- Email: [email protected]
Economic valuation of informal care: Lessons from the application of the opportunity costs and proxy good methods
ARTICLE IN PRESS
Social Science & Medicine 62 (2006) 835–845 www.elsevier.com/locate/socscimed
Economic valuation of informal care: Lessons from the application of the opportunity costs and proxy good methods Bernard van den Berga,, Werner Brouwerb,c, Job van Exelb,c, Marc Koopmanschapb,c, Geertrudis A.M. van den Bosd, Frans Ruttenb,c a
Centre for Health Economics Research and Evaluation, Faculty of Business, University of Technology, Sydney b Department of Health Policy and Management, Erasmus MC, Rotterdam, The Netherlands c Institute for Medical Technology Assessment, Erasmus MC, Rotterdam, The Netherlands d Department of Social Medicine, Academic Medical Centre, University of Amsterdam, The Netherlands Available online 30 August 2005
Abstract This paper reports the results of the application of the opportunity costs and proxy good methods to determine a monetary value of informal care. We developed a survey in which we asked informal caregivers in The Netherlands to indicate the different types of time forgone (paid work, unpaid work and leisure) in order to be able to provide care. Moreover, we asked informal caregivers how much time they spent on a list of 16 informal care tasks during the week before the interview. Data were obtained from surveys in two different populations: informal caregivers and their care recipients with stroke and with rheumatoid arthritis (RA). A total of 218 care recipients with stroke and their primary informal caregivers completed a survey as well as 147 caregivers and their care recipients with RA. The measurement of care according to both methods is more problematic compared to the valuation. This is especially the case for the opportunity costs method and for the housework part in the proxy good method. More precise guidelines are necessary for the consistent application of both methods in order to ensure comparability of results and of economic evaluations of health care. r 2005 Elsevier Ltd. All rights reserved. Keywords: Informal care; Economic evaluations; Opportunity costs method; Proxy good method; Stroke; Rheumatoid arthritis; The Netherlands
Introduction Informal care plays a substantial role in the total care provided, especially in cases of care for people with chronic diseases and the terminally ill (Norton, 2000). Corresponding author. Centre for Health Economics
Research and Evaluation, Faculty of Business, University of Technology, Sydney, PO Box 123, Broadway NSW 2007, Australia. Tel.: +61 2 9514 4753; fax: +61 2 9514 4730. E-mail address: [email protected] (B. van den Berg).
Because caregivers sacrifice (amongst other resources) time to provide care, informal care should be incorporated in an economic evaluation taking a societal perspective (Drummond, O’Brien, Stoddart, & Torrance, 1997; Luce, Wanning, Siegel, & Lipscomb, 1996). Despite the recommendation to include informal care in economic evaluations, in practice it is often neglected (Stone, Chapman, Sandberg, Liljas, & Neumann, 2000). It is quite common to consider informal care as a cost in an economic evaluation and it is therefore suggested to incorporate changes in the use of informal caregiver time as direct non-health care costs into the numerator of the
0277-9536/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2005.06.046
ARTICLE IN PRESS 836
B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
cost-effectiveness ratio (Luce et al., 1996, p. 177).1 This implies that informal caregivers’ time should be valued in monetary terms. It is often recommended to use either the opportunity costs method or the proxy good method (also known as the replacement cost method) to value the time investment in informal care (Drummond et al., 1997; Luce et al., 1996; Posnett & Jan, 1996). Both methods have their strengths and weaknesses (McDaid, 2001; Van den Berg, Brouwer, & Koopmanschap, 2004). However, from a theoretical point of view the opportunity costs method is preferred (Posnett & Jan, 1996). Although alternative monetary valuation methods such as contingent valuation and conjoint measurement are proposed and applied to value informal care (Van den Berg, Al, Brouwer, Van Exel, & Koopmanschap, 2005; Van den Berg, Bleichrodt, & Eeckhoudt, 2005; Van den Berg, Brouwer, Van Exel, & Koopmanschap, 2005), the opportunity costs and proxy good methods are most advocated and most often used. One important reason for recommendations to use either one of these methods may be their relatively straightforward application. In economic evaluations, where the focus is on the care recipients rather than on informal caregivers, this may be considered an advantage. Informal care in practice is often neglected in economic evaluations where it is an important input. This may be due to various factors which include: (1) informal care is simply overlooked; (2) it is not overlooked but disregarded; (3) many health technology assessment guidelines recommend conducting economic evaluations from more narrow perspectives than the societal perspective or (4) informal care is considered relevant but researchers may have difficulties with measuring or valuing it, because guidelines and handbooks are quite short about these issues and recommended valuation methods are less straightforward to apply than they first appear. In terms of measurement of informal care as an input in health care, some important problems exist. One problem is the difficulty in measuring time forgone in order to provide informal care. Especially, when proving care started many years ago, as is often the case in chronic diseases, the normal activities forgone are difficult to indicate for caregivers. Another problem concerns the distinction between ‘‘normal’’ housework and additional housework due to the health problems of the care recipient. If this distinction is not properly made it is easy to overestimate the time spent on informal care. Regarding the valuation of informal care, it may be difficult to find appropriate opportunity costs estimates for all different time uses and groups of caregivers. In 1 This is not problematic unless informal care is the focus of the intervention for instance respite care programs for informal caregivers, see, e.g. (Mohide, Torrance, Streiner, Pringle, & Gilbert, 1988 or Drummond et al., 1991).
using the proxy good method, problems may arise in finding appropriate wages of professional substitutes who might perform the care activities if no informal caregiver was available. In this paper we discuss the application of the opportunity costs method and proxy good method in two caregiver populations—informal caregivers of care recipients with stroke and caregivers of care recipients with rheumatoid arthritis (RA). Our aim is to assess the costs of informal care in these two populations using both the opportunity costs and proxy good methods. Moreover, we wish to detect the major problems in using these often recommended methods. Application of these two methods in such distinct populations is useful in this context. Stroke is an acute condition with a clear starting point, while RA is a slowly progressive chronic disease without a clear starting point. A starting point is important for the measurement of time forgone and time spent on informal care and therefore has important implications for the application of the opportunity costs method and proxy good method. We also propose solutions for the problems in measuring time forgone when a clear starting point is unavailable and for the distinction between ‘‘normal’’ housework and informal care.
