1867
The Cost of Cancer Home Care to Families
Manfred Stommel, Ph.D.,*,t Charles W . Given, Ph.D.,$
atid Barbara A. Given, Ph.D., F.A.A.N.*
Background. For the most part, previous research on
costs of cancer care has focused on the formal medical
care costs. Research on home care for patients with
cancer has emphasized direct care costs (expenditures).
Among indirect costs, only loss of income to family
members has been studied. However, a major component
of indirect costs, the family labor expended to care for the
patient with cancer, needs to be included for a more real-
istic appreciation of home care costs.
Methods. The costs of family labor are estimated by
imputing monetary values for the time spent caring for
the patient with cancer. The assigned monetary cost ei-
ther is equated with income losses of the helper in ques-
tion or is based on a putative market value of the
expended labor time. In addition, out-of-pocket expendi-
tures examined in this study cover all cancer care-re-
lated expenses for which the patient was not reimbursed
by third parties. Data were obtained from a convenience
sample of 192 patients with cancer and their families in
lower Michigan.
Results. When family labor is included in the cost
calculations, average cancer home care costs for a 3-
month period ($4563)are not much lower than the costs
of nursing home care. The substantial variation in home
care costs (standard deviation [SD] = $4313)appears to be
unrelated to the type of cancer diagnosis, type of treat-
ment, or time since diagnosis but seems to be driven by
the functional status of the patient and the family living
arrangements.
Conclusions. Outpatient care for patients with
cancer coupled with greater reliance on home care ap-
pear to be economically attractive because costs to fami-
lies usually are underestimated. Cancer 1993;71:1867-
74.
Key words: costs, home care, cancer, family labor.
From the *College of Nursing, tSocial Science Research Bureau,
and the $Department of Family Practice, College of Human Medi-
cine, Michigan State University, East Lansing, Michigan.
Supported by a grant, "Family Homecare for Cancer-A Com-
munity-Based Model" (NCNR#l R 0 1 NR01915) to Barbara A.
Given and Charles W. Given, Principal Investigators; and a grant
from the American Cancer Society, "Family Home Care for Cancer
Patients," to Barbara A. Given, Principal Investigator.
Address for reprints: Manfred Stommel, Ph.I)., College of Nurs-
ing, Michigan State University, A-230 Life Sciences Building, East
Lansing, MI 48824.
Accepted for publication September 21, 1992.
With the exception of a few studies on the costs of
childhood cancer to families, past research on the cost
of cancer largely has been restricted to studying treat-
ment, screening, or prevention costs. This focus not
only leads to a neglect of the nonmedical costs asso-
ciated with cancer, but also implicitly assumes that cost
examinations should be confined to the patients alone.
In contrast, a more comprehensive view of costs would
include the financial impact on the family of the patient
and go beyond the costs of medical treatment, screen-
ing, or prevention.
One reason for the focus on medical care costs is
that such research can be conducted with reliance on
data from insurance companies, hospitals, or Medicare
files.' However, these data sources are less useful for
describing the costs borne by patients and their fami-
lies.' In particular, three important components of fam-
ily costs often are neglected, although they provide a
much better picture of the impact of an illness on the
patients and their families: (1) direct costs such as out-
of-pocket cash expenditures for services not or only par-
tially covered by third-party payers; (2) indirect costs in
the form of forgone earnings opportunities resulting
from the illness; and (3) indirect costs in the form of
forgone household production and/or leisure time. It is
especially the latter component of cost for which almost
no information is available.
If the cost of family labor input is considered at all,
it usually is evaluated only insofar as it may lead to a
reduction in market earnings. However, because time
devoted to the care of a patient with cancer in the home
is no longer available for other household activities and
leisure, there are definite costs associated with caring
even if the people providing that care are not engaged
in outside work for pay.3 Estimates of such costs are
needed urgently, especially given the current trend to-
ward outpatient cancer care that has resulted in ever-
increasing amounts of care being provided at home.
However, it is difficult to assess the costs of family labor
accurately in money terms. As a result, home care costs
tend to be underestimated ~ubstantially,~ especially
when compared with the costs of institutional care.
An additional complication arises from the defini-
tion of the relevant family unit. Often "family" is de-
1868 CANCER March 1, 1993, Volume 7 1 , No. 5
fined as members of the household in which the patient
with cancer resides, whether they are relatives or not5
However, a complete description of family help would
include relatives (and others) residing in different
households, if they provide tangible help.
Microlevel studies that have focused on the costs of
