Offodile AC, 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
Hospital readmissions are a marker of surgical care delivery and quality that are progressively more scrutinized.
To provide a comprehensive analysis of 30-day readmissions for patients with head and neck cancer who underwent free flap reconstruction to highlight the rate, causes, and associated patient risk factors.
Design, Setting, and Participants
Retrospective cohort study at a single tertiary care academic institution. The study consisted of 249 patients who underwent microvascular reconstruction of a presumed head and neck oncologic defect from January 1, 2000, through June 30, 2014. Follow-up continued through July 30, 2014.
Microvascular reconstruction of an oncologic head and neck defect.
Main Outcomes and Measures
Incidence of 30-day all-cause readmissions, patient risk factors, and readmission indications. Regression analyses were conducted to discern patient-level risk factors related to 30-day readmissions.
Among the 249 patients, the 30-day all-cause readmission rate was 14.5%, while the unplanned readmission rate was 11.6%. The most common reason for readmission was neck wound complications. Predictors of readmission following multivariable analysis were T4 pathologic stage (odds ratio [OR], 11.68; 95% CI, 1.37-99.81; P = .02) and having a tumor located in the oropharynx (OR, 4.64; 95% CI, 1.89-11.38; P = .001), hypopharynx (OR, 8.30; 95% CI, 1.52-45.24; P = .01), or larynx (OR, 10.97; 95% CI, 2.27-52.98; P = .003). Patients who were readmitted were more likely to experience neck wound complications (OR, 5.07; 95% CI, 1.31-19.57; P = .02) and undergo reoperation (OR, 47.20; 95% CI, 8.33-267.33; P < .001).
Conclusions and Relevance
In this study, advanced pathologic tumor staging and tumor location were associated with 30-day readmissions in patients with head and neck cancer who receive free flaps. Our results provide a benchmark for risk stratification that can be used in system-based practice improvements, health care cost savings, and postoperative patient counseling.
Health care systems today have embraced early readmissions, defined as admission within 30 days of discharge, as a valid metric for measuring health care quality.1 Prior research2 has established the deleterious effect of hospital readmissions on postoperative quality of life. Quiz Ref IDThe Center for Medicare & Medicaid Services embarked on public reporting of individual hospital readmission rates in 2009 and instituted financial penalties in 2013 for hospitals with 30-day readmissions that were associated with heart attacks, pneumonia, and heart failure.3,4 The expectation is that this heightened awareness of hospital readmissions as a key element of quality improvement will affect surgical departments in the near future. In fact, in 2015, the Hospital Readmissions Reduction Program of the Affordable Care Act will expand the aforementioned list of reportable diagnoses to include surgical procedures.5,6 All of these diagnoses have yielded a sizeable increase in research efforts aimed at identifying risk factors for early readmissions across a breadth of surgical populations.3,5- 12
Prior work8 evaluated the risk factors for readmission after free tissue transfer using a national data set and noted an association with reoperation and surgical and medical complications. Patients who underwent ablation of head and neck cancers have been identified as a high-risk population for readmission.7,9 To our knowledge, there has been no study specifically addressing patient-level readmission risk factors, which are known to differ by specific patient groups, following free flap reconstruction for head and neck cancer. This study analyzed 30-day readmissions in a subset of patients with head and neck cancer who received free flap reconstruction at a tertiary care center, with the aim of highlighting the rate, reasons, and predictive risk factors for early readmissions.
This is an institutional review board–approved retrospective medical record review of all patients who underwent microvascular reconstruction of a head and neck defect following a presumed oncologic extirpation at Brigham and Women’s Hospital from January 1, 2000, through June 30, 2014, with follow-up through July 30, 2014. Cases that involved microvascular reconstruction of known traumatic, congenital, or benign head and neck defects were excluded. Patients who died during the index hospitalization (2 patients) were also excluded from the study. A consecutive series of 249 patients were identified and comprised the final study cohort. Institutional review board approval was obtained from Brigham and Women’s Hospital.
