Association between symptom burden and physical function in older patients with cancer

Abstract

Objective:

To evaluate the independent association between symptom burden and physical function impairment in older adults with cancer.

Design:

Cross-sectional.

Setting:

Two university-based geriatric oncology clinics.

Participants:

Patients with cancer aged ≥65 years who underwent evaluation with geriatric assessment (GA).

Methods:

Symptom burden was measured as summary score of severity ratings (range 0-10) of ten commonly reported symptoms using a Clinical Symptom Inventory (CSI). Functional impairment was defined as presence of ≥1 Instrumental Activity of Daily Living (IADL) impairment, any significant physical activity limitation on Medical Outcomes Survey (MOS), ≥1 any fall in the previous six months, or a Short Physical Performance Battery (SPPB) score ≤ 9. Multivariate analysis evaluated the association between symptom burden and physical function impairment, adjusting for other clinical and sociodemographic variables.

Results:

From 2011-2015, 359 patients with cancer with median age of 81 years (range 65-95) consented. The mean CSI score was 23.2±20.5 with an observed range of 0-90. Patients in the highest quartile of symptom burden (N=91; CSI score: 52±13) had a higher prevalence of IADL impairment (91% vs. 51%), physical activity limitation (93% vs. 65%), falls (55% vs. 21%), and SPPB score ≤ 9 (92% vs. 69%) (all p-values <0.01) when compared to those in the bottom quartile (N=81; CSI score: 2±2). With each unit increase in CSI score, the odds of having IADL impairment, physical activity limitations, falls, and SPPB scores ≤ 9 increased by 4.8%, 4.4%, 2.9% and 2.5% respectively (p<0.05 for all results).

Conclusions:

In older patients with cancer, higher symptom burden is associated with functional impairment. Future studies are warranted to evaluate if improved symptom management can improve function in older cancer patients.

INTRODUCTION:

Most older patients with cancer experience a large number of physical and psychological symptoms as a result of the disease and its treatments.¹ Commonly reported symptoms include fatigue, pain, drowsiness, insomnia, dyspnea, and anorexia.^2,3 Compared to individuals without cancer, patients with cancer have a high prevalence of symptoms such as loss of sensation abnormality, pain, and fatigue which are significantly associated with worse physical function.⁴ Prior work has shown that in younger patients with cancer, high symptom burden can negatively affect physical function, quality of life, and survival.^5,6 Impaired physical function has been shown to predict worse outcomes such as higher mortality and lower quality of life in community-dwelling older adults⁷ as well as in individuals with cancer.^8,9

In oncology clinical trials the most frequently used outcome variables are progression-free and overall survival. However, older patients with cancer often prioritize preservation of physical function over survival,¹⁰ limiting the applicability of clinical trial data for use in treatment decision-making.¹¹ Additionally, oncology clinical trials tend to include a “fitter” population of patients and may not be generalizable to more frail older patients with cancer. Thus, physical function limitation should be evaluated as one of the outcomes in older patients with cancer while choosing appropriate treatment options.¹¹

Older patients constitute a significant group of patient population in oncology practices, and require special consideration for symptom assessment and management.¹² Despite this, the majority of studies on symptom burden focus on younger adults with cancer or enroll a more general group of older patients with cancer rather than those referred to a geriatric oncology clinics who may be frailer.^13–15 A better understanding of relationship between symptom burden and physical function may allow identifying older patients who may benefit from symptom management and interventions to prevent/postpone decline in physical function. Moreover, there is heterogeneity in the methods used to evaluate physical function impairment in patients with cancer, with most studies using patient-reported measures of physical function,^15,16 which may underestimate physical function limitations.¹⁷ Validated objective measures of physical function such as the Short Physical Performance Battery (SPPB) may further clarify the relationship between symptom burden and physical function in older adults with cancer. The purpose of this study is to examine the relationship between symptom burden and physical function in older adults with cancer seen in geriatric oncology clinics, evaluated through a variety of validated patient-reported and objective measures of physical function. We hypothesized that higher symptom burden is independently associated with loss of physical function. Establishing this relationship would inform multi-component clinical pathways and interventions for physical function in older patients with cancer who have burdensome symptoms.¹⁸

