https://orcid.org/0000-0003-0043-3853
Figures
Abstract
Introduction
Frailty leads to disability, morbidity, and mortality in older persons. The Fried’s Frailty Phenotype (FFP), derived in the American Cardiovascular Health Study (CHS), is widely used around the world to define frailty, but lacks adaptation in African populations.
Objective
To derive a modified FFP definition which best identifies frailty in a Southern African context.
Methods
A population-based cross-sectional study of adults aged ≥40 years collected data from questionnaires and physical assessments. Original CHS, population-dependent, European Working Group on Sarcopenia in Older People2 (EWGSOP2) and Sarcopenia Definitions and Outcomes Consortium (SDOC) and independent thresholds were all applied to the five FFP criteria (weight loss, exhaustion, low physical activity [PA], low grip strength [GS] and slow walking speed [WS]) to assess non-differentiality, internal consistency, and plausibility.
Results
The 919 participants had a median age of 59 years [IQR 50–70], 53.3% were female. Self-reported exhaustion was reported by 37.5%nd self-reported weight loss by 34.9%. Using the lowest quintile of body mass index (BMI), missed 15.2% of overweight and obese participants who reported weight loss. Using CHS thresholds, low PA was present in 36.7%. Grip strength correlated better with age (r = −0.45) than BMI (r = −0.19). Therefore, the sex-specific tenth percentile of the 40–49-years-age band of the study population was used rather than the CHS approach. The modified SDOC threshold identified slow WS in almost all (85.8%) and was therefore non-differential. The EWGSOP2 and CHS thresholds identified slow WS in 52.9% and 22.9%, respectively, compared to 34.5% using the study population’s lowest quintile.
Conclusion
Culture and language sensitive questions for self-reported exhaustion and weight loss, CHS thresholds for low PA, and population dependent thresholds for GS and WS were the most suitable modifications in a Southern African setting, highlighting the need for region-specific adaptations when diagnosing frailty.
Citation: Madela EIY, Gregson CL, Paruk F, Burton AJ, Patel R, Wilson H, et al. (2026) Defining frailty using a modified Fried’s Frailty Phenotype in a Southern African context. PLoS One 21(2): e0340723. https://doi.org/10.1371/journal.pone.0340723
Editor: Mario Ulises Pérez-Zepeda, Instituto Nacional de Geriatria, MEXICO
Received: July 16, 2025; Accepted: December 24, 2025; Published: February 4, 2026
Copyright: © 2026 Madela et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data will be stored on a secure password protected University of Bristol network filestore space. Data are available upon reasonable request. Third party researchers wishing to access individual data records will be able to apply following processes outlined at http://www.bristol.ac.uk/staff/researchers/data/accessing-research-data/, subject to a data transfer agreement and informed consent procedures. Data for sharing will be anonymised and contain no information that could directly identify any of the participants. Data users will need to acknowledge data sources and ensure regulatory requirements of relevant ethical bodies are met.
Funding: This work was supported by the National Institute for Health Research (NIHR) (using the UK’s Official Development Assistance (ODA) Funding) and Wellcome (217135/Z/19/Z) under the NIHR-Wellcome Partnership for Global Health Research. CLG is funded by the National Institute for Health and Care Research (NIHR302394). The work was further supported by the UK Medical Research Council (MR/W003961/1). The views expressed are those of the authors and not necessarily those of the Wellcome Trust, NIHR or the Department of Health and Social Care. The funders had no role in the conceptualization, design, data collection, analysis, decision to publish, or preparation of the manuscript. There was no additional external funding received for this study.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Life expectancy (LE) is increasing globally [1], especially in sub-Saharan Africa (SSA) where improvements in survival have added nearly 11 years to the average LE from 1990 to 2021 [1]. Reflecting this trend, in South Africa (SA) the proportion of persons aged 60 years and over increased from approximately 7.7% (3.7 million) in 2006 to 9.2% (5.6 million) in 2022 [2]. With this increase in LE a rapid rise in age-related conditions, such as frailty, is expected.
Ageing is accompanied by physiological changes across multiple systems [3], leading to a steady decline in homeostatic reserves, and increasing risk of non-communicable diseases and geriatric syndromes, including frailty. Frailty is defined clinically as an increased vulnerability towards dependency and/or death when exposed to a stressor, such as acute illness, injury, or psychological stress [4,5]. It is a major cause of disability, morbidity, and mortality in older persons [6,7] and is associated high healthcare costs [6].
The prevalence of frailty differs by geographic location, socio-economic status, and how frailty is measured [6,8–12]. Numerous clinical and research scales have been developed to assess frailty. The Fried’s Frailty Phenotype (FFP) defines frailty by its physical characteristics, or ‘phenotype’, namely the presence of three or more of: exhaustion (self-reported), shrinking (weight loss), weakness (low grip strength), slowness (slow walking speed) and low physical activity [4]. Pre-frail equates to the presence of one or two of these features. The relative simplicity of the assessment needed to define the FFP makes is feasible and practical to use in a resource-constrained setting, such as Southern Africa.
