patients on a national scale. Material and methods: A retrospective cohort study of head and neck cancer patients in the Nationwide Readmissions Database (2013). Our main outcome was 30-day readmission. Statistical analysis included 2-sided t tests, v2, and multivariate logistic regression analysis. Results: Within 30 days, 16.1% of 11,832 patients were readmitted and 20% of readmissions were at nonindex hospitals, costing $31 million. Hypopharyngeal cancer patients had the highest readmission rate (29.6%), followed by laryngeal (21.8%), oropharyngeal (18.2%), and oral cavity (11.6%) cancers (P < 0.001). Half of readmissions occurred within 10 days and were often associated with infections (27%) or wound complications (12%). Patients from lower household income areas were more likely to be readmitted (odds ratio [OR], 1.54; 95% confidence interval [CI], 1.16–2.05). Patients with valvular disease (OR, 2.07; 95% CI, 1.16–3.69), rheumatoid arthritis/collagen vascular disease (OR, 2.05; 95% CI, 1.27–3.31), liver disease (OR, 2.02, 95% CI, 1.37–2.99), and hypothyroidism (OR 1.30; 95% CI, 1.02–1.66) were at highest risk of readmission. Conclusion: The true rate of 30-day readmissions after head and neck cancer surgery is 16%, capturing non-index hospital readmissions which make up 20% of readmissions. Readmissions after head and neck cancer surgery are most commonly associated with infections and wound complications. Ó 2017 Elsevier Ltd. All rights reserved.
Introduction Hospital readmissions cost Medicare alone more than $17 billion annually in avoidable expenses, with over 17.5% of patients getting readmitted within 30-days of hospital discharge [1–3]. Only 10% of these Medicare readmissions are considered planned readmissions [1]. In order to encourage hospitals to reduce readmissions, the Affordable Care Act, which was passed in March 2010, created the Hospital Readmissions Reduction Program. Since October 2012, the Centers for Medicare and Medicaid Services (CMS) has levied a penalty on hospitals with readmissions in excess of the expected national rate. For fiscal year 2017, the penal-
⇑ Corresponding author. E-mail address: [email protected] (V. Divi). http://dx.doi.org/10.1016/j.oraloncology.2017.06.010 1368-8375/Ó 2017 Elsevier Ltd. All rights reserved.
ties are expected to total $528 million, which is $108 million more than the prior year [4]. The current program targets six medical conditions. Initially, it targeted acute myocardial infarctions, heart failure, and pneumonia. In 2015, total knee and hip surgery and chronic obstructive pulmonary disease were added; and in 2017, coronary artery bypass surgery was added [2,3]. Preliminary data since the introduction of the program has shown a reduction in the rates of avoidable readmissions, making the expansion of this program increasingly likely [2]. Oncology patients in particular have high rates of readmission following discharge, making them a possible target of the program. The concept of avoidable admissions in also present in the CMS Oncology Care Model, piloted on July 1, 2016, which seeks to capture the rates of emergency department visits and readmissions during chemotherapy.
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
Given the unique challenges faced by oncology patients, and head and neck cancer surgery patients in particular, there is a need to understand the risk factors for readmission so interventions can be effectively targeted and hospital benchmarks can be appropriately risk-adjusted. In addition, it is important for providers to understand the causes of these readmissions to help design programs that reduce avoidable readmissions. Prior studies on readmissions in otolaryngology are primarily institutional studies that are unable to capture readmissions that occur outside of a single institution. The purpose of our study was to evaluate the causes of and risk factors for readmission in head and neck cancer surgery patients utilizing a national database that covers readmissions at both the index hospital and outside facilities. This database also captures socioeconomic and comorbidity data to allow for evaluation of patient-level characteristics that influence readmission.
