| Kalra et al. (1997) (UK) [15] GRADE C |
Aim – RCT to determine whether poor outcome in patients with visual neglect (VN) was due to greater stroke severity or non-specialist management Setting – Acute, stroke unit Sample (47 HI+, 99 HI−) Mean age 77 (SD = 8) Time to 1st obs. 1–2 weeks post-stroke onset Follow-up at discharge Before and after controlled intervention (conventional versus spatio-motor cueing and early emphasis on restoration of function) |
VN assessed by Line bisection supplemented by functional observation at admission 1a Outcome BI (scale 0–20) and Thumb finding test 2nd Outcome Mortality Discharge-destination LOS Therapy intensity |
Median statistic Chi squared test, Mann–Whitney U, t-test Multiple linear regression (n = 146), DV = BI at admission Modelled IVs Age, gender, muscle power, balance, proprioception, cognition, pre-stroke ADL status, HI level |
Patients with or without visual neglect (VN) had similar destination, slightly lower median BI scores at admission and discharge (4 versus 5 and 16 versus 14) resp. Greater LOS/days (64 HI+ versus 36 HI−) and therapy input/h PT (30 HI+ versus 19 HI−) and OT (18 HI+ versus 10 HI−) HI negatively associated with admission BI [β = −0.17, p = 0.011, R2 = 0.16] All other IV’s not associated with DV |
Confirmed stroke Clear selection criteria Validated ADL assessment Statistically modelled variety of factors associated with ADL besides HI Reported attrition due to death (n = 3 extended stroke, 1 pulmonary embolus, 1 myocardial infarction) Intention to treat analysis Corrected for small sample size |
Wrongly labelled as RCT Recruited only patients with Partial Anterior Circulation Infarct of moderate stroke severity with potential for rehabilitation Line bisection does not distinguish between visual neglect and other sub-types BI version excluded psycho-social dysfunction & cognitive measure No community follow-up Different patient LOS so exposure to therapy uncontrolled Did not model outcome data at discharge No sensitivity analysis |
| Ring et al. (1997) [28] (Israel) GRADE C |
Aim – To measure function and determine gain between admission and discharge Design – Prospective comparative Setting – Acute General Rehabilitation facility Sample (28 HI+, 56 HI−) Mean age 60.8 Time to 1st observation was 29 days (±17) Follow-up at discharge |
BIT at admission to detect “neglect” 1a Outcome FIM 2nd Outcomes LOTCA Type and site of lesion LOS Discharge destination |
t-test Chi square test Repeated measures ANOVA Multiple linear regression with FIM gain (DV) Modelled IV’s LOS, admission FIM, age, gender, risk factors (not clear which) |
FIM admission score, LOS and age predicted functional gain [β = −0.034, 0.13, 0.49, p = 0.011, 0.03, 0.05] resp. 24/28 patients with HI discharged home after considerably longer period of rehab and LOS/days (137 HI+ versus 102 HI− days) Total FIM gain HI+ 33 versus HI− 21 units |
Confirmed stroke by CT scan Validated functional ability scale and test battery for detection of HI Statistically adjusted for age and gender Clear distinction between RHS and LHS, lesion site and type Reported attrition due to death (n = 1) |
Selection criteria not clear what behavioural conditions were excluded Variable obs. time-point No community follow-up Not adjusted for differences in stroke severity or time post-stroke No sensitivity analysis No data on cognitive function from LOTCA published |
| Paolucci et al. (2001) [16] (Italy) GRADE C |
Aim – Assess influence of unilateral spatial neglect (USN) on rehabilitation outcome Matched by Age (69 ± 10) and stroke onset admission time (38 ± 17 days) Setting – Acute, in-patient rehabilitation hospital Sample – (89 HI+, 89 HI−) Time to 1st observation (38 ± 17 days) Follow-up at discharge Intervention; special training in visual scanning, reading and copying script, line drawings, dot matrix and description of scene 5h/week for 8 weeks |
USN detection – Letter cancellation, line bisection, sentence reading and Wundt–Jastrow area illusion test at admission 1a Outcome BI (0–100) 2nd outcome LOS Rate of gain and amount of progress Other RMI CNS Hamilton Depression Rating scale |
Eight multiple linear regression (forward stepwise) Six logistic regressions Five DV’s, CNS, BI, RMI, LOS, rate of gain and amount of progress Modelled IV’s Admission CNS, gender, type of lesion, hypertension, diabetes, heart disease, unilateral spatial neglect, depression, epileptic seizures post-stroke, family support, education level, discharge destination |
