Cross-cultural translation, validation, and responsiveness of the Pittsburgh Fatigability Scale - Italian version (PFS-I) in a mixed-sample of older adults

Abstract

Introduction

Older adults experience fatigue which impacts health-related quality of life. The Pittsburgh Fatigability Scale (PFS) was specifically designed to assess perceived physical and mental fatigability in older adults. The aim of this study was to translate the PFS into Italian (PFS-I) and to investigate its psychometric properties.

Methods

The PFS-I was translated in accordance with international standards. The following properties were evaluated: structural validity, internal consistency, hypotheses testing for construct validity, test-retest reliability using intraclass correlation coefficients (ICC), and minimal clinically important difference (MCID) for responsiveness.

Results

The study included 87 older adults with cardiovascular/respiratory diseases (CVRD), 46 with Parkinson disease (PD), and 67 healthy controls (HC). The PFS-I Physical and Mental subscales scores were significantly different in these populations, with HC reporting the lowest fatigability. Exploratory and confirmatory factor analysis showed a two-factor structure for both subscales, with good internal consistency (Cronbach’s alpha: 0.86 and 0.89, respectively). The PFS-I Physical subscale showed weak goodness-of-fit of the confirmatory factor analysis models, whereas the PFS-I Mental subscale was deemed acceptable. Construct validity of the PFS-I was excellent with 75% (18 out of 24) of hypotheses accepted. Test-retest reliability was analyzed in a subset of 23 patients with CVRD and showed excellent results for both the PFS-I Physical and Mental subscales (ICC = 0.93 and 0.92, respectively). MCID ranged between 6 and 7 points for the Physical and 7–9 points for the Mental subscale.

Conclusions

The PFS-I is a valid, reliable, and responsive instrument to assess perceived fatigability for healthy older adults as well as those with CRVD and PD.

Background

Fatigue is a perception, characterized by feelings of tiredness and weakness, in which both physical and cognitive functions are impaired due to interactions between performance and perceived fatigability [1]. Unlike fatigue, which is related to an instantaneous state of lack of energy, fatigability is a trait that contextualizes an individual’s fatigue relative to level of activity [2]. Associating fatigue to specific activities - anchored to intensity and duration - enhances sensitivity and facilitates comparisons among individuals and between different cohorts, and allows for the assessment of changes over time and treatment effects [3].

Perceived fatigability is highly prevalent in older adults and strongly associated with age, mortality, and risk factors including greater body mass index, worse physical and cognitive function, and lower physical activity [3–5]. Fatigability can be assessed by means of performance-based tests or self-report tools [6]. Performance fatigability is measured by observing a decrement in performance during a standardized physical task or activity [7, 8]. On the other hand, perceived fatigability can be measured either by asking the actual self-reported fatigue immediately after performing a physical task of a standardized intensity and duration, or by querying one’s expected fatigue if they were to perform an activity or tasks [9]. The latter is especially suitable for populations in which performing activities is too challenging or it is not practical. Several self-administered instruments to measure perceived fatigue are available, including the Multidimensional Fatigue Inventory, the Fatigue Severity Scale, the Borg scale, and the Visual Analogue Scale for fatigue [10–12]. However, a major drawback of these tools is that they do not account for the respondent’s activity level or physical functioning, which can result in self-pacing bias. For example, sedentary individuals might report lower fatigability levels compared to more active individuals, making it difficult to compare and interpret the results. To overcome these issues, the Pittsburgh Fatigability Scale (PFS) was specifically developed for the assessment of both perceived physical and mental fatigability in older adults [13, 14]. The PFS consists of 10 items that anchors fatigue to activities of varying intensity and duration. The PFS has been validated in several languages [15–19] and clinical populations, such as persons with multiple sclerosis, fibromyalgia and cancer survivors [19–21]. Perceived physical fatigability measured by the PFS has been demonstrated to be a strong patient-reported prognostic indicator of risk of death in older adults [4]. The PFS translations showed strong internal consistency (Cronbach’s alpha of the identified subscales between 0.80 and 0.93), moderate-to-good validity (construct and convergent validity were demonstrated in the English version and in the Chinese, Dutch, Korean, and Spanish translations) and good test-retest reliability (ICC between 0.78 and 0.86 in the English and Dutch versions), indicating that it can be used as an effective clinical tool [22]. To date, the PFS has not yet been translated into Italian.

Although fatigue has been frequently reported as a debilitating symptom that affects older adults with Parkinson’s disease (PD) [23] or chronic cardiovascular/respiratory diseases (CVRD) [24], the impact of fatigability is understudied in these populations [25]. Moreover, evaluating fatigability can offer important insights into the functional status of older adults that are not captured by traditional fatigue measures. Therefore, the purpose of our study was to translate the PFS into Italian and to evaluate validity and internal consistency of the PFS-Italian version (PFS-I) in older adults with CVRD residing in an inpatient rehabilitation centre, older adults with PD, and community-dwelling healthy older adults as controls (HC). Furthermore, test-retest reliability and responsiveness was evaluated in individuals with CVRD.

Materials and methods

Study design

The study was carried out in two phases: a), the English version of the PFS was translated into Italian according to standardized guidelines [26]; b) a cross-cultural adaptation and multicenter validation study was conducted using a convenience sampling method to test psychometric properties of the PFS-I.

Ethical considerations

This study was ethically approved by the Ethics Committee of Canton Ticino, Bellinzona, Switzerland (No. 2022 − 01909 CE 4220), Ethics Committee of Istituti Clinici Scientifici Maugeri, Italy (No. 2769 CE), and Ethics Committee of Fondazione IRCCS Policlinico San Matteo Pavia, Italy (No. 0040059/22). Before enrolment, all participants were fully informed of the testing procedures and asked to sign an informed written consent form. Participant confidentiality and data protection were maintained. In addition, permission was granted by the original developer of the PFS (NWG) to translate and validate it into Italian.

Translation process

After receiving permission, the PFS was translated in accordance with international recommended standards [26] of a stepwise, forward-backward approach including cross-cultural adaptation with a multidisciplinary expert committee. First, two forward translations (Trl and Tr2) of the original English PFS into the Italian language were conducted by two Italian mother-tongue, independent translators (one with a background in physiology and an expertise in assessment of fatigue, and one naïve on the construct measured). These translations were then compared and reconciled into one Italian version (Tr1-2) by a physiotherapist and professional translator. The synthesized Italian translation was independently back translated by two bilingual persons, naïve to the diseases involved and the construct of fatigue, to produce comparative English versions (BTr1 and BTr2). Then, an expert committee that included methodologists, health professionals, and the original developer of the PFS evaluated and reviewed the English and Italian versions for face and content validity to produce the pre-final PFS-I version. To ensure the comprehensibility of instructions, response options, recall period, and that all questions were understood by Italian-speaking individuals, the pre-final version was piloted on ten older adults. During the pilot test, a physiotherapist assisted the participants in the completion of the questionnaire, interviewing them at the end of the administration to investigate any doubts and verify the exact understanding of the terms used. Feedback from participants and translators was discussed by the expert committee. At this stage, the text of the PFS-I was finalized, and is available upon request from the original developer (NWG) through https://www.sph.pitt.edu/pittsburgh-fatigability-scale.