METHODS:

Setting and sample:

Our sample included patients aged 65 and over referred to the Specialized Oncology Care and Research in the Elderly (SOCARE) geriatric oncology clinics at two sites (University of Rochester and University of Chicago) between May 2011 and October 2015. The SOCARE clinics are multidisciplinary geriatric oncology consultative clinics where older patients with cancer are referred for GA, consultation, and care management. Medical oncologists referred patients to the geriatric oncology clinics who needed evaluation for starting new treatment or had age-related concern during on-going treatment. Further details on the clinic structure, care processes, services provided, and patient consent are available elsewhere.¹⁹ Human subject approval was obtained from the institutional review boards of both institutions.

Measures:

Independent Variable:

Symptom burden was measured using the Clinical Symptom Inventory (CSI).¹² The CSI assesses ten symptoms (pain at its worst, other pain, distress, memory problem, feeling drowsy, dyspnea, sleep problem, anorexia, dry mouth, feeling sad) rated on a 0-10 scale, with 0 indicating that the symptom is not present and 10 indicating the symptom is “as bad as can be imagined”.³ Total CSI score range is 0-100.

Outcome Variables:

The outcomes of interest were physical function measures captured in the GA including: Instrumental Activities of Daily Living (IADL) status measured by difficulty using a telephone, managing money, managing medications, grocery shopping, preparing meals, doing housework, and using transportation; ²⁰ Medical Outcomes Survey (MOS) physical activity (PA) survey eliciting significant limitations in performing ten activities such as lifting or carrying groceries, climbing several flights of stairs, and walking more than one mile;²¹ Short Physical Performance Battery (SPPB) as an objective measure of physical performance including lower-extremity physical function by performing 4-meter walk, repeated chair stands and a balance test; ⁹ and falls in the previous six months. Four separate measures of functional impairment were created based on the above instruments: 1 -any assistance required for IADL task, 2- any significant limitation on MOS PA survey, 3- a history of at least 1 fall in the previous 6 months, and 4 - SPPB score ≤9.²²

Demographic and clinical variables:

We collected data on age, race (white, non-white), education (<high school, high school, >high school), and marital status (married, widowed, single/separated/divorced). The clinical characteristics included depression (scores ≥5 on 15-item Geriatric Depression Scale)²³, non-cancer comorbidities (presence of at least one illness that affected respondents a “great deal”, or at least three illnesses that affected them “somewhat” on the modified Older Americans Resources and Services Questionnaire Physical Health subscale), cancer type (gastrointestinal, genitourinary, lung, and others) and cancer stage (limited, advanced, and unknown).

Statistical analysis:

Patient demographics, clinical characteristics, and functional measures were summarized using descriptive statistics as means (± standard deviation) or frequencies (%), as appropriate. Missing values of the study variables were tested using Little’s missing-at-random test.²⁴ Imputation was not undertaken for data missing-at-random (available case analysis). Bivariate analyses using t-test for continuous variables and chi-square tests for categorical variables were performed to test the associations between the predictors and covariates with each of the functional outcomes (i.e., IADL, MOS physical activity, falls and SPPB). Separate multivariate logistic regression models were used to assess the independent association between symptom burden score and each functional outcomes, after adjusting for covariates based on clinical relevance in literature review and significant (p <0.05) association in bivariate analyses. Model fit was evaluated using the Hosmer-Lemeshow goodness of fit test.²⁵ To determine whether greater symptom burden was associated with functional impairment, we also compared the presence of each functional impairment measure by top versus bottom quartile of the total CSI score. All statistical tests were two-tailed with p-values of <0.05 considered statistically significant and were performed using SAS 9.4 (SAS Institute Inc., Cary, NC, USA).

