Barthel Index

Brain injury: (Liu, 2004; n=175 with n=101 with vascular brain injury, n=49 with traumatic brain injury, n=27 with other brain injuries; median age=45, inpatient rehabilitation; length of stay=95 days).

Predictive Validity:

Brain injury: (Liu, 2004; n=175 with n=101 with vascular brain injury, n=49 with traumatic brain injury, n=27 with other brain injuries; median age=45, inpatient rehabilitation; length of stay=95 days).

• Adequate predictive validity: The admission score on the Barthel was predictive of the discharge score. The lower the admission score, the greater the change with rehabilitation. (Rho=-0.42)

Brain injury: (Liu, 2004; n=175 with n=101 with vascular brain injury, n=49 with traumatic brain injury, n=27 with other brain injuries; median age=45, inpatient rehabilitation; length of stay=95 days).

Inpatient rehab: (Castiglia et al., 2017; n=53; mean age 72.71 (12.79); orthopedic diseases n=21; PD or stroke n=32; mean hospital duration: 45.11(16.53) days, Italian version)

• MCID = 35 points
• Sensitivity = 0.65; specificity = 0.75

Inpatient rehab: (Castiglia et al., 2017)

• Mean IcaBI (Italian version of the Barthel Index) scores at admission: 43.7(22.9)
• Mean IcaBI scores at discharge: 80.5(21.5)

Neurologic Rehab Patients:

• Excellent : (Rollnik et al., 2011; n=273) Inter-rater reliability was r=0.849 (p<0.001). The findings suggest that the Early rehabilitation Barthel Index (ERBI) is a reliable and valid scale to assess early neurological rehabilitation patients (Modified version of BI)

Inpatient Rehab: (Castiglia et al., 2017; mean age 73.79 (9.9); orthopedic diseases n=22; stroke or Parkinson’s disease n=26, Italian version)

• Adequate to Excellent interrater reliability (ICC range from 0.74-0.96 depending on item)[TA1] 
• Moderate to substantial interrater agreement (Cohen’s K value range from 0.64-0.85 depending on item)

Neurologic Rehabilitation: (Hobart, 2001; n=149; female= 81; mean age= 46.2 (14.8) ; stroke n=45, head injury n=9, ms n=64, other n=34)

• Excellent internal consistency alpha = 0.94

Inpatient Rehab: (Castiglia et al., 2017; n=264; mean age 73.84 (13.3); internal diseases (cardiorespiratory, metabolic, oncological, urological) n=115; neurological disorders (Parkinson’s disease, stroke) n=83; orthopedic pathologies (hip/femur surgery, knee surgery, vertebral problems w/o neurological sequelae) n=66, Italian version)

• Excellent: Cronbach’s alpha = 0.94
• Excellent: when item 1 (“feeding”) removed, Cronbach’s alpha = 0.95

Neurologic Rehabilitation: (Hobart, 2001; n=149; female= 81; mean age= 46.2 (14.8) ; stroke n=45, head injury n=9, ms n=64, other n=34)

• Excellent correlation for between the disability FIM motor and FIM + FAM motor and 10 item BI (r = 0.84) for all 3

• The bladder item for the Barthel Index had fit residuals greater than 2, indicating the likely measurement of another construct (Morton et al, 2008)

• Multiple items indicated item redundancy and observed proportions deviated significantly from the Rasch model (Morton et al, 2008)

Neurological rehab patients : (Houlden, 2006; n=259 vascular brain injury due to single cerebral infarction (n = 75), spontaneous intracerebral haemorrhage (n = 34) and subarachnoid haemorrhage (n = 43), and 107 patients who had sustained traumatic brain injury, setting= regional neurologic rehabilitation unit)

• Highly responsive to detecting changes:
  • In all patient groups there was a significant improvement (Wilcoxon's rank sum, P<0.0001) in the Barthel Index (mean change score: vascular 3.9, traumatic 3.95). Barthel Index (effect size: vascular 0.65, traumatic 0.55)

Inpatient rehab: (Castiglia et al., 2017; n=53; mean age 72.71 (12.79); orthopedic diseases n=21; PD or stroke n=32; mean hospital duration: 45.11(16.53) days, Italian version)

• Adequate: AUC = 0.72

Older Adults in Nursing Homes: (Bouwstra et al., 2018; n=244; mean age: 77 (9.7) Years, Collin and Wade’s BI)

• SEM =  1.1 points

Older adults in Nursing Homes: (Bouwstra et al., 2018)

