Support Care Cancer

DOI 10.1007/s00520-014-2122-6

ORIGINAL ARTICLE

Quantifying the burden of informal caregiving for patients

with cancer in Europe
Amir Goren & Isabelle Gilloteau & Michael Lees &
Marco daCosta DiBonaventura

Received: 5 September 2013 / Accepted: 6 January 2014

# Springer-Verlag Berlin Heidelberg 2014

Abstract health utilities, 1.46 times as much work impairment, and

Purpose Informal caregivers for patients with cancer provide 1.97 times the odds of anxiety).
critical emotional and instrumental support, but this role can Conclusions Caregivers for patients with cancer experienced
cause substantial burden. This study expands our understand- significant impairments. These findings reinforce the need for
ing of cancer-related caregiving burden in Europe. enhancing our understanding of the caregiving experience and
Methods Caregivers (n=1,713) for patients with cancer and developing supportive and personalized multicomponent in-
non-caregivers (n=103,868) were identified through the 2010 terventions for caregivers, given their pivotal role in providing
and 2011 European Union National Health and Wellness support for patients.
Survey, administered via the Internet to adult populations in
France, Germany, Italy, Spain, and the United Kingdom. Keywords Caregiving . Health-related quality of life .
Respondents completed measures of sociodemographics and Cancer . Resource use . Productivity
health behaviors, health-related quality of life (using SF-12v2),
work productivity and activity impairment (using WPAI),
healthcare resource use (emergency room visits, hospitaliza- Introduction
tions, and traditional provider visits), and reported diagnosis of
stress-related comorbidities (depression, anxiety, insomnia, The burden of cancer worldwide is substantial and expected to
headache, migraine, and gastrointestinal problems). Two- increase significantly in the future [1–4]. The global incidence
sided tests of means or proportions compared caregivers against of cancer may rise more than 75 % by 2030, led by developing
non-caregivers. Multivariable regression models, comparing countries, according to recent research by the World Health
caregivers for patients with any cancer vs. non-caregivers on Organization’s International Agency for Cancer Research [5].
all health outcomes, adjusted for covariates (age, sex, college, This increase in incidence is due primarily to population
income, marital status, employment, body mass index, alcohol, growth and aging, and coupled with the implementation of
smoking, and Charlson comorbidity index). effective cancer treatments, it has led to an increase in the
Results Caregivers for patients with cancer vs. non-caregivers number of survivors [1, 4, 6–9]. The considerable economic
reported significant (P<0.05) impairment across all health and humanistic burden of cancer care is well recognized, yet
outcomes, even after adjusting for several confounds (e.g., estimates often neglect to account for the vital role played by
3.26-point lower mental health status, 0.043-point lower informal caregivers (typically family or friends) in providing
critical instrumental and emotional support for cancer patients
A. Goren (*) : M. daCosta DiBonaventura across the disease trajectory [10–13].
Health Outcomes Practice, Kantar Health, 11 Madison Ave, 12th Informal caregivers of patients with cancer, who provide
Floor, New York, NY 10010, USA
unpaid long-term voluntary care, are increasingly responsible
e-mail: amir.goren@kantarhealth.com
for the provision of support to patients due to healthcare
I. Gilloteau changes and the shift from inpatient to outpatient care [11].
Global HEOR, Bristol-Myers Squibb, Princeton, NJ, USA While this relieves the healthcare and welfare systems of the
cost of caring for these patients, evidence suggests that the
M. Lees
Global HEOR, Bristol-Myers Squibb, Rueil-Malmaison Cedex, provision of informal care can be associated with significant
France burden and encompasses multiple dimensions: impaired
 Support Care Cancer

