Inpatient Cancer Rehabilitation: A Retrospective Comparison of Transfer Back to Acute Care Between Patients With Neoplasm and Other Rehabilitation Patients

Article Outline

• Abstract
• Methods
  • Data
  • Analysis
    • Transfer rate
    • Adjusted analysis of transfers
    • Reason for transfer
  • Statistical Analysis
• Results
• Discussion
  • Study Limitations
• Conclusions
• Acknowledgments
• References
• Copyright

Abstract 

Alam E, Wilson RD, Vargo MM. Inpatient cancer rehabilitation: a retrospective comparison of transfer back to acute care between patients with neoplasm and other rehabilitation patients.

Objective

To determine whether patients with diagnoses of neoplasm undergoing acute rehabilitation differ from other patients in frequency of acute care transfer and type of medical complications.

Design

Retrospective cohort analysis.

Setting

Acute rehabilitation hospital located within an academic medical center.

Participants

Patients with diagnosis of neoplasm (n=40) and patients without neoplasm (n=253) requiring transfer were identified from a database of 2801 rehabilitation discharges over nearly a 4-year period.

Interventions

Not applicable.

Main Outcome Measures

Frequency of unplanned transfer and reasons for the transfer.

Results

Significant difference occurred in overall rate of transfer between patients with neoplasm (21%) and controls (9.7%; P<.001). When evaluated separately for type of neoplasm (with patients receiving corresponding type of rehabilitation as controls), a significantly higher rate of transfer to acute care was found for brain tumor (25% vs 12%; P=.004) and spinal cord neoplasms (23% vs 10%; P=.009), but statistical significance was not reached for other tumor types (12.5% vs 7.4%; P=.19). Patients with stroke with neoplasm as a comorbidity, analyzed separately, with the other patients with stroke as controls, had significantly higher risk of transfer (22% vs 10%; P=.012). Logistic regression analysis found an odds ratio (OR) of 2.5 for unplanned transfer among patients with diagnosis of neoplasm (OR=2.5 for malignancy; OR=2.4 for benign neoplasm). Patients with neoplasm had infection as the most common reason for transfer (28% of the neoplasm transfers vs 18% of other transfers; P=.01), whereas in the nonneoplasm group, cardiopulmonary factors predominated (12% of patients with tumor vs 31% of patients without tumor transfers; P<.001).

Conclusions

In the present sample, patients with neoplasm were more likely to require transfer than patients without neoplasm, although this pattern did not reach statistical significance for noncentral nervous system cases. Overall, patients with neoplasm appear more likely than those without neoplasm to have an infectious cause for unplanned transfer. Increased awareness of this difference may lead to improved medical management on rehabilitation units.

Key Words: Cancer, Rehabilitation

List of Abbreviations: CNS, central nervous system, CI, confidence interval, ICD-9, International Classification of Diseases, 9th Revision, OR, odds ratio, SCI, spinal cord injury

BECAUSE OF THE INTENSITY and comprehensive nature of acute rehabilitation, expectations are very high to maximize patients' function and help them become as independent as possible. Therefore, the ability to tolerate such a program is an important criterion in rehabilitation admission decision-making; however, medical complexity is also considered supportive justification for acute rehabilitation over less intensive settings of care.¹ When medical complexity reaches a point that necessitates transfer back to acute care, it is at best an unavoidable frustration to patients and staff and a strain on resources, and at worst a failure of care or decision-making.

The identification of factors surrounding unplanned transfer from acute rehabilitation has received limited attention. Patients with stroke undergoing acute rehabilitation have been most extensively studied, with risk factors for unplanned transfer including greater disability at time of rehabilitation admission, older age, and possibly race or ethnicity.², ³ One large multicenter study found greater risk of transfer in blacks than whites,² whereas another found similar incidence in black and white men, with both groups at significantly higher risk of unplanned transfer than Asian or Hispanic men; a similar pattern was not observed in women.³ The incidence of medical complications in patients with stroke correlates with severity of disability.⁴ In a large American single-institution series, specific underlying medical problems associated with transfer risk in the stroke rehabilitation population included hypoalbuminemia, hypertension, elevated admission white blood count, and history of cardiac arrhythmia.⁵ Actual reasons for transfer have varied among studies and are difficult to compare because of small numbers. In a Taiwanese series, recurrent stroke predominated,⁶ and in a Canadian series, rare onset of surgical complications.⁷

