Frailty in community-dwelling older people and nursing home residents: An adaptation and validation study

Abstract

Aim:

Psychometrically adapt and evaluate the Tilburg Frailty Indicator to assess frailty among older people living in Slovenia’s community and nursing home settings.

Design:

A cross-cultural adaptation and validation of instruments throughout the cross-sectional study.

Methods:

Older people living in the community and nursing homes throughout Slovenia were recruited between March and August 2021. Among 831 participants were 330 people living in nursing homes and 501 people living in the community, and all were older than 65 years.

Results:

All items were translated into the Slovene language, and a slight cultural adjustment was made to improve the clarity of the meaning of all items. The average scale validity index of the scale was rated as good, which indicates satisfactory content validity. Cronbach’s α was acceptable for the total items and subitems.

Conclusions:

The Slovenian questionnaire version demonstrated adequate internal consistency, reliability, and construct and criterion validity. The questionnaire is suitable for investigating frailty in nursing homes, community-dwelling, and other settings where older people live.

Impact:

The Slovenian questionnaire version can be used to measure and evaluate frailty among older adults. We have found that careful translation and adaptation processes have maintained the instrument’s strong reliability and validity for use in a new cultural context. The instrument can foster international collaboration to identify and manage frailty among older people in nursing homes and community-dwelling homes.

Reporting Method:

The Strengthening the Reporting of Observational Studies in Epidemiology checklist for reporting cross-sectional studies was used.

No Patient or Public Contribution:

No patient or public involvement in the design or conduct of the study. Head nurses from nursing homes and community nurses helped recruit older adults. Older adults only contributed to the data collection and were collected from nursing homes and community-dwelling.

INTRODUCTION

Frailty is associated with loss of biological reserves, failure of physiological mechanisms, vulnerability related to many negative consequences (Clegg, Young, Iliffe, Rikkert, & Rockwood, 2013) and ageing (Feng et al., 2017). It is not the same as multi-morbidity but is often related to chronic disease or disability and potentially to late-life dependency (Vetrano et al., 2019). A frail person has a higher risk of daily living limitations (Liu et al., 2019). Frailty is a public health priority worldwide (Hoogendijk et al., 2019). Defining frailty is challenging for several reasons, including the complex aetiology (Clegg et al., 2013); independent researchers’ work (Karunananthan, Wolfson, Bergman, Béland, & Hogan, 2009); and the complexity of ageing frailty and disability (Vetrano et al., 2019). The World Health Organization (WHO) and the International Association of Geriatrics and Gerontology (IAGG) are working toward developing an internationally acceptable definition of frailty (Berrut et al., 2013).

Frailty is associated with age, gender, economic status (Feng et al., 2017), and chronic disorders (James et al., 2018). The incidence rate of frailty varies across countries and is expected to increase based on the projected demographics (Ofori-Asenso et al., 2019). Based on a literature review (Collard, Boter, Schoevers, & Oude Voshaar, 2012), of 21 cohorts with 61,500 participants, 10.7% of community-dwelling older people (65+) are frail, and another 41.6% are prefrail. Prevalence of frailty in nursing homes ranged from 19% to 75.6%, and approximately 40% were prefrail (Kojima, 2015). From another literature review, prevalence rates for adults ≥ 50 years of age from 62 countries resulted in an overall estimate of 12% (11–13%) for physical frailty (O’Caoimh et al., 2021).

Despite the importance of frailty for understanding the effects of population ageing, global fragility prevalence remains unclear (O’Caoimh et al., 2021). Fortunately, frailty is a dynamic geriatric syndrome that can be mitigated with appropriate intervention strategies (Kidd et al., 2019).

The Tilburg Frailty Index (TFI) is one of the self-administered questionnaires (Gobbens, van Assen, Luijkx, & Schols, 2012; Gobbens, van Assen, Luijkx, Wijnen-Sponselee, & Schols, 2010) that has been used increasingly to assess frailty amongst older people since 2010 (Gobbens & Uchmanowicz, 2021). It contains fifteen physical, psychological and social items (Gobbens & Uchmanowicz, 2021).