• MDC = 3.0 points

Older Adults Acutely Hospitalized: (Wales et al. 2018; n=66; mean age = 78.05 (4.13) years; currently hospitalized)

• MDC = 27.26 

Older Adults with Femoral Neck Fracture: (Unnanuntana et al., 2018; n = 81; mean age = 76.0 (8.2) years; 2 weeks, 3 months, and 12 months post-op)

• MCID = 9.8 points

Older Adults with Hip Fracture: (Gonzalez et al, 2017; n=1094; mean 

age=76.6 (7) years)

• <85 indicates risk of mortality 6 months post fracture (sensitivity: 63.64%, specificity: 64.77%)

Older Adults in a Nursing Home: (Bouwstra et al., 2018, Collin & Wade’s BI)

• Median score at admittance: 11
• Median score at discharge: 16.3
• Mean change score between assessments: 4.8 (4.0)

Older Adults with Fractures: (Gonzalez et al., 2018)

*The worsened and unchanged groups were derived from the change score calculated in the Lawton-Brody Index (LBI). Patients with LBI at baseline equal or lower than at follow-up were classified as “unchanged or improved”. Patients with LBI at baseline higher than at follow-up were considered “worsened”.

Outpatient Rehabilitation - Older Adults: (Hormozi, et al., 2019; n=395; mean age = 67.3 (6.04), Iranian version)

• Mean BI Score: 73.25 (SD = 31.55)
• 41.3% were fully independent and scored 100 points; 15.4% were physically dependent in at least one area
• Mean BI score for those with arthritis: 85
• Mean BI score for those with stroke: 67
• Mean BI score for those with falling: 70
• Mean BI score for those with incontinence: 77

Osteoporotic Hip Fracture:

(Mayoral, et al., 2019; n = 208; 1, 3, and 6 months post-injury; Spanish sample)

Community-Dwelling Elderly People in Japan:

(Saito, et al., 2016; n = 346; Mean Age = 76.0 (6.9); Japanese sample)

• Mean BI score (HE group): 99.4 ± 2.1
• Mean BI score (FE group): 83.6 ± 18.2

Older Adults with Femoral Neck Fracture: (Unnanuntana et al., 2018)

• Baseline range = 15-95 (80)
• Baseline median (IQR) 60 (50, 70)

Hospitalized Older Adults:  (Wales et al., 2018)

• Baseline mean: 68.64 (16.77)
• Baseline range: 25-100

Outpatient Rehabilitation - Older Adults: (Hormozi, et al., 2019)

• ICC = .936 (95% CI: .895-.965)

Elderly: (Richards et al, 1998; n = 94; mean age = 78.4 years)

• Fair to Good interrater reliability depending on the activity assessed.

Older Adults in Nursing Homes: (Bouwstra et al., 2018)

• Excellent: Cronbach’s alpha = .83

Older Adults with Hip or Wrist Fracture: (Gonzalez et al., 2017; 

n=2102; Mean Age: 80.1 (7.02) years)

• Excellent: Cronbach’s alpha = .91-.92

Older Adults in Residential Care Settings: (Gonzalez et al.,

 2017; n=234; Mean Age: 81 (7.06) years)

• Excellent: Cronbach’s alpha = .92

Community-Dwelling Older Adults: (Gonzalez et al., 2017; n=1106; Mean Age: 72.07 (7.83) years)

• Excellent: Cronbach’s alpha = .88-.92

Geriatric Outpatients: (Hormozi, et al., 2019)

• Excellent: Cronbach’s alpha = .938, p<.001
• Item- Total Correlation (ITC) - Cronbach’s alpha if Item Deleted:
  • Bowel: .936
  • Bladder: .933
  • Grooming: .932
  • Toilet use: .924
  • Feeding: .930
  • Transfer: .924
  • Mobility: .920
  • Dressing: .928
  • Stairs: .930
  • Bathing: .931

Hospitalized older adults: (Wales et al., 2018)

• Poor Cronbach’s alpha = .66
• “Tools showed poor-to-excellent internal consistency (α = 0.27–0.92).”