Among patients with nontraumatic SCI (including a 28% subset of tumor cases) undergoing acute rehabilitation in India, Nair et al⁸ found incidence of medical complications to correlate with more involved neurologic status. Comparing nontraumatic and traumatic spinal cord cases, McKinley et al⁹ found the nontraumatic group to have a lower incidence of many common complications, including pneumonia, deep vein thrombosis, pressure ulceration, autonomic dysreflexia, and orthostatic hypotension, despite older average age, possibly related to lower incidence of complete injury than seen in patients with traumatic injury. A notable exception was that the nontraumatic group had a higher rate of wound infection. Neither of these studies reported rate of transfer back to acute care.

Among a mixed population of brain injury, SCI, stroke, amputation, and other cases, Carney et al¹⁰ found infections (22% of transfers) followed by pulmonary complications (14%) as the most common reasons for unanticipated transfer from acute inpatient rehabilitation, with an 8% overall transfer rate. Age greater than 64 years, SCI, and amputation were risk factors. Deshpande et al¹¹ studied reasons for transfer to acute units among patients with traumatic brain injury and found history of pneumonia or recent surgery during the acute hospital stay to be the most strongly associated factors.

Although previous studies have found similar functional outcomes among patients with and without neoplasm in acute rehabilitation units,¹², ¹³, ¹⁴, ¹⁵ wide disparity has been reported in the rate of transfer for patients with neoplasm. Reported transfer rates range from as high as 33%¹⁶, ¹⁷ to as low as 5%.¹⁸ In a multicenter Dutch series of patients with metastatic epidural spinal cord compression, a 27% acute transfer rate was found; in long-term follow-up, prolonged survival was more likely to be seen in those who had fewer medical complications (and did not need transfer) during the rehabilitation stay.¹⁹ Most previous studies have not compared rate of transfer between patients with and without neoplasm within the same institution and in particular have not examined the reasons for these transfers with significant detail. Identifying the reasons for transfer is a first step toward developing appropriate systems to screen for and manage the complications, especially if they differ in significant ways from complications for other patients in acute rehabilitation. The present study aims to compare the rate of transfer among patients with and without diagnosis of neoplasm, and also compares the reasons for these unanticipated transfers, employing the null hypothesis of no difference in transfer rate or transfer reason.

Back to Article Outline

Methods 

Data 

The dataset consists of first-time, acute rehabilitation discharges at a large, academic rehabilitation facility from 2001 to 2005. Information was obtained from an institutional database that included impairment group, ICD-9 codes, age, sex, race, comorbid conditions, rehabilitation type (derived from the case-mix group number), tier status, admission FIM motor and cognitive scores, and discharge destination. The first author (EA) conducted chart reviews to determine the reason for transfer among the neoplasm group and controls (patients without neoplasm who required unplanned transfer). Cases that were reviewed beyond the level of the departmental database included all patients with neoplasm, and the nonneoplasm cases requiring transfer back to acute care. This study was approved by MetroHealth Medical Center's institutional review board.

Analysis 

All transferred patients were separated into neoplasm and nonneoplasm groups. ICD-9 codes 140 through 239 were used to identify patients with either primary diagnosis or comorbidity of neoplasm. The 2 groups were then divided into categories of brain, spinal cord, stroke, or other areas of rehabilitation type. The controls (patients without neoplasm) were assigned to corresponding categories. Patients with brain impairment codes (traumatic or nontraumatic brain injury) served as controls for the patients with brain neoplasm, and those with traumatic or nontraumatic SCI served as controls for spinal cord neoplasm cases. Because it is conceptually indeterminate whether patients with stroke with comorbidity of neoplasm should be considered brain or other cases, the decision was made to analyze this group separately, using patients with stroke without comorbidity of cancer as controls. The remainder of the transferred patients served as controls for the other neoplasm cases. A secondary analysis of effect of benign or malignant neoplasm on transfer rate within the 4 subgroups was also performed.