The FFP was originally derived in 2001 from two cohorts (n = 5,317) of American subjects enrolled in the Cardiovascular Health Study (CHS) [4]. It has since been successfully applied in multiple epidemiological studies globally where it is predictive of adverse clinical outcomes, including mortality [13–15]. Due to socio-cultural differences, the FFP has been modified in several countries [16–25]. An example is the use of sarcopenia thresholds from the Sarcopenia Definitions and Outcomes Consortium (SDOC) [26] and the European Working Group on Sarcopenia in Older People2 (EWGSOP2) [27] to define slow walking speed. Another example relates to mean grip strength (GS), which varies with ethnicity with lower values reported in African populations compared to European and North American populations [28], invalidating the use of CHS thresholds in a Southern African context. Hence, there have been efforts to modify the FFP for use in SSA, but none have approached this in a systematic way. Inconsistent definition modifications make prevalence comparisons difficult [16]. A recent scoping review on application of screening instruments for frailty in SSA identified 17 studies, which used seven different frailty screening methods, with only one developed and validated in SSA [29]. The brief frailty screening tool in Tanzania (BFIT) was developed in older population in a rural low-resource setting. The study population was characterized by low educational attainment, traditional lifestyles and diet, subsistence farming as the main form of employment, and limited access to healthcare services [30]. The tool has not been used in SA.
In SA, the prevalence of frailty in community-dwelling older adults ranges between 5.4% and 38% depending on frailty definitions and age range [12,17–26]. Hence, the aim of this study was to determine, from a range of different FFP modifications, the most internally consistent, differential, and plausible definition of frailty in the Southern African context.
Materials and methods
Study population and recruitment
This study utilised data from a population-based cross-sectional study conducted in KwaMashu, an urban area of KwaZulu-Natal, SA [31], which lies 12 kilometres north of eThekwini (Durban) and is home to 175,663 persons, 37,633 (21.4%) of whom are aged 40 years or older [2]. Recruitment began in April 2022 and ended in February 2023. Satellite images and OpenStreetMap, available through the software Aeronautical Reconnaissance Coverage Geographic Information System (ARCGIS), were used to map all of KwaMashu. Household blocks were randomly selected to be representative of the underlying population and maximise generalisability of findings. Population denominator data by sex and five-year age-bands, obtained through national census data (2011), were used to build a stratified random quota sampling frame. Participants (1 male: 1 female) were recruited by fieldworkers from households. The study aimed to recruit 168 men and 168 women in each of three age strata 40–54, 55–69 and ≥70 years, to reach a total of 504 women and 504 men. Adults were eligible if they were aged 40 or older and had lived and shared meals in the household for at least four weeks. Participants provided written or thumbprint informed consent. For those who lacked capacity to consent due to cognitive impairment, legal guardian (proxy) written/ thumbprint informed consent was sought. Pregnant women were excluded. Consenting participants were then assessed at their local clinic [31].
Data collection and measurements
Data were collected using questionnaires (translated into isiZulu) and standardized examination forms, with data entered directly into Research Electronic Data Capture (REDCap)® [32], with inbuilt data validation checks. Sociodemographic data included sex and age, verified by the date of birth on identification documents. A single question from the Shona Symptom Questionnaire (SSQ) “I felt run down” was asked to identify self-reported exhaustion [33]. A question “Have you, or those close to you, noticed that you have lost weight or become thinner in the last 12 months?” was asked to determine the presence of weight loss (because home scales are not widely available and so measured weight loss is not possible in this setting). Physical activity was assessed using the short form of the International Physical Activity Questionnaire (IPAQ) [34]. Metabolic equivalent (MET) minutes were calculated for walking, moderate intensity and vigorous intensity activities in accordance with IPAQ instructions and converted to kilocalories (Kcal) per week [35]
All measurements were performed by trained research staff, following standardised operating procedures. Grip strength (GS) was measured in kilograms (kgs) using a Jamar dynamometer (ASP Global, Atlanta Industrial Parkway, USA). Contraindications included recent (within 12 weeks) hand trauma or surgery, excessive pain, swelling or hand inflammation. Grip strength was measured three times in each hand, and the maximal GS was used. Walking speed (WS) in seconds was measured by timing participants’ walking at their usual pace over a 4-metre (13.1 feet) flat surface. Participants who used walking aids at home performed the test with their walking aids. Balance was assessed by 10 second tandem, semi-tandem and full tandem stands (not using walking aids) and chair stand time (five sit-to-stands as fast as possible) with arms folded across the chest. To derive the Short Physical Performance Battery (SPPB), WS, balance, and chair stand times were each allocated between 0 and 4 points. A score of >9–12 was regarded as normal and 0–9 as low [36]. Weight was measured in kilograms (kg) using a calibrated Seca® scale (SECA Gesellschaft mit beschränkter Haftung (GMBH) & Co., Hamburg, Germany) (participants wearing minimal clothing, shoes and socks removed, pockets emptied). Standing height was measured in millimetres (mm) using a portable Seca® stadiometer (shoes, socks, hats, head scarves removed). Mid-upper arm circumference (MUAC) was measured on the non-dominant arm using an ergonomic Seca® measuring tape; < 24 cm was considered low [37]. All measurements were repeated and the mean calculated. If there was a difference of >0.5 kgs or >5 mm between measurements, measurements were repeated until the difference was < 0.5. Body mass index (BMI) (kg/m2) was calculated by dividing weight (kg) by height (metres) squared. World Health Organization (WHO) categories were used to classify BMI as underweight: < 18.5 kg/m2, normal: 18.5–24.9 kg/m2, overweight: 25–29.9 kg/m2, obese class I: 30–34.9 kg/m2, obesity class II: 35–39.9 kg/m2 and morbidly obese (class III): ≥ 40 kg/m2.
Development of diagnostic criteria for frailty
Original Fried frailty criteria.