quality of life (QoL), increased psychological distress, mor- annual, cross-sectional, Internet-based survey of self-
bidity and mortality risk, and other sources such as lost time reported healthcare attitudes and behaviors among adults
and productivity [14–28]. The financial impact of informal (≥18 years old). Respondents, invited and recruited from
care for any condition in the USA has been estimated at more online, opt-in consumer panels maintained by Lightspeed
than $300 billion in 2004 [14]. In cancer specifically, one Research (LSR), provided informed consent to participate,
estimate of the total annual cost associated with cancer care and received points that may be exchanged for prizes. LSR
in 2009 (including productivity loss, healthcare costs, and panel members are recruited via email, e-newsletters, online
informal caregiving) in Europe comes out to 117 billion euros banners, and co-registration with panel partners. Respondents
[15]. This cost is expected to grow as treatment advances shift are limited in the number of surveys they may complete per
care to the outpatient and home settings [16], with informal year. The NHWS surveys were approved by Essex IRB
caregiving expected to contribute significantly and increas- (Lebanon, NJ, USA).
ingly to overall costs given these shifts. Samples were representative (using stratified random sam-
Despite their prominent and critical role in the care of pling with age and gender according to population parameters
patients, knowledge of cancer caregivers’ burden is limited, in the International Database of the US Census) of adult
especially in Europe, with most research in this domain pub- populations in France, Germany, Italy, Spain, and the United
lished in the USA [10, 17, 25, 29, 30]. This literature is Kingdom (UK). In cases where respondents completed both
predominantly focused upon Alzheimer’s disease, dementia, the 2010 and 2011 EU NHWS, only 2011 data were analyzed.
multiple sclerosis, or on other chronic diseases like diabetes, The total sample included 114,962 respondents. All NHWS
with little attention given to cancer. The rapid increase of the respondents were asked, “Are you currently caring for an
incidence of cancer patients contributes to cancer being one of adult relative with any of the following conditions?” Several
the health conditions requiring greater caregiver support and response conditions were listed (e.g., Alzheimer’s disease,
recognition. epilepsy, and cancer). Data were analyzed for respondents
The current study investigated the burden associated with who self-reported being caregivers for an adult relative with
informal care for patients with cancer in several countries in cancer or not being caregivers (for an adult relative with any
Europe, exploring its impact on stress-related comorbidities condition). The inclusive nature of the caregiving question
(depression, anxiety, insomnia, headache, migraine, and gas- allows for a heterogeneous sample of caregivers.
trointestinal problems), healthcare resource utilization (emer-
gency room (ER) visits, hospitalizations, and visits to tradi- Measures and survey instruments
tional providers), health-related quality of life (HRQoL), and
work productivity and activity impairment. Specifically, it was Sociodemographic and health characteristic measures includ-
hypothesized that those who provide informal care for a ed age, country, gender, education, household income catego-
patient with cancer would report worse HRQoL, greater im- ry, marital status, employment, alcohol, exercise (i.e., at least
pairment in work and daily activities, greater healthcare re- 20 min of vigorous activity at least 12 times in the past month
source utilization, and a higher prevalence of stress-related vs. less), smoking (i.e., said “yes” to currently smoking vs.
comorbidities compared with a control group of non- never having smoked or not currently smoking), and Charlson
caregivers. Caregiver roles may vary across type and severity comorbidity index (CCI) category [31]. The CCI is a summary
of disease, but the current study was intended to examine score weighting the presence of several comorbidities, with
overall burden associated with caring for patients with any higher scores indicating increased degree of mortality risk;
type of cancer and to understand better the relative burden as because ulcers were not assessed in the NHWS, they were not
experienced by a representative sample of caregivers in the included in the calculation. Respondents also provided their
population (i.e., regardless of the specific distributions of height and weight (with an option to decline to provide their
subtypes and stages). These results can contribute to the weight). This information was then converted into a body
growing recognition of the deleterious impact on a heteroge- mass index (BMI) category.
neous range of caregivers, providing important insight into the Stress-related comorbidities informed by the literature (i.e.,
caregiving experience in Europe. comorbidities potentially triggered or exacerbated by stressful
experiences such as caregiving or other factors) included self-
reported diagnosis with depression, anxiety, insomnia, head-
Methods ache, migraine, and gastrointestinal (GI) comorbidities (i.e.,
diagnosis with gastroesophageal reflux disease, heartburn,
Study design and/or irritable bowel syndrome) [23].
HRQoL was assessed via the Short Form-12 (SF-12v2)
Data were provided from the 2010 and 2011 European Union questionnaire, which provided mental (MCS) and physical
(EU) National Health and Wellness Survey (NHWS): an (PCS) component summary scores (varying, conceptually,
Support Care Cancer

between 0 and 100, with higher scores indicating better health with a 1-unit change in the covariate. Significant differences
status) as well as specific subcomponent scales [32]. SF-6D from bivariate analyses helped inform the selection of control
health state utilities were also calculated from responses to the variables used in multivariable analyses.
SF-12v2, representing the value of a respondent’s given health Multivariable regression models then examined health out-
state and varying conceptually between 0 and 1, with higher come differences between caregivers and non-caregivers,
scores indicating better HRQoL [32, 33]. Differences in PCS adjusting for select demographic and health characteristic
and MCS scores that exceed 3 points are considered minimal- covariates (i.e., age, sex, college degree, income, marital
ly important differences (MIDs) [32], as are differences ex- status, employment, BMI, alcohol, exercise, smoking, and
ceeding 0.03 points on SF-6D health utilities [32, 34]. MIDs CCI). For all analyses, a P value<0.05 was considered statis-
represent the smallest differences on a measure associated tically significant.
with meaningful (as opposed to trivial) differences on clinical
and other relevant aspects of a respondent’s health status,
considered in terms of importance to the respondent or soci-
ety; they are not a fixed quantity, but rather a guideline based Results
on consideration of, e.g., known causes and indicators of
meaningful differences [32, 35]. A total of 1,713 caregivers and 103,868 non-caregivers were
Work productivity and activity impairment (WPAI) identified via NHWS. Among the 2010 EU NHWS respon-
questionnaire-derived scores included absenteeism (percent- dents, five caregivers also participated in the 2011 EU NHWS;
age work productivity lost due to missed work days), only their 2011 responses were analyzed. In this total sample
presenteeism-related impairment (percentage impairment of 105,581 adults in the EU, 27,368 (25.9 %) were in France,
while working), overall work impairment (percentage impair- 27,898 (26.4 %) in Germany, 13,329 (12.6 %) in Italy, 9,090
ment combining absenteeism and presenteeism), and activity (8.6 %) in Spain, and 27,896 (26.4 %) in the UK (Table 1).
impairment (percentage impairment during daily activities) Demographic characteristics of the study sample are sum-
due to respondents’ general health during the past week marized in Table 1. Caregivers were on average 46.3 years
[36]. Absenteeism, presenteeism, and overall work impair- old, 51.5 % were either overweight or obese, and the majority
ment were assessed only for employed respondents, whereas was employed (57.4 %). Compared with non-caregivers, care-
activity impairment was assessed for all respondents. givers were significantly more often female (58.7 vs. 50.4 %),
Healthcare resource use was operationalized as self- married or living with a partner (68.5 vs. 63.1 %) and current
reported number of hospitalizations, emergency room visits, smokers (30.9 vs. 26.9 %), and they experienced more co-
and visits to traditional providers in the past 6 months, for morbidities (23.1 vs. 17.3 % with CCI=1+).
respondents’ own medical condition. Bivariate analyses revealed that caregivers reported signifi-
cant impairments across all health outcome variables, when
Statistical analyses compared with non-caregivers (Table 2). In particular, care-
givers reported a significantly higher frequency of stress-
Descriptive analyses were conducted to provide an overview related comorbidities than non-caregivers: depression (16.0
of respondents’ characteristics, with means and standard devi- vs. 10.4 %), anxiety (24.0 vs. 12.8 %), insomnia (20.7 vs.
ations, or percentages and sample sizes, for continuous or 10.6 %), headaches (13.7 vs. 9.3 %), migraines (21.6 vs.
categorical variables, respectively. Bivariate and multivariable 13.1 %), and GI conditions (26.9 vs. 17.2 %). Caregivers,
analyses were conducted to test the potential impact of care- compared with non-caregivers, also reported significant impair-
giving on several outcomes. The independent variable was the ments in HRQoL, particularly with respect to the mental com-
respondent’s self-identification as a caregiver for a patient with ponent summary score (43.23 vs. 46.95), as well as significant-
cancer vs. non-caregiver. Dependent variables in bivariate ly lower health utilities (0.681 vs. 0.732), indicating lower
analyses consisted of health outcomes, sociodemographics, overall health status. Impairments in work productivity and
and health behaviors. activity were also observed, with caregivers reporting a signif-
Bivariate analyses compared across groups using two-sided icantly greater percentage of absenteeism (9.0 vs. 5.1 %),
tests of equality for column means or proportions for continu- presenteeism-related impairment (23.5 vs. 15.2 %), overall
ous and categorical variables, respectively. Multivariable re- work impairment (28.7 vs. 18.6 %), and activity impairment
gression models compared across caregiver vs. non-caregiver (31.8 vs. 23.6 %) than non-caregivers over the previous week.
(control) groups, adjusting for covariates and providing adjust- Finally, caregivers reported significantly greater use of
ed means (except from logistic regressions), unstandardized healthcare resources in the past 6 months than non-caregivers,
betas (B) for linear regressions, rate ratios for negative binomial with higher mean number of reported hospitalizations (0.39 vs.
regressions, and odds ratios (ORs) for logistic regressions. 0.12), ER visits (0.43 vs. 0.17), and visits to healthcare profes-
Betas represent the change in the outcome variable associated sionals (7.12 vs. 4.98).
 Support Care Cancer