RESULTS:

Characteristics of the study sample (Table 1):

Table 1:

Patient characteristics (N=359)

	Frequency	Percentage
Gender
Female	163	45.40
Male	196	54.60
Race
White	265	74.44
Black/African American	77	21.94
Others	9	3.62
Marital status
Married	181	51.42
Widowed	107	30.40
Divorced/separated/single	64	18.18
Education
<High School	168	48
=High School	105	30
>High School	77	22
Depressiona
No	244	69.12
Yes	109	30.88
Comorbidityb
Absent	108	30.51
Present	246	69.49
Total number of comorbidities
0	30	8.48
1	41	11.55
2	54	15.18
3	57	15.92
4	55	15.46
5	43	12.11
6	39	10.99
7	20	5.69
8	11	3.17
9	2	0.66
10	2	0.66
Symptomsc
Feeling drowsy	220	63.77
Memory problem	215	62.14
Pain at its worst	191	55.52
Dry mouth	187	54.36
Disturbed Sleep	179	53.12
Distress	179	52.65
Dyspnea	177	52.06
Other pain	162	51.27
Feeling sad	175	50.87
Lack of Appetite	172	49.86
Cancer type
Lung	66	18.38
GI	92	25.63
GU	106	29.53
Others	95	26.46
Cancer stage
Limited	133	37.05
Advanced	171	47.63
Unknown	55	15.32
IADL limitationsd
No	112	31.55
Yes	243	68.45
Short Physical Performance Batterye
SPPB>9	72	21.62
SPPB≤9	261	78.38
Physical Activity limitationsf
No	77	22.19
Yes	270	77.81
Fallsg
No	218	62.11
Yes	133	37.89

SD - Standard Deviation, GI - Gastrointestinal, GU - Genitourinary

^aDepression was defined as present if patients scored ≥ 5 on the 15-item Geriatric Depression Scale.

^bComorbidity was defined as present if patients reported at least one illness that affected them by a “great deal”, or at least three illnesses that affected them by “somewhat” on the modified Older Americans Resources and Services Questionnaire Physical Health subscale.

^cSymptom frequency is the number of patients who reported the symptoms and percentage is calculated based on the number of patients who responded to the corresponding symptom items.

^dInstrumental Activities of Daily living (IADL) limitations was defined as patients reporting needing assistance with at least one of the IADLs (use of telephone, managing money, preparing meals, shopping for groceries, doing housework, and transportation)

^eShort Physical Performance battery based physical function limitation was defined as a score of 9 or less out of a total score of 12 on SPPB.

^fPhysical activity limitations was present if patients self-report of at least one activity that was “limited a lot” on the Medical Outcomes Study Short Form Questionnaire.

^gFalls in the preceding 6 months

Of the patients referred to the two SOCARE clinics during the study period, 359 patients had values for the pre-determined study variables. The average age was 81 years (range 65-95); 55% were male, 74% were white, 51% were married, and 22% had at least a high school education. Most had genitourinary (30%), gastrointestinal (26%) or lung (18%) cancers. Nearly half the study population had an advanced cancer stage. At the time of GA, 69% (215/310) had received prior surgery, 41% (130/320) received radiation, and 32% (99/312) had received chemotherapy. Approximately 70% had non-cancer comorbidities with most prevalent being high blood pressure (59%), arthritis (58%), and heart disease (28.5%). Feeling drowsy (64%), memory problem (62%), and pain (56%) were the most commonly reported symptoms; the mean (±SD) CSI score was 23.2±20.6 (range: 0-90)

Association between symptom burden and physical function

The study sample had significant physical impairment (Table 1). For clinical relevance, patients in the highest quartile of symptom burden (N=91; CSI score: 52±13) were compared to those in the bottom quartile (N=81; CSI score: 2±2). Figure 1A shows that those in the top quartile had a higher prevalence of IADL impairment (91% vs. 51%), physical activity limitation (93% vs. 65%), and falls (55% vs. 21%); and SPPB scores ≤ 9 (92% vs. 69%) (all p-values <0.01).

