Matching the Assessment Method to the Goal
Over the past two decades, efforts to distinguish frail from nonfrail older adults have led to the introduction of dozens of operational definitions with varied theoretical basis, clinical feasibility, included domains and assessment items. Thus, deciding which frailty assessment is appropriate in a given context can be challenging.
A recent literature review identified 67 distinct published frailty instruments; of these, nine were highly cited (200 or more citations; Buta et al., 2016):
- Physical Frailty Phenotype (PFP, also called CHS frailty phenotype) (Fried et al., 2001)
- Deficit Accumulation Index (DAI, also called Frailty Index) (Mitnitski et al., 2001; Mitnitski et al, 2004; Rockwood et al, 2007; Rockwood & Mitnitski, 2007; Rockwood et al., 2006)
- Gill Frailty Measure (Gill et al., 2002)
- Frailty/Vigor Assessment (Speechley & Tinetti, 1991)
- Clinical Frailty Scale (Rockwood et al., 2005)
- Brief Frailty Instrument (Rockwood et al., 1999)
- Vulnerable Elders Survey (VES-13) (Saliba et al., 2001)
- FRAIL Scale (Abellan van Kan et al., 2008; Abellan van Kan et al., 2008)
- Winograd Screening Instrument (Winograd et al., 1991)
Which assessment method is appropriate for a particular research project, clinical setting or use depends on the study goal and setting as well as the frailty definition one intends to measure. Eight major uses for frailty instruments have been identified in the literature: risk assessment for adverse health outcomes (31% of all uses); etiologic studies of frailty (22%); methodology studies (14%); biomarker studies (12%); study inclusion/exclusion criteria (10%); estimating prevalence as primary goal (5%); clinical decision making (2%); and interventional targeting (2%) (Buta et al., 2016). The most common assessment context has been observational studies of older community-dwelling adults. Better guidance is needed for matching the selection of assessment method to the conceptual basis and goals of one’s consideration of frailty. Please visit our section on Guidance on Selecting a Frailty Instrument to learn more.
Developing Frailty Assessment Methods
If an assessment method is to be strongly recommended for capturing a given definition of frailty or accomplishing a given set of goals, it is important to ground the method in evidence that it measures what it intends or performs as it is designed to (Bandeen-Roche et al., 2019). The process of producing such evidence is known as “validation.”
Many frailty instruments were developed and considered validated based on a tendency for older adults the instruments identify as “frail” to more often experience adverse health outcomes than those identified as “not frail.” This is one aspect of validity—that a frailty measure should identify individuals at high risk for adverse outcomes. Other aspects of validity may also be important, however—for example:
- Are the instrument’s component measures specific to one’s definition of frailty, or do they capture extraneous information?
- Do the component measures co-occur or segregate as they should if they do indeed measure one’s intended concept?
- Does the assessment identify individuals who suffer from frailty’s underlying causes as “frail”?
- Do individuals identified as “frail” experience worse outcomes following stressors than those identified as “not frail”?
Only by answering such questions can clinical care be guided appropriately to frail persons, or interventions designed to manage risk of adverse outcomes or prevent frailty in the first place.
Some readers will be familiar with validation based on expertise or exposure to the psychometric field, where the process has been most fully developed. They will recognize that aspects of validity have formal labels: “Content validity” for aspect #1 exemplified above; “internal construct validity” for aspect #2; “external construct validity” for aspects #3 and #4; and “criterion (or specifically, predictive) validity” for the aspect exemplified in the first sentence of the last paragraph. Relatively few validation studies of frailty assessments have considered aspects beyond predictive validity. PFP components were selected for content validity. PFP has been shown to have internal construct validity with respect to the hypothesized clinical presentation of frailty as a medical syndrome (Bandeen-Roche et al., 2006). It has been externally validated for its association with multisystem dysregulation as a hypothesized etiology (Fried et al., 2009) and shown to distinguish those who do poorly from those who do well following surgeries and other stressors (Makary et al., 2010).