Research-Human-Clinical Studies: Editor's Choice
Does 30-Day Readmission Affect Long-term Outcome Among Glioblastoma Patients?
Nuño, Miriam PhD; Ly, Diana MPH; Ortega, Alicia BS; Sarmiento, J. Manuel BA; Mukherjee, Debraj MD, MPH; Black, Keith L. MD; Patil, Chirag G. MD, MS
Center for Neurosurgical Outcomes Research, Maxine Dunitz Neurosurgical Institute, Department of Neurosurgery, Cedars-Sinai Medical Center, Los Angeles, California
Correspondence: Chirag G. Patil, MD, MS, Center for Neurosurgical Outcomes Research, Department of Neurosurgery, Cedars-Sinai Medical Center Advanced Health Sciences Pavilion, 127 S. San Vicente Blvd., Suite A6229, Los Angeles, CA 90048. E-mail: firstname.lastname@example.org
* These authors have contributed equally to this article.
Supplemental Digital Content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of the article on the journal's Web site (www.neurosurgery-online.com).
Received June 12, 2013
Accepted October 28, 2013
BACKGROUND: Research on readmissions has focused mainly on the economic and resource burden it places on hospitals.
OBJECTIVE: To evaluate the effect of 30-day readmission on overall survival among newly diagnosed glioblastoma multiforme (GBM) patients.
METHODS: A nationwide cohort of GBM patients diagnosed between 1991 and 2007 was studied using the Surveillance, Epidemiology and End Results Medicare database. Multivariate models were used to determine factors associated with readmission and overall survival. Odds ratio, hazard ratio, 95% confidence interval, and P values were reported. Complete case and multiple imputation analyses were performed.
RESULTS: Among the 2774 newly diagnosed GBM patients undergoing surgery at 442 hospitals nationwide, 437 (15.8%) were readmitted within 30 days of the index hospitalization. Although 63% of readmitted patients returned to the index hospital where surgery was performed, a significant portion (37%) were readmitted to nonindex hospitals. The median overall survival for readmitted patients (6.0 months) was significantly shorter than for nonreadmitted (7.6 months; P < .001). In a confounder-adjusted imputed model, 30-day readmission increased the hazard of mortality by 30% (hazard ratio, 1.3; P < .001). Neurological symptoms (30.2%), thromboembolic complications (19.7%), and infections (17.6%) were the leading reasons for readmission.
CONCLUSION: Prior studies that have reported only the readmissions back to index hospitals are likely underestimating the true 30-day readmission rate. GBM patients who were readmitted within 30 days had significantly shorter survival than nonreadmitted patients. Future studies that attempt to decrease readmissions and evaluate the impact of reducing readmissions on patient outcomes are needed.
ABBREVIATIONS: CI, confidence interval
GBM, glioblastoma multiforme
ICD-9-CM, International Classification of Diseases, Ninth Revision, Clinical Modification
SEER, Surveillance, Epidemiology and End Results
Thirty-day hospital readmission has emerged as one of the most important pay-for-performance benchmarks and is being presumed by policy makers to be an indicator of hospital quality.1 Under the auspices of the Patient Protection and Affordable Care Act, hospitals with higher standardized readmission rates for medical conditions such as heart failure and pneumonia are already being penalized by the Centers for Medicare & Medicaid Services.2 It is expected that in the near future the focus will expand beyond medical diagnoses and include postsurgical readmission rate assessments and penalties. However, studies reporting associations between 30-day readmission after surgical procedures and long-term patient outcomes are limited.
Before policy makers implement 30-day readmission as a quality measure, it is important to validate readmissions as being associated with worse patient outcomes in each subspecialty area. Unfortunately, most of the past research on readmissions has mainly focused on the economic and resource burden it places on hospitals, and its association with clinically relevant parameters such as patient outcomes has not been emphasized.3-7 A handful of studies in surgical oncology have revealed associations between 30-day readmission and overall survival.7-14 For example, Reddy et al14 demonstrated that 30-day readmissions for patients with pancreatic cancer were associated with early mortality. To date, no study has investigated the relationship between readmission after brain tumor surgery and long-term outcomes such as overall survival.
Given that glioblastoma multiforme (GBM) is the most common primary malignant brain tumor, we selected this relatively homogeneous diagnosis to assess whether readmission after initial surgery was associated with worse overall survival. Using the Surveillance, Epidemiology and End Results (SEER) Medicare-linked database, we controlled for known prognostic factors to determine whether 30-day readmission is independently associated with shorter survival. We further detailed the reasons for readmissions and attempted to determine factors associated with readmissions.
