Patient outcomes and length of stay in a stroke unit offering both acute and rehabilitation services

Yan H Ang, Daniel K Y Chan, Qing Shen and Derrick M K Heng
Med J Aust 2003; 178 (7): 333-336. || doi: 10.5694/j.1326-5377.2003.tb05225.x
Published online: 7 April 2003


Objectives: To compare hospital length of stay (LOS) and outcome after stroke between patients in a stroke unit offering combined acute and rehabilitation services and patients treated elsewhere in New South Wales.

Design: Retrospective audit of two hospital databases (Diagnosis-Related Groups [DRG] database and Australian National Subacute Non-Acute Patient Classification System [AN-SNAP] database), with comparison with DRG and AN-SNAP data for NSW.

Setting and participants: 242 episodes of acute stroke in patients admitted to the stroke unit of a metropolitan teaching hospital between July 1999 and November 2000, 113 of whom also underwent rehabilitation in the unit; 9777 episodes of acute stroke in the NSW DRG database, and 2350 in the NSW AN-SNAP database.

Main outcome measures: Acute and rehabilitation LOS; mortality in acute care; FIM (Functional Independence Measure) score at discharge and change in FIM score; and discharge destination.

Results: Patients in the combined stroke unit had shorter LOS and better functional outcome in all DRG and AN-SNAP groups, with both higher discharge FIM scores and greater gain in FIM scores than NSW patients. Acute stroke mortality of 12% and nursing home admission rate of 15.5% in the combined stroke unit were not significantly different from rates for NSW (15.7% and 11.2%, respectively).

Conclusions: Combining acute and rehabilitation services in a stroke unit may reduce LOS and improve functional outcome of patients with acute stroke.

It is well established that treating patients with stroke in specialised stroke units reduces mortality,1-7 institutionalisation1,2,6 and length of hospital stay,1,5 and improves functional status.2,5-7 However, there are many different models of stroke unit, and few data on the model that gives the best outcomes. The latest analysis by the Stroke Unit Trialists' Collaboration (Cochrane Library) found greater benefits in units based in a discrete ward,8 but these results were not statistically convincing, and more information is required.

Bankstown–Lidcombe Hospital is unique in New South Wales in having a comprehensive, dedicated stroke unit that provides both acute and rehabilitative care. Most other stroke rehabilitation units in the State are mixed rehabilitation wards (which treat a range of patients with disabling illness, not only stroke patients), and no others combine acute stroke management and rehabilitation. Comprehensive, dedicated stroke units have the potential to reduce length of stay and improve functional outcome, as rehabilitation can be started earlier, without the need for patient transfer from acute to rehabilitation facilities. However, the potential advantages of this model of stroke unit have not been adequately explored.

Our study aimed to compare length of stay and outcomes of rehabilitation between patients managed in the stroke unit at Bankstown–Lidcombe Hospital and those managed in other stroke rehabilitation units in NSW. A secondary aim was to compare length of stay and mortality between patients receiving acute care in the stroke unit and those managed elsewhere in NSW.


This was a retrospective study, drawing on casemix information. It was approved by the Ethics Committee of the Southwestern Sydney Area Health Service.

Bankstown–Lidcombe Stroke Unit

A comprehensive, dedicated stroke unit was established in May 1998 at Bankstown–Lidcombe Hospital (a 450-bed, metropolitan teaching hospital in Sydney, NSW). The unit has 20 beds and is staffed by a multidisciplinary team of neurologists, geriatricians, a rehabilitation physician, nurses, physiotherapists, occupational therapists, speech pathologists, dietitians, social workers and an orthoptist. Patients diagnosed with stroke are admitted directly to the unit from the emergency department, unless the unit is already full. As well as acute medical management, the unit provides rehabilitation as soon as the patient's medical condition allows, often starting on Day 1 or 2 of admission. Multidisciplinary meetings are held weekly to plan rehabilitation goals, discuss patient progress and coordinate discharge planning.

