Comparison of Diagnostic Accuracies for FRAIL, PRISMA-7, and GFI scale to identify frailty in Older People: A diagnostic meta-analysis

doi:10.21203/rs.3.rs-3084856/v1

Download PDF

Research Article

Comparison of Diagnostic Accuracies for FRAIL, PRISMA-7, and GFI scale to identify frailty in Older People: A diagnostic meta-analysis

https://doi.org/10.21203/rs.3.rs-3084856/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

Frailty has become one of the most serious global public health challenges. Whether the assessments tools are useful for frailty needs further evaluation.

Objectives

This study aims to evaluate the diagnostic accuracy of measurement tools for frailty in older people.

Design

This study was registered on PROSPERO (registration number: CRD: 42023415074.) Medline, Web of Science, Embase, China Knowledge Resource Integrated Database (CNKI), Wanfang databases, and VIP databases were searched from their inception to March 31, 2022.

Measurements

Two reviewers independently scored quality of the studies based on the Quality Assessment of Diagnostic Accuracy Studies 2(QUADAS-2) tool in Review manager 5.4.1 software, a summary receiver operating characteristics (SROC) curve was plotted and area under curve (AUC)was calculated to evaluate the accuracy of the frailty assessment tool.

Results

A total of 23 articles were included. The sample size ranges from 45 to 1037. The pooled sensitivity, specificity and AUC of FRAIL was 0.75 (95% CI: 0.56 - 0.88), 0.81 (95% CI: 0.71 - 0.88) and 0.88 (95% CI: 0.85 - 0.91), respectively; the pooled sensitivity, specificity and AUC of PRISMA-7 scale was 0.81 (95% CI: 0.71 - 0.88), 0.78 (95% CI: 0.69 - 0.85) and 0.86 (95% CI: 0.82 - 0.88) , respectively; the pooled sensitivity, specificity and AUC of GFI was 0.72 (95% CI: 0.62 - 0.80), 0.74 (95% CI: 0.67-0.81) and AUC of 0.79 (95% CI: 0.76 - 0.83) , respectively.

Conclusion

Our study suggests that, compared with the FP scale, the FRAIL and PRISMA-7 scales have a wide range of applications in the frail older population.

Frailty

Diagnostic value

Older people

Scale

Frailty is a clinical syndrome that the ability of maintaining the body's internal homeostasis reduces while facing stressful events, thereby increasing the risk of adverse outcomes (such as mortality and hospitalization) and seriously affecting the quality of life of older people(1, 2). Risk factors for frailty mainly include demographic and social factors, clinical factors, lifestyle factors as well as biological factors. Compared with the rest of the world, Asian populations are exposed to more risk factors for frailty(3, 4). The most common of these are clinical factors. For example, patients with chronic conditions such as diabetes(5, 6), chronic kidney disease(7), heart disease(8) and lung disease(9, 10) often feel weak and feeble. These illnesses can lead to restricted bodily functions and lack of energy, which can affect daily activities. In addition, lack of physical activity is also a major cause of debilitation. Lack of moderate physical activity and exercise can lead to weakness and reduced function, exacerbating debilitating symptoms(11). According to studies(12–14), the prevalence of frailty in the Chinese community is about 6% for people aged 60 years and above, 7.4% in the Japanese community and 20% in the South American community. The prevalence of frailty increases with increasing age, decreasing literacy, decreasing annual household income, improving marital status of widowed spouses and increasing burden of chronic diseases. As a result, frailty has become one of the most serious global public health challenges and its harms are quite widespread, with multiple adverse health outcomes that may occur(15, 16).

A frailty assessment tool should be a suitable tool for assessing the extent of frailty in older people, an appropriate frailty diagnostic tool for comprehensive screening of community populations becomes essential concern. The more we emphasize early detection and prevention, coupled with increased implementation of targeted interventions for older people, the more effective we will be in slowing the development of frailty in old age and improving quality of life, while saving healthcare resources. However, there is no uniform diagnostic criteria for frailty, although more than 60 assessment tools are available (17). In 2019, international clinical practice guidelines recommend several validated measures for frailty in older adults, including the Frailty Phenotype (FP), the Frailty Screening Inventory (FRAIL scale) and the Edmonton Frailty Scale (EFS). Frailty Profile (FP) is the most widely used screening tool in current research(18). Studies have been conducted on the diagnoses of various frailty screening tools in community-dwelling older adults and older inpatients(19–21), though there are some deficiencies in previous studies. As the older population grows and the health care system strives to provide comprehensive care for the older individuals, a thorough examination of the diagnostic performance of frailty assessment tools customized for this population is imperative. However, to our knowledge, meta-analyses of the accuracy of frailty assessment tools for older adults are lacking, possibly due to sample size and population adaptation.

This study summarizes evidence on the accuracy of different frailty assessment tools and diagnostic features in the older participants. Therefore, the purpose of our study was to systematically review current evidence on the accuracy of frailty assessment tools in identifying and quantifying frailty in older adults.

2.1 | Protocol and registration

The protocol for this this study has been registered in PROSPERO (CRD: 42023415074). The reporting adhered to the Preferred Reporting Items for Systematic reviews and Meta-analyses of DTA studies (PRISMA-DTA) guideline(22).