The primary outcome of interest was all-cause hospital readmission, defined as an admission to any service in the hospital within 30 days of discharge from the surgical service (otolaryngology or plastic surgery). The overall sample was categorized into nonreadmitted and readmitted patients for the purposes of analysis. Individual-level data were collected for more than 40 variables related to patient demographics, comorbidity, ablative procedure performed during the index hospitalization, pathologic tumor information, postoperative course, length of stay, reason for readmission, and time between discharge and readmission. Length of stay was defined as the time from the date of the index operation to the date of discharge from the hospital. Early discharge, as described in the literature, is a length of stay below the 25th percentile for the cohort.12
In secondary analysis, we investigated the incidence of postoperative complications among readmitted patients. Postoperative complications, for the purposes of this study, were categorized as follows: any flap complication (suture line dehiscence at the flap inset, bleeding, venous congestion, flap revision, or flap loss), total flap loss, partial flap loss, fistula (orocutaneous or pharyngocutaneous), donor site complications, neck wound complications (defined as incidence of neck hematoma, neck incision dehiscence, cellulitis, or neck abscess), reoperation and medical complications (pneumonia, respiratory insufficiency, renal failure after surgery, urinary tract infection, myocardial infarction, stroke, or delirium).
Standard descriptive statistics (mean [SD], median, and percentage) were used to characterize patient cohorts (nonreadmitted vs readmitted patients). Univariate analysis with the χ2 or independent-sample, 2-sided, unpaired t test, when appropriate, was used to identify risk factors related to 30-day readmission (dependent variable). Only variables with a significant univariate test (P <.05) were used as independent variables in the creation of a multivariable logistic regression model. To identify the patient-level characteristics that would reliably predict early readmission, we excluded postoperative complications from the logistic regression of independent variables that entailed preoperative and intraoperative characteristics. Postoperative complications have been shown to be an independent risk factor for early readmissions and were considered separately during our secondary analysis.13 This approach is consistent with findings from published studies.6 All analyses were performed using SAS, version 9.4 (SAS Institute Inc). Two-tailed P <.05 was considered to indicate statistical significance.
Clinical characteristics and demographic data for the study sample (249 patients) are summarized in Table 1. The overall patient cohort had a mean (SD) age of 59 (12) years, a mean (SD) body mass index of 25.7 (5.1) (calculated as weight in kilograms divided by height in meters squared), and was predominantly male (145 [58.2%]). The most common tumor location was the oral cavity (189 of 249 [75.9%]), and the most common histologic diagnosis was squamous cell carcinoma (228 [91.6%]). A total of 168 of 249 cases (67.5%) were due to a primary or new cancer diagnosis. The radial forearm (170 [68.3%]) and osteocutaneous fibula free flaps (51 [20.5%]) comprised most of the microvascular reconstructive cases. The mean length of stay was 12.3 (8.6) days, and the median follow-up duration was 25 months. The breakdown of patients according to T stage and discharge status is also presented in Table 1.
Of the 249 patients identified in this study, 36 (14.5%) were readmitted within 30 days after discharge. Seven of the 36 readmissions (19.4%) were planned readmissions; of these, 5 were for the placement of a percutaneous endoscopic gastrostomy tube, and 2 were for the division of a coincident pedicled pectoralis myocutaneous flap to the neck. The incidence of unplanned readmissions in the present study was 11.6% (29 of 249), and the most common reasons for unplanned readmissions were neck wound complications (n = 7) and orocutaneous and haryngocutaneous fistula (n = 6). The complete list of readmission diagnoses is outlined in Table 2, and no unifying diagnosis was appreciated among patients with early readmissions.
As detailed in Table 3, Quiz Ref IDreadmitted patients were not found to significantly differ from nonreadmitted patients with respect to age, body mass index, sex predominance, baseline comorbidity, or cancer type. However, readmitted patients had a significantly longer mean (SD) postoperative stay during the index hospitalization (17.0 [17.7] days vs 11.5 [5.5] days; P = .009). The mean duration between discharge and readmission was 12.8 days. Early discharge and destination at time of discharge (ie, rehab vs skilled nursing facility) were not associated with readmission.
Advanced pathologic tumor stages T3 (odds ratio [OR], 9.92; P = .04) and T4 (OR, 8.92; P = .04) were associated with early hospital readmission on univariate analysis. Tumor presentation in the oropharynx (OR, 4.05; P = .001), larynx (OR, 10.12; P = .002), and hypopharynx (OR, 10.12; P = .007) were also significantly related to an early readmission on univariate analysis. After adjustment, having a tumor located in the oropharynx (OR, 4.64; 95% CI, 1.89-11.38; P = .001), larynx (OR, 10.97; 95% CI, 2.27-52.98; P = .003), or hypopharynx (OR, 8.30; 95% CI, 1.52-45.24; P = .01), along with T4 pathologic staging (OR, 11.68; 95% CI, 1.37-99.81; P = .02), were still statistically significant (Table 4). The type of microvascular flap used in the reconstruction was not significantly associated with readmission.
