See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/332078529

Quality of Life of Caregivers of Older Patients with Advanced Cancer

Article in Journal of the American Geriatrics Society · March 2019

DOI: 10.1111/jgs.15862

CITATIONS READS

73 331

20 authors, including:

Lee A Kehoe Huiwen Xu

University of Rochester Medical Center Emory University
26 PUBLICATIONS 200 CITATIONS 103 PUBLICATIONS 2,349 CITATIONS

SEE PROFILE SEE PROFILE

Kah Poh Loh Eva Culakova

University of Rochester University of Rochester Medical Center
363 PUBLICATIONS 5,063 CITATIONS 298 PUBLICATIONS 10,873 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Huiwen Xu on 02 April 2019.

The user has requested enhancement of the downloaded file.

 CLINICAL INVESTIGATION

Quality of Life of Caregivers of Older Patients with Advanced

Cancer
Lee A. Kehoe, MS, LMHC,* Huiwen Xu, MHA,*† Paul Duberstein, PhD,‡
Kah Poh Loh, MBBCh, BAO,* Eva Culakova, PhD,* Beverly Canin,§ Arti Hurria, MD,¶#
William Dale, MD, PhD,k Megan Wells, MPH,* Nikesha Gilmore, PhD,*
Amber S. Kleckner, PhD,* Jennifer Lund, PhD,k Charles Kamen, PhD,* Marie Flannery, PhD,*
Mike Hoerger, PhD, MSCR,** Judith O. Hopkins, MD,†† Jane Jijun Liu, MD,‡‡ Jodi Geer,§§
Ron Epstein, MD,* and Supriya G. Mohile, MD, MS*

See related editorial by Li-Wen Huang et al.

the number and type of patient GA impairments with

OBJECTIVES: To evaluate the relationships between aging- caregiver outcomes, controlling for patient and caregiver
related domains captured by geriatric assessment (GA) for covariates.
older patients with advanced cancer and caregivers’ emo- RESULTS: A total of 541 patients were enrolled, 414 with a
tional health and quality of life (QOL). caregiver. Almost half (43.5%) of the caregivers screened
DESIGN: In this cross sectional study of baseline data from a positive for distress, 24.4% for anxiety, and 18.9% for
nationwide investigation of older patients and their care- depression. Higher numbers of patient GA domain impair-
givers, patients completed a GA that included validated tests ments were associated with caregiver depression (adjusted
to evaluate eight domains of health (eg, function, cognition). odds ratio [aOR] = 1.29; P < .001], caregiver physical health
SETTING: Thirty-one community oncology practices through- on SF-12 (regression coefficient [β] = −1.24; P < .001), and
out the United States. overall caregiver QOL (β = −1.14; P < .01). Impaired patient
PARTICIPANTS: Enrolled patients were aged 70 and function was associated with lower caregiver QOL (β =
older, had one or more GA domain impaired, and had an −4.11; P < .001). Impaired patient nutrition was associated
incurable solid tumor malignancy or lymphoma. Each could with caregiver depression (aOR = 2.08; P < .01). Lower
choose one caregiver to enroll. caregiver age, caregiver comorbidity, and patient distress
were also associated with worse caregiver outcomes.
MEASUREMENTS: Caregivers completed the Generalized
Anxiety Disorder-7, Distress Thermometer, Patient Health CONCLUSION: Patient GA impairments were associated
Questionnaire-2 (depression), and Short Form Health Survey- with poorer emotional health and lower QOL of caregivers.
12 (SF-12 for QOL). Separate multivariate linear or logistic J Am Geriatr Soc 00:1–9, 2019.
regression models were used to examine the association of
Key words: caregivers; geriatric assessment; emotional
health; quality of life
From the *URCC NCORP Research Base, University of Rochester Medical
Center; †Department of Public Health Sciences, University of Rochester
School of Medicine, Rochester, New York; ‡Rutgers School of Public
Health, New Brunswick, New Jersey; §SCOREboard Advisory Group,
University of Rochester Medical Center, Rochester, New York; ¶City of
Hope National Medical Center, Duarte, California; kUniversity of North
Carolina, Chapel Hill, North Carolina; **Tulane University, New Orleans,
Louisiana; ††Southeast Clinical Oncology Research (SCOR), Consortium
NCI Community Oncology Research Program (NCORP), Winston Salem,
North Carolina; Novant Health-GWSM; ‡‡Heartland NCORP, Decatur,
T he number of caregivers of older adults with cancer is
on the rise.1 An informal caregiver was defined as a rel-
ative, partner, or friend who provides assistance across mul-
Illinois; and the §§Metro-Minnesota NCORP, St. Louis Park, Minnesota. tiple areas of functioning and living.2,3 Most older patients
Address correspondence to Supriya G. Mohile, MD, MS, Departments of with cancer live at home and depend on informal caregivers
Medicine and Surgery, University of Rochester Medical Center, for support with cancer treatment, symptom management,
601 Elmwood Avenue, Box 704, Rochester, NY 14642. and activities of daily living.4,5 Clinicians often focus on the
E-mail: supriya_mohile@urmc.rochester.edu health of the patients, whereas informal caregivers are sub-
#
Arti Hurria is now deceased. jected to a significant amount of stress that can adversely
DOI: 10.1111/jgs.15862 affect their own physical and emotional health.6–8

JAGS 00:1–9, 2019

© 2019 The American Geriatrics Society 0002-8614/18/$15.00
2 KEHOE ET AL. MONTH 2019–VOL. 00, NO. 00 JAGS