Materials and methods Our study was granted an exemption from our institutional review board. Patient data was drawn from the Healthcare Cost and Utilization Project’s Nationwide Readmissions Database (NRD) from January 1, 2013 to December 31, 2013. The NRD is a database of all-payer hospital inpatient stays from 21 states that accounts for 49.1% of all United States hospitalizations [5]. This stratified sample can be used to produce nationally weighted estimates. In 2013, the NRD covered 14 million discharges, resulting in a nationally weighted sample of 36 million discharges. All adult patients (age 18) with a primary diagnosis of head and neck cancer were chosen based on their International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes for oral cavity cancer (141.1–141.4, 141.8–141.9, 143.0–145.9), oropharyngeal cancer (141.0–141.6, 146.0–146.9), hyopharyngeal cancer (148.0–148.9), and laryngeal cancer (161.0–161.9). We limited our cohort to those who had surgery for the primary site based on their ICD-9-CM procedure codes. We also limited our index hospitalizations from January 1, 2013 to November 30, 2013, in order to ensure we were able to capture all 30-day readmissions. We excluded patients that died during the index hospitalization and any records that were combined transfer records that combined multiple hospitalizations into one. Demographic and socioeconomic status variables included age, sex, patient location, state residence, insurance status, household income, comorbidities, and number of chronic conditions. Age was grouped into four categories (18–49, 50–59, 60–69, and 70 years of age). Patient location was grouped into large metropolitan areas (populations 1 million), small metropolitan areas (populations from 50,000 to 999,999), and micropolitan and other areas. State residence was a binary characteristic that depended on whether the patient was a resident of the state of the index hospitalization. Insurance status was categorized as Medicare, Medicaid, private insurance, self-pay, and other (including no charge). Income was the median household income for the patient’s ZIP code and grouped into quartiles ($1–$37,999; $38,0 00–$47,999; $48,000–$63,999; and $64,000 or more). Comorbidity measures were identified by the Agency for Healthcare Research and Quality’s comorbidity software that identifies coexisting medication conditions that are likely present prior to hospital admission. The comorbidity measures included alcohol abuse, anemia (chronic blood loss and deficiency anemias), congestive heart failure, coagulopathy, depression, diabetes (complicated and uncomplicated), pulmonary disease (chronic lung disease and pulmonary circulation disorders), drug abuse, hypertension (complicated and uncomplicated), hypothyroidism, liver disease, fluid and electrolyte disorders, obesity, renal failure, weight loss, valvular heart disease, peripheral vascular disease, rheumatoid arthri-
107
tis/collagen vascular disease, other neurological disorders, and psychoses. Tobacco use was a binary characteristic created for any patient with an ICD-9-CM diagnosis code of tobacco use. Clinical variables for index hospitalization included tumor site, use of a flap during the index surgery, hospital length of stay, discharge location, costs, and primary diagnosis on readmission. Tumor site was classified as oral cavity, oropharynx, hypopharynx, and larynx. Discharge location was grouped into home, home with home health, discharge to skilled nursing facility (SNF) or other facilities, and leaving against medical advice (AMA). Total costs were determined by total charges and hospital-specific cost to charge ratios. Primary diagnosis on readmission was based on ICD-9-CM diagnosis codes and grouped into twelve categories (infection, other medical, head and neck cancer, other head and neck, wound complication, hematoma/hemorrhage, tracheostomy complication, electrolyte/nutrition, gastrostomy complication, pulmonary embolism/deep venous thromboembolism/stroke, chemotherapy/radiotherapy, or pain). ‘‘Other medical” encompassed diagnoses such as cardiovascular (heart failure, atrial fibrillation, chest pain, cardiac arrest, coronary artery disease), pulmonary (dyspnea), psychologic (anxiety, depression), renal, gastrointestinal, and genitourinary disorders. ‘‘Other head and neck” included diagnosis codes referring to edema in the head and neck, laryngeal stenosis, vocal fold paralysis, foreign bodies in the head and neck, and other diseases of the head and neck not classified elsewhere. Infections of tracheostomy and gastrostomy sites were grouped under the tracheostomy and gastrostomy complication categories rather than under infection. Hospital level variables included hospital size, ownership, location, and teaching status. Hospital size is grouped into small, medium, and large based on bedside and hospital location with larger cut-offs for hospitals in urban locations. Hospital ownership was categorized into government, private non-profit, and private investor-owned hospitals. Hospital location was classified as large metropolitan area (1 million residents), small metropolitan area (<1 million residents), and other (includes micropolitan and non-urban areas). Teaching status was divided into metropolitan non-teaching hospitals, metropolitan teaching hospitals, and non-metropolitan hospitals. Our primary outcome of interest was 30-day readmission defined as any hospital admission that occurred within 30 days of discharge from the index hospitalization. All statistical analysis was performed using STATA/SE software (version 14.2; StataCorp, College Station, TX, USA). Bivariate analysis using chi-square and t tests were used to analyze our categorical and continuous variables, respectively. Multivariate logistic regression analysis was used to identify factors associated with 30-day readmission. Odds ratios (OR) and 95% confidence intervals (95% CIs) were calculated for the strength of association. All tests were 2-sided, and a P value < 0.05 was considered to be statistically significant.