Table 1 shows the original CHS thresholds for each of the five FFP criteria. A positive response to either or both questions from the Centre for Epidemiologic Studies Depression (CES-D) Scale [38] defined exhaustion. Weight loss was defined as unintentional loss of >10 pounds in the last year, or ≥5% measured weight loss when compared to the previous year’s body weight. Physical activity was measured in kcal using the Minnesota Leisure Time Physical Activity Questionnaire (MLTPA) [39]. Low GS and slow WS were determined using CHS study-specific lowest sex-specific quintiles, stratified by BMI and height, respectively. Frailty was considered to be present if three or more criteria were met and pre-frailty if one or two criteria were met. The absence of any criteria classified the person as being robust [4]
Assessment of frailty and context-specific modifications to the FFP CHS criteria.
In the current study (KwaMashu study) the following modifications were applied to the original FFP criteria (CHS study).
- 1.. Exhaustion
A positive response to a single statement from the SSQ “I felt run down” was used to identify exhaustion.
- 2.. Weight loss
To quantify unintentional weight loss participants were asked: “Have you, or those close to you, noticed that you have lost weight or become thinner in the last 12 months?”. Due to concerns over recall bias, we further considered defining weight loss based on the lowest sex-specific quintile of BMI.
- 3.. Physical activity
The IPAQ Short Form [34], calculated MET minutes/week, which were converted to Kcal per week. The threshold for defining low PA used in the CHS (the sex-specific lowest quintile) was compared to the sex-specific lowest quintile in the KwaMashu population.
- 4.. Grip strength
For GS, the following were used as potential thresholds: (i) the lowest sex-specific quintile of CHS GS stratified by BMI, (ii) the lowest sex-specific quintile of GS in the KwaMashu study’s youngest age band (40–49 years) and (iii) the lowest sex-specific 10th percentile of GS in the KwaMashu study’s youngest age band (40–49 years).
- 5.. Walking speed
Four different thresholds were applied to potentially identify slow WS (Table 1), based on the existing study population and international guidelines developed for the diagnosis of sarcopaenia, the: (i) original FFP CHS threshold, ii) lowest sex-specific KwaMashu study quintile stratified by median standing height, (iii) 2023 modified SDOC threshold of <1 m/s [26], and (iv) <0.8 m/s as recommended by the EWGSOP2 [27].
As per the original FFP, the presence of three or more criteria classified the participants as frail, one or two as pre-frailty and the absence of all criteria indicated someone was robust.
Ethical and governance approvals
Ethical approval was obtained from the University of KwaZulu-Natal’s Biomedical Research Ethics Committee; reference numbers BREC/00002513/2021and BREC/00005591/2023. Permission to conduct the study was granted by the Health Committee of the local area council and the District and Provincial Departments of Health (NHRD Ref: KZ_202106_024).
Statistical analysis
Statistical analysis was performed with Stata® SE version 17. Data were assessed for normality using histograms. Participant characteristics were summarised using descriptive statistics; continuous data were summarised using means with standard deviation (SD) if normally distributed, or medians with interquartile ranges (IQR) if not normally distributed. Categorical values were summarised using frequency counts and percentages. Participants with missing data required to calculate FFP criteria were excluded from analyses. The percentage of participants meeting each criterion, using the different thresholds, was reported.
The appropriateness of the different approaches at defining the FFP criteria were sequentially assessed as follows; a threshold needed to (i) be differential (i.e., not be ubiquitous and not rare, and instead be able to identify a proportion of the study population >5% or < 80%), (ii) demonstrate internal consistency by correctly identify individuals, in whom other characteristics, known to be associated with frailty, are associated, and (iii) be clinically plausible. For example, to assess internal consistency, self-reported weight loss and BMI were cross tabulated and the relationships between GS, age and BMI were assessed by linear regression. Frail participants were expected to be older, with lower PA, GS, WS, BMI, MUAC, and SPPB scores compared to robust participants [4,44,45], thus these associations were investigated.
Results
Characteristics of the study population
In total, 5,136 households were approached; 1,440 people aged ≥40 years were identified of whom 1,035 were invited and accepted to participate, and 968 then consented and were recruited into the study. Of these 919 (94.9%) had complete data to permit analyses (S1 Fig). Their median [IQR] age was 59 [50–70] years, and 52.8% were female (Table 2). Females had a higher median BMI and MUAC than males 34.1 kg/m2 [28.5–39.9] vs. 24.3 kg/m2 [21.0–28.8], and 28.0 cm [24.4–33.4] vs. 24.6 cm [21.0–28.3] respectively, with a lower median SPPB score 9.5 [7–11] vs. 11 [8–12].
Criteria for frailty: proportions using different thresholds
- 1.. Exhaustion
The prevalence of self-reported exhaustion was 37.5% (n = 345); with 34.3% (n = 149) males and 40.4% (n = 196) females reporting feeling run down (Table 2).
- 2.. Weight loss
Overall 34.9% (n = 321) self-reported weight loss, whilst 19.7% (n = 181) fell within the lowest sex-specific quintile of BMI (<20.4 kg/m2 in men and <27.4 kg/m2 in women) and 15.2% were either overweight or obese (Table 3). Only 7.5% (n = 24) participants self-reporting weight loss were underweight, with the majority being either overweight (21.1%; n = 71) or obese (38.9%, n = 125), suggesting that using BMI alone missed weight loss in this population, thus self-reported weight loss was used (S1 Table).
- 3.. Low physical activity
Physical activity consumed and estimated median [IQR] of 726.2 [0–2556.1] and 197.6 [0–1654.8] Kcal per week in men and women, respectively. Almost half (49%) of study participants reported not walking for more than 10 minutes a day in the last week, thus both the lowest and second lowest quintiles were characterized by (almost) zero minutes of PA in a week, invalidating an approach based on the lowest quintile. Therefore, the CHS thresholds (lowest of sex-specific quintiles) were applied, which classified 36.7% of participants as having low PA (Table 3).