2.2 | Eligibility criteria

There were four inclusion criteria for the systematic review and meta-analyses. Firstly, participants must have included the older adults, with all people in the sample being older than 60 years. Secondly, diagnostic studies, cross-sectional observational studies, prospective studies were included in our research. Thirdly, the research tools include one or more assessments of frailty in older adults, excluding clinical measurement tools or a single index test. The reference tool is the Frailty Phenotype Scale. Finally, the outcome measure is the accuracy of frailty assessment tools, including sensitivity, specificity, and the area under the receiver operating characteristic curve.

2.3 | Information source

In this study, a joint electronic search was performed in four Chinese and three English databases, including MEDLINE, Embase, and Web of science, CNKI, CBM, VIP and Wanfang from inception to Mar,31,2022. NoteExpress was used for removal of duplicates and screening. We conducted manual forwards and backwards reference searching for all studies to identify additional relevant studies. Titles, abstracts, and full texts were screened independently by two members of the study team (ZYH and LL) in NoteExpress. In case where there was disagreement, any discrepancies were resolved through discussion until consensus was reached or with a third reviewer was consulted (ZCY), if necessary.

2.4 | Search

The following Medical Subject Heading (MeSH) were used in MEDLINE, Embase, and Web of science, CNKI, CBM, VIP and Wanfang, respectively: (“fraily OR “frail elderly” OR “aged,60 and over” OR “aged” OR “aging”) AND (“frailty screening” OR “frailty recognition” OR “frailty identification” OR “frailty assessment”). The following search strategy was used for Medline and modified to suit other databases(the detailed retrieval strategy pf other databases in Supplementary Documents) (Fig. 1).

2.5 | Study selection

NoteExpress was used for study selection due to its advantages of multi-language support, advanced search capabilities and automated PDF management. Duplicate literature was removed based on title, author, and publication date. Then, the eligible literature was screened independently by two members of the study team (ZYH and LL). Any disagreement in the process of screening were solved through discussion until consensus was reached with a third researcher.

2.6 | Data collection process

We developed an initial data extraction template, the number of true positives, false positives, false negatives and true negatives of each included studies were extracted by two reviewers independently using 2 × 2 contingency table. If necessary, we sought additional clarification or data from study authors, especially with respect to 2 × 2 data to allow calculation of diagnostic characteristics.

2.7 | Risk of bias and applicability

Included studies were assessed against Quality Assessment of Diagnostic Accuracy Studies(QUADAS-2)(23).Two review authors assessed the risk of bias and applicability concerns independently, using QUADAS-2, which includes four domains: a. selection of research subjects; b. diagnostic tools to be evaluated; c. gold standard for diagnosis; d. case flow and progress. According to the specific content of each study, the study will be assessed as "low risk", "high risk" or "uncertain risk" on each item. When all items in each dimension are low risk, it can be evaluated as low risk; when one item is high risk, it can be evaluated as high risk; when there is one item with uncertain risk, it can be evaluated as uncertain risk.

2.8 | Diagnostic accuracy measures

According to the total sample size, true positives, false positives, true negatives and false negatives in the research results, construct a 2×2 table of diagnostic results, and then calculate the sensitivity, specificity and corresponding 95% of each tool to be evaluated % confidence interval (confidence interval, CI).

2.9 | Synthesis of results

When the same frailty assessment tool to be evaluated has more than 3 studies evaluating its diagnostic performance, the results of the corresponding literature were combined.

2.10 | Meta-analysis

When there are three or more studies evaluating the diagnostic performance of the same frailty assessment tool, the results of the relevant literature are pooled. First, the threshold effect test was carried out. When the Spearman correlation coefficient test results for the true positive rate and false positive rate do not show statistical significance, it indicated that there was no threshold effect; when there was no threshold effect, the Q test was used to test the heterogeneity among studies. I² was used to estimate the extent of heterogeneity(24) ; a random-effects model is used to calculate the pooled sensitivity and specificity, If the I² for either sensitivity or specificity is greater than 50%, along with their 95% confidence intervals. The comprehensive Receiver operating characteristic curve (summary receiver operating characteristic curve, SROC) was fitted to evaluate the accuracy of the frailty assessment tool, using the area under the curve (area under curve, AUC) as the evaluation criterion. All statistical analyses were done in Stata (version 16.0) and Meta-Disc (version 1.4).

2.11 | Additional analyses

We did Deek's funnel plot and Egger's linear regression, p values of less than 0.05 signified evidence of publication bias among the included studies(25).

For sensitivity analysis, the leave-one-out analysis is used to explore the influence of individual studies on the stability of this results(26).

3.1 | Study selection

The search results returned 4568 results in total. Results of included studies are shown in the flow diagram in Fig. 2. Of the total records identified, 1894 duplicates were identified and removed, leaving 2674 records to be screened by title and abstract. After title and abstract screening, 1742 studies were excluded due to lack of relevance or full text. After reading the full text, 906 articles were excluded because they did not meet the inclusion criteria, and 26 articles were found to meet the inclusion criteria. Ultimately, 23 articles were included, consisting of 4 Chinese articles and 19 English articles.

3.2 | Study characteristics

Across the 23 included studies, the sample size ranged from 45 to 1037, with a total of 7423 participants. The mean age of participants, which stated, ranges from 61.0 to 85.2 years. Based on the frailty phenotype scale as a reference standard, diagnostic performance of 22 frailty assessment tools were reported. The characteristics of included studied are depicted in Table 1.