Data sources and extraction

Information on all patients admitted to the stroke unit with acute stroke between July 1999 and November 2000 was extracted from two databases of Bankstown–Lidcombe Hospital. Information on acute stroke episodes was obtained from the DRG (Diagnosis-Related Groups) database and on rehabilitation from the AN-SNAP (Australian National Subacute Non-Acute Patient Casemix Classification System) database.

The Australian Refined Diagnosis Related Group (AR-DRG) classification system9 is used to describe the acute inpatient casemix of a hospital. Version 4.1 of the system, introduced in July 1999, defines acute stroke as classes B70A–D (Box 1). These incorporate the principal ICD-10 diagnoses of all types of ischaemic stroke (including cortical, subcortical, brainstem and lacunar infarcts) and non-traumatic primary haemorrhagic strokes.10 Data were extracted from the DRG database on patient demographic and clinical characteristics, DRG class, LOS, mortality and hospital readmissions within 28 days.

The AN-SNAP casemix classification system11 has been used by the 48 rehabilitation units in NSW to classify rehabilitation patients since July 1998. It classifies stroke episodes on the basis of patient age and Functional Independence Measure (FIM) scores12 on admission (Box 1). The FIM is similar to the commonly used Barthel Index, but includes a cognition subscale (five items — comprehension, expression, social interaction, problem solving and memory), as well as a motor subscale (13 items, including self-care [bathing, grooming, toileting, eating and dressing], sphincter control, transfers, and mobility). Each item is scored on a seven-point ordinal scale (1 = total assistance, 7 = complete independence), giving a total FIM score of 18–126. The FIM has high reported reliability and validity.13-14 We identified patients with acute stroke who went on to have rehabilitation in the stroke unit in the AN-SNAP database by their unique medical record numbers, and extracted information on AN-SNAP class, FIM scores, rehabilitation LOS and discharge destination. Patients in the stroke unit were reclassified as rehabilitation patients when they were medically stable with regard to neurological deficits and complications, and deemed fit for intensive rehabilitation.

To compare acute stroke outcomes, we obtained AR-DRG data for NSW for the period July 1999 to June 2000 from the NSW Health Department. To compare rehabilitation outcomes, we extracted data on stroke rehabilitation episodes, excluding those at Bankstown–Lidcombe Hospital, for the period July 1999 to November 2000 from the NSW AN-SNAP database.

Statistical analysis

Data were analysed using SPSS version 9.0.15 Patient characteristics were compared using χ2 or t tests, while outcomes were compared using χ2 tests (categorical outcomes) and the non-parametric Mann–Whitney U test (continuous outcomes). Linear regression was used to adjust and test for statistical interaction. P values for linear regression model comparisons were obtained using the likelihood ratio test.

Acute stroke episodes

Between July 1999 and November 2000, 233 patients were admitted to our stroke unit with a total of 242 acute stroke episodes; 148 were admitted directly from the emergency department (61% of episodes), 81 were transferred from general medical wards (34%) and 11 from other hospitals (5%), and two were admitted directly from home.

During the year July 1999 to June 2000, there were 9777 acute stroke episodes among patients admitted to NSW hospitals.

Patient characteristics: Patients receiving acute care in our stroke unit did not differ significantly from NSW patients with stroke in age (mean age [SD], 74.0 [12.2] v 72.7 [14.9] years for NSW) or sex distribution (50% male v 49% for NSW). However, a smaller proportion of patients in our unit were married (34% v 48% for NSW; χ= 16.85; 1 df; P < 0.001). There were also no significant differences in frequencies of hypertension, diabetes mellitus, atrial fibrillation/flutter, hyperlipidaemia and dementia, but a smaller proportion of patients at our unit had ischaemic heart disease (17% v 24% for NSW; χ= 6.73; 1 df; P = 0.01).

