Table 1 Results: Community, acute care and residential aged care settings

Community

Clegg et al., 2016

UK [7]

19.42

Cohort

931,541

75.0 (7.2)

55.0

Development and internal validation cohort = 0.14 (0.09)

External validation cohort = 0.15 (0.10)

<  0.12 (fit)

>  0.12–0.24 (mild frailty)

>  0.24–0.36 (moderate frailty)

>  0.36 (severe frailty)

Study authors derived FI quartiles using 99th centile as upper limit.

Wallace et al., 2019

USA [8]

17.94

Cross-sectional

456

89.7 (6.1)

69.0

0.42 (0.18)

≥ 0.41 (median) (high frailty)

0.24–0.41 (low frailty)

0.42 (intermediate frailty)

0.43–0.60 (high frailty)

Study authors derived FI categories utilising the median and then using the mean + −  1 SD.

Rockwood et al., 2011

Canada [9]

17.78

Cohort

14,127

44.3 (18.3)

54.2

0.07 (0.08)

≤ 0.03 (relatively fit)

0.03 < FI ≤ 0.10 (less fit)

0.10 < FI ≤ 0.21 (least fit)

>  0.21 (frail)

≥ 0.45 (most frail)

Referenced Rockwood et al.’s study [4], which demonstrated construct and predictive validity of Clinical Frailty Scale (CFS) categories in study of community-dwellers. CFS categories ≥4 (‘apparently vulnerable’- ‘severely frail’) corresponded with a mean FI > 0.21.

Theou et al., 2013

11 European countries [10]

11.00

Cohort

27,527

65.3 (10.5)

54.8

Not reported

≥ 0.25 (frail)

Referenced Rockwood et al.’s study [3], which demonstrated the construct and predictive validity of FI > 0.25 in community-dwellers.

Blodgett et al., 2015

Canada [11]

8.52

Cross-sectional

4096

63.4 (10.3)

53.3

0.20 men

0.17 women

≤ 0.10 (non-frail)

0.10 < FI ≤ 0.21 (vulnerable)

0.21 < FI ≤ 0.45 (frail)

>  0.45 (most frail)

Referenced Hoover et al.’s study [12], which demonstrated the predictive validity of FI > 0.21 cut-off as well as four frailty categories (as listed here).

Thompson et al., 2018

Australia [13]

7.23

Cohort

909

74.4 (6.2)

55.0

0.23 (0.15)

≤ 0.21 (non-frail and prefrail)

>  0.21 (frail and most frail)

Referenced Hoover et al. [12]

Ntanasi et al., 2018

Greece [14]

6.76

Cross-sectional

1740

73.4 (5.4)

59.0

Not reported

> 0.25 (frail)

Referenced Rockwood et al. [3]

Song et al., 2010

Canada [15]

6.51

Cohort

2740

74.0 (6.6)

60.8

Not reported

≤ 0.08 (non-frail)

0.09–0.24 (prefrail)

≥ 0.25 (frail)

Referenced Rockwood et al. [3]

Ravindrarajah et la., 2017

UK [16]

6.00

Cohort

144,403

85.1 (4.9) – 88.0 (5.4)

50–68

Not reported

<  0.12 (fit)

>  0.12–0.24 (mild frailty)

>  0.24–0 36 (moderate frailty)

>  0.36 (severe frailty)

Referenced Clegg et al.’s study [7], which demonstrated the predictive validity of these eFI categories in UK community-dwellers.

Lansbury et al., 2017

UK [17]

5.22

Cross-sectional

589

82.7

58.1

0.23 (0.12)

<  0.12 (fit)

>  0.12–0.24 (mild frailty)

>  0.24–0.36 (moderate frailty)

>  0.36 (severe frailty)

Referenced Clegg et al. [7]

Acute Care

Joseph et al., 2014

USA [18]

15.46

Cohort

250

77.9 (8.1)

30.8

0.21 (0.10)

<  0.25 (non-frail)

≥ 0.25 (frail)

Referenced Searle et al.’s study [2], which did not report FI categories.

Chong et al., 2018

Singapore [19]

5.49

Cohort

210

89.4 (4.6)

69.5

Not reported

≥ 0.25 (frail)

Nil

Joseph et al., 2016

USA [20]

5.03

Cohort

220

75.5 (7.7)

44.0

0.28 (0.13)

<  0.25 (non-frail)

≥ 0.25 (frail)

Referenced study by co-authors [18] and a conference abstract.