As the cancer progresses, the level of care burden increases understanding of the English language. Patients had at least
for the caregiver and can profoundly worsen caregivers’ quality one GA domain impairment excluding polypharmacy (due to
of life (QOL).9,10 The role of caregiving itself impacts the emo- the known high prevalence of polypharmacy in functionally
tional health of the caregivers; many studies demonstrate that fit patients); this eligibility criterion was designed to capture
caregivers experience even more emotional health challenges patients who are frailer than the fit older patients tradition-
(eg, anxiety, depression, distress) than the patients they are car- ally enrolled in clinical trials.23 If patients did not have
ing for.11–14 Furthermore, caregiver distress increases as the decision-making capacity, a healthcare proxy was required
patient with cancer declines functionally.15 to sign the consent. One caregiver was chosen by the patient
The geriatric assessment (GA) provides a framework that to enroll using this question: “Is there a family member, part-
can be incorporated into clinical care to improve decision mak- ner, friend, or caregiver (age 21 or older) with whom you dis-
ing and guide interventions for vulnerable older adults with cuss or who can be helpful in health-related matters?” It was
cancer.16 The GA assesses, with patient-reported and objective not required for a patient to have a caregiver to participate.
validated measures, aging-related domains known to influence Caregivers had to be 21 years of age or older, have an ade-
morbidity and mortality in older patients with cancer: func- quate understanding of English, and be able to provide
tion, physical performance, comorbidities, polypharmacy, cog- informed consent. This study was approved by the UR
nition, nutrition, psychological health, and social support.17 A Research Subjects Review Board and review boards of each
2015 Delphi consensus statement from geriatric oncology NCORP affiliate.
experts18 concluded that all of these GA domains are useful for
guiding nononcologic interventions and cancer treatment deci-
Study Procedures and Measures
sions. Eliciting support from caregivers is often a GA-guided
recommendation for older patients with cancer.18 Surveys were used to obtain sociodemographic characteristics
Although previous studies demonstrated that caregiving of each participant and caregiver and to assess their health.
for patients with cancer is burdensome,3,7,19–21 no large study Clinical information was collected by research staff. At base-
has evaluated if impaired GA domains in older patients with line, patients completed a GA consisting of validated mea-
advanced cancer are associated with caregivers’ emotional sures to evaluate the health of older adults in eight domains:
health and QOL in a national cohort. In this analysis of base- physical performance, functional status, comorbidity, cogni-
line data from a large multicenter study that enrolled patients tion, nutrition, social support, polypharmacy, and psycholog-
aged 70 and older with advanced cancer who had at least one ical status.18 If a patient met a cutoff score for a measure, they
impaired GA domain, we describe the characteristics of study were considered impaired in that domain (Supplementary
patients with a caregiver and evaluate the relationships Table S1, Table 1). At baseline, caregivers completed multiple
between impaired GA domains of the patients with the emo- validated measures of emotional health and QOL including
tional health and QOL of their caregivers. Our primary the two-item Patient Health Questionnaire (PHQ-2), General-
hypothesis was that a higher number of impaired GA ized Anxiety Disorder 7-Item Scale (GAD-7), Distress Ther-
domains would be associated with poorer caregiver emo- mometer, and Short Form Health Survey-12 (SF-12). A score
tional health and QOL. These results will inform clinical of 2 on the PHQ-2 suggested depression; a score of 5 or higher
practice and the development of interventions designed to on the GAD-7 suggested anxiety.24–27 Distress was measured
improve the QOL of both frail older patients with advanced for both patients and caregivers using a distress thermometer
cancer and their caregivers. with a score of 4 or higher (0-10) suggesting at least moderate
distress.28 QOL was captured with total SF-12 score and SF-
12 subscales that capture mental and physical health; SF-12
METHODS
scores and subscales range from 0 to 100, with higher scores
indicating better QOL, mental health, and physical health.29
Study Design
Our primary independent variables were number of
This cross-sectional study used baseline data from older impaired patient GA domains and specific GA domain impair-
patients with advanced cancer and their caregivers from ments. The number of GA domain impairments was the sum
31 community oncology practice clusters enrolled in the of all GA domains that were impaired (range = 1-8).23
Improving Communication in Older Cancer Patients and Their
Caregivers (COACH) study (clinicaltrials.gov NCT02107443;
Statistical Analysis
URCC13070) conducted through the University of Roches-
ter (UR) National Cancer Institute (NCI) Community Oncol- Descriptive statistics were used to examined demographics,
ogy Research Program (NCORP) Research Base between GA impairments, and clinical information. Bivariate ana-
October 2014 and April 2017. COACH is a cluster random- lyses compared characteristics of patients enrolled with a
ized trial to evaluate if a GA summary plus GA-guided rec- caregiver with those patients without one. Bivariate analyses
ommendations improve communication between older were also used to select significant caregiver and patient cov-
patients with cancer, their oncologists, and their caregivers ariates, based on P < .1, to enter a stepwise regression
about age-related concerns.22 model. The final multivariate models included information
from patient and caregiver dyads. These models included
covariates with P < .1 from stepwise procedures in addition
Study Participants
to caregiver age, sex, race, and patient cancer type. Multi-
Patients were eligible if they were diagnosed with an advanced variate logistic regressions and linear regressions were per-
solid tumor or lymphoma, were considering or currently formed for binary outcomes (depression, anxiety, and
receiving any type of cancer treatment, and had an adequate distress) and continuous outcomes (physical health, mental
JAGS MONTH 2019–VOL. 00, NO. 00 GERIATRIC ASSESSMENT AND CAREGIVER HEALTH 3

Table 1. Demographics, Clinical Characteristics, and Geriatric Assessment Impairments of Patients with Caregiversa

Patients with Patients with

caregivers (n = 414) caregivers (n = 414)

Variables N (%) Variables N (%)

Age, mean (SD), y 76.8 (5.4) Physical performance 389 (94.0)