Psychometric evaluation

Participants

A convenience mixed sample of 200 older adults affected by CVRD, PD, and healthy peers was consecutively recruited between October, 2023 and March, 2024. Individuals aged 70 + years with intact intellectual functioning and Italian language skills were eligible. Participants were excluded in case of acute or persistent infection, acute or uncontrolled heart or lung failure, unstable angina and myocardial infarction during the previous month, recent major surgery, acute or uncontrolled renal failure or hepatic dysfunction, severe hearing problem, drug treatments that may interfere with the fatigue level, or were unable or refuse to participate. Participants with CVRD were patients admitted to a rehabilitation centre; those with PD were community dwelling (Hoehn & Yahr score ranging from 1 to 3) [27] and were recruited voluntarily through a local association for people with PD. Healthy control (HC) participants were recruited among active older people attending a fitness centre.

Procedures

Participants were assessed at three different sites, located in Switzerland (Clinica Moncucco, Lugano, site 1) and Italy (Istituti Clinici Scientifici Maugeri IRCCS, Istituto di Tradate, site 2; Criams-Sport Medicine Centre, Voghera, site 3). At each site, participants were assessed using the same procedures. Psychometric properties and responsiveness of the PFS-I were analyzed according to the definition and criteria proposed by the COnsensus-based Standards for the selection of health Measurement Instruments (COSMIN) methodology [28]. Demographic and clinical characteristics of participants were collected. Participants underwent a cognitive health screening using the Short Portable Mental Status Questionnaire (SPMSQ) [29]. Only participants with normal cognitive status were included (i.e., SPMSQ score = 0).

During the first visit (T1) and under the supervision of a researcher, enrolled participants were asked to self-administer the PFS-I, the Fatigue Severity Scale (FSS), the Geriatric Depression Scale (GDS), and the Self-care Ability Scale for Elderly (SASE). In addition, the modified Barthel Index (mBI), the Short Physical Performance Battery (SPPB), and the 6 min Walking Test (6MWT) were also administered by a trained researcher to CVRD and PD groups.

To assess test-retest reliability, a sub-group of participants with CVRD completed the PFS-I four to six days later (T2). To ensure stability in that period, the Global Rating of Change (GROC) scale was used.

To estimate the minimal clinically important difference (MCID), another sub-group of patients with CVRD completed the PFS-I before discharge (T3) and were also asked to independently assess on a GROC scale the overall change in their perceived fatigability compared to initial self-administration at study enrollment.

Measures and tests

Pittsburgh Fatigability Scale (PFS)

The original PFS was developed to rate the perceived physical and mental fatigability in older adults aged ≥ 60 years [13, 14]. Participants rated on a scale from 0 (“no fatigue”) to 5 (“extreme fatigue”) the physical and mental fatigue that they expected or imagined they would experience immediately after completing activities of various intensity and duration. The scores for PFS Physical and Mental fatigability were summed separately across the 10-items (range 0 to 50, higher scores indicate greater fatigability). Three severity levels were set to interpret the PFS scores: most severe (≥ 25 and ≥ 20 points for the Physical and Mental subscales, respectively), more severe (ranges between 15 and 24 and 13–19 points for the Physical and Mental subscales, respectively) and less severe (< 15 and < 13 points for the Physical and Mental subscales, respectively) [7, 30, 31].

Fatigue Severity Scale (FSS)

The FSS is a self-administered questionnaire with 9 items that assess the intensity of fatigue in various scenarios over the last week [32]. In this study, the Italian version was adopted [33]. Each question is graded on a 7-point Likert scale ranging from 1 (“completely disagree”) to 7 (“completely agree”). The final FSS score is the average of the 9 items. The higher the score, the more severe the fatigue is and the more it affects a person’s activities. A threshold of ≥ 4 points is commonly applied in research studies to denote clinically significant fatigue [34].

Geriatric Depression Scale (GDS-15)

The GDS is considered a valid screening tool for depressive symptoms [35] and is available in a full [36] or short version [37]. For this study, the Italian short version (GDS-15) consisting of 15 dichotomous questions (Yes/No) was used [38]. Total score ranges from 0 to 15. Scores ≥ 6 are considered a sign of depression.

Self-care Ability Scale for Elderly (SASE)

The SASE is used to assess self-care abilities in older people [39]. It includes 17 items covering various aspects of daily life, such as well-being, loneliness, and dressing. Items are rated on a scale of 1 to 5, with 1 indicating strong disagreement and 5 strong agreement. Scores for items with negative statements are reversed before summing. The overall score ranges from 17 to 85, with higher values reflecting greater self-care capabilities. In this study, the Italian version of the SASE was used [40].

Modified Barthel Index (mBI)

The mBI was developed to assess functional outcomes [41]. It includes 10 items and uses 5 different levels of rating scales. Score ranges from 0 to 5 for bathing and grooming, 0 to 10 for feeding, dressing, bowel and bladder control, using the toilet, and climbing stairs, and 0 to 15 for transferring from a chair or bed, and walking. A higher score indicates a greater level of independence in activities of daily living. The Italian version of the mBI was used [42].

Short Physical Performance Battery (SPPB)

The SPPB [43] includes three physical demanding tests: (i) hold standing balance for 10 s in 3 positions: side-by-side, semi-tandem, and tandem stand; (ii) two-timed 4-m walks at one’s usual pace; and (iii) rising 5 times from a seated position to an upright position as quickly as possible with the arms folded across the chest. Total SPPB scores range from 0 to 12; higher score indicates better function.

The 6 min Walking Test (6MWT)

The 6MWT [44] quantifies the distance covered over a time of 6 min, along a 25-metre section of unimpeded walkway. Two cones are positioned on the ground to mark the distance that needs to be covered and as turning points. Greater distance indicates better functional exercise capacity.

Global Rating of Change Scale (GROC)

The GROC is a reliable self-reported tool used to measure changes in a patient’s health status between two points in time, whether it has improved, worsened, or remained stable [45, 46]. In this study, patients with CRVD were asked to respond to the question: “Compared to the time of your admittance to this Clinic, how would you describe your current level of physical and mental fatigue?” GROC scores ranged from − 3 to + 3, where − 3 represents very much worsened, + 3 signifies very much improved, and zero denotes no change. To test reliability, only patients with GROC scores ranging from − 1 to 1 were categorized as unchanged and included in the analysis.