The review found that 27 studies reported the psychometric properties of the TFI in terms of reliability, validity, or both. Twenty-seven focused on older people in shared accommodations (Gobbens & Uchmanowicz, 2021). In 2020, internal consistency, convergent and divergent validity were assessed as simultaneous validity of the TFI for older people in community-dwelling in Spain, Greece, Croatia, the Netherlands and the United Kingdom (Zhang et al., 2020). The questionnaire is valid and reliable for older people in community-dwelling (Gobbens & Uchmanowicz, 2021; Gobbens & van Assen, 2014; Gobbens et al., 2010) but not for older people in a nursing home.

This study aimed to adapt and evaluate, psychometrically, the Tilburg Frailty Indicator to assess frailty among older people living in both community and nursing home settings in Slovenia.

METHODOLOGY

Content, face validity, and internal consistency reliability were assessed along with exploratory factor analysis on the Tilburg Frailty Indicator in Slovenia (TFI-SI). For conducting psychometric testing, we follow the recondition of Polit and Beck (2021), and for reporting a cross-sectional study, we follow the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist.

Instrument

TFI is a 15-item questionnaire measuring frailty in older people. TFI consists of three domains: (1) Physical components (Ph) (eight questions); (2) Psychological components (Ps) (four questions); and (3) Social components (So) (three questions). The score range of physical components is from 0–8, psychological components are scored from 0–4, and social components are scored from 0–3. Eleven items had two possible answers (yes and no), and four items had three possible answers (‘yes’, ‘sometimes’ and ‘no’). Total scores ranged from 0 to 15 (Gobbens et al., 2010), with ≥ 5 points indicating frailty in an older person (Gobbens et al., 2010; Vetrano et al., 2019).

Validation regarding the experts’ agreement

The process of translating the questionnaire

The TFI questionnaire was translated from the first author into Slovene with the author’s permission. An independent bilingual translator and a nursing language expert conducted the back translation. After back translation, both questionnaire versions were reviewed, harmonised, and formulated the final version. Other authors (an expert in caring for older people) reviewed and discussed both versions and approved the final questionnaire. No problems were encountered in translating the questionnaire. To achieve semantic equivalence, we used back translation so that the meaning of each item in the target culture after translation would be the same as in the original (Polit & Beck, 2021). The items in the translated version are easy to understand, and the questionnaire can be completed in less than 15 minutes, which is comparable to other versions of questionnaires (Dent, Kowal, & Hoogendijk, 2016; Gobbens & Uchmanowicz, 2021; Gobbens & van Assen, 2014; Gobbens et al., 2012; Gobbens et al., 2010).

Content validity

To assess the validity of the content of the translated questionnaire, ten people with expertise in the care of older people reviewed the questionnaire and rated each element for its relevance using a four-point scale. Item content validity index (I-CVI), scale content validity index (S-CVI) (Shi, Mo, & Sun, 2012) and average scale validity index (S-CVI/Ave) were calculated. The scores were rated good when I-CVI was >.78 and S-CVI/Ave was >.90 (Polit & Beck, 2021). We calculated a modified kappa (κ*) statistic to reduce the influence of chance agreement. To evaluate the modified kappa, we followed the recommendations of Cicchetti and Sparrow (1981), and Polit, Beck, and Owen (2007). They divided the values into three groups, namely moderate (.40-.59), good (.60-.75) and excellent (>.75).

Face validity and cultural adaptation

Face validity and cultural adaptation were tested with the same convenience sample of 10 experienced nurses in older people’s nursing care. Participants were asked to suggest better wording if items were identified as unclear. If comments or corrections were given, the authors discussed this, and after reaching a consensus, the finished version of the questionnaire was confirmed (see SuppInfo).

Cross-sectional study

The internal consistency of the TFI-SI was assessed in a cross-sectional study using a convenience sample (Polit & Beck, 2021). The inclusion criteria for the study were older adults (aged 65 and over) living in nursing homes and community-dwelling. Participants were required to have the cognitive capability to assess their frailty using the TFI (cognitive ability was assessed so that the healthcare professional considered which participants could be offered to participate in the study. In doing so, they looked at the absence of known organic or psychiatric affecting cognitive ability). Exclusion criteria were younger adults than 65 years living in nursing homes and community-dwelling and cognitive impairment to answer the question.