Readmitted patients were found to have significantly higher rates of the following complications: any flap complication (18 [50.0%] vs 54 [25.4%]; P = .003), total flap loss (8 [22.2%] vs 7 [3.3%]; P < .001), fistula (13 [36.1%] vs 24 [11.3%]; P < .001), neck wound complications (17 [47.2%] vs 49 [23.0%]; P = .002), and reoperation (20 [55.6%] vs 2 [0.9%]; P < .001). Quiz Ref IDFollowing multivariable analysis, only reoperation (OR, 47.20; 95% CI, 8.33-267.33; P <.001) and neck wound complications (OR, 5.07; 95% CI, 1.31-19.57; P = .02) retained statistical significance (Table 5).
Today’s health care and economic climate mandates that early hospital readmission be closely monitored both as a cost-containment strategy and a metric of health care quality. Accordingly, considerable research effort in recent years has focused on enumerating surgical readmissions and identifying their root causes.10Quiz Ref ID Prior studies11 have quantified 30-day readmission rates by specialty, yielding readmission rates of 11.9% for vascular operations, 11.1% for thoracic operations, and 5.0% after general surgery procedures (hernia, breast, gallbladder, and appendix). Unfortunately, there is a limited representation in reconstrictive surgery publications of reliable data on early hospital readmissions.6,8 With microvascular techniques increasingly becoming the de facto reconstructive modality at many head and neck cancer centers,14,15 it is imperative that an evaluation of the readmission risk factors for this resource-intensive enterprise be performed. Risk factor analysis for early readmission after microvascular surgery was conducted by Kim et al,8 but the study population contained a high proportion of breast reconstruction cases (60.3%); consequently, its findings are not equally representative for this population of patients who were already identified as being at high risk for early readmission.
In this study, the 30-day all-cause readmission rate for a 14-year period was 14.5%, and the unplanned readmission rate was 11.6%. Given the inherent complexity of head and neck reconstruction, these data are comparable with the reported value of 7.9% for unplanned readmissions in previous work.8Quiz Ref ID One of our major findings, hitherto novel to our knowledge, is that advanced pathologic T stage and tumor subsite (location in the pharynx or larynx) yielded the most significant increases in likelihood for early readmission. A similar association between advanced T stage and higher rates of readmission had been noted following other high-risk operations, such as pancreaticoduodenectomy.16 Our results make intuitive sense given the context that advanced T stage dictates the extent of surgical resection and reconstruction and has been convincingly shown in studies17- 19 to be related to poorer postoperative outcomes, functionality, and quality of life following free flap reconstruction for head and neck cancer. Although more studies are indicated, T-category grouping in head and neck cancer may have the ability to reliably prognosticate readmission risk following microvascular reconstruction.
Handling of the upper aerodigestive tract, which is requisite for the resection and reconstruction of tumors located in the pharynx and larynx, is more likely to produce pronounced changes in speech, breathing, and swallowing compared with oral cavity cancers.9 These life-altering physiologic changes, temporary or permanent, often pose increased demands on patients and their caretakers during the early postoperative period.9 Also, the relative inaccessibility of the laryngeal and pharyngeal locations to potential local wound and suture line care, as well as the continuous physiologic movements across them, might predispose these patient groups to more significant wound issues and thus earlier or more hospital readmissions for necessary management. Total laryngectomy was found to independently increase the risk of unplanned readmission 5-fold in the study by Graboyes et al.9 The study authors asserted that the intricacies of these physiologic changes placed recipients at higher risk for early readmissions. Most oropharyngeal, hypopharyngeal, and laryngeal cancers are initially treated with primary radiotherapy relative to oral cavity cancers; it is possible that this process might also contribute to a higher incidence of wound complications and fistulas, leading to more readmissions. In our study, 66.1% of patients with pharyngeal or laryngeal cancers received radiation therapy before definitive surgical resection compared with only 29.1% of patients with oral cavity cancers.