To estimate the MCID, those who declared a change were also asked to judge “What importance would you assign to changes in your level of physical and mental fatigue?” The possible answers were no importance, of little importance, moderately important, and very important.

Statistical analysis

Demographic characteristics and distribution of scores were analyzed with descriptive statistics. The statistically significant level was fixed at p < 0.05. In case of missing items, participants were excluded from the analysis. The following measurement properties were evaluated: validity (structural validity, internal consistency, cross-cultural validity/measurement invariance, hypotheses testing for construct validity), reliability (test-retest reliability, measurement error), floor or ceiling effects, acceptability, and responsiveness (MCID). The content validity of the PFS was investigated for the original version [13], and was not analyzed again in this study. Factorial analysis was conducted using Stata/IC 16.0 (StataCorp, 4905 Lakeway Drive, College Station, Texas, USA), and other statistics were computed by SPSS software (version 29 for OS; SPSS Inc., Chicago, IL, USA).

Validity

First, the internal structure of the PFS-I was considered. Internal structure refers to how different items are related, and it is important to determine how to combine them into a scale or subscale. In addition, hypothesis testing for construct validity was conducted. The minimum sample size of 100 participants suggested by COSMIN for studying structural validity through factor analysis was met [28].

Structural validity

The PFS-I Physical and Mental scales’ validity was calculated on the full sample (T1) through an Exploratory Factor Analysis (EFA) with direct oblimin rotation and a subsequent confirmatory factor analysis (CFA) based on the EFA results. The robustness of the estimates was assessed by comparing the results of the overall CFA model with those of the population-specific models. As model fit indicators we considered an omnibus fit index (the 𝒳²) and four incremental fit indices, namely the comparative fit index (CFI), the Tucker-Lewis index (TLI), the root mean square error of approximation (RMSEA) and the standardized root mean square residual (SRMR). We considered values above 0.90 for the CFI and the TLI and below 0.08 for the SRMR and 0.06 for the RMSEA to indicate a good fit between the theoretical model and the observed data [47].

Internal consistency

Internal consistency refers to the degree of interrelatedness among items, indicating the extent to which these items measure the same construct [28]. Cronbach’s alpha, item-to-total correlation, and corrected inter-items correlations were used to assess the PFS-I Physical and Mental scales internal consistency on the whole sample. Cronbach’s alpha values between 0.70 and 0.95, Cronbach’s alpha if item deleted lower than the total alpha, item-to-total correlation values ≥ 0.25, and average inter-item correlations ideal range between 0.20 and 0.5, were considered adequate [48–50].

Cross-cultural validity/measurement Invariance

This measurement property investigates whether items of a patient-related outcome measure behave similarly in diverse populations (i.e., ethnicity, language, gender or disease populations) [28]. Known-group validity was evaluated by comparing mean differences (ANOVA) of the PFS-I Physical and Mental fatigability scores in the HC group versus the two groups of patients (CVRD and PD). In addition, female and male participants were compared in the whole sample and within each group to explore potential gender effects.

Hypotheses testing for construct validity

No gold standard for measuring fatigability are available. In this study, construct validity was investigated by testing hypotheses on expected associations of the PFS-I with the outcome measures administered, and expected differences between relevant groups.

Spearman’s rank correlation coefficient (r_s) was used to test correlations, and ANOVA to compare groups. Correlation was considered strong with r_s ≥0.60, moderate with r_s ≥0.30 and < 0.60, and weak with r_s <0.30 [51]. The following hypotheses were formulated: (1) we expected to observe moderate-to-strong correlation between PFS-I Physical and Mental subscales and the FSS in each group because the two questionnaires share similar constructs (i.e., fatigability and fatigue, respectively); (2) the correlation with GDS was presumed to be moderate for the PFS-I Mental scale, and weak-to-moderate for the PFS-I Physical scale in all the three groups because there is evidence about connections between mental fatigability and depressive symptoms in older adults [14, 18]; (3) the inverse correlation between PFS-I Physical score and SASE was thought to be moderate in the two groups of patients but not in HC group because fatigability can have an adverse effect on the self-care abilities; (4) the inverse correlation between PFS-I Physical score and mBI score was expected to be moderate because physical fatigability and level of independence in activities of daily living are two inter-related dimensions; (5) the inverse correlation between PFS-I Physical score and the 6MWT was presumed to be moderate, because PFS physical fatigability score was closely associated with measures of walking performance in older adults [13]; (6) the inverse correlation between PFS-I Physical score and SPPB test was presumed to be weak (i.e., r_s < 0.30) because fatigability can affect motor performance, but the SPPB includes also balance tasks, which may be less affected by physical fatigability; and (7) the mean scores of the CVRD and PD groups were expected to be higher than in the HC.

The construct validity was evaluated as satisfactory, mild, and insufficient if values ≥ 75%, between ≥ 50% and < 75%, and < 50% of the a-priori hypotheses were satisfied, respectively [28, 52].

Reliability

Reliability refers to the degree to which an assessment is free from measurement error. In this study, test-retest reliability and measurement error were assessed for both the PFS-I Physical and Mental scores in a subsample of people with CVRD.

Test-retest reliability

Test-retest reliability was calculated with the Intraclass Correlation Coefficient using a 2-way random effects model (ICC_2,1) with 95% confidence interval (CI). Cut-off values to interpret ICC were fixed as follows: poor < 0.50, moderate ≥ 0.50 and < 0.75, good ≥ 0.75 and < 0.90, and excellent reliability ≥ 0.90 [53]. Assuming a null attrition rate due to consecutive sampling, confidence of 90% and an expected reliability with ICC = 0.90, a sample size of 23 participants was required to test reliability [54].

Measurement error

Measurement error was assessed using the Standard Error of Measurement (SEM) and the Minimum Detectable Change (MDC). SEM was computed using the formula SD*√ (1 – ICC), with SD being the standard deviation of the baseline measurement. To interpret the SEM, cut-off percentage values of the scale range suggested by Ostelo et al. [55] were used as follows: ≤5% very good, ≤ 10% good, < 20% moderate, ≥ 20% poor.

The MDC, defined as the smallest unit of change that can be detected by the instrument beyond measurement error, was calculated by multiplying the SEM by 1.64 and 1.96 - corresponding to the z-score associated with 90% (MDC₉₀) and 95% (MDC₉₅) confidence level - and the square root of 2 [48]. Both the absolute and percentage of the scale range were calculated for SEM and MDC.

Floor and ceiling effects

A floor or ceiling effect was detected if ≥ 15% of the respondents achieved the lowest or highest possible score [48]. A floor effect in participants belonging to the HC group was considered normal.

Acceptability

The time needed to complete the PFS-I was recorded, and participants were asked about any problems they encountered while fulfilling the questionnaire.