Figure 1 –

A. Physical function impairments by highest versus lowest quartile of symptom burden scores. *All test of proportions p-values <0.05. B. Multivariate analyses showing associations between symptom burden and physical function impairment measures. Covariates in each regression models: Age, Gender, Race, Marital Status, Education, Depression, Comorbidity, Cancer types, and Cancer Stage. *p-value < 0.05

In bivariate analyses, patients who reported higher symptom burden scores had a higher prevalence of impaired physical function. For example, patients who required assistance with IADLs reported higher symptom burden scores than those who were independent in IADLs (27.97 ± 21.47 vs. 12.89 ± 13.41, p<0.01) (Table 2). In multivariate analyses, after adjusting for age, gender, race, education, marital status, depressive symptoms (GDS score), non-cancer comorbidities, cancer type and stage, the associations between symptom burden and functional limitations persisted (Supplemental Table). As shown in Figure 1B, each unit increase in symptom burden score is associated with greater odds of IADL impairment (4.8%), physical activity limitations (4.4%), falls (2.9%), and SPPB scores ≤ 9 (2.5%) (p<0.05 for all results). Little’s missing-at-random test on the variables included in the multivariate analysis showed p-values >0.05, suggesting no statistically significant patterns in the missing data (Results not shown).

Table 2:

Symptom burden score for patients with and without physical function impairment

Functional Measures	Impaired		Not Impaired
	CSI Mean	CSI SD	CSI Mean	CSI SD	p-value
IADL	27.97	21.47	12.89	13.41	<0.001
SPPB≤9	25.28	21.29	14.37	15.01	<0.001
Physical Activity	26.36	20.98	12.58	13.11	<0.001
Falls	29.89	20.95	19.01	19.03	<0.001

IADL - Instrumental Activities of Daily Living, PA - Physical Activity, SPPB - Short Physical Performance Battery, CSI – Clinical Symptom Inventory

DISCUSSION:

In this sample of older adults with cancer, rates of physical function impairment ranged from 39% to 78%, depending on the physical function measure. Functional impairment was more prevalent when patients reported greater symptom burdens (top quartile) compared to those with less symptom burden (bottom quartile). After adjusting for covariates, a higher symptom burden was found to be associated with impairments in all four measures of physical function.

Physical functioning is critical for older adults with cancer to optimize independence.²⁶ Furthermore, functional impairment has been shown to predict worse outcomes such as lower survival and poor quality of life in older patients with cancer who are undergoing cancer treatment.⁹ Compared to older patients without cancer, those with cancer are at high risk of decline in functional status due to intensive treatments and comorbidities.²⁷ More recently, a geriatric oncology guideline from the American Society of Clinical Oncology (ASCO) recommends IADL evaluation and a question about falls for all older adults considering or receiving chemotherapy.²⁸ Our study findings of high prevalence of functional impairment, across both objective and subjective physical function measures, in a large sample of older patients with cancer referred for geriatric oncology consultation highlight the need to incorporate physical function assessment in routine care to help guide cancer treatment recommendations in this patient population.²⁹ Our findings further indicate that cancer type was not associated with physical function outcomes after controlling for covariates. While previous cross-sectional studies have shown higher physical function impairment in lung cancer patients versus other cancers³⁰, recent prospective studies have also suggested findings similar to our study such that function decline or trajectory are not associated with the type of cancer diagnosis, but may be associated with type of cancer treatments and new cancer diagnosis.^31,32 Future studies should investigate whether the association between symptoms burden and physical function is moderated by cancer type.

This study expands upon prior literature which has demonstrated an association between symptom burden and physical function in younger patients with cancer.¹⁵ In a longitudinal study of 93 patients (average age 55.4 years), Dodd et al. found that between two consecutive time periods in the study, cancer-related symptoms of pain and fatigue contributed to 10.7% and 7.3% respectively decline in Karnofsky scores.¹⁴ We also found an independent association of symptom burden with physical impairment after adjusting for covariates. Specific symptoms such as depressive symptoms have been shown to be associated with lower physical function in older patients with non-cancer chronic conditions such as osteoarthritis, heart failure and diabetes.^33–35 While we did not conduct a symptom cluster analysis, a higher prevalence of feeling drowsy, memory problems, and pain symptoms out of total 10 symptoms in the study sample may reflect priority cancer-related symptoms clusters associated with physical function that could be investigated as potential target for interventions in older patients with cancer. Although we cannot infer causation in our cross-sectional study, prior work using longitudinal study design in younger cancer population has established a strong basis for symptoms as causal factors in physical function impairment.^14–16 Moreover, older adults often have other comorbidities that may cause pre-existing impairments in strength and balance, rendering them vulnerable to further decline with cancer- and cancer treatment-related symptom burden.³⁶ Kurtz and colleagues evaluated 420 patients age 65-98 years with cancer and found that pre-diagnosis physical functioning, symptom severity, and days since surgery were significant predictors of subsequent deficit in physical function, measured by MOS SF-36 survey.³⁷ The majority of these studies used self-reported general measures of physical function assessment³⁸, our study adds to the literature by showing association between symptom burden and lower physical function measured using multiple methods, including an objective measure of physical performance, the SPPB.