Results We identified a weighted total of 11,832 patients who had primary surgery for oral cavity, oropharyngeal, hypopharyngeal, and laryngeal cancer with a 30-day readmissions rate of 16.1%. Over half of these patients were readmitted through the Emergency Department and 20% were readmitted to a different hospital (Table 1). Just over half (51%) of initial 30-day readmissions occurred within 10 days of hospital discharge (see Fig. 1). Other than a diagnosis of head and neck cancer, the most common primary diagnosis on 30-day readmission was an infectious etiology (23.5%) (Fig. 2). In the first 10 days after hospital discharge, nearly half of readmissions were infectious (27.2%), wound-related (11.9%), and
108
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
Table 1 30-day readmissions for head and neck cancer surgery patients, national weighted sample. Totala
Readmissionsa
Emergency department readmissionsa
Mean proportion of readmissions at index hospitalb
Site
N
N
% of Total
N
% of Readmissions
% of Readmissions (SE)
$ (SE)
Oral Cavity Oropharynx Hypopharynx Larynx Total
6183 2090 452 3108 11832
717 380 134 678 1908
11.6 18.2 29.6 21.8 16.1
417 128 67 362 974
58.2 33.7 50.3 53.4 51.1
73.1 90.1 78.9 82.8 80.3
13,587 10,609 15,795 18,445 14,895
(3.5) (3.0) (6.6) (2.3) (2.2)
Mean cost per readmissionb
(1097) (834) (2031) (1815) (926)
SE, standard error. a National weighted totals. b National weighted means with linearized standard errors.
Fig. 1. Over half of head and neck cancer surgery patients are readmitted within the first 10 days after discharge.
tracheostomy (5.6%) or gastrostomy tube-associated (3.0%). For patients readmitted after 10 days, the most common diagnoses on readmission were other medical diagnoses (19.9%) or a head and neck cancer diagnosis (28.6%). For admissions occurring from January to November 2013, readmissions accounted for $31.1 million in total cost. The mean cost per readmission was $14,895 (SEM, $926), which was 44.9% of the average cost of the index hospitalization. The readmission cost was highest for laryngeal cancer patients (mean cost, $18,445) and lowest for oropharyngeal cancer patients (mean cost, $10,609). There was no difference in cost between readmission at the initial hospital and at a non-index hospital (P = 0.43). Patients with laryngeal and hypopharyngeal primaries had the highest rate of readmission at 21.8% and 29.6%, respectively; while patients with oropharyngeal cancer (18.2%) and oral cavity cancer (11.6%) had significantly lower readmission rates (P < 0.001). There was no difference in likelihood of readmission across age groups (P = 0.06), but Medicare (18.0%) and Medicaid patients (18.5%) were more likely to be readmitted than patients with private insurance (13.7%) or who self-paid for their care (10.9%) (P < 0.001) (Table 2). Compared with patients discharged home, patients dis-
charged to skilled nursing facilities and other facilities were more likely to be readmitted within 30 days (13.3% vs 22.3%, P < 0.001). Readmitted patients had longer index hospitalizations and more chronic conditions. On multivariate analysis, tumor site, a flap during the index procedure, lower median household incomes, and teaching hospital status were independently associated with increased odds of readmission (Table 3). Of the 21 comorbidities analyzed, valvular heart disease (odds ratio [OR], 2.07; 95% confidence interval [CI], 1.16–3.69), rheumatoid arthritis or collagen vascular disease (OR, 2.05; 95% CI, 1.27–3.31), liver disease (OR, 2.02, 95% CI, 1.37–2.99), and hypothyroidism (OR 1.30; 95% CI, 1.02–1.66) were independent risk factors for readmission. Hospital size, ownership, and location were not associated with readmission. The only hospital level factor associated with readmission was teaching status. Relative to non-teaching hospitals, metropolitan teaching hospitals were associated with increased odds of readmission (OR, 1.50; 95% CI, 1.04–2.16). We analyzed a subset of patients with only one 30-day readmission in order to investigate factors associated with index hospital readmission. Larger index hospitals (medium hospital: OR, 3.20; 95% CI, 1.23–8.32; large hospital: OR, 3.72; 95% CI, 1.75–7.92) were
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
109
Fig. 2. For all 30-day readmissions, infection (23.5%) was the most common primary diagnosis on readmission. For early readmissions (10 days after discharge), infection continued to be the most common diagnosis on readmission (27.2%), but for late readmissions (11–30 days after discharge), head and neck cancer (28.6%) and other medical diagnoses (19.9%) were more common. PE, pulmonary embolism; DVT, deep venous thromboembolism.