Other than exploring the effect of benign versus malignant status, the transfer rate methodology does not formally include the activity of the neoplasm, including whether the cancer was the underlying reason for the rehabilitation admission versus a comorbidity or even remote issue. However, much of this information was obtained in the chart reviews and descriptively summarized.

To determine whether a difference in transfer rate between neoplasm and nonneoplasm groups relates specifically to cancer diagnosis versus confounding factors, we used logistic regression analyses to control for selected major comorbidities (heart disease, diabetes, chronic obstructive pulmonary disease, peripheral vascular disease, deep vein thrombosis, pulmonary embolus, urinary tract infection, pneumonia), type of rehabilitation (brain injury, stroke, SCI, major multiple trauma, neurologic, bone and joint, miscellaneous), tier classification (0, 1, 2, 3), admission FIM (cognitive, motor), and demographic variables including age, sex, and race. This analysis was performed for those with malignant neoplasm (ICD-9 codes 140–208), for benign neoplasm (ICD-9 codes 210–239), and for the group as a whole.

Eight categories of potential medical reasons for transfer were defined, and the medical records of all transferred patients were reviewed by the primary author to identify the reason for transfer in each group. These categories for reason of transfer were cardiopulmonary, infectious, neurologic, surgical, hematologic, endocrine or metabolic, mixed or other, and planned transfer. In cases in which more than 1 category might apply for a given event, the more specific assignment, according to the physiology of the event, was given. For example, pneumonia would be included in the infectious, not cardiopulmonary category, and a neurologic decline requiring a craniotomy would be listed under neurologic, not surgical. In the setting of more than 1 active medical problem at the time of transfer, the case was assigned to the clinical category that represented the major reason for transfer. If this was indeterminate, the case was assigned to the mixed or other category.

Statistical Analysis 

Chi-square analysis was performed to determine the difference in the rate of transfer between the groups as a whole and subcategories (brain, spinal cord, stroke with neoplasm comorbidity, other), where P less than .05 represents a significant difference. For analyses involving small numbers (<10 subjects in a cell), a 2-tailed Fisher exact test was used. Chi-square analysis was employed for the reason for transfer data. For the comorbidity or severity factors, logistic regression analysis was performed, and ORs applying 95% CI were calculated for risk of transfer between neoplasm and nonneoplasm cases. Logistic regression analysis was performed for the neoplasm group as a whole, malignancy alone, benign neoplasm alone, and neoplasm coded by the attending physician in the first 2 diagnostic spots. All analyses were completed with SAS.^a

Back to Article Outline

Results 

There were 3160 discharges during the study period, of which 359 were removed because they were not the initial rehabilitation stay. The final dataset included 2801 first-time discharges (table 1). The malignant neoplasm group was older than controls and also older than patients with neoplasm with benign tumors; no other major demographic differences were found. The prevalence of heart disease was significantly higher in the malignant cases, as was bladder infection in all neoplasm cases. Tier status and admission cognitive and motor FIM scores were comparable between groups.

Table 1. Patient Demographics, Comorbidities, Rehabilitation Type, and Initial Case Severity