Our study finding of association of symptom burden with physical function could help health care providers target efforts to monitor patients with higher symptom burden and consider physical function as effectiveness indicator of interventions targeting symptom burden. Kroenke et al. evaluated the impact of longitudinal symptom burden trajectories on patient-reported physical function outcomes in relatively young patients with cancer (age: 58.8 years, range- 23-96) enrolled in a clinical trial.¹⁶ They reported that reduction in symptom burden predicted improvement in functional status and decline in disability. This suggests that it may be possible to modify the impact of symptom burden on physical function outcomes with improvements in symptom control. Early supportive care focusing on comprehensive symptom control through lifestyle management may maximize physical function and improve quality of life of older patients with cancer.³⁹ Symptoms have also been shown to moderate the effect of cognitive behavioral theory guided self-efficacy enhancing intervention to improve the physical function of cancer patients.^40,41 Based on the study findings of association between higher symptom burden and greater functional limitation, oncologists caring for older cancer patients could target such interventions towards patients with different levels of symptom burden and increase the effectiveness of intervention on physical functioning. The study results have the potential to guide clinical care at the intersection between oncology, palliative care, and geriatrics. For example, geriatricians who best recognize functional impairment should also attend to symptoms that are burdensome since symptoms and functional impairment are associated with one another, in addition to the oncologists and palliative care physicians who have more limited training in evaluating and intervening on age-related conditions in older patients with cancer.¹⁸

This study has several strengths, including a large sample of older adults with cancer who underwent GA using standardized tools including subjective and objective measures of physical function at two different geriatric oncology clinics. The study also has limitations. The cross-sectional design precludes causal inference. Since the patients included in the study consists of those referred to the geriatric oncology clinics at the two sites, there is a potential of selection and referral bias arising from physician-, patient-, disease- and other related factors. It included patients at any time point of their cancer treatment, some of whom had cancer surgery, radiation, hormonal therapy or chemotherapy which can affect their symptom burden and functional status.¹⁰ A symptom inventory more specific to patients on cancer therapy may clarify which symptoms are associated with the most impairment and are highest priorities for intervention. Due to limited availability of cancer treatment information in our study population, the effect of treatment on outcome was not evaluated in multivariate models which may bias the estimates. Future studies could evaluate a larger cohort of older patients with cancer in different health care settings and include a longitudinal design with a broad array of risk factors for symptom burden and functional limitations.

In conclusion, in real-world sample of older patients with cancer referred for geriatric oncology evaluation, we observe a significant association of higher symptom burden with physical functional impairment. While caring for older patients with cancer, it is important for oncologists, geriatricians, and palliative care providers to consider targeting specific interventions towards patients with different levels of symptom burden to improve physical function. Our findings could inform the priorities for future research to identify causal mechanisms between symptom burden and physical function and evaluate if optimal symptom management can improve physical function in older patients with cancer.

Supplementary Material

Supplemental Table

Supplemental Table -

Multivariable regression models showing association between symptoms score and physical function measures

Impact Statement:

We certify that this work is novel clinical research. The potential impact of this research on clinical care includes the following: our study findings of high prevalence of functional impairment, in a relatively large sample of older patients with cancer highlight the need to incorporate physical function assessment in routine care to help guide treatment recommendations in this patient population. Moreover, the association of symptom burden with physical function could help health care providers target efforts to monitor patients with higher symptom burden and consider physical function as an outcome for interventions targeting symptom burden.