- 4.. Low grip strength
The median [IQR] grip strength was 38 [32−44] kg for males and 28 [22−33] kg for females. Using the sex-specific CHS thresholds, stratified by BMI, low GS was prevalent in 20.5%. Given the high prevalence of obesity, the relationship between GS and BMI was investigated to internally validate the CHS approach; however, we found a poor correlation (overall r = −0.19, r = 0.14 in males and r = 0.09 in females), invalidating the CHS sex-specific, BMI-stratified approach. Instead, GS was more closely correlated with age (overall r = −0.45, r = −0.42 in males and r = −0.56 in females) (S2 Fig), supporting the use of sex-specific age-stratified thresholds. The lowest sex-specific quintile (20%) of GS in the KwaMashu study’s 40–49-year age group categorized 52.7% as having low GS, while 30.5% met the criterion when a 10th percentile threshold was used (Table 3). The 10th percentile (which was also equivalent to ≤2 SDs below the mean) was favoured over the lowest quintile approach as it is more clinically plausible for the younger age group to have good GS, with only 10% at most having low GS.
- 5.. Slow walking speed
The median [IQR] WS was 0.8m/s [0.7–1.0] and 0.7m/s [0.6–0.9] m/s for men and women respectively. Almost all of the study population met the criterion for slow WS (85.8%) when the modified SDOC threshold of <1.0 m/s was used; suggesting this was a non-differential approach. Using the CHS WS thresholds, a lower than seemed clinically plausible prevalence of slow WS (22.9%) was identified. The lack of applicability of the CHS approach may be explained by the shorter height of the KwaMashu study population compared to the CHS population (168.2 cm vs 173 cm for males and 156.5 cm vs 159 cm for females respectively).
Using the KwaMashu study population’s sex-specific lowest quintile, stratified by median height, 34.5% had a slow WS, whilst 52.9% did when using the EWGSOP2 thresholds. As both were clinically plausible, both these thresholds were then used in two frailty modifications for further comparison.
Frailty modifications
The prevalence of individual frailty criteria within the two remaining potential FFP modifications, differing only by the WS threshold used, are shown in Supplementary Table 2, with mean age, anthropometric characteristics, and SPPB scores summarised in (Table 4). In both modifications, exhaustion and weight loss were self-reported, CHS thresholds defined low PA, and the sex-specific 10th percentile of the 40–49 years age group of the KwaMashu study defined low GS. In the first modification the EWGSOP2 threshold was applied to WS, generating prevalence of pre-frailty and frailty of 57.7% and 29.7% respectively. In the second modification low WS speed was defined as the KwaMashu population’s sex-specific lowest quintile stratified by median height; this generated prevalence of prefrailty and frailty of 58.2% and 25.5%, respectively, so the modifications were similar in this regard (S2 Table).
When comparing the two modifications, frail participants in modification two were older, with lower SPPB score, lower MUAC and took longer to walk over the 4-meters as compared to modification one. Interestingly in both modifications, there was no consistency between the number of frailty criteria and BMI. For consistency we opted to use the study specific threshold for WS in the final model, namely modification 2, as summarized in Fig 1.
Discussion
The purpose of this study in KwaMashu, SA, was to determine the most suitable modifications and thresholds for FFP criteria in the local population. We modified the FFP criteria for frailty by using culturally and linguistically appropriate questions for exhaustion and weight loss. For PA, although we used the IPAQ rather than the MLTPA questionnaire to derive kcal/week, we used the CHS threshold, whereas for GS and WS population-dependent thresholds were applied. Thus, the modified FFP, mostly using population specific thresholds, was identified as the most suitable, internal consistent modification in this setting.
The original FFP criteria and thresholds, derived in the CHS study using their population specific thresholds, have been validated in high income countries. However, these criteria and thresholds are influenced by ethnic variations, socio-cultural differences, levels of health awareness and poverty, triggering the need for modifications in SSA [8,43,46–49]. A prerequisite of the translation of questionnaires developed in English into local languages (in this instance IsiZulu) is that the original meaning and intent of questions is maintained. However, a previous study in SA found that translating one of the questions from the CES-D used to denote exhaustion in the original FFP, (“everything you did was an effort”) was found to be ambiguous [43]. We therefore opted to use a more culturally and linguistically appropriate question from the SSQ which has been validated in this setting to detect common mental health symptoms, including exhaustion [40,50].
Quantifying weight loss is uncommon in our population, due to a lack of health awareness, non-availability of weighing-scales, as well as differences in appreciation of body image [51]. While BMI was explored as a potential as a marker of weight loss, weight loss was masked by the high prevalence of obesity [52,53]. A more suitable alternative in this context was self-reported weight loss significant enough to be noticed by those closest to the participant. Although intentional weight loss is uncommon currently, this may change in the future in response to public health interventions. In which case, the question will need to be revised to capture unintentional weight loss specifically.
Quantifying PA in the KwaMashu population was challenging even when substituting the MLTPA questionnaire (used in the CHS) with the IPAQ – Short Form which has been validated in SA [41]. Despite this contextual validation, vigorous and moderate activities like heavy lifting, digging, fast bicycling, are activities that may require access to equipment or a gym subscription, which are uncommon practices for older persons in this resource constrained setting. Surprisingly, 49% of the study population reported not walking for more than 10 minutes in the past week, scoring 0 Kcals/week, which suggests that the IPAQ may not be sensitive enough to capture minimal to mild activity, as most participants reported no disability or walking difficulties. We were therefore unable to use the lowest quintile of the KwaMashu study population as a threshold for low PA and defaulted to use the CHS thresholds instead.