Data collection

Data collection took place between March and August 2021 and took place in Slovenian nursing homes and community-dwelling after the participants gave written consent. Nurses working in community nursing and who visit older people living at home helped us distribute questionnaires to community-dwelling older adults.

Five researchers distributed and assisted older people in completing the questionnaires. All researchers received standardised data collection training for this study and were familiarised with the questionnaire, including the collection of demographic information and item responses. The survey was completed using a pencil and paper.

At each sampling location, older people were recruited with the help of the head nurse and community health nurses. After obtaining consent for the implementation, the interviewers introduced themselves to the participants in their homes and nursing homes and then explained the purpose of this study. Verbal informed consent was obtained before the respondents filled out the questionnaire. We asked the participants to fill out the questionnaire independently and allowed them to have the questions explained to them if necessary. For those who were visually impaired or illiterate, the researchers made it possible to complete the questionnaire by reading the questions to the participants. Participants could skip a question if they did not understand it or did not want to answer it. When data was being collected, participants could quit at any time.

The sample size was determined using the Cochran formula and estimated that the representative sample should be 384 (e=95%; z=5%). The researchers distributed 1,010 questionnaires, of which 888 were returned. Due to missing data (over 50 % of the questionnaire), 22 questionnaires were removed; therefore, we had 866 questionnaires included in the analysis (response rate: 87.9%).

Analysis of the data

Using IBM SPSS v28.0, we created a database and used it for descriptive and inferential analysis and checked for accuracy. Descriptive statistics (mean and standard deviation) were estimated with 95% confidence intervals. Internal consistency was assessed with Cronbach’s coefficient ɑ, omega coefficient and item-total correlations. The Cronbach’s coefficient ɑ was judged based on the Nunnally (1978) recommendations (adequate [>.7], excellent [>.9]). With the calculation of intraclass correlation coefficients, the test-retest reliability was estimated. As part of the test-retest procedure, the participants were asked to answer the TFI-SI twice, approximately three months apart. The intraclass correlation coefficients were elevated based on recommendations from Cicchetti and Sparrow (1981) (good [>.6], excellent [> .75]). The previous step helped us calculate the reproducibility of the Slovenian, and Pearson’s correlation coefficient enabled us to present the reproducibility (Vaz, Falkmer, Passmore, Parsons, & Andreou, 2013). Adjusted correlations of each item with their respective scales were also assessed, with values of .2<r<.3 considered acceptable (Mahieu, de Casterlé, Van Elssen, & Gastmans, 2013).

Exploratory factor analysis was performed using the Kaiser-Meyer-Olkin and Bartlett’s test for sampling adequacy. Factor analysis was used to determine the covariance of the scale items, which warranted factor analysis. In addition, factor analysis using Direct Oblimin rotation was used. Factor extraction was based on parallel analysis for Eigenvalues equal to or greater than one and a scree plot (Patil, Singh, Mishra, & Donavan, 2007).

Ethical considerations

We obtained the relevant Ethics Committee for approval (038/2018/2510–1/504). Participants were informed in writing of the purpose of the study and objectives, highlighting confidentiality, anonymity, and voluntary withdrawal from participation at any research stage. On request, results obtained from the study will be shared with participants. The authors follow the Declaration of Helsinki (World Medical Association, 2001) and the Oviedo Convention (Council of Europe, 2001).

RESULTS

Validation based on experts’ agreement

We encountered no problems with the translated items throughout the expert review, and no changes or deletions were necessary.

Content validity

The content validity of all items of the TFI-SI questionnaire is presented below (Table 1). All items had an I-CVI score of at least .60. Based on the kappa coefficient the items Ph1 (I-CVI = .80), Ph3 (I-CVI = .80), Ph4 (I-CVI = .90), Ph6 (I-CVI = .80), Ph7 (I-CVI = .90), Ph8 (I-CVI = 1.00), Ps1 (I-CVI = .80), Ps2 (I-CVI = .80), Ps4 (I-CVI = .80), and So1 (I-CVI = .80) were evaluated as excellent. Items Ph2 (I-CVI = .70), Ph5 (I-CVI = .70), Ps3 (I-CVI = .70) and So2 (I-CVI = .60) evaluated as good. The S-CVI/Ave was estimated at .793.