Table 1

The basic characteristics of included studied
Study	Country	Source of sample	Sample size	Age(SD)	Sex (male %)	Index test	Cut-off (point)	TP	FP	FN	TN	AUC
Zhu 2022(39)	China	nursing home	216	82.4(7.6)	74%	EFS	4.5	54	34	22	106	0.80
Zhu 2022	China	nursing home	216	82.4(7.6)	74%	GFI	4.5	46	22	27	118	0.80
Cardoso 2021(42)	Portugal	Community	128	71.09(7.85)	23.4%	SUNFTRAIL	2	59	16	17	36	0.67
Yang 2020(27)	China	Community	343	NR	NR	FRAIL	3	76	10	15	242	0.96
Yang 2020	China	Community	343	NR	NR	Tilburg	5.5	81	117	10	135	0.89
Hou 2019(28)	China	Inpatient	308	NR	56.5%	FRAIL	2	55	67	9	177	0.86
Braun 2017(29)	Germany	Outpatient	52	73.0(6.0)	37.0%	PRISMA-7	3	2	10	0	40	0.96
Braun 2017	Germany	Outpatient	52	73.0(6.0)	37.0%	FRAIL	3	1	4	1	46	0.90
Braun 2017	Germany	Outpatient	52	73.0(6.0)	37.0%	GFI	4	2	10	0	40	0.96
Thompson 2019(30)	Australia	Community	846	74.3(6.3)	45.2%	FRAIL	2	137	252	6	451	0.80
Xiang 2019(40)	China	nursing home	190	77.7(9.1)	43.2%	GFI	4	97	28	20	45	0.82
Rachel 2019(31)	Australia	Community	228	79.0(6.0)	54.8%	EFS	8	6	3	34	185	0.83
Rachel 2019	Australia	Community	228	79.0(6.0)	54.8%	FRAIL	3	12	11	28	117	0.82
Rachel 2019	Australia	Community	228	79.0(6.0)	54.8%	GFI	4	27	53	13	135	0.79
Rachel 2019	Australia	Community	228	79.0(6.0)	54.8%	Kihon Checklist	7	34	50	6	138	0.81
Rachel 2019	Australia	Community	228	79.0(6.0)	54.8%	PRISMA-7	3	31	48	9	140	0.83
Chen 2019(43)	Japan	Community	858	70.9(3.2)	48.3%	FRAIL	2	14	93	4	747	0.86
Aprahamian 2017(33)	Brazil	Outpatient	124	78.6(7.1)	17.0%	FRAIL	3	7	22	11	83	0.68
Hoogendijk 2012(36)	Netherlands	Primary institutions	102	78.6(7.1)	43.1%	GFI	4	7	25	5	65	0.64
Hoogendijk 2012	Netherlands	Primary institutions	102	78.6(7.1)	43.1%	PRISMA-7	3	10	15	2	75	0.85

Table 1

(continued)
Soysal 2019(37)	Turkey	Out patient	1003	74.2(8.5)	26.5%	MNAS-SF	11	223	50	90	640	0.906
Nixon 2019(38)	England	Out patient	90	69.0(13.0)	NR	CFS	5	15	9	4	62	0.90
Nixon 2019	England	Out patient	90	69.0(13.0)	NR	PRISMA-7	3	17	28	2	43	0.83
Nixon 2019	England	Out patient	90	69.0(13.0)	NR	SPPB	10	19	34	0	37	0.92
Mirabelli 2017(44)	America	Out patient	159	61(NR)	45.0%	FiND	1	20	49	2	37	0.73
Sukkriang 2020(34)	Thailand	Out patient	214	NR	50.0%	CFS	5	14	3	11	186	0.93
Sukkriang 2020	Thailand	Out patient	214	NR	50.0%	FRAIL	3	22	26	3	163	0.86
Sukkriang 2020	Thailand	Out patient	214	NR	50.0%	PRISMA-7	3	19	26	6	163	0.85
Sukkriang 2020	Thailand	Out patient	214	NR	50.0%	GFST	3	22	30	3	159	0.84
Bréchemier2019(45)	France	Out patient	418	82.8(5.5)	51.9%	CGA	4	164	66	49	139	0.80
Alqahtani 2019(35)	Saudi Arabia rabbin	Community	47	70.0(4.0)	66.0%	FRAIL	3	8	9	5	25	0.71
Kajsa2016(46)	Sweden	Inpatient	161	82.0(5.5)	45.0%	FRESH	2	95	9	22	35	0.86
Lilamand 2015(47)	France	Out patient	267	81.5(5.8)	67.0%	MNAS-SF	11	130	5	8	124	0.95
Bari 2014(48)	Italy	Community	1037	NR	NR	PQ	4	270	200	110	457	0.76
Cesari 2014(49)	France	Community	45	72.5(8.2)	62.2%	FiND	1	19	1	6	19	0.84
Saionara 2012(50)	Brazil	Community	124	69.5(2.9)	41.1%	SPPB	9	22	46	2	54	0.78
Dent 2012(51)	Australia	Inpatient	100	85.2(6.1)	25.0%	MNA	17	60	14	6	20	0.78
Dent 2012	Australia	Inpatient	100	85.2(6.1)	25.0%	MNA-SF	17	50	8	10	32	0.80

3.3 |Risk of bias and applicability

We assessed the 23 included articles against the QUADAS-2. The results of the summary of the overall methodological quality assessment using the QUADAS-2 tool for all 23 included studies are displayed in Fig. 2. The details of a quality assessment at the study level are presented in Fig. 3.