Grip strength, as a measure of muscle strength, decreases with age and is associated with disability and increased mortality [54,55]. Ethnic variations in GS invalidate the use of CHS thresholds [28]. Furthermore, the normative data derived for African populations in the Prospective Urban Rural Epidemiology (PURE) study have an upper age limit of 70 years and were therefore not appropriate for this study [28]. The CHS thresholds for low GS were stratified by BMI. There is conflicting evidence of the relationship between GS and BMI, potentially as BMI is unable to distinguish between muscle and fat components of body composition. In an Australian study greater GS was weakly related to higher BMI in adults under the age of 30 years and those over the age of 70 years, but inversely related to higher BMI between these ages [56]. In India, Dhananjaya et al noted that BMI and GS were negatively correlated in male participants with normal BMI and weakly negatively correlated in obese males. In overweight females, GS and BMI were weakly negatively correlated, but not in those with normal BMI [57]. Thus BMI might influence the interpretation of low GS in countries with high levels of obesity such as SA. We found that in our population where the prevalence of obesity is high, age correlated better with GS than did BMI, justifying using the lowest sex-specific 10th percentile of the 40–49 years age group as the threshold rather than BMI based approach.
Walking speed, while easy to measure and an important indicator of poor health outcomes in older persons, is affected by several factors including age, lower extremity muscle power and leg-length [58]. Thus, WS thresholds were stratified by sex and median standing height in the CHS. Compared to the CHS, the SA participants were shorter (173 cm vs 168.2 cm and 159 cm vs 156.5 cm, for males and females, respectively), invalidating the use of the CHS threshold in this setting. Using WS thresholds recommended by SDOC and EWGSOP (mostly derived from high-income populations) of <1m/s and <0.8m/s, respectively [26,27], overestimated the prevalence of slow WS in this setting, highlighting the need for population-specific thresholds. We therefore adopted the slowest sex-specific quintile of the KwaMashu study population stratified by median standing height.
In a systematic review of 262 studies investigating the impact of modifications to the FFP on frailty classification, prevalence, and predictive ability, modifications were commonly made to weight loss and low PA definitions [16]. Experiencing similar limitations when quantifying weight loss, studies have used either the lowest quintile of BMI for their population, or a BMI < 18.5 kg/m2 to define weight loss [16–18,22,25]. This approach was not effective in our setting where SA is within the top 10 countries with the highest levels of obesity [53]. The prevalence of obesity in SA was reported to be 32.1%, 15.3% in males and 41.3% in females in 2023 with KZN having the highest obesity prevalence compared to other provinces [59]. Obesity is common in other SSA countries such as Gabon, Botswana, Lesotho and Swaziland [53]. Using low BMI as a marker of weight loss underreported prefrailty and frailty in SA. This was supported by the lack of a relationship between self-reported weight loss and BMI and the lack of an inverse relationship between BMI and the number of frailty criteria in our study. More participants reported weight loss, than had low BMI, supporting our rationale that weight loss observed by the participant or someone close to them was differential in identifying frailty. Findings were similar for MUAC, where more participants reported weight loss than had low MUAC. Hence, in the absence of an objective measurement of weight-change, we recommend the use of self-reported weight loss rather than BMI in this setting.
The prevalence of frailty differed in our study population depending upon the modifications used to define the five criteria of FFP, as has been reported previously [16]. The prevalence of frailty was 25.5% using the final modification. This prevalence was significantly higher than in the previous studies conducted in SA, using the FFP [8,43]. The Study on global AGEing and adult health (SAGE) reported a prevalence of 14.5% in persons aged 50 years and older [8] the Health and Aging in Africa: Longitudinal Studies of an INDEPTH Community study (HAALSI) conducted in a rural population aged 40 years and above reported a prevalence between 5.4% and 13.2% [43]. The higher prevalence in our study, may reflect a higher proportion of older people, as well as the definitions of frailty.
To our knowledge, this is first study to explore the challenges and necessary modifications to the FFP definition of frailty in a Southern African context. Recruitment was designed to be able to study a representative sample of the general population. Our study had several limitations. The insensitivity of the IPAQ to capture mild PA limited application of different thresholds to identify those with low PA. The median age of our study population was younger than the original FFP validation study. This reflects the lower average life expectancy of SA at 65 years, compared to 77 years in the United States in 2001. Even though the African population is relatively younger (chronological age), African populations tend to age faster biologically than in European populations [60]. Hence, the SAGE and HAALSI studies on frailty conducted in SA were conducted in populations aged 50 years and over and aged 40 years and over, respectively [8,43]. Our approach gave us the opportunity to compare our findings with these previous studies. Self-reported weight loss was most suitable in this setting as low BMI and/ or MUAC were not sensitive markers of weight loss due to high obesity prevalence. The intention to lose weight was not explored in this study, therefore, self-reported weight loss may have included those with healthy weight loss and overestimated the frailty prevalence. Future studies using self-reported weight loss should consider adding a question on intent. Data were all cross-sectional, so we were unable to validate the FFP definition against longitudinal outcomes such as hospital admission and mortality. Furthermore, we lacked access to electronic health records, that might have enabled computation of an electronic frailty index, against which could have validated our approach. The FFP thresholds proposed remain to be validated and in other provinces and ethnic groups in Southern Africa.