Table 1 Sociodemographics and health characteristics as a function of caregiver status (bivariate analyses)

Caregiver status Total

Non-caregiver Cancer

(n=103,868) (n=1,713) (n=105,581)

Mean SD Mean SD Mean SD

Age 46.28 15.82 46.43 15.02 46.28 15.81

% n % n % n

Age category
18–29 19.5 20,260 18.0 308 19.5 20,568
30–39 18.0 18,676 16.5 283 18.0 18,959
40–49 19.7 20,488 21.2 363 19.7 20,851
50–64 25.1 26,041 29.2 500* 25.1 26,541
65+ 17.7 18,403 15.1 259* 17.7 18,662
Country
France 25.8 26,829 31.5 539* 25.9 27,368
Germany 26.5 27,520 22.1 378* 26.4 27,898
Italy 12.6 13,036 17.1 293* 12.6 13,329
Spain 8.6 8,891 11.6 199* 8.6 9,090
UK 26.6 27,592 17.7 304* 26.4 27,896
Sex
Female 50.4 52,357 58.7 1,005* 50.5 53,362
Male 49.6 51,511 41.3 708* 49.5 52,219
Education
Some college or less 67.2 69,824 67.7 1,160 67.2 70,984
College degree+ 32.8 34,044 32.3 553 32.8 34,597
Household income
Low: <€20 k/£20 k 27.7 28,814 27.7 475 27.7 29,289
Medium: €20 k/£20 k to<€50 k/£40 k 42.5 44,141 44.9 769* 42.5 44,910
High: €50 k/£40 k+ 15.6 16,199 15.3 262 15.6 16,461
Declined to answer 14.2 14,714 12.1 207* 14.1 14,921
Marital status
Single/divorced/separated/widowed 36.9 38,304 31.5 539* 36.8 38,843
Married/living with partner 63.1 65,564 68.5 1,174* 63.2 66,738
Employment status
Employed FT/PT/Self 57.4 59,600 57.8 990 57.4 60,590
Disabled 2.5 2,613 3.2 55 2.5 2,668
Unemployed 40.1 41,655 39.0 668 40.1 42,323
BMI category
Underweight 2.7 2,850 3.4 59 2.8 2,909
Normal weight 43.2 44,848 42.8 733 43.2 45,581
Overweight 33.1 34,363 32.3 553 33.1 34,916
Obese 18.4 19,077 19.7 337 18.4 19,414
Unknown 2.6 2,730 1.8 31* 2.6 2,761
Drink alcohol, per week
Once or less 69.8 72,517 68.9 1,181 69.8 73,698
Twice or more 30.2 31,351 31.1 532 30.2 31,883
Exercise 20+min, in the past month
0–11 times 79.3 82,318 79.0 1,354 79.2 83,672
12+ times 20.7 21,550 21.0 359 20.8 21,909
Support Care Cancer

Table 1 (continued)

Caregiver status Total

Non-caregiver Cancer

(n=103,868) (n=1,713) (n=105,581)

% n % n % n

Smoke cigarettes
No 73.1 75,936 69.1 1,183* 73.0 77,119
Yes 26.9 27,932 30.9 530* 27.0 28,462
CCI, excluding ulcers
0 82.7 85,894 76.9 1,318* 82.6 87,212
1 12.4 12,835 14.2 243* 12.4 13,078
2 3.5 3,586 5.3 91* 3.5 3,677
3+ 1.5 1,553 3.6 61* 1.5 1,614

BMI body mass index, CCI Charlson comorbidity index, FT employed full-time, PT employed part-time, SD standard deviation, Self self-employed
*P<0.05 (findings within a given row represent a significant difference from non-caregivers)