Across all 23 included studies, we found of bias based on concerns about the selection of participants to be high in 1(4.3%) and unclear in 21 (91.3%) studies; the main concern in this domain was high risk of bias due to case-control design (n = 1).

Among the 23 studies awaiting evaluation, 14 studies (60.9%) had a high risk of bias, and 2 studies (8.7%) had unclear bias risk, the 14 studies with a high risk of bias used Non-predefined threshold.

Risk of bias based on concerns related to participant flow and timing was unclear in four (17.4%) studies.

Concerns about the applicability of the evidence to participants were high in one study (4.3%) and unclear in 20 (87.0%) studies. Concerns about the applicability of the evidence to the index test were high in 14 (60.7%) and unclear in two (8.7%) studies. Concerns about the applicability of the evidence to the reference standard were unclear in 23 (100%) studies. Additional details about risk of bias and applicability assessment are presented in Fig. 2–3.

3.4 | Synthesis of results

3.4.1 | Threshold effect and Heterogeneity test

Nine of 23 included studies(27–35), consisting of 3020 patients, evaluated the FRAIL scale. The calculation of the Spearman correlation coefficient (ρ = 0.450, P > 0.05) suggested the absence of a threshold effect. The heterogeneity of the included studies was tested using sensitivity and specificity as effect sizes. The results showed that I² (sensitivity) was 93.98% (P < 0.01), and I² (specificity) was 96.64% (P < 0.01). Five studies(29, 31, 34, 36–38), consisting of 686 patients, evaluated the PRISMA-7 scale. The calculation of the Spearman correlation coefficient (ρ = 0.667, P > 0.05) suggested the absence of a threshold effect. The heterogeneity of the included studies was tested using sensitivity and specificity as effect sizes. The results showed that I² (sensitivity) was 0% (P = 0.69), and I² (specificity) was 82.77% (P < 0.01). Five studies(29, 31, 36, 39, 40), consisting of 650, patients evaluated the GFI scale. The calculation of the Spearman correlation coefficient (ρ = 0.200, P > 0.05) suggested the absence of a threshold effect. The heterogeneity of the included studies was tested using sensitivity and specificity as effect sizes. The results showed that I² (sensitivity) was 67.40% (P = 0.06), and I² (specificity) was 73.62% (P = 0.16).

3.4.2 | Pooled effect size and SROC

A random-effects model was used to combine the effect sizes, since the significant heterogeneity among the studies. For the FRAIL scale, the combined sensitivity was 0.75, with a 95% confidence interval of 0.56–0.88, and the combined specificity was 0.85, with a 95% confidence interval of 0.77–0.91. For the PRISMA-7 scale, the combined sensitivity was 0.81, with a 95% confidence interval of 0.71–0.88, and the combined specificity was 0.78, with a 95% confidence interval of 0.69–0.85. For the GFI scale, the combined sensitivity was 0.72, with a 95% confidence interval of 0.62–0.80, and the combined specificity was 0.74, with a 95% confidence interval of 0.67–0.81. All forest plots are showed in Fig. 4.

The combined diagnostic odds ratio, positive likelihood ratio, negative likelihood ratio, and diagnostic score for FRAIL scale were 17.18(95%CI,7.30-40.42), 5.08(95%CI,3.30–7.81), 0.30(95%CI,0.16–0.55), 2.84(95%CI,1.00-3.70), respectively. The pooled diagnostic odds ratio, positive likelihood ratio, negative likelihood ratio, and diagnostic score for PRISMA-7 scale were 14.70(95%CI,8.24–26.24),3.66(95%CI,2.63–5.09), 0.25(95%CI,0.16–0.38), 2.69(95%CI,2.11–3.27), respectively. The combined diagnostic odds ratio, positive likelihood ratio, negative likelihood ratio, and diagnostic score for GFI scale were 7.37(95%CI,5.08–10.70), 2.80(95%CI,2.23–3.51), 0.38(95%CI,0.29-0.0.50), 2.00(95%CI,1.63–2.37), respectively. All forest plots are showed in Fig. 5 and Fig.S1-2.

Figures S3 present the summary receiver operating characteristic (SROC) curves for the diagnosis of frailty using the FRAIL scale, with a combined area under the curve (AUC) of 0.88 and a 95% confidence interval (CI) of 0.85–0.91. Figures S5(b) show the SROC curves for the diagnosis of frailty using the PRISMA-7 scale, with a combined AUC of 0.86 and a 95% CI of 0.82–0.88. Figure S5(c) displays the SROC curve for the diagnosis of frailty using the GFI scale, with a combined AUC of 0.79 and a 95% CI of 0.76–0.83. The pooled sensitivity, specificity, AUC and DOR were showed in Table 2.