Conclusion
In conclusion, with an increasing longevity in SA, the prevalence of frailty is likely to increase and contribute to disability, morbidity, and mortality in older persons with associated high healthcare costs. Frailty can be assessed in a number of ways and the FFP is one of the most widely used phenotypic approaches; however, contextual application is needed. We have identified thresholds for each of the five FFP criterion that are suitable for the Southern African context. These thresholds show that one in four older, urban-dwelling South African men and women are frail.
Supporting information
S1 Fig. Participant flow chart describing household-based identification and recruitment of study participants.
https://doi.org/10.1371/journal.pone.0340723.s001
(TIF)
S2 Fig. Relationship between body mass index and grip strength and age and grip strength in the total population and in men and women.
Relationship between body mass index and grip strength in the total cohort (A), in men (C) and women (D) and age and grip strength in the total population (B) and in men (E) and women (F).
https://doi.org/10.1371/journal.pone.0340723.s002
(TIF)
S1 Table. Body Mass Index in the participants who self-reported weight loss.
https://doi.org/10.1371/journal.pone.0340723.s003
(DOCX)
S2 Table. The prevalence of frailty according to the three modifications of the FFP.
https://doi.org/10.1371/journal.pone.0340723.s004
(DOCX)
Acknowledgments
A heartfelt gratitude to the eThekwini municipality officers and KwaMashu ward councilors who approved access to the area, assisted with community sensitization and approved use of the community hall for clinical assessments. Fieldworkers and nurses who were involved in recruitment and data collection. The KwaMashu Community Clinic management for allowing us to use clinic premises for clinical assessments and accepting referrals for participants who required continued care.
References
- 1.
United Nations. United Nations World Population Prospects 2022. Department of Economic and Social Affairs, Population Division; 2022. https://population.un.org/wpp/
- 2.
Statistics South Africa. Statistics South Africa. 2022. [Cited 2025 March 26]. https://www.statssa.gov.za/?p=15601
- 3. Fried LP, Ferrucci L, Darer J, Williamson JD, Anderson G. Untangling the concepts of disability, frailty, and comorbidity: implications for improved targeting and care. J Gerontol A Biol Sci Med Sci. 2004;59(3):255–63. pmid:15031310
- 4. Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci. 2001;56(3):M146-56. pmid:11253156
- 5. Morley JE, Vellas B, van Kan GA, Anker SD, Bauer JM, Bernabei R, et al. Frailty consensus: a call to action. J Am Med Dir Assoc. 2013;14(6):392–7. pmid:23764209
- 6. Shamliyan T, Talley KMC, Ramakrishnan R, Kane RL. Association of frailty with survival: a systematic literature review. Ageing Res Rev. 2013;12(2):719–36. pmid:22426304
- 7. Hoogendijk EO, Afilalo J, Ensrud KE, Kowal P, Onder G, Fried LP. Frailty: implications for clinical practice and public health. Lancet. 2019;394(10206):1365–75. pmid:31609228
- 8. Hoogendijk EO, Rijnhart JJM, Kowal P, Pérez-Zepeda MU, Cesari M, Abizanda P, et al. Socioeconomic inequalities in frailty among older adults in six low- and middle-income countries: Results from the WHO Study on global AGEing and adult health (SAGE). Maturitas. 2018;115:56–63. pmid:30049348
- 9. Collard RM, Boter H, Schoevers RA, Oude Voshaar RC. Prevalence of frailty in community-dwelling older persons: a systematic review. J Am Geriatr Soc. 2012;60(8):1487–92. pmid:22881367
- 10. Siriwardhana DD, Hardoon S, Rait G, Weerasinghe MC, Walters KR. Prevalence of frailty and prefrailty among community-dwelling older adults in low-income and middle-income countries: a systematic review and meta-analysis. BMJ Open. 2018;8(3):e018195. pmid:29496895
- 11. Bandeen-Roche K, Seplaki CL, Huang J, Buta B, Kalyani RR, Varadhan R, et al. Frailty in Older Adults: A Nationally Representative Profile in the United States. J Gerontol A Biol Sci Med Sci. 2015;70(11):1427–34. pmid:26297656
- 12. Chen X, Mao G, Leng SX. Frailty syndrome: an overview. Clin Interv Aging. 2014;9:433–41. pmid:24672230
- 13. Damluji AA, Chung S-E, Xue Q-L, Hasan RK, Moscucci M, Forman DE, et al. Frailty and cardiovascular outcomes in the national health and aging trends study. Eur Heart J. 2021;42(37):3856–65. pmid:34324648
- 14. Brighton LJ, Nolan CM, Barker RE, Patel S, Walsh JA, Polgar O, et al. Frailty and mortality risk in COPD: a cohort study comparing the fried frailty phenotype and short physical performance battery. Int J Chron Obstruct Pulmon Dis. 2023;18:57–67. pmid:36711228
- 15. Fan JN, Sun ZJ, Yu CQ, Guo Y, Sun DJY, Pei P, et al. Comparison of Fried phenotype and frailty index and their associations with risk of mortality. Zhonghua Liu Xing Bing Xue Za Zhi. 2021;42(7):1179–87. pmid:34814528