Table 1:

Content validity of the TFI-SI questionnaire

No.	Item	N	A	I-CVIi	Pcii	κ*iii	Evaluationiv
Ph1	Do you feel physically healthy?	10	8	.80	.00	.80	Excellent
Ph2	Have you lost a lot of weight?	10	7	.70	.00	.70	Good
Ph3	Do you experience problems due to difficulty in walking?	10	8	.80	.00	.80	Excellent
Ph4	Do you experience problems due to difficulty in maintaining your balance?	10	9	.90	.00	.90	Excellent
Ph5	Do you experience problems due to poor hearing?	10	7	.70	.00	.70	Good
Ph6	Do you experience problems due to poor vision?	10	8	.80	.00	.80	Excellent
Ph7	Do you experience problems due to a lack of strength in your hands?	10	9	.90	.00	.90	Excellent
Ph8	Do you experience problems due to physical tiredness?	10	10	1.00	.00	1.00	Excellent
Ps1	Problems with your memory?	10	8	.80	.00	.80	Excellent
Ps2	Felt down during the last month?	10	8	.80	.00	.80	Excellent
Ps3	Nervous or anxious during the last month?	10	7	.70	.00	.70	Good
Ps4	Are you able to cope with problems well?	10	8	.80	.00	.80	Excellent
So1	Do you live alone?	10	8	.80	.00	.80	Excellent
So2	Miss having people around you?	10	6	.60	.01	.60	Good
So3	Enough support from other people?	10	8	.80	.00	.80	Excellent
					vS-CVI/Ave: .793

ⁱitem content validity index/ number giving a rating of 3 or 4/number of experts.

ⁱⁱP_c (probability of a chance occurrence) = $\left[Pc=\left[\frac{N!}{A!}\right]\times \left(N-A\right)\times {.5}^N\right]$

ⁱⁱⁱk* = kappa designating agreement on relevance.

^ivEvaluation criteria for kappa: moderate = k of .40–.59; good = k of .60–.74; and excellent = k >.75.

^vS-CVI/Ave (average scale validity index) = mean of I-CVI.

A - No. of agreement; N - No. of experts; Ph - Physical domain; Ps - Psychological domain; So - Social domain

Face validity

Experts suggested clarifying the items’ meaning with minor cultural adaptations, and no items were removed.

Psychometric testing based on a cross-sectional study

Participants

We distributed 1010 questionnaires, and 888 were returned; therefore, there were 122 non-responders. The analysis included 866 questionnaires after 22 were removed owing to missing data. Among our overall sample of 866 older people, 35.5% (n = 307) were male, and a majority lived in the community (60.6%; n = 525), and 39.4% (n = 341) lived in nursing homes. The mean age of participants was 76.5 (SD = 9.2) (95%, CI = 75.1–77.1). The mean age of community-dwelling older people was 74.9 years (SD = 9.5) (95%, CI = 74.1–75.7) and 79.0 years (SD = 9.0) (95%, CI = 78.1–80.0) for older people living in a nursing home. Fifty-two-point three per cent (n = 453, 95%, CI=56–63%) of all participants had one chronic disease, including 57.9% (95%, CI = 53–63%) of community-dwelling older people and 47.2% (95%, CI = 55–66%) of those from nursing homes (Table 2). The estimated prevalence of frailty was 45% overall (n = 390; 95%, CI = 39.8–50.3); among older adults from nursing homes, the estimated prevalence was 59% (n = 193; 95%, CI = 53.3–63.6) and among community-dwelling older adults estimated prevalence was 36% (n = 183; 95%, CI = 32.5–40.9). The prevalence of frailty among women was 50.5% (n = 279; 95%, CI = 46.3–54.8) and among men 35% (n = 97; 95%, CI = 29.9–40.6).

Table 2.