USN was a negative prognostic factor. USN patient group had low ADL and mobility outcomes at discharge (∼50% less mean scores) HI+ had longer LOS/days (117 ± 61 versus 81 ± 38), ↑rate of discharge to institution (18% versus 5%), ↑ discharge continence rates (21% versus 5%) USN, stroke severity, heart disease and type of lesion appear to be important explanatory variables in the acute phase (∼3 months) |
Confirmed stroke (CT scan) Validated tools BI supplemented by data from RMI Screened for depression and neurological severity Reported attrition, (9% HI−, 6.7% HI+) Modelled broader range of factors e.g. psych-social factors and comorbidity Adjusted for stroke severity in some models |
Probable patient overlap with earlier sample (Paolucci et al. [27]) Probably excluded severe stroke included (mean CNS = 7) Highly variable T0 observations Complicated paper to follow due to large number of factors and combinations modelled Did not measure cognition which is strongly associated with USN (neglect) Not adjusted for or modelled age which is associated with USN High variability in LOS and exposure to in-patient care likely source of bias No information on handling of missing data |
| Buxbaum et al. (2004) [11] (Italy and USA) GRADE D |
Aim – Assess occurrence of subtypes and related deficits in RHS Design – cross-section, retro and prospective data Setting – Acute and community Sample – 623 RHS recruited from four rehab hospitals in Philadelphia and two in Italy. 268 met selection criteria 166 consented; 86 had acute and 80 chronic lesions, (88 HI+, 78 HI−) Mean age – Acute 66, range (37–89) chronic 67, range (33–88) Time to 1st and only observation – Acute (5–41) and chronic (94–1272) days |
Personal and Peri-personal Bells test and 4 Behavioural Inattention (BIT) sub-tests (letter cancellation, picture scan, menu reading and line bisection) Motor and perceptual neglect measured by response latencies in two stimulus and response tasks Motor and Sensory exam visual fields and extinction by means of confrontation method Sustained and divided auditory attention Test (SART) Anosognosia 5 questions adapted from Cutting’s questionnaire 1a Outcome FIM Family Burden Scale |
Chi square test Mann–Whitney U test Correlation tests Repeated measures ANOVA Regression analyses |
Neglect severity significantly explained FIM scores and carer burden but not lesion size Similar rate of gain in HI± but lower FIM scores in HI+ (estimates not reported in paper) Acute patient lesions were not restricted to cortical areas Variation in associated deficits but higher frequencies in HI+ Variation in occurrence of HI sub-types |
Attempted to document frequency of various HI subtypes and related deficits Included burden of care assessment Acknowledged significant limitations in sensitivity and specificity of tests used to identify neglect sub-types and anosognosia Also acknowledged lack of statistical adjustment for multiple tests |
Significant heterogeneity in sample and variation in time to 1st observation complicate interpretation of results Recruited patients deemed to benefit from rehabilitation, i.e. Excluded severe attention and cognitive deficits, previous stroke or neurological disorder and dementia Combined analysis of patients from different culture and health care systems – can be strength but also weakness Inter-rater reliability not performed FIM mean scores not directly reported |
| Gillen et al. (2005) [12] (USA) GRADE D |
Aim – Examine the relationship between left unilateral spatial neglect (USN) and rehabilitation outcomes in RHS patients Design - Retrospective Setting – Acute in-patient rehabilitation hospital Sample – (50 HI+ 125 HI−) Mean age 72 (SD = 11.0) Time to 1st observation was 15 ± 10 days Follow-up observation at discharge |
“USN” assessed by Letter cancellation test (LCT) at admission 1a Outcome FIM Other Cognistat at admission Geriatric Depression Scale (GDS) at admission LOS |
Univariate correlation Multivariate regression analyses (n = 98) FIM discharge scores (DV) regressed on FIM admission and USN |
Longer mean LOS in HI+ 31 versus 25 in HI− HI+ progressed at slower rate. Mean admission FIM score 50 (SD = 16) versus 69 in HI− (SD = 16) Greater cognitive impairment in HI+ (p < 0.001), higher GDS scores and depression levels (p < 0.01) “USN” predicted social-cognitive domain (β = −0.29, p < 0.001) |
Included depression and cognitive function Used validated measures Modelled rate of progress (change in FIM score/LOS) |