Even after adjusting for various demographic and patient Caregivers also reported increased rates of stress-related
characteristics, caregivers revealed a significant (P<0.001) im- conditions, including depression, anxiety, insomnia, headaches,
pairment across all health outcome variables when compared migraines, and GI problems. These results are consistent with
with non-caregivers. Caregivers’ HRQoL was worse, particu- the previous research on the comorbid profile of caregivers for
larly with respect to the mental component summary score and those with cancer [23], and they contribute to an emerging
health state utilities, where the deterioration exceeded the MID literature documenting the impact of stress on the cellular level
and therefore represented a meaningful decrease in HRQoL among caregivers across various conditions [37]. Frequency of
(Table 3). For example, MIDs on MCS have been associated self-reported diagnosis with headaches was lower than that of
with increased probability of utilizing mental health services migraines, likely reflecting relative reluctance of respondents to
and increased risk of mortality [32]. Caregivers also reported seek out medical care (and hence a formal diagnosis) for
significantly greater rates of absenteeism (69.1 %) and in- headaches vs. migraines, even though the experience of
creases of 44.6 % in presenteeism-related impairment, 46.3 % nondiagnosed (and presumably less severe and chronic) head-
in overall work impairment, and 31.7 % in activity impairment aches should be higher among both caregivers and non-
(Fig. 1). Caregivers reported significant increases of 73.9 % in caregivers, These high rates of stress-related comorbidities
visits to the ER, 106.6 % in hospitalizations, and 33.4 % in may be particularly notable given the possible reluctance of
healthcare professional visits (Fig. 2). Finally, caregivers had many caregivers to seek medical care themselves. Although
significantly increased odds of depression (OR=1.50), anxiety higher rates of healthcare resource utilization than non-
(OR=1.97), insomnia (OR=2.01), migraines (OR=1.66), and caregiver controls were reported, consistent with previous find-
GI problems (OR=1.63) (Fig. 3). Caregivers had increased ings [27], many caregivers may delay or avoid seeking medical
odds of headaches (OR=1.37) relative to non-caregivers, but care due to a lack of time or unwillingness to shift focus away
this was not statistically significant. from the patients’ own physical condition.
These results are noteworthy given the large, representative
sample surveyed across several EU countries, the use of self-
Discussion reported caregiver status (as opposed to unreliable proxy
measures such as having spouses in claims databases identi-
The current study provides important insight into the relative fied as “caregivers”), and comparisons with a non-caregiver
burden of informal caregivers of those with cancer in Europe. control group, thus helping to appreciate the relative burden
The results highlight the breadth of caregiver burden, ranging associated with caregiving, as opposed to studies showing
from serious comorbidities to deterioration in HRQoL, to only absolute impairments and costs. These results are further
reductions in work productivity and increases in healthcare bolstered by the use of patient-reported outcomes examined in
resource use. These impacts are typically examined from the prior studies showing impairments among caregivers [38].
perspective of the patient rather than from the perspective of Impairments across multiple domains are consistent with the
those who are affected secondhand by the disease. general consensus that caregiver burden is multidimensional
 Support Care Cancer

Table 2 Health outcomes as a function of caregiver status (bivariate analyses)

Caregiver status Total

Non-caregiver Cancer

(n=103,868) (n=1,713) (n=105,581)

% n % n % n

Stress-related comorbidities
Depression 10.4 10,787 16.0 274* 10.5 11,061
Anxiety 12.8 13,284 24.0 411* 13.0 13,695
Insomnia 10.6 10,967 20.7 354* 10.7 11,321
Headache 9.3 9,691 13.7 235* 9.4 9,926
Migraine 13.1 13,564 21.6 370* 13.2 13,934
Gastrointestinal (GERD, heartburn, IBS) 17.2 17,847 26.9 461* 17.3 18,308

Mean SD Mean SD Mean SD

Work productivity and activity impairment (WPAI)
Absenteeism 5.13 18.59 8.97* 22.83 5.20 18.67
Presenteeism-related impairment 15.25 22.60 23.47* 26.18 15.39 22.68
Overall work impairment 18.62 27.29 28.67* 31.32 18.79 27.39
Activity impairment 23.64 28.01 31.76* 29.93 23.77 28.06
Healthcare resource use
ER visits 0.17 0.99 0.43* 2.33 0.18 1.03
Hospitalized 0.12 1.04 0.39* 2.68 0.13 1.09
Visits to healthcare professionals 4.98 6.92 7.12* 8.94 5.01 6.96
HRQoL (SF-12v2)
Physical Component Summary 48.93 9.65 46.94* 9.90 48.90 9.66
Mental Component Summary 46.95 10.51 43.23* 10.81 46.89 10.53
SF-6D Health Utilities 0.732 0.135 0.681* 0.127 0.731 0.135
HRQoL (SF-12v2 subcomponents)
Bodily Pain 47.23 10.74 44.18* 11.07 47.18 10.75
General Health 46.50 10.95 44.30* 11.29 46.47 10.96
Mental Health 47.54 10.44 44.06* 10.64 47.48 10.45
Physical Functioning 50.38 9.49 48.12* 10.44 50.34 9.51
Role Emotional 45.85 11.31 41.64* 12.15 45.79 11.34
Role Physical 47.54 9.85 44.44* 10.19 47.49 9.86
Social Functioning 46.95 10.75 42.64* 11.32 46.88 10.78
Vitality 50.96 9.76 49.78* 9.86 50.94 9.76

WPAI scores ranged from 0 to 100 %, with higher scores indicating worse outcomes. Higher HRQoL scores indicate better health status
ER emergency room, GERD gastroesophageal reflux disease, HRQoL health-related quality of life, IBS irritable bowel syndrome, SD standard deviation
*P<0.05 (findings within a given row represent a significant difference from non-caregivers)

