Original Research

Transferring Inpatient Rehabilitation Facility CancerPatients Back to Acute Care (TRIPBAC)Arash Asher, MD, Pamela S. Roberts, PhD, OTR/L, SCFES, FAOTA, CPHQ,Catherine Bresee, MS, Garret Zabel, Richard V. Riggs, MD, Andre Rogatko, PhD

A.A. Samuel Oschin Comprehensive CancerInstitute, Cedars-Sinai Medical Center, 8700Beverly Blvd, Suite AC 1109, Los Angeles, CA90048. Address correspondence to: A.A.;e-mail: arash.asher@cshs.orgDisclosures related to this publication: grant(money to institution), Donna & Jesse GarberAward in Cancer ResearchDisclosures outside this publication: grants/grants pending (money to institution), Co-I fora National Cancer Institute RO1 grant studyingcancer-related fatigue

P.S.R. Department of Rehabilitation, Cedars-Sinai Medical Center, Los Angeles, CADisclosures outside this publication: boardmembership, California Hospital AssociationePost Acute Care, American Congress of

Objective: To determine predictive factors for TRansferring Inpatient rehabilitationfacility (IRF) cancer Patients Back to Acute Care (TRIPBAC).Design: A retrospective chart review of patients with cancer admitted to an IRF from2009 to 2010 because of a functional impairment that developed as a direct consequence oftheir cancer or its treatment.Setting: IRF of a community-based, academic, tertiary care facility.Methods: The characterization of patients with cancer in the IRF was primarily based onanalysis of the IRF Patient Assessment Instrument and other internal IRF data logs.Main Outcome Measurement: Frequency and reasons for TRIPBAC.Results: The TRIPBAC rate in our IRF was 17.4%. The most common reasons forTRIPBAC were postneurosurgical complications (31%). Factors associated with TRIPBACwere a motor Functional Independence Measure score of 35 points or lower on admission(odds ratio 4.01, 95% confidence interval 1.79-8.98; P ¼ .001) and the presence ofa feeding tube or a modified diet (odds ratio 3.18, 95% confidence interval 1.44-7.04;P ¼ .004).Conclusions: Motor Functional Independence Measure score on admission is the bestpredictor for TRIPBAC in patients with cancer admitted to our IRF, followed by thepresence of a feeding tube or a modified diet.

PM R 2014;-:1-6

Rehabilitation Medicine; consultancy, UniformData System for Medical Rehabilitation, com- mission on accreditation for rehabilitation fa-cilities; employment, Cedars-Sinai MedicalCenter, University of Southern California(USC), Trident University International; grants/grants pending (money to institution), sub-contract National Institutes of Health-ICAREstudy, USC primary grant

C.B. Samuel Oschin Comprehensive CancerInstitute, Cedars-Sinai Medical Center, LosAngeles, CADisclosure: nothing to disclose

G.Z. Department of Rehabilitation, Cedars-Sinai Medical Center, Los Angeles, CADisclosure: nothing to disclose

R.V.R. Department of Rehabilitation, Cedars-Sinai Medical Center, Los Angeles, CADisclosure: nothing to disclose

A.R. Samuel Oschin Comprehensive CancerInstitute, Cedars-Sinai Medical Center, LosAngeles, CADisclosures related to this publication: grant(money to author and institution), NIHDisclosures outside this publication: employ-ment, Cedars Sinai Medical Center; grants/grants pending (money to author and institu-tion), NIHResearch support: Donna & Jesse GarberAward in Cancer Research

Submitted for publication October 24, 2013;accepted January 12, 2014.

INTRODUCTION

Cancer and its treatment sequelae can have a profound influence on a person’s functionalstatus and quality of life. Historically, the rehabilitation of patients with cancer wasmarginalized because of the perception that cancer represented an inevitably progressiveand terminal disease [1]. In 2012, the incidence of cancer cases in the United States wasestimated at more than 1.6 million, with approximately 14 million cancer survivors alivetoday [2]. As cancer treatments have become more sophisticated and mortality rates inisolation are no longer an adequate outcome measure, issues related to quality of life andfunction have become increasingly more relevant for patients and clinicians. For manypatients, the fear of death is outweighed by the fear of loss of independence and functionaldecline [3]. Rehabilitation services can help patients with cancer and long-term survivorsregain and improve physical, psychosocial, and vocational functioning within the limita-tions imposed by the disease and its treatment [4-12].