Responsiveness

Responsiveness is the ability of a measure to detect change over time in the construct to be measured, and was estimated longitudinally in a sub-sample of patients with CVRD. In this study, the MCID ranges are proposed as a threshold range for determining meaningful change, which allows for distinguishing patients who have changed versus those who have not according to an external criterion, i.e., the GROC scale [56]. The GROC questions administered after treatment were used as an external criterion. To proceed with the analysis, the external anchor was correlated with the PFS-I change values obtained between T1 and T3, and a value ≥ 0.40 was required [57]. In accordance with previous studies [58, 59] our receiver-operating characteristic (ROC) analysis used the entire sample, rather than just those participants with ratings adjacent to the dichotomization point, to increase precision and obtain more logical estimates of the MCID. The scores on the GROC were dichotomized as “changed” if the response to the first question was ≥ 2 (the level of physical and mental fatigue was moderately or much improved), and changes were judged at least of little importance. The patients were classified as “unchanged” if the score to the first question was equal to 1 (little improved) or 0 (unchanged) or if they declared that changes were of no importance.

First, the ability of PFS-I to detect a clinically important change was determined by the area under the ROC curve (AUC). The AUC can be interpreted as the probability of correctly identifying a patient who has improved in randomly selected pairs of patients who have and have not shown an improvement. As a general rule, AUC values between ≥ 0.7 and < 0.8 are judged as acceptable, and ≥ 0.8 are considered representative of good to excellent discriminative ability [60].

Then, the ROC cut-off was calculated by maximizing Youden’s index (sensitivity + specificity – 1), i.e., associated with the least amount of misclassification. The MCID, i.e., the smallest score change that informed patients consider necessary, whether beneficial or harmful [48], was then estimated by matching ROC cut-off and MDC values.

Results

Translation process

The entire adaptation process took about 45 days to produce a culturally adapted Italian version of the PFS. The forward/backward translation went smoothly and there were no marked disagreements between the translators during this process. Furthermore, the expert committee reached full agreement when evaluating the two translated versions in terms of semantic, idiomatic, experiential, and conceptual equivalence. The pre-final version of the PFS-I was given to ten older adults, who had no difficulty understanding and answering the questions, confirming the translation’s comprehensibility and cognitive equivalence.

Descriptive statistics

A total of 228 eligible participants were screened for inclusion, of whom 200 (mean age = 76 ± 5.1 years, 53% male) had no exclusion criteria and were included in the study. Participants were divided in three groups: 87 were older adults with CRVD residing in an inpatient rehabilitation centre, 46 were individuals with PD, and 67 were healthy, community-dwelling participants included in the HC group. Complete data with no missing values were available for all measurements.

Detailed demographic characteristics of the participants and descriptive statistics of the outcome measures are presented in Table 1, while Table 2 shows descriptive statistics for each PFS-I item.

Table 1.

Demographic characteristics and descriptive statistics of the study samples

	Overall [N = 200]	HC [n = 67]	CVRD [n = 87]	PD [n = 46]
Age, years	76.0 ± 5.1	75.3 ± 4.7	77.2 ± 5.5	74.5 ± 4.1
Sex, women	94 [47%]	44 [66%]	37 [43%]	13 [28%]
Comorbidities (overall)			371
Trauma and orthopedics			40 [11%]
Vascular and hepatobiliary			64 [17%]
Internal medicine			111 [30%]
Cardiology			77 [21%]
Pulmonology			79 [21%]
Number of comorbidities
0–1			2 [2%]
2–4			44 [51%]
≥5			41 [47%]
T1
PFS Physical score, 0–50  <15 less severe  15–24 more severe  ≥25 most severe	19.4 ± 9.9 65 [33%] 74 [37%] 61 [30%]	11.8 ± 6.8 44 [66%] 21 [31%] 2 [3%]	23.1 ± 8.1 12 [14%] 36 [41%] 39 [45%]	23.5 ± 10.4 8 [17%] 17 [37%] 21 [46%]
PFS Mental score, 0–50  <13 less severe  13–19 more severe  ≥20 most severe	12.2 ± 10.7 119 [60%] 32 [16%] 49 [24%]	6.9 ± 6.7 53 [79%] 9 [13%] 5 [8%]	12.9 ± 11.0 49 [56%] 16 [18%] 22 [25%]	18.6 ± 10.9 17 [37%] 7 [15%] 22 [48%]
Total mBI score, 0-100	96.1 ± 6.8	99.2 ± 1.7	95.2 ± 6.8	92.9 ± 9.6
Total GDS-15 score, 0–15	3.9 ± 2.7	2.8 ± 2.5	4.6 ± 2.6	4.3 ± 2.6
Total SASE score, 17–85	62.9 ± 8.5	67.4 ± 5.1	61.6 ± 6.8	58.7 ± 11.8
Total FSS score, 9–63	31.1 ± 14.2	22.0 ± 10.9	35.5 ± 13.6	36.1 ± 13.4
Total SPPB score, 0–12	9.2 ± 2.2	10.4 ± 1.6	8.8 ± 2.0	8.2 ± 2.5
6MWT, m	376.0 ± 99.8	428.3 ± 73.2	362.1 ± 103.6	324.1 ± 90.5
T2 (N=22)
PFS Physical score, 0–50  <15 less severe  15–24 more severe  ≥25 most severe			25.8 ± 7.7 2 [9%] 8 [36%] 12 [55%]
PFS Mental score, 0–50  <13 less severe  13–19 more severe  ≥20 most severe			17.2 ± 21.2 8 [36%] 4 [19%] 10 [45%]
T3 (N=58)
PFS Physical score, 0–50  <15 less severe  15–24 more severe  ≥25 most severe			17.7 ± 7.2 20 [34%] 26 [45%] 12 [21%]
PFS Mental score, 0–50  <13 less severe  13–19 more severe  ≥20 most severe			8.3 ± 7.8 40 [69%] 12 [21%] 6 [10%]
GROC 1			1.7 ± 0.8
GROC 2			2.0 ± 0.8
Total SPPB score, 0–12			8.8 ± 3.8
6MWT, m			377.1 ± 119.1

PFS, Pittsburgh Fatigability Scale; mBI, Barthel Index Modified; GDS, Geriatric Depression Scale; SASE, Self-care Ability Scale for Elderly; FSS, Fatigue Severity Scale; SPPB, Short Physical Performance Battery; 6MWT, 6 min Walking Test; GROC 1, Global Rating of Change scale answering the question: Compared to the time of your admittance to this Clinic, how would you describe your current level of physical and mental fatigue?; GROC 2, Global Rating of Change scale answering the question: What importance would you assign to changes in your level of physical and mental fatigue?; HC, healthy controls (older adults); CVRD, older adults with cardiovascular and respiratory diseases; PD, older adults with Parkinson’s disease

Table 2.

