Abstract
Objective To examine the accuracy of individual Fried frailty phenotype measures in identifying the Fried frailty phenotype in primary care.
Design Retrospective chart review.
Setting A community-based primary care practice in Kitchener, Ont.
Participants A total of 516 patients 75 years of age and older who underwent frailty screening.
Main outcome measures Using modified Fried frailty phenotype measures, frailty criteria included gait speed, hand-grip strength as measured by a dynamometer, and self-reported exhaustion, low physical activity, and unintended weight loss. Sensitivity, specificity, accuracy, and precision were calculated for single-trait and dual-trait markers.
Results Complete frailty screening data were available for 383 patients. The overall prevalence of frailty based on the presence of 3 or more frailty criteria was 6.5%. The overall prevalence of individual Fried frailty phenotype markers ranged from 2.1% to 19.6%. The individual criteria all showed sensitivity and specificity of more than 80%, with the exception of weight loss (8.3% and 97.4%, respectively). The positive predictive value of the single-item criteria in predicting the Fried frailty phenotype ranged from 12.5% to 52.5%. When gait speed and hand-grip strength were combined as a dual measure, the positive predictive value increased to 87.5%.
Conclusion There is a need for frailty measures that are psychometrically sound and feasible to administer in primary care. While use of gait speed or grip strength alone was found to be sensitive and specific as a proxy for the Fried frailty phenotype, use of both measures together was found to be accurate, precise, specific, and more sensitive than other possible combinations. Assessing both measures is feasible within primary care.
Frailty has been defined as a state of increased vulnerability due to age-associated decline in reserve and function resulting in reduced ability to cope with stressors.1,2 Currently, frailty is recognized as a multidimensional concept with dynamic interrelated factors in the physical, psychological, social, and environmental domains that affect the physiologic equilibrium of the person.3 Frailty has been associated with higher risk of adverse health outcomes,4 functional impairment and mortality,5,6 emergency department visits and hospitalizations,7 and postoperative complications and reduced survival.8,9 Early frailty is easily overlooked because its manifestations can be subtle or dismissed as normal aging, and physicians’ training has been focused on identifying specific medical diseases rather than overall vulnerability and the overall effect of diseases on homeostasis.
Although a number of conceptualizations of frailty have been proposed,10,11 frailty has been frequently described as a clinical phenotype of slowed walking speed, low physical activity, unintentional weight loss, low energy, and low grip strength (weakness), where the presence of 3 of 5 criteria indicates frailty.5 Multiple frailty scales have been developed to date.12 Although several systematic reviews have not conclusively identified a preferred instrument for measuring frailty in the elderly, the Fried frailty phenotype measure has been extensively tested for its validity and is a widely used instrument in frailty research.12–17
Recognition of frailty is important because there is evidence that the degree of frailty can be improved with interventions such as high-intensity exercise training and nutritional supplementation.18,19 The presence of frailty can affect the potential risks and benefits of medical interventions, as well as alter appropriate treatment targets when managing comorbid conditions such as hypertension and diabetes.18,20,21 When frailty is identified, management can be directed at identifying and addressing conditions that might underlie frailty and mitigating stressors that might precipitate adverse outcomes. Yet in the busy clinical setting of primary care practice, there is currently a lack of consensus as to how best to screen for, assess, and diagnose frailty.22 Care processes in typical primary care practice are not conducive to the administration of many existing measures of frailty that are time intensive. There is a need to identify measures that are both psychometrically sound and feasible to administer in the primary care setting. The feasibility of assessing frailty in primary care might be facilitated by the use of single-trait markers, rather than composite measures such as the Fried criteria. Commonly used single traits include gait speed and grip strength.23,24 Single-trait studies23–25 have measured outcomes associated with frailty such as morbidity, mortality, hospitalizations, and falls. To our knowledge, no studies have assessed the test accuracy of Fried frailty phenotype measures in identifying the Fried frailty phenotype itself.
The purpose of this study was to examine the relative accuracy of individual Fried frailty phenotype measures in identifying the Fried frailty phenotype in a primary care setting.