Opportunity costs and proxy good methods Opportunity costs method Conceptually, the opportunity costs method values the inputs of the production process. However, in practice it often values informal care according to Eq. (1): Value informal care ¼ ti wi ,
(1)
where ti is the time spent on care tasks by caregiver i, and wi the net market wage rate of i. If i is unemployed, a proxy for wi is used, e.g., a modified opportunity costs method to find out the reservation wage of the caregiver: the wage rate for what an individual is willing to supply at least 1 h on the labour market (Kooreman & Wunderink, 1996, p. 113). Another solution is the imputation of the actual wage of similar individuals (for example those of the same gender, education and age). Eq. (1) implies that the value of leisure and unpaid work equal the (would be) wage rate. But, when caregivers e.g. derive disutility from work, the shadow price of unpaid work may differ from the marginal wage, see Posnett and Jan (1996) for a detailed discussion. Therefore, it would be better to distinguish between different sources and amounts of time forgone, e.g., according to Eq. (2): Value informal care ¼ ni wi þ hi si þ l i ti ,
(2)
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
where ni is the i’s hours of forgone paid work, hi the i’s hours of forgone unpaid work, si the shadow price of unpaid work, li the i’s hours of forgone leisure, and ti the shadow price of leisure. According to Eq. (2), the amount and sources of time forgone should be measured instead of just the time spent on caregiving. In addition, shadow prices for unpaid work and leisure need to be determined, which poses another challenge. Often, these shadow prices are based on (an arbitrarily adjusted) wage rate, which makes the distinction between Eqs. (1) and (2) rather cosmetic. For the measurement of time according to Eq. (2), one would ideally use panel data. This is often not feasible and sub-optimal solutions are required, e.g., asking respondents how their time allocation has changed since engaging in informal care or comparing their time allocation to that of a comparable sample from the general public.
837
Both methods differ conceptually: the opportunity costs method valuing inputs and the proxy good method valuing output. Therefore, from a theoretical point of view, the opportunity costs method is the benchmark. If caregivers derive direct utility from caring, the proxy good method involves an underestimation.
Applying the methods in two populations The populations were approached as part of larger studies: an evaluation of stroke units for care recipients with stroke and a study on health and health care utilization among care recipients with RA. Some of the information gathered in the two studies was therefore not fully symmetrical, but this mainly pertains to additional information. Populations
Proxy good method The proxy good method, also called the market cost method or replacement costs method is an alternative for the opportunity costs method. It values time spent on caregiving at the (labour) market price of a close substitute: conceptually it values the output of production. This requires the availability of a market substitute for the non-market commodity, which is assumed to be almost perfect, e.g. same quality. Informal care time is e.g. valued at the wage rate or market price of a professional caregiver. The value can differ per task: e.g. housework valued at the market wage of a housekeeper and personal care valued at the market wage of a nurse.2 The measurement of time spent on informal care, instead of the time forgone, is the cornerstone of the application of the proxy good method. Two important measurement methods are the diary and recall method. Because a diary, the gold standard (Juster & Stafford, 1991, p. 473), is time-consuming for respondents, a recall method is often applied. Respondents are retrospectively asked how much time they spent on different care tasks during a certain time period. To ensure comparison of results between different populations, it is important to standardize the concept of informal care, e.g. due to the development of a standard list of care tasks. If one defines only additional housework as part of caregiving, as would be reasonable, it is crucial to distinguish ‘‘normal’’ housework from additional housework due to caregiving. This is troublesome when the caregiver and care recipient live together or if care has been provided for several years already (as is often the case in RA). 2
One can debate whether this should be the gross wage (the real opportunity costs to society) or the net wage (the wage rate for which the professional is willing to sacrifice leisure).
Informal care for care recipients with stroke The stroke data was collected as a supplement to the EDISSE study (Huijsman et al., 2001). Care recipients with stroke were included at hospital admission and followed for a period of 6 months. They were asked whether or not they received informal care. If they received informal care, we asked them to indicate their primary caregiver and caregiver was asked to fill in a written survey. If the caregiver was not present, a survey was left behind which could be posted at a later date. About 597 stroke care recipients were included in the EDISSE study. Over 181 care recipients died in the period up to 6 months after stroke and 336 of the care recipients were discharged. A total of 255 caregivers completed the survey. Due to item non-response, especially on care time questions, we used a net sample of 218 stroke-caregivers. Table 1 provides the descriptive statistics of caregivers and their care recipients. It shows that the majority of caregivers are females mostly caring for their partners and 42% have a paid job. Stroke care recipients have mean EQ-5D scores of 0.47. Informal care for care recipients with RA The data for the RA part of this study were collected as a supplement of the RA+ study, a panel study on health and health care utilization among people with RA (Jacobi, Triemstra, Rupp, Dinant, & Van den Bos, 2001; Jacobi et al., 2003). In the 2001 wave, 365 of 683 care recipients indicated they received informal care. We approached the 365 receiving informal care and asked them to pass our mail survey on to their primary informal caregiver. We also asked care recipients to complete a mail survey and included a question for the 318 care recipients without informal care asking if they currently received informal care. If so, we asked them to
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
838
Table 1 Characteristics informal caregivers and their care recipients Characteristics
Stroke (n between brackets)
RA (n between brackets)
Informal caregivers Age (mean) Female (%) Partner (%)
58.5 (218) 61.6 (218) 54.2 (218)
62.4 (147) 23.7 (147) 93.2 (147)
Education (%) Primary school Lower vocational Medium vocational Higher vocational University Duration of care (mean)
11.8 (218) 44.1 (218) 27.2 (218) 12.7 (218) 4.2 (218) n/a
13.3 45.1 28.3 9.1 4.2 11.5
24.2 (218) 22.4 (218)
17.7 (147) 58.5 (147)
Occupation (%)a Housework Disability pension/ retired Paid job Other Monthly income (mean)b Care recipients Age (mean) Female (%)
41.8 (218) 11.6 (218) 1635.03 (67)
72.0 (218) 57.5 (218)
Education (%) Primary school Lower vocational Medium vocational Higher vocational University
n/a
Occupation (%) Housework Disability pension/ retired Paid job EQ-5D (mean) EQ-VAS (mean) Professional care (%) Waiting list professional care (%) Other informal care (%)
n/a
(147) (147) (147) (147) (147) (147)
36.7 (147) 4.2 (147) 1506.55 (114)
62.1 (147) 83.9 (147) 20.7 52.6 16.4 6.7 3.6
(147) (147) (147) (147) (147)
older than the stroke caregivers and more than 90% are the partners of the care recipients. Less RA caregivers have a paid job compared to those caring for people with stroke. The duration of providing care for RA care recipients is 11.5 years, while the disease duration was 13.0 years. This indicates that providing care often starts soon after the diagnosis of RA. For stroke we assume that providing care starts directly after the diagnosis.3 The EQ-5D scores of the RA care recipients are with 0.48 similar to the stroke care recipients. Care recipients’ partners were predominantly male in the case of RA compared with mainly females in case of stroke. The latter is quite common (caregivers are often females) and the former reflects the relatively high prevalence of RA in females. The valuation methods Opportunity costs method We asked informal caregivers retrospectively what types and amount of time of paid work, unpaid work and/or leisure they gave up in order to be able to provide care. The validity of these questions increases if they are asked soon after the caregiving episode started. This makes a comparison between stroke and RA without a clear starting point particularly interesting. As we anticipated the absence of a clear starting point the case of RA, we also asked some additional questions. These questions pertain to the best alternative time use, not retrospectively, but in terms of which activity the respondents would prefer to spend time on as well as how many hours a week if they could reduce their caregiving time. These questions were phrased as follows: Suppose, you do not have to spend time on caring anymore: Would you prefer to spend this time on paid, unpaid work or leisure and how many hours per week per category?