care to the family are s ~ a r c e .Most
~ - ~ of these studies are
based on very small convenience samples (except for
the study of Bloom et a1.,6with 569 patients), and none
of them take a comprehensive view of costs of care to
the family. Those studies containing an estimate of fam-
ily care ~ o s t s ~
only
, ~ take
. ~ into account cash outlays for
hired labor, whereas labor provided directly by family
members remains unassessed.
Few generalizations have emerged about the distri-
bution of available cost data, except that in all studies
costs appear to vary substantially from one family to
the next, with some families facing truly catastrophic
expenditures4 It is not yet clear what governs these
variations in costs to families. There is some evidence
that the distance of a family residence from a treatment
center adds to transportation and time cost^.^,^ Family
costs also may be related to type of diagnosis, but evi-
dence on this point is quite weak6
In addition to the more customary estimates of cash
outlays (out-of-pocket expenditures) for services (in-
cluding hired labor) and supplies, estimates of the costs
of family labor will be provided in this article. Family
labor will be broken down according to its source: the
primary caregiver and/or other family members. The
price of care-giving activities is evaluated in two ways:
for those who had to give up gainful employment to
provide care, the cost is evaluated in terms of loss of
income (forgone earnings); for caregivers who were not
gainfully employed before the care-giving situation
commenced, the value of their services is equated to the
wage rate of a "typical" provider of such services. Con-
ceptually, this deviates from the opportunity cost frame-
work (because the value of their work actually equals
the value of forgone home-production and leisure
time). However, in the absence of data on usual activi-
ties of family members, the cost of a home health aide
may be considered a reasonable approximation, espe-
cially because such caregivers always could command a
wage in the labor market equivalent to that of a home
health aide. This procedure also deals with the problem
of assigning a value to forgone leisure time in that it
substitutes the value of services produced during that
time.
After providing the more comprehensive estimates
of the cost of cancer care to patients and their families,
we will offer some preliminary examinations of factors
thought to influence costs of care. In particular, the ef-
fects of medically relevant diagnostic variables are
compared with variables usually predictive of varia-
tions in home care needs and arrangements'0,": the
functional status of the patient and the family living
arrangements.
Methods
This report is based on information from a convenience
sample of 192 patients with cancer and their primary
caregivers. The data are part of the first wave of a larger
panel study of 303 caregiver-patient dyads who were
recruited by nurses and oncologists in six community-
based cancer treatment centers covering cities ranging
in size from 20,000 to 500,000 and their surrounding
rural areas in lower Michigan. Eligibility criteria for the
study included the following: (1)patient age between
20 and 80 years; (2) diagnosis of a solid tumor or lym-
phoma; (3) at least one dependency in an activity of
daily living (ADL) (e.g., dressing, eating) or an instru-
mental activity (e.g., shopping, cooking), or patient had
symptoms appearing on a standard list of symptoms";
(4) current treatment for new or recurring disease; and
(5) the existence of a "primary" caregiver (a person re-
sponsible for most caregiving tasks). All information
was obtained through telephone interviews or ques-
tionnaires mailed to both the patient and the primary
caregiver. Mainly because many patients did not return
their self-administered questionnaires (response rate,
79%; compared with a caregiver response rate of 92%),
only 192 cases with complete data on the variables of
interest could be included in the analysis.
As can be seen from the background characteristics
displayed in Table 1, most of the study participants
could be characterized as white, middle-class, Midwes-
tern couples of widely varying age. The patients in this
study had a variety of cancers, with breast cancer af-
fecting the largest group (26.6%). However, no infor-
mation was collected on the stages of the respective
cancer, although there is a measure of the time elapsed
since the first cancer diagnosis. Table 1 also contains
information on the functional status of the patients, in-
dicated by a count of the number of dependencies as
related to 11 ADL, such as eating, dressing, or bathing.
As the data indicate, most of the patients with cancer in
this sample are not highly dependent in their daily
functioning (mean, 1.4), with 50% of the patients expe-
riencing no functional limitations in ADL at all.
It is difficult to obtain accurate and reliable data on
the cost of care. In studies of medical care costs in which
survey results have been compared with insurance
claim files and other secondary data sources, underre-
porting of use of services often has been noticed.13In
addition, income and expenditure data seem more
likely to be subject to missing item response^.'^,'^ This
Cancer Home Care Costs/Stommel et al. 1869