METHODS
This was a retrospective chart review of consecutive patients aged 75 years and older who were assessed through a comprehensive screening program to identify at-risk seniors at the Centre for Family Medicine Family Health Team (CFFM FHT) in Kitchener, Ont, from April 1, 2013, to April 31, 2014. The CFFM FHT is a primary care setting comprising 18 family physician practices with a combined population of 27 997 patients, 1291 of whom are 75 years of age or older. The Case-finding for Complex Chronic Conditions in Seniors 75+ (C5-75) program at the CFFM FHT is designed to identify seniors who are frail and might have unrecognized comorbid conditions underlying their frailty.
Before regularly scheduled medical appointments, all patients 75 years of age or older were assessed through the C5-75 screening program by specially trained nurses. There are 2 levels to the C5-75 screening, the first of which screens for frailty. Patients identified as frail proceed to the second level, which is a more comprehensive screening for common geriatric issues associated with frailty such as polypharmacy, fall risk, cognitive impairment, mood disorders, and social isolation. When concerns are identified, the patient is referred to the appropriate care professional for assessment and intervention. Any patients for whom 1 or more frailty criteria were not measured during the period of April 1, 2013, to April 31, 2014, were excluded from this analysis.
This study was approved by the McMaster University Research Ethics Board in Hamilton, Ont.
Measures
The Fried frailty phenotype was selected to identify frailty because it has been validated in a number of studies12,26,27 and is widely used in studies examining frailty associated with various health conditions and outcomes.16,17,28 The 5 Fried frailty phenotype elements were assessed.5 Gait speed was calculated as the time (in seconds) to walk 4 m at a usual pace. The fastest time of 2 trials was recorded.29 Grip strength (in kilograms) was measured as the higher score of 2, 3-second trials (with each hand) using a hand-held dynamometer (Jamar Hydraulic Hand Dynamometer, model 281-12-0600, J.A. Preston Corporation, Clifton, NJ). Self-reported exhaustion was measured using the Center for Epidemiologic Studies Depression Scale item “I could not get going,” with response choices for frequency in the past week: rarely or none of the time (< 1 day), some or a little of the time (1 to 2 days), a moderate amount of the time (3 to 4 days), or most of the time (5 to 7 days).5,30 Weight loss was measured as self-reported unintentional weight loss in the previous year. Physical activity was measured by self-reported descriptions (I am physically active; I do 30 minutes or more of moderate intensity physical activities on 5 or more days per week; I am physically active occasionally or during some seasons much more than others; I am not physically active beyond moving around or walking during activities of daily living).31
Patients were classified as frail if they met at least 3 of the following 5 criteria.5
Gait speed measured as 6 seconds or more to walk 4 m, independent of sex.
Grip strength measured as a score within the lowest 20%, stratified by sex.32
Exhaustion measured as not being able to get going a moderate amount of the time or all of the time.
Weight loss measured as unintentional loss of 4.5 kg or more in the past year.
Activity level measured as not being physically active beyond walking around during activities of daily living.
Data analysis
Descriptive statistics were used to summarize the data. Sex differences in frailty indicators were determined using Fisher exact tests (2-tailed). Sensitivity, specificity, accuracy, and positive predictive value were calculated for single-trait and dual-trait markers by constructing 2 × 2 contingency tables. Standard epidemiologic definitions were used for the calculations. The 95% CIs were calculated using the Wilson procedure with continuity correction.33 Analyses were performed using R, version 3.0.1.
RESULTS
A total of 516 patients were screened in the C5-75 program. Complete data on the frailty criteria measured in this study were available for 383 patients (Table 1). Patients ranged in age from 75 to 94 years of age; 53.5% were female.
Characteristics of patients excluded from and included in the study
The prevalence of frailty based on the Fried criteria (ie, positive on 3 or more indicators) was 6.5% (n = 25). As shown in Table 2, the overall incidence of each frailty marker ranged from 2.1% (weight loss) to 19.6% (lack of exercise). Frailty in grip strength was defined as those in the lowest quintile and by definition should be 20%. Deviations from 20% are a result of multiple patients near the lowest quintile having the same measured grip strength. Female patients were significantly more likely than male patients were to meet the Fried frailty criteria for slow gait (P = .026) and low activity level (P = .030). There were no significant sex differences for the other frailty components.