Measure	All Neoplasms	Malignant	Benign	No Neoplasm
n (%)	188 (6.8)	130 (4.6)	61 (2.2)	2613 (93.3)
Mean age ± SD (y)	57.7±16.2‡	61.0±15.8‡	50.2±18.8	52.5±18.5
Female (%)	50.3⁎	48.5	54.1	41.9
Race (%)
White	65.2	68.5	59.0	64.1
Black	25.9	22.3	32.8	27.0
Asian	1.0	1.5	0.0	1.0
Hispanic/Latino	1.6	1.5	1.7	3.5
Other	6.3	6.2	6.5	4.4
Comorbidity (%)
Heart disease	65.6	70.0⁎	55.8	58.7
Stroke	36.0	34.6	37.7	39.3
DM	19.1	18.5	19.7	20.3
COPD	12.2	15.4	6.6	11.3
PVD	2.7	3.9	0.0⁎	5.9
Kidney failure	0.5	0.8	0.0	0.5
GI ulcer	2.1	3.1	0.0	1.2
DVT	9.5	9.2	11.5	8.0
PE	2.7	2.3	3.3	1.0
UTI	32.8‡	32.3†	34.4⁎	22.2
Pneumonia	2.7	3.1	1.7	4.8
Rehabilitation type (%)
Brain injury	38.1‡	32.3‡	50.8‡	17.4
SCI	23.3†	20.0	31.2‡	14.6
Stroke	17.5†	20.8	9.8†	26.6
Major polytrauma	1.6‡	2.3‡	0.0‡	15.2
Neurologic	2.1	3.1	0.0	2.0
Bone or joint	10.6†	13.1	4.9†	19.3
Miscellaneous	6.9	8.5	3.3	5.0
Tier (%)
0	65.1	65.4	63.9	71.6
1	5.8	6.2	4.9	4.9
2	13.8	13.9	14.8	11.5
3	15.3	14.6	16.4	11.9
Mean FIM admission score ± SD
Cognition	24.8±9.7	24.5±9.8	25.4±9.5	24.5±10.1
Motor	37.2±11.6	37.5±11.3	36.7±12.5	38.4±14.3

Abbreviations: COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; DVT, deep venous thrombosis; GI, gastrointestinal; PE, pulmonary embolus; PVD, peripheral vascular disease; UTI, urinary tract infection.

Of the 2801 discharges, 188 had a neoplasm diagnosis, and 21% of those required transfer, whereas 253 (9.7%) of the 2613 patients without neoplasm required transfer (P<.001) (table 2). The neoplasm was malignant in 69% of cases, of which 22.8% required transfer, compared with 19.7% of benign neoplasm cases (P=.57). Patients with neoplasm who had unplanned transfer did so after an average of 9.3 days, and controls with unplanned transfer averaged 12.6 days (P=.01). Among 37 patients with metastatic disease, 8 (21.6%) required unplanned transfer, including only 1 of 14 patients with metastatic lung cancer (table 3).

Table 2. Unadjusted and Adjusted Unplanned Transfers

Table 3. Tumor Diagnoses

NOTE. Cases total more than 188 because some patients had history of >1 type of neoplasm.

The neoplasm subcategories consisted of 72 brain cases, 44 spinal cord cases, 32 stroke with neoplasm comorbidity cases, and 40 other cases. There was a difference between those who had rehabilitation needs related to the CNS and those patients with other rehabilitation needs. Significantly higher transfer rate compared with the respective controls without neoplasm was found in the brain (25% vs 12%, P=.004), stroke (22% vs 10%, P=.001), and spinal cord (23% vs 10%, P=.009) groups (fig 1). The other group with neoplasm did not have a significantly higher rate of unplanned transfer (12.5% vs 7.4%, P=.19). The other group, although highly heterogeneous, was composed mainly of nonneurologic cases. Across all subcategories, for both patients with neoplasm and controls, those with more severe disability, as measured by FIM, were more likely to require transfer; severity of disability itself did not account for the differences in transfer rate (table 4).

• 
  • View Large Image
  • Download to PowerPoint

• Fig 1. 

  The percentage of transfers in each group according to rehabilitation type.