more likely to have patients return upon readmission than small index hospitals. Patients readmitted to non-index hospitals were more likely to be readmitted from the emergency department (OR, 0.21; 95% CI, 0.12–0.35). There was no difference in index hospital readmission between patients discharged home and those who were discharged to skilled nursing facilities. Discussion This study utilizes a national inpatient discharge database to provide national weighted estimates of readmissions after head and neck cancer surgery. We demonstrate that there is a 16.1% national 30-day readmission rate and a fifth of the readmissions were not at the index hospital. The total cost of readmissions for head and neck cancer surgery patients over 11 months was $31 million. Surgical site, flaps, areas with lower median household incomes, and hospital type were associated with 30-day readmission. Over half of all initial readmissions occurred within 10 days of discharge and early readmissions had a higher rate of infections and wound complications than late readmissions. Prior studies on otolaryngologic readmissions have been primarily single institution studies, state-specific studies, or focused on otolaryngologic procedures in aggregate. While readmissions after otolaryngologic procedures in aggregate have been shown to range from 3.1 to 9.5% [6–11], cancer patients represent a complex subset of patients who tend to be more frequently readmitted. Evaluating 155 total laryngectomy patients treated at one institution, Graboyes et al. found a 26.5% 30-day readmission rate [12]. Chaudhary et al. examined 1518 elderly oropharyngeal and laryngeal cancer Medicare patients and 14.1% were readmitted within 30 days of hospital discharge [13]. Similarly, our study reports a 16.1% overall readmission rate with the highest readmission rates for patients with laryngeal (21.8%) and hypopharyngeal (29.6%) cancer. Head and neck cancer patients are complex and have a higher rate of unavoidable readmissions than other otolaryngology patients. Readmissions are particularly important in this patient population since they can delay initiation of adjuvant therapy and thus potentially lead to decreased survival [14].