Incidence of single-trait frailty indicators
Diagnostic accuracy
The individual traits comprising the Fried frailty phenotype all showed sensitivity and specificity of more than 80%, with the exception of weight loss, which appears to be a poor indicator of frailty in this sample (Table 3). The predictive value of the single traits in predicting the Fried frailty phenotype ranged from 12.5% to 52.5%, suggesting a large number of false positives. All possible dual-trait combinations were tested in this population (data not shown); of these, gait speed and hand-grip strength yielded the most precise results. When gait speed and hand-grip strength were combined as a dual measure, the positive predictive value increased to 87.5%.
Diagnostic accuracy of frailty markers
DISCUSSION
This study found that while use of either gait speed or grip strength alone was sensitive and specific as a proxy for the Fried frailty phenotype, the dual-trait measure of gait speed with grip strength was accurate, precise, specific, and more sensitive than individual traits and other possible dual-factor combinations. To our knowledge, this is the first study to empirically examine the accuracy and precision of the individual Fried phenotype elements as a proxy for the original Fried frailty phenotype in a primary care setting. A recent systematic review concluded that in predicting risk of adverse outcomes, use of gait-speed measurements alone was as good as use of other composite tools and that there was sufficient evidence to support the use of gait speed as a single-item assessment tool for screening for frailty in community-dwelling older adults.23 The intent of finding a single-trait marker for frailty is predicated on the desire for a rapid assessment tool without a substantial loss of diagnostic accuracy. However, if intended as a proxy for the Fried frailty phenotype, then one must consider the positive predictive value of the tests. Other researchers15 have corroborated our finding that single-trait markers of frailty have moderate to low predictive value; that is, they produce too many false positives. In our study, gait speed alone identified greater than 60% more patients as frail compared with using both gait speed and hand-grip strength. In clinical practice, this is an important consideration because identifying a patient as frail can initiate a cascade of investigations and follow-up testing that can be both time-consuming and intensive.34
Screening for all 5 frailty markers within primary care practice might be impractical and a barrier to widespread frailty screening. As a proxy measure of the Fried frailty phenotype, we suggest that adding grip strength to gait speed would substantially increase the precision of screening without a large additional investment of time, training, or cost.29 This is feasible, as the required time for a nurse to measure gait speed is estimated to be less than 5 minutes, and measuring grip strength using a dynamometer takes approximately 2 minutes. Dynamometer use requires minimal staff training, and the cost is relatively affordable for most medical practices. With substantially reduced false positives, the potential benefits for patients associated with identifying and managing frailty and reducing adverse outcomes through proactive interventions might be important enough to justify the time and resource investments required for assessing gait speed and grip strength in seniors.
In primary care practice, recognition of frailty offers the opportunity to identify and optimize the management of coexisting conditions that might contribute to, or be affected by, frailty and to mitigate stressors that might precipitate adverse outcomes. Management of frail seniors might include medication review35; more frequent outpatient visits with the primary care physician36,37; exercise interventions for strength, endurance, and balance training18,38; and informed discussion about risks associated with surgical procedures.39 Because frailty is a predictor of survival,10 identification of frailty might help to determine the appropriateness of preventive interventions that require years for benefit, helping physicians to individualize goals of care for their geriatric patients or refer them for specialized assessment.40 Future studies will be aimed at examining clinical outcomes associated with identifying frailty using this dual-trait measure of gait speed with grip strength.
Limitations
Our study population was older than that studied by Fried and colleagues5 (100% vs 33% 75 years of age or older) and had a lower prevalence of frailty (6.5% vs 13%), which might indicate population differences beyond that of age; this is likely affected by slightly different cutoffs for the lowest quintiles for measuring some criteria. Compared with Fried and colleagues’ study5 there were some differences in the measurement of the frailty criteria; however, we do not expect these differences to affect the results greatly. Slight differences in the measurement of these frailty criteria are prevalent in the published literature.