38.9 (147) 51.8 (147)
0.47 (218) n/a n/a n/a n/a
15.4 0.48 55.61 26.1 5.9
(147) (147) (147) (147) (147)
68.0 (147)
a In case of RA, the percentages add up to over 100% due to the fact that some respondents reported different occupations. b Note that in case of stroke the net monthly income is private income, while in case of RA the net monthly income is family income.
hand a mail survey over to their primary informal caregiver. Table 1 also shows the characteristics of the net sample of 147 RA caregivers. RA caregivers are slightly
Proxy good method There are no instruments that distinguish between ‘‘normal’’ housework and additional housework due to caregiving. Existing instruments, e.g. caregiver activities time survey (CATS) (Clipp & Moore, 1995), caregiver activity survey (CAS) (Davis et al., 1997) and resource utilization in Dementia (RUD) (Wimo, von Strauss, Nordberg, Sassi, & Johansson, 2002), seem to neglect this difference. We asked caregivers to report whether, and if so, how much time they spent on a list of 16 activities (see Table 3 for the complete list) in the week preceding the interview. We distinguished between (1) 3
Information about co-morbidities is lacking, so we have to assume that the provision of informal care is due to stroke or RA.
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
household activities of daily living (HDL), (2) activities of daily living (ADL), and (3) instrumental activities of daily living (IADL). We also added a question about travel time due to caregiving, because this might be an important aspect of total time use.4 In order to derive the time spent on different care tasks, we asked caregivers two different types of questions. First: ‘‘How much time did you spend on assistance with y’’? These questions pertained to support activities, e.g., assisting the care recipient in visiting the toilet or with mobility outside. Given their nature, answers to these questions would inevitably indicate time spent on caregiving (correct responses assumed for the moment). Secondly, we asked them questions like: ‘‘How much time did you spend on y’’? These questions mainly pertained to time spent on different types of housework. Because our study focussed on time investment in the context of informal care, we were sceptical about the respondents’ ability to separate ‘‘normal’’ from additional housework, particularly for caregivers living with the care recipients. But also measuring additional housework for caregivers not living with care recipients may be troublesome e.g. separating shopping time due to caregiving from shopping for their own purposes. Comparing the indicated time spent on housework of caregivers with that of the general population could indicate whether or not caregivers are expected to have indicated ‘additional time’ spent on these tasks or rather ‘normal time’. Information on time allocation of the Dutch general public was derived from the Dutch TimeAllocation Survey 2000 (TBO’2000) (SCP, 2000). The data were collected using time budget diaries. Respondents (n ¼ 1813) reported their time spent on a broad range of activities including HDL tasks every 15 min per day. With the results from TBO’ 2000, we forecast the expected time spent on several HDL activities in our two populations corrected for age and gender to get an idea of the correctness of the answers provided in our sample. Finally, one could argue that the time invested in caregiving should equal the total time forgone due to providing care. A possible difference may be due to the fact that one of the two methods is easier to complete. It is also possible that this is due to the neglect of joint production in one way of questioning or because the figures do not adequately reflect that certain household tasks (e.g. house maintenance) are sacrificed to perform more urgent household tasks (cleaning or cooking). Possible differences between the two populations may 4
Some questions had answer categories in minutes per day, while others had answer categories in hours per week, depending on the expected weekly time investment.