Descriptive statistics of the Pittsburgh Fatigability Scale – Italian version (PFS-I) Physical and Mental scale items

PFS-I Physical scale items	Population				Statistical test result1
	Total [N = 200]	HC [n = 67]	CVRD [n = 87]	PD [n = 46]
Leisurely walk for 30 min	1.81 ± 1.63	0.78 ± 1.03	2.44 ± 1.60	2.13 ± 1.71	z = 41.32***
Brisk or fast walk for 1 h	3.27 ± 1.69	1.96 ± 1.42	4.17 ± 1.24	3.46 ± 1.59	z = 69.43***
Light household activity for 1 h	1.72 ± 1.54	0.90 ± 1.00	1.87 ± 1.48	2.65 ± 1.70	z = 34.11***
Heavy gardening or outdoor work for 1 h	3.02 ± 1.64	2.13 ± 1.61	3.45 ± 1.40	3.50 ± 1.60	z = 28.59***
Watching TV for 2 h	0.52 ± 1.14	0.21 ± 0.69	0.40 ± 0.99	1.20 ± 1.59	z = 26.04***
Sitting quietly for 1 h	0.42 ± 1.06	0.18 ± 0.55	0.33 ± 0.97	0.91 ± 1.53	z = 12.45**
Moderate to high intensity strength training for 30 min	2.84 ± 1.53	2.21 ± 1.47	3.25 ± 1.37	2.96 ± 1.63	z = 17.69***
Participating in a social activity for 1 h	0.88 ± 1.34	0.43 ± 0.91	0.90 ± 1.39	1.48 ± 1.55	z = 17.68***
Hosting a social event for 1 h	1.63 ± 1.57	1.01 ± 1.16	1.93 ± 1.70	1.96 ± 1.59	z = 14.21***
High-intensity activity for 30 min	3.32 ± 1.68	1.96 ± 1.49	4.41 ± 0.93	3.24 ± 1.57	z = 83.67***
PFS-I Mental scale items	Population				Statistical test result 1
	Total [N = 200]	HC [n = 67]	CVRD [n = 87]	PD [n = 46]
Leisurely walk for 30 min	0.62 ± 1.21	0.15 ± 0.53	0.63 ± 1.32	1.28 ± 1.41	z = 33.40***
Brisk or fast walk for 1 h	1.41 ± 1.81	0.49 ± 0.93	1.67 ± 2.00	2.26 ± 1.87	z = 28.40***
Light household activity for 1 h	1.05 ± 1.44	0.70 ± 1.18	0.80 ± 1.28	2.00 ± 1.66	z = 26.28***
Heavy gardening or outdoor work for 1 h	1.60 ± 1.72	1.23 ± 1.42	1.39 ± 1.70	2.52 ± 1.85	z = 16.17***
Watching TV for 2 h	0.98 ± 1.33	0.37 ± 0.97	1.28 ± 1.36	1.28 ± 1.44	z = 25.87***
Sitting quietly for 1 h	0.46 ± 1.04	0.15 ± 0.50	0.48 ± 1.06	0.87 ± 1.39	z = 12.44**
Moderate to high intensity strength training for 30 min	1.50 ± 1.69	1.10 ± 1.54	1.43 ± 1.72	2.22 ± 1.66	z = 14.38***
Participating in a social activity for 1 h	1.08 ± 1.33	0.54 ± 1.00	1.22 ± 1.39	1.61 ± 1.36	z = 21.37***
Hosting a social event for 1 h	1.60 ± 1.60	1.22 ± 1.46	1.59 ± 1.63	2.13 ± 1.61	z = 9.57**
High-intensity activity for 30 min	1.84 ± 1.78	0.96 ± 1.27	2.22 ± 1.93	2.41 ± 1.68	z = 23.86***

All values are expressed as mean ± standard deviation

¹ Kruskal-Wallis test with two degrees of freedom

*** p < 0.001, ** p < 0.01, * p < 0.05

Detail about skewness, kurtosis and normality regarding the distributions of each PFS-I item is reported in Table 3. Evaluation of the skewness data indicate that the distributions of 13 out of 20 items are low or moderately skewed (skewness parameter between − 1 and + 1), while the remaining 7 are highly asymmetric. The results of the Jarque-Bera test clearly indicate that none of the items are normally distributed.

Table 3.

Distribution of descriptive statistics of the Pittsburgh Fatigability Scale – Italian version (PFS-I) Physical and Mental scale items

PFS Physical scale items	Skewness	Kurtosis	Jarque-Bera test result
Leisurely walk for 30 min	0.38	1.99	𝒳2(2)1 = 31.27***
Brisk or fast walk for 1 h	-0.58	2.12	𝒳2(2) = 24.24***
Light household activity for 1 h	0.52	2.23	𝒳2(2) = 16.90***
Heavy gardening or outdoor work for 1 h	-0.39	2.01	𝒳2(2) = 28.95***
Watching TV for 2 h	2.30	7.50	𝒳2(2) = 76.63***
Sitting quietly for 1 h	2.84	10.63	𝒳2(2) = 98.74***
Moderate to high intensity strength training for 30 min	-0.16	2.02	𝒳2(2) = 24.95***
Participating in a social activity for 1 h	1.44	4.05	𝒳2(2) = 38.43***
Hosting a social event for 1 h	0.66	2.41	𝒳2(2) = 15.12***
High-intensity activity for 30 min	-0.64	2.19	𝒳2(2) = 21.56***
PFS Mental scale items
Leisurely walk for 30 min	2.03	6.24	𝒳2(2) = 64.59***
Brisk or fast walk for 1 h	0.93	2.35	𝒳2(2) = 23.75***
Light household activity for 1 h	1.18	3.24	𝒳2(2) = 26.60***
Heavy gardening or outdoor work for 1 h	0.64	2.06	𝒳2(2) = 29.82***
Watching TV for 2 h	1.14	3.21	𝒳2(2) = 25.54***
Sitting quietly for 1 h	2.36	7.78	𝒳2(2) = 78.99***
Moderate to high intensity strength training for 30 min	0.74	2.18	𝒳2(2) = 24.57***
Participating in a social activity for 1 h	0.93	2.69	𝒳2(2) = 19.28***
Hosting a social event for 1 h	0.63	2.20	𝒳2(2) = 20.94***
High-intensity activity for 30 min	0.45	1.83	𝒳2(2) = 63.23***

¹ The degrees of freedom (df) of the statistical test are reported in parenthesis

*** p < 0.001, ** p < 0.01, * p < 0.05

Structural validity

The EFA showed a two factors structure for the PFS-I Physical and Mental scores, accounting for approximately 60% of the variance in both cases (Table 4). Item factor loadings showed a very similar structure for the two scales: basically, the first factor represents activities involving all intensities of physical exercise, while the second factor accounts for sedentary and social activities. The only difference between the PFS-I Physical and Mental subscales concerns the item “Hosting a social event for 1 h”, that has the highest loading on the first factor in the PFS-I Physical but on the second factor in the PFS-I Mental subscale. However, the difference in the loadings on the two factors is very small, and this item could reasonably be included in both factors.