Characteristics of the participants

Variables		Descriptive statistics
		Total (n = 866)	Nursing home (n = 341)	Community-dwelling (n = 525)
Gender %(n)		—	—	—
	Male	35.5(307)	30.5(104)	36.4(191)
	Female	64.5(559)	69.5(237)	63.6(334)
Age (Y; M±SD)		76.47±1.15	79.01 ± 9.04	74.86 ± 9.48
No. of CD %(n)		—	—	—
None		14.8(128)	14.4(49)	14.7(77)
One		52.3(453)	47.2(161)	57.9(304)
Two to three		31.5(273)	36.7(125)	26.7(140)
Four or more		1.4(12)	1.7(6)	0.7(4)

Internal consistency

The Cronbach’s coefficient α of the TFI-SI questionnaire was evaluated as adequate (.79), and the McDonald Omega coefficient was evaluated as adequate (.76). Table 3 presents Pearson’s correlations between item scores and the total score. All items showed significant item-total correlations and ranged between .129 and .702 (p<.001). The strongest correlations between components were from the Physical component (r_p = .878; p<.001), the Psychological component (r_p = .588, p<.001), and the lowest item-total correlation value was the Social component (r_p = .515, p<.001).

Table 3.

Pearson’s correlation item-total correlation of the TFI-SI questionnaire

Component of TFI-SI	Corrected - item total correlation	Total alpha if item is deleted
Physical component	.878	.795
Do you feel physically healthy?	.439	.783
Have you lost a lot of weight?	.212	.792
Do you experience problems due to difficulty in walking?	.673	.772
Do you experience problems due to difficulty in maintaining your balance?	.637	.773
Do you experience problems due to poor hearing?	.580	.776
Do you experience problems due to poor vision?	.541	.778
Do you experience problems due to a lack of strength in your hands?	.702	.771
Do you experience problems due to physical tiredness?	.593	.777
Psychological component	.588	.770
Problems with your memory?	.300	.789
Felt down during the last month?	.592	.775
Nervous or anxious during the last month?	.129	.795
Are you able to cope with problems well?	.252	.808
Social component	.515	.780
Do you live alone?	.585	.789
Miss having people around you?	.363	.786
Enough support from other people?	.453	.797
Cronbach’s Alpha	.79 (95%, CI= .772–.813)
Mean ± Std. Deviation	7.02 ± 3.15 (95%, CI=6.80–7.23)

Among community-dwelling older people, a corrected item-total correlation ranged from .54 to 1.00; if an item is deleted, it ranged from .48 to .81 (Table 4). Among older people living in nursing homes, item-total correlations ranged from .53 to 1.00; if an item is deleted from .38 to .76. Cronbach’s coefficient ɑ for community-dwelling older people was .77 (95%, CI = .74-.80), and for those in a nursing home was also .76 (95%, CI = .67-.76). Intraclass correlation coefficients for the total TFI score was .79 (95%, CI = .72 to .81, p<.001). The TFI-SI instrument has high within-subject reliability based on the intraclass correlation coefficients above .70.

Table 4.

Pearson’s correlation item-total correlation coefficients differences between older people living in nursing homes and community-dwelling of the TFI-SI questionnaire

	Community-dwelling		Nursing homes
Component of TFI-SI	Corrected-item total correlation	Total alpha if item is deleted	Corrected-item total correlation	Total alpha if is item deleted
Physical component	.875	.521	.865	.741
Psychological component	.585	.794	.582	.733
Social component	.538	.808	.533	.756
Total TFI-SI	1.000	.475	1.000	.384
Cronbach Alpha	.77 (95%, CI = .738–.803)		.76 (95%, CI = .723–.798)
Mean ± Std. Deviation	6.39±3.23 (95%, CI=6.12–6.68)		7.98±2.76 (95%, CI = 7.68–8.27)

Construct validity

Exploratory factor analysis of the TFI-SI questionnaire among older people suggested three factors (Table 5). The Kaiser-Meyer-Olkin (KMO index is adequate at .829) and Bartlett’s sphericity test (χ² = 334.81, df = 105, p<.001) showed acceptable values. Then, we assessed commonalities, i.e. the proportion of variance explained by the common factors. In general, we can claim that all three components have no value close to zero depending on the factors. Principal axis factoring showed three factors that explained 65.6% of the variance. The first factor explained 41.3% and included eight questions focusing on coping with everyday life, with factor loadings between .56 and .77. The second explained 16.3% and consisted of four questions that focused on psychological well-being and had factor loadings of .53 and .78. The last factor explained 8.0%, consists of three questions that focus on coping with problems and support and had factor loadings of .60 and .82 (Table 5).