MATERIALS AND METHODS
We used the National Cancer Institute's SEER Medicare-linked database for this analysis. The SEER program of cancer registries prospectively collects incident cancer cases with information on patient demographics, tumor characteristics, first course of treatment, and survival data. Individuals from this data set who are ≥65 years of age are matched to Medicare master enrollment file; 93% of persons ≥65 years old in the SEER database have been matched to the Medicare database.15 The 17 registries of the SEER program comprise approximately 28% of the US population; 16 of the registries are participating in the SEER-Medicare linkage. In this analysis, the Medicare Provider Analysis and Review file was merged with the Patient Entitlement and Diagnosis Summary File to retrieve hospital claims for patients with a diagnosis of GBM.
Patients who met the following criteria were included in the study: (1) International Classification of Diseases, Oncology, Third Revision histology codes for GBM of 9440, 9441, and 9442; (2) age ≥66 years at the time of diagnosis (to ensure that patients had a full year of inpatient hospitalization data up to the index admission to identify preexisting comorbidities); (3) GBM as the only primary tumor; (4) enrollment in both Medicare Parts A and B and not members of a health maintenance organization during the 3-month period after diagnosis; (5) microscopic confirmation of tumor; (6) age eligibility as the sole reason for Medicare entitlement; (7) reporting source for the cancer case was hospital, laboratory, private doctor, or nursing home; (8) known month of diagnosis; and (9) biopsy or surgery was performed on the basis of the Patient Entitlement and Diagnosis Summary File variable for extent of resection and International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) procedure codes (see Table, Supplemental Digital Content 1, http://links.lww.com/NEU/A595, which gives ICD-9-CM and Healthcare Common Procedure Coding System codes for procedures and diagnoses during index admission.). Of the 3277 individuals who fulfilled these inclusion criteria, we identified 3082 patients who had their first procedure date related to a biopsy or surgical resection during the month of or the next month after diagnosis. We then excluded 154 patients who died within a month after their index admission and 154 patients who had planned readmissions. We defined “planned” readmission as any readmission in which a patient was admitted as elective. According to the Medicare database, an elective admission is defined as any admission in which the patient's condition permitted adequate time to schedule the availability of suitable accommodations. As a result, we identified a total of 2774 GBM patients diagnosed between January 1991 and December 2007 who met our study criteria.
An unplanned readmission denoted a readmission to an inpatient hospital that occurred within 30 days of discharge from index hospitalization in which a biopsy or surgical resection was performed. Readmissions that occurred on the same day as the index discharge were identified as being part of the index hospitalization. Reasons for readmission were obtained from Medicare Provider Analysis and Review's fields for ICD-9 diagnosis codes (see Table, Supplemental Digital Content 2, http://links.lww.com/NEU/A596, which gives ICD-9 diagnosis codes identifying reasons for readmission). The principal diagnosis code was used to identify the primary reason for readmission, but when more information was needed (eg, if the principal diagnosis code indicated brain tumor), the ICD-9-CM diagnosis code next highest in rank was examined to derive the reason for readmission.
Details were obtained on patients' sociodemographic characteristics, including age, sex, race, marital status, and median household income. Clinical characteristics included year of diagnosis, index length of stay, index admission type (emergent, urgent, elective, or trauma), index discharge disposition, patient comorbidities, and complications during index admission. Years were grouped according to the following categories: 1991 to 1995, 1996 to 1999, 2000 to 2003, and 2004 to 2007. Discharge disposition was classified as routine (discharged to home/self-care) vs nonroutine (discharged to other inpatient, outpatient, home health, hospice services). Elixhauser comorbidities were calculated on the basis of a full year of inpatient hospitalization data up to the index admission. Complications assessed during index admission included neurological, pulmonary, thromboembolic, cardiac, urinary/renal, and other procedural complications and complications of medical care. Additionally, tumor size and location were ascertained. Patients were categorized as undergoing biopsy, partial resection, or gross total resection on the basis of the Patient Entitlement and Diagnosis Summary File variable for extent of resection.
Descriptive statistics were reported for all patients (n = 2774) and for cohorts classified according to whether they were readmitted within 30 days of the index surgery. Continuous variables were described using means and medians, whereas categorical variables were reported with frequencies. Univariate comparisons by readmission were conducted using Wilcoxon rank sum, χ2, and Fisher exact tests when appropriate. Factors associated with a readmission were analyzed with logistic regression models. Demographic and/or clinical variables that were associated with readmission at the univariate level, reaching significance of 0.20 or lower, were included in the multivariate model. In this analysis identifying predictors of readmission, we excluded 26 patients who were discharged to hospice. The Kaplan-Meier estimation method was used to obtain median survival times and estimated probability rates. We used the Cox proportional hazards model to analyze the effect of readmission on survival while adjusting for confounders. Complete data and multiple imputation data analysis were reported. A value of P ≤ 0.05 was considered to be statistically significant. All statistical analyses were conducted in SAS 9.2 (SAS Institute, Cary, North Carolina).