Predictive validity: 

Older Adults with Hip Fracture: (Gonzalez et al., 2018)

• Excellent correlation between BI score and mortality (r = .67)

Older Adults with Hip and Wrist Fracture: (Gonzalez et al., 2017; n=2102; mean age: 80.1 (7.06) years)

Convergent validity 

• Excellent correlation between LBI and Barthel Index (r = .76 and .58)
• Excellent correlation between FD-W and Barthel Index (r = .73)
• Adequate correlation between PCS-12 and Barthel Index (r =.57) 

Discriminant validity

• Poor correlation between MCS-12 and Barthel Index (r = .20)
• Poor correlation between PD-W and Barthel Index (r = -.25)

Community-Dwelling Older Adults:  

Discriminant validity (Gonzalez et al., 2017; n=1106; mean age: 72.07 (7.83) years)

• Poor correlation between DUFFS and Barthel Index (r=.10)

Community-Dwelling Elderly People in Japan

Convergent validity (Saito et al., 2016)

• Poor correlation with FIDS (r = 0.25) in HE group
• Excellent correlation with FIDS (r = 0.81) in FE group

Older Adults with Femoral Neck Fracture: 

Convergent validity (Unnanuntana et al., 2018) 

• Adequate to Excellent correlation with DEMMI at baseline, 3-months, and 12 months post-op (r = 0.580, 0.688, 0.678)
• Poor to Adequate correlation with EQ-VAS at baseline, 3-months, and 12-months post-op (r = 0.278, 0.323, 0.317)
• Adequate to Excellent correlation with 2-min walk test at baseline, 3-months, and 12-months post-op (r = 0.594, 0.630, 0.651)
• Adequate to Excellent correlation with TUG at baseline, 3-months, and 12-months post-op (r = -0.490, -0.743, -0.843)

Elderly: (Morton,  et al. 2008)

• The range of person abilities exceeded the range of item difficulty at both ends of the scale

• The original Barthel Index had a ceiling effect at hospital admission and hospital discharge

Older Adults in a Nursing Home: (Bouwstra et al., 2018)

Floor effect:

• Adequate: At baseline: 0.5%
• Adequate: At discharge: 0.6%

Ceiling effect:

• Adequate: At baseline: 2.0%
• Poor: At discharge: 22%

Older Adults with Fractures: (Gonzalez et al., 2018)

*The worsened and unchanged groups were derived from the change score calculated in the Lawton-Brody Index (LBI). Patients with LBI at baseline equal or lower than at follow-up were classified as “unchanged or improved”. Patients with LBI at baseline higher than at follow-up were considered “worsened”.

Community-Dwelling Elderly People in Japan: (Saito et al., 2016)

Floor Effect:

• Excellent, no floor effect (0%)

Ceiling Effect:

• HE group 
  • Poor ceiling effect on TMIG-IC score 0-10 (72.0%)
  • Poor ceiling effect on TMIG-IC score 11-12 (88.9%)
  • Poor ceiling effect on TMIG-IC score 13 (94.9%)
• FE group 
  • Adequate ceiling effect on those requiring long-term care 3-5 (4.2%)
  • Poor ceiling effect on those requiring long-term care 1-2 (21.7%)
  • Poor ceiling effect on those requiring support 1-2 (63.2%)

Older Adults with Femoral Neck Fracture: (Unnanuntana et al., 2018) 

Floor Effect:

• Excellent, no floor effect (0% at 3-months and 12-months post-op)

Ceiling Effect:

• Poor (25.9% 3-months post-op, 51.9% 12-months post-op)

Older Adults with Femoral Neck Fracture: (Unnanuntana et al., 2018) 

• Large effect estimate 
  • Standardized response mean = 2.65
  • Effect size = 2.13

Chronic Stroke: (Hsieh et al, 2007; n = 56; mean time post-stroke = 1197.1 (1281.8) days; mean age = 59.4 (14.6) years; Taiwanese sample)

• SEM = 1.45 points

Chronic Stroke: (Hsieh et al, 2007)

• MDC = 4.02 points

Acute Stroke: (Hsieh et al, 2007; n = 43; mean age = 55.4 (14.6) years; mean time since stroke 7.04 (64.1) days; Taiwanese sample)

• MCID = 1.85 points

Acute Stroke: (Uyttenboogaar et al, 2005; n = 1034; mean age = 69.1 (12.8) years; Median BI score = 80 (IQR = 40 to 100); assessed 12 weeks post stroke)

• Cutoff scores that indicate a favorable outcome:

  • > 95 (sensitivity 85.6%; specificity 91.7%)

  • > 90 (sensitivity 90.7%; specificity 88.1%)

  • > 75 (sensitivity 95.7%; specificity, 88.5%)

Acute Stroke: (Hsueh et al, 2002; n = 118, mean age = 67.5 (10.9) years; mean number of days in rehab = 26; Taiwanese sample)