With improved survival rates and an increasing number of cancer survivors, a myriad ofissues related to the rehabilitation of patients with cancer have arisen. For patients whoexperience the most profound functional decline as a result of their cancer or its treatment,evidence and guidelines for the most appropriate and cost-effective use of rehabilitationservices are needed. The most intensive level of rehabilitation, known as “acute rehabili-tation,” is furnished at an inpatient rehabilitation facility (IRF) [13]. Care at an IRF providesan intense level of daily rehabilitation that typically focuses on improving overall functionand integrates comprehensive medical care such as pain management, wound care, res-piratory care, and psychological services.

PM&R1934-1482/14/$36.00

Printed in U.S.A.

ª 2014 by the American Academy of Physical Medicine and RehabilitationVol. -, 1-6, - 2014

http://dx.doi.org/10.1016/j.pmrj.2014.01.0091

2 Asher et al TRANSFERRING IRF CANCER PATIENTS BACK TO ACUTE CARE

Insurers and IRFs have admission criteria to identifypatients who would most likely benefit from care at the IRFlevel [14]. Two key requirements include the need for closemedical supervision by a physician with specialized trainingin rehabilitation and the ability of the patient to toleratea therapy program consisting of at least 3 hours of therapyper day, 5 days per week (ie, the “3-hour rule”). Theserequirements can result in a difficult predicament for phys-iatrists, in that the medical problem must be significantenough to require physician supervision but stable enoughto not interfere with the patient’s ability to attend andtolerate an average of 3 hours of rehabilitation per day.

It is thus important for physicians who make decisionsregarding the candidacy of patients with cancer for IRFadmission to identify patients who can complete the pro-grammatic requirements for an IRF. Compared with patientswho have diagnoses other than cancer, research hasdemonstrated a higher rate of transfer of patients with cancerfrom the IRF back to acute care as the result of medicalproblems and complications, with rates ranging from21%-41% [9,11,15,16]. In contrast, 10%-19%, 10%-11%,and 12%-22% of patients with stroke, traumatic spinal cordinjury, and traumatic brain injury, respectively, requiredtransfer back to acute care [15,17-19].

A higher transfer rate from an IRF is an indication ofmedical instability or specialized medical or proceduralinterventions. Patients with cancer generally tend to have amore inherently unpredictable medical course [5,9,11,13].Identifying risk factors that might predict a poor likelihoodof successfully completing an inpatient rehabilitation coursewould be very valuable for physiatrists, allowing them tomake evidence-based decisions regarding successful inpa-tient rehabilitation admission for patients with cancer. Thisability is also important because of its impact on coordina-tion of care and decision making regarding the use of limitedresources.

A retrospective study completed within a rehabilitationunit of a comprehensive cancer center identified severalsignificant risk factors associated with TRansferring IRFcancer Patients Back to Acute Care (TRIPBAC): a low albu-min level, an elevated creatinine level, tube feeding, and useof a Foley catheter [9]. However, this tertiary cancer center isatypical in that it carries a diagnostic criteria waiver.Therefore the IRF at this hospital did not have to follow the75% rule (later changed to the 60% rule). In brief, the 75%rule put forth by the Centers for Medicare and MedicaidServices (CMS) stipulated that at least 75% of the IRF’soverall patient population must have any one of 13 specifiedconditions [20]. Cancer is not one of the 13 conditions,although some patients such as those with brain tumors,spinal cord tumors, amputation, or pathologic hip fracturesfall within the 13 CMS categories. Many patients with cancerexperience problems such as severe deconditioning, gait andmobility problems, deficits in activities of daily living,generalized weakness, and asthenia that are not included

among the 13 specified conditions [6]. Therefore thesepatients are less often candidates for IRF admission in mosthospitals throughout the United States that do not have anexemption for the diagnostic criteria waiver.