70-79 299 (72.4%) TUG > 13.5 s 161 (39.0)
80-89 103 (24.9%) SPPB ≤ 9 points 325 (78.5)
≥90 11 (2.7%) Falls history ≥ 1 in previous 6 mo 107 (25.8)
Sex OARS Physical Health ≥ 1 316 (76.3)
Female 176 (42.6%) Functional status 254 (61.4)
Male 237 (57.4%) ADL ≥ 1 115 (27.8)
Race/Ethnicity IADLs ≥ 1 243 (58.7)
Non-Hispanic white 371 (89.8%)
African American 30 (7.3%)
Others 12 (2.9%) Comorbidity
Education OARS Comorbidity ≥ 3 or ≥ 1 263 (63.5)
<High school 57 (13.8%)
High school graduate 142 (34.4%)
Some college or above 214 (51.8%)
Income Cognition 144 (34.8)
≤$50 000 193 (46.8%) BOMC ≥ 11 or 12 (2.9)
>$50 000 219 (53.2%) Mini-Cog 0 words recalled or 1-2 words
Living arrangements recalled and abnormal clock 144 (34.8)
Independent living 172 (41.7%)
(more than one story) Nutrition 259 (62.6)
Independent living 223 (54.1%) BMI < 21.0 kg/m2 or 45 (10.9)
(one story)
Others 17 (4.1%) Weight loss > 10% in the past 6 mo or 62 (15.0
Cancer type MNA ≤ 11 points 248 (59.9)
Gastrointestinal 103 (24.9%) Social support
Lung 109 (26.4%) OARS Medical Social Support ≥ 1 91 (22.0)
Other 201 (48.7%)
Cancer stage
Stage III 35 (8.5%)
Stage IV 365 (88.4%) Polypharmacy 350 (84.5)
Others 13 (3.1%) Polypharmacy Log ≥ 5 regularly
Cancer treatment scheduled prescription or medications or
Any treatment (≥1) 404 (97.8%) Polypharmacy high-risk drug review
Multiple treatments (≥2) 136 (32.9%) ≥1 high-risk medication or
Chemotherapy 282 (68.3%) Labs Creatinine clearance or GFR <60
Monoclonal antibodies 102 (24.7%) mL/min
Hormonal treatment 66 (16.0%) Psychological status 112 (27.1)
Orally administered cancer treatment 73 (17.7%) GAD-7 ≥ 10 points 39 (9.4)
Radiation therapy 40 (9.8%) GDS ≥ 5 points 100 (24.2)

Abbreviations: ADLs, activities of daily living; BMI, body mass index; BOMC, Blessed Orientation Memory Concentration test; GAD-7, General Anxiety
Disorder 7-item scale; GDS = Geriatric Depression Scale; GFR, glomerular filtration rate; IADLs, instrumental activities of daily living; MNA, mini nutri-
tional assessment; OARS, Older Americans Resources and Services; SPPB, Short Physical Performance Battery; TUG, timed up-and-go.
a
There were some missing data (no more than six missing data points for any question); percentages and statistics are calculated from available data.

health, and total score on SF-12), respectively. In the models RESULTS

evaluating the number of patient GA domain impairments
as primary variable of interest, the number was included as Patient Demographics
a continuous variable. For the models evaluating specific GA
domains, each domain was included if associated with the In total, 414 eligible older patients with advanced cancer
outcome at P < .1. Likelihood ratio tests from linear or gen- who were enrolled with a caregiver were included in this
eralized mixed models with practice oncology site as random analysis. On average, the patients were 76.8 years of age
effects were not statistically significant (all P > .1), suggest- (standard deviation [SD] = 5.4; range = 70-96 y). Most of
ing a weak clustering effect of practice site; therefore, the the cohort was non-Hispanic white (89.8%) and had stage
results from the original multivariate models were presented. IV cancer (88.4%). The mean number of GA domain
Two-sided P < .05 were considered statistically significant. impairments for the sample was 4.48 (SD = 1.53); 89.6%
All analyses were conducted with SAS v.9.4 (SAS Institute had three or more domains impaired (Supplementary Fig-
Inc., Cary, NC) and Stata v.13.0 (StataCorp LLC, College ure S1). More than 80% of the patients had polypharmacy,
Station, TX). and nearly all the patients had physical performance
4 KEHOE ET AL. MONTH 2019–VOL. 00, NO. 00 JAGS

problems (94.0%) (Table 1). Just over one-third (34.8%) of

Table 2. Caregiver Demographics and Clinical
patients had an abnormal screen for cognitive impairment, Characteristicsa
63.5% had significant comorbidities, and 27.1% had a pos-
itive screen for depression or anxiety. N = 414

Variables N (%)

Age, mean (SD), y 66.5 (12.5)

