Racial Disparities in Longitudinal Frailty State Transitions Among Older Cancer Survivors

Abstract

Black older cancer patients are at high risk of frailty during cancer treatment, yet racial disparities in long-term frailty state transitions during the survivorship phase are understudied. We conducted a longitudinal analysis of older cancer survivors (≥65 years) using rounds 1–9 (2011–2019) of the National Health and Aging Trends Study. A multistate Markov model assessed one-year transitions across Fried frailty phenotypes states (robust, prefrail, frail) and death, stratified by self-reported race (Black vs. non-Black). Among 1,303 survivors, 55% were female, 24% Black, and 76% non-Black. Black survivors were less likely to remain robust (50% vs. 57%), more likely to stay frail (46% vs. 39%), more likely to die, and more likely to transition to poorer frailty states (e.g. robust to prefrail or frail) than non-Black survivors. Black older cancer survivors experience poorer frailty trajectories than non-Black counterparts, highlighting the need for targeted survivorship care to improve long-term outcomes.

The population of older cancer survivors is expanding, highlighting the need to identify survivors at risk of adverse health outcomes.¹ Frailty, an age-related syndrome of reduced physiologic reserve to maintain homeostasis, is an important marker of independence and well-being in older adults. In older cancer survivors, frailty is associated with mortality, morbidity, and diminished quality of life.^2,3 Despite its importance to patients, frailty is understudied in clinical cancer studies, which often focus on morbidity, mortality, and toxicity endpoints.

Frailty is dynamic and can improve or worsen over time.^4,5 Older adults with newly diagnosed cancer who are Black are more likely to experience a worsening in frailty during the acute period surrounding cancer diagnosis and treatment.^6,7 However, longitudinal population-based studies of racial disparities in frailty state transitions among cancer survivors are limited. Understanding these disparities is essential for identifying subgroups at risk of physical decline and designing equitable healthcare interventions.

Our study describes racial disparities in nine-year frailty state transitions among older cancer survivors. We hypothesized that Black older cancer survivors would experience poorer frailty state transitions than non-Black counterparts.

We used data from Rounds 1–9 (2011–2019) of the National Health and Aging Trends Study (NHATS), a longitudinal, nationally representative cohort of older Medicare beneficiaries aged 65 and older designed to study trends in disability, physical functioning, and aging in the U.S.^8,9 NHATS collects data through in-person interviews, self-report surveys, and performance- based assessments that cover a wide range of domains, including health conditions, functional abilities, cognitive status, living arrangements, social support, and healthcare utilization. We restricted our study population to older adults who reported a history of cancer during Round 1. We excluded older adults with missing cancer history (N=5) and those who reported a history of skin cancer (N=648).

Race and ethnicity were self-reported and categorized within NHATS as non-Hispanic White, non-Hispanic Black, Hispanic, and non-Hispanic Other. Due to sample size constraints, we grouped participants as Black or non-Black (combined other racial and ethnic groups), as our primary hypothesis was that Black participants would experience worse frailty transitions than their counterparts.^7,10 Given evidence suggesting that Hispanic older adults have a higher frailty prevalence than non-Hispanic White older adults, we conducted a sensitivity analysis restricting the comparison to non-Hispanic Black versus non-Hispanic White participants.¹¹

Frailty was assessed annually using the Fried Frailty Phenotype,^2,11 which describes frailty based on five symptoms: unintentional weight loss, weakness, exhaustion, slowness, and low physical activity. Those with zero symptoms are classified as robust, those with one or two as prefrail, and those with three to five as frail. Other covariates included demographics (age, gender, metropolitan vs. non-metropolitan residence), socioeconomic status (wealth, home ownership, educational attainment),^12,13 and medical history (falls, hospitalizations, mobility devices, sleep disorders, comorbidities). We accounted for missing data using multiple imputation with fully conditional specification.^14,15