839
also have to do with the starting point, which is clear in stroke while it is lacking in RA or with a relatively long caregiving history. Results from the two valuation methods Opportunity costs—measuring time forgone We distinguished three types of time forgone in order to be able to provide care: paid and unpaid work and leisure. Table 2 presents types and time forgone in both populations. Firstly, it shows the percentage of respondents involved in the three different categories before they became a caregiver, just like the mean hours per week in column 3. Next, it gives the percentage of respondents that gave up paid work, unpaid work or leisure in order to provide informal care. Column five provides the total number of weekly hours forgone per activity. The results show that 16.5% of stroke caregivers with paid work reduced their time spent on paid work, resulting in a mean reduction of 3 h paid work per week. RA caregivers with paid work indicated a slightly lower amount of paid work forgone, that is, 2.2 h/week, whereas the amount of unpaid work forgone was higher (1.2 versus 0.6 h/week). In the case of stroke we collected no information about the amount of leisure before the informal care episode started, because of expected difficulties in the measurement of leisure time. Based on the stroke experience, we redesigned the survey (just a question with one table involving a complete overview of the different sources and amount of time forgone instead of complicated routing via many different questions) and included the question on the amount of leisure before providing care. A majority of RA caregivers indicated they spent time on leisure before the caregiving episode started. Assuming that all stroke caregivers enjoyed leisure before becoming a caregiver, they indicated more often than the RA group that they had given up leisure in order to provide informal care (37.6% versus 25.2%). Moreover, the average number of hours of leisure forgone was higher in stroke (8.8 versus 6.1). In total, stroke caregivers indicated higher opportunity costs than RA caregivers: 12.4 versus 9.5 h/ week. It is worth noting that Table 2 indicates increasing item non-response when asking about numbers of hours forgone, i.e. n ¼ 30 in case of RA. As mentioned above it could be difficult for caregivers to indicate the amount and sources of time forgone if a clear starting point is lacking as in RA, because the caregiving episode may have started many years ago, e.g. before retirement. Moreover, the number of care tasks provided as well as the time spent on caregiving may slowly increase, without clear starting points. Such increases may go unnoticed. This makes it expectedly difficult for the RA respondents to indicate time spent on other activities forgone. We tested for this hypothesis
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
840
Table 2a Informal caregivers’ opportunity costs of time Percentage respondents performing activity before caregiving episode
Mean hours a week before (n between brackets)
Percentage respondents having forgone activity due to informal care
Mean hours a week forgone (n between brackets)
Stroke Paid work Unpaid work Leisure Total (218)
44.5 13.3 n/a 47.2
27.7 (103) 1.7 (103) n/a 29.4 (103)
16.5 9.7 37.6a 67.7
3.0 0.6 8.8 12.4
(70) (70) (70) (70)
RA Paid work Unpaid work Leisure Total (147)
36.7 25.2 60.5 72.8
15.9 (107) 2.5 (107) 7.2 (107) 25.6 (107)
8.4 9.3 25.2 28.0
2.2 1.2 6.1 9.5
(30) (30) (30) (30)
Sub samples and activities (n between brackets)
a
Percentage based on total group (n ¼ 218).
Table 2b Alternative question posed to RA informal caregivers Activities (n between brackets)
Percentage respondents indicating they would spend freed time on activity
Mean hours a week (n between brackets)
Paid work Unpaid work Leisure Total (147)
6.8 9.5 61.9 68.0
0.6 1.0 5.7 7.3
(100) (100) (100) (100)
by means of a simple correlation coefficient between respondents indicating both their opportunity costs of time and the number of years they had provided care. The relatively strong correlation coefficient of 0.50 supports this hypothesis. Because we expected difficulties in indicating their opportunity costs of time, we also asked RA caregivers which activities they would spend their time on if they no longer had to fulfil care tasks. Table 2b shows the results. Most caregivers preferred to spend freed time on leisure. The percentages of caregivers indicating that they preferred to spend freed time on paid work or unpaid work resemble those in Table 2a. For leisure however, there is a substantial difference compared to the results reported in Table 2a. In terms of hours per week, both methods yield similar results for unpaid work and leisure. However, the amount of paid work is in the alternative question (Table 2b) lower compared with Table 2a. This might be related to the fact that some caregivers have retired within the long time
interval of 11.5 years between the start of providing informal care and date of survey completion. Therefore, although the alternative method may be useful in a context of long term care and slowly progressive diseases, the validity of provided answers remains to be established. Proxy good method—measuring time investment In applying the proxy good method, time investments on different care tasks needs to be assessed. We distinguished HDL, ADL and IADL tasks and travel time. Table 3 presents the results both in percentages of caregivers performing specific activities as well as the number of hours per week they spent on these activities. Most caregivers performed HDL tasks, and the time spent on these tasks was relatively large. The majority of caregivers were also involved in IADL tasks, yet the time involved in these tasks was substantially lower compared to HDL. About one third of the caregivers were involved in ADL tasks, which required around 2 h/ week. More stroke than RA caregivers had to travel, probably because more RA caregivers were partners of the care recipients. The overall percentage of caregivers performing tasks was similar in both groups, but the RA caregivers provided about 7 h more care per week. This difference is mainly related to the performance of HDL tasks, probably because more RA caregivers are men who indicate HDL more often as informal care compared to women. The large amount of time invested in HDL tasks may reflect the fact that ‘‘normal’’ HDL tasks are not fully separated from additional HDL tasks. Therefore, we compared the time allocation of the caregivers to that of the general public, using the results from the TBO’ 2000
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
841
Table 3 Informal care time in mean hours a week Strokea
Preparation of food and drinks Time investments in shopping, groceries etc. Housecleaning Washing, ironing or sowing Caring for and playing with own children Chores, gardening, maintenance HDL (Total)
Mean hours a week (n between brackets)
56.8 85.3 62.5 62.9 4.8 43.8
3.3 4.6 2.3 0.2 2.4 1.3
RAa
Mean hours a week (n between brackets)
76.7 87.2 86.0 43.0 11.3 74.2
5.0 6.3 2.0 0.9 4.2 2.9
89.0
14.0
94.1
21.2
18.2 8.5 18.6 19.2
0.6 0.1 0.4 0.5
34.9 6.7 17.7 10.3
1.3 0.1 0.5 0.2
ADL (Total)
32.9
1.6
37.3
2.1
Aiding the patient in travelling outside the house Aiding the patient with visiting and in excursions Aiding the patient in contacting health care suppliers Aiding patient in organising home adaptations, etc. Aiding patient in financial matters (insurance, rent)
38.5 45.5 59.7 34.4 55.1
0.6 0.9 0.6 0.2 0.6
38.8 51.4 60.5 23.2 39.7
1.1 1.2 0.9 0.1 0.4
IADL (Total) Travelling to and from care recipient Total
76.9 42.4 94.5
Aiding Aiding Aiding Aiding
a
patient patient patient patient
with personal care in visiting the toilet moving around within the house with eating and drinking
2.9 1.7 20.2 (218)
68.6 8.8 96.1
3.7 0.3 27.4 (147)
Percentage indicating that they spend time on the activity.