Caregiver Demographics <70 210 (51.1)
The average caregiver age was 66 years (range = 26-92 y); 70-79 151 (36.7)
≥80 50 (12.2)
48.9% of caregivers were aged 70 and older (Table 2). Most
Sex
of the caregivers were female (75.4%), non-Hispanic white Female 310 (75.4)
(89.8%), and the patient’s spouse or cohabiting partner Male 101 (24.6)
(67.2%). Close to 40% of caregivers had significant comor- Education
bidities of their own; 43.5% reported moderate to high dis- Less than high school 30 (7.3)
tress, 18.9% reported depressive symptoms, and 24.4% were High school graduate 118 (28.7)
anxious. Mean SF-12 scores were 98.0 (SD = 14.2); 46.9 Some college or above 263 (64.0)
(SD = 10.5) for the physical health subscale and 51.1 Race
Non-Hispanic white 369 (89.8)
(SD = 9.8) for the mental health subscale.
African American 27 (6.6)
Other 15 (3.6)
Relationship
Spouse/Cohabiting partner 276 (67.2)
Multivariable Analyses Son/Daughter 94 (22.9)
Other 41 (10.0)
Several caregiver characteristics were associated with care- Income, annual
giver outcomes (Table 3, Table 4). Increasing caregiver age <$50 000 151 (36.8)
was associated with less anxiety and depression, as well as >$50 000 259 (63.2)
better SF-12 mental health but poorer SF-12 physical health. Living arrangements
Being female was associated with less distress (adjusted odds Independent living (more than one story) 188 (45.9)
ratio [aOR] = .43; 95% confidence interval [CI] = .25-.74; Independent living (one story) 215 (52.4)
P < .01). An income higher than $50 000/year was associated Other 7 (1.7)
Comorbidityb
with a higher SF-12 physical subscale and total scores. In the
Yes 162 (39.4)
models evaluating the number of GA domains, caregiver No 249 (60.6)
comorbidities were associated with caregiver anxiety (aOR = Anxiety (GAD-7) (≥5)
2.94; 95% CI =1.70-5.09; P < .001), depression (aOR = Yes 97 (24.4)
3.13; 95% CI = 1.74-5.60; P < .001), poorer SF-12 physical No 300 (75.6)
health (regression coefficient [β] = −8.11; 95% CI = −10.09 Distress (≥4)
to −6.13; P < .001), poorer SF-12 mental health (β = 3.99; Yes 177 (43.5)
95% CI = −6.00 to −1.97; P < .001), and poor overall QOL No 230 (56.5)
Depression (PHQ-2) (≥2)
(β = −11.86; 95% CI = −14.57 to −9.16); P < .001).
Yes 75 (18.9)
In the models evaluating the number of GA domains, No 322 (81.1)
patient distress was associated with caregiver anxiety (aOR =
2.07; 95% CI = 1.22-3.52; P < .01), caregiver distress Abbreviations: GAD-7, Generalized Anxiety Disorder 7-item scale; PHQ,
(aOR = 2.79; 95% CI) = 1.76-4.44; P < .01), caregiver men- Patient Health Questionnaire; SD, standard deviation.
tal health on SF-12 (aOR = −2.62; 95% CI = −4.70 to a
Missing data ≥3% for any variable; percentages are calculated from avail-
−0.54; P < .05), and overall QOL on SF-12 (β = −3.51; 95% b
able data.
CI) = −6.28 to −.74; P < .05) Defined using the Older American Resources and Services Comorbidity Form
that assesses the presence of 13 illnesses and how much each problem inter-
Our primary independent variables of interest were the
feres with his or her function; caregiver was noted to have the domain
number of GA domain impairments for the patient and impaired if she or he answered “yes” to three illnesses or answered that one
individual domain impairments (Tables 3 and 4; Figure 1.). illness interferes “a great deal.”
In the multivariate analysis, the number of patient GA
domain impairments was associated with caregiver depres-
DISCUSSION
sion (aOR = 1.29; 95% CI = 1.07-1.55; P < .001), lower
caregiver physical health (β = −1.24; 95% CI = −1.85 to In this large cohort of older patients with advanced cancer and
−.63; P < .001), and lower caregiver QOL (β = −1.14; their caregivers, patient GA measures were associated with
95% CI = −2.01 to −.27; P < .01). In separate models for emotional health and QOL of informal caregivers. Specifically,
individual GA domains, impaired patient functional status a higher number of patient GA impairments was associated
was associated with significantly worse caregiver physical with caregiver depression and lower caregiver QOL.
health (β = −2.55; 95% CI = −4.45 to −.56; P < .05) and Informal caregivers provide essential support for older
overall QOL (β = −4.11; 95% CI = −6.73 to −1.48; patients with advanced cancer receiving treatment including
P < .001). Impaired patient nutrition was significantly asso- assisting with activities of daily living, performing medical-
ciated with caregiver depression (aOR = 2.08; 95% CI = and nursing-related tasks, and providing direct physical and
1.15-3.77; p < .01). emotional assistance.30 Our descriptive results are similar to
Table 3. Association of Caregiver/Patient Predictors with Caregiver Emotional Health and Quality-of-Life Outcomes in Models with Number of Geriatric
JAGS

Assessment Domains Impaired

Anxiety (GAD-7) Distress Depression (PHQ-2) SF-12 Physical health SF-12 Mental health SF-12 Total score

Independent variables OR (95% CI) OR (95% CI) OR (95% CI) Coef. (95% CI) Coef. (95% CI) Coef. (95% CI)

Caregiver predictors
Age .96 (.94-.99)* .98 (.96-1.00)a .97 (.94-.99)* −.13 (−.21 to −.05)* .23 (.14-.31)** .08 (−.4 to .19)
Female (vs male) .93 (.50-1.73) .43 (.25-.74)* .72 (.37-1.42) 2.14 (.06-4.34) .30 (−1.95 to 2.54) 2.15 (−.84 to 5.14)
Non-Hispanic white (vs other) 1.17 (.51-2.67) .90 (.45-1.84) .8 (.36-1.79) .12 (−2.93 to 3.16) −1.78 (−4.90 to 1.34) −1.59 (−5.74 to 2.56)
Income > $50 000 + .62 (.37-1.03) ... ... 2.40 (.48-4.32)a ... 4.03 (1.39-6.66)*
“Decline to answer” (vs < $50 000)
Comorbidity 2.94 (1.69-5.09)** ... 3.13 (1.74-5.60)** −8.11 (−10.09to −6.13)** −3.99 (−6.00 to −1.97)** −11.86 (−14.57 to −9.16)**
Patient predictors
MONTH 2019–VOL. 00, NO. 00

Cancer type (vs GI)

Lung 1.22 (.62-2.39) 1.04 (.58-1.88) .84 (.41-1.71) −1.16 (−3.72 to 1.41) .13 (−2.51 to 2.77) −1.03 (−4.54 to 2.47)
Others .84 (.45-1.56) .57 (.34-.97)a .68 (.36-1.27) −2.48 (−4.73 to −.24)a .64 (−1.66 to 2.94) −1.85 (−4.91 to –1.21)
Distress (≥4) 2.07 (1.22-3.52)* 2.79 (1.76-4.44)** ... ... −2.62 (−4.70 to −.54)a −3.51 (−6.28 to −.74)a
Number of GA domains impaired .98 (.82-1.16) 1.11 (.96-1.29) 1.29 (1.07-1.55)** −1.24 (−1.85 to −.63)** .00 (−.66 to .65) −1.14 (−2.01 to .27)*

Note: Lower scores for the GAD-7, distress thermometer, and PHQ-2 indicate lower anxiety, distress, and depression, respectively; while higher scores on the SF-12 indicate greater quality of life.
a
P < .5, *P < .1, **P < .001.
Predictors with empty cells were excluded in the stepwise procedure.
Abbreviations: CI, confidence interval; Coef, coefficient; GA, geriatric assessment; GAD-7, Generalized Anxiety Disorder 7-item scale; GI, gastrointestinal; OR, odds ratio; PHQ, Patient Health Questionnaire; SF-12,
12-item Short Form Health Survey.