[39] as well as with emerging literature documenting the with a meaningful decline in mental health status and overall
different negative health effects associated with informal care- HRQoL. This finding is particularly interesting given the
giving in cancer, which can be similar to the burdens (phys- generality of the SF-12v2 vs. other, caregiving-specific instru-
ical, emotional, financial) faced by caregivers for patients with ments that might have been more sensitive to differences in
other diseases (e.g., dementia) [30]. In particular, impaired aspects specific to caregiver roles (e.g., instrumental support,
HRQoL that is associated with caregiving in the context of financial considerations) [40].
cancer [17, 19, 23] is confirmed in this study with mental Individuals who experience burden across multiple do-
health status and health utilities impairments exceeding the mains, such as demonstrated in the current study, may become
cutoff for MIDs, suggesting that caregiving was associated less effective in their caregiving support, thus placing strain on
Support Care Cancer

Table 3 Adjusted health-related quality of life as a function of caregiving for patients with cancer, after controlling for covariates (multivariable
analyses)

Unstandardized Adjusted means

coefficients

B SE Wald P value B 95 % LCL B 95 % UCL Non-caregiver Caregiver

HRQoL (SF-12v2) Physical Component Summary −1.29 0.20 40.80 <0.001 −1.69 −0.89 48.92 47.63
Mental Component Summary −3.26 0.24 182.06 <0.001 −3.74 −2.79 46.94 43.68
SF-6D Health Utilities −0.043 0.003 191.23 <0.001 −0.049 −0.037 0.732 0.689
HRQoL (SF-12v2 Bodily Pain −2.32 0.24 89.84 <0.001 −2.79 −1.84 47.22 44.90
subcomponents) General Health −1.42 0.23 36.40 <0.001 −1.88 −0.96 46.49 45.07
Mental Health −2.95 0.24 150.76 <0.001 −3.43 −2.48 47.53 44.57
Physical Functioning −1.64 0.21 62.68 <0.001 −2.05 −1.23 50.37 48.73
Role Emotional −3.65 0.26 191.48 <0.001 −4.17 −3.13 45.85 42.19
Role Physical −2.47 0.22 125.72 <0.001 −2.91 −2.04 47.53 45.06
Social Functioning −3.77 0.25 230.52 <0.001 −4.26 −3.29 46.94 43.17
Vitality −0.68 0.22 9.22 0.002 −1.12 −0.24 50.95 50.27

Presented is the association of caregiving for patients with cancer vs. non-caregiving with health-related quality of life. All comparisons involved 1,713
caregivers vs. 103,868 non-caregivers. Multivariable (normal regression) models adjusted for the following covariates (reference categories shown in
parentheses): age, sex (female), college degree+(less), income: medium (low), income: high (low), income: declined (low), married/partnered (single/
divorced/separated/widowed), employed: full-time/part-time/self-employed (unemployed), employed: disabled (unemployed), BMI: overweight (nor-
mal/underweight/unknown), BMI: obese (normal/underweight/unknown), alcohol twice per week or more (less), exercise 12 times per month or more
(less), smoke (no), and Charlson comorbidity index
B beta, BMI body mass index, LCL lower confidence limit, SE standard error, UCL upper confidence limit

Non-caregivers the patient-caregiver relationship and potentially resulting in

Cancer Caregivers the eventual need for greater formal care resources to be
30 28.7
implemented. Moreover, the notable impairments in work
Percentage impairment (%)

26.4
25 productivity reported by caregivers suggest not only immedi-
21.5 21.8
ate financial burdens but also long-term financial losses for
20 18.1
14.8 families, employers, and society as a whole. The substantial
15 burden of caregiving and its multifactorial nature, along with
10 8.1 its potential impact on the patient’s outcomes, reinforces the
4.8 need for continued development, testing, and implementation
5
of customized evidence-based supportive services and
0 multidimensional interventions for caregivers of patients with
cancer [41–44].

Limitations and future directions

Although this study extends our understanding of the burden of

Fig. 1 Adjusted productivity-related impairment as a function of care-
caregiving in Europe, the following limitations should be noted.
giving for patients with cancer, after controlling for covariates (multivar- Cross-sectional data do not provide a basis for definitive claims
iable analyses). Presented is the association of caregiving for patients with regarding causal relationships (e.g., between caregiving and
cancer vs. non-caregiving with work productivity and activity impair- HRQoL), as variables are assessed contemporaneously and
ment, along with 95 % confidence intervals. All impairments were
significantly worse for caregivers vs. non-caregivers, at P<0.001. Activ-
retrospectively (e.g., caregiving vs. non-caregiving may lead
ity impairment involved 1,713 caregivers vs. 103,868 non-caregivers. to impairments, impaired respondents may be more likely to
Absenteeism and overall work impairment were assessed only for the become caregivers vs. non-caregivers, or other unmeasured
subsample of employed respondents: 946 caregivers vs. 57,242 non- variables may lead to both caregiving and impairment).
caregivers. Presenteeism was assessed only for the subsample of
employed respondents who worked nonzero hours in the past week:
Therefore, longitudinal studies may help strengthen the current
922 caregivers vs. 55,899 non-caregivers. Multivariable (negative bino- findings. The survey did not collect information on patient
mial regression) models adjusted for the covariates noted previously characteristics such as relationship with care recipient, disease
 Support Care Cancer