Table 4. FIM Characteristics by Subgroup

Subgroup	Motor FIM	Cognitive FIM
Brain
No transfer	37.7±15.2	15.8±7.6
Transfer	26.4±13.8	11.9±7.3
P	<.001⁎	<.001⁎
No neoplasm	36.3±15.9	14.9±7.6
Neoplasm	35.5±12.4	18±8.1
P	.60†	.004†
Spinal cord
No transfer	33.2±14.2	32.3±4.7
Transfer	27.3±11.8	30.8±7.3
P	<.01⁎	<.04†
No neoplasm	32.2±14.3	32.0±5.2
Neoplasm	35.3±11.4	33.1±3.6
P	.09†	.20†
Stroke
No transfer	36.9±14.2	19.9±8.8
Transfer	32.2±13.5	16.4±8.7
P	<.006⁎	<.001⁎
No neoplasm	36.4±14.3	19.5±8.9
Neoplasm	35.6±12.1	19.0±7.6
P	.70⁎	.70⁎
Other
No transfer	47.2±9.3	31.4±5.6
Transfer	35.8±14.4	27.6±10.1
P	<.001⁎	<.03†
No neoplasm	46.5±10.2	31.1±6.1
Neoplasm	42.2±9.9	31.3±6.6
P	.20†	.97†

NOTE. Values are mean ± SD.

There were no significant differences between the rates of unplanned transfer for malignant or benign neoplasm within categories of rehabilitation type. For brain cases, the rates of unplanned transfer between malignant and benign cases were 26.2% versus 25.8% (P=.71), and for spinal cord cases, the rates were 28% versus 15.8% (P=.30) (fig 2). For patients with stroke and comorbidity of neoplasm, those with malignant tumor were transferred at a 22% rate, and benign disease at a 16.4% rate, the difference between which was not significant, likely a result of sample size; the benign group contained only 6 patients, only one of whom transferred. In the other group, 14.3% of those with malignant disease required transfer, and none of the 5 patients with benign disease.

• 
  • View Large Image
  • Download to PowerPoint

• Fig 2. 

  Malignant versus benign tumor transfer rate.

In the adjusted analyses, the OR for unplanned discharge among patients with any type of neoplasm (malignant or benign) was 2.5 (95% CI, 1.7–3.7) (see table 2). Using the same logistic regression but dividing the neoplasm cases into malignant and benign diagnoses, the OR was 2.5 (95% CI, 1.6–4.0) for the malignant cases and 2.4 (95% CI, 1.2–4.8) for the benign cases. Analyzing within groups, the OR for unplanned transfer for stroke cases with comorbidity of neoplasm group was 3 (95% CI, 1.2–7.7), for brain cases 2.8 (95% CI, 1.4–5.6), for spinal cord cases 4 (95% CI, 1.6–9.7), and for the other group not significant, with an OR of 1.2 (95% CI, 0.3–4.5). If any neoplasm was listed among the first 2 diagnostic coding spots by the attending physician, the OR for unplanned transfer was 3.3 (95% CI, 2.0–5.5), and the OR increased to 4.1 (95% CI, 2.3–7.5) for malignant neoplasm being in the first 2 diagnostic coding spots. There was not a significant increase in the odds of unplanned transfer if a benign neoplasm diagnosis was listed in the first 2 diagnostic coding spots (OR=1.9; 95% CI, 0.6–5.9).

The largest differences found for reason of transfer occurred in the infectious and cardiopulmonary categories. Infection was the most common reason for transfer among neoplasm cases (28% of the neoplasm transfers vs 18% of control transfers; P=.01), whereas cardiopulmonary factors predominated in the control group (12% of neoplasm vs 31% of control transfers; P<.001) (fig 3). There were no significant differences seen among the other reasons for transfer. Among all admitted patients, 11 (5.8%) of the 188 patients with neoplasm required transfer for an infectious reason, compared with 1.7% of the 2613 controls (P<.001). With this approach, the neurologic category also showed a significant difference (4.2% of neoplasm cases vs 1.5% of controls; P<.01), but the cardiopulmonary category did not (2.6% of neoplasm cases and 5.8% of controls; P, not significant).

• 
  • View Large Image
  • Download to PowerPoint

• Fig 3. 

  Reason for transfer: neoplasm versus no neoplasm. Abbreviations: Card/pulm, cardiopulmonary; Endo/Metab, endocrine or metabolic; Hemat, hematologic; Neurol, neurologic; Planned, planned transfer.