106/281 eligible patients excluded due to poor visual acuity. Perceptual deficits and difficulty completing LCT at 1st observation Depression assessed probably too early when patients are likely to be depressed due to stroke event No FIM or cognitive discharge score reported. |
| Odell et al. (2005) [29] (USA) GRADE D |
Aim – To document selected functional outcomes at the termination of in-patient treatment Design – Retrospective Setting – Acute in-patient rehabilitation hospital Sample – (60 HI+ 41 HI−) Mean age 70 years Range (40–99) Time to 1st observation not known Follow-up observation at discharge |
No formal assessment of HI (relied on mention of condition in medical records) 1a Outcome FIM scores at admission and discharge 2nd Outcome Amount and efficiency of gain, LOS Discharge placement |
Mann–Whitney U test Regression analysis Modelled IV’s 12 predictor variables made up of initial motor score, cognitive items plus age, gender, previous neurological episodes, no. of comorbidities, lesion site and presence/absence of HI |
Admission, discharge FIM median HI+ (57 and 88), HI− (66 and 104); similar gains in motor ∼24 units, cognitive domains HI+ (3.5), HI− (2). 1 unit gain in FIM cognitive scores by in HI± groups When modelled, functional outcome was predicted by age, memory, problem solving and motor function Mean LOS, HI± 29 versus 22 (3–75) days; >75% home discharge Therapy sessions HI± 61 versus 27 (range 1–194) |
Transformed data by means of Rasch method to increase accuracy of estimates Adjusted for variation in age Recorded number of comorbidities and therapy sessions Categorised descriptive statistics by age range (40–92); younger age group were less impaired and made highest gains overall |
Highly selective criteria, i.e. included only patients referred to speech therapy (reduces generalisation of findings) Stroke severity not known No formal assessment of HI Variable follow-up observation point Limitations of retrospective studies, e.g. reliability and accuracy of data cannot be checked, consistency of assessment methods and data collection cannot be guaranteed Missing data not reported |
| Di Monaco et al. (2011) (Italy) [30] GRADE C |
Aim – To investigate the relationship between severity of unilateral spatial neglect (USN) and functional recovery in ADL after a RHS Design – Prospective Setting – Acute in-patient, physical medicine and rehabilitation hospital Sample – (54 HI+, 53 HI−) Mean age 70 (range 63–80) Time to 1st observation was 23 days post-stroke onset Follow-up observation 80 days post-stroke onset |
Detection of USN – BIT at admission only and Diller’s test (cancellation task) 1a Outcome Admission and discharge FIM scores Other BI prior stroke by anamnesis Mini-Mental (MMSE) LOS |
Data analysis on 107/131 Bivariate correlation FIM × BIT scores Mann–Whitney U test for group differences Chi square test Three multiple regressions Three DV’s = discharge FIM, FIM efficiency and effectiveness Modelled IV’s Age, MMSE score, time to 1st observation, gender, education, BI , FIM admission and discharge |
Admission, discharge FIM median HI+ (45 and 91), HI− (55 and 110) but >30 units of variation within each group at all times MMSE median group score (HI+ 24, HI− 27). FIM admission best predicted FIM discharge score Model explained 49% of variance in DV; of these “USN” explained 5%; FIM 44% High variability in and LOS (37–72 days) |
Reported missing data (n = 5) Statistically adjusted for age, gender, education level, time to 1st observation and FIM admission Transformed FIM scores to ∼ normal distribution Recognised limitations of the study, i.e. assessing limited no. factors associated with HI and function and limitations of BIT in distinguishing between sensory motor HI, visual-spatial and motor Modelled education level |
Excluded 19 with severe stroke No intention to treat analysis – possible bias towards milder stroke severity (MMSE scores at admission indicate mild cognitive impairment) FIM cognitive score not provided to compare with MMSE No adjustment for stroke severity or carer status Different patient exposure to in-patient care likely source of bias |
| Timbeck et al. (2013) [32] (Canada) GRADE D |
Aim – Evaluate effect of visuo-spatial neglect (VSN) on functional outcome and discharge destination in RHS Design – Prospective Setting – Stroke rehabilitation programme Sample - (6 HI+, 10 HI−) Mean age 76 (SD = 10) Time to 1st observation was 7 days from admission to rehabilitation Follow-up observation prior to discharge |