According to the SEER-Medicare database, between 1991 and 2007, a total of 2774 GBM patients were diagnosed at 442 hospitals in the United States. Among these patients, 437 (15.8%) were readmitted within 30 days; of these 437 patients, 152 (34.8%), 105 (24.0%), and 81 (18.5%) were hospitalized within 7, 14, and 21 days after discharge, respectively. Whereas more patients (n = 275, 62.9%) were readmitted to the same hospital of the index surgery, a significant proportion of patients (n = 162, 37.1%) were readmitted elsewhere within those 30 days. We found a 30-day same-hospital readmission rate of 9.9% with an additional 5.9% of patients being readmitted to hospitals other than the operating hospital.
Overall, the median age of patients was 73 years; 52.1% of patients were male, 94.1% were white, and 67.3% were married (Table 1). Most tumors (65.3%) were ≥4 cm. Gross total resection was achieved in 42.4%; partial resection was achieved in 36.2%; and biopsy was performed in 21.4% of the cases. Most patients (84.4%) had 1 or more comorbidities. Table 1 describes missing data for variables of interest. Table 2 depicts differences in length of stay (P = .01), discharge disposition (P = .03), and complications at index admission (P = .03) between the readmitted and nonreadmitted cohorts.
The median overall survival for the entire cohort of 2774 patients was 7.4 months (95% confidence interval [CI], 7.1-7.6) with survival rates at 3, 6, 12, and 24 months of 92.8%, 60.5%, 24.0%, and 4.9%, respectively (Table 3). Median overall survival was reduced from 7.6 months (95% CI, 7.3-7.9) to 6.0 months (95% CI, 5.7-6.7) in patients with a 30-day readmission (log-rank P < .001; Figure 1). Patients who experienced a 30-day readmission had 3-, 6-, 12-, and 24-month survival rates of 91.0%, 52.7%, 18.3%, and 1.9%, respectively (Table 3).
In a confounder-adjusted multiple imputation analysis of overall survival, we found that readmission increased the hazards of mortality by 30% (hazard ratio [HR] 1.3; P < .001; Table 4). Furthermore, a 10-year increment in age at diagnosis (HR, 1.3; P < .001), nonroutine discharge (HR, 1.1; P = .01), and larger tumors (HR, 1.1; P = .01) were significantly associated with increased hazards of mortality. Factors that were significantly associated with lower mortality hazards included higher median income (HR, 0.97; P = .02), later year of diagnosis (HR, 0.97; P < .001), postoperative radiotherapy (HR, 0.6; P < .001), and gross total resection compared with biopsy (HR, 0.9; P < .001). Although readmitted patients appeared to have higher levels of comorbidities than nonreadmitted patients at the univariate level (Table 1), comorbidities were not significantly associated with survival or readmission in the multivariate setting (Tables 4 and 5). Multivariate analysis aimed at determining factors associated with 30-day readmissions yielded nonsignificant predictors (Table 5).
Reasons for Readmission
A detailed analysis of the 437 readmitted patients showed that neurological (30.2%), thromboembolic (19.7%), and infectious complications (17.6%) were the top 3 reasons for readmission (Figure 2); the remaining reasons included fluid and electrolytes (8.9%), gastrointestinal (6.0%), cardiac (5.3%), miscellaneous (5.0%), general/failure to thrive (3.9%), respiratory (1.8%), and psychiatric (1.6%) complications. Table 2, Supplemental Digital Content 2 delineates the specific ICD-9-CM diagnosis codes that determined each of these categories.
Thirty-Day Readmission and Survival
Possible associations between 30-day patient readmissions, cost, resource burden, and its implications to the process of quality of care have been widely studied for a multitude of conditions.6,7,11,16,17 Jencks et al18 estimated that the cost to Medicare for unplanned rehospitalizations in 2004 was $17.4 billion. More recently, Lawson et al16 noted a savings of $620.3 million per year to Medicare with the prevention of complications (and readmissions) among surgical patients. Fewer studies have investigated the effect of early readmissions on patient outcomes, especially long-term outcomes such as overall survival. Schneider et al10 assessed the impact of readmission in the survival of 9957 hepato-pancreato-biliary malignancy patients undergoing surgery. Compared with nonreadmitted patients, they found that overall survival was reduced by 10.4 months (21.3 vs 10.9 months, P < .001) among patients who were readmitted within 30 days of index hospitalization. Tuggle et al11 showed that unplanned rehospitalization was significantly associated with death at 1 year compared with nonhospitalized patients (18% vs 6%; P < .001). Reddy et al14 similarly found that patients readmitted within 30 days had shorter median survival compared with those who did not experience a readmission (11.8 vs 16.5 months; P = .04).