Table 4. Association of Caregiver/Patient Predictors with Caregiver Emotional Health and Quality of Life Outcomes in Models with Individual Geriatric
Assessment Domains
Anxiety (GAD-7) Distress Depression (PHQ-2) SF-12 Physical health SF-12 Mental health SF-12 Total score

Independent variables OR (95% CI) OR (95% CI) OR (95% CI) Coef. (95% CI) Coef. (95% CI) Coef. (95% CI)

Caregiver predictors
Age .96 (.94-.99)* .98 (.96-1.00)* .96 (.94-.98)** −.13 (−.21 to −.05)* .23 (.14-.31)** .07 (−.4 to .18)
Female (vs male) .92 (.50-1.71) .44 (.26-.75)* .74 (.38-1.46) 2.43 (.23-4.64)a .30 (−1.93 to 2.52) 2.35 (−.64 to 5.34)
Non-Hispanic white (vs other) 1.17 (.51-2.67) .86 (.42-1.75) .72 (.32-1.60) .15 (−2.88 to 3.19) −1.78 (−4.89 to 1.33) −1.50 (−5.62 to 2.63)
Income > $50 000 + .61 (.37-1.03) ... ... 2.41 (.50-4.33)a ... 4.27 (1.65-6.89)*
“Decline to answer” (vs < $50 000)
Comorbidity 2.91 (1.69-5.04)** ... 3.46 (1.94-6.18)** −7.86 (−9.85- -5.87)** −3.99 (−5.99 to −1.98)** −11.56 (−14.28 to −8.85)**
Patient predictors
Cancer type (vs GI)
Lung 1.22 (.62-2.39) 1.06 (.59-1.92) .86 (.42-1.74) −1.30 (−3.85 to 1.26)* .13 (−2.50 to 2.76) −1.02 (−4.51 to 2.47)
Others .84 (.45-1.56) .60 (.35-1.01)a .71 (.38-1.35) −2.47 (−4.71 to −.24)a 0.64 (−1.66-2.94) −1.85 (−4.90-1.19)
Distress (≥4) 2.03 (1.22-3.38)* 2.95 (1.89-4.61)** ... ... −2.62 (−4.61 to −.64)* −3.97 (−6.64 to −1.31)*
Impaired function ... ... ... −2.50 (−4.45 to −.56)a ... −4.11 (−6.73 to −1.48)**
Impaired nutrition ... 1.45 (.92-2.26) 2.08 (1.15-3.77)* ... ... ...
GERIATRIC ASSESSMENT AND CAREGIVER HEALTH

Note: Lower scores for the distress thermometer and PHQ-2 indicate distress and depression, respectively, whereas higher scores on the SF-12 indicate greater quality of life.
a
P < .05, *P < .01, **P < .001.
Predictors with empty cells were excluded in the stepwise procedure.
5

Abbreviations: Coef, coefficient; GA, geriatric assessment; GAD-7, Generalized Anxiety Disorder 7-item scale; GI, gastrointestinal; OR, odds ratio; PHQ, Patient Health Questionnaire; SF-12, 12-item Short Form Health Survey.
6 KEHOE ET AL. MONTH 2019–VOL. 00, NO. 00 JAGS

Figure 1. Association between Number of Impaired Patient Geriatric Assessment Domains and Caregiver Outcomes. Note: Besides care-
giver age, sex, race, and patient cancer type, the following covariates were also included in the multivariate models if they had a P value
<.1 in the stepwise models: caregiver education, family income, living arrangement, comorbidity, distress; patient cancer treatments.

previous studies.7,31,32 Hsu et al33 found that in 100 patients the Zarit Burden Interview, a measure that captures physical
aged 65 years and older (70% with advanced cancer) and their and mental health constructs in the context of caregiving.
caregivers, caregivers were mostly female (73%) and spouses SF-12, used in this study, captures physical and mental health
(68%); 79% lived with the patient. Jones et al32 found that in more globally and does not ask about these constructs in the
76 caregivers of older patients with cancer, 19.1% and 23.6% context of caregiving. Caregivers may self-report global QOL
reported moderate or greater anxiety and depression, respec- deficits, without communicating caregiver burden. Other
tively. In this study, lower caregiver age was associated with potential reasons for differences in outcomes between studies
higher prevalence of emotional health issues (ie, anxiety, could be related to patient sample; our sample included only
depression), and caregiver comorbidities were adversely asso- patients with advanced cancer who had at least one GA
ciated with all caregiver outcomes except for distress. Clini- domain impairment, which is less robust than studies that
cians should consider caregiver comorbid conditions when also included patients undergoing curative intent therapy.
evaluating the caregiver’s emotional health and QOL. This study is the first to show the association between the
Our study adds to evidence supporting an interdependent number of GA impairments and caregiver health (specifically
relationship between patient and caregiver health. Patient dis- caregiver depression, poorer physical health, and poorer
tress is associated with caregiver distress.34 In a study of QOL) in older patients with advanced cancer. In this nation-
43 caregiver/patient dyads, caregivers of patients with depres- wide study, 89.6% had three or more GA domains impaired.
sion experienced greater emotional distress.35 In this study, we This number is likely high due to our eligibility criteria,
also showed that patient function is associated with caregiver although comparable with some studies enrolling “real-
outcomes. In the study by Hsu et al,32 caregivers reported that world” patients.17 The number of GA domain impairments
patients had poorer physical function and mental health than was independently associated with caregiver outcomes
the patients reported for themselves. In multivariate analysis, beyond other patient and caregiver clinical and demographic
those caring for patients who required more help with instru- factors. In addition to the number of GA domains, two spe-
mental activities of daily living were more likely to experience cific GA domains had strong independent associations: nutri-
high caregiver burden. Germain et al7 showed that in close to tion with caregiver depression and impaired functional status
100 older patients with cancer and their caregivers, older with poorer caregiver physical health and QOL. Previous
patient age, perceived burden by caregiver, and patient func- studies showed that patient function is associated with care-
tional status were associated with lower caregiver QOL. giver burden and QOL, and caregivers have expressed that
In contrast, Rajasekaran et al36 did not find an associa- nutritional concerns (eg, anorexia, cachexia) can affect their
tion between patient GA measures and caregiver burden using emotional health.13,37–39,54 These findings suggest that the
JAGS MONTH 2019–VOL. 00, NO. 00 GERIATRIC ASSESSMENT AND CAREGIVER HEALTH 7