Although the primary analysis showed a higher rate of bladder infection documented at the time of admission in the patients with neoplasm, and infection was the prevailing reason for transfer, the presence of bladder infection at time of admission was actually not a risk factor for unplanned transfer (19.4% transfer in patients with neoplasm and documented bladder infection vs 22.1% of patients with neoplasm without documented bladder infection; P=.67).

Back to Article Outline

Discussion 

The main aim of this study was to compare the rate of transfer from acute rehabilitation between patients with and without neoplasm, which we found to be approximately 21% and 10%, respectively. The adjusted analysis found patients with neoplasm had about 2.5 times the odds of unplanned transfer to acute medical-surgical floors compared with controls, whether the neoplasm was malignant (OR=2.5) or benign (OR=2.4). Although primary analysis showed that patients with malignant neoplasm were significantly older than controls, this does not explain the difference in transfer rate, because the regression analysis controlled for age. Some interesting patterns were found, including that the increased risk of transfer was greater for patients with neoplasm with neurologic manifestations. This contrasts with data reported by Marciniak et al,¹⁶ who did not find major differences in transfer rate across 4 subgroups, although the groups in that study were defined differently from ours. One possible explanation is that for the large majority of the neurologic patients (all the spinal cord cases and 87% of the brain cases), the neoplasm was the major reason for the rehabilitation admission, whereas in the other and stroke groups, the neoplasm was more variably prominent in the clinical picture.

Little difference was found between those with malignant and benign disease and the need for transfer. Possibly, patients in the other group, some of whom had neoplasm as a remote issue, may have diluted the overall transfer rate for malignant disease. However, the similarity in malignant and benign transfer rate holds true for the brain and spinal cord categories as well, where the neoplasm presented the immediate need for rehabilitation in the vast majority of cases. In the other and stroke groups, the number of patients with benign disease was too low for meaningful examination. Overall, the impact of neurologic disease and its treatment on medical and functional status seems to outweigh any effect of benign or malignant disease.

The similar transfer rate between those with metastatic disease and patients with neoplasm as a whole is also noteworthy. In particular, the ability of the patients with metastatic lung cancer, often considered a sicker population with particularly poor prognosis, to complete their rehabilitation course at a rate well above average was surprising. It is unclear whether selection bias in the referral and admitting process may have been a factor. However, the comparable average admission total FIM scores between the lung cancer cases (60.9) and neoplasm cases overall (62) argue against this possibility.

The second aim was to compare the reasons for these unanticipated transfers between the neoplasm and nonneoplasm populations. Patients with neoplasm had a higher risk of infection as a reason for transfer, whereas in the nonneoplasm group, cardiopulmonary factors predominated. Regarding infection risk, given the high number of neurologically based tumors, there were many patients receiving corticosteroids. It is suspected that facilities with a patient mix weighted toward greater immune compromise (hematologic malignancies, concurrent chemotherapy) might face an even higher incidence of infection and risk of transfer. The cardiopulmonary findings, with patients with neoplasm not having a significantly higher risk in this regard, are reassuring, because patients with malignancy are considered to be in a hypercoagulable state, and on that basis might be expected to be at greater risk for circulatory complications. In addition, our comorbidity data indicate a higher degree of underlying heart disease present in the malignant neoplasm cases than controls, raising concern for cardiopulmonary complications. Although patients with neoplasm were at higher risk of needing transfer because of neurologic decline than controls, it is possible that this figure was skewed by the relative proportion of neurologic cases in the neoplasm versus nonneoplasm groups. Nearly 149 (79%) of 188 neoplasm cases in our population fell into the neurologic rehabilitation categories (brain, spinal cord, or stroke), compared with 1530 (59%) of 2612 control cases. The reason patients with neoplasm and previously documented urinary tract infections do not have a higher rate of unplanned transfer than those without history of urinary tract infection is unclear. Possibly, people with previously diagnosed infections receive greater vigilance, even though those without previously documented infection may be at comparable risk. Alternatively, the antibiotic treatment for the bladder infection may be effectively suppressing new, potentially more serious infections.