study described above. We predicted the time spent on different HDL tasks by Tobit-regression, because the time spent on an activity is constrained to be greater than zero (Scott Long, 1997). See Appendix A for the exact estimation results. Table 4 shows that in both samples total reported time spent on HDL is greater than the predicted HDL time. The total differences are significant. However, in the case of housecleaning, the reported time is less than the forecasted time. This suggests that caregivers re-allocate their time spent on housework due to their caregiving responsibilities: sacrificing around 2 h of housecleaning in order to be able to spend time on other activities. Comparing the two methods—measurement Looking at measurement of time, the proxy good and the opportunity costs methods yield different results. The average weekly time spent on caregiving is 20.2 h in the case of stroke and 27.4 h in the case of RA using the proxy good method (Table 3). The opportunity costs method yields substantially lower estimates: 12.4 and 9.5 h, respectively (Table 2a). The alternative opportunity costs used in the RA population leads to even lower time estimates (7.3 h). These substantial differences
demand more research in terms of validation of the measurement methods. Another important aspect of the comparison is the number of respondents that complete the questions. As can be derived from Tables 2 and 3, the measurement questions related to the proxy good method appear to perform much better than those related to the opportunity costs method. Money value using the opportunity costs method In the opportunity costs method, we use the caregiver’s hourly wage to value the provided care. According to Eq. (2), we also use hourly wages as shadow prices for unpaid work and leisure. Table 5 shows the wage rates, while those for RA are somewhat overestimated, because they represent hourly household income. The subgroup with available information about income from paid work is small (see Table 1). Combined with item non-response on the opportunity costs method time questions, this results in very small n’s (stroke: n ¼ 23; and RA: n ¼ 7) making weekly average costs for stroke of 204.64 euro and for RA of 49.18 euro. The opportunity costs method is often applied by combining
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
842
Table 4 Housework part of informal care Variable
Stroke forecast normal
Preparation of food and drinks Time investments in shopping, groceries etc. Housecleaning Washing, ironing or sowing Caring for and playing with own children Chores, gardening, maintenance Total N
Reported additional
Significance
RA forecast normal
Reported additional
Significance
1.9
3.3
o0.0001
2.5
5.0
o0.0001
0.8
2.4
0.0009
1.0
4.3
o0.0001
7.3 0.2 2.3
4.6 2.3 0.2
o0.0001 o0.0001 o0.0001
8.1 0.5 2.0
6.2 1.8 1.0
0.0071 o0.0001 o0.0001
0.5
1.0
0.0211
0.4
3.0
o0.0001
8.3 218
13.8
o0.0001
10.5 147
21.3
o0.0001
Table 5 Opportunity costs and proxy good methods compareda
Opportunity costs method
Proxy good method
a
Stroke (n between brackets)
RA (n between brackets)
Per hour Per week Sign.
Per hour Per week Sign.
17.34 (60) Diff. 0.00 Proxy good time with housework 17.34 (60) Diff. 0.90 Without housework 18.24 (218) Diff. 4.73 With housework 13.51 (218)
204.64 10.64 (23) (37) b 0.00 131.56 336.20 10.64 (59) (37) 216.40 o 0.0001 9.60 119.80 20.24 (n ¼ 218) (147) 119.44 o 0.0001 8.05 239.24 12.19 (218) (147)
49.18 (7) 129.66 178.84 (37) 25.33 153.51 (147) 181.25 334.76 (147)
b
o 0.0001 o 0.0001
In euro. No statistical test because of the low n.
b
time input rather than activities forgone with an hourly wage rate. If we adopt that approach, e.g. if wage rates were combined with the time investment as indicated in Table 3, costs per week would amount to 336.20 euro for stroke and 178.84 euro for RA.5 Money value using the proxy good method The proxy good method requires a close market substitute as the value of informal care. In The Netherlands professional home care seems to be a good proxy. 5
Leaving out HDL activities (if one considers these to result in an overestimation of time investment) would change the results to 72.38 euro for stroke and 27.66 euro for RA.
The tariff of a professional for HDL tasks is approximately 8.53 euro/h, for ADL tasks and for IADL tasks 32.67 euro/h. Using these figures, the cost estimates can be derived, as shown in Table 5. The hourly wage rate differs between stroke and RA because the combination of different types of tasks is different for the two groups. Again, two estimates are shown, one with and one without HDL time investment resulting in statistically significant differences in favor of estimates with housework. Comparing the two methods—valuation When we compare both methods, it is clear that important differences arise, both between as well as
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
within methods when using different ways of application. Some of the differences relate back to the differences in measured time investment (measurement according to Eqs. (1) or (2)). Other differences relate to distinct hourly values of professional care (housework is cheaper compared to personal care) or even to the use of hourly private (in the case of stroke) versus hourly household income (in the case of RA). It is also worth noting that we only value the informal caregivers’ opportunity costs of time with the opportunity costs method. But in comparing the opportunity costs method with the proxy good method one should also take into account other caregiver opportunity costs like financial outlays. This is because these costs are included in the tariffs of professional caregivers as used with the proxy good method. See Netten (1990) for an overview of other caregiver opportunity costs. In sum, weekly cost estimates range from 204.64 to 336.20 euro for stroke caregivers and 49.18 to 334.76 euro for RA caregivers.