Similar to previous investigations, our present study demonstrates a significant reduction in the median overall survival from 7.6 to 6.0 months (P < .001) in patients who are readmitted within 30 days of surgery. The implementation of temozolomide along with radiation in 2005 as standard treatment for newly diagnosed glioblastoma patients improved their median overall survival to 14.6 months.19 The median survival of about 7 months in this SEER-Medicare study is illustrative of an older (>65 years), nationwide Medicare cohort and is consistent with other SEER publications. To the best of our knowledge, this is the first report on the association of readmission and overall survival among brain tumor patients. Furthermore, after adjusting for possible confounders such as age, preexisting comorbidities, treatment, and income, among others, we found that readmitted patients had a 30% increase in the hazards of mortality. The most common reasons for readmission were neurological, thromboembolic, and infectious complications.
As with any administrative database, the Medicare portion of our data are particularly limited by the quality and accuracy of medical chart documentation and coding. Assessing the preventability of the complications that led to readmission is obviously quite important. Given the limitations of the SEER-Medicare database, a more detailed analysis of the reasons and preventability of readmission beyond what we have presented was not feasible. Neurological complications were documented as the reason for readmission in 30% of the patients. Given the aggressive nature of glioblastoma, some or many of the neurological complications may be due to tumor progression and related consequences and hence may not be preventable. Furthermore, our analysis is limited by our inability to capture known prognostic factors of survival such as Karnofsky performance scores, MGMT methylation, and IDH1 status. Although Karnofsky performance score has been shown to be an independent and significant predictor of survival among GBM patients, preliminary findings from our internal cohort of 354 GBM patients indicate that even after adjustment for Karnofsky performance scores, readmission remains strongly associated with poor survival. In addition, studies in other disease processes have reported that even after adjustment for functional status, readmissions continued to be independently associated with increased mortality.20
No previous studies have focused on readmissions after brain tumor surgery. Readmissions after neurosurgical spinal surgery have been described.21 Amin et al21 reported an index hospital readmission rate of 3.6% in an internal hospital cohort that involved detailed chart view of cases reported. In an effort to determine the validity of the readmission rate documented in this administrative database, we reviewed our internal hospital cohort of newly diagnosed glioblastoma patients (2003-2011) and found a 7.5% 30-day readmission rate (unpublished data). Although the SEER-Medicare readmission rate in this study may appear higher than reported among our institutional series or the spinal surgery patients, our 9.9% rate reflects older patients treated at a multitude of hospitals nationwide.
Readmission by Index Hospital
A multitude of studies across different disease states have reported on hospital readmissions (see Table, Supplemental Digital Content 3, http://links.lww.com/NEU/A597, which lists the characteristics of selected studies assessing factors associated with readmission for several disease conditions). However, very few studies involve nationwide cohorts that facilitate the analysis of patient readmissions across multiple hospitals. In this study, we found that although 63% of readmitted patients returned to the index hospital where they had surgery, more than one-third (37.%) were readmitted to nonindex, outside hospitals. Similarly, Saunders et al22 have reported that among patients undergoing abdominal aortic aneurysm repair, 29.3% were readmitted to nonindex hospitals. This finding suggests that single-institution studies, which are generally limited to analyzing readmissions back to the treating hospital, are likely to significantly underestimate the true readmission rate. Although numerous studies have pointed to the limitations of using administrative databases for the purpose of analyzing readmission rates, these databases can have certain advantages in health services research.21,23,24 Our evaluation of index vs nonindex hospital readmissions provides an excellent sample of the useful insight that national database studies can provide. Given the inherent nature of this multiple hospital cohort study (n = 442), the findings in this study can be generalizable to hospitals across the nation.
Factors Predictive of 30-Day Readmissions
Neurological (30.2%), thromboembolic (19.7%), and infectious (17.6%) complications were the top three reasons for readmission. We found that unlike previous studies in which age,3,25,26 sex,9 complications,8,11,12,14 income,4,27,28 type of admission,9,21,29 mental health,25 length of stay,21 insurance,6,30,31 and hemoglobin/sodium and number of admissions32 were associated with readmission, our analysis did not reveal any significant predictors (see Table, Supplemental Digital Content 3, http://links.lww.com/NEU/A597). Our inability to capture significant predictors of readmission in this patient cohort may be partially explained by different disease condition of study or unavailable data that may be relevant to determining readmissions among glioblastoma patients. For example, the SEER-Medicare database has limitations with no information on functional status such as Karnofsky performance scale (functional status) that may be very relevant to the risk of readmission.