Table 1. Characteristics of Patients with Cancer, Their comprehensive information on the cost of care (espe-
Households, and Their Caregivers* (n = 192) cially the nonmedical cost of care) to families can be
Standard obtained directly only from the families.
Characteristic Mean deviation Range To obtain a picture of the cost of care to families of
Patient's household income patients with cancer that is more complete than is cus-
($) 34,473 19,041 2500-120,000 tomary in the literature, four categories of cost have
Caregiver's age (yr) 55.2 12.4 23-81 been used in the following way: out-of-pocket expendi-
Patient's age (yr) 58.7 12.2 22-83 tures for services, labor costs of caregiver services, labor
Months since first diagnosis costs of other family members, and patient loss of earn-
of cancer 31.2 46.2 1-236
ings.
Patient's functional status
(no. of ADL
dependencies)t 1.4 2.0 0-1 1 Out-of-Pocket Expenditures for Services

No. % The primary caregivers were asked to report the dollar

Living arrangement
amounts for each of nine specific care expenditures for
Caregiver and patient alone 107 55.7% the patient. The specified expenditure categories were
Caregiver, patient, and others 68 35.4% derived from the literature and included the following:
Caregiver and patient in different households 17 8.9% hospital and physician services, nursing homes, medica-
Patient's sex tions, visiting nurses, home health aides, and purchases
Female 97 50.5% of special equipment, supplies, and foods and supple-
Male 95 49.5% ments. Caregivers also were encouraged to list and re-
Primary site of cancer port any additional care-related expenditures. Each
Breast 51 26.6% time, they were reminded to only include expenditures
Colon/rectal 24 12.5%
for which they were not reimbursed. All expenditure
Gastrointestinal 14 7.3%
questions referred to the last 3 months before the inter-
Gynecologic 10 5.2%
Lung 29 15.1%
view, which presents a maximum time period for accu-
Prostate 11 5.7% rate reconstruction and recall of such information.16
Lymphoma 24 12.5% Conceptually, our aim was to ascertain incremental
Other 29 15.1% costs (i.e., new additional expenditures that resulted
Type of current treatment from the illness or even comorbidities) if they were exac-
Chemotherapy (only) 147 76.6% erbated by the ~ a n c e rIdeally,
.~ this would require the
All other treatment combinations 45 23.4% subtraction of savings resulting from the illness. How-
Caregiver's sex ever, in only one expenditure category are such savings
Female 116 60.4% likely: Purchases of special foods should be counterba-
Male 76 39.6%
lanced by possible savings on regular foods no longer
Caregiver's relation to patient
consumed by patients. Data on such savings were not
Spouse 159 82.8%
Daughter/son 33 17.2%
available in this study, but because expenditures for
Caregiver's education special foods do not represent a large portion of the
Some high school or less 24 12.5% total out-of-pocket expenditures (on average, less than
High school graduate 69 35.9% 5%), the error introduced is negligible.
Some college 60 31.3%
College graduate/graduate degree 39 20.3% Labor Costs of Caregiver Services
Caregiver's race/ethnicity
White/Caucasian 183 95.3%
Caregivers were asked to indicate how many hours per
Black/African American 6 3.1% day they spent providing tangible care (excluding time
HisDanic/Mexican American
1 ,
3 1.6% of "keeping company"). When caregivers responded
ADL: activity of daily living. that they spent "24 hours" giving care to the patient,
* "Caregiver" refers to primary caregiver additional probing occurred to establish the meaning of
t Patient's functional status is a count of the number of limitations in the fol- "giving care." Caregivers also provided information on
lowing areas: eating, dressing, bathing, grooming, toileting, walking, getting
around the house, transferring in and out of bed, mouth care, and incontinence how long they had been giving care at the current level.
of urine and stool (Cronbach's alpha: 0.89) Multiplication of the number of days by the "typical"
number of hours spent per day yielded an estimate of
may be the result of memory problems and an unwil- hours spent providing care during the last 3 months
lingness on the part of some survey respondents to di- before the interview. The reliability of this measure was
vulge financial information. Despite these difficulties, estimated by correlations between it and ordinal mea-
1870 CANCER March 1, 2993, Volume 71,No. 5
sures of caregiver involvement in ADL (r = 0.42) and
health care activities (r = 0.38), which suggests that it
captures at least some systematic variation in hours of
care provided. The final step in constructing estimates
for the cost of caregiver labor involved the assignment
of a market value to the services provided by the care-
giver. Two situations had to be distinguished. For care-
givers who sacrificed paid employment, the value of
their caregiving services provided during the weeks of