Data from 133 out of 516 patients (25.8%) were removed from further analysis because they were missing 1 or more Fried criteria measures, usually exhaustion, as this was added during the course of the study to allow us to make comparisons with the Fried frailty criteria. Patients whose data were removed were significantly more likely to be female (P = .02) and older (P = .003 for women, P = .04 for men) than the patients who were left in the study. However, we believe the significance of our findings is unaffected because cutoffs for gait speed and grip strength were calculated based on the entire sample of 516 patients, so the determination of frailty markers is unaffected by removing part of the sample, particularly as the patients who were removed from the study tended to be missing exhaustion or weight-loss assessments; more than 84% of the 133 people who were removed had both gait-speed and grip-strength measurements.
During the study time frame of 13 months, 516 of the 1291 patients identified in this medical practice as being 75 years of age or older underwent screening. It is possible that this might have affected the frailty prevalence rate, as those who did not visit their family physicians during the study time period might have been either too frail to visit the clinic or too well to schedule a visit with their family physicians. Further, some of the screening items rely on self-reporting, and it is possible responses to these questions might have been affected by unrecognized cognitive impairment in this population.
Conclusion
Despite its limitations, this study contributes to knowledge about frailty measures that might be feasibly administered in primary care settings. Although single-trait measures of gait speed or grip strength alone were found to be sensitive and specific as proxies for the Fried frailty phenotype, use of gait speed with grip strength was found to be accurate, precise, specific, and more sensitive than other possible combinations. Future studies aimed at identifying frailty in seniors based on Fried frailty phenotype criteria might consider using gait speed with grip strength as a feasible means of identifying this condition within primary care settings.
Acknowledgments
This work was supported by the Centre for Family Medicine Family Health Team, the Ontario Ministry of Health and Long-Term Care and Ontario Medical Association Medically Complex Patient Demonstration Project, and the Canadian Mental Health Association Waterloo Wellington Dufferin. We thank Veronique Boscart and Niraj Maulkhan for their contributions to this study.
Notes
EDITOR’S KEY POINTS
Little attention has been given to the concept of frailty in primary care medicine relative to its importance. Many existing measures for assessing frailty are not feasible in the busy primary care environment. This study aimed to examine the relative accuracy of individual frailty markers to identify measures that might be useful for screening in the primary care setting.
The dual-trait measure of gait speed with grip strength is accurate, precise, specific, and more sensitive than individual traits and other possible dual-factor combinations; the measurement of gait speed and grip strength is feasible in primary care settings.
With the rapidly aging population, primary care physicians will be increasingly required to identify and manage frail seniors and their associated complex chronic conditions with judicious use of the limited available geriatric specialist resources.
POINTS DE REPÈRE DU RÉDACTEUR
Malgré son importance, le concept de fragilité n’a pas suscité beaucoup d’intérêt dans les milieux de soins primaires. Bon nombre des méthodes actuellement utilisées pour en évaluer l’incidence ne sont pas applicables dans le milieu trop exigeant des soins primaires. Cette étude visait à vérifier la précision relative de certains marqueurs de fragilité afin d’identifier ceux qui pourraient être utilisés dans un milieu de soins primaires.
La mesure combinée de la vitesse de marche et de la force de la poigne est précise, spécifique et plus sensible que tout critère isolé ou que toute autre combinaison de deux critères; elle peut être effectuée dans un contexte de soins primaires.
Étant donné le vieillissement rapide de la population, les médecins de première ligne devront de plus en plus identifier et soigner les problèmes de santé chroniques complexes de personnes âgées fragiles, et ce, en utilisant de façon judicieuse des ressources gériatriques limitées.
Footnotes
This article has been peer reviewed.
Cet article a fait l’objet d’une révision par des pairs.
Contributors
Drs Lee and Patel contributed to the study concept and design; data analysis, review, and interpretation; and final manuscript preparation and approval. Dr Costa contributed to the study concept and design, data analysis and interpretation, and final manuscript preparation and approval. Dr Bryce and Ms Hillier contributed to data analysis and interpretation, and final manuscript preparation and approval. Drs Slonim, Hunter, Heckman, and Molnar contributed to the study concept and design, data analysis and interpretation, and final manuscript preparation and approval.
Competing interests
None declared
- Copyright© the College of Family Physicians of Canada
References
In this issue