Table 5.

Exploratory factor analysis of the TFI-SI questionnaire

Component	Factor loadings
	Factor 1	Factor 2	Factor 3
Do you feel physically healthy? (Q11)	.636	—	—
Have you lost a lot of weight recently without wishing to do so? (Q12)	.573	—	—
Do you experience problems in your daily life due to difficulty in walking? (Q13)	.714	—	—
Do you experience problems in your daily life due to difficulty maintaining your balance? (Q14)	.742	—	—
Do you experience problems in your daily life due to poor hearing? (Q15)	.770	—	—
Do you experience problems in your daily life due to poor vision? (Q16)	.708	—	—
Do you experience problems in your daily life due to a lack of strength in your hands? (Q17)	.672	—	—
Do you experience problems in your daily life due to physical tiredness? (Q18)	.560	—	—
Do you have problems with your memory? (Q19)	—	.535	—
Have you felt down during the last month? (Q20)	—	.734	—
Have you felt nervous or anxious during the last month? (Q21)	—	.781	—
Are you able to cope with problems well? (Q22)	—	.771	—
Do you live alone? (Q23)	—	—	.823
Do you sometimes miss having people around you? (Q24)	—	—	.595
Do you receive enough support from other people? (Q25)	—	—	.739
Total variance explained (%)	41.3	16.3	8.0
Cronbach’s α 0.90 (each factor)	.84	.60	.57

DISCUSSION

The aim of this article was the translation and cross-cultural adaptation of the TFI for the Slovenian population. This study evaluated the physical, psychological, and social components of a Slovene language version of the TFI questionnaire, enabling an analysis of frailty in older people living in nursing homes and community-dwelling. The translation process is a crucial aspect of test adaptation in general and assuring, in particular, semantic equivalence (Polit & Beck, 2021). The TFI questionnaire was translated into Slovenian and validated. To ensure semantic equivalence, we used independent translation and back-translation to ensure the equality of meaning of each item of the original text according to the translated items. The first author translated the questionnaire items to the target language; a second bilingual translator independently back-translated the questionnaire from the target language to the original language. The two versions of the questionnaire were compared for equivalence. This procedure continued until a research team agreed that the conceptual meaning of the two questionnaire versions was identical. The Slovenian translation of the TFI contained no unclear items. The questionnaire took less than 15 minutes to complete and was equivalent to reviewing other frailty questionnaires (Gobbens et al., 2012; Dent et al., 2016). This questionnaire has good face and content validity, internal consistency, and reliability. Following Polit and Beck’s (2021) recommendations, we tested content validity, which is not commonly used but recommended for the survey’s psychometric testing and cultural adaptations. Content validity indicates the adequacy of the TFI-SI questionnaire. The same applies to the content validity of each item. The content and face validity of TFI-SI were found to be adequate. We confirm the adequacy of the TFI-SI questionnaire psychometric validation procedure based on our reliability and validity results.

The internal consistency of the TFI-SI assessed among older people was adequate (α=.79). Gobbens and Uchmanowicz (2021) found that Cronbach’s alpha of the TFI, was between .66 to .80. Furthermore, other researchers reported Cronbach’s alpha of frailty physical, psychological, and social components, ranging between .57 to .79 (Mulasso, Roppolo, Gobbens, & Rabaglietti, 2016; Santiago et al., 2018), .37 to .63 (Gobbens et al., 2010; Uchmanowicz, Jankowska-Polańska, Uchmanowicz, Kowalczuk, & Gobbens, 2016), and .25 to .59 (Dong et al., 2017; Uchmanowicz et al., 2016). We found that Cronbach’s alpha for physical frailty was .88, psychological frailty was .59, and social frailty was .52, which is in line with or higher (e.g., for psychological frailty) than in previous studies. Our study complements existing research by a scale for frailty’s physical, social and psychological components among older people living in nursing homes and community-dwelling in Slovenia.