unpaid leave of absence was equated to their lost earn-
ings. Caregivers were asked how many weeks of wages
or salary they lost because of the cancer. Lost wages
were computed by multiplying caregiver annual in-
come by 1/52 times the number of weeks of lost in-
come. For caregivers who did not lose any earnings
(only 38.5% in this sample worked for pay outside the
home), the value of their caregiving activities was esti-
mated in the following way. Given the similarity of the
caregiving tasks to those of a home health aide, the cost
of 1 hour of caregiving was equated to the average
hourly wage-rate ($7.82) of a home health aide em-
ployed by home care agencies in lower Michigan during
1989-1990 (Michigan Home Health Association). Be-
cause the available data did not allow for separation of
skilled home care services that caregivers might have
provided, the imputed wage rate must be considered a
low estimate.
Labor Costs of Other Family Members
Caregivers were asked to list all family members who
provided help with patient care. For each person listed,
caregivers were requested to estimate the hours of care
provided during the last week. Because most of the pri-
mary caregivers in this sample were spouses (82.8%) or
children living in the same household as the patient
(8.4%), they may be assumed to have sufficient knowl-
edge to provide this kind of information. This assump-
tion also seems warranted because the caregivers were
required to reconstruct schedules of other helpers for
only 1 week. Because information on lost earnings of
other family members was not available, the market
price for a home health aide ($7.82/hour) again was
used to estimate the value of the patient care services
rendered by family members. This procedure assumes
that the nature of the tasks performed by primary care-
givers and other family members is essentially the same
when, in fact, a specialization of tasks among care pro-
viders is more likely.”~’*However, in the absence of
any concrete measures of the division of labor between
various informal care providers, any attempt to differ-
entiate the market value of services provided by pri-
mary caregivers and other family members would ap-
pear to be arbitrary.
Patient Loss of Earnings
Like the caregivers, patients also were asked about the
number of weeks during the last 3 months that they
took an unpaid leave of absence or quit working alto-
gether. (Note: Because our focus is on the costs of care
to families, we do not include paid leave of absence,
which is a cost borne by the employer but not the fam-
ily.) As before, lost wages were calculated by multipli-
cation of the annual earnings of the patient by 1/52
times the number of weeks of lost income.
The four categories of cost just discussed do not
encompass all components of cost to a family. For in-
stance, if other family members reduce their paid em-
ployment as a result of the cancer in the family, this
would not be captured by these measures. Furthermore,
there is no accounting of such elusive consequences as
promotions that might have been lost because of re-
duced work effort or a decline in home-production of
goods and services not related directly to patient care
(e.g., an increase in restaurant expenditures resulting
from reduced time available for cooking). Finally, no
attempt has been made to monetize the psychosocial
costs of the illness, which include reduced patient com-
fort and likely family stress. However, the list of cost
components used in this study clearly goes beyond the
customary efforts of only counting direct cash outlays
and may be viewed as a first (conservative) step toward
approximating the total cost to families involved in car-
ing for a patient with cancer.
Results
The data presented in Table 2 show the cost estimates
of each of the four cost components for the 3-month
period before the interviews with the patients and care-
givers. Loss of earnings for patients appeared to be the
least important cost, averaging only $131 in this sam-
ple. However, 95% of the patients in this sample did
not report any forgone earnings. For those who did (10
patients, or 5%), the losses amounted to an average of
$2524 or 37% of their total family costs. (Earnings of
the 10 patients involved ranged from $2500 to $37,500,
with a mean of $13,800.) Similarly, the lost earnings of
caregivers amounted to an overall sample average of
$311, but for the 26 families (14%) in which caregivers
gave up paid employment, this loss accounted for an
average of $2299, representing 42% of their total family
costs. (Earnings of the 26 caregivers involved ranged
from $12,500 to $60,000, with a mean of $35,192.)
Clearly, income losses due to reduced employment oc-
cur only in a minority of families of patients with
cancer, especially because cancer is much more likely to
occur during retirement age. (Overall cancer rates in the
population older than 64 years of age are 10 times
Cancer Home Care Costs/Stonrmel et al. 1871