The predictive value of the factors identified in the studyby Guo et al [9] at the tertiary cancer center has not beenevaluated for patients with cancer in a hospital that isincluded in the CMS 60% rule. Therefore more research isneeded to determine whether these clinical factors areassociated with transfer out of an IRF in a setting outside ofan exclusive cancer institution. Similarly, categorizing andconfirming the most common reasons for TRIPBAC mayimprove awareness of the most significant medical compli-cations that can affect the rehabilitative potential of theirpatients with cancer. In addition, anticipatory strategies canbe implemented to prevent and manage these complications.

The goal of this retrospective study was to answer thefollowing questions regarding patients with cancer in atypical IRF:

1. What are the characteristics of patients with cancer in theIRF?

2. What is the frequency of TRIPBAC?3. What are the reasons for TRIPBAC?4. Are any laboratory and clinical factors associated with

TRIPBAC?

We hypothesized that among cancer rehabilitationpatients of a typical IRF, an elevated level of creatinine, a lowlevel of albumin, tube feeding, or use of a Foley catheter canpredict transfer to acute care from the IRF.

METHODS

After obtaining approval from our institutional review board,we used a retrospective design to perform this study. The IRFPatient Assessment Instrument (IRF-PAI), which representsthe source of the majority of the information for this study, isrequired to be fully completed per Medicare regulations forevery licensed IRF in the United States [21]. All staff memberswho enter data into the IRF-PAI at our institution have beencredentialed through the Uniform Data System for MedicalRehabilitation [22]. Etiologic diagnosis and comorbid con-ditions from the IRF-PAI are completed by certified coderswith the use of standard coding guidelines.

Study Population

Inclusion Criteria. Patients with any cancer diagnosis whowere admitted for the first time to the IRF at our institutionfrom 2009 through 2010 were reviewed for study eligibility.Our institution is a community-based academic tertiary carefacility that includes an inpatient rehabilitation unit withinthe hospital. Patients were screened for eligibility in reversechronologic order from December 31, 2010. We extracted

PM&R Vol. -, Iss. -, 2014 3

these diagnostic data from the IRF-PAI by reviewing theetiologic diagnosis and comorbid conditions.

Exclusion Criteria. Only patients who were admitted to theIRF because of a functional impairment that developed as adirect consequence of their cancer or its treatment wereanalyzed. For example, a patient with a previous history ofbreast cancer in remission who was admitted to the IRF fortraumatic brain injury rehabilitation unrelated to her previ-ous cancer diagnosis did not qualify. The inclusion andexclusion criteria were applied by a physiatrist, whoreviewed each case individually. A sample of patients whowere screened for the study was reviewed by a secondphysiatrist to confirm that the exclusion and inclusioncriteria were met. A 100% concordance rate was foundbetween the 2 clinicians.

Data Collected

To characterize the patients with cancer in the IRF, thefollowing information was collected from the IRF-PAI:primary cancer diagnosis, complications during the IRFstay, age, ethnicity, gender, and marital status. The fre-quency and reasons for TRIPBAC also were obtained fromthe IRF-PAI. Reasons for TRIPBAC were grouped into thefollowing categories: oncology treatment (chemotherapy,surgery, or radiation therapy) or an unanticipated med-ical complication attributable to cardiac complications,neurosurgical complications, infection, seizure, venousthromboembolism, respiratory distress, gastrointestinalcomplications, wound complications, change in mentalstatus, tumor recurrence or progression, or other. Thereason for transfer was determined by a review of the pa-tient chart. The following information was collected todetermine whether any laboratory and clinical factors wereassociated with TRIPBAC:

1. Length of stay, in days, in acute care before admission tothe IRF; these data were collected from the IRF-PAI.

2. Laboratory values (ie, blood urea nitrogen, creatinine,albumin, prealbumin, hemoglobin, sodium, potassium,chloride, bicarbonate, calcium, absolute neutrophilcount, white blood cell count, and platelet count) uponadmission to the IRF; these data were obtained from thepatient chart.