We found that each 30-day readmission costs $14,895 on average for all-payers, nearly half of the cost of the initial surgical hospitalization. This corroborates prior data by Chaudhary et al. on Medicare patients that demonstrated oropharyngeal and laryngeal cancer patients who were readmitted had mean costs that were $15,123 greater than those who were not readmitted [13]. Readmissions are a small, but significant proportion of healthcare costs for head and neck cancer surgery. Over half of patients were readmitted within 10 days of discharge. Of these, 27% had a primary diagnosis of infection and 12% had a diagnosis of a wound complication. This supports findings by Bur et al. who demonstrated that causes of unplanned readmissions after head and neck surgery were mostly infectious, including wound infection or breakdown (24%) and pneumonia (6.4%). However this study was limited because it looked at readmissions starting from 30 days of the surgical date instead of the discharge date, which may underestimate the actual 30-day readmissions for complex head and neck patients with prolonged hospital stays [11]. These admissions may present a unique opportunity for quality improvement. More aggressive assessment of potential infections and wound complications prior to discharge may help identify patients who require further treatment and close follow-up. This may be done with a combination of improved caregiver and patient education on early signs of infection or wound complications, post-discharge phone calls with targeted questions, and close follow-up in outpatient clinics. In our cohort, while infection was the most common diagnosis for early readmissions (10 days after discharge), other medical diagnoses and head and neck cancer diagnoses were more common for readmissions after 10 days. This suggests that perhaps a shorter time frame for readmissions may better capture related and avoidable readmissions. The 30-day readmission metric was initially chosen as a quality metric for both medical and surgical patients since readmissions after 30-days were less likely to be due to the acute care received in the hospital. It has been shown previously that half of all post-discharge complications in otolaryngology occur soon after discharge [15]. Readmissions that
110
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
Table 2 Baseline characteristics for head and neck cancer surgery patients by 30-day readmission status. 30-day readmissiona No N (%)
P value
177 518 633 579
0.06
Age
18–49 50–59 60–69 70+
1250 2909 3032 2733
Sex
Male Female
6768 (83.0) 3156 (85.8)
1386 (17.0) 522 (14.2)
0.03
Flap
No Yes
7654 (84.6) 2270 (81.6)
1395 (15.4) 513 (18.4)
0.02
Discharge Location
Home Home Health SNF/Other AMA
5724 (86.7) 3153 (81.3) 1029 (77.7) 19
877 (13.3) 725 (18.7) 295 (22.3) 10
<0.001
Insurance
Medicare Medicaid Private Self-Pay Other
4366 (82.0) 1191 (81.5) 3669 (86.3) 281 (89.1) 405 (87.0)
959 (18.0) 270 (18.5) 584 (13.7) 34 (10.9) 61 (13.0)
<0.001
Patient Location
Large metropolitan area Small metropolitan area Micropolitan/Other
4615 (83.8) 3100 (84.6) 2185 (83.0)
892 (16.2) 566 (15.4) 448 (17.0)
0.73
Patient Residence in State of Index Hospitalization
Nonresident Resident $1–$37,999 $38,000–$47,999 $48,000–$63,999 $64,000
1274 8651 2423 2695 2414 2235
216 (14.5) 1691 (16.4) 550 (18.5) 571 (17.5) 449 (15.7) 312 (12.2)
0.39
Length of stay, days (SEM)b Chronic conditions (SEM)b
Hospital length of stay Number of chronic conditions
7.3 (0.2) 5.1 (0.1)
8.9 (0.5) 5.8 (0.1)
<0.001 <0.001
Hospital size
Small Medium Large
647 (84.0) 1433 (84.2) 7844 (83.8)
123 (16.0) 269 (15.8) 1517 (16.2)
0.96
Hospital ownership
Government, nonfederal Private, not-profit Private, invest-own
1766 (82.9) 7522 (84.0) 636 (84.7)
363 (17.1) 1429 (16.0) 115 (15.3)
0.77
Hospital Location
Large metropolitan area Small metropolitan area Other
6084 (83.7) 3638 (84.0) 203 (86.7)
1186 (16.3) 691 (16.0) 31 (13.3)
0.81
Teaching status of Hospital
Metropolitan non-teaching Metropolitan teaching Non-metropolitan hospital
978 (87.9) 8743 (83.4) 203 (86.7)
134 (12.1) 1743 (16.6) 31 (13.3)
0.06
Median Household Income
(87.6) (84.9) (82.7) (82.5)
Yes N (%)
(85.5) (83.6) (81.5) (82.5) (84.3) (87.8)
(12.4) (15.1) (17.3) (17.5)
0.002
SNF, skilled nursing facility; AMA, against medical advice; SEM, standard error of mean. a Weighted national sample. b Weighted national means.