We pooled all nine years of data to describe one-year transitions across the frailty categories (robust, prefrail, frail) and death using a multi-state Markov model. Our models relied on the Markov assumption that frailty state transitions only depended on the current frailty state. We estimated one-year transitional probabilities across frailty states in each multiply imputed dataset for Black and non-Black older cancer survivors separately, adjusting for age, gender, wealth, and educational attainment. Ninety-five percent confidence intervals were estimated using 1000 bootstrapped samples.^16,17 We used a likelihood ratio test to assess whether differences in transitional probabilities between Black and non-Black participants were statistically significant (α=0.05) by comparing a Markov model assuming equal transition probabilities across racial groups to a stratified model that allowed transitions to vary by race. We used the NHATS survey sampling weights to reweight the population to be representative of Medicare beneficiaries with cancer in 2011. Analyses were conducted using SAS Version 9.4 (SAS Institute Inc., Cary, NC) and results were visualized using Biorender.¹⁸

Among the 1,303 older cancer survivors, 70% were non-Hispanic White, 24% were non-Hispanic Black, 4% were Hispanic, and 3% were another non-Hispanic race or ethnicity (Table 1). Fifty-five were female and 34% were 65–74 years, 43% were 75–84 years, and 23% were 85+ years. Black cancer survivors had lower median wealth ($47,000 vs. $95,800), were less likely to own a home (62% vs. 73%) and were less likely to have a high school degree (64% vs. 79%) than non-Black cancer survivors. Hypertension (83% vs. 67%) and diabetes (37% vs. 22%) were more prevalent in Black cancer survivors compared to non-Black survivors.

Table 1.

Characteristics of NHATS participants with a history of cancer, stratified by Black and non-Black race, during the Round 1 (2011) interview, after multiple imputation with fully conditional specification.

Characteristics	Overall N=1303		Black N=306 (23%)		Non-Black N=997 (77%)
Demographics
Self-reported race and ethnic category, n (%)
Non-Hispanic White	910	(70)	0	(0)	910	(91)
Non-Hispanic Black	306	(24)	306	(100)	0	(0)
Hispanic	51	(4)	0	(0)	51	(5)
Non-Hispanic Other	36	(3)	0	(0)	36	(4)
Age group, n (%)
65–74	444	(34)	122	(40)	322	(32)
75–84	561	(43)	130	(43)	431	(43)
85+	298	(23)	54	(18)	244	(25)
Gender, n (%)
Male	584	(45)	166	(54)	418	(42)
Female	719	(55)	139	(46)	580	(58)
County of residence, n (%)
Metropolitan area	1,081	(83)	273	(89)	808	(81)
Non-metropolitan area	222	(17)	33	(11)	189	(19)
Socioeconomic status
Wealth, median (IQR)	$95,800	($3,750, $339,250)	$47,000	($300, $166,002)	$130,500	($12,000, $432,000)
Home ownership, n (%)	921	(71)	190	(62)	731	(73)
Educational attainment, n (%)
≤ High school degree or equivalent	323	(25)	111	(36)	212	(21)
≥ High school degree or equivalent	981	(75)	195	(64)	785	(79)
Medical History
Falls during past yr, n (%)	439	(34)	89	(29)	350	(35)
Hospital stay in past yr, n (%)	402	(31)	106	(35)	295	(30)
Used mobility devices in prior month, n (%)	434	(33)	113	(37)	321	(32)
Sleep disorder in prior month, n (%)	714	(55)	150	(49)	564	(57)
Comorbid conditions
Myocardial infarction, n (%)	216	(17)	57	(19)	159	(16)
Hypertension, n (%)	922	(71)	253	(83)	669	(67)
Arthritis, n(%)	745	(57)	167	(55)	577	(58)
Diabetes, n (%)	331	(25)	114	(37)	217	(22)
Depression symptoms, n (%)	221	(17)	66	(22)	155	(16)
Anxiety symptoms, n (%)	200	(15)	48	(16)	152	(15)

Abbreviations: IQR= Interquartile range

Figure 1 depicts one-year frailty transition probabilities for Black and non-Black survivors (confidence intervals are provided in Supplemental Table 1). Robust Black cancer survivors were less likely to remain robust (50% vs. 57%), more likely to transition to prefrailty (34% vs. 30%), and more likely to die (13% vs. 9%) than robust non-Black survivors. Prefrail Black cancer survivors were more likely to worsen to frailty than pre-frail non-Black cancer survivors (19% vs. 14%), although mortality was similar. Among frail older cancer survivors, transitional probabilities to better states (robust, prefrail) were similar across racial groups. Black frail older cancer survivors were less likely to die than non-Black older cancer survivors (26% vs. 33%). Differences in transitional probabilities were statistically significant (p<0.001). Findings were similar in the sensitivity analysis comparing frailty transitions between non-Hispanic Black and non-Hispanic White cancer survivors (Supplemental Table 2).