3. The presence or absence of metastatic disease status wasobtained by reviewing documentation from the patientchart.

4. The presence of a Foley catheter and use of tube feedingupon admission to the IRF were obtained from the IRF-PAI.

5. Functional Independence Measure (FIM) total, motor,and cognitive scores upon admission to the IRF from theUDS-PRO database (The UDS-PRO System; Amherst,NY).

Statistical Methods

All data were first inspected graphically to confirm overalldata distributions and investigate for outliers. Data are pre-sented as counts and percentages or means and standarddeviations. Correlations between variables were investigatedby way of either Pearson or Spearman Rank correlationcoefficients where data were parametric or nonparametric,respectively. Differences between groups were tested by wayof a Fisher exact test for categorical data or t-tests for dif-ferences in group means. Logistic regression modeling wasused to examine variables predictive of TRPIBAC incidenceby the use of exact methods where low counts wereobserved. Investigation of receiver operating characteristiccurves was used to determine the accuracy of predictingTRIPBAC from patient variables on admissions. A poweranalysis performed before data collection indicated thatour sample size should have 80% power to detect oddsratios of >2.3 for categorical variables and >1.6 forcontinuous measures. All tests of significance were set toP < .05. Analysis was performed with use of SAS, version9.3 (Cary, NC).

RESULTS

Data were available from 184 unique patients with first-timeadmissions to the IRF from 2009 to 2010. The patients’demographic and clinical characteristics are summarized inTable 1. The greatest number of patients (49%) had a pri-mary brain tumor. Nevertheless, no increased risk of TRIP-BAC was identified among any specific category of patientswith cancer. The average age of patients admitted to the IRFwas 61.4 years (standard deviation, 15.9 years). The medianlength of stay for previous acute care was 8 days, with arange of 0-137 days. The median length of stay in the IRFwas 10.5 days, with a range of 0.5-44 days. A total of 31% ofthe patients had metastatic disease, 22% had a Foley catheterupon admission, and 5% had a feeding tube upon admis-sion. Table 2 summarizes the laboratory values of patients atthe time of admission to the rehabilitation unit.

Overall, we observed a TRIPBAC rate of 17.4% (95%confidence interval [CI], 11.9-22.9), which is significantlygreater than the overall national 2010 rate of 10.3%(P ¼ .003) and our hospital TRIPBAC rate of 9.9% [22].Reasons for TRIPBAC are summarized in Table 3. The mostfrequent reason for TRIPBAC was related to post-neurosurgical complications (eg, hematoma, cerebrospinalfluid leak, and hydrocephalus).

Table 4 summarizes the risks associated with TRIPBACfrom a variety of clinical factors. None of the followingfactors was associated with TRIPBAC: gender, tumor diag-nosis, presence of metastatic disease or the use of a Foleycatheter. No laboratory value at the time of admission to therehabilitation unit, including albumin, sodium, potassium,white blood cell count, hemoglobin, or platelet count, was

Table 1. Patient characteristics (n ¼ 184)

Characteristics

Age, years, mean (SD) 61 (16)GenderMale 95 (51)Female 92 (49)

EthnicityWhite 151 (82)Black 19 (10)Asian 13 (7)Pacific 1(1)

Marital statusMarried 111 (60)Never married 34 (18)Divorced/separated 24 (13)Widowed 15 (8)

Tumor diagnosisPrimary brain 90 (49)Lung 16 (9)Primary spinal cord 14 (8)Gastrointestinal 12 (7)Hematologic 12 (7)Prostate 9 (5)Melanoma 8 (4)Breast 7 (4)Renal cell 5 (3)Other 11 (6)

Metastatic disease 56 (31)Feeding tube or modified diet 46 (25)Foley catheter 41 (22)Rehabilitation admission impairmentBrain dysfunction 110 (60)Spinal cord dysfunction 30 (16)Orthopedic disorder 17 (9)Stroke 9 (5)Neurologic condition 7 (4)Debility 5 (3)Other 6 (3)

FIM scores (mean, SD)*Total 63.1 (15.5)Motor 39.4 (12.1)Cognition 23.7 (7.0)

Previous hospital admission, days (mean, SD) 12 (15)

Values are n (%) unless otherwise specified.SD ¼ standard deviation; FIM ¼ Functional Independence Measure.*Note that lower scores indicate a greater burden of care.