occur closer to hospital discharge are also more likely to be classified as avoidable readmissions [16]. Chin et al. found that overall hospital level variation in readmissions was highest on the first day after discharge and decreased until it reached its lowest point at seven days after discharge [17]. A shorter time frame for capturing readmissions, such as 7–10 days, could be a better representative of a hospital’s quality of surgical care. Evaluation of detailed patient-level comorbidities showed that valvular heart disease, rheumatoid arthritis or collagen vascular disease, liver disease, and hypothyroidism were independently associated with readmission. Given that most readmissions are for infectious and wound healing issues, these comorbidities become particularly important. Rheumatoid arthritis and collagen vascular disease are markers of long-term steroid use and chronic immunosuppression. This is consistent with a previous study which found long-term steroid use (OR, 7.78; 95% CI, 1.50–41.83) was associated with readmission in laryngectomy patients [12]. Hypothyroidism and liver disease can contribute to difficulties in wound healing [18,19]. There is a paucity of knowledge about
how to optimize post-operative care for these immunosuppressed and coagulopathic patients, who might be more susceptible to infection and wound breakdown. Our finding that diabetes and hypertension were not associated with readmission suggests that using an aggregate comorbidity index may not be as accurate in risk-adjusting patients. This is in contrast to prior studies in the otolaryngology literature, which have been limited to the Medicare population and have shown that increases in the aggregate Charlson comorbidity score is associated with readmission [13]. Insurance status was not associated with readmission, but patients in areas with lower median household incomes were more likely to be readmitted. This is an important factor to consider when risk-adjusting for a hospital’s patient population, as failure to do so would unfairly punish hospitals in low-income areas which typically have fewer resources. Carey et al. investigated the impact of HRRP on safety-net hospitals with a high proportion of low-income patients and found that safety-net hospitals were able to reduce their readmissions, but not to the extent of nonsafety-net hospitals that started with similar readmission rates
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112 Table 3 Multivariate logistic regression for risk factors associated with 30-day readmission. 30 Day Readmission OR (95% CI)
P Value
Site Oral Cavity Oropharynx Hypopharynx Larynx
1 [Reference] 2.17 (1.32–3.57) 2.76 (1.94–3.92) 2.07 (1.55–2.76)
0.002 <0.001 <0.001
Age group <50 50–59 60–69 >=70
1 [Reference] 1.04 (0.73–1.50) 1.14 (0.77–1.70) 1.21 (0.77–1.91)
0.83 0.51 0.40
Sex Male Female
1 [Reference] 0.92 (0.73–1.16)
0.47
Flap No Yes
1 [Reference] 1.32 (1.07–1.63)
0.009
Discharge location Home Home Health SNF/Other AMA
1 [Reference] 1.17 (0.90–1.51) 1.26 (0.91–1.73) 3.55 (0.85– 14.78)
0.23 0.16 0.08
Insurance Medicare Medicaid Private Insurance Self-Pay Other
1 [Reference] 1.04 (0.74–1.45) 0.91 (0.67–1.24) 0.60 (0.31–1.13) 0.77 (0.50–1.19)
Median Household Income (Quartiles) $64,000 $48,000–$63,999 $38,000–$47,999 $1–$37,999 Length of Stay Number of chronic conditions
1 [Reference] 1.34 (1.01–1.78) 1.50 (1.18–1.90) 1.54 (1.16–2.05) 1.00 (0.99–1.02) 1.01 (0.96–1.06)
0.04 0.001 0.003 0.55 0.70
Teaching status Metropolitan non-teaching Metropolitan teaching Non-metropolitan hospital
1 [Reference] 1.50 (1.04–2.16) 0.94 (0.48–1.84)
0.03 0.85
1.12 (0.78–1.63) 1.37 (0.88–2.14) 1.27 (0.97–1.66) 1.10 (0.88–1.39) 1.30 (1.02–1.66) 2.02 (1.37–2.99) 0.97 (0.78–1.21) 1.16 (0.88–1.53) 2.07 (1.16–3.69) 1.28 (0.86–1.90) 2.05 (1.27–3.31)
0.53 0.17 0.08 0.40 0.03 <0.001 0.78 0.28 0.01 0.22 0.003
Comorbidities Anemia Coagulopathy Diabetes Pulmonary Disease Hypothyroidism Liver Disease Fluid/Electrolyte Disorders Weight Loss Valvular Disease Peripheral Vascular Disease Rheumatoid Arthritis/Collagen Vascular Disease
