Skip to main content

Table 3 Criterion validity results; the predictive ability of the frailty index for adverse health outcomes

From: Screening for frailty in primary care: a systematic review of the psychometric properties of the frailty index in community-dwelling older people

Study Outcome variable with events (n) Model Factors controlled for in model Effect measure 95% CI/SE Interpretation effect measure
Armstrong et al. [21] Mortality: 1676 Cox proportional hazards regression Age, gender FI: HR = 1.93 1.79-2.08 Most frail (15%) vs. least frail (60%) group
Institutionalization: 4550 (EFS: HR = 2.49) (2.32-2.68)
(CHESS: HR = 2.32) (2.15-2.51)
Drubbel et al. [22] Mortality/ED visits/institutionalization/out-of-hours GP surgery visits: 508 Cox proportional hazards regression Age, gender, consultation gap HR = 1.166 1.129-1.210 Per deficit increase in FI score
Fang et al. [32] Recurrent falls: 109 Logistic regression Age, gender, education OR = 1.54 1.34-1.76 Per one-unit increment in FI score
Recurrent fractures: 174 Logistic regression Age, gender, education OR = 1.07 0.94-1.22 Per one-unitincrement in FI score
Mortality: 1101 Cox proportional hazards regression Age, gender, education, falls, fractures HR = 1.29 1.25-1.33 Per one-unit increment FI score
García-González et al. [33] Mortality: 279 Cox proportional hazards regression Age, gender HR = 6.45 4.10-10.14 Most frail (FI 0.35-0.65) vs. least frail group (0.00-0.07)
Gu et al. [23] Mortality: 5,753 Weibull proportional hazards regression Age, ethnicity, urban–rural residence, SES, family/social connection and support, health practices Men (65–79):   Most frail vs. least frail quartile
HR = 4.56 0.96
Women (65–79):  
HR = 3.84 1.01
Hogan et al. [37] Mortality: 170 Logistic regression Age, gender, co-morbidity RR = 2.35 1.56-3.54 All analyses: most frail (FI > 0.30) vs. least frail group (FI < 0.20)
≥ 1 hospitalization: 424 Logistic regression Age, gender, co-morbidity RR = 1.28 1.04-1.57
Institutionalization: 204 Logistic regression Age, gender, co-morbidity RR = 3.30 2.29-4.76
Kulminski et al. [24] Mortality: 421 Cox proportional hazards regression Age, gender, FP FI: RR = 1.035 1.026-1.045  
(FP: RR = 1.014) (1.009-1.019) Per 1% increment in FI score (or FP)
Kulminski et al. [25] Mortality: 2146 Cox proportional hazards regression Age, gender RR = 1.029 1.001 Per 1% increment in FI score
Lucicesare et al. [26] Mortality: 147 Cox proportional hazards regression Age, gender, CSBA score FI: HR = 5.26 1.05-26.42 ?
(CSBA score: HR = 1.52) (1.28-1.81)
Mitnitski et al. [34] Mortality (%/yr) 3.7-20.6 Cox proportional hazards regression Age, gender CSHA-s: HR = 1.031 0.003 Per deficit increase in FI score
CSHA-c: HR = 1.054 0.007
CSHA-i: HR = 1.046 0.009
SOPSA: HR =1.079 0.022
NHANES: HR = 1.011 0.003
Searle et al. [19] Mortality: ? Cox proportional hazards regression Age, gender HR = 1.03 1.02-1.04 Per 0.01 increase in FI score
Shi et al. [35] Mortality: 1,155 Cox proportional hazards regression Age, gender HR = 1.13 1.09-1.47 Per deficit increase in FI score
Song et al. [29] Mortality: 1,208 Cox proportional hazards regression Age, gender FI: RR = 1.57 1.41-1.74 Per FI level (FI ≤ 0.08; FI between 0.08-0.25; FI ≥ 0.25).
Theou et al. [36] Mortality: 1002 Cox proportional hazards regression Age, gender, nr. of ADL disabilities, nr. of chronic diseases FI 1: HR = 1.11 1.06-1.17 Per 0.1 increase in FI score
Woo et al. [31] Change in ADL score 0–3 yrsa Linear regression Age, gender, ADL score at baseline B = −4.99 −7.68 - −2.30 Per 1.0 increase in FI score
Change in mental score 0-3 yrsa Linear regression Age, gender, mental score at baseline B = −2.23 −4.11 - −0.35 Per 1.0 increase in FI score
Change in hospital days 0–3 yrsa Linear regression Age, gender, hospital days at baseline B = 45.74 28.16 – 63.33 Per 1.0 increase in FI score
New diseases at three yrsa Ordinal logistic regression - For FI = 0.00, predicted probability ≥ 1 new disease = 17.4% Predicted probabilities for new diseases at 3 years
For FI = 0.50, predicted probability ≥ 1 new disease = 52.2%
  1. aRegression models with 3-year outcomes reported due to excess LTFU at 10 years. 95% CI = 95% Confidence Interval; adm. = admission; ADL = Activities of Daily Living; B = beta; CHESS = Changes in Health, End-Stage Disease and Signs and Symptoms Scale; CSBA = Conselice Study of Brain Ageing; CSHA = Canadian Study of Health and Ageing; DI = Deficit Index (Frailty Index); EFS = Edmonton Frail Scale; FI = Frailty Index; FP = Frailty Phenotype; HR = hazard ratio; NHANES = National Health and Nutrition Examination Survey; OR = odds ratio; PBA = Personal Biological Age; RR = relative risk; SE = standard error; SOPSA Sydney Older Persons Studies on Aging.