The corrected item-total correlation of the TFI-SI met the required criterion of .2<r<.3. The internal consistency of our psychological component increased slightly (.80) when eliminating one item (3rd) with a small item-total correlation (r<.13). The third component had the lowest correlations with the total score among older people in community-dwelling (its removal increased to α=.81) and for older people living in nursing homes (its removal increased to α=.76). The results suggest that the TFI-SI demonstrates reasonable internal consistency among older people.

We found that the average TFI score was 7.02±3.15 (in community-dwelling, 6.39±3.23; in nursing homes, 7.98±2.76). Compared with results from a study conducted in five European countries and including healthy older adults in community-dwelling, our results indicate higher mean scores in frailty (6.39±3.23) than found in Greece (5.80±3.09), Spain (4.64±2.88), the Netherlands (4.25±3.01), and the UK (4.47±2.01) and a lower score only according to Croatia (6.92±3.20) (Zhang et al., 2020). In Brazil, the mean score of the TFI for older adults aged 60 years and more was 4.40±3.0, and frailty prevalence was 44.2% (Santiago et al., 2018). In our study, the prevalence of frailty was 45% among older people >65 years old living in nursing homes and community-dwelling, which is comparable with the studies which involved older people aged 60 years or more residing in community-dwelling in Brazil (44.2%) (Santiago et al., 2018), but not with the results from China (17.2%) (Dong et al., 2017) and UK (14%) (Gale, Cooper, & Sayer, 2015). One meta-analysis (O’Caoimh et al., 2021) found that the prevalence in older people living in the community was 12% and 45%, and an overall estimate for pre-frailty was 46% (45%–48%). In our study, the frailty prevalence was 36% of older people living in community-dwelling and 59% of older people living in nursing homes.

TFI includes the physical, social and psychological components of frailty (Gobbens et al., 2010). Therefore, the questionnaire is crucial and applicable to studying frailty in nursing homes and community-dwelling older people. To assess frailty in older people accurately, it was crucial to undertake psychometric validation of TFI-SI questionnaires.

Strengths and limitations

One strength of the study is its inclusion of older adults from both community and nursing home settings. Gobbens and Uchmanowicz (2021) found in their review that only one study validated the questionnaire in older people, among which two included older people who do not live in the community-dwelling. One study included older people admitted to the hospital (Gobbens & Andreasen, 2020), and the other included older people living in assisted living facilities (Gobbens & van Assen, 2014). This study shows that a TFI-SI questionnaire helps assess frailty in older adults. The questionnaire items are easily understood and answered by older people in nursing homes or community-dwelling. Due to its simplicity, the questionnaire can assess the frailty of older people in various settings. It can help caregivers develop a quality approach to identifying and reducing frailty.

This study further supports the content validity and internal reliability of the TFI-SI; however, due to some limitations of the study, they should be interpreted with caution, as the study does not provide additional information about health status, pharmacotherapy, and other interventions for older adults. In addition, the study’s results may be specific to the gender of women. In the data analysis, we did not include the characteristics of those who produced incomplete questionnaires. Due to the higher number of distributed questionnaires and the number of researchers involved, we do not have the information about the reasons for not returning. We also recognise that three months may be a long time to re-test, as we did, because participants’ views may change. Another limitation is that we could not ensure the same participant for the re-test; however, we used the same inclusion criteria and survey location to match the two-time data sets. Finally, one of the limitations is also using the same sample for content and face validity. Notwithstanding certain limitations, we have shown the questionnaire is essential and suitable for investigating frailty in nursing homes and community-dwelling where older people live.

CONCLUSION

Although frailty is not an expected part of primary ageing, it is frequently overlooked, particularly in long-term care. However, studies monitoring fragility often indicate that the prevalence of this potentially treatable condition is high in older adults, adversely affecting the overall quality of life. Psychometrically validated tools are most important in assessing the general, more ageing adult population’s frailty. The study has demonstrated that the TFI-SI questionnaire is an uncomplicated instrument with good psychometric properties in the older population.

Continuous monitoring of frailty in clinical practice can enable healthcare professionals to identify and manage the condition early. Understanding how social, environmental, and biological factors influence frailty and ongoing monitoring of frailty can help prevent the negative health consequences of frailty and develop strategies for better health care in old age and early life.