Discussion This paper discusses, applies and compares two often recommended methods to value informal care in economic evaluations: the opportunity costs and proxy good methods. Valuing informal care firstly implies a valid measurement of the amount and sources of time forgone in order to be able to provide care (opportunity costs method), or measurement of the amount of time invested in informal care (proxy good method). Secondly, it implies economic valuation: determining valid shadow prices per hour of provided informal care. Results show that the two methods do not differ very much with respect to the valuation step. Differences between the methods are quite small or explained by the relatively low prices of housework in the case of the proxy good method. The measurement step, however, seems to be more problematic and crucial, as the opportunity costs and proxy good methods yield quite different results. On an average the total weekly time spent on caregiving is 20.2 h in the case of stroke and 27.4 h in the case of RA using the proxy good method, whereas the opportunity costs method yields much lower estimates (12.4 and 9.5 h, respectively). The measurement questions related to the proxy good method appear to perform better than those related to the opportunity costs method, at least from a response point of view. This does not necessarily imply that the answers are valid. For example, Van den Berg & Spauwen (Accepted) compared the results of a retrospective way of questioning similar to those proposed in the proxy good method using a diary (within subject comparison). They concluded that a retrospective way of questioning involves an overestimation of the provided care. A comparison of our results with national time allocation
843
data revealed that there might be a re-allocation of caregivers’ time spent on housework due to caregiving responsibilities, sacrificing around 2 h of house cleaning in order to be able to spend time on other activities. Future research could focus on measurement of care using diaries with a focus on informal care (between subject comparisons). In general, clarification of terms used in the survey measuring time seems to be crucial. Our analysis seems to indicate that the terms ‘unpaid work’ and ‘leisure’ in the opportunity costs method were not entirely clear for all respondents. This may be because the 60% of RA caregivers indicating that they spent time on leisure before the caregiving episode started is strikingly low (Table 2a). One would expect a percentage close to 100%, as almost everybody enjoys leisure now and then. The same holds for unpaid work because one would expect that almost everybody performs some unpaid work (in and around the house). For the measurement of time we recommend the addition of an open answer category in order to give respondents the chance to indicate other possible opportunity costs. This holds also, but to a lower extent, for the proxy good method. Giving respondents the opportunity to indicate other informal care tasks could provide additional insights. In applying the proxy good method, one could also use the market prices of house workers. On the one hand this would be a better proxy because the market for house workers is not as heavily regulated as the health care market. On the other hand the quality of home keepers could be less than the quality of the professional caregivers due to education and training. So, using the salaries of house workers could underestimate the value of informal care. There is also an institutional argument in favor of the professional caregiver. In The Netherlands if no informal caregiver was available, the care recipient would get professional care as a consequence of his/her social insurance and therefore would not need to hire a house worker. It is conspicuous if we compare our results with that of Blackwell, O’Shea, Moane, and Murray (1992) where our caregivers reported lower opportunity costs of paid work forgone despite the fact that their caregivers are older. They found that on an average 24% of informal care provided was at the cost of paid work, 37% at the cost of unpaid work and 32% at the cost of leisure. Blackwell et al. (1992) also added a category of voluntary activity forgone and found that 7% was at the cost of voluntary activity. Moreover, their average amount of time spent on providing informal care was around twice as much compared to our estimates (50.5 h a week versus 20.2 and 27.4 for stroke and RA, respectively). Our paper adds to this literature by estimating the opportunity costs of caring directly from informal caregivers involved in the caregiving episode instead of indirectly through the general population as
ARTICLE IN PRESS 844
B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
Blackwell et al. (1992) did. They justified their approach by assuming that care recipients were not able to indicate informal caregivers’ opportunity costs of care and they only interviewed care recipients. We interviewed both care recipients and their informal caregivers. Timmermans (2003) found that caregivers (a sample of the general population in The Netherlands) on an average worked 9.9 h a week less due to providing care (Timmermans, 2003). This is much higher compared to our findings. However, Timmermans (2003) also included respondents who stated that they rejected additional paid work due to their caregiving responsibilities. Informal caregivers who indicated that they had given up paid work reported an average of 9.1 h/week, while caregivers with a paid job reported 3.6 h/week less paid work due to the provision of care. Future research could collect more details about caregivers’ ability to choose freely changing their hours in the labor market. Labor market regulations often restrict people in choosing their optimal hours of labor supply, but all kinds of institutions like care leave facilities support caregivers’ choices. Restrictions on caregivers’ choices could have implications for the valuation of informal care using marginal wages. This
would provide an extra illustration of the complexity of the application of the opportunity costs method. In order to assure the incorporation of informal care in economic evaluations of health care, it would be useful to develop more precise guidelines for the use of both methods instead of just the recommendation to apply one of them. A consistent application is necessary in order to ensure comparison in results between different studies that provide economic valuations of informal care and also between different economic evaluations.
Acknowledgments The stroke part of this paper was presented at the Third World Conference of the international Health Economics Association (iHEA) 2001 in York. We would like to thank the editor and two anonymous referees for useful comments on an earlier version of the paper and Liz Chinchen for editing the English. We acknowledge The Netherlands Organization for Health Research and Development (ZON-MW) (Grant no. 945-10-044) for their funding.
Appendix A Table Tobit-regression to predict time spend on six different housework tasksa Independent variables
Eat and drink
Cleaning
Washing
Playing children
Shopping
Chores, gardening
Intercept Age Dummy sex (1 ¼ male) Pseudo R2 N
0.44 0.04 1.13 0.11 1813
0.41 0.01 0.98 0.13 1813
0.28 0.01 0.80 0.22 1813
0.69 0.04 1.25 0.04 1813
0.60 0.01 0.35 0.06 1813
1.69 0.03 1.49 0.05 1813
a
All estimates are statistically significant at the 5% significance level: po0:0001.
References Blackwell, J., O’Shea, E., Moane, G., & Murray, P. (1992). Care provision and cost measurement: Dependent elderly people at home and in geriatric hospitals. Dublin: Economic and Social Research Institute. Clipp, E. C., & Moore, M. J. (1995). Caregiver time use: An outcome measure in clinical trial research on Alzheimer’s disease. Clinical Pharmacology & Therapeutics, 58(2), 228–236. Davis, K. L., Marin, D. B., Kane, R., Patrick, D., Peskind, E. R., Raskind, M. A., & Puder, K. L. (1997). The caregiver activity survey (CAS): Development and validation of a new measure for caregivers of persons with Alzheimer’s disease. International Journal Geriatric Psychiatry, 12(10), 978–988.
Drummond, M. F., Mohide, E. A., Tew, M., Streiner, D. L., Pringle, D. M., & Gilbert, J. R. (1991). Economic evaluation of a support program for caregivers of demented elderly. International Journal Technology of Assessment in Health Care, 7(2), 209–219. Drummond, M. F., O’Brien, B. J., Stoddart, G. L., & Torrance, G. W. (1997). Methods for the economic evaluation of health care programmes. Oxford: Oxford University Press. Huijsman, R., Klazinga, N. S., Scholte op Reimer, W. J. M., Van Wijngaarden, J. D. H., Van Exel, N. J. A., Van PutteBoon, C., Presholt, F. T., Koopmanschap, M. A., & Niessen, L. W. (2001). Beroerte, beroering en borging in de keten: Resultaten van de edisse-studie van drie regionale experimenten met stroke service. Den Haag: ZON-MW.