Although it was beyond the scope of the present report to study systematically the impact of neoplasm as the primary versus secondary diagnosis in transfer risk, the analysis examining whether the neoplasm diagnosis fell within the top 2 codes documented by the attending physician may provide some approximation. We assumed the attending physician's rankings of the neoplasm diagnosis would provide an estimation of the prominence of the neoplasm, and if the diagnosis fell within the first 2 diagnostic spots, it would indicate a larger role in the clinical picture. The adjusted analysis provided some support, with higher odds of unplanned transfer if any neoplasm was in the first 2 diagnoses (OR=3.3) or for malignant neoplasm within the first 2 diagnostic spots (OR=4.1). An advantage of our methodology is that it allows for potential effects of neoplasm as a comorbidity to manifest, a consideration that is not trivial in this time of improving cancer survivorship.²⁰

Study Limitations 

It is worth noting that the present data reflect a single acute rehabilitation institution, located within a full service academic medical center, and containing an American College of Surgeons−accredited cancer center. The rehabilitation care operates under a model of physiatrist primary management with a very rich range of consultative supports, beyond what is likely available in most freestanding facilities, but without internist or oncologist concurrent care. Although the physiatrists (and the care units as a whole) are subspecialized in brain injury, stroke, SCI, and orthopedic rehabilitation, they do not have specialization toward cancer. In addition, although the range of neoplastic diagnoses is this setting is diverse (see table 3), the patient mix is likely quite different than would be seen at a cancer hospital. Because transfer risk may of course be sensitive to such factors as the service delivery model and patient mix, data from other institutions would be quite useful in determining the most effective care systems for this population.

The central question raised by the present data, given the higher risk of patients with neoplasm requiring transfer back to acute care, is the following: How should rehabilitation care systems respond? At the least, rehabilitation providers must be aware of the higher risk of infection and possibly neurologic change as a reason for transfer compared with other rehabilitation patients, with attention to appropriate surveillance and systems of care mechanisms to meet the need when such complications do occur. Appropriate expectations must also be set among patients, referring providers, and rehabilitation providers. Although rehabilitation providers strive to prevent unanticipated transfers, perhaps they must simultaneously alter the mindset of acute transfer as a failure of care or decision-making, and rather view it as a sometimes unavoidable risk in providing necessary care to complicated patients.

Back to Article Outline

Conclusions 

Rehabilitation patients with neoplasm are more likely to require transfer than patients without neoplasm, with the exception of non-CNS cases. Patients with neoplasm appear more susceptible to infection as the reason for transfer, whereas cardiopulmonary reasons predominate among patients without neoplasm. Development of care strategies that more aggressively prevent and/or treat infection may be warranted. Further study is needed to determine whether these results can be generalized to other inpatient rehabilitation settings, where case mix or service delivery model may be different.

Supplier

Back to Article Outline

Acknowledgments 

We thank John Chae, MD, for his assistance in the development of this project and statistical guidance, and Theresa Fitzgerald, OTR, for assistance with the database.