Non-caregivers
history, symptom burden, tumor type and staging, cancer treat-
Cancer Caregivers ment (e.g., chemotherapy vs. non-chemotherapy), or receipt of
Number of visits in past 6 months 7
6.18 formal or other supportive care, which are expected to be key
6 variables explaining the caregiver burden in cancer and provide
fertile ground for future research efforts. These variables may
5 4.64
be confounded with caregiving in the current study. However,
4 the study did control for other likely, more general confounds
such as sex, income, and marital status of the caregiver in
3 multivariable analyses, thus helping to focus results on the
2 burden of caregiving.
Whereas the NHWS dataset mimics the demographic com-
1 position of the respective populations, the Internet-based, self-
0.16 0.27 0.11 0.22
0 report format can lead to sample biases (e.g., lack of repre-
sentation of institutionalized respondents or those with other-
wise limited access to the Internet). Importantly, this may
render the burden in this study an underestimate of the full
burden, to the extent that those with ready access and interest
in participating in the NHWS are likely to be more urban,
healthier, with greater resources and access to supportive
Fig. 2 Adjusted healthcare resource use as a function of caregiving for services. Self-selection on the part of caregivers interested in
patients with cancer, after controlling for covariates (multivariable anal- participating may result in a subsample of respondents whose
yses). Presented is the association of caregiving for patients with cancer
vs. non-caregiving with healthcare resource use, along with 95 % confi-
characteristics differ from those of nonparticipating NHWS
dence intervals. All resource use was significantly higher for caregivers caregivers (e.g., greater motivation and ability to participate,
vs. non-caregivers, at P<0.001. All comparisons involved 1,713 care- likelihood of being caregivers for a longer period of time,
givers vs. 103,868 non-caregivers, except for the non-caregivers group in etc.). Data across European countries were pooled together to
the traditional provider visits analysis, which excluded 1 respondent with
missing data (n=103,867). Multivariable (negative binomial regression)
maximize statistical power. However, this may have hidden
models adjusted for the covariates noted previously. ER emergency room underlying heterogeneity of the burden across countries, es-
pecially Italy and Spain that were less represented in our
sample. This heterogeneity across countries may manifest in
different mean levels of resource use due to healthcare system
Depression differences, different normative thresholds and barriers to
utilizing healthcare resources (e.g., reluctance to seek out a
healthcare professional when sick), and different experiences
Anxiety
and ratings of HRQoL (e.g., lower average mental health
status).
Insomnia A small proportion of caregivers (and non-caregivers) had
cancer themselves, which could potentially inflate the burden
Headache of caregiving, especially as cancer rates were slightly higher
among caregivers (9.0 %) than among non-caregivers (5.6 %).
However, multivariable regression analyses controlled for
Migraine
CCI (which includes the experience of metastatic tumors or
any tumor) and supplementary analyses conducted on restrict-
Gastrointestinal (GERD,
heartburn, or IBS) ed, cancer-free samples of caregivers vs. non-caregivers
showed that the magnitude and statistical significance
0.0 0.5 1.0 1.5 2.0 2.5 (P<0.001) of the differences on three major outcomes (health
Fig. 3 Adjusted relative odds of stress-related comorbidities as a func- utilities, overall work productivity impairment, and traditional
tion of caregiving for patients with cancer, after controlling for covariates
(multivariable analyses). Presented in bars are the adjusted odds ratios of
provider visits) were virtually unchanged, with only a slight
stress-related comorbidities associated with caregiving for patients with shift to lower impairment observed across groups.
cancer vs. non-caregiving, along with 95 % confidence intervals. Care- Future research is needed to further clarify the important
givers vs. non-caregivers had significantly higher odds of each comorbid role family members and informal caregivers play in the
condition, at P<0.001. All comparisons involved 1,713 caregivers vs.
103,868 non-caregivers. Multivariable (logistic regression) models ad-
cancer care trajectory. The caregiving relationship is com-
justed for the covariates noted previously. GERD gastroesophageal reflux plex and is influenced by both patient and caregiver charac-
disease, IBS irritable bowel syndrome teristics as well as additional factors such as whether formal
Support Care Cancer