Table 2. Cost of Care to Families With a Member With Cancer During the 3-Month Period
Before the Interview (n = 192)
Cost category as
Percent of
percent of total
Standard families
Mean deviation Range with cost Mean Range
Categories of cost'
(1) Out of-pocket expenditures $660 $624 $60-$3380 100 29 0.5-100
(2) Carpgiver labor cost $2612 $3091 $0-$15,000 80 49 0-99
($3253) ($38 3 6) ($0-$15,000)
Of which:
lost earnings $311 $1400 $0-$15,000 14 6 0-96
(3) Family labor cost $1160 $2517 $0-$15,656 46 20 0-97
($1483) ($3219) ($0-$18,200)
(4) Patient loss of income $131 $811 $0-$9375 5 2 0-94
(5) Total cost to family $4563 $4413 $60-$20,195 100 100
($5527) ($5447) ($60-$25,340)
Cost of 3-month stay in Michigan $5704 $3235-$10,038
nursing home

Correlations between percentages of individual cost categories and total cost

Out-of -pocket Caregiver lost Patient lost
expenditures Caregiver labor cost earnings Family labor cost earnings
(Yo of total) (YO of total) (YO of total) (YO of total) (Yo of total)

Total cost -0.49t 0.21t 0.07 0.22t 0 09

* Computations of caregiver and family labor costs are based on assumed hourly costs of $7 82 (or $10 00).
tP<@@1

higher than among people younger than 65 years of
age.'') However, when such income losses do occur,
they represent a substantial part of the costs to the fami-
lies.
Out-of-pocket expenditures for services and sup-
plies have been studied more often in previous studies.
In this sample, out-of-pocket expenditures were in-
curred by all families. However, on average, these costs
have only a moderate impact on the families of patients
with cancer: mean expenditures amounted to $660 over
3 months, although for 11.5% of the families, out-of-
pocket expenditures exceeded $1500, up to a maximum
of $3380. Such skewed distributions of expenditure
data with few high spenders and many families spend-
ing moderate sums are quite common'' and can be ex-
pected in a sample of patients in which some may be in
a terminal stage. More noteworthy is that, with higher
overall costs to families, cash expenditures decline
sharply as a proportion of total family costs (r = -0.49
in Table 2). This linear association between total family
costs and the proportion of costs accounted for by cash
expenditures actually underestimates the strength of
the relationship. As Figure 1 shows, cash expenditures
represent a large percentage of costs only when overall
family costs are less than $2000. (A square root trans-
formation of the total cost variable changes the correla-