Certain complications were noted among patients with early readmission in our study population following multivariable analysis; these were the occurrence of a neck wound complication (OR, 5.07) and reoperation (OR, 47.20). Neck wound complications, defined by the presence of a hematoma, incision dehiscence, cellulitis, or abscess, were also the presenting diagnosis for 24.1% of unplanned readmissions. Similarly, other studies have shown that reoperation8 and postoperative complications7,8 were associated with unplanned readmissions. All of these outcomes can have serious consequences regarding the creation of transition-of-care policies7; in other words, a reasonable argument can be made for the redirection of individual and systems-based practices toward averting and recognizing complications promptly, eg, earlier subspecialist follow-up or standardization of postoperative care processes.8,20 If readmissions are indeed the “final common pathway for most surgical complications,”10(p197) as proposed by Lucas and Pawlik in their 2014 review article, then such an intervention strategy may represent the most effective avenue for reducing hospital readmissions. A potential area for future research is to examine the temporal association between the recognition of complication onset during the postoperative period (ie, before or after discharge) and incidence of readmissions. This information would greatly augment our current understanding of the contribution of postoperative complications to readmissions after microvascular surgery. A 2014 study by Morris et al21 revealed that postdischarge complications were associated with a 58.9% readmission rate compared with an 18.3% readmission rate for predischarge complications after arthroplasty as well as colorectal, vascular, and gynecologic procedures (P < .001). Furthermore, our data showed that those in the readmissions group experienced a prolonged postoperative stay during the index hospitalization. Prior studies have also noted a similar association following complex cardiac, vascular, spinal, and bariatric operations.22- 25
Regarding planned readmissions, percutaneous endoscopic gastrostomy placement was the underlying diagnosis in 71.4% of cases. In our institution, the indication is for the creation of enteral access for nutritional support before the initiation of adjuvant therapy.26 A better estimation on the part of treating physicians regarding the suitability or role of percutaneous endoscopic gastrostomy placement in a patient’s oncologic treatment strategy (ie, more precise tumor staging) might reduce the risk for planned readmissions.
Remarkably, we did not find comorbidities (in aggregate and when weighed individually) or destination at the time of discharge to be significantly associated with 30-day readmissions. These findings are largely congruent with the results of prior studies8 and speak to a likely uniform representation of head and neck cancer–related comorbidities in our study population (eg, poor nutrition, body mass index, or smoking). It is also possible that these comorbid conditions were medically optimized during the index hospital course.
This study has several methodological limitations beyond its retrospective design and reliance on the exactitude and comprehensiveness of the medical record. We were unable to study patients who were readmitted at other hospitals, possibly leading to an underestimation of the true rate of readmissions. However, it is our opinion and experience that most patients who received a significant portion of their index hospital care at our tertiary institution are transferred back to us in instances when a readmission is deemed necessary. Also, although we are affiliated with a regional cancer care referral center, the single-institution nature of our study means that our results may not be representative of all microvascular reconstruction practices, that is, perioperative care pathways or thresholds for readmissions. Last, the relatively small number of readmissions may lead to imprecise and wide estimates of the 95% CI on multivariable analysis. Additional multicenter, prospective studies are needed to expand the panel of readmission risk factors.
The first step in any meaningful effort to address the health care quality concerns surrounding early readmissions should be the identification of predictive risk factors. We have found that advanced pathologic tumor staging, tumor location, and certain postoperative complications are risk factors for unplanned readmissions. By providing more robust and detailed clinical data, the results of our study complement the findings of existing studies. It is clear that the complete eradication of readmissions is unlikely given that some readmissions are clinically indicated and represent appropriate health care delivery.5 However, our results provide a basis for risk stratification, which can be incorporated in system-based practice improvements. For instance, physicians can now focus additional time and resources on their highest-risk patients, such as those with T4-stage disease or laryngeal tumor subsites, because they are about to enter the postdischarge setting.7 The data presented here may also prove useful in postoperative care counseling and managing patient expectations.
Submitted for Publication: February 13, 2015; final revision received April 8, 2015; accepted June 2, 2015.
Corresponding Author: Lifei Guo, MD, PhD, Department of Plastic Surgery, Lahey Hospital & Medical Center, 41 Mall Rd, Burlington, MA 01805 (firstname.lastname@example.org).
Published Online: July 23, 2015. doi:10.1001/jamaoto.2015.1323.
Author Contributions: Drs Offodile and Guo had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Offodile, Orgill, Guo.
Acquisition, analysis, or interpretation of data: Offodile, Pathak, Wenger, Guo.
Drafting of the manuscript: Offodile, Pathak, Wenger, Guo.
Critical revision of the manuscript for important intellectual content: Pathak, Orgill, Guo.
Statistical analysis: Offodile, Wenger, Guo.
Administrative, technical, or material support: All authors.
Study supervision: Offodile, Orgill, Guo.
Conflict of Interest Disclosures: None reported.
Previous Presentation: This article was presented as a poster at the 94th Annual Meeting of the American Association of Plastic Surgeons; April 12, 2015; Scottsdale, Arizona.