VSN detected by BIT 1a Outcome FIM Other MMSE Berg balance scale (BBS) CMSA LOS |
MANOVA to compare between VSN± patients DV – age, time to 1st observation, LOS. MMSE, admission–discharge FIM, BBS and CMSA Independent t-tests for univariate analyses and Fisher’s exact for categorical variables |
VSN+ (n = 6) tended towards supported living FIM admission–discharge score; HI+ 60 and 73, HI− 86 and 102 units High SD in both groups at all FIM observations ∼20 admission, 28 discharge LOS average VSN+ 48, VSN− 38 days Differences in BBS within groups (SD = 16), between groups; HI+ scored 12 and 22 versus 28 and 41 BBS units in HI− at admission and discharge resp |
Included balance measure Supplemented motor activity on the FIM scale with another impairment measure Evaluated multivariate effect by Pillai’s trace (ensure robustness against non-normal distributions and heterogeneity of variance particularly with small samples and groups) Acknowledged significant study limitations |
Very small sample unlikely to be fully representative of RHS has implications for study power and validity of results Tight selection criteria excluded patients with chronic co-morbidity (not clear what), English as 2nd language and cognitive impairment – has implication for generalisation of results Not accounted for changes due to spontaneous recovery effects occurring in average 28 days (SD 19.23) delay in starting rehabilitation programme. This has implications for findings and conclusions based on results No adjustment for multiple testing especially on a small sample |
| Paolucci et al. (1996) [27] (Italy) GRADE D |
Aim – to test whether specific neglect training improved hemi-spatial neglect (HSN) and functional outcome Design – Prospective, cross over design for HSN+ (divided into two groups) + HSN- (3rd group) Setting – Community rehabilitation facility Sample n = 59 RHS (23 HI+, 36 HI−) Mean age 65 (SD = 13) Time to 1st observation was 2–6 months post-stroke onset Follow-up at 2 and 4 months whilst in rehabilitation Intervention – 40 h of visual scanning, auditory cueing, reading, copying, line drawing, picture description |
“HSN” assessed once at admission to rehabilitation facility by Letter cancellation, line bisection, sentence reading and Wundt–Jastrow area illusion test at admission 1a Outcome BI (0 to 100) Other RMI CNS Lesion size |
Three ANOVA’s to estimate differences between three groups in BI, RMI and CNS scores at follow-up (2 and 4 months) Four ANOVA’s for differences in USN tests One ANOVA difference in lesion size by group (n = 3) |
Specific USN training improved functional ability of USN+ group but gains not maintained by end of study Similar magnitude of difference between USN± patients in mean functional ability and mean RMI (1st, 2nd and 3rd observation = 20%, 30% and 30%, respectively) No group difference in lesion size |
Screened for stroke severity but data not reported Validated measures Test-battery used to assess HI but not standardised Used RMI to supplement information on functional ability not provided by BI scale, e.g. walking outside house Community follow-up |
No radiologic confirmation of stroke Excluded patients over 78, multiple lesions, haemorrhage or chronic CNS pathologies Crossover intervention for USN+ group sizes were small (n = 11 and 12) Stroke severity not known No fixed assessment time-points Not adjusted for multiple testing No statistical adjustment for confounding factors Attrition not reported |
| Katz et al. (1999) [10] (Israel) GRADE D |
Aim – To evaluate impact of unilateral spatial neglect (USN) on functional outcome in long term Design – Prospective, longitudinal Setting – Acute, General Rehabilitation Hospital Sample – (19 HI+, 21 HI−) Mean age 57 (SD = 10) Time to 1st observation was ∼30 days Follow-up at discharge, 6/12 after discharge, up to 1 year post-stroke onset No intervention but USN+ patients received special attention and care for USN |
USN detected by BIT at admission and discharge 1a Outcome FIM Other LOTCA at admission and discharge RKE at discharge LOS |
t-test Chi squared test Repeated measures ANOVA Multiple linear- regression – FIM (DV) Modelled IV’s (stepwise entry method) BIT score, sitting balance, thinking operations (not defined) and tactile sensation |