E A 0 60%

7
x s F 55%'
P 50%
E P T 45%
N E 0 40%
D R T 35%
I C A 30%
T E L 25%

Figure 1. Unreimbursed out-of-

R T C
E
s
0
s
T
5%
O%u
under
1 2000 4000 6000 8000 10000 12000 14000 16000 over
pocket expenditures as a percentage 2000 -4000 -6000 -8000 -10000 -12000 -14000 -16000 -18000 -18000
of total cost of cancer home care to
families. TOTAL COST OF CANCER HOKE CARE TO FAMILIES
1872 CANCER March 1, 2993, Volume 71, No. 5

tion to -0.63.) Thus, higher home care costs to families
usually are not the result of larger cash expenditures.
By far the largest cost component is the labor cost of
primary caregivers (mean, $26 12), accounting, on aver-
age, for 49% of all family costs. This result is even more
remarkable because approximately 20% of these care-
givers do not appear to be engaged in any “hands-on”
caregiving tasks. In case other family members provide
direct care (which occurred in 54% of the families of
this sample), the market value of their contribution is
also substantial (mean, $2162) and accounts for 37% of
the total family costs in those families. The monetiza-
tion of these labor services offers a clear demonstration
of the value of the services provided by families. It is
this component of the costs of care that usually is
treated as a free good.
Because the estimates of total costs of cancer care
for the families rely heavily on labor costs, we provide
one alternative calculation based on an assumed market
value of $10.00 per hour of caregiving activity. As the
figures in parentheses in Table 2 indicate, this increase
in the assumed labor cost by 28% (from $7.82 to
$10.00) leads to an increase in estimated total costs by
21% (from $4563 to $5527). The sensitivity of the esti-
mates of total family costs to changes in assumed labor
costs needs to be taken into account when comparing
the costs of cancer home care by families with the costs
of nursing home care. Nonetheless, it is instructive to
take a look at the costs of nursing home care. In the state
of Michigan, the average annual cost (1989) of a nurs-
ing home bed (in mostly intermediate-care facilities)
was $22,813, starting from a low of $12,939 to a high of
$40,150.21This translates into average 3-month costs of
$5704 per nursing home resident in Michigan. Al-
though the similarity to the family care costs is striking,
it is important to observe the much greater range in the
costs of care to the families: $60-20,195 for families
versus $3235-10,038 for nursing home residents.
In a preliminary effort to identify at least some cor-
relates of family care cost, two types of variables were
introduced as possible predictors of family care cost: (1)
medical status indicators such as primary site of cancer,
current treatment modality, and time elapsed since the
diagnosis of the cancer; and (2) indicators of long-term
care needs such as patient functional status and family
living arrangements (see Table 1 for the descriptive sta-
tistics).
Table 3 shows the results from a regression analy-
sis, with all categoric variables (primary site, treatment,
living arrangement) converted to dummy variables2’
Although the dependent variable of total family cost
has a skewed distribution (skewness = 1.61), examina-
tion of the residuals did not lead to the conclusion that
the normality assumptions were violated severely. In
Table 3, the unique effect of each independent variable
is displayed. (For the categoric variables-primary site,
current treatment, or living arrangement-the effects of
the associated dummies are grouped and reported as a

Table 3. Patient Conditions and Living Arrangements as Predictors of Cost

of Family Care During a 3-Month Period (n = 191)
Dependent variable
Total cost of family care
Sum of: out-of-pocket expenditures, caregiver labor costs, family labor cost, patient loss of income
DF F ratio Significance of F
Independent variables
Primary site 7 0.65 0.732
Type of treatment 8 0.31 0.962
Time since diagnosis 1 1.80 0.182
Dependencies in ADL 1 20.28 0.000
Living arrangement 2 7.15 0.003
YO Variance accounted for (R’) 21.0%