Outcomes: Patients in our stroke unit had shorter LOS for acute stroke than NSW patients, reaching statistical significance for DRG class B70A (Box 2). There was also a trend for a lower mortality rate for acute stroke episodes in our stroke unit (12% v 15.7% for NSW; χ= 2.5; 1 df; P = 0.11) and a slightly lower 28-day readmission rate (5% v 5.8% for NSW; χ= 0.31; 1 df; P = 0.58). However, further analysis by DRG class showed no significant differences between stroke-unit and NSW patients in mortality and readmission rates.

Rehabilitation episodes

Patient characteristics: Patients undergoing rehabilitation in our stroke unit did not differ significantly from those undergoing rehabilitation elsewhere in NSW in either age (mean age [SD], 73.5 [11.1] v 72.1 [12.5] years) or sex distribution (% male: 47% v 50%). However, our patients had lower mean FIM score on admission than NSW patients in most AN-SNAP classes (Box 3), although the differences were significant only for classes 203 and 208.

Outcomes: Patients in our stroke unit had shorter mean rehabilitation LOS and greater change in FIM scores than patients elsewhere in NSW, even after adjustment for patient age and FIM score at admission to rehabilitation (Box 4). Our patients also had higher mean FIM scores at discharge after adjustment for age. There was no significant statistical interaction with AN-SNAP class, implying that the differences seen were consistent across all AN-SNAP classes. There were no significant differences in the proportion of patients discharged to nursing homes between patients undergoing rehabilitation in our unit and elsewhere in NSW (15.5% v 11.2%; χ= 1.9; 1 df; P = 0.17).

Non-stroke-unit patients

During the study period, 171 patients with 177 acute stroke episodes were admitted to wards of Bankstown–Lidcombe Hospital other than the stroke unit. Mean age was 70.9 years (SD, 14.1), and 48% were male. DRG classification is shown in Box 2. About half (85) of these episodes were classed as B70D ("died or transferred out of the hospital in under five days"). Non-stroke-unit patients had shorter mean LOS (mean LOS for DRGs B70A, B70B and B70C were 10.0, 7.1 and 5.6 days, respectively), but a higher acute stroke mortality rate (16%) and a higher 28-day readmission rate (19%) than the stroke unit.


Our study showed that combining acute and rehabilitation services in a stroke unit may improve patients' functional outcomes and reduce length of hospital stay. Possible reasons for the shorter LOS in acute care in our stroke unit compared with NSW as a whole were that patients did not need to wait for a rehabilitation bed or that their classification was shifted early from acute to rehabilitation. However, rehabilitation LOS was also shorter in our unit, demonstrating an overall benefit.

LOS in our stroke unit may have been reduced because of the immediate availablity of intensive physiotherapy during acute care. As our patients were not transferred for rehabilitation, they did not need to adjust to a new environment, which may hinder recovery, especially among elderly patients, or undergo assessment by a new team of doctors and therapists. Although availability of outpatient and domiciliary rehabilitation services may allow earlier patient discharge, this was probably not a factor in our patients' shorter LOS, as Bankstown does not have a strong domiciliary rehabilitation team, and most rehabilitation in the region is done on an inpatient basis.

Despite our patients' shorter LOS in rehabilitation, their functional outcomes at discharge were better than those in patients elsewhere in NSW, probably because of the early initiation of rehabilitation in our unit.16-18 The greater increase in FIM score in our patients was expected, as their scores at admission were lower, attributable to their earlier reclassification as rehabilitation rather than acute patients. However, our patients also had higher mean FIM scores at discharge. Coupled with their shorter length of stay, this increases the robustness of our results, as functional gain is expected to increase with longer hospital stay.19

Confounding factors that may affect rehabilitation outcome are stroke severity (which is reflected in the level of function on admission),17,20-22 patient age,18,20-22 presence of medical comorbidities,20,23 and intensity of rehabilitation.17 We adjusted for age in our analysis, and also attempted to control for stroke severity by comparing outcomes within AN-SNAP classes and adjusting for FIM score on admission. Although AN-SNAP class and FIM score on admission may not be adequate measures of stroke severity, no better measures (such as score on the Scandinavian Stroke Scale24) are included in the AN-SNAP database. This is a limitation of our study. The lack of data on comorbidities and intensity of rehabilitation is a further limitation.