Table 2

Pooled sensitivity, specificity, and AUC for FRAIL, PRISMA-7 and GFI
Tools	No. of studies	Sensitivity(95%CI)	Specificity (95%CI)	Area under SROC^a curve	DOR^b (95%CI)
FRAIL scale	9	0.75 (0.56–0.88)	0.85 (0.77–0.91)	0.88 (0.85–0.91)	17.18 (7.30-40.42)
PRISMA-7 scale	5	0.81 (0.71–0.88)	0.78 (0.69–0.85)	0.86 (0.82–0.88)	14.70 (8.24–26.24)
GFI scale	5	0.72 (0.62–0.80)	0.74 (0.67–0.81)	0.79 (0.76–0.83)	7.37 (5.08–10.70)
a. SROC = Summary receiver operating characteristic; b. DOR = diagnostic odds ratio.

3.5 | Additional analyses

3.5.1 | Publication bias

The Egger's linear regression test and the Deeks’ funnel plot(Figure S4) for the FRAIL scale yielded a P-value of 0.15, suggesting that there was no publication bias in included studied. However, no publication bias tests were conducted for the PRISMA-7 and GFI scales due to the small number of studies included.

3.5.2 | Sensitivity analysis

The combined sensitivity, specificity, and AUC with 95% CIs were calculated after each excluded study, and the results are shown in table S1-3. The results indicated that the combined effect sizes did not change significantly after each study was excluded, suggesting the stability of the meta-analysis results.

4.1 | Summary

This study, a total of 25 articles were included, systematically evaluated 18 frailty assessment tools using the Fried phenotype (FP) as a reference standard and conducted a meta-analysis on three out of 18 tools (namely FRAIL, PRISMA-7, and GFI) to compare their diagnostic accuracy. These tools exhibit a wide range of features, varying from concise and quick screening tools to more intricate and time-consuming comprehensive assessments. The simple frailty assessment methods mentioned in this study include hand grip strength, gait speed testing, timed up and go, and polypharmacy, etc. For instance, in Ma's study(41), it was found that hand grip strength showed favorable sensitivity and specificity among hospitalized older individuals, although the precision of the findings was limited due to the small sample size.. Nixon et al(38) found that hand grip strength showed good sensitivity but poor specificity among hospitalized older individuals, but their study also suffered from a small sample size. This highlights the need for further research to corroborate these findings and enhance the generalizability of the results. Two studies used a diagnostic threshold of 0.8 m/s for gait speed testing. Nixon's study demonstrating favorable sensitivity and specificity, whereas Rachel's study(31) exhibited only modest level of sensitivity (0.70) and specificity (0.77). On the one hand, the limited sample size in Nixon's study may have contributed to lower reliability in the estimation of results. On the other hand, differences in the populations between the two studies may have also impacted the findings. Gait speed testing may not be an effective tool for assessing frailty in community-dwelling older individuals and may require a more comprehensive assessment approach. Timed-Up and Go showed good sensitivity but poor specificity in community-dwelling older individuals, whereas the opposite was observed among outpatient older individuals. Since both studies used 10 seconds as the diagnostic threshold for Timed-Up and Go. Consequently, additional research is required to determine the optimal threshold. In contrast to other simple tools, polypharmacy does not require examination of participants, but only requires evaluation based on their medication utilization. Both related studies used five medications as the threshold for diagnosing frailty. Despite the method's simplicity, its sensitivity and specificity are low. The results of this study also imply that the application of simple assessment tools has its limitations and may require the combination of additional assessment tools for comprehensive evaluation, which mirrors the multifarious aspects of frailty to some extent.

FP was chosen as the reference standard in this study mainly because it is based on the frailty cycle model and has been validated by multiple studies(27–35), which ensures the consistency and uniformity of frailty diagnosis standards in different studies.

Meta-analysis showed that the combined sensitivity of FRAIL was 0.75, specificity was 0.85, DOR was 17.18, and AUC was 0.88, respectively, which indicates that FRAIL has a good diagnostic value for frailty. However, there was substantial heterogeneity (I² > 75%) in terms of sensitivity and specificity. Diagnostic studies tend to show greater heterogeneity compared to other types of clinical studies. Our study showed that the heterogeneity may have also been caused by the study population, sample size, study type and the quality of study. These factors may lead to differences in study outcomes, resulting in heterogeneity in the results of FRAIL's efficacy to assessment, specifically. For instance, there may be differences in the population participating in the study in terms of age, gender, geography and culture, and these differences may lead to inconsistencies in assessing the effects for frailty. Additionally, variations in sample sizes among different studies may result in differences in the stability of evaluation results. Furthermore, differences in study design and quality may also contribute to heterogeneity in evaluation results. Differences in sample selection, diagnostic thresholds of assessment criteria among studies may lead to heterogeneity in results.

4.2 | Limitations

Our research focuses on including more self-reported tools to assess geriatric frailty, evaluating their feasibility and accuracy. What’s more, we place emphasis on expanding the scope of research so as to include more specific populations and environments, such as cancer and surgical patients, long-term care, emergency departments, etc., in order to more comprehensively evaluate the accuracy and practicality of geriatric frailty assessment tools.

