Background
Frailty is highly prevalent in older people. Up to 40% of older people can be considered as frail and an increasing trend can be expected [
1]. Next to its high prevalence, frailty is characterized by its seriousness as it is related to an increased risk of adverse health outcomes such as disability [
2‐
4]. Disability is defined as difficulty or dependency in the execution of activities of daily living and it is associated with increased health service utilization and related costs. Frailty and disability are separate but overlapping concepts. On the one hand, frailty predicts disability. On the other hand, disability may well exacerbate frailty [
5]. With regard to a growing frail population and limited health care expenditures, disability in community-dwelling frail older people is suggested to be a public health problem [
3]. Therefore disability prevention in community-dwelling frail older people is considered to be a priority for research and clinical practice in geriatric care [
6].
Several authors emphasize a two-step approach in preventive interventions for community-dwelling frail older people, in which screening is followed by extensive assessment. With valid (screening) instruments to identify frail older people this approach may avoid costs and the unnecessary assessment of healthy people [
7,
8]. During the last few decades, various instruments, based on various definitions, have been developed to detect frailty. This has lead to a diversity of prevalence estimates of frailty [
4]. Little is yet known about the reliability and validity of these instruments and no gold standard exists. Therefore, more insight into the psychometric properties of frailty instruments is relevant for geriatric care and research in this area [
4].
Frailty instruments have been developed from the point of view of different perspectives on frailty [
9]. From a physiological perspective physical frailty markers, such as unintentional weight loss or weakness (grip strength), are used to identify frail older people [
5]. Next to physical factors, a multifactorial perspective on frailty also takes psychological, social and environmental factors into account [
10]. An example of such an instrument is the Frailty Index [
11,
12], which combines, for example, physical frailty markers such as weight loss and grip strength with other factors such as cognition, mood or limitations in (instrumental) activities of daily living. Frailty may be elaborated more sharply if it is described from a physiological perspective, however, the usefulness of this perspective in daily practice is questioned, as frailty cannot be separated from other factors such as cognition, mood or social support [
9].
Frailty instruments can be divided into self-report and performance-based instruments [
9]. It is assumed that performance-based instruments provide more precise and valid answers [
13,
14]. Although they are less influenced by socio-demographic variables, personality and cognitive and affective factors, they are more sensitive to non-response, changes in time and differences in the execution of activities. Furthermore, they are less easy to conduct and time-consuming [
13‐
15]. Self-report measures are believed to be an efficient method for reaching large groups and for providing high response rates and reliable and valid answers [
15].
In this study, we present the psychometric properties of frailty instruments that define frailty from a multifactorial perspective and are applicable for postal screening of community-dwelling older people. Given this objective and the target population, the Groningen Frailty Indicator (GFI) [
16], the Tilburg Frailty Indicator (TFI) [
17] and the Sherbrooke Postal Questionnaire (SPQ) [
8] were chosen. The GFI and the SPQ have been used in previous studies for the purpose of postal screening [
18‐
22], however, empirical evidence about the psychometric properties of the GFI, the TFI, and the SPQ is still scarce. The purpose of the present study was to evaluate and compare their psychometric properties.
Discussion
The purpose of the present study was to evaluate and compare the psychometric properties of three screening instruments that define frailty from a multifactorial perspective and which are applicable for postal screening in community-dwelling frail older people. The chosen instruments were the Groningen Frailty Indicator (GFI), the Tilburg Frailty Indicator (TFI) and the Sherbrooke Postal Questionnaire (SPQ).
From the present study we may conclude that: (1) prevalence estimates of frailty ranged between 40.2% (TFI), 46.3% (GFI) and 59.1% (SPQ); (2) the agreement in identifying frailty between the GFI and the TFI was satisfactory (kappa = 0.74) and the agreements between the SPQ and the GFI and the TFI, respectively, were much lower; (3) both the GFI and the TFI had high internal consistency in contrast to the SPQ; (4) the GFI and the TFI had better construct validity in comparison with the SPQ.