Table 2. Laboratory values at rehabilitation admission

No. of Patients (%)

AboveNormal Normal

BelowNormal

Albumin 0 (0) 92 (50) 92 (50)Sodium 0 (0) 156 (85) 28 (15)Chloride 7 (4) 166 (90) 11 (6)Potassium 0 (0) 168 (91) 16 (9)Bicarbonate 10 (5) 160 (87) 14 (8)Calcium 4 (2) 141 (77) 39 (21)Blood urea nitrogen 36 (20) 145 (79) 3 (2)Creatinine 6 (3) 173 (94) 5 (3)WBC count 0 (0) 171 (93) 13 (7)Absolute neutrophilcount

62 (34) 116 (63) 6 (3)

Hemoglobin 1 (1) 40 (22) 143 (78)Platelets 8 (4) 142 (77) 34 (18)

WBC ¼ white blood cell.

Table 3. Reasons for transferring inpatient rehabilitation facilitycancer patients back to acute care (n ¼ 32)

n %

Neurosurgical complication 10 31Pathologic fracture 3 9Mental status change 3 9Oncology treatment 3 9Infection 3 9Seizure 2 6Gastrointestinal 2 6Cardiac 2 6Other* 4 13

*Other includes venous thromboembolism, hematoma, wound complication,and anemia.

4 Asher et al TRANSFERRING IRF CANCER PATIENTS BACK TO ACUTE CARE

associated with TRIPBAC. The presence of a feeding tube ora modified diet was predictive of TRIPBAC (odds ratio [OR]3.18, 95% CI 1.44-7.04; P ¼ .004).

The motor FIM score at the time of admission was themost predictive of TRIPBAC (P ¼ .002). On the basis of theexamination of the receiver operating characteristic curve, acutoff of 35 points or lower on the motor FIM score was thebest predictor of TRIPBAC in our sample, with an area underthe curve of 0.68. Although the only other strong predictivefactor of TRIPBAC was the presence of a feeding tube ora modified diet on admission, neither this nor any otherfactor in combination with motor FIM score <35 in a multi-variable model improved the prediction of TRIPBAC morethan motor FIM score alone (OR 4.01, 95% CI 1.79-8.95;

P ¼ .001). This finding is likely attributable to the strongintercorrelation of low motor FIM scores with the frequencyof modified feeding methods (rs ¼ 0.30, P < .001).

DISCUSSION

Our study demonstrated that the motor FIM score alone isthe best predictor of transferring patients with cancer from atypical IRF back to acute care. Unlike Guo et al [9], we didnot find that a low level of albumin, an elevated creatininelevel, or the presence of a Foley catheter was predictive ofTRIPBAC. We did find that the presence of a feeding tube ora modified diet was predictive of TRIPBAC, similar to thefindings of Guo et al. However, no other factors in combi-nation were more predictive than a low motor FIM score. Toour knowledge, this study is the first to evaluate admissionFIM scores as an independent risk factor for TRIPBACamong patients with cancer in the IRF.