Another potential limitation is the concern that DRG and AN-SNAP data may not be totally robust. To ensure quality control of FIM scoring, weekly one-hour training sessions have been held for staff in the stroke unit since its implementation. These are conducted by a senior occupational therapist who is an accredited FIM trainer. Similar training programs have been held in rehabilitation units throughout NSW.

We found no statistical difference in rates of nursing home admission between rehabilitation patients in our stroke unit and elsewhere in NSW. However, being unmarried and not having a caregiver are both associated with institutionalisation,23,25 and acute patients in our stroke unit were less likely to be married than NSW patients (the AN-SNAP database does not record marital status).

Although the policy at Bankstown–Lidcombe Hospital is to admit patients with stroke to the stroke unit whenever possible, in 177 episodes of acute stroke patients were admitted to other wards because insufficient beds were available in the stroke unit. About half these patients were classified DRG class B70D ("died or transferred out of the hospital in under five days"), compared with 5% of stroke-unit patients. Non-stroke-unit patients also had higher acute mortality. However, this would not affect LOS comparisons between stroke-unit and NSW patients, as data were stratified by DRG class, with class B70D excluded from the analyses. Furthermore, in other DRG classes (B70A–C), LOS was uniformly shorter in patients treated outside the stroke unit. Therefore, the exclusion of these patients did not account for the lower LOS in the stroke unit compared with NSW as a whole. There were no rehabilitation data for non-stroke-unit patients, as AN-SNAP data are collected only for those in the stroke unit. However, it was not our aim to compare outcomes between the stroke unit and general medical wards.

A controlled trial with randomisation at onset of acute stroke would provide the best comparison of different types of stroke services, but has practical difficulties. Despite the limitations of our study, it provides preliminary evidence of the benefits of a comprehensive, dedicated stroke unit in reducing length of hospital stay and improving functional outcomes in patients with stroke.

1: Definitions of DRG and AN-SNAP classes of stroke

DRG classes*

B70A: With severe or complicating diagnosis/procedure (eg, septicaemia, pneumonia or pulmonary embolism)

B70B: With other complications/comorbidities

B70C: Without other complications/comorbidities

B70D: Died or transferred < 5 days

AN-SNAP classes

203: Motor score, 13

204: Motor, 63–91; cognition, 20–35

205: Motor, 63–91; cognition, 5–19

206: Motor, 47–62

207: Motor, 14–46; age ≥ 75 years

208: Motor, 14–46; age ≤ 74 years

* From version 4.1 of the Australian Refined Diagnosis Related Group (AR-DRG) classification system.9
† From the Australian National Subacute Non-Acute Patient Casemix Classification System (AN-SNAP),11 based on Functional Independence Measure (FIM) scores.12

2: Characteristics and acute length of stay of patients with acute stroke in Bankstown–Lidcombe stroke unit and throughout New South Wales

Number of patients (%)

Mean length of stay in acute care (days)

DRG class

Bankstown– Lidcombe stroke unit
(n = 242)

Bankstown–Lidcombe non-stroke unit
(n = 177)

(n = 9777)

Bankstown–Lidcombe stroke unit
(n = 242)

(n = 9777)

(95% CI)



117 (48%)

37 (21%)

3261 (33%)



− 1.8 (−3.5 to −0.2)



  93 (38%)

39 (22%)

3121 (32%)



− 0.7 (−1.9 to 0.4)  



19 (8%)

16 (9%)  

1391 (14%)



−1.1 (−3.2 to 1.0) 



13 (5%)

85 (48%)

2004 (21%)





NA = not assessed. * Difference in mean LOS between the stroke unit and throughout NSW. † For comparison of mean LOS between the stroke unit and throughout NSW.