- 16. Theou O, Cann L, Blodgett J, Wallace LMK, Brothers TD, Rockwood K. Modifications to the frailty phenotype criteria: Systematic review of the current literature and investigation of 262 frailty phenotypes in the Survey of Health, Ageing, and Retirement in Europe. Ageing Res Rev. 2015;21:78–94. pmid:25846451
- 17. Strandberg TE, Sirola J, Pitkälä KH, Tilvis RS, Strandberg AY, Stenholm S. Association of midlife obesity and cardiovascular risk with old age frailty: a 26-year follow-up of initially healthy men. Int J Obes (Lond). 2012;36(9):1153–7. pmid:22614054
- 18. Szanton SL, Seplaki CL, Thorpe RJ Jr, Allen JK, Fried LP. Socioeconomic status is associated with frailty: the women’s health and aging studies. J Epidemiol Community Health. 2010;64(1):63–7. pmid:19692719
- 19. Tribess S, Virtuoso Júnior JS, Oliveira RJ de. Physical activity as a predictor of absence of frailty in the elderly. Rev Assoc Med Bras (1992). 2012;58(3):341–7. pmid:22735227
- 20. Xue Q-L, Walston JD, Fried LP, Beamer BA. Prediction of risk of falling, physical disability, and frailty by rate of decline in grip strength: the women’s health and aging study. Arch Intern Med. 2011;171(12):1119–21. pmid:21709116
- 21. Batista FS, Gomes GA de O, Neri AL, Guariento ME, Cintra FA, Sousa M da LR de, et al. Relationship between lower-limb muscle strength and frailty among elderly people. Sao Paulo Med J. 2012;130(2):102–8. pmid:22481756
- 22. Bandeen-Roche K, Xue Q-L, Ferrucci L, Walston J, Guralnik JM, Chaves P, et al. Phenotype of frailty: characterization in the women’s health and aging studies. J Gerontol A Biol Sci Med Sci. 2006;61(3):262–6. pmid:16567375
- 23. Aranda MP, Ray LA, Snih SA, Ottenbacher KJ, Markides KS. The protective effect of neighborhood composition on increasing frailty among older Mexican Americans: a barrio advantage?. J Aging Health. 2011;23(7):1189–217. pmid:21948774
- 24. Alvarado BE, Zunzunegui M-V, Béland F, Bamvita J-M. Life course social and health conditions linked to frailty in Latin American older men and women. J Gerontol A Biol Sci Med Sci. 2008;63(12):1399–406. pmid:19126855
- 25. Avila-Funes JA, Amieva H, Barberger-Gateau P, Le Goff M, Raoux N, Ritchie K, et al. Cognitive impairment improves the predictive validity of the phenotype of frailty for adverse health outcomes: the three-city study. J Am Geriatr Soc. 2009;57(3):453–61. pmid:19245415
- 26. Westbury LD, Beaudart C, Bruyère O, Cauley JA, Cawthon P, Cruz-Jentoft AJ, et al. Recent sarcopenia definitions-prevalence, agreement and mortality associations among men: Findings from population-based cohorts. J Cachexia Sarcopenia Muscle. 2023;14(1):565–75. pmid:36604970
- 27. Cruz-Jentoft AJ, Bahat G, Bauer J, Boirie Y, Bruyère O, Cederholm T, et al. Sarcopenia: revised European consensus on definition and diagnosis. Age Ageing. 2019;48(1):16–31. pmid:30312372
- 28. Leong DP, Teo KK, Rangarajan S, Kutty VR, Lanas F, Hui C, et al. Reference ranges of handgrip strength from 125,462 healthy adults in 21 countries: a prospective urban rural epidemiologic (PURE) study. J Cachexia Sarcopenia Muscle. 2016;7(5):535–46. pmid:27104109
- 29. Dzando G, Ward PR, Asante D, Ambagtsheer RC. Application of frailty screening instruments for older people in Sub-Saharan Africa: A scoping review. Ageing Res Rev. 2024;102:102571. pmid:39476936
- 30. Gray WK, Orega G, Kisoli A, Rogathi J, Paddick S-M, Longdon AR, et al. Identifying frailty and its outcomes in older people in rural Tanzania. Exp Aging Res. 2017;43(3):257–73. pmid:28358296
- 31. Burton A, Drew S, Cassim B, Jarjou LM, Gooberman-Hill R, Noble S, et al. Fractures in sub-Saharan Africa: epidemiology, economic impact and ethnography (Fractures-E3): study protocol. Wellcome Open Res. 2023;8.
- 32. Patridge EF, Bardyn TP. Research Electronic Data Capture (REDCap). jmla. 2018;106(1).
- 33. Patel V, Simunyu E, Gwanzura F, Lewis G, Mann A. The shona symptom questionnaire: the development of an indigenous measure of common mental disorders in harare. Acta Psychiatr Scand. 1997;95(6):469–75. pmid:9242841
- 34. Hallal PC, Victora CG. Reliability and validity of the International Physical Activity Questionnaire (IPAQ). Med Sci Sports Exerc. 2004;36(3):556. pmid:15076800
- 35.
IPAQ. Guidelines for data processing and analysis of the International Physical Activity Questionnaire (IPAQ) – short and long forms. 2005. https://www.physiopedia.com/images/c/c7/Quidelines_for_interpreting_the_IPAQ.pdf
- 36. Metanmo S, Simo-Tabue N, Kuate-Tegueu C, Bonnet M, Gbessemehlan A, Metanmo F, et al. Short physical performance battery and study of osteoporotic fractures index in the exploration of frailty among older people in cameroon. Int J Public Health. 2023;68:1605900. pmid:37609077
- 37. Van Tonder E, Mace L, Steenkamp L, Tydeman-Edwards R, Gerber K, Friskin D. Mid-upper arm circumference (MUAC) as a feasible tool in detecting adult malnutrition. South African Journal of Clinical Nutrition. 2018;32(4):93–8.