The FIM was created to develop a universal language fordescribing function and outcomes. Today, the FIM is oneof the most widely used tools in the United States [23]. Thetool is designed to measure the “burden of care,” with

Table 4. Risk factors for transferring inpatient rehabilitation fa-cility cancer patients back to acute care

OR 95% CI P Value

Age, y 0.99 (0.96-1.01) .299Male (vs female) 1.10 (0.51-2.37) .800Tumor diagnosisBrain Ref .651Thoracic 1.25 (0.32-4.97)Spinal 0.42 (0.05-3.45)Gastrointestinal 2.71 (0.72-10.25)Lymphoma/myeloma 1.09 (0.22-5.50)All others 1.36 (0.52-3.55)

Metastatic disease 1.72 (0.78-3.78) .179Feeding tube or modified diet 3.18 (1.44-7.04) .004*Foley catheter 1.77 (0.76-4.13) .184Rehabilitation admission impairmentBrain dysfunction Ref .364Spinal cord dysfunction 1.86 (0.72-4.83)Orthopedic disorder 0.32 (0.04-2.56)All other conditions 1.16 (0.39-3.47)

FIM instrument scoresTotal 0.97 (0.94-0.99) .006*Motor 0.96 (0.91-0.98) .002*Cognition 0.98 (0.93-1.04) .544

Previous hospital stay, days 1.01 (0.99-1.03) .301Blood chemistry concentrationAlbumin 1.21 (0.52-2.84) .656Sodium 0.93 (0.82-1.05) .217Chloride 0.91 (0.82-1.01) .083Potassium 0.81 (0.37-1.78) .597Bicarbonate 1.04 (0.92-1.18) .562Calcium 1.33 (0.78-2.29) .294Blood urea nitrogen 1.01 (0.97-1.05) .760Creatinine 1.15 (0.48-2.75) .753WBC 1.09 (0.94-1.25) .257Absolute neutrophil count 0.98 (0.89-1.07) .593Hemoglobin 0.92 (0.74-1.16) .488Platelets 1.00 (1.00-1.00) .437

OR ¼ odds ratio; CI ¼ confidence interval; Ref ¼ reference; FIM ¼ FunctionalIndependence Measure; WBC ¼ white blood cell.*Statistically significant.

PM&R Vol. -, Iss. -, 2014 5

lower scores indicating a greater level of dependence. Anextensive body of research has found the FIM to be reli-able, valid, and responsive to change [23-25]. A low motorFIM score may therefore represent a reasonable tool toidentify high-risk cancer rehabilitation patients. A numberof explanations could account for this observation. Patientswho are less mobile may be more likely to experiencecomplications related to immobility, such as deep venousthrombosis, decubitus ulcers, autonomic dysfunction, andfalls. In addition, patients who are less mobile are alsolikely to have other medical comorbidities that may makethem more medically unstable, such as bone pain that mayimpede walking but also increase the chance of pathologicfractures. Although the reasons why a low motor FIM scoreincreases TRIPBAC are not evaluated in this study, theutility in recognizing a low motor FIM as a tool to identifyhigh-risk cancer rehabilitation patients warrants furtherstudy.

This study also underscores the differences between thecancer population of an IRF from a hospital that has adiagnostic criteria waiver and is part of a national compre-hensive cancer center compared with a typical IRF. Becauseof the 60% rule, fewer patients with asthenia are generallyadmitted to a typical IRF. In this study, 85% of the patientswere admitted because of a neurologic consequence of theircancer such as brain or spinal cord metastasis, underscoringthe neurorehabilitative nature of this population. This situ-ation may possibly explain why the predictive factors of thestudy by Guo et al [9] were not confirmed with our popu-lation, which may be more representative of IRFs across theUnited States. In addition, our results may be more repre-sentative for IRFs within a tertiary academic institution withreadily available oncologic support. The TRIPBAC rate andreasons for TRIPBAC may be unique in a freestanding IRF,depending on the availability of supportive and oncologicacute care services within this medically complex patientpopulation. Nevertheless, this study adds to the recentliterature that is demonstrating that the vast proportioncancer rehabilitation patients are being successfully dis-charged home from the IRF [26].