Prevalence estimates of 40% to 60% found in the present study can be considered as high. It is important to bear in mind that prevalence estimates strongly depend on the interpretation of the concept of frailty and the approach that is chosen to measure it [
32]. In a recent study by Santos-Eggimann and colleagues [
33], a distinction was made between frail and pre-frail older people based on the frailty phenotype of Fried and colleagues [
5,
34]. In a Dutch sample of community-dwelling older people, Santos-Eggimann and colleagues [
33] found a frailty prevalence of 11.3%, while 38.5% were identified as pre-frail. These results indicate that the instruments in our study, based on the proposed cut-off points, may identify pre-frail instead of frail older people. Further research is needed to provide a better view on relevant cut-off points for frailty instruments. Longitudinal studies are needed to investigate the predictive power of instruments to identify older people who are at risk for adverse health outcomes in the near future.
Steverink and colleagues [
16] suggested that the GFI is an internally consistent scale with positive indications for construct and clinical validity. The present study supports these findings. Similar results for the TFI may be explained by seven out of fifteen items of the TFI being identical with the GFI. These items are about hearing and vision capacity, unintentional weight loss and psychosocial and cognitive functioning. Please see Additional file
1:
Frailty Instruments: Overview of all items for more information about the instruments. Scores on the Sherbrooke Postal Questionnaire were higher for males compared with females. This finding is inconsistent with the literature [
27]. However, other findings on the Sherbrooke Questionnaire (higher score with higher age, lower educational level and lower incomes) are well in line with the literature [
27]. Previous studies about the SPQ have reported positive results regarding the predictive validity of the SPQ [
8,
19,
20,
22], however, in the present study the SPQ showed less reliability and construct validity. Conclusions about predictive validity can not be drawn for any of the three instruments.
The findings of the present study should be interpreted in the context of potential limitations. First, little is known about the test-retest reliability of the instruments. Second, there is no gold standard available as an external criterion of frailty. Future studies could analyse the predictive validity of the frailty instruments with respect to disability, health service utilization and mortality. Last, the SPQ was not fully used according to the protocol, as non-respondents were excluded from analyses. According to the protocol of the SPQ [
8], non-respondents should also be considered at risk (which would have resulted in a prevalence estimate of 67.0% instead of 59.1%). The strengths of the present study are the comparisons of the psychometric properties of the frailty instruments, the proven feasibility of the postal procedure [
24] and the response rate of 77.4%, which is as good as, or even better than, previous studies in which postal screening procedures were applied [
24,
35,
36].
Although most older people may visit their GP regularly, primary care often fails in the identification of the health care needs of older people [
37]. Screening has the potential to identify older people at risk, followed by comprehensive assessment when needed [
7,
8]. Frailty instruments have to provide reliable and valid answers and have to be feasible [
15]. The psychometric properties of the TFI were slightly better than those of the GFI. However, the number of missing values was lower for GFI items than for TFI items, indicating a higher feasibility of the GFI. Based on these findings it is not yet possible to conclude whether the GFI or the TFI should be preferred for postal screening. The SPQ is less appropriate with regard to its psychometric quality and missing values.
The frailty index [
8,
11] is a simple measure that is based on self-reports. However, less is known about its feasibility for postal screening. Investigating the feasibility and validity of the frailty index as a postal screening instrument may be a point of interest for future research.
Future (longitudinal) research into the psychometric properties of the GFI and the TFI is urgently needed with regard to predictive validity and test-retest reliability of the GFI and the TFI. In addition, comparing the GFI and the TFI with other frailty-related constructs would lead to more insight into their construct validity.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
SFM and RD designed the study and collected the data. SFM analysed the data and drafted the manuscript with contributions of RD, EvR, LdW, WJAvdH and GIJMK. All authors read and approved the final manuscript.