We had to acknowledge that several limitations constrain the generalizability and robustness of the results. First of all, although a comprehensive search strategy was used in this study, the number of studies on some tools was limited, making it difficult to provide sufficient evidence. Furthermore, it is noteworthy that most of the relevant studies came from European and American countries, while there is a dearth of evaluation studies conducted in developing countries. This geographical difference may limit the transferability and generalizability of the findings to regions with different socio-economic, cultural, and healthcare contexts. Therefore, it is urgent to conduct more relevant study in developing countries to promote the development of effective interventions on frailty. This not only benefits the individuals affected by frailty in developing countries but also contributes to the global understanding of frailty, helping to improve care and support for persons 65 years and older around the world. Ultimately, prioritizing research on frailty in developing countries is an urgent and necessary step towards achieving better health outcomes and promoting healthy aging globally.

4.3 | Conclusions

In conclusion, our meta-analysis suggests that the FRAIL scale exhibits the highest diagnostic accuracy, with reasonable sensitivity and good specificity, followed by the PRISMA-7 scale, which demonstrates good sensitivity and reasonable specificity. The GFI scale shows the lowest diagnostic accuracy. Limited research evidence exists for other assessment tools, emphasizing the need for further research. These findings can inform the selection of frailty assessment tools and highlight areas for future investigation.

Ethical Standards This study did not include any animal or human experiments.

Experimental subjects/animals None

Conflict of interest The authors declare that no competing interest.

Consent for publication We give the publisher license of the copyright which provides the publisher with the exclusive right to publish and sell the research findings in all languages, in whole or in part. All authors guarantee that the research findings have not been previously published.

Availability of data and materials The datasets and materials analyzed during the current study are available from the corresponding author upon reasonable request. The authors are committed to making the data and materials accessible to other researchers to facilitate replication and extension of the study results.

Funding This study was supported by National Clinical Research Center for Geriatrics, West China Hospital, Sichuan University (Z20192005), and 1.3.5 project for disciplines of excellence, West China Hospital, Sichuan University.

Authors' contributions Study concept and design LL and KDY. Acquisition of data: LL and ZYH. Analysis and interpretation of data: LL, ZCY, and ZYH. Preparation of manuscript: LL and DC. Revision of manuscript: KDY and ZCY.

Acknowledgements This research was supported by grants from National Clinical Research Center for Geriatrics, West China Hospital, Sichuan University.

