You are here

Frailty as a Predictor

Chapter 04 – Frailty in the Elderly

Frailty measured in a variety of ways is associated with a number of different clinical outcomes. In a UK study of over 11,000 emergency admissions in people aged over 75, the clinical frailty scale (CFS) was an independent predictor of inpatient mortality (Odds Ratio [OR] = 1.6), transfer to a geriatric ward (OR = 1.33) and length of stay > 10 days (OR = 1.19), but not a predictor of 30 day readmission. Its use as a predictor was demonstrated after adjustment for age, gender, Charlson Co-morbidity index and history of dementia and/or current cognitive issues. A limitation of this study by Wallis et al (2015) was that the CFS was completed within 72 hours of admission in 81% of cases but 23.5% of patients had missing CFS information.

A study from the USA, measuring frailty in almost 600 patients aged 65 or older and admitted for elective surgery, used a validated scale (0-5) which included weakness, exhaustion, low physical activity, slowed walking speed and weight loss (patients with a score ≥ four were classified as frail). Pre-operative frailty was associated with an increased risk of post-operative complications (OR = 2.54), prolonged length of stay (OR = 1.69) and discharge to a skilled or assisted living facility after previously residing at home (OR = 20.48). Those scoring two to three were classified as having intermediate frailty and were also found to have increased risk of post-operative complications, prolonged length of stay, and discharge to residential care.

Within oncology there have been varying results using frailty to predict outcomes. A 2010 pilot study from Canada studied 110 patients aged 65 or older. This study found those patients requiring cancer-related hospitalisations were more often those with colorectal or lung cancer and less often breast cancer. In addition, and not surprisingly, those with more advanced disease who had received more extensive treatment, were more likely to have cancer-related hospitalisations. However, they also found that none of the frailty markers predicted the outcome of cancer-related hospitalisation or GP visits. It is important to note that 60% of those included in the sample were under the age of 75, in addition 65% of the study sample were classified as fully active. With this in mind, it is not surprising that this relatively fit study group failed to identify frailty as a marker of adverse outcomes.

In a variety of specific tumour types, frailty is seen to be an independent predictor. Ommundsen et al (2014) categorised a cohort of 178 colorectal cancer patients, aged 70 years or older, into being frail or non-frail. The frailty was assessed through geriatric assessment and was found to be present in 43% of individuals and predicted both one-year and five-year survival post-surgery. In localised and regional disease, the impact of frailty on five-year survival was comparable with that of TNM staging. In this study the geriatric assessment took between 20 and 60 minutes and included a Barthel Index, a medication review, comorbidity, nutritional status, cognitive function and identification of depression.

A systematic review of the prevalence and outcomes of frailty in older cancer patients identified data from 20 studies including almost 3,000 participants. The median reported prevalence of frailty was 42% with a further 43% being pre-frail. A median of 32% of patients were classified as fit (range 11% to 78%). Frailty was independently associated with all-cause mortality (HR = 1.97) post-operative mortality in both frail (HR = 2.67) and pre-frail (HR = 2.33) patients. Both treatment complications were more frequent (OR = 4.86) and post-operative complications at 30 days (HR = 3.19) in those classified as frail. These studies were heterogeneous in that; only 16 used CGA for the diagnosis of frailty, in nine studies the member of the multidisciplinary team who completed the CGA was identified, and in nine studies there was a retrospective review of medical records to calculate the CGA. Seven studies carried out face to face interviews, two used telephone consultations and five used self-reported questionnaires. Studies either dichotomised patients into frail or fit (N=8); four used impairments in >2 GCA domains to define frailty; two used >3 as a cut off for frailty and one study defined frailty as >2 CGA impairments or cognitive impairment alone. A final study reported that two independent physicians defined frailty on the basis of CGA but this was not fully defined. In two studies a detailed description of the definition of frailty was not described. The majority of studies were from the US (7), Canada (2), Belgium (2), Norway (2) and Australia (2).

It is clear that frailty requires definition in a consistent manor and that its correlation with the results of CGA are defined. A systematic review by Hamaker et al (2012) identified a number of screening methods for predicting those patients who need to have their frailty defined by a CGA. They showed a variety of sensitivity and specificity (Table below).

Sensitivity and specificity of comprehensive geriatric assessment

 Comprehensive geriatric assessmentSensitivity for predicting frailtySpecificity for predicting frailty
Vulnerable Elders Survey (VE13)68%78%
Geriatric 8 (G8)87%61%
Triage risk screening tool (TRST 1+)92%47%
Groningen Frailty Index (GFI)57%86%
Emed Criteria31%91%
Barber59%79%
Abbreviated CGA (aCGA)51%97%

The G8 and TRST 1+ had the highest sensitivity for frailty, but poor specificity and negative predictive value.  This review supports the fact that available screening methods have insufficient discriminative power to select patients for further assessment and therefore patients should receive a CGA.

« Previous Page Next Page »