clinical team should address caregiver needs especially when Additionally, to minimize the burden of this study on partici-
the patient’s GA shows a high number of domain impair- pants, only broad screening tools, as opposed to more refined
ments and/or when patients have significant nutritional, func- diagnostic tools, were used to assess caregiver burden, anxi-
tional, or mental health concerns. Audio recordings of clinical ety, depression, and QOL, which may lead to some error in
encounters between older patients, caregivers, and oncologists the measurement of these constructs. Although the relation-
showed that although caregivers are unlikely to bring up their ships between patient GA factors and caregiver outcomes are
own emotional and physical health needs, they do provide reasonably strong, we did not adjust for multiple compari-
clues when they bring up patients’ age-related concerns such sons51 The study’s results should be considered hypothesis
as medication, functional, and nutritional issues that increase generating and require validation in other cohorts.
their own distress.40,41 These conversations are opportunities In conclusion, this study indicates that caregivers for
for oncology teams to offer support for caregivers. older patients with advanced cancer are a vulnerable group.
A symposium of experts convened by the NCI and Caregivers are often older themselves, and their own
National Institute of Nursing Research in 2016 highlighted comorbidities are associated with poor emotional health
the need for developing and testing interventions for care- and QOL. Future studies should explore GA-guided inter-
givers.1 Other research reports have discussed the need to ventions that include not only the older patient with cancer
develop interventions to improve psychosocial care for but also their caregivers, as a dyadic or triadic (with the
older patients and their caregivers.42,43 In one large study, oncologist)52 approach to interventions. Given that poor
lack of formal training in medical/nursing skills was associ- caregiver emotional and self-rated health is associated with
ated with greater levels of caregiver burden.30 Skills training patient-perceived quality of care, interventions may not
is a potential area for interventions, but research on how only improve clinical outcomes but also patient and care-
best to provide training for caregivers (ie, the content, mode giver satisfaction with care delivery.53
of delivery, and timing) is needed.30 In another study,
unhealthy behaviors (ie, low physical activity, binge drink-
ing) were associated with worse emotion-focused coping of ACKNOWLEDGMENTS
caregivers; interventions that provide support for promot- We thank the patients and caregivers who participated in the
ing healthier behaviors for caregivers may improve their study as well as the research staff in the University of Roches-
emotional health.44 Early and integrated palliative care and ter NCI Community Oncology Research Base network. In
psychosocial interventions for both patients and caregivers addition, we want to acknowledge the patient and caregiver
were shown to improve outcomes, although more work on stakeholders in SCOREboard. This work would not have
dissemination and implementation is needed.45 been possible without them.
Although the American Society of Clinical Oncology17 Financial Disclosure: The work was funded through a
and other guidelines46,47 have recommended GA for older Patient-Centered Outcomes Research Institute (PCORI) Pro-
patients with cancer receiving chemotherapy, limited data are gram contract (4634), UG1 CA189961, R01 CA177592,
available on how GA can help guide interventions to improve and K24 AG056589. All statements in this report, including
QOL and emotional health in caregivers of older adults with its findings and conclusions, are solely those of the authors,
cancer. Given the aging of both patients with cancers and their do not necessarily represent the official views of the funding
caregivers, a GA-guided dyadic approach to interventions agencies, and do not necessarily represent the views of the
should be studied.48 Engaging both older patients and their Patient-Centered Outcomes Research Institute (PCORI), its
caregivers in the research process from design to dissemina- Board of Governors, or Methodology Committee.
tion of interventions may improve the successful implementa- Conflicts of Interest: Arti Hurria received research fund-
tion and integration of interventions for vulnerable caregivers ing from Celgene, Novartis, GSK, and she was a consultant
at high risk for poor emotional health and QOL. In a series of to Behringer Ingelheim Pharmaceuticals, Carevive, Sanofi,
focus groups with older adults and caregivers, Puts et al49,50 GTX, Pierian Biosciences, and MJH Healthcare Holdings,
found that the stakeholders were motivated to work with a LLC. The other authors have declared no conflicts of interest
research team, but there are logistical considerations (such as for this article.
accessibility of technology and transportation) that need to be Author Contributions: Study concept and design:
addressed to support engagement. Trevino et al held a 1-day Mohile, Hurria, Dale, Flannery, Duberstein, and Epstein.
conference with older patient and caregiver stakeholders and Acquisition of subjects/data: Xu, Wells, Gilmore, Mohile,
found that tailoring interventions for older adults and modify- Hopkins, Greer, and Liu. Analysis and interpretation of data:
ing institutional-level factors to allow for ease of implementa- All authors. Preparation of manuscript: All authors.
tion was important to them.43 Sponsor’s Role: The sponsor provided funding only and
Strengths of this study include its large sample size of had no other role in collecting data or its analysis and
older frail patients with advanced cancer and their caregivers. interpretation.
Limitations of this study involve the use of a cross-sectional
design that prevents determination of causation. Furthermore,
cross-sectional designs present limitations by using only one REFERENCES
time point to assess outcomes. Patients and caregivers were
enrolled in a clinical trial from community oncology clinics 1. Kent EE, Rowland JH, Northouse L, et al. Caring for caregivers and
that may not be as representative as a population-based sam- patients: research and clinical priorities for informal cancer caregiving. Can-
cer. 2016;122(13):1987-1995.
ple. There is also a potential sample bias because all partici- 2. Roth DL, Fredman L, Haley WE. Informal caregiving and its impact on
pants were required to have a GA impairment that could lend health: a reappraisal from population-based studies. Gerontologist. 2015;55
to more frail older adults being included in the analysis. (2):309-319.
8 KEHOE ET AL. MONTH 2019–VOL. 00, NO. 00 JAGS