Table 4.

Results of the exploratory factor analysis for the Pittsburgh Fatigability Scale-Italian version Physical and Mental scales

PFS-I items1	Physical score factor loadings		Mental score factor loadings
	Factor 1	Factor 2	Factor 1	Factor 2
Leisurely walk for 30 min	0.71	0.26	0.71	0.17
Brisk or fast walk for 1 h	0.82	0.08	0.81	0.17
Light household activity for 1 h	0.63	0.46	0.71	0.35
Heavy gardening or outdoor work for 1 h	0.8	0.09	0.83	0.1
Watching TV for 2 h	0.1	0.76	0.13	0.81
Sitting quietly for 1 h	-0.01	0.82	0.18	0.73
Moderate to high intensity strength training for 30 min	0.72	0.11	0.75	0.22
Participating in a social activity for 1 h	0.28	0.69	0.44	0.6
Hosting a social event for 1 h	0.52	0.46	0.49	0.58
High-intensity activity for 30 min	0.83	-0.02	0.8	0.15
Eigenvalue	4.4	1.6	5	1.18
Percentage of explained variance	37.7	22.4	40.5	21.4

¹ Items with loading greater than 0.5 are highlighted in bold. Factor 1 represents all activities involving any level of physical exercise; Factor 2 accounts for sedentary and social activities

On the basis of the EFA results, we conducted a CFA by estimating a two-factor model, within which the two factors were allowed to covary and all cross-loadings were fixed at zero. Table 5 illustrates the goodness-of-fit of the CFA models estimated. The PFS-I Physical scale reported an acceptable goodness-of-fit for HC, but less robust overall for patients with CVRD and particularly for those with PD. Conversely, the PFS-I Mental scale was evaluated as acceptable overall and for those with PD, but less robust for HC and patients with CVRD.

Table 5.

Goodness-of-fit of the confirmatory factor analysis models for the Pittsburgh Fatigability Scale-Italian version Physical and Mental scales

Model	Likelihood ratio test		RMSEA1			CFI2	TLI3	SRMR4
	𝒳2(df5)	p-value	Value	90% CI6	p-value
PFS Physical scale
Total [N = 200]	169.79 (34)	0.000	0.14	(0.12–0.16)	0.000	0.83	0.77	0.09
HC [n = 67]	45.81 (34)	0.085	0.07	(0.00-0.12)	0.242	0.91	0.88	0.08
CVRD [n = 87]	71.21 (34)	0.000	0.11	(0.08–0.15)	0.005	0.85	0.80	0.09
PD [n = 46]	98.70 (34)	0.000	0.21	(0.16–0.25)	0.000	0.68	0.58	0.15
PFS Mental scale
Total [N = 200]	127.55 (34)	0.000	0.12	(0.10–0.14)	0.000	0.90	0.87	0.05
HC [n = 67]	76.69 (34)	0.000	0.14	(0.10–0.18)	0.001	0.78	0.71	0.10
CVRD [n = 87]	102.68 (34)	0.000	0.15	(0.12–0.19)	0.000	0.86	0.81	0.09
PD [n = 46]	51.12 (34)	0.030	0.11	(0.03–0.16)	0.082	0.90	0.87	0.08

¹ RMSEA = Root mean square error of approximation

² CFI = Comparative fit index

³ TLI = Tucker-Lewis index

⁴ SRMR = Standardized root mean square residual (not reported in case of missing values)

⁵ df = degrees of freedom

⁶ CI = Confidence interval

HC, healthy older adults; CVRD, older adults with cardiovascular and respiratory diseases; PD, older adults with Parkinson’s disease

Internal consistency

Cronbach’s alpha for the PFS-I Physical and Mental scales were 0.86 and 0.89 respectively, indicating an overall good level of internal consistency. The average inter-items correlations (0.37 and 0.44 respectively, ideal range between 0.20 and 0.5) confirmed good reliability. Item-to-total correlation values were all ≥ 0.25 (PFS-I Physical range: 0.31–0.69; Mental range: 0.43–0.72). Cronbach’s alpha if item deleted was lower or equal to the total alpha (PFS-I Physical range: 0.83–0.86; Mental range: 0.87–0.89).

Cronbach’s alpha for the PFS-I Physical scale composed of variables loading on Factor 1 was 0.87 with an average inter-item correlation of 0.49, showing a good and reliable internal consistency. However, the coefficient for the variables loading on factor 2 was 0.68 (average inter-item correlation = 0.42), slightly lower than the 0.7 acceptability threshold for internal consistency.

Cronbach’s alpha for the PFS-I Mental scale composed of variables loading on Factor 1 was 0.88 with an average inter-item correlation of 0.56, indicating a good level of internal consistency but an average inter-item correlation slightly above the ideal interval. The coefficient for the scale composed of variables loading on factor 2 was 0.74 (average inter-item correlation = 0.42), indicating an acceptable level of internal consistency.

Cross-cultural validity/measurement invariance

The scores of the HC group were significantly lower (p < 0.001) from those observed in the two patient groups, confirming known-group validity. No gender differences in PFS-I scores were observed in the entire sample. However, in CVRD group, females showed greater fatigue in the Physical subscale than males (p < 0.01).

Hypotheses testing for construct validity

The correlation matrix in Table 6 shows the association between the PFS-I and convergent measures. Construct validity of the PFS-I was excellent with approximately 86% (18 out of 21) of overall hypotheses accepted. However, the PFS-I Physical scale was not significantly correlated with the two physical tests (SPPB and 6MWT) in participants affected by PD, and no correlation was found between PFS-I Physical scale and SASE in HC. In 3 cases, the correlations were only slightly below the expected cutoff. Patients with CVRD or PD had higher PFS-I Physical scores than in HC (p < 0.001).

Table 6.