Our study tends to support the growing literature on 30-day readmission as being a critical indicator that is associated with long-term patient outcomes. Independently of treatment, patient risk factors, or extent of resection, the findings in this study demonstrate that 30-day readmission is associated with an average 6-week reduction in overall survival. Future studies that aim to reduce postdischarge complications and to prevent readmissions should also assess the impact of reducing these readmissions on short-term and long-term patient outcomes. Future exploration of factors influencing early hospitalizations and survival among cancer patients may involve assessment of end-of-life care. For example, studies have showed that palliative care services can help reduce hospitalization by as much as 80%, and these services have also been shown to extend survival.33,34 Although not evaluated in this study, reduction in hospitalizations by early implementation of end-of-life care may improve survival because patients have fewer readmissions and thereby are less exposed to the risks associated with a hospitalization such as hospital-acquired conditions (catheter-associated urinary tract infections, Clostridium difficile colitis, hospital-acquired pneumonia, catheter-associated bloodstream infections).35
Is 30-day Readmission a Hospital Quality Indicator?
The goal of this study was to establish a possible association between 30-day readmission and survival. This study was not designed to evaluate whether readmission is a hospital quality indicator. Therefore, hospital and treatment factors such as hospital volume, centers of excellence designation, availability of clinical trials, and adherence to National Comprehensive Cancer Network guidelines were not compared between hospitals. Despite the growing awareness of readmissions as a possible indicator of poor hospital care, little evidence exists linking readmissions to substandard inpatient care.36-39 Research on factors influencing the complex relationship of the quality-of-care process and early readmissions is limited.40 As noted by Marks et al,40 patient outcomes are influenced by a complex adaptive system that involves patient-specific factors, caregivers, physicians, and social service agencies, among others. Current methodology that attempts to adjust readmission rates by comorbidities and severity of illness may not be adequate. Future studies are necessary to assess whether correlation exists between well-adjusted 30-day readmission rates and other more established quality indicators and patient outcome indicators.
Our study supports the growing literature on 30-day readmission as being a critical indicator that is associated with long-term patient outcomes. After adjusting for multiple confounders, we have demonstrated that 30-day readmission is significantly associated with shorter overall survival among GBM patients. Neurological, infection, and thromboembolic complications were the leading reasons for readmission. Future studies that aim to decrease readmissions by decreasing these complications need also to evaluate the impact of reducing readmissions on patient short- and long-term outcomes.
The authors have no personal financial or institutional interest in any of the drugs, materials, or devices described in this article.
1. Vaduganathan M, Bonow RO, Gheorghiade M. Thirty-day readmissions: the clock is ticking. JAMA. 2013;309(4):345–346.
2. TPPaACAtUSCM. 2010.
3. Anderson MA, Clarke MM, Helms LB, Foreman MD. Hospital readmission from home health care before and after prospective payment. J Nurs Scholarsh. 2005;37(1):73–79.
4. Arbaje AI, Wolff JL, Yu Q, Powe NR, Anderson GF, Boult C. Postdischarge environmental and socioeconomic factors and the likelihood of early hospital readmission among community-dwelling Medicare beneficiaries. Gerontologist. 2008;48(4):495–504.
5. Dignan FL, Potter MN, Ethell ME, et al.. High readmission rates are associated with a significant economic burden and poor outcome in patients with grade III/IV acute GvHD. Clin Transplant. 2013;27(1):E56–E63.
6. Friedman B, Basu J. The rate and cost of hospital readmissions for preventable conditions. Med Care Res Rev. 2004;61(2):225–240.
7. Kent TS, Sachs TE, Callery MP, Vollmer CM. Readmission after major pancreatic resection: a necessary evil? J Am Coll Surg. 2011;213(4):515–523.
8. Emick DM, Riall TS, Cameron JL, et al.. Hospital readmission after pancreaticoduodenectomy. J Gastrointest Surg. 2006;10(9):1243–1252; discussion 1252-1243.
9. Greenblatt DY, Weber SM, O'Connor ES, LoConte NK, Liou JI, Smith MA. Readmission after colectomy for cancer predicts one-year mortality. Ann Surg. 2010;251(4):659–669.
10. Schneider EB, Hyder O, Wolfgang CL, et al.. Patient readmission and mortality after surgery for hepato-pancreato-biliary malignancies. J Am Coll Surg. 2012;215(5):607–615.
11. Tuggle CT, Park LS, Roman S, Udelsman R, Sosa JA. Rehospitalization among elderly patients with thyroid cancer after thyroidectomy are prevalent and costly. Ann Surg Oncol. 2010;17(11):2816–2823.
12. van Geenen RC, van Gulik TM, Busch OR, de Wit LT, Obertop H, Gouma DJ. Readmissions after pancreatoduodenectomy. Br J Surg. 2001;88(11):1467–1471.