USN was major predictor of functional outcome from admission to follow-up Despite special attention given to USN+ group, they had higher disability levels, slower improvement rate Most progress occurred within the in-patient facility Longer LOS/days for USN+ (119 ± 49) versus (78 ± 52) for USN− 39/40 patients were discharged home, one patient with USN discharged to nursing home USN+ needed high levels of support at home compared to USN− USN could be predicted from pen and paper tests alone (no advantage in giving functional sub-section) |
Confirmed stroke by CT scan Long term follow-up 2/4 fixed observation points Modelled also cognitive, IADL score, tactile factors, sitting balance Reported therapy time 45–60 min of OT and PT/patient Tracked recovery of function up to a year post-onset |
Small sample size, possibly underpowered for regression analysis (increased risk of type 1 error) Excluded severe stroke and psychiatric disorders not clear which, restricted inclusion to 1st stroke only with no comorbidities Inconsistent assessment protocol (BIT and LOTCA not repeated at follow-up) to assess recovery No attrition reported Observations from same patients not independent – invalidates regression assumption No statistical adjustment of confounding factors FIM is a multi-disciplinary tool (not clear how this was completed in the community? |
| Cherney et al. (2001) [14] (USA) GRADE D |
Aim – To evaluate relationships between unilateral spatial neglect (USN) and cognitive-communicative functional outcomes in RHS Design – Prospective, repeated measures Setting – Acute rehabilitation facility Sample (36 HI+, 16 HI−) Mean age – 66 (SD = 14.0) Time to 1st observation at facility was 33 ± 68 days after stroke Follow-up at discharge and 3 months post-discharge |
USN detected by (BIT) at admission 1a Outcome FIM Other RIC-FAS LOS |
ANOVA Mann–Whitney U test Pearson’s coefficient of correlation |
Statistically significant differences were found in overall FIM and motor sub-score but not cognitive score. USN+ patients scored 10 FIM units (8%) less at each observation point High correlation between pen and paper tests and behavioural section on BIT (r = 0.89) Moderate correlation (r = 0.51) between FIM and BIT scores at 1st and 2nd observation points which weakened by 3rd (r = 0.36) LOS/days for USN+ versus USN− (38 ± 9 versus 31 ± 10). No impact of USN severity on LOS reported |
Evaluated cognitive function and communication (not previously included) Reported attrition (n = 4) due to incomplete documentation at discharge and (n = 12) lost to 3 month follow-up |
Small sample size for sub-group analysis by USN severity Highly variable time to 1st observation (source of bias) Stroke severity not known No fixed observation point – limits comparison of results No intention to treat analysis FIM scores at 3 month follow-up obtained by telephone interview – reliability of data? |
| Stein et al. (2009) [31] (UK) GRADE D |
Aim – To compare and evaluate basic functional mobility in patients with and without visual neglect (VN) Design – Prospective, repeated measures Setting – Acute inpatient (stroke unit) and community rehabilitation Sample – (14 HI+, 14 HI−) Mean age 76 (SD = 11) Time to 1st observation was 7–28 days post-stroke onset Follow-up observation at discharge and 4 weeks post-discharge |
VN detected by BIT 1a Outcome BI (0–20) Other Elderly mobility scale (EMI) Middlesex elderly assessment of mental status (MEAMS) Postural assessment scale for stroke (PASS), LOS Discharge destination Continence status Carer status |
Mann–Whitney U test Kruskal–Wallis Wilcoxon matched pairs Bonferroni correction for multiple testing |
Mean LOS/days was 79 and 52 for HI±, respectively Seven VN+ discharged home versus 12 VN−. VN+ increased risk for institution discharge. Mean difference of 7 BI units (35%) at discharge (p = 0.013) Patients with mild VN and independent mobility tended to be discharged home Relationship between carer presence and discharge destination was not clear |
Included community follow-up Included range of severity of VN levels Included separate measure of posture relevant to functional mobility Included data on discharge destination and continence status Reported number of deaths (n = 3) and outliers (n = 4) Corrected for multiple testing to minimise type 1 error |
BIT, MEAMS, BI were not assessed post-discharge, therefore unable to track change especially in functional mobility Possibility that differences observed between patients could be due to type 1 and II errors largely due to small sample size No correlation statistics to study association of factors with functional mobility No fixed observation points limits comparison to other studies |