Hypothesis testing for Spearman’s rank correlations between the Pittsburgh Fatigability Scale-Italian version and other outcome measures

	Pittsburgh Fatigability Scale-Physical scale			Pittsburgh Fatigability Scale-Mental scale
	HC  [n = 67]	CVRD [n = 87]	PD [n = 46]	HC [n = 67]	CVRD [n = 87]	PD [n = 46]
Fatigue Severity Scale	0.38**	0.59***	0.35*	0.30*	0.41***	0.48***
Geriatric Depression Scale-15	0.28*	0.51***	0.59***	0.45***	0.34**	0.60***
Self-care Ability Scale for Elderly	-0.01	-0.48***	-0.42**
modified Barthel Index		-0.29**	-0.37*
Short Physical Performance Battery		-0.23*	-0.26
6 min Walking Test		-0.50***	-0.291

* p < 0.05, ** p < 0.01, *** p < 0.001, ¹p = 0.051

HC, healthy older adults; CVRD, older adults with cardiovascular and respiratory diseases; PD, older adults with Parkinson’s disease

Reliability

The test-retest reliability was calculated in a subsample of 23 patients with CVRD. Test-retest reliability was excellent in both PFS-I subscales (Physical scale, ICC = 0.93, 95% CI = 0.84–0.97; Mental scale, ICC = 0.92, 95% CI = 0.83–0.97).

The SEM was very good for the PFS Physical scale (2.2 points, equal to 4.4% of the total score), and good for the PFS Mental scale (2.9 points, 5.8%). The MDC₉₀ and MDC₉₅ beyond measurement error were 5.1 (10%) and 6.1 points (12.2%) for the PFS-I Physical scale, and 6.8 (13.6%) and 8.1 points (16.2%) for the PFS-I Physical and Mental scales, respectively.

Acceptability and floor/ceiling effects

The time needed to complete the PFS-I was 5 to 10’. All individuals in the sample responded to every item. No participant achieved the maximum score of 50 points (most severe fatigability status). As expected, a floor effect (least severe fatigability status) was observed in the HC group in the PFS-I Mental scale (18% of participant scored 0), but not for the PFS-I Physical scale (4.5% of participant scored 0). In the two groups of patients, no participant scored 0 for the PFS-I Physical scale (score range = 4–44). In contrast, the proportions of participants that scored 0 were 13% (score range = 0–42) and 2% (score range = 0–39) in the CVRD and PD groups, respectively.

Responsiveness

The responsiveness was calculated in a subsample of 54 patients with CVRD. The mean PFS-I Physical and Mental scales scores were 23.2 (SD 8.6) and 12.3 (SD 10.23) at admission, and 17.8 (SD 6.9) and 7.7 (SD 7.3) at discharge, respectively. Mean PFS-I Physical and Mental scores improved (i.e., scores have decreased) between admission and discharge by -5.4 (SD 5.7) and -4.6 points (SD 6.9), respectively (p < 0.0001). The correlation coefficient between the difference in PFS-I scores and the GROC scale was 0.40 (p < 0.01), reaching the required value.

The AUC for the ROC curve was 0.84 (95% CI 0.72–0.96) for the Physical and 0.74 (95% CI 0.60–0.89) for the Mental scale. Cut-off values were 5.5 points (sensitivity 70.6%, specificity 85%) and 2.5 points (sensitivity 73.5%, specificity 75%) for the PFS-I Physical and Mental scales, respectively. By examining the ROC cut-offs and MDC values together, we determined that appropriate MCID ranges were 6 to 7 points for the PFS-I Physical scale and 7 to 9 points for the Mental scale. The lower MCID values were set just above the MDC₉₀, while the higher values correspond to the MDC₉₅.

Discussion

The PFS was specifically designed to assess perceived physical and mental fatigability in older adults. This is particularly important when studying older people, who - in an attempt to reduce or avoid fatigue - may modify their activities (i.e., self-pacing) to maintain a tolerable effort [13]. The PFS improves and overcomes shortcomings in existing fatigue self-report tools by anchoring fatigue to set demand activities [13]. In addition, the PFS was designed to be user-friendly, both for older adults and healthcare professionals.

This study outlines the success of the cross-cultural adaptation of the PFS-I and is the first to have investigated a comprehensive set of psychometric properties and responsiveness of the scale with a classical approach. The questions of the PFS-I were clear and easy to understand, allowing for a quick (less than 10’) and accurate assessment of perceived physical and mental fatigability. Additionally, feasibility is supported by the high completion rate of all items, confirming its potential to be used in large population-based studies.

Overall, physical fatigability was low (below 1 point on average) for sedentary activities and participation in a social activity, higher (between 1.63 and 1.81 on average) for lower intensity activities and hosting a social event, and highest (between 2.84 and 3.32 on average) for moderate-to-high intensity activities, namely brisk or fast walk, heavy gardening or outdoor work, moderate to high intensity strength training and high-intensity activity. This supports the validity of the scale.

As expected, PFS-I Physical scores significantly differed among the three populations analyzed, with HC reporting the lowest perceived fatigability. The CVRD and PD groups showed overlapping mean fatigability scores and severity levels. However, patients with CVRD reported higher fatigability scores for activities involving walking or sports compared to patients with PD, while it was the opposite for sedentary activities, light household activities and participating in a social activity. In previous studies, patients with PD experienced higher perceptions of fatigability compared to healthy individuals, but physical activity levels have revealed no significant impact on this perception [61]. This might explain why those with PD experience relatively greater fatigability in domains including social activities or light physical activities.

As a whole, the lowest PFS-I Mental fatigability levels (around 0.5 points on average) were reported for the items leisurely walk and sitting quietly, while slightly higher mental fatigue levels (around 1 on average) were found for light household activities, watching TV and participating in a social activity. The highest PFS-I Mental fatigability (between 1.41 and 1.84 on average) was reported for moderate to high intensity activities and hosting a social event. Again, PFS-I Mental fatigability scores significantly differed among the three populations considered, with HC self-reporting the lowest scores. This observation further supports the construct validity of the scale. Compared to patients with CVRD (mean total score 12.9 ± 11.0), patients with PD reported higher mental fatigability for almost all items (mean total score 18.6 ± 10.9).

It is challenging to compare the mean fatigability scores to other published studies, as our sample is very heterogeneous in terms of age, disease and socio-cultural conditions. Nonetheless, our results showed that on average the PFS-I Mental scores were lower than the Physical score (overall mean scores and SD: Mental = 12.2 ± 10.7; Physical = 19.4 ± 9.9 points), in line with populations living in similar cultures [15, 20]. A different trend with mental higher than physical scores was observed in a Chinese population, where individuals have a tendency to use somatization in expressing suppressed mental distress [18]. Moreover, the average PFS-I Mental scores of the HC group were very similar to those of a large epidemiological study [62].