Figure 1.

One-year frailty state transitional probabilities among Black cancer survivors (A) and non-Black cancer survivors (B) over the age of 65 in the National Health and Aging Trends Study (2011–2019). The probability of remaining in the same frailty state is demonstrated within the circle. The arrow size is proportional to the magnitude of the transition probability. The proportion within the death circle is 100% since once older adults die, they remain in the death state for the remainder of follow-up and cannot transition to other frailty states.

This longitudinal study examined racial disparities in one-year frailty state transitions and survival for Black and non-Black older cancer survivors using nationally representative data. Black cancer survivors were less likely to remain robust, more likely to stay frail, and had a higher probability of transitioning to poorer frailty states (e.g. robust to prefrail, robust to death, prefrail to frail, and prefrail to death) than non-Black survivors. These findings may be explained by several underlying mechanisms. Black older cancer survivors have disproportionately lower socioeconomic status and less access to healthcare and survivorship care, largely due to systemic and interpersonal racism.^6,7 We found Black cancer survivors had lower median wealth, educational attainment, and home ownership than non-Black survivors, and a higher prevalence of comorbid conditions. Chronic inflammation due to the stressors of systemic racism may also lead to immune system decline and physiologic dysfunction.¹⁹

Surprisingly, we found that Black and non-Black survivors who were prefrail had similar mortality and Black survivors who were frail were less likely to die than non-Black frail survivors. This likely reflects heterogeneity in frailty and other characteristics by race in our sample, rather than a true survival benefit. For example, the Black cancer survivors in our study who were frail were younger than frail non-Black cancer survivors. If Black cancer survivors become prefrail or frail at younger ages, they may have lower all-cause mortality relative to their non-Black counterparts, despite experience poorer frailty transitions.

Other studies describing racial disparities in frailty have typically focused on frailty measured at cancer diagnosis or treatment initiation, with Black patients having a higher frailty prevalence, worse functional status, and poorer health-related quality of life than non-Black survivors.^6,20 Literature on racial disparities in longitudinal frailty state transitions is sparse. A recent longitudinal study using a nationwide cancer registry of older women with early-stage breast cancer found that Black patients were more likely to experience worsening frailty without signs of resilience than women of other races.⁷ Our findings add to the literature by describing racial disparities in frailty state transitions during the survivorship period for a broad population of survivors using annual objective frailty assessments. Future research should describe disparities in long-term frailty outcomes within specific tumor types.

A strength of our study is the use of NHATS, which oversampled Black participants to improve inference in this historically underrepresented group. However, the number of Black participants was smaller than the non-Black group, which impacted the precision of our estimates. We were unable to conduct subgroup analyses in other racial and ethnic groups due to sample size constraints. We grouped all non-Black participants in the primary analysis to avoid exclusion of other groups historically underrepresented in research, such as Hispanic participants; however, this likely masked important heterogeneity in frailty outcomes. To account for this, we conducted a subgroup analysis comparing outcomes between non-Hispanic Black and non-Hispanic White survivors, which yielded similar findings. NHATS began oversampling Hispanic participants in 2022, which will facilitate future research in this area.

While NHATS captures information on cancer type, these variables are not available in the public use files, preventing stratification by cancer type in our analysis. Other cancer-related information, such as timing of diagnosis, stage, and treatments, is not available in NHATS. We excluded older adults living in nursing homes during Round 1 as they did not participate in the primary NHATS data collection instrument, limiting the generalizability of our findings. Additionally, there was a large amount of missing data in key variables like frailty (15%) and wealth (42%). We addressed this with multiple imputation, but residual bias may remain.

This study uncovered important racial disparities in frailty state transitions for Black older cancer survivors. Future research is necessary to understand the mechanisms behind these disparities so that targeted interventions can be developed and implemented to improve equitable cancer survivorship care and outcomes.

Data Availability

This study used data from Rounds 1–9 (2011–2019) of the National Health and Aging Trends Study (NHATS). NHATS data are publicly available at https://nhats.org/