3: Characteristics of patients undergoing rehabilitation at the Bankstown–Lidcombe stroke unit and elsewhere in NSW

AN-SNAP class

Number of patients (%)

Mean FIM score on admission (SD)

Stroke unit (n = 113)

NSW (n = 2350)

Stroke unit (n = 113)

NSW (n = 2350)



29 (26%)

83 (4%)

20.9 (7.9)  

23.2 (6.7)



13 (12%)

655 (28%)

106.4 (8.3)    

105.7 (9.6)  



4 (4%)

82 (4%)

85.0 (6.1)  

86.8 (9.1)



16 (14%)

535 (23%)

81.3 (10.0)

80.4 (9.4)



28 (25%)

503 (21%)

48.3 (15.7)

50.8 (14.9)



23 (20%)

492 (21%)

44.0 (14.5)

51.5 (15.2)


FIM = Functional Independence Measure. * For comparison of mean FIM score on admission between patients undergoing rehabilitation at the stroke unit and elsewhere in NSW.

4: Rehabilitation length of stay (LOS) and functional outcome in the stroke unit at Bankstown–Lidcombe Hospital compared with elsewhere in New South Wales

AN-SNAP class

Mean LOS in rehabilitation

Mean FIM score at discharge

Mean change in FIM score

Stroke unit


(95% CI)


Stroke unit


(95% CI)


Stroke unit


(95% CI)





(−23.5 to 2.7)




(3.1 to 24.2)




(8.2 to 26.9)

< 0.001




(−8.6 to 3.0)




(−1.3 to 10.0)




(− 0.9 to 9.0)





(−25.8 to 3.3)




(−6.8 to 21.0)




(− 3.3 to 20.6)





(−16.9 to −0.34)




(0.9 to 19.1)




(1.2 to 17.1)





(−25.0 to −7.2)

< 0.001



(2.1 to 23.5)




(7.9 to 24.0)

< 0.001




(−24.6 to −1.5)




(1.9 to 23.8)




(10.1 to 28.9)

< 0.001




(−17.6 to −9.7)

< 0.001



− 6.7 
(−12.3 to −1.0)




(6.8 to 13.8)

< 0.001

FIM = Functional Independence Measure. *Difference in means between the stroke unit and elsewhere in NSW (adjusted for age and admission FIM score for rehabilitation LOS and change in total FIM score, and adjusted for age for discharge total FIM score). † For comparison of means between the stroke unit and elsewhere in NSW.

  • Yan H Ang1
  • Daniel K Y Chan2
  • Qing Shen3
  • Derrick M K Heng4

  • 1 Aged Care and Rehabilitation, Bankstown–Lidcombe Hospital, Sydney, NSW.
  • 2 Clinical Trials and Epidemiological Research Unit, National Medical Research Council, Singapore.



We thank Natalia Alechna (Information Manager, Bankstown–Lidcombe Hospital) for extracting Bankstown data. We also thank David Perabo and Greg Perry (New South Wales Health Department) for providing NSW data.

Competing interests:

None identified.