0.83 0.55 0.11 0.24
111
non-index hospital readmission and these were associated with a 2.1-fold increased risk of in-hospital mortality [22]. Among general surgery patients, 25.0% of readmissions are to non-index hospitals and these admissions are associated with 48% higher odds of mortality than patients admitted to the index hospital [23]. Care fragmentation does not lead to higher costs, but it can potentially lead to increased mortality, delay in initiation of adjuvant therapy, and redundancy in care. Limitations to our study include coding and misclassification errors that are inherent to database research. In particular, we relied on this database to capture comorbidities, which required accurate diagnosis and coding by the hospital. We cannot validate that the coding of the primary diagnosis was the critical reason for the readmission. We cannot definitively exclude all oropharyngeal cancer patients who received a biopsy rather than definitive surgery based on procedure codes, however, we believe that the majority of our patients are receiving definitive surgery since they were all admitted post-operatively and the average length of stay was 5.1 days (SEM, 0.2 days). We cannot separate unplanned from planned readmissions. The database does not contain any cancerspecific variables therefore we cannot determine or adjust for clinical staging or pathologic characteristics. Teaching hospital status may be associated with readmission due to having more complex cases and salvage cases, however, the NRD does not include reliable data on history of prior head and neck chemoradiation. We do not have mortality data to know if lack of readmission was due to patient death. Conclusion This is the first national study that includes both readmissions at index and non-index hospitals to capture the true rate of readmission after head and neck cancer surgery in the United States. Readmissions are most commonly associated with infections and wound complications and occur soon after hospital discharge. Quality improvement initiatives targeting avoidable readmissions should concentrate on early detection and prevention of infection and wound complications. The use of 30-day all-cause unplanned readmission rates as a global metric for hospital performance should be reconsidered as a shorter time period (10 days) may better capture the majority of avoidable readmissions for surgical patients. Further research is needed to define what is considered an avoidable readmission for head and neck cancer surgery patients. Meetings A portion of this data was presented at the American Head and Neck Society 2017 Annual Meeting, San Diego, CA. Financial disclosures/conflicts of interest None declared.
[20]. Herrin et al. demonstrated that 58% of the national variation in hospital readmissions rates is explained by the county in which the hospital is located [21]. Risk-adjustment for socioeconomic status in readmission reduction programs can help account for this variability or perhaps safety-net hospitals should be exempted from readmission reduction program penalties. This is the first study to capture a true national 30-day readmission rate that encompasses both index and non-index hospital readmissions. Nearly 20% of readmitted head and neck cancer surgery patients have non-index hospital readmissions. Costs were similar between index and non-index readmissions. In head and neck cancer patients treated in California, there is a 37.4% rate of
References [1] Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med 2009;360:1418–28. [2] Zuckerman RB, Sheingold SH, Orav EJ, Ruhter J, Epstein AM. Readmissions, observation, and the hospital readmissions reduction program. N Engl J Med 2016;374:1543–51. http://dx.doi.org/10.1056/NEJMsa1513024. [3] American Hospital Association. Rethinking the Hospital Readmssions Reduction Program. TrendWatch; 2015. [4] Rau J. Medicare’s Readmission Penalties Hit New High. Kais Health News; 2016.