- 38. Stansbury JP, Ried LD, Velozo CA. Unidimensionality and bandwidth in the Center for Epidemiologic Studies Depression (CES-D) Scale. J Pers Assess. 2006;86(1):10–22. pmid:16436016
- 39. Pereira MA, FitzerGerald SJ, Gregg EW, Joswiak ML, Ryan WJ, Suminski RR, et al. A collection of Physical Activity Questionnaires for health-related research. Med Sci Sports Exerc. 1997;29(6 Suppl):S1-205. pmid:9243481
- 40. Mthiyane N, Harling G, Chimbindi N, Baisley K, Seeley J, Dreyer J, et al. Common mental disorders and HIV status in the context of DREAMS among adolescent girls and young women in rural KwaZulu-Natal, South Africa. BMC Public Health. 2021;21(1):478. pmid:33691665
- 41. Bohlmann IM, MacKinnon? S, Kruger S, Leach L, Heerden JV, Cook I, et al. Is the international physical activity questionnaire (IPAQ) valid and reliable in the south african population? Medicine & Science in Sports & Exercise. 2001;33(5):S119.
- 42. Gregson CL, Madanhire T, Rehman A, Ferrand RA, Cappola AR, Tollman S, et al. Osteoporosis, rather than sarcopenia, is the predominant musculoskeletal disease in a rural south african community where human immunodeficiency virus prevalence is high: a cross-sectional study. J Bone Miner Res. 2022;37(2):244–55. pmid:34694025
- 43. Payne CF, Wade A, Kabudula CW, Davies JI, Chang AY, Gomez-Olive FX, et al. Prevalence and correlates of frailty in an older rural African population: findings from the HAALSI cohort study. BMC Geriatr. 2017;17(1):293. pmid:29281995
- 44. Lewis EG, Coles S, Howorth K, Kissima J, Gray W, Urasa S, et al. The prevalence and characteristics of frailty by frailty phenotype in rural Tanzania. BMC Geriatr. 2018;18(1):283. pmid:30445919
- 45. Perracini MR, Mello M, de Oliveira Máximo R, Bilton TL, Ferriolli E, Lustosa LP, et al. Diagnostic accuracy of the short physical performance battery for detecting frailty in older people. Phys Ther. 2020;100(1):90–8. pmid:31612228
- 46. Witham MD, Davies JI, Bärnighausen T, Bountogo M, Manne-Goehler J, Payne CF, et al. Frailty and physical performance in the context of extreme poverty: a population-based study of older adults in rural Burkina Faso. Wellcome Open Res. 2019;4:135. pmid:32280791
- 47. Vecchio A, Nakigozi G, Nakasujja N, Kisakye A, Batte J, Mayanja R, et al. Assessment, prevalence, and correlates of frailty among middle-aged adults with HIV in rural Uganda. J Neurovirol. 2021;27(3):487–92. pmid:33788138
- 48. Pathai S, Gilbert C, Weiss HA, Cook C, Wood R, Bekker L-G, et al. Frailty in HIV-infected adults in South Africa. J Acquir Immune Defic Syndr. 2013;62(1):43–51. pmid:23018372
- 49. Edwards A, Siedner MJ, Nash S, Neuman M, Danaviah S, Smit T, et al. HIV serostatus, inflammatory biomarkers and the frailty phenotype among older people in rural KwaZulu-Natal, South Africa. Afr J AIDS Res. 2020;19(3):177–85. pmid:32892699
- 50. Skeen S, Tomlinson M, Macedo A, Croome N, Sherr L. Mental health of carers of children affected by HIV attending community-based programmes in South Africa and Malawi. AIDS Care. 2014;26 Suppl 1(0 1):S11-20. pmid:24766642
- 51. Mchiza ZJ, Parker W-A, Makoae M, Sewpaul R, Kupamupindi T, Labadarios D. Body image and weight control in South Africans 15 years or older: SANHANES-1. BMC Public Health. 2015;15:992. pmid:26423378
- 52. van Vollenstee FA, van der Merwe M-T. Obesity and its implications for COVID-19 pandemic in South Africa. S Afr J Infect Dis. 2021;36(1):228. pmid:34192123
- 53.
World Obesity Federation. Obesity: missing the 2025 global targets. 2020. https://data.worldobesity.org/publications/WOF-Missing-the-2025-Global-Targets-Report-FINAL-WEB.pdf
- 54. Cai Y, Liu L, Wang J, Gao Y, Guo Z, Ping Z. Linear association between grip strength and all-cause mortality among the elderly: results from the SHARE study. Aging Clin Exp Res. 2021;33(4):933–41. pmid:32524391
- 55. Malhotra R, Tareque MI, Tan NC, Ma S. Association of baseline hand grip strength and annual change in hand grip strength with mortality among older people. Arch Gerontol Geriatr. 2020;86:103961. pmid:31704626
- 56. Massy-Westropp NM, Gill TK, Taylor AW, Bohannon RW, Hill CL. Hand Grip Strength: age and gender stratified normative data in a population-based study. BMC Res Notes. 2011;4:127. pmid:21492469
- 57. Dhananjaya JR, Veena HC, Mamatha BS, Sudarshan CR. Comparative study of body mass index, hand grip strength, and handgrip endurance in healthy individuals. Natl J Physiol Pharm Pharmacol. 2017;7(6):594–8.
- 58. Bohannon RW. Comfortable and maximum walking speed of adults aged 20-79 years: reference values and determinants. Age Ageing. 1997;26(1):15–9. pmid:9143432
- 59.
National food and nutrition security survey. Human Sciences Research Council; 2024. https://foodsecurity.ac.za/publications/national-food-and-nutrition-security-survey-national-report/
- 60. Hansen CW, Strulik H. Biological age across the globe: 1990–2019. The Journal of the Economics of Ageing. 2025;31:100573.