3. Hsu T, Loscalzo M, Ramani R, et al. Factors associated with high burden in 29. Ware J Jr, Kosinski M, Keller SD. A 12-item short-form health survey: con-
caregivers of older adults with cancer. Cancer. 2014;120(18):2927-2935. struction of scales and preliminary tests of reliability and validity. Med Care.
4. Rha SY, Park Y, Song SK, Lee CE, Lee J. Caregiving burden and health- 1996;34(3):220-233.
promoting behaviors among the family caregivers of cancer patients. Eur J 30. Mollica MA, Litzelman K, Rowland JH, Kent EE. The role of medical/nursing
Oncol Nurs. 2015;19(2):174-181. skills training in caregiver confidence and burden: a CanCORS study. Cancer.
5. Wadhwa D, Burman D, Swami N, Rodin G, Lo C, Zimmermann C. Quality 2017;123(22):4481-4487.
of life and mental health in caregivers of outpatients with advanced cancer. 31. Jayani R, Hurria A. Caregivers of older adults with cancer. Semin Oncol
Psychooncology. 2013;22(2):403-410. Nurs. 2012;28(4):221-225.
6. Grande G, Rowland C, van den Berg B, Hanratty B. Psychological morbidity 32. Jones SB, Whitford HS, Bond MJ. Burden on informal caregivers of elderly
and general health among family caregivers during end-of-life cancer care: a cancer survivors: risk versus resilience. J Psychosoc Oncol. 2015;33(2):
retrospective census survey. Palliat Med. 2018;32(10):1605-1614. 178-198.
7. Germain V, Dabakuyo-Yonli TS, Marilier S, et al. Management of elderly 33. Hsu T, Loscalzo M, Ramani R, et al. Are disagreements in caregiver and
patients suffering from cancer: assessment of perceived burden and of quality patient assessment of patient health associated with increased caregiver bur-
of life of primary caregivers. J Geriatr Oncol. 2017;8(3):220-228. den in caregivers of older adults with cancer? Oncologist. 2017;22(11):1383-
8. Haley WE. Family caregivers of elderly patients with cancer: understanding and 1391.
minimizing the burden of care. J Support Oncol. 2003;1(4 Suppl 2):25-29. 34. Hodges LJ, Humphris GM, Macfarlane G. A meta-analytic investigation of
9. Given BA. Caregiver burden. In: Holland JC, Weisel TW, Nelson CJ, the relationship between the psychological distress of cancer patients and
Roth AJ, Alici Y, eds. Geriatric Psycho-Oncology: A Quick Reference on the their carers. Soc Sci Med. 2005;60(1):1-12.
Psychological Dimensions of Cancer Symptom Management. New York, 35. Tan JY, Molassiotis A, Lloyd-Williams M, Yorke J. Burden, emotional dis-
NY: Oxford University Press; 2015. tress and quality of life among informal caregivers of lung cancer patients:
10. Kim MH, Dunkle RE, Lehning AJ, Shen H-W, Feld S, Perone AK. Caregiver an exploratory study. Eur J Cancer Care. 2018;27(1):e12691. https://doi.org/
stressors and depressive symptoms among older husbands and wives in the 10.1111/ecc.12691.
United States. J Women Aging. 2017;29(6):494-504. 36. Rajasekaran T, Tan T, Ong WS, et al. Comprehensive geriatric assessment
11. Braun M, Mikulincer M, Rydall A, Walsh A, Rodin G. Hidden morbidity in (CGA) based risk factors for increased caregiver burden among elderly Asian
cancer: spouse caregivers. J Clin Oncol. 2007;25(30):4829-4834. patients with cancer. J Geriatr Oncol. 2016;7(3):211-218.
12. Revenson TA, Griva K, Luszczynska A, et al. Caregiving outcomes. In: 37. Rhondali W, Chisholm GB, Daneshmand M, et al. Association between
Revenson TA, Griva K, Luszczynska A, et al., eds. Caregiving in the Illness body image dissatisfaction and weight loss among patients with advanced
Context. London, UK: Palgrave Pivot; 2016:15-24. cancer and their caregivers: a preliminary report. J Pain Symptom Manage.
13. Mellon S, Northouse LL, Weiss LK. A population-based study of the quality 2013;45(6):1039-1049.
of life of cancer survivors and their family caregivers. Cancer Nurs. 2006;29 38. Paek MS, Nightingale CL, Tooze JA, Milliron BJ, Weaver KE, Sterba KR.
(2):120-131. Contextual and stress process factors associated with head and neck cancer
14. Siefert ML, Williams A-L, Dowd MF, Chappel-Aiken L, McCorkle R. The caregivers’ physical and psychological well-being. Eur J Cancer Care. 2018;
caregiving experience in a racially diverse sample of cancer family caregivers. 27(2):e12833.
Cancer Nurs. 2008;31(5):399-407.
39. Mislang AR, Di Donato S, Hubbard J, et al. Nutritional management
15. Northouse L, Williams AL, Given B, McCorkle R. Psychosocial care for
of older adults with gastrointestinal cancers: an International Society of
family caregivers of patients with cancer. J Clin Oncol. 2012;30(11):1227-
Geriatric Oncology (SIOG) review paper. J Geriatr Oncol. 2018;9(4):
1234.
382-392.
16. Magnuson A, Allore H, Cohen HJ, et al. Geriatric assessment with manage-
40. Lowenstein L, Mohile SG. Missed Opportunities to Discuss Patient and
ment in cancer care: current evidence and potential mechanisms for future
Caregiver Aging Concerns in Oncology Clinical Encounters. Annals of
research. J Geriatr Oncol. 2016;7(4):242-248.
Behavioral Medicine. 2015;49:S130-S130.
17. Mohile SG, Dale W, Somerfield MR, et al. Practical assessment and manage-
41. Ramsdale E, Lemelman T, Loh KP, et al. Geriatric assessment-driven poly-
ment of vulnerabilities in older patients receiving chemotherapy: ASCO