Back to Article Outline

References 

1. Centers for Medicare and Medicaid Services. Medicare hospital manual. Section 211: inpatient hospital stays for rehabilitation care. http://www.cms.hhs.govAccessed September 28, 2005
   • View In Article
2. Stineman MG, Ross R, Maislin G, Fiedler RC, Granger CV. Risks of hospital transfer and mortality during stroke rehabilitation. Arch Phys Med Rehabil. 2003;84:712–718
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (68 KB)
   • CrossRef
3. Ottenbacher KJ, Smith PM, Illig SB, Fiedler RC, Gonzalez V, Granger CV. Characteristics of persons rehospitalized after stroke rehabilitation. Arch Phys Med Rehabil. 2001;82:1367–1374
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (110 KB)
   • CrossRef
4. Dromerick A, Reding M. Medical and neurological complications during inpatient stroke rehabilitation. Stroke. 1994;25:358–361
   • View In Article
   • MEDLINE
   • CrossRef
5. Roth EJ, Lovell L, Harvey RL, Heinemann AW, Semik P, Diaz S. Incidence of and risk factors for medical complications during stroke rehabilitation. Stroke. 2001;32:523–529
   • View In Article
   • CrossRef
6. Hung JW, Tsay TH, Chang HW, Leong CP, Lau YC. Incidence and risk factors of medical complications during inpatient stroke rehabilitation. Chang Gung Med J. 2005;28:31–38
   • View In Article
7. McLean DE. Medical complications experienced by a cohort of stroke survivors during inpatient, tertiary-level stroke rehabilitation. Arch Phys Med Rehabil. 2004;85:466–469
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (52 KB)
   • CrossRef
8. Nair KP, Taly AB, Maheshwarappa BM. Nontraumatic spinal cord lesions: a prospective study of medical complications during acute rehabilitation. Spinal Cord. 2005;43:558–564
   • View In Article
   • MEDLINE
   • CrossRef
9. McKinley WO, Tewksbury MA, Godbout CJ. Comparison of medical complications following traumatic and nontraumatic spinal cord injury. J Spinal Cord Med. 2002;25:88–93
   • View In Article
   • MEDLINE
10. Carney ML, Ullrich P, Esselman P. Early unplanned transfers from inpatient rehabilitation. Am J Phys Med Rehabil. 2006;85:453–460
    • View In Article
    • MEDLINE
    • CrossRef
11. Deshpande AA, Miller SR, Zafonte RD, Hammond FM, Wood DL. Risk factors for acute care transfer among traumatic brain injury patients. Arch Phys Med Rehabil. 1997;78:350–352
    • View In Article
    • Abstract
    • Full-Text PDF (454 KB)
    • CrossRef
12. Greenberg E, Treger I, Ring H. Rehabilitation outcomes in patients with brain tumors and acute stroke. Am J Phys Med Rehabil. 2006;85:568–573
    • View In Article
    • MEDLINE
    • CrossRef
13. Huang ME, Cifu DX, Keyser-Marcus L. Functional outcome after brain tumor and acute stroke: a comparative analysis. Arch Phys Med Rehabil. 1998;79:1386–1390
    • View In Article
    • Abstract
    • Full-Text PDF (631 KB)
    • CrossRef
14. McKinley WO, Huang ME, Brunsvold KT. Neoplastic versus traumatic spinal cord injury: an outcome comparison after inpatient rehabilitation. Arch Phys Med Rehabil. 1999;80:1253–1257
    • View In Article
    • Abstract
    • Full-Text PDF (698 KB)
    • CrossRef
15. O'Dell ME, Barr K, Spanier D, Warnick RE. Functional outcomes of inpatient rehabilitation in persons with brain tumors. Arch Phys Med Rehabil. 1998;79:1530–1534
    • View In Article
    • Abstract
    • Full-Text PDF (704 KB)
    • CrossRef
16. Marciniak CM, Sliwa JA, Spill G, Heinemann AW, Semik PE. Functional outcome following rehabilitation of the cancer patient. Arch Phys Med Rehabil. 1996;77:54–57
    • View In Article
    • Abstract
    • Full-Text PDF (1647 KB)
    • CrossRef
17. O'Toole D, Golden AM. Evaluating cancer patients for rehabilitation potential. West J Med. 1991;155:384–387
    • View In Article
    • MEDLINE
18. Cole RP, Scialla SJ, Bednarz L. Functional recovery in cancer rehabilitation. Arch Phys Med Rehabil. 2000;81:623–627
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (110 KB)
    • CrossRef
19. Eriks IE, Angenot EL, Lankhorst GJ. Epidural metastatic spinal cord compression: functional outcome and survival after inpatient rehabilitation. Spinal Cord. 2004;42:235–239
    • View In Article
    • MEDLINE
    • CrossRef
20. Committee on Cancer Survivorship. In:  Hewitt M,  Greenfield S,  Stovall E editor. Improving Care and Quality of Life, Institute of Medicine, National Research Council (From cancer patient to cancer survivor, lost in transition). Washington (DC): Natl Acad Pr; 2006;
    • View In Article

• a Version 9.1; SAS Institute, 100 SAS Campus Dr, Cary, NC 27513-2414.