Hoogendijk EO, van der Horst HE, Deeg DJH, Frijters DHM, Prins BAH, Jansen APD, Nijpels G, van Hout HPJ (2013) The identification of frail older adults in primary care: comparing the accuracy of five simple instruments. Age Ageing 42:262–265.
Morley JE, Vellas B, van Kan GA, et al. (2013) Frailty Consensus: A Call to Action. J Am Med Dir Assoc 14:392–397.
Hoogendijk EO, Afilalo J, Ensrud KE, Kowal P, Onder G, Fried LP (2019) Frailty: implications for clinical practice and public health. Lancet 394:1365–1375.
Qin Y, Hao X, Lv M, Zhao X, Wu S, Li K (2023) A global perspective on risk factors for frailty in community-dwelling older adults: A systematic review and meta-analysis. Arch Gerontol Geriatr 105:104844.
MacKenzie HT, Tugwell B, Rockwood K, Theou O (2020) Frailty and Diabetes in Older Hospitalized Adults: The Case for Routine Frailty Assessment. Can J Diabetes 44:241-+.
Aguayo GA, Hulman A, Vaillant MT, Donneau A-F, Schritz A, Stranges S, Malisoux L, Huiart L, Guillaume M, Sabia S, Witte DR (2019) Prospective Association Among Diabetes Diagnosis, HbA(1c), Glycemia, and Frailty Trajectories in an Elderly Population. Diabetes Care 42:1903–1911.
Wilhelm-Leen ER, Hall YN, Tamura MK, Chertow GM (2009) Frailty and Chronic Kidney Disease: The Third National Health and Nutrition Evaluation Survey. Am J Med 122:664-U86.
Akasaki Y, Tabira T, Maruta M, Makizako H, Miyata M, Han G, Ikeda Y, Nakamura A, Shimokihara S, Hidaka Y, Kamasaki T, Kubozono T, Ohishi M (2022) Social Frailty and Meaningful Activities among Community-Dwelling Older Adults with Heart Disease. Int J Environ Res Public Health 19:15167.
Montgomery E, Macdonald PS, Newton PJ, Jha SR, Malouf M (2020) Frailty in lung transplantation: a systematic review. Expert Rev Respir Med 14:219–227.
Lahousse L, Ziere G, Verlinden VJA, Zillikens MC, Uitterlinden AG, Rivadeneira F, Tiemeier H, Joos GF, Hofman A, Ikram MA, Franco OH, Brusselle GG, Stricker BH (2016) Risk of Frailty in Elderly With COPD: A Population-Based Study. J Gerontol Ser -Biol Sci Med Sci 71:689–695.
Tolley APL, Ramsey KA, Rojer AGM, Reijnierse EM, Maier AB (2021) Objectively measured physical activity is associated with frailty in community-dwelling older adults: A systematic review. J Clin Epidemiol 137:218–230.
Xu W, Li Y-X, Wu C (2019) Incidence of frailty among community-dwelling older adults: a nationally representative profile in China. Bmc Geriatr 19:378.
Coelho-Junior HJ, Marzetti E, Picca A, Calvani R, Cesari M, Uchida MC (2020) Prevalence of Prefrailty and Frailty in South America: A Systematic Review of Observational Studies. J Frailty Aging 9:197–213.
Ofori-Asenso R, Chin KL, Mazidi M, Zomer E, Ilomaki J, Zullo AR, Gasevic D, Ademi Z, Korhonen MJ, LoGiudice D, Bell JS, Liew D (2019) Global Incidence of Frailty and Prefrailty Among Community-Dwelling Older Adults A Systematic Review and Meta-analysis. Jama Netw Open 2:e198398.
Chittrakul J, Siviroj P, Sungkarat S, Sapbamrer R (2020) Physical Frailty and Fall Risk in Community-Dwelling Older Adults: A Cross-Sectional Study. J Aging Res 2020:1–8.
Liu HX, Ding G, Yu WJ, Liu TF, Yan AY, Chen HY, Zhang AH (2019) Association between frailty and incident risk of disability in community-dwelling elder people: evidence from a meta-analysis. Public Health 175:90–100.
Buta BJ, Walston JD, Godino JG, Park M, Kalyani RR, Xue Q-L, Bandeen-Roche K, Varadhan R (2016) Frailty assessment instruments: Systematic characterization of the uses and contexts of highly-cited instruments. Ageing Res Rev 26:53–61.
Dent E, Morley JE, Cruz-Jentoft AJ, et al. (2019) Physical Frailty: ICFSR International Clinical Practice Guidelines for Identification and Management. J Nutr Health Aging 23:771–787.
Clegg A, Rogers L, Young J (2015) Diagnostic test accuracy of simple instruments for identifying frailty in community-dwelling older people: a systematic review. Age Ageing 44:148–152.
Lu J-L, Ding L-Y, Xu Q, Zhu S, Xu X-Y, Hua H-X, Chen L, Xu H (2021) Screening Accuracy of SARC-F for Sarcopenia in the Elderly: A Diagnostic Meta-Analysis. J Nutr Health Aging 25:172–182.
Ambagtsheer RC, Thompson MQ, Archibald MM, Casey MG, Schultz TJ (2020) Diagnostic test accuracy of self‐reported screening instruments in identifying frailty in community‐dwelling older people: A systematic review. Geriatr Gerontol Int 20:14–24.
McInnes MDF, Moher D, Thombs BD, et al. (2018) Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies: The PRISMA-DTA Statement. JAMA 319:388–396.
Whiting PF, Rutjes AWS, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, Leeflang MMG, Sterne JAC, Bossuyt PMM (2011) QUADAS-2: A Revised Tool for the Quality Assessment of Diagnostic Accuracy Studies. Ann Intern Med 155:529-U104.
Higgins JPT, Thompson SG, Deeks JJ, Altman DG (2003) Measuring inconsistency in meta-analyses. Br Med J 327:557–560.
Irwig L, Macaskill P, Berry G, Glasziou P (1998) Bias in meta-analysis detected by a simple, graphical test - Graphical test is itself biased. Br Med J 316:470–470.
Thompson SG, Sharp SJ (1999) Explaining heterogeneity in meta-analysis: A comparison of methods. Stat Med 18:2693–2708.
Yang Y, Liao L, Deng Q (2020) Comparison of FRAIL Scale and Tilburg Frailty Indicator Scale in assessment of frailty status of elderly patients with diabetes in the community. J Clin Nurs Pract 6:1–4.
Hou P, Xue H, Li Y, Mao X, Sun K, 薛玲玲, 吴琳凤, 刘永兵 (2019) Performance of the FRAIL Scale in Screening Frailty among Elderly Patients with Coronary Heart Disease. Chin Gen Pract 22:1052–1056.
Braun T, Grueneberg C, Thiel C (2018) German translation, cross-cultural adaptation and diagnostic test accuracy of three frailty screening tools. Z Gerontol Geriatr 51:282–292.
Thompson MQ, Theou O, Tucker GR, Adams RJ, Visvanathan R (2020) FRAIL scale: Predictive validity and diagnostic test accuracy. Australas J Ageing 39:E529–E536.