112
M.M. Chen et al. / Oral Oncology 71 (2017) 106–112
[6] Graboyes EM, Liou T-N, Kallogjeri D, Nussenbaum B, Diaz JA. Risk factors for unplanned hospital readmission in otolaryngology patients. Otolaryngol – Head Neck Surg 2013;149:562–71. http://dx.doi.org/10.1177/ 0194599813500023. [7] Dziegielewski PT, Boyce B, Manning A, Agrawal A, Old M, Ozer E, et al. Predictors and costs of readmissions at an academic head and neck surgery service. Head Neck 2016;38:E502–10. http://dx.doi.org/10.1002/hed.24030. [8] Danino JF, Taylor T, Metcalfe CW, Muzaffar SJ, Sinha A. Readmission rates and financial penalties after ear, nose and throat surgery: how can we improve? Br J Hosp Med Lond Engl 2005;2015(76):655–7. http://dx.doi.org/10.12968/ hmed.2015.76.11.655. [9] Han S, Smith TS, Gunnar W. Descriptive analysis of 30-day readmission after inpatient surgery discharge in the veterans health administration. JAMA Surg 2014;149:1162. http://dx.doi.org/10.1001/jamasurg.2014.1706. [10] Rajasekaran K, Revenaugh P, Benninger M, Burkey B, Sindwani R. Development of a quality care plan to reduce otolaryngologic readmissions early lessons from the Cleveland clinic. Otolaryngol – Head Neck Surg 2015;153:629–35. http://dx.doi.org/10.1177/0194599815570025. [11] Bur AM, Brant JA, Mulvey CL, Nicolli EA, Brody RM, Fischer JP, et al. Association of clinical risk factors and postoperative complications with unplanned hospital readmission after head and neck cancer surgery. JAMA Otolaryngol Neck Surg 2016. http://dx.doi.org/10.1001/jamaoto.2016.2807. [12] 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 Neck Surg 2014;140:1157. http://dx.doi.org/ 10.1001/jamaoto.2014.1705. [13] Chaudhary H, Stewart CM, Webster K, Herbert RJ, Frick KD, Eisele DW, et al. Readmission following primary surgery for larynx and oropharynx cancer in the elderly. The Laryngoscope; 2016, n/a-n/a. http://dx.doi.org/10.1002/lary. 26311. [14] Tribius S, Donner J, Pazdyka H, Münscher A, Gröbe A, Petersen C, et al. Survival and overall treatment time after postoperative radio(chemo)therapy in patients with head and neck cancer. Head Neck 2016;38:1058–65. http://dx. doi.org/10.1002/hed.24407.
[15] Chen MM, Roman SA, Sosa JA, Judson BL. Postdischarge complications predict reoperation and mortality after otolaryngologic surgery. Otolaryngol – Head Neck Surg 2013:0194599813505078. http://dx.doi.org/10.1177/ 0194599813505078. [16] van Walraven C, Jennings A, Taljaard M, Dhalla I, English S, Mulpuru S, et al. Incidence of potentially avoidable urgent readmissions and their relation to all-cause urgent readmissions. Can Med Assoc J 2011;183:E1067–72. http:// dx.doi.org/10.1503/cmaj.110400. [17] Chin DL, Bang H, Manickam RN, Romano PS. Rethinking thirty-day hospital readmissions: shorter intervals might be better indicators of quality of care. Health Aff (Millwood) 2016;35:1867–75. http://dx.doi.org/10.1377/ hlthaff.2016.0205. [18] Zimmermann E, Ribas-Filho JM, Malafaia O, Ribas CAPM, Nassif PAN, Stieven Filho E, et al. Tracheal suture in rats with hypothyroidism: wound healing study. Acta Cir Bras 2009;24:282–9. [19] Karalis M, Pavlidis TE, Psarras K, Ballas K, Zaraboukas T, Rafailidis S, et al. Effect of experimentally induced liver cirrhosis on wound healing of the postextraction tooth socket in rats. Eur Surg Res 2007;40:190–6. http://dx.doi.org/ 10.1159/000110860. [20] Carey K, Lin M-Y. Hospital readmissions reduction program: safety-net hospitals show improvement, modifications to penalty formula still needed. Health Aff (Millwood) 2016;35:1918–23. http://dx.doi.org/10.1377/ hlthaff.2016.0537. [21] Herrin J, St. Andre J, Kenward K, Joshi MS, Audet A-MJ, Hines SC. Community factors and hospital readmission rates. Health Serv Res 2015;50:20–39. http:// dx.doi.org/10.1111/1475-6773.12177. [22] Graboyes EM, Kallogjeri D, Saeed MJ, Olsen MA, Nussenbaum B. Postoperative care fragmentation and thirty-day unplanned readmissions after head and neck cancer surgery. The Laryngoscope; 2016:n/a-n/a. http://dx.doi.org/10. 1002/lary.26301. [23] Tsai TC, Orav EJ, Jha AK. Care fragmentation in the postdischarge period: surgical readmissions, distance of travel, and postoperative mortality. JAMA Surg 2015;150:59. http://dx.doi.org/10.1001/jamasurg.2014.2071.