As with most retrospective study designs, the difficulty incontrolling bias and confounders needs to be consideredwith the interpretation of the results. First, establishing causeand effect is not possible. For example, it is impossible toconclude that a low motor FIM score on admission directlyresults in poor IRF candidacy. However, low functioning orlow FIM scores have been shown to be a predictor forreadmission in multiple other IRF populations [18,27-29].Second, although general guidelines exist to guide physiat-rists’ clinical decision regarding IRF candidacy, this decisionstill represents a highly subjective process. It can be influ-enced by many factors such as the individual physiatrist’sclinical experience with patients who have cancer, his or herworking relationship with the referring oncology team, andthe physiatrist’s sense of the patient’s and the family’smotivation in the rehabilitation process. These factors mayrepresent a selection bias that may influence admissioncandidacy. Finally, some physiatrists may be more com-fortable in managing the acute medical issues that may occurduring the IRF stay and may have a higher threshold beforetransferring a patient back to acute care. These more intan-gible factors cannot be controlled for in this retrospectivestudy and need to be considered in the interpretation ofthe results.

CONCLUSION

Inpatient rehabilitation services can be highly effectivein improving functional independence and communityreintegration. In the current economic milieu that demandsevidence for quality and cost-effective medical care, pro-spective studies are needed to help guide clinicians in theevidence-based use of rehabilitative services across the

6 Asher et al TRANSFERRING IRF CANCER PATIENTS BACK TO ACUTE CARE

cancer trajectory. This study lays a foundation for learningmore about historically higher TRIPBAC rates in a typicalIRF.

On the basis of the results of this study, anticipatorystrategies may be developed and integrated into the reha-bilitation plan for patients with cancer who have knownpredictors for a poor IRF outcome. These anticipatory stra-tegies will then need to be evaluated in prospective studiesand may prove to both improve clinical care via the imple-mentation of preventative strategies to minimize complica-tions in the IRF and decrease cost by refining appropriatepatient selection for the IRF. In addition, future prospectivestudies will need to compare functional outcomes, long-termquality of life measures, and costs of rehabilitating patientswith cancer in various settings: IRFs, skilled nursing facil-ities, or as consultative model within the oncology unit.Furthermore, prospective studies are needed to evaluate theeffectiveness of various screening methods to trigger reha-bilitation assessment and therefore improve rehabilitationaccess for oncology patients. Finally, more research isneeded to evaluate the access of minority and socioeco-nomically disadvantaged patients with cancer for inpatientrehabilitative services. Even though clinicians may berecognizing their rehabilitation needs, it is unclear whetherthese patients have adequate access because of issues relatedto lack of transportation for family members, conflicting liferesponsibilities, adequate social support to ensure safedischarge, and insurance denials.

REFERENCES1. Cheville AL. Cancer rehabilitation. Semin Oncol 2005;32:219-224.2. Cancer Facts and Figures 2012. American Cancer Society. Available at

http://www.cancer.org/acs/groups/content/@epidemiologysurveilance/documents/document/acspc-031941.pdf. Accessed January 27, 2014.

3. Cheville A. Rehabilitation of patients with advanced cancer. Cancer2001;92:1039-1048.

4. Cromes G. Implementation of interdisciplinary cancer rehabilitation.Rehabil Counseling Bull 1978;21:230-237.

5. DeLisa JA. A history of cancer rehabilitation. Cancer 2001;92:970-974.6. Movsas SB, Chang VT, Tunkel RS, Shah VV, Ryan LS, Millis SR.

Rehabilitation needs of an inpatient medical oncology unit. Arch PhysMed Rehabil 2003;84:1642-1646.

7. Fialka-Moser V, Crevenna R, Korpan M, Quittan M. Cancer rehabili-tation: Particularly with aspects on physical impairments. J RehabilMed 2003;35:153-162.

8. Gerber L, Vargo N, Smith R. Rehabilitation of the cancer patient. In:DeVita VT Jr, Lawrence LS, Rosenberg SA, eds. Cancer: Principles andPractice of Oncology. 7th ed. Philadelphia, PA: Lippincott, Williams &Wilkins; 2005, 2719-2746.

9. Guo Y, Persyn L, Palmer JL, Bruera E. Incidence of and risk factors fortransferring cancer patients from rehabilitation to acute care units. Am JPhys Med Rehabil 2008;87:647-653.