  • 1. Jorgensen HS, Nakayama H, Raaschou HO, et al. The effect of a stroke unit: reductions in mortality, discharge rate to nursing home, length of hospital stay, and cost. Stroke 1995; 26: 1178-1182.
  • 2. Indredavik B, Bakke F, Solberg R, et al. Benefit of a stroke unit: a randomised controlled trial. Stroke 1991; 22: 1026-1031.
  • 3. Langhorne P, Williams BO, Gilchrist W, Howie K. Do stroke units save lives? Lancet 1993; 342: 395-398.
  • 4. Ronning OM, Guldvog B. Stroke units versus general medical wards, I: Twelve- and eighteen-month survival: a randomised, controlled trial. Stroke 1998; 29: 58-62.
  • 5. Strand T, Asplund K, Eriksson S, et al. A non-intensive stroke unit reduces functional disability and the need for long-term hospitalisation. Stroke 1985; 16: 29-34.
  • 6. Stevens RS, Ambler NR, Warren MD. A randomised controlled trial of a stroke rehabilitation ward. Age Ageing 1984; 13: 65-75.
  • 7. Garraway WM, Akhtar AJ, Prescott RJ, Hockey L. Management of acute stroke in the elderly: preliminary results of a controlled trial. BMJ 1980; 280: 1040-1043.
  • 8. Stroke Unit Trialists' Collaboration. Organised inpatient (stroke unit) care for stroke (Cochrane Review). In: The Cochrane Library 2002; 2. Oxford: Update Software, 2002.
  • 9. Department of Human Services and Health. Casemix in Australia: an overview. Canberra: Department of Human Services and Health, 1994.
  • 10. International statistical classification of diseases and related health problems, 10th revision, Australian modification (ICD-10-AM). Sydney: National Centre for Classification, 2000.
  • 11. Lee LA, Eagar KM, Smith MC. Subacute and non-acute casemix in Australia. Med J Aust 1998; 169 Suppl: S22-S25.
  • 12. Granger CV, Hamilton BB, Keith RA, et al. Advances in functional assessment for medical rehabilitation. Top Geriatr Rehabil 1986; 1: 59-74.
  • 13. Dodds TA, Martin DP, Stolov WC, Deyo RA. A validation of the functional independence measurement and its performance among rehabilitation inpatients. Arch Phys Med Rehabil 1993; 74: 531-536.
  • 14. Ottenbacher KJ, Hsu Y, Granger CV, Fiedler RC. The reliability of the functional independence measure: a quantitative review. Arch Phys Med Rehabil 1996; 77: 1226-1232.
  • 15. Statistical Package for the Social Sciences. SPSS for Windows. Release 9.0.1. Chicago: SPSS Inc, 1999.
  • 16. Indredavik B, Bakke F, Slordahl SA, et al. Treatment in a combined acute and rehabilitation stroke unit: which aspects are most important? Stroke 1999; 30: 917-923.
  • 17. Cifu DX, Stewart DG. Factors affecting functional outcome after stroke: a critical review of rehabilitation interventions. Arch Phys Med Rehabil 1999; 80 Suppl 1: S35-S39.
  • 18. Ottenbacher KJ, Jannell S. The results of clinical trials in stroke rehabilitation research. Arch Neurol 1993; 50: 37-44.
  • 19. Carey RG, Seibert JH, Posavac EJ. Who makes the most progress in inpatient rehabilitation? An analysis of functional gain. Arch Phys Med Rehabil 1988; 69: 337-343.
  • 20. Shah S, Vanclay F, Cooper B. Predicting discharge status at commencement of stroke rehabilitation. Stroke 1989; 20: 766-769.
  • 21. Jongbloed L. Prediction of function after stroke: a critical review. Stroke 1986; 17: 765-776.
  • 22. Inouye M. Predicting models of outcome stratified by age after first stroke rehabilitation in Japan. Am J Phys Med Rehabil 2001; 80: 586-591.
  • 23. Ween JE, Alexander MP, DiEsposito M, Roberts M. Factors predictive of stroke outcome in a rehabilitation setting. Neurology 1996; 47: 388-392.
  • 24. Scandinavian Stroke Study Group. Multicenter trial of hemodilution in ischemic stroke. Background and study protocol. Stroke 1985; 16: 885-890.
  • 25. Rockwood K, Stolee P, McDowell I. Factors associated with institutionalization of older people in Canada: testing a multifactorial definition of frailty. J Am Geriatr Soc 1996; 44: 578-582.


remove_circle_outline Delete Author
add_circle_outline Add Author

Do you have any competing interests to declare? *

I/we agree to assign copyright to the Medical Journal of Australia and agree to the Conditions of publication *
I/we agree to the Terms of use of the Medical Journal of Australia *
Email me when people comment on this article

Online responses are no longer available. Please refer to our instructions for authors page for more information.