Ambagtsheer RC, Visvanathan R, Dent E, Yu S, Schultz TJ, Beilby J (2020) Commonly Used Screening Instruments to Identify Frailty Among Community-Dwelling Older People in a General Practice (Primary Care) Setting: A Study of Diagnostic Test Accuracy. J Gerontol Ser -Biol Sci Med Sci 75:1134–1142.
Chen S, Chen T, Kishimoto H, Susaki Y, Kumagai S (2020) Development of a Fried Frailty Phenotype Questionnaire for Use in Screening Community-Dwelling Older Adults. J Am Med Dir Assoc 21:272-+.
Aprahamian I, de Castro Cezar NO, Izbicki R, Lin SM, Vianna Paulo DL, Fattori A, Biella MM, Jacob Filho W, Yassuda MS (2017) Screening for Frailty With the FRAIL Scale: A Comparison With the Phenotype Criteria. J Am Med Dir Assoc 18:592–596.
Sukkriang N, Punsawad C (2020) Comparison of geriatric assessment tools for frailty among community elderly. Heliyon 6:e04797.
Alqahtani BA, Nasser TA (2019) Assessment of frailty in Saudi community-dwelling older adults: validation of measurements. Ann Saudi Med 39:197–204.
Hoogendijk EO, van der Horst HE, Deeg DJH, Frijters DHM, Prins BAH, Jansen APD, Nijpels G, van Hout HPJ (2013) The identification of frail older adults in primary care: comparing the accuracy of five simple instruments. Age Ageing 42:262–265.
Soysal P, Veronese N, Arik F, Kalan U, Smith L, Isik AT (2019) Mini Nutritional Assessment Scale-Short Form can be useful for frailty screening in older adults. Clin Interv Aging 14:693–699.
Nixon AC, Bampouras TM, Pendleton N, Mitra S, Dhaygude AP (2019) Diagnostic Accuracy of Frailty Screening Methods in Advanced Chronic Kidney Disease. Nephron 141:147–155.
Zhu L, Xu H, Dong A, Xu H, Li J, Li P (2022) Comparison of the Edmonton Frailty Scale and the Groningen Frailty Indicator in the assessment of frailty among elderly in nursing homes. Chin Nurs Manag 22:83–87.
Xiang W, Cheng Y, Li Z, Han J, Li K (2020) Cross-cultural adaptation and validation of the Groningen Frailty Indicator in Chinese nursing home residents. Aging Clin Exp Res 32:1035–1042.
Ma Y, Wang Y, Sun C (2019) Application of hand grip strength as a screening tool for frailty in patients with liver cirrhosis.
Cardoso AF, Bobrowicz-Campos E, Teixeira-Santos L, Cardoso D, Couto F, Apostolo J (2021) Validation and Screening Capacity of the European Portuguese Version of the SUNFRAIL Tool for Community-Dwelling Older Adults. Int J Environ Res Public Health 18:1394.
Chen S, Chen T, Kishimoto H, Susaki Y, Kumagai S (2020) Development of a Fried Frailty Phenotype Questionnaire for Use in Screening Community-Dwelling Older Adults. J Am Med Dir Assoc 21:272-276.e1.
Mirabelli LG, Cosker RM, Kraiss LW, Griffin CL, Smith BK, Sarfati MR, Al-Dulaimi R, Brooke BS (2018) Rapid Methods for Routine Frailty Assessment during Vascular Surgery Clinic Visits. Ann Vasc Surg 46:134–141.
Brechemier D, Sourdet S, Girard P, Steinmeyer Z, Mourey L, Gerard S, Balardy L (2019) Use of comprehensive geriatric assessment (CGA) to define frailty in geriatric oncology: Searching for the best threshold. Cross-sectional study of 418 old patients with cancer evaluated in the Geriatric Frailty Clinic (GFC) of Toulouse (France). J Geriatr Oncol 10:944–950.
Kajsa E, Katarina W, Sten L, Synneve I-D (2016) Screening for frailty among older emergency department visitors: Validation of the new FRESH-screening instrument. BMC Emerg Med 16:27.
Lilamand M, Kelaiditi E, Cesari M, Raynaud-Simon A, Ghisolfi A, Guyonnet S, Vellas B, Van Kan GA (2015) Validation of the Mini Nutritional Assessment-Short Form in a population of frail elders without disability. Analysis of the Toulouse Frailty Platform population in 2013. J Nutr Health Aging 19:570–574.
Di Bari M, Profili F, Bandinelli S, Salvioni A, Mossello E, Corridori C, Razzanelli M, Di Fiandra T, Francesconi P (2014) Screening for Frailty in Older Adults Using a Postal Questionnaire: Rationale, Methods, and Instruments Validation of the INTER-FRAIL Study. J Am Geriatr Soc 62:1933–1937.
Cesari M, Demougeot L, Boccalon H, Guyonnet S, Van Kan GA, Vellas B, Andrieu S (2014) A Self-Reported Screening Tool for Detecting Community-Dwelling Older Persons with Frailty Syndrome in the Absence of Mobility Disability: The FiND Questionnaire. Plos One 9:e101745.
Da Câmara SMA, Alvarado BE, Guralnik JM, Guerra RO, Maciel ÁCC (2013) Using the Short Physical Performance Battery to screen for frailty in young-old adults with distinct socioeconomic conditions: SPPB to identify frail young-old adults. Geriatr Gerontol Int 13:421–428.
Dent E, Visvanathan R, Piantadosi C, Chapman I (2012) Use of the Mini Nutritional Assessment to Detect Frailty in Hospitalised Older People. J Nutr Health Aging 16:764–767.

SupportingInformationDTAoffrailassessmenttools.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Comparison of Diagnostic Accuracies for FRAIL, PRISMA-7, and GFI scale to identify frailty in Older People: A diagnostic meta-analysis

Status:

Version 1

Abstract

Figures

1 Introduction

2 Methods

2.1 | Protocol and registration

2.2 | Eligibility criteria

2.3 | Information source

2.4 | Search

2.5 | Study selection

2.6 | Data collection process

2.7 | Risk of bias and applicability

2.8 | Diagnostic accuracy measures

2.9 | Synthesis of results

2.10 | Meta-analysis

2.11 | Additional analyses

3 Results

3.1 | Study selection

3.2 | Study characteristics

3.3 |Risk of bias and applicability

3.4 | Synthesis of results

3.4.1 | Threshold effect and Heterogeneity test

3.4.2 | Pooled effect size and SROC

3.5 | Additional analyses

3.5.1 | Publication bias

3.5.2 | Sensitivity analysis

4 Discussion

4.1 | Summary

4.2 | Limitations

4.3 | Conclusions

Declarations

References

Supplementary Files

Status:

Version 1