Structural validity

A theoretical four-factor structure of the PFS was originally proposed to ensure the consistency of the structure of item loadings with the hypothesized constructs [13]. More recently, the CFA performed on this solution resulted in weak model fit [15, 19], and a clean two-factor structure was found in the Dutch PFS Physical scale [15]. The same two factors (social and moderate-to-high intensity physical activity) has been later confirmed by the original developers in the Mental scale [14] and in the traditional Chinese version [18] of the PFS. In the current data, EFA confirmed previous results showing a two-factor structure for each PFS-I subscale, with the first factor representing activities involving physical activity, and the second factor accounting for sedentary and social activities. The two-factor result is nearly identical for the two scales. One cross-loading existed for the item “Hosting a social event for 1 hour” that could reasonably be included in both factors, particularly in the Physical scale. In the original USA [13], Dutch [15], and traditional Chinese [18] PFS Physical scale, this item loaded only on the factor of social activity. This could be likely explained by cross-cultural difference: Italians are known as a very convivial people, who hold events that can involve many people and foresee activities such as dancing. For this reason, the physical effort required to host social events can be perceived as more physically demanding than in other cultures.

Internal consistency

Internal consistency of the PFS-I was good, providing support for reliability and demonstrating that all items were closely correlated with perceived fatigability. Our results are comparable to previous studies [16, 18–20]. With respect to the sub-domains, only the first factor (activities involving any level of physical exercise) showed good internal consistency, while the second factor (sedentary and social activities) was acceptable (Mental scale) or slightly lower (Physical scale) than the acceptability threshold for internal consistency. Similar results were reported also for the Korean versions [19]. However, this can be attributable to the small number of items, and do not necessarily translate into poor internal consistency [19].

Cross-cultural validity/measurement invariance

No significant gender differences were found in this study, with the exception of the CVRD group in which females showed higher levels of physical fatigability than males. The mechanisms underlying the possible gender difference have not yet been fully elucidated: the results are often conflicting, especially with advanced aging when physical attributes decrease dramatically for both sexes. As expected, lower levels of fatigability were detected in the HC group.

Hypotheses testing for construct validity

The construct validity of the PFS-I was supported by our analyses. Our hypothesis that PFS-I would be moderately correlated with the FSS was met. Although the PFS-I and FSS share the same construct, the wording of the questions make the two scales very different, explaining why the correlation between the two instruments is not strong. In the FSS, questions are not anchored to the intensity and duration of activities: the scale measures the impact of fatigue on specific types of functioning and should therefore be regarded as a measure of state fatigue status. Conversely, the PFS was conceived to measure perceived fatigability and can be considered a trait measure.

The PFS-I showed moderate-to-strong correlation with a measure of depression. In addition, the PFS-I Physical scale showed acceptable concurrent validity with self-care abilities and activities of daily living. These findings were consistent with previous literature. For example, fatigability was correlated with depression in older adults living in Taiwan [18], United States and Denmark [14]. Moreover, physical disability in self-care and activities of daily living was a risk factor associated with greater perceived physical fatigability [5, 17]. However, the pattern of correlations was not always as expected. It was rather surprising to find only a weak-to-moderate correlation between physical fatigability and physical performance measures. Our results are different from those of previous studies conducted in community-dwelling or retirement community populations of older adults, showing an inverse correlation between PFS and SPPB (r = -0.50 to -0.51) or gait speed (r = -0.38 to -0.55) [5, 16, 17]. However, this disagreement may be explained by the fact that our sample consisted of participants with PD or CVRD, which may have affected performance on the physical tasks such as walking, sit-to-stand, or balance.

Test-retest reliability and measurement error

This study demonstrated excellent test-retest reliability of the PFS-I in a sub-population of participants with CVRD. Our results are higher than those previously reported for the Physical (ICC = 0.80) (15) and Mental (ICC = 0.78 to 0.81) scale [14, 15]. Likewise, the SEM and MDC values were clinically acceptable (i.e., SEM < 6% of the total scale range) and lower than those of the Dutch [15] and the English PFS versions [20]. This showed that the PFS remains a reliable scale for measuring perceived fatigue even in different cultures.

Responsiveness

To our knowledge, this is the first study to explore the responsiveness and interpretability of the PFS. Our results revealed that PFS-I attained good responsiveness (AUC > 0.70) in older people with CVRD. With a follow-up of about three to four weeks post-hospitalization, we were able to establish MCID ranges of 6 to 7 points for the PFS-I Physical and 7 to 9 points for the Mental scale. These thresholds could optimize the application of the PFS-I for interpreting observed changes in clinical settings and inform clinical decision-making in individuals with CVRD affected by perceived fatigability. In clinical practice, it is necessary to balance decisions between the risk of discharging patients deemed to be improved who may not actually be so, and the risk of continuing treatment even after improvements have already been achieved [56]. The narrow MCID ranges obtained for both PFS-I subscales allow clinicians to limit the risks of misinterpretation of scores. In addition, the MCIDs can be used to compare effectiveness of treatment within clinical trials, integrating measures of statistical significance. Moreover, they can be considered when designing randomized clinical trials for sample size estimation.

Limitations

This study has some limitations. We used a convenience sampling method which can introduce selection bias and limit the generalizability of the findings. This sampling technique may not adequately represent the broader population of older adults affected by CVRD and PD or the healthy peers. Furthermore, the inclusion of specific populations may decrease the generalizability of these results to other older adult populations not represented in the sample.

Furthermore, the sample size was considered ‘adequate’ according to the COSMIN recommendations for the analysis of test-retest reliability and measurement error was not achieved, although it is sufficient according to the formula reported by Zou [54].

Conclusions

This study provides evidence of the psychometric properties and responsiveness of the PFS-I in a mixed population of older adults. The PFS-I is a simple-to-use tool that assesses physical and mental fatigability by considering both the intensity and duration of physical activities, effectively overcoming the issue of self-pacing hampering general fatigue assessments. The PFS also serves as a prognostic or outcome indicator in epidemiological and clinical research, improving our understanding of phenotypic aging.

Author contributions

MBP, SV, NWG, and MB conceived and designed the work; EZ, FN, CP, GDA, and SP acquired the data; SV, ES, and MBP analyzed and interpreted the data; all authors made substantial contribution to the drafting of the manuscript; all authors read, revised, and approved the submitted version of the manuscript.

Funding

Financial support has been granted by Araldi-Guinetti Foundation, Vaduz (LI). The National Institute on Aging Claude D. Pepper Older Americans Independence Center, Research Registry and Developmental Pilot Grant (NIH/NIA P30 AG024827), and the Intramural Research Program, National Institute on Aging supported NWG to develop and validate the Pittsburgh Fatigability Scale.

Data availability

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was ethically approved by the Ethics Committee of Canton of Ticino, Bellinzona, Switzerland (approval number: 2022 − 01909 CE 4220), Ethics Committee of “Istituti Clinici Scientifici Maugeri SpA-SB”, Pavia, Italy (approval number: 2769 CE), and Ethics Committee of Pavia, IRCCS Policlinico San Matteo Foundation, Italy (approval number: 00440059/22). Before enrolment, all participants were fully informed of the testing procedures and asked to sign an informed written consent form. Participant confidentiality and data protection were maintained.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Clinical trial number

Not applicable.