13. Yermilov I, Bentrem D, Sekeris E, et al.. Readmissions following pancreaticoduodenectomy for pancreas cancer: a population-based appraisal. Ann Surg Oncol. 2009;16(3):554–561.
14. Reddy DM, Townsend CM, Kuo YF, Freeman JL, Goodwin JS, Riall TS. Readmission after pancreatectomy for pancreatic cancer in Medicare patients. J Gastrointest Surg. 2009;13(11):1963–1974; discussion 1974-1965.
15. Warren JL, Klabunde CN, Schrag D, Bach PB, Riley GF. Overview of the SEER-Medicare data: content, research applications, and generalizability to the United States elderly population. Med Care. 2002;40(suppl 8):IV3–IV18.
16. Lawson EH, Lee Hall B, Louie R, et al.. Association between occurrence of a postoperative complication and readmission: implications for quality improvement and cost savings. Ann Surg. 2013.
17. Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med. 2009;360(14):1418–1428.
18. Jencks SF, Bobula JD. Does receiving referral and transfer patients make hospitals expensive? Med Care. 1988;26(10):948–958.
19. Stupp R, Mason WP, van den Bent MJ, et al.. Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med. 2005;352(10):987–996.
20. Ma HM, Yu HY, Woo J. Recurrent hospitalization with pneumonia is associated with a higher 1-year mortality in frail older people [published online ahead of print August 14, 2013]. Intern Med J. 2013. doi: 10.1111/imj.12258.
21. Amin BY, Tu TH, Schairer WW, et al.. Pitfalls of calculating hospital readmission rates based on nonvalidated administrative data sets: presented at the 2012 Joint Spine Section Meeting: clinical article. J Neurosurg Spine. 2013;18(2):134–138.
22. Saunders RS, Fernandes-Taylor S, Decker MR, et al.. Rehospitalization to index vs. non-index facility following abdominal aortic aneurysm (AAA) repair. Vascular Annual Meeting, May 30-June 1 2013; San Francisco, PS10.
23. Lindenauer PK, Lagu T, Shieh MS, Pekow PS, Rothberg MB. Association of diagnostic coding with trends in hospitalizations and mortality of patients with pneumonia, 2003-2009. JAMA. 2012;307(13):1405–1413.
24. Sarrazin MS, Rosenthal GE. Finding pure and simple truths with administrative data. JAMA. 2012;307(13):1433–1435.
25. Goldfield NI, McCullough EC, Hughes JS, et al.. Identifying potentially preventable readmissions. Health Care Financ Rev. 2008;30(1):75–91.
26. Weeks WB, Lee RE, Wallace AE, West AN, Bagian JP. Do older rural and urban veterans experience different rates of unplanned readmission to VA and non-VA hospitals? J Rural Health. 2009;25(1):62–69.
27. Kim H, Ross JS, Melkus GD, Zhao Z, Boockvar K. Scheduled and unscheduled hospital readmissions among patients with diabetes. Am J Manag Care. 2010;16(10):760–767.
28. Rathore SS, Masoudi FA, Wang Y, et al.. Socioeconomic status, treatment, and outcomes among elderly patients hospitalized with heart failure: findings from the National Heart Failure Project. Am Heart J. 2006;152(2):371–378.
29. Timmers TK, Verhofstad MH, Moons KG, Leenen LP. Patients' characteristics associated with readmission to a surgical intensive care unit. Am J Crit Care. 2012;21(6):e120–128.
30. Hasan O, Meltzer DO, Shaykevich SA, et al.. Hospital readmission in general medicine patients: a prediction model. J Gen Intern Med. 2010;25(3):211–219.
31. Novotny NL, Anderson MA. Prediction of early readmission in medical inpatients using the Probability of Repeated Admission instrument. Nurs Res. 2008;57(6):406–415.
32. Donzé J, Aujesky D, Williams D, Schnipper JL. Potentially avoidable 30-day hospital readmissions in medical patients: derivation and validation of a prediction model. JAMA Intern Med. 2013:173:632–638.
33. Barrett M, Wheatland B, Haselby P, Larson A, Kristjanson L, Whyatt D. Palliative respite services using nursing staff reduces hospitalization of patients and improves acceptance among carers. Int J Palliat Nurs. 2009;15(8):389–395.
34. Temel JS, Greer JA, Muzikansky A, et al.. Early palliative care for patients with metastatic non-small-cell lung cancer. N Engl J Med. 2010;363(8):733–742.
35. Saliba W, Fediai A, Edelstein H, Markel A, Raz R. Trends in the burden of infectious disease hospitalizations among the elderly in the last decade. Eur J Intern Med. 2013;24:536–540.