care was being administered to the patient. Formal care (e.g., in shaping policy and practice for cancer patients and their
administered by a nurse, therapist, or social worker) was not caregivers, and implementing cancer care models that involve
assessed in the current study, and the presence of comple- cancer caregivers.
mentary formal or informal care may have resulted in an
underestimation of burden. This information is important for Acknowledgments The authors would like to acknowledge Errol J.
evaluating caregiver burden more precisely in future research. Philip, PhD, who provided support with literature review and manuscript
editing as a paid consultant to Kantar Health, with funding from Bristol-
The role of patient disease characteristics (e.g., tumor type) in Myers Squibb, Inc.
guiding caregiver duties and potential burden remains to be
fully elucidated in cancer, while researchers have begun Disclosures Amir Goren and Marco daCosta DiBonaventura are em-
identifying factors that may buffer against the negative im- ployees of Kantar Health, a global healthcare consulting and research
pact of caregiver burden [45], including the role of social organization, and contributed to the study design, analysis, and reporting
with funding from Bristol-Myers Squibb. Isabelle Gilloteau and Michael
support [46]. Further, the burden associated with caregiving is Lees are paid employees of Bristol-Myers Squibb, a pharmaceutical
not a static entity and is expected to change over time in company whose mission to provide innovative medicines (helping pa-
response to the disease trajectory and characteristics of the tients prevail over serious diseases such as cancer) informed the current
patient-caregiver relationship [10]. study and its aim of exploring different dimensions of cancer-related
burden and the broader community affected by the disease. A portion of
The need for longitudinal research is therefore critical in this work was presented at the 2012 Congress of the European Society for
identifying opportunities to intervene and to help guide the Medical Oncology (ESMO) in Vienna, Austria. Bristol-Myers Squibb,
development of targeted, effective interventions to support Inc. provided funding for the execution, analysis, reporting, and publica-
caregivers. The substantial burden identified in the current tion of the current study.
study (especially in terms of mental health impairments) sug-
gests that the impact of cancer on informal caregivers represents
an important additional outcome to be assessed in the develop- References
ment of comprehensive models of cancer care. Given the rise in
the substantial costs associated with cancer treatment [2–5, 47], 1. ACS (2012) Cancer facts and figures 2012. American Cancer
cost-based analyses of treatments and interventions may need Society, Atlanta
to consider not only costs incurred by payers and patients but 2. Yabroff KR, Lund J, Kepka D, Mariotto A (2011) Economic burden
of cancer in the United States: estimates, projections, and future
also those incurred by informal caregivers as critical members research. Cancer Epidemiol Biomarkers Prev 20:2006–2014. doi:
of the care team. On the flipside, cost-based analyses should 10.1158/1055-9965.EPI-11-0650
account for the cost savings associated with effective treatments 3. Soerjomataram I, Lortet-Tieulent J, Parkin DM, Ferlay J, Mathers C,
evaluated over time, as caregivers no longer need to provide Forman D et al (2012) Global burden of cancer in 2008: a systematic
analysis of disability-adjusted life-years in 12 world regions. Lancet
extended care for cancer survivors or patients with mitigated 380:1840–1850. doi:10.1016/S0140-6736(12)60919-2
symptoms. Caregiver-based costs associated with work produc- 4. Jönsson B, Wilking N (2007) The burden and cost of cancer. Ann
tivity impairment, treatment of comorbid conditions, and in- Oncol 18(Supplement 3):iii2–iii7
creased care needs for the patient are likely considerable, and 5. Bray F, Jemal A, Grey N, Ferlay J, Forman D (2012) Global cancer
transitions according to the Human Development Index (2008–
efforts should be made to quantify these factors in future 2030): a population-based study. Lancet Oncol 13:790–801
research endeavors. We revealed the extent of the caregiver 6. Maddams J, Brewster D, Gavin A, Steward J, Elliott J, Utley M et al
burden in cancer in Europe but its financial impact is still (2009) Cancer prevalence in the United Kingdom: estimates for
unknown. The research does indicate, however, that while 2008. Br J Cancer 101:541–547. doi:10.1038/sj.bjc.6605148
7. Ferlay J, Autier P, Boniol M, Heanue M, Colombet M, Boyle P
caregivers may be unpaid and therefore save the healthcare (2007) Estimates of the cancer incidence and mortality in Europe in
system costs, the impact on caregivers will result in other 2006. Ann Oncol 18:581–592
hidden costs to the healthcare system. 8. Alexe DM, Albreht T, McKee M, Coleman MP (2008) Cancer control
The current study provides important insight into the relative in Europe today: challenges and policy options. Eurohealth 14:12–15
9. Stewart BW, Kleihues P (2003) World cancer report. WHO
and multidimensional burden associated with informal caregiv- International Agency for Research on Cancer, IARC Press, Lyon,
ing for patients with cancer in Europe. Caregivers experienced pp 1–342
a significantly higher burden than non-caregivers, thus likely to 10. Given BA, Given CW, Sherwood PR (2012) Family and caregiver
add to the already high societal costs of cancer and its treatment. needs over the course of the cancer trajectory. J Support Oncol 10:
57–64. doi:10.1016/j.suponc.2011.10.003
Special attention should be given to caregivers based on their 11. van Ryn M, Sanders S, Kahn K, van Houtven C, Griffin JM, Martin
pivotal role in maintaining the health and well-being of patients M et al (2011) Objective burden, resources, and other stressors among
with cancer, fulfilling a need not routinely addressed by the informal cancer caregivers: a hidden quality issue? Psychooncology
healthcare system. As is the case in research on dementia 20:44–52. doi:10.1002/pon.1703
12. Hassett MJ (2010) The full burden of cancer. Oncologist 15:793–795.
caregiving [30], this growing body of literature on cancer doi:10.1634/theoncologist.2010-0211
caregiver burden can be influential in developing a comprehen- 13. Girgis A, Lambert S, Johnson C, Waller A, Currow D (2012)
sive understanding of the unique cancer caregiver experience, Physical, psychosocial, relationship, and economic burden of caring
 Support Care Cancer