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845 Jacobi, C. E., Triemstra, M., Rupp, I., Dinant, H. J., & Van den Bos, G. A. (2001). Health care utilization among rheumatoid arthritis patients referred to a rheumatology center: Unequal needs, unequal care? Arthritis & Rheumatism, 45, 324–330. Jacobi, C. E., Van den Berg, B., Boshuizen, H. C., Rupp, I., Dinant, H. J., & Van den Bos, G. A. (2003). Dimensionspecific burden of caregiving among partners of rheumatoid arthritis patients. Rheumatology, 42, 1–8. Juster, F. T., & Stafford, F. P. (1991). The allocation of time: Empirical findings, behavioral models, and problems of measurement. Journal of Economic Literature, XXIX, 471–522. Kooreman, P., & Wunderink, S. (1996). The economics of household behaviour. Houndmills: MacMillan Press. Luce, B. R., Wanning, W. G., Siegel, J. E., & Lipscomb, J. (1996). Estimating costs in cost-effectiveness analysis. In M. R. Gold, J. E. Siegel, L. B. Russell, & M. C. Weinstein (Eds.), Cost-effectiveness in health and medicine (pp. 176–213). New York: Oxford University Press. McDaid, D. (2001). Estimating the costs of informal care for people with alzheimers’s disease: Methodological and practical challenges. International Journal of Geriatric Psychiatry, 16, 400–405. Mohide, E. A., Torrance, G. W., Streiner, D. L., Pringle, D. M., & Gilbert, R. (1988). Measuring the wellbeing of family caregivers using the time trade-off technique. Journal of Clinical Epidemiology, 41, 475–482. Netten, A. (1990). An approach to costing informal care. Canterbury: University of Kent. Norton, E. C. (2000). Long-term care. In A. J. Culyer, & J. P. Newhouse (Eds.), Handbook of health economics (pp. 955–994). Amsterdam: Elsevier. Posnett, J., & Jan, S. (1996). Indirect cost in economic evaluation: The opportunity cost of unpaid inputs. Health Economics, 5(1), 13–23.
845
Scott Long, J. (1997). Regression models for categorical and limited dependent variables. SAGE Publications: Londen. SCP. (2000). Tijdsbestedingsonderzoek 2000—TBO’2000. Amsterdam: Steinmetz-archief. Stone, P. W., Chapman, R. H., Sandberg, E. A., Liljas, B., & Neumann, P. J. (2000). Measuring costs in cost-utility analyses. Variations in the literature. International Journal of Technology Assessment in Health Care, 16(1), 111–124. Timmermans, J.M. (Ed.). (2003). Informal care: On the help from and at informal caregivers. Den Haag: SCP [in Dutch]. Van den Berg, B., Al, M. J., Brouwer, W. B. F., Van Exel, J. A. J., & Koopmanschap, M. A. (2005). Economic valuation of informal care: The conjoint measurement method applied to informal caregiving. Social Science & Medicine, 61, 1342–1355. Van den Berg, B., Bleichrodt, H., & Eeckhoudt, L. (2005). The economic value of informal care: A study of informal caregivers’ and patients’ willingness to pay and willingness to accept for informal care. Health Economics, 14(4), 363–376. Van den Berg, B., Brouwer, W. B. F., & Koopmanschap, M. A. (2004). Economic valuation of informal care: An overview of methods and applications. The European Journal of Health Economics, 5(1), 36–45. Van den Berg, B., Brouwer, W. B. F., Van Exel, J. A. J., & Koopmanschap, M. A. (2005). Economic valuation of informal care: The contingent valuation method applied to informal caregiving. Health Economics, 14(2), 169–183. Van den Berg, B., Spauwen, P. Measurement of informal care: An empirical study into the valid measurement of time spent on informal caregiving. Health Economics, accepted for publication. Wimo, A., von Strauss, E., Nordberg, G., Sassi, F., & Johansson, L. (2002). Time spent on informal and formal care giving for persons with dementia in Sweden. Health Policy, 61(3), 255–268.
Social Science & Medicine 62 (2006) 835–845 www.elsevier.com/locate/socscimed
Economic valuation of informal care: Lessons from the application of the opportunity costs and proxy good methods Bernard van den Berga,, Werner Brouwerb,c, Job van Exelb,c, Marc Koopmanschapb,c, Geertrudis A.M. van den Bosd, Frans Ruttenb,c a
Centre for Health Economics Research and Evaluation, Faculty of Business, University of Technology, Sydney b Department of Health Policy and Management, Erasmus MC, Rotterdam, The Netherlands c Institute for Medical Technology Assessment, Erasmus MC, Rotterdam, The Netherlands d Department of Social Medicine, Academic Medical Centre, University of Amsterdam, The Netherlands Available online 30 August 2005
Abstract This paper reports the results of the application of the opportunity costs and proxy good methods to determine a monetary value of informal care. We developed a survey in which we asked informal caregivers in The Netherlands to indicate the different types of time forgone (paid work, unpaid work and leisure) in order to be able to provide care. Moreover, we asked informal caregivers how much time they spent on a list of 16 informal care tasks during the week before the interview. Data were obtained from surveys in two different populations: informal caregivers and their care recipients with stroke and with rheumatoid arthritis (RA). A total of 218 care recipients with stroke and their primary informal caregivers completed a survey as well as 147 caregivers and their care recipients with RA. The measurement of care according to both methods is more problematic compared to the valuation. This is especially the case for the opportunity costs method and for the housework part in the proxy good method. More precise guidelines are necessary for the consistent application of both methods in order to ensure comparability of results and of economic evaluations of health care. r 2005 Elsevier Ltd. All rights reserved. Keywords: Informal care; Economic evaluations; Opportunity costs method; Proxy good method; Stroke; Rheumatoid arthritis; The Netherlands
Introduction Informal care plays a substantial role in the total care provided, especially in cases of care for people with chronic diseases and the terminally ill (Norton, 2000). Corresponding author. Centre for Health Economics
Research and Evaluation, Faculty of Business, University of Technology, Sydney, PO Box 123, Broadway NSW 2007, Australia. Tel.: +61 2 9514 4753; fax: +61 2 9514 4730. E-mail address: [email protected] (B. van den Berg).