36. Ashton CM, Kuykendall DH, Johnson ML, Wray NP, Wu L. The association between the quality of inpatient care and early readmission. Ann Intern Med. 1995;122(6):415–421.
37. Ashton CM, Wray NP. A conceptual framework for the study of early readmission as an indicator of quality of care. Soc Sci Med. 1996;43(11):1533–1541.
38. Kossovsky MP, Sarasin FP, Perneger TV, Chopard P, Sigaud P, Gaspoz J. Unplanned readmissions of patients with congestive heart failure: do they reflect in-hospital quality of care or patient characteristics? Am J Med. 2000;109(5):386–390.
39. Press MJ, Scanlon DP, Ryan AM, et al.. Limits of readmission rates in measuring hospital quality suggest the need for added metrics. Health Aff (Millwood). 2013;32(6):1083–1091.
40. Marks E. Complexity science and the readmission dilemma. JAMA Intern Med. 2013;173(8):629–631.
The article is a stimulating, timely, well-written, novel, and extensive analysis of a large number (n = 2774) of glioblastoma patients cared for in 442 hospitals nationwide using the Surveillance, Epidemiology and End Results and Medicare registries (1991-2007). The study proposes to evaluate the simple thesis that 30-day readmission rates reflect, and possibly shape, long-term outcomes among glioblastoma patients. If this hypothesis is valid, the authors reason that outcomes for glioblastoma patients could improve by avoiding readmissions, thereby supporting current stipulations in the Affordable Care Act to penalize hospitals financially if the 30-day readmission rate exceeds a preset norm. The article, as written, illustrates the pitfalls of using data to support a preconceived policy decision to save dollars in the name of promoting health care. Beyond the numbers and underlying political implications, the complex data require integration with our current understanding of glioblastoma prognosis and best practices.
There are 3 fundamental problems with the interpretation of the data. First, the 30-day readmission metric as a tool to penalize poorly performing hospitals is perilous, risking punishment of first-rate institutions that provide care for patients in the greatest need of advanced care.1-3 Second, the heterogeneity of glioblastoma “multiforme” in terms of biology and clinical care that influences outcomes3-6 are not considered. Third, the care and outlook of patients with glioblastoma have changed significantly since 19976-10; therefore, the data may be outdated, a historical benchmark, but of limited value to inform current treatment policy.
First, using readmission rates as a marker of suboptimal care is highly controversial,1,2 an example of Oscar Wilde's quotation, “No good deed goes unpunished.” The readmission rate can be a surrogate marker of the disease severity and possibly the intensity of treatment, not necessarily an indictment of poor care.1,2 The breakdown of 4 common causes for readmission (neurological symptoms, thromboembolism, and infections) is interesting, but given the retrospective nature of the study, there is no distinction between readmissions preventable from superior, upfront planning vs nonpreventable comorbidities of disease progression. It is well known that multiple factors (older age, poor Karnofsky performance score, medical comorbidities, diffuse spread of tumor, eloquent location, steroid dependency) portend a poor outcome. These patients, immune-compromised or hemiparetic, may well require a medically justified readmission for an intercurrent infection, thromboembolism, or neurological symptoms (hydrocephalus, seizures, etc) based on best practices.
It is estimated that in 2014, under section 3025 of the Affordable Care Act (the Hospital Readmission Program), 2000 hospitals will face penalties and lose $300 million in payments.1 Large teaching and safety-net hospitals face steeper penalties under the proposed program.1 Certain readmissions paradoxically reflect exceptionally good care: keeping patients alive who might have otherwise died or providing access to hospitals to those patients who otherwise faced certain death.1 Severity of disease, categorized in various ways, is commonly identified as a major contributor to the risk of readmission; documented adherence to specific interventions designed to improve quality and safety has not been shown to consistently affect readmission frequency.2 The incidence/prevalence of readmissions during the past 5 years, despite efforts to reduce the rate, has remained between 16% (as in the present report) and 20%; a better approach to a study of hospital readmission might be through the use of a complexity science analysis.2 In a complex adaptive system model, there is a necessity to look beyond single causative factors and the expectation of linear responses to interventions.2
Second, the report confirms numerous known prognostic factors but ignores many other factors, biological and clinical, known to determine outcomes.3-6 The majority of patients (84.4%) had 1 or more comorbidities. There were, not surprisingly, statistically significant differences in the readmitted vs nonreadmitted groups in terms of initial length of stay, discharge disposition (routine, home, self-care vs nonroutine, inpatient or outpatient facility, or home health care), and complications at the initial admission. Most tumors (65.3%) were ≥4 cm. Surgery consisted of a gross total resection in 42% of patients and a biopsy in 21.4%. Predictors of long-term survival for patients with glioblastoma include age <65 years (all of the patients in the current report were by definition >65 years of age), Karnofsky performance score of 70, gross total resection, combined chemoradiation (with temozolomide), avoidance of postoperative complications, reoperation on recurrence, low Ki-67 rate, and favorable location (eg, away from the corpus callosum).6 Overall survival for readmitted patients was only 6.0 months; in contrast, the nonreadmitted patients survived 7.6 months (P < .0001). The authors make a dubious comparison to this (marginal) survival advantage of 1.6 months to the gains brought about by the introduction of temozolomide and suggest that if patients could be treated without readmission, the resulting increased survival would be similar to the benefits of adding temozolomide and at a much lower cost.