Poudel et al., 2016

Australia [21]

4.93

Cohort

1418

81 (6.8)

55.0

0.32 (0.15)

<  0.25 (low)

0.26–0.39 (medium)

≥ 0.4 (high)

Referenced Rockwood et al. [3, 4]

Also referenced Singh et al.’s study [22], which utilised similar categories and referenced Rockwood et al. [3, 4]

Andrew et al., 2017

Canada [23]

4.87

Case control

884

78.8 (7.9) – 80.6 (9.0)

55.0–56.9

Cases = 0.2 (0.11)

Controls = 0.22 (0.13)

<  0.10 (non-frail)

>  0.10–0.21 (prefrail)

>  0.21–0.45 (frail)

Referenced Hoover et al. [12]

Dent et al., 2014

Australia [24]

4.22

Cohort

172

Not reported

72.0

Not reported

<  0.2 (robust)

0.2–0.45 (prefrail)

>  0.45 (frail)

Referenced Rockwood et al. [4]

Mueller et al., 2016

USA [25]

4.16

Cohort

102

61.9 (15.8)

39.2

0.23 (0.12)

<  0.25 (non-frail)

≥ 0.25 (frail)

Referenced Joseph et al. [18]

Zeng et al., 2015

China [26]

2.92

Cohort

155

82.7 (7.1)

12.9

Not reported

<  0.22 (least frail)

>  0.46 (least fit)

Authors determined FI scores below which all participants survived and above which all participants died.

Hao et al., 2019

China [27]

2.86

Cohort

271

81.1 (6.6)

20.3

0.26 (0.16)

>  0.25 (frail)

Referenced Rockwood et al. [3]

Also referenced several other studies that utilised the same categories and referenced Rockwood et al. [3, 4].

Arjunan et al., 2019

Australia [28]

2.83

Cohort

258

79.0 (8.0)

54.0

0.42 (0.13)

≤ 0.40 (less frail)

>  0.40 (more frail)

Authors determined the FI cut point for optimal sensitivity and specificity for four adverse outcomes.

Residential Aged Care

Theou et al., 2018

Spain [29]

4.00

RCT

50

75.3 (7.3)

70.0

0.23 (0.1)

<  0.20 (non-frail)

0.20–0.30 (vulnerable/mildly frail)

>  0.30 (moderately/severely frail)

Study authors categorised the FI in 0.1 groups then combined groups due to the small number of participants. They referenced two studies [30, 31], which both categorised the FI into 0.1 increments to facilitate regression analyses.

Shaw et al., 2019

Canada [32]

3.84

Cohort

116

84.2 (0.9)

56.0

0.36 (0.01)

<  0.27 (non-frail)

≥ 0.27 (frail)

Study authors demonstrated a bimodal distribution of the continuous FI with ‘crossing points’ at an FI = 0.27.

Theou et al., 2018

Australia [33]

3.26

Cohort

383

Median 88.0

IQR 4.0

77.6

0.33 (0.24–0.46)

≤ 0.10 (non-frail)

0.10–0.21 (vulnerable)

0.21–0.44 (mild/moderate frailty)

≥ 0.45 (most frail)

Referenced study by co-authors [34], which utilised the same categories and referenced Hoover et al. [12]

Maclagan et al., 2017

Canada [35]

2.33

Cohort

41,351

Not reported

64.7

Not reported

<  0.20 (robust / non-frail)

0.20–0.30 (pre-frail)

>  0.30 (frail)

Referenced study by co-authors [36], which utilised the same FI categories, referencing Searle et al. [2], co-authors Hogan et al. [37] (see below) and Kulminski et al. [38]

Kulminski et al.’s study [38] demonstrated the predictive validity of similar FI categories in community-dwellers.

Hogan et al., 2012

Canada [37]

2.03

Cohort

1066

84.9 (7.3)

76.7

Not reported

<  0.20 (robust / non-frail)

≥ 0.20 ≤ 0.30 (prefrail)

>  0.30 (frail)

Referenced Searle et al. [2] and Kulminski et al. [38]

Buckinx et al.,

2017

Belgium [39]

1.24

Cohort

662

83.2 (9.0)

72.5

Not reported

<  0.25 (robust)

≥ 0.25 (frail)

Referenced a review article [40] and Mitnitski et al.’s study [41], which based ‘FI-Biomarker’ categories on maximum separation of mortality curves in community-dwellers.

Ambagtsheer et al., 2020

Australia [42]

1.23

Cross-sectional

592

Median 88.0

IQR 9.0

66.6

0.20 (0.08)

≤ 0.10 (non-frail)

> 0.10 ≤ 0.21 (pre-frail)

>  0.21 (frail)

Referenced Hoover et al. [12]

Ambagtsheer et al., 2020

Australia [43]

1.03

Cross-sectional

592

Median 88.0

IQR 9.0

66.6

Not reported

≤ 0.21 (non-frail)

>  0.21 (frail)

Referenced Hoover et al. [12]

Ge et al., 2019

China [44]

0.72

Cross-sectional

302

82.7 (8.5)

71.2

0.27 (0.11)

<  0.21 (non-frail)

0.22–0.44 (frail)

≥ 0.45 (frailest)

Referenced Hoover et al. [12]

Stock et al.,

2017

Canada [45]

0.54

Cohort

1066

84.4 (7.3)

76.7

Not reported

<  0.20 (non-frail)

0.20–0.30 (prefrail)

>  0.30 (frail)

Referenced study by co-authors [37] (see above).