pharmacy discussions between oncologists, older patients, and their care-
guideline for geriatric oncology. J Clin Oncol. 2018;36(22):2326-2347.
givers. J Geriatr Oncol. 2018;9(5):534-539.
18. Mohile SG, Verlarde C, Hurria A, et al. Geriatric assessment-guided care
42. Given B, Given CW. Cancer treatment in older adults: implications for psy-
processes for older adults: a Delphi consensus of geriatric oncology experts.
chosocial research. J Am Geriatr Soc. 2009;57(Suppl 2):S283-S285.
J Natl Compr Canc Netw. 2015;13(9):1120-1130.
43. Trevino KM, Healy C, Martin P, et al. Improving implementation of psycholog-
19. Given BA, Given CW, Sherwood PR. Family and caregiver needs over the
ical interventions to older adult patients with cancer: convening older adults,
course of the cancer trajectory. J Support Oncol. 2012;10(2):57-64.
caregivers, providers, researchers. J Geriatr Oncol. 2018;9(5):423-429.
20. Kurtz ME, Kurtz JC, Given CW, Given B. Relationship of caregiver reactions
44. Litzelman K, Kent EE, Rowland JH. Interrelationships between health
and depression to cancer patients’ symptoms, functional states and
behaviors and coping strategies among informal caregivers of cancer survi-
depression--a longitudinal view. Soc Sci Med. 1995;40(6):837-846.
vors. Health Educ Behav. 2018;45(1):90-100.
21. Sherwood PR, Given CW, Given BA, von Eye A. Caregiver burden and
depressive symptoms: analysis of common outcomes in caregivers of elderly 45. Hoerger M, Cullen BD. Early integrated palliative care and reduced emo-
patients. J Aging Health. 2005;17(2):125-147. tional distress in cancer caregivers: reaching the “hidden patients.”. Oncolo-
22. Mohile SG, Epstein RM, Hurria A, et al. Improving communication with gist. 2017;22(12):1419-1420.
older patients with cancer using geriatric assessment (GA): a University of 46. Hurria A, Wildes T, Blair SL, et al. Senior adult oncology, version 2.2014:
Rochester NCI Community Oncology Research Program (NCORP) cluster clinical practice guidelines in oncology. J Natl Compr Canc Netw. 2014;12
randomized controlled trial (CRCT). J Clin Oncol. 2018;36(18_suppl): (1):82-126.
LBA10003-LBA10003. 47. Wildiers H, Heeren P, Puts M, et al. International Society of Geriatric Oncol-
23. Cohen HJ, Smith D, Sun CL, et al. Frailty as determined by a comprehensive ogy consensus on geriatric assessment in older patients with cancer. J Clin
geriatric assessment-derived deficit-accumulation index in older patients with Oncol. 2014;32(24):2595-2603.
cancer who receive chemotherapy. Cancer. 2016;122(24):3865-3872. 48. Mohile SG, Hurria A, Cohen HJ, et al. Improving the quality of survivorship
24. Jakobsson U. Using the 12-item short form health survey (SF-12) to measure for older adults with cancer. Cancer. 2016;122(16):2459-2568.
quality of life among older people. Aging Clin Exp Res. 2007;19(6):457-464. 49. Puts MTE, Sattar S, Fossat T, et al. The senior Toronto oncology panel
25. Kroenke K, Spitzer RL, Williams JBW. The patient health questionnaire-2: (STOP) study: research participation for older adults with cancer and care-
validity of a two-item depression screener. Med Care. 2003;41(11):1284- givers. J Natl Compr Canc Netw. 2017;15(10):1208-1215.
1292. 50. Puts MTE, Sattar S, Ghodraty-Jabloo V, et al. Patient engagement in
26. Spitzer RL, Kroenke K, Williams JBW, Löwe B. A brief measure for assessing research with older adults with cancer. J Geriatr Oncol. 2017;8(6):391-396.
generalized anxiety disorder: the GAD-7. Arch Intern Med. 2006;166(10): 51. Feise RJ. Do multiple outcome measures require p-value adjustment? BMC
1092-1097. Med Res Methodol. 2002;2:8.
27. Ransom S, Jacobsen PB, Booth-Jones M. Validation of the distress thermom- 52. Laidsaar-Powell R, Butow P, Charles C, et al. The TRIO framework: concep-
eter with bone marrow transplant patients. Psychooncology. 2006;15(7): tual insights into family caregiver involvement and influence throughout
604-612. cancer treatment decision-making. Patient Educ Couns. 2017;100(11):2035-
28. National Comprehensive Cancer Network. NCCN Distress Thermometer and 2046.
Problem List for Patients. 2016. https://www.nccn.org/patients/resources/ 53. Litzelman K, Kent EE, Mollica M, Rowland JH. How does caregiver well-
life_with_cancer/pdf/nccn_distress_thermometer.pdf. NCCN: Plymouth Meeting, being relate to perceived quality of care in patients with cancer? Exploring
PA. Accessed February 28, 2019. associations and pathways. J Clin Oncol. 2016;34(29):3554-3561.
JAGS MONTH 2019–VOL. 00, NO. 00 GERIATRIC ASSESSMENT AND CAREGIVER HEALTH 9

54. Monin J, Doyle M, Levy B, Schulz R, Fried T, Kershaw T. Spousal associa- Supplementary Figure S1: Percentage of patients who
tions between frailty and depressive symptoms: longitudinal findings from
exhibited any given number of Geriatric Assessment
the cardiovascular health study. J Am Geriatr Soc. 2016;64(4):824-830.
Domain Impairments (N = 414).
SUPPORTING INFORMATION Supplementary Table S1: Geriatric Assessment Domains,
Tools, Descriptions, and Definitions of Impairmenta
a
Additional Supporting Information may be found in the Impairment is considered present within each domain if there
online version of this article. is one impairment noted on at least one tool.

View publication stats