Third, the authors underestimate the benefit of modern advances in the care of patients with glioblastoma, including the introduction of temozolomide, the impact of clinical trials, and advances in neuroimaging and magnetic resonance—guided neurosurgery. For example, the 2-year survival rate of patients with glioblastoma accrued to research studies increased from 10% to nearly 40% in the decade from 2000 to 2010.7 Changing patterns of care may be responsible for superior outcomes in the post-temozolomide era.9 Two recent studies demonstrate the positive impact of enrolling patients in clinical trials, even when the experimental agent may prove itself to be ineffective, with current median survivals for glioblastoma ranging from 14.5 months10 to 16.1 months8 for patients enrolled in clinical trials, doubling the survival of the nonreadmitted cohort in the present study.
The present report should be a wakeup call for the brain tumor community to develop national consensus, to determine best practices, and to spur clinical translational and cost-effective research to optimize outcomes for patients with glioblastoma. Otherwise, those who will reduce costs and compromise health care, justified by incomplete, stale data, will form policy.
1. Kamerow D. Reassessing hospital readmission penalties. BMJ. 2013;346:f1043. View Full Text | PubMed | CrossRef Cited Here... |
2. Marks E. Complexity science and the readmission dilemma. JAMA Intern Med. 2013;173(8):629–631. Cited Here...
3. Long DM, Gordon T, Bowman H, et al.. Outcome and cost of craniotomy performed to treat tumors in regional academic referral centers. Neurosurgery. 2003;52(5):1056–1063. View Full Text | PubMed | CrossRef Cited Here... |
4. Zinn PO, Sathyan P, Mahajan B, et al.. A novel volume-age-KPS (VAK) glioblastoma classification identifies a prognostic cognate microRNA-gene signature. PLoS One. 2012;7(8):e41522. PubMed | CrossRef Cited Here... |
5. Chaichana KL, Chaichana KK, Olivi A, et al.. Surgical outcomes for older patients with glioblastoma multiforme: preoperative factors associated with decreased survival. J Neurosurg. 2011;114(3):587–594. PubMed | CrossRef Cited Here... |
6. Chaudhry NS, Shah AH, Ferraro N, et al.. Predictors of long-term survival in patients with glioblastoma multiforme: advancements from the last quarter century. Cancer Invest. 2013;31(5):287–308. PubMed | CrossRef Cited Here... |
7. Holdhoff M, Grossman SA. Controversies in the adjuvant therapy of high-grade gliomas. Oncologist. 2011;16(3):351–358. PubMed | CrossRef Cited Here... |
8. Shahar T, Nossek E, Steinberg DM, et al.. The impact of enrollment in clinical trials on survival of patients with glioblastoma. J Clin Neurosci. 2012;19(11):1530–1534. PubMed | CrossRef Cited Here... |
9. Grossman SA, Ye X, Piantadosi S, et al.. Survival of patients with newly diagnosed glioblastoma treated with radiation and temozolomide in research studies in the United States. Clin Cancer Res. 2010;16(8):2443–2449. PubMed Cited Here... |
10. Field KM, Drummond KJ, Yilmaz M, et al.. Clinical trial participation and outcome for patients with glioblastoma: multivariate analysis from a comprehensive dataset. J Clin Neurosci. 2013;20(6):783–789. PubMed | CrossRef Cited Here... |
1. The National Cancer Institute's Surveillance, Epidemiology and Ends Results (SEER) cancer registry is often sued for epidemiological studies of cancer. What proportion of the US population does SEER include?
1. What is the most common type of complication that results in unexpected readmission of GBM patients within 30 days of surgery?
A. Pulmonary embolism
E. Failure to thrive
1. What is the clinical impact of 30-day readmission on survival for GBM patients?
A. Longer overall survival
B. Shorter overall survival
C. Longer progression free survival
D. Shorter progression free survival
Glioblastoma multiforme; Overall survival; 30-Day readmission
Figure. No available...Image Tools
Supplemental Digital Content
Copyright © by the Congress of Neurological Surgeons
Highlight selected keywords in the article text.