for people with cancer: a review. J Oncol Pract 9:197–202. doi:10. 30. Kim Y, Schulz R (2008) Family caregivers’ strains: comparative
1200/JOP.2012.000690 analysis of cancer caregiving with dementia, diabetes, and frail
14. Arno PS (2006) Economic value and informal caregiving. Paper elderly caregiving. J Aging Health 20:483–503. doi:10.1177/
presented at the Care Coordination and Caregiving Forum. 0898264308317533
Department of Veterans Affairs, Bethesda 31. Charlson ME, Pompei P, Ales KL, MacKenzie CR (1987) A new
15. Luengo-Fernandez R, Leal J, Sullivan R (2012). Economic burden of method of classifying prognostic comorbidity in longitudinal studies:
malignant neoplasms in the European Union. ESMO 1415PD_PR. development and validation. J Chronic Diseases 40:373–383
ESMO 2012. http://oncologypro.esmo.org/meeting-resources/ 32. Maruish ME (ed) (2011) User’s manual for the SF-36v2 Health
meeting-abstracts/european-society-for-medical-oncology-esmo- Survey, 3rd edn. QualityMetric Incorporated, Lincoln
2012/economic-burden-of-malignant-neoplasms-2513.aspx. 33. Brazier J, Roberts J, Deverill M (2002) The estimation of a
Accessed 7 February 2013 preference-based measure of health from the SF-36. J Health Econ
16. Ferrall SM (2006) Caring for the family caregiver. In: Carroll- 21:271–292
Johnson RM, Gorman LM, Bush NJ (eds) Psychosocial nursing care 34. Walters SJ, Brazier JE (2005) Comparison of the minimally impor-
along the cancer continuum, 2nd edn. Oncology Nursing Society, tant difference for two health state utility measures: EQ-5D and SF-
Pittsburgh, pp 603–610 6D. Quality Life Res 14:1523–1532
17. Kim Y, Given BA (2008) Quality of life of family caregivers of 35. King MT (2011) A point of minimal important difference (MID): a
cancer survivors: across the trajectory of the illness. Cancer 112(11 critique of terminology and methods. Expert Rev Pharmacoecon
Suppl):2556–2568. doi:10.1002/cncr.23449 Outcome Res 11:171–184
18. Chen ML, Chu L, Chen HC (2004) Impact of cancer patients’ quality 36. Reilly MC, Zbrozek AS, Dukes EM (1993) The validity and repro-
of life on that of spouse caregivers. Support Care Cancer 12:469–475. ducibility of a work productivity and activity impairment instrument.
doi:10.1007/s00520-004-0636-z PharmacoEconomic 4:353–365
19. Kitrungroter L, Cohen MZ (2006) Quality of life of family caregivers 37. Damjanovic AK, Yang Y, Glaser R, Kiecolt-Glaser JK, Nguyen H,
of patients with cancer: a literature review. Oncol Nurs Forum 33: Laskowski B et al (2007) Accelerated telomere erosion is associated
625–632. doi:10.1188/06.onf.625-632 with a declining immune function of caregivers of Alzheimer’s
20. Rhee YS, Yun YH, Park S, Shin DO, Lee KM, Yoo HJ et al (2008) disease patients. J Immunol 179:4249–4254
Depression in family caregivers of cancer patients: the feeling of 38. Gupta S, Goren A, Phillips AL, Stewart M (2012) Self-reported
burden as a predictor of depression. J Clin Ocol 46:5890–5895 burden among caregivers of patients with multiple sclerosis. Int J
21. Rivera HR (2009) Depression symptoms in cancer caregivers. Clinl J MS Care 14:179–187
Oncol Nurs 13:195–202 39. Yueh-Feng Lu Y, Austrom MG (2005) Distress responses and self-
22. Hearson B, McClement S (2007) Sleep disturbance in family care- care behaviors in dementia family caregivers with high and low
givers of patients with advanced cancer. Int J Palliat Nurs 13:495–501 depressed mood. J Am Psychiatr Nurse Assoc 11:231–240
23. Stenberg U, Ruland CM, Miaskowski C (2010) Review of the 40. Deeken JF, Taylor KL, Mangan P, Yabroff KR, Ingham JM (2003)
literature on the effects of caring for a patient with cancer. Care for the caregivers: a review of self-report instruments developed
Psychooncology 19:1013–1025. doi:10.1002/pon.1670 to measure the burden, needs, and quality of life of informal care-
24. Yabroff KR, Kim Y (2009) Time costs associated with informal givers. J Pain Symptom Manage 26:922–953
caregiving for cancer survivors. Cancer 115(18 Suppl):4362–4373. 41. Pasacreta JV, McCorkle R (2000) Cancer care: impact of interven-
doi:10.1002/cncr.24588 tions on caregiver outcomes. Annu Rev Nurs Res 18:127–148
25. Given B, Wyatt G, Given C, Sherwood P, Gift A, DeVoss D et al 42. Sorensen S, Pinquart M, Duberstein P (2002) How effective are
(2004) Burden and depression among caregivers of patients with interventions with caregivers? An updated meta-analysis.
cancer at the end of life. Oncol Nurs Forum 31:1105–1117. doi:10. Gerontologist 42:356–372
1188/04.ONF.1105-1117 43. Northouse LL, Mood DW, Schafenacker A, Kalemkerian G,
26. Tsigaroppoulos T, Mazaris E, Chatzidarellis E, Skolarikos A, Zalupski M, Lorusso P et al (2013) Randomized clinical trial of a
Varkarakis I, Deliveliotis C (2009) Problems faced by relatives caring brief and extensive dyadic intervention for advanced cancer patients
for cancer patients at home. Int J Nurs Pract 15:1–6. doi:10.1111/j. and their family caregivers. Psychooncology 22:555–563. doi:10.
1440-172X.2008.01725.x 1002/pon.3036
27. Sjovall K, Attner B, Lithman T, Noreen D, Gunnars B, Thome B et al 44. Northouse LL, Katapodi MC, Song L, Zhang L, Mood DW (2010)
(2009) Influence on the health of the partner affected by tumor Interventions with family caregivers of cancer patients: meta-analysis
disease in the wife or husband based on a population-based register of randomized trials. CA Cancer J Clin 60:317–339
study of cancer in Sweden. J Clin Ocol 27:4781–4786. doi:10.1200/ 45. Scherbring M (2002) Effect of caregiver perception of preparedness
JCO.2008.21.6788 on burden in an oncology population. Oncol Nurs Forum 29:E70–
28. Mazanec SR, Daly BJ, Douglas SL, Lipson AR (2011) Work pro- E76. doi:10.1188/02.ONF.E70-E76
ductivity and health of informal caregivers of persons with advanced 46. Goldstein NE, Concato J, Fried TR, Kasl SV, Johnson-Hurzeler R,
cancer. Res Nurs Health 34:483–495. doi:10.1002/nur.20461 Bradley EH (2004) Factors associated with caregiver burden among
29. Palos GR, Mendoza TR, Liao KP, Anderson KO, Garcia-Gonzalez caregivers of terminally ill patients with cancer. J Palliat Care 20:38–43
A, Hahn K et al (2011) Caregiver symptom burden: the risk of caring 47. Sullivan R, Peppercorn J, Sikora K, Zalcberg J, Meropol NJ, Amir E
for an underserved patient with advanced cancer. Cancer 117:1070– et al (2011) Delivering affordable cancer care in high-income countries.
1079. doi:10.1002/cncr.25695 Lancet Oncol 12:933–980. doi:10.1016/s1470-2045(11)70141-3