Mean family cost by living arrangement*

Out-of-pocket Caregiver Family labor Patient loss Total
exvenditures labor cost cost of income cost
Living arrangement
(1) Patient + caregiver $622 $2792 $616 $27 $4069
(2) Patient, caregiver, and others $664 $2624 $1667 $181 $5121
+
(3) Patient caregiver live apart $913 $1587 $2592 $587 $5683
Significance level (Univ. F test) 0.048 0.019 0.001 0.029 0.025
DF: degrees of freedom; ADL: activities of daily living.
* Adjusted for the effects of ADL dependencies of patients.
Cancer Home Care CostslStommel et al.
single F-test.) Most interesting among the results is the
apparent fact that the type of cancer, current treatment
modality (chemotherapy, surgery, radiation therapy, or
any combination thereof ), or the time elapsed since the
cancer diagnosis did not have any influence on the cost
of care to the families.This clearly differs from the wide-
spread finding’ that the medical costs of treating cancer
vary according to diagnosis and indicates that the costs
of caring borne by families do not present a mirror
image of the costs incurred in the medical care system.
As it turns out, the strongest predictor of the family
cost of cancer care is the functional limitation of the
patient in terms of ADL. This highly significant predic-
tor alone accounts for more than 10% of the variance in
family cost. This result is, of course, consistent with
much of the caregiving literature that shows ADL de-
pendencies to be the best predictor of both service use
and informal a s ~ i s t a n c e . Finally,
~ ” ~ ~ the living arrange-
ments of the family also appear to influence their cost of
care. As the means (adjusted for the effects of differ-
ences in ADL dependencies) in the second part of Table
3 seem to indicate, family costs are highest when care-
giver and patient live in different households and low-
est when they live together with no one else present.
The reasons for the high costs of living apart include the
following: highest out-of-pocket expenditures and
highest family labor contribution. Reasons for the lower
costs of couples living alone together in the same house-
hold include the following: small income losses result-
ing from reduced employment (a disproportionate num-
ber of these couples are of retirement age) and little help
from other family members. To interpret the latter cir-
cumstance as a cost saving is somewhat problematic,
however. If other family members are not willing or
able to provide much care for these couples, this simply
may mean that the needs of the patients are not at-
tended adequately. This, in itself, is a cost not captured
by the measures used here.
Conclusions
The analysis and data presented in this article represent
a first step in exploring the cost of cancer care to fami-
lies. Differing from the previous literature on this sub-
ject, our approach has been to monetize the informal
labor care contributions of family members and the pri-
mary caregiver. When added to out-of-pocket expendi-
tures and lost earnings, estimates of the market value of
the family and caregiver labor yield a more comprehen-
sive picture of the total costs incurred by families with a
member with cancer. To be sure, not all opportunity
costs are included in this article. We also did not attempt
to estimate the costs associated with diminished utility
(quality of life) resulting from the illness because mea-
surement difficulties still pose a substantial problem in
1873
addressing these issues. As the literature on the costs of
acquired immune deficiency syndrome shows, studies
are usually weakest in assessing indirect costs, includ-
ing family labor
One result of our investigation indicates that varia-
tion in home care costs seems to be much greater than
the variation in the cost of nursing home care. This is
not surprising because the cost of nursing home care
contains substantial fixed cost components such as
shelter and minimum 24-hour personnel. However, it is
also likely that the smaller variation in nursing home
costs results from the restricted (higher) range in ADL
dependencies among nursing home residents. By con-
trast, a major reason for the low home care costs of
some patients with cancer is that they can carry on an
almost normal life and certainly do not need much tan-
gible care. The data presented provide little evidence
that the costs of home care vary systematically by type
of cancer. It is possible that the absence of detailed in-
formation on stages of the cancer has contributed to this
finding. Bloom et aL6 had found the cost of cancer care
and treatment to vary with cancer diagnosis, prognosis,
and year since diagnosis. However, most of their cost
data refer to the costs of medical treatment; only their
