“Frail”—More Than a Descriptor for Patients Enrolled in the Program of All-Inclusive Care for the Elderly

Accompanying the growth of the older population worldwide is a surge in the number of those in need of long-term care (LTC) services and supports. In fact, the number of older adults using long-term services in any US setting is expected to increase from 8 million in 2000 to 19 million by 2050.¹

The Program for All-Inclusive Care for the Elderly (PACE) is a community-based LTC model serving high-risk older adults in the United States who meet state criteria for nursing home (NH) eligibility but prefer to live in the community for as long as they can.^2,3 PACE provides comprehensive medical care and support services through an interdisciplinary team with the goal of maintaining or improving functional independence in order to maximally prevent or delay institutionalization. PACE operates within a capitated payment structure and operates at full risk for all health-related expenses. As of August 2019, more than 50,000 participants were enrolled in 130 PACE programs in 31 states.⁴ Despite its success in reducing rates of hospitalization, NH admission, and overall health care expenditures compared with usual fee-for-service care models, studies of PACE’s performance and cost-effectiveness have shown considerable variability across PACE sites.^5-10 Varying rates of cost-effectiveness may at least partially result from a payment model based primarily on medical diagnosis, such as the US Centers for Medicare & Medicaid Services’ Hierarchical Condition Categories (HCC).¹¹ Studies have shown that diagnosis-based risk algorithms do not fully predict expenditures or health outcomes of older adults.¹² 

In the geriatrics literature, frailty is theorized to result from age-related decline in multiple physiological systems, leading to depletion of homeostatic reserves and increased vulnerability to adverse health outcomes after acute stressor events.¹³ Frailty manifests clinically as a medical syndrome, is more prevalent with increasing age, and confers elevated risk of adverse health outcomes, including mortality, institutionalization, falls, hospitalization, and increased caregiver burden.^14,15 Prior to 2000, studies generally treated frailty, comorbidity, and disability as synonyms to define a subset of vulnerable older adults who require higher levels of care.¹⁶ Recently, there has been consensus that these are interrelated but distinct clinical entities, supported by clinical observations that neither individual diseases nor disability alone is sufficient for frailty identification.¹⁷ 

The important questions of which older adults would benefit the most from participating in PACE and from receiving particular PACE interventions remain without clear answers. This study represents a first attempt to assess the feasibility of incorporating frailty assessment in PACE and to obtain preliminary data on the potential prognostic value of frailty for health care utilization among program participants. This pilot study also provides estimates needed to power a future study on a larger scale for confirmation. We hypothesize the predictive power afforded by frailty assessment extends beyond burden of comorbidity and disability. Understanding the predictors of health outcomes and expenditure in PACE participants, other than comorbidity and disability, is important for improving patient-specific outcomes as well as controlling health care expenditures.  

We performed a small pilot study at one PACE center in Baltimore, MD, to measure frailty and subsequently collected data over 2 years on health care utilization among those study participants. This study was approved by our institution’s internal board review committee. 

This convenience sample of 30 cognitively healthy participants was recruited from 147 PACE enrollees. Participants underwent history and physical exam followed by frailty assessment at the PACE medical offices in 2016. We used the physical frailty phenotype originally developed in the Cardiovascular Health Study and validated in the Women’s Health and Aging Studies.^13,18 The frailty phenotype includes five criteria: (1) weak grip strength; (2) slow walking speed; (3) exhaustion; (4) low physical activity; and (5) unintentional weight loss (Table 1). People with three or more indicators are classified as frail; those with one to two are deemed prefrail; and those with none are deemed non-frail. Frailty assessment took 15 minutes on average. Some of the data needed for this assessment, such as gender, body weight, and height, were readily available through the electronic medical record. We used negative binomial regression models to analyze associations between frailty and 2-year incidence rates of  emergency department (ED) visits, hospitalizations, ambulance use, and NH admissions after adjusting for age and HCC. HCC scores were included to assess the independent effects of frailty above and beyond disease burden. 

Of the 30 PACE participants in the sample, 83% were female; 70% were African American and 3%, 47%, and 50% were identified as non-frail (n=1), prefrail (n=14), and frail (n=15), respectively (Table 2). The non-frail participant was excluded from subsequent analysis due to small sample size. 