Chronic Stroke: (Lin et al, 2010; n = 45, mean age = 60.0 (12.6) years; mean time since stroke = 9 months; outpatient therapy by a trained physiotherapist at 1 week, 2 months, and 5 months of out patient rehab; Taiwanese sample)

Chronic Stroke: (Lin et al, 2009)

Ischemic Stroke: (Galeoto, et al., 2019; n = 99; Mean age = 73.82 (11.93); Mean time from ischemic stroke event to first administration = 3 days (1 day), Italian culturally adapted version)

• Mean administration time=5 minutes (SD=2.5)

Stroke: (Ghandehari, et al., 2012)

• Median BI score 57.52 (SD: 27.567)

Ischemic Hemispheric Stroke: (Kwakkel, et al., 2011):

• Median BI at 2.18 (1.19) days post-stroke: 7 points (IQ 3 to 12)
• Median BI at 6 months post-stroke: 19 points (IQ 16.75-20). 60.7% showed full independence

• Mean difference (between first and second assessment)=0.4 (95% CI .0.01 - .90)
• Reliability coefficient=2 (out of 20)
• Percentage agreement >75%

Acute Stroke: (Hsueh et al, 2001; n = 121; assessed at 14, 30, 90 and 180 days)

• Adequate to Excellent item level agreement among raters (kappa value range, 0.53-0.94)
• Excellent total score agreement (ICC = 0.94)

All Stroke Survivors: (Duffy et al. 2013; n = 543)

• Good to Excellent Interrater Reliability: (k = .62-.99)
• “Meta-analysis suggested overall very good reliability for standard face-to-face BI (κw, 0.95; 95% CI, 0.94–0.96 fixed effects modeling; κw, 0.93; 95% CI, 0.90–0.96 random effects modeling); there were insufficient suitable data to perform meta-analysis for other methods of BI assessment”

Ischemic Stroke: (Galeoto, et al., 2019)

• Intrarater reliability: Excellent - ICC = .987 (95% CI .975 - .993); Cohen’s Kappa = .69
• Interrater reliability: Excellent - ICC = .909 (95% CI .852-.948); Light’s Kappa = .79

Stroke: (Ghandehari, et al., 2012)

• Interrater variability was not significant (df = 4; F = 1.061; 95% confidence interval = 52.639-62.400; P = 0.379).

Acute stroke: (Hsueh et al, 2001)

• Excellent internal consistency (alpha = 0.89 to 0.90)

Acute Stroke: (Hsueh et al, 2002)

Barthel Index:

• Excellent internal consistency at admission (alpha = 0.84)
• Excellent internal consistency at discharge (alpha = 0.85)

BI-5 (a modified 5-item version):

• Excellent internal consistency at admission (alpha = 0.71)
• Excellent internal consistency at discharge (alpha = 0.73)

Ischemic Stroke: (Galeoto, et al., 2019):

• Excellent: Cronbach’s Alpha= .901 (p<.001)

Acute Stroke: (Hsueh et al, 2002)

• Excellent correlation between the FIM motor and 10 item BI at both admission and discharge (r > 0.92)
• Excellent agreement between the FIM motor and 10 item BI at both admission and discharge (ICC >0.83)

Acute Stroke: (Wade and Hewer, 1987; n = 976; age not reported; assessed within 7 days of onset)

• Excellent concurrent validity Between the modified BI and measure of motor ability using the Motricity Index (r = 0.73 to 0.77)

Ischemic Hemispheric Stroke: (Kwakkel, et al., 2011)

• Positive predictive value

            Post-Stroke Day 2: Excellent - .696 (95% CI, .645-.739)

                         Post-Stroke Day 5:  Excellent -.816 (95% CI, .766-.859)

                         Post-Stroke Day 9:  Excellent -.864 (95% CI, .815-.905)

• Negative predictive value

             Post-Stroke Day 2: Excellent- .778 (95% CI, .699-.844)

                          Post-Stroke Day 5: Excellent- .704 (95% CI, .626-.769)

                          Post-Stroke Day 9: Excellent- .613 (95% CI, .536-.676)

• Odds Ratio: 
  • Post-Stroke Day 2: 8.013 (95% CI, 4.192-15.316)
  • Post-Stroke Day 5: 10.533 (95% CI, 5.458-20.325)
  • Post-Stroke Day 9: 10.042 (95% CI, 5.082-29.842) 
  • *Used a cutoff of 7 points on the BI 

• Overall accuracy for correctly predicting outcome increased from 72.8% on day 2 to 77.2% on day 5, down to 76.6% on day 9 post-stroke

Acute Stroke: (Duncan et al, 1997; n = 304; mean age = 62.89 (13.17) years; time since stroke not specified)

• BI not sensitive to change among mild stroke/transient ischemic attack patients scoring at least 100 points.