out-of-pocket expenditure category is comparable to
the one presented here. In any case, it is by no means
clear why one should expect family care costs to vary
according to diagnostic criteria that are indispensable
tools for medical care but do not translate easily into
home care needs.
If the estimates of the family costs of cancer care
provided here can be confirmed in larger, representa-
tive samples, they point to a potentially growing prob-
lem. Home care appears relatively inexpensive com-
pared with institutional care because the labor contri-
bution of the family (and the cost of shelter) usually is
not counted in comparisons with institutional care.
Given the aging of the cancer population and the con-
tinuing trend toward smaller households,5 the impor-
tant policy question is whether we can continue to rely
on family labor to cover a growing need for home care.
With cancer rapidly developing into a continuous care
problem because of increasing incidence rates, longer
average survival lengths, and a trend toward outpatient
treatment, a realistic appreciation of the costs involved
in caring for a patient with cancer at home seems neces-
sary.
References
1. Scheffler RM, Andrews NC, editors. Cancer care and cost:
DRGs and beyond. Ann Arbor (MI): Health Administration
Press, 1989.
2. Hodgson TA, Meiners MR. Cost-of-illness methodology: a
guide to current practices and procedures. Milbank Q 1982;
60:429-62.
1874 CANCER Murch 2, 2993, Volume 71, No. 5
3. Becker GS. A theory of the allocation of time. Ecor?onric1 1965;
75:493-517.
4. Jacobs P, McDermott S. Family caregiver costs of chronically ill
and handicapped children: method and literature review. Public
Health Rep 1989; 104:158-63.
5. Stommel M. Patient position, family structure, and the burden
of cancer care to families. lnvitational Workshop on the Costs of
Continuing Cancer Care for American Families; 1992 April 22-
24: Rockville (MD): Applied Research Branch of the National
Cancer Institute, 1992.
6. Bloom BS, Knorr RS, Evans AE. The epidemiology of disease
expenses: the costs of caring for children with cancer. I A M A
1985; 253:2393-7.
7. Bodkin CM, Pigott TJ, Mann JR. Financial burden of childhood
cancer. Br Med ] 1982; 284:1542-4.
8. Houts PS, Lipton A, Harvey HA, Martin B, Simmonds MA,
Dixon RH, et al. Nonmedical costs to patients and their families
associated with outpatient chemotherapy. Cancer 1984;
5312388-92.
9. Lansky SB, Cairns NU, Clark GM, Lowman J, Miller L, True-
worthy R. Childhood cancer: non-medical costs of the illness.
Cancer 1979; 43:403-8.
10. Kemper P. Living arrangement and the demand for home care:
an illustrative theoretical model. Rockville (MD): Agency for
Health Care Policy and Research; 1989 Reprint: U.S. Depart-
ment of Health and Human Services.
11. Manton KG. Epidemiological, demographic, and social corre-
lates of disability among the elderly. Milhank Q 1989; 67(2
SUPPI,Pt 1):13-58.
12. McCorkle R, Quint-Benoliel J . Symptom distress, current con-
cerns, and mood disturbance after diagnosis of life-threatening
disease. Sac Sci Med 1983; 7:431-8.
13. Cox BG, Cohen SB. Methodological issues for health care sur-
veys. New York: Marcel Dekker, 1985.
14. Lansing JB, Morgan IN. Economic survey methods. Ann Arbor
(MI): University of Michigan, 1971.
15. Rossi PH, Wright JD, Anderson AB. Handbook of survey re-
search. Orlando (FL): Academic Press, 1983.
16. Anderson RA. Total survey error. San Francisco: Jossey-Bass,
1979.
17. Matthews SH. Provision of care to old parents: division of re-
sponsibility among adult children. Res Aging 1987; 9:45-60.
18. Penning MJ. Receipt of assistance by elderly people: hierarchical
selection and task specificity. GeronfoIogist 1990; 30:220-7.
19. Gloeckler-Ries LA, Hankey BF, Edwards BK, editors. Cancer
statistics review 1973-1987. Bethesda (MD): National Cancer
Institute; 1990 NIH Publication No. 90-2789.
20. Berki SE, Wyszewianski L, Magdavy LJ, Lepkowski JH. Families
with high out-of-pocket health services expenditures relative to
their income. Ann Arbor (MI): University of Michigan, School of
Public Health; 1985 Final report for NCHSR Contract #233-81-
3032.
21. Zonia S. The effects of nursing home decertification on resi-
dents. East Lansing (MI): Michigan State University, Social
Science Research Bureau, 1991.
22. Neter J, Wasserman W, Kutner MH. Applied linear statistical
models. 2nd ed. Homewood (IL): RD Irwin, 1985.
23. Horowitz A. Family caregiving to the frail elderly. Ann Rev Ger-
ontol Geriatr 1985; 5:194-246.
24. Wolinsky FD, Arnold CL. A different perspective on health and
health services utilization. AJV?Rev Gerontol Geriatr 1988; 8:71-
101.
25. Bennett CL, Cvitanic M, Pascal A. The costs of AIDS in Los
Angeles. Iournal of Acquired Inrmune Deficiency Syndrome 1991;
4: 197- 203.