Frail patients were more likely to be older, African-American, and have greater mobility limitation (a proxy for disability) and disease burden as reflected by higher prevalence of slow walking speed and higher HCC score, respectively (Table 3).¹⁹ However, none of these differences reached statistical significance. All frail participants except one had mobility limitation. among those with mobility limitation, however, only 56% (n=14) qualified as frail (Table 3). In our study, frail participants had between 27% and 92% increased risk of several outcomes, after adjustment for age and HCC, but none of the estimates reached statistical significance (Table 4).

The number of incident hospitalization days over 2 years of follow up in participants categorized as frail was 1.5 (95% CI, 0.43-5.08) times that of the prefrail; however, the association was not statistically significant (P=.543). There was no association between frailty and incidence of ED visits (incidence rate ratio = 0.99; P=.977).

PACE enrollees are high utilizers of federal and state healthcare dollars, and a significant amount of those funds are spent on hospitalization.²⁰ Many studies of PACE have used the term frail elderly loosely. When used in PACE literature, the term is not based on formal frailty assessment but rather a characterization of generalized vulnerability with disability and dependency.²¹ Our pilot study is unique in its formal assessment of frailty for each participant using a validated tool. 

In this pilot, frailty was not found to be synonymous with high disease burden and disability proxy in PACE participants, evidenced by the fact that about half of the study participants were not frail despite a high prevalence of mobility limitation and disease burden. This study provides further support to the argument that frailty, disability, and chronic diseases are overlapping yet distinct concepts.¹⁶ In our pilot study, prefrail participants consistently trended toward lower levels of health care utilization than frail participants. Although the causality of the association remains to be seen, it can be postulated that if frailty risk is diagnosed early, and given that frailty is a dynamic condition characterized by frequent transitions between frailty states, multidomain interventions such as resistance training and nutritional support may reverse the course of frailty and its sequelae in older adults.^22-24 

If it is proven that frailty assessment is both feasible and value-added for the PACE-eligible population, PACE programs should consider performing frailty assessment on every participant. Assessment should occur at the time of PACE enrollment for early detection and subsequently during the semiannual health assessment to provide additional prognostic information. Our pilot findings suggest PACE may have the potential to enhance cost-effectiveness by targeting the needs of frail participants, providing targeted rehabilitation therapy (physical, occupation, speech, and recreational), nutrition counseling, dental care, medication reduction, and caregiver support. This assessment could potentially be extended to participants with cognitive impairment using sensor-based technology to allow more objective and sensitive measurement of frailty.²⁵ 

The study is not without limitations. Limitations include the pilot nature of the study, the small sample, a single PACE location, a lack of participants with cognitive impairment at the time of frailty assessment, and a lack of non-frail participants, which may limit the generalizability of this study.

However,  recent efforts by the authors and others to focus more on function in risk screening, care planning, and care delivery in PACE may generate new opportunities for the program to achieve its maximum potential.^14,15,26 As the PACE model considers expansion to serve more populations, mechanisms to determine eligibility and opportunities to improve outcomes could be valuable.²⁷ Incorporation of frailty assessment in PACE may provide an additional tool to help identify those most in need and allocate resources accordingly. 

Authors: Shaista U Ahmed, MD¹ • Matthew K McNabney, MD¹ • Jeremy S Barron, MD, MPH^1,2 • Qian-Li Xue, PhD^1,3,4

Affiliations: ¹Department of Medicine Division of Geriatrics and Gerontology, Johns Hopkins School of Medicine, Baltimore, MD; ²Herzog Medical Center, Department of Ventilatory Care, Jerusalem, Israel; ³Johns Hopkins Center on Aging and Health, Johns Hopkins Medical Institutions, Baltimore, MD; ⁴Departments of Biostatistics and Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD

Disclosures: This work was supported by a grant from NIH/NIA P30AG021334. The authors report no other relevant financial relationships.

Correspondence:

Shaista U Ahmed, MD
Clinical Associate Department of Medicine Division of Geriatrics and Gerontology
Johns Hopkins School of Medicine
MFL Building Center Tower, Suite 2200
5200 Eastern Avenue
Baltimore, MD, 21224

Email: sahmed49@jh.edu