Acute Stroke: (Hsueh et al, 2001)

Mild Stroke:  (Duncan et al, 1997)

• The Barthel Index, which has a ceiling effect and which captures only physical domains of health status, is not adequate for assessing the full impact of stroke-related disability

Acute Stroke: (Hsueh et al, 2002):
Responsiveness (Standardised Response Mean)

• The B1 and BI-5 are highly responsive in detecting changes:
• BI (Original Measure) = 1.2
• Bi-5 (Modified version) = 1.2

Acute Stroke: (Salbach et al, 2001; n = 53; mean age = 68 (13) years)

• The 5MWT (at a comfortable pace) at 5 weeks post-stroke was more responsive than the Berg Balance Scale or the Barthel Index
• All three measures were found to be responsive to change

Acute Stroke: (Tilling et all, 2001)

• Created an Excel Spreadsheet to calculate recovery trajectories using the modified BI; download it here: http://www.epi.bris.ac.uk/user/jons/stroke/

Acute Stroke: (Duncan et al, 1997; n = 304; mean age = 62.89 (13.17) years; time since stroke not specified):

• BI not sensitive to change among mild stroke/transient ischemic attack patients scoring at least 100 points (maximal score is 100 points).
• “Patients who score 100 on the Barthel Index have widely varying scores on the physical function subscale of the MOS-36; for example, fewer than 20% scored the maximum possible value on this subscale. Thus, if the Barthel Index is the only stroke outcome measure used, a decline in many domains of health status will be missed. The Barthel Index will also be ineffective in detecting the psychosocial dimensions of impaired function.”

Ischemic Hemispheric Stroke: (Kwakkel, et al., 2011)

• Day 2 Post-Stroke: Adequate AUC = 0.785 (SE = 0.035;

p < 0.001; 95% CI, 0.715-0.854)

• Day 5 Post-Stroke: Adequate AUC = 0.837 (SE = 0.031;

p < 0.001; 95% CI, 0.776-0.899)

• Day 9 Post-Stroke: Adequate AUC = 0.848 (SE = 0.030;

p < 0.001; 95% CI, 0.788- 0.908). 

• AUC was significantly different between Day 2 and Day 5 (z = 3.537, P < 0.001) and between Day 2 and 9 (z = 3.621, p < 0.001). No significant difference was found between the AUC of the ROC curves of Days 5 and 9 (z = 1.416, p=0.08).

Parkinson’s Disease: (Taghizadeh, et al., 2019; n = 260, mean age = 60.3 (12.3) years; female=28.1%; Mean time since diagnosis (SD): 6.6 (5.5); Stage 1: 33.9%; Stage 2: 36.5%; Stage 3: 20%; Stage 4: 9.6%, Iranian version) 

• SEM ON medication (1 hour after levodopa) for all (n=260): 3.41
• SEM OFF medication (12 hours after last evening dose) (n=260): 9.90

Parkinson’s Disease: (Taghizadeh, et al., 2019): 

• ON Medication Mean BI (SD) =94.0(12.6)
• OFF Medication Mean BI (SD) =80.0(23.3)

Parkinson’s Disease: (Taghizadeh, et al., 2019): 

• ON Medication Phase (n=260): ICC (95% CI): .84 (.78-.91)
• OFF Medication Phase (n=260): ICC (95% CI): .77 (.65-.85)

Parkinson’s Disease: (Taghizadeh, et al., 2019): 

• Inter-rater Reliability 
  • Excellent Reliability: ON Medication Phase (n=260): ICC (95% CI): .91 (.87-.94)
  • Excellent Reliability: OFF Medication Phase (n=260): ICC (95% CI): .90 (.86-.93)

Parkinson’s Disease: (Taghizadeh, et al., 2019): 

Convergent validity 

Parkinson’s Disease: (Taghizadeh, et al., 2019): 

• Floor Effect
  • Excellent: ON Medication Phase (n=260): 0%
  • Excellent: OFF Medication Phase (n=260): 0%
• Ceiling Effect
  • Poor: ON Medication Phase (n=260): 47.7%
  • Poor: OFF Medication Phase (n=260): 32.7%