10. Ganz PA. The status of cancer rehabilitation in the late 1990s. MayoClin Proc 1999;74:939-940.

11. Marciniak CM, Sliwa JA, Spill G, Heinemann AW, Semik PE. Func-tional outcome following rehabilitation of the cancer patient. Arch PhysMed Rehabil 1996;77:54-57.

12. Dietz JH Jr. Rehabilitation of the cancer patient. Med Clin North Am1969;53:607-624.

13. Hewitt M, Maxwell S, Vargo MM. Policy issues related to the rehabil-itation of the surgical cancer patient. J Surg Oncol 2007;95:370-385.

14. Centers for Medicare and Medicaid Services. Chapter 15: CoveredMedical and Other Health Services. In: Medicare Benefit Policy Manual.Available at: http://www.cms.gov/Regulations-and-Guidance/Guidance/Manuals/downloads/bp102c15.pdf. Accessed January 27, 2014.

15. Alam E, Wilson RD, Vargo MM. Inpatient cancer rehabilitation: Aretrospective comparison of transfer back to acute care between pa-tients with neoplasm and other rehabilitation patients. Arch Phys MedRehabil 2008;89:1284-1289.

16. Fu JB, Lee J, Smith DW, Guo Y, Bruera E. Return to primary serviceamong bone marrow transplant rehabilitation inpatients: An index forpredicting outcomes. Arch Phys Med Rehabil 2013;94:356-361.

17. Roth EJ, Lovell L, Harvey RL, Heinemann AW, Semik P, Diaz S.Incidence of and risk factors for medical complications during strokerehabilitation. Stroke 2001;32:523-529.

18. Hammond FM, Horn SD, Smout RJ, et al. Acute rehospitalizationsduring inpatient rehabilitation for spinal cord injury. Arch Phys MedRehabil 2013;94(suppl 4):S98-S105.

19. Deshpande AA, Millis SR, Zafonte RD, Hammond FM, Wood DL. Riskfactors for acute care transfer among traumatic brain injury patients.Arch Phys Med Rehabil 1997;78:350-352.

20. Clark J, Ford S, Hegedus P. Developing a comprehensive cancer centerrehabilitation program. J Oncol Manag 2004;13:13-21.

21. Centers for Medicaid and Medicare Services. Inpatient RehabilitationFacility-Patient Assessment Instrument Training Manual, 2004. Avail-able at http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/InpatientRehabFacPPS/downloads/irfpaimanual040104.pdf. AccessedJanuary 27, 2014.

22. Uniform Data System for Medical Rehabilitation. The UDS-PRO�System. Amherst, NY: Uniform Data System for Medical Rehabilitation;2010.

23. Cournan M. Use of the functional independence measure for outcomemeasurements in acute inpatient rehabilitation. Rehabil Nurs 2011;36:111-117.

24. Cohen ME, Marino RJ. The tools of disability outcomes researchfunctional status measures. Arch Phys Med Rehabil 2000;81:S21-S29.

25. Stineman MG, Shea JA, Jette A, et al. The Functional IndependenceMeasure: Tests of scaling assumptions, structure, and reliability across20 diverse impairment categories. Arch Phys Med Rehabil 1996;77:1101-1108.

26. Shin KY, Guo Y, Konzen B, Fu J, Yadav R, Bruera E. Inpatient cancerrehabilitation: The experience of a national comprehensive cancercenter. Am J Phys Med Rehabil 2011;90(suppl 5):S63-S68.

27. Roberts PS, Divita MA, Riggs RV, Niewczyk P, Bergquist B, Granger CV.Risk factors for discharge to an acute care hospital from inpatientrehabilitation among stroke patients. PM R 2014;6:50-55.

28. Hoyer EH, Needham DM, Miller J, Deutschendorf A, Friedman M,Brotman DJ. Functional status impairment is associated with un-planned readmissions. Arch Phys Med Rehabil 2013;94:1951-1958.

29. Schneider JC, Gerrard P, Goldstein R, et al. Predictors of transfer fromrehabilitation to acute care in burn injuries. J Trauma Acute Care Surg2012;73:1596-1601.