Because caregivers sacrifice (amongst other resources) time to provide care, informal care should be incorporated in an economic evaluation taking a societal perspective (Drummond, O’Brien, Stoddart, & Torrance, 1997; Luce, Wanning, Siegel, & Lipscomb, 1996). Despite the recommendation to include informal care in economic evaluations, in practice it is often neglected (Stone, Chapman, Sandberg, Liljas, & Neumann, 2000). It is quite common to consider informal care as a cost in an economic evaluation and it is therefore suggested to incorporate changes in the use of informal caregiver time as direct non-health care costs into the numerator of the
0277-9536/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.socscimed.2005.06.046
ARTICLE IN PRESS 836
B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
cost-effectiveness ratio (Luce et al., 1996, p. 177).1 This implies that informal caregivers’ time should be valued in monetary terms. It is often recommended to use either the opportunity costs method or the proxy good method (also known as the replacement cost method) to value the time investment in informal care (Drummond et al., 1997; Luce et al., 1996; Posnett & Jan, 1996). Both methods have their strengths and weaknesses (McDaid, 2001; Van den Berg, Brouwer, & Koopmanschap, 2004). However, from a theoretical point of view the opportunity costs method is preferred (Posnett & Jan, 1996). Although alternative monetary valuation methods such as contingent valuation and conjoint measurement are proposed and applied to value informal care (Van den Berg, Al, Brouwer, Van Exel, & Koopmanschap, 2005; Van den Berg, Bleichrodt, & Eeckhoudt, 2005; Van den Berg, Brouwer, Van Exel, & Koopmanschap, 2005), the opportunity costs and proxy good methods are most advocated and most often used. One important reason for recommendations to use either one of these methods may be their relatively straightforward application. In economic evaluations, where the focus is on the care recipients rather than on informal caregivers, this may be considered an advantage. Informal care in practice is often neglected in economic evaluations where it is an important input. This may be due to various factors which include: (1) informal care is simply overlooked; (2) it is not overlooked but disregarded; (3) many health technology assessment guidelines recommend conducting economic evaluations from more narrow perspectives than the societal perspective or (4) informal care is considered relevant but researchers may have difficulties with measuring or valuing it, because guidelines and handbooks are quite short about these issues and recommended valuation methods are less straightforward to apply than they first appear. In terms of measurement of informal care as an input in health care, some important problems exist. One problem is the difficulty in measuring time forgone in order to provide informal care. Especially, when proving care started many years ago, as is often the case in chronic diseases, the normal activities forgone are difficult to indicate for caregivers. Another problem concerns the distinction between ‘‘normal’’ housework and additional housework due to the health problems of the care recipient. If this distinction is not properly made it is easy to overestimate the time spent on informal care. Regarding the valuation of informal care, it may be difficult to find appropriate opportunity costs estimates for all different time uses and groups of caregivers. In 1 This is not problematic unless informal care is the focus of the intervention for instance respite care programs for informal caregivers, see, e.g. (Mohide, Torrance, Streiner, Pringle, & Gilbert, 1988 or Drummond et al., 1991).
using the proxy good method, problems may arise in finding appropriate wages of professional substitutes who might perform the care activities if no informal caregiver was available. In this paper we discuss the application of the opportunity costs method and proxy good method in two caregiver populations—informal caregivers of care recipients with stroke and caregivers of care recipients with rheumatoid arthritis (RA). Our aim is to assess the costs of informal care in these two populations using both the opportunity costs and proxy good methods. Moreover, we wish to detect the major problems in using these often recommended methods. Application of these two methods in such distinct populations is useful in this context. Stroke is an acute condition with a clear starting point, while RA is a slowly progressive chronic disease without a clear starting point. A starting point is important for the measurement of time forgone and time spent on informal care and therefore has important implications for the application of the opportunity costs method and proxy good method. We also propose solutions for the problems in measuring time forgone when a clear starting point is unavailable and for the distinction between ‘‘normal’’ housework and informal care.
Opportunity costs and proxy good methods Opportunity costs method Conceptually, the opportunity costs method values the inputs of the production process. However, in practice it often values informal care according to Eq. (1): Value informal care ¼ ti wi ,
(1)
where ti is the time spent on care tasks by caregiver i, and wi the net market wage rate of i. If i is unemployed, a proxy for wi is used, e.g., a modified opportunity costs method to find out the reservation wage of the caregiver: the wage rate for what an individual is willing to supply at least 1 h on the labour market (Kooreman & Wunderink, 1996, p. 113). Another solution is the imputation of the actual wage of similar individuals (for example those of the same gender, education and age). Eq. (1) implies that the value of leisure and unpaid work equal the (would be) wage rate. But, when caregivers e.g. derive disutility from work, the shadow price of unpaid work may differ from the marginal wage, see Posnett and Jan (1996) for a detailed discussion. Therefore, it would be better to distinguish between different sources and amounts of time forgone, e.g., according to Eq. (2): Value informal care ¼ ni wi þ hi si þ l i ti ,
(2)
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
where ni is the i’s hours of forgone paid work, hi the i’s hours of forgone unpaid work, si the shadow price of unpaid work, li the i’s hours of forgone leisure, and ti the shadow price of leisure. According to Eq. (2), the amount and sources of time forgone should be measured instead of just the time spent on caregiving. In addition, shadow prices for unpaid work and leisure need to be determined, which poses another challenge. Often, these shadow prices are based on (an arbitrarily adjusted) wage rate, which makes the distinction between Eqs. (1) and (2) rather cosmetic. For the measurement of time according to Eq. (2), one would ideally use panel data. This is often not feasible and sub-optimal solutions are required, e.g., asking respondents how their time allocation has changed since engaging in informal care or comparing their time allocation to that of a comparable sample from the general public.
837
Both methods differ conceptually: the opportunity costs method valuing inputs and the proxy good method valuing output. Therefore, from a theoretical point of view, the opportunity costs method is the benchmark. If caregivers derive direct utility from caring, the proxy good method involves an underestimation.
Applying the methods in two populations The populations were approached as part of larger studies: an evaluation of stroke units for care recipients with stroke and a study on health and health care utilization among care recipients with RA. Some of the information gathered in the two studies was therefore not fully symmetrical, but this mainly pertains to additional information. Populations
Proxy good method The proxy good method, also called the market cost method or replacement costs method is an alternative for the opportunity costs method. It values time spent on caregiving at the (labour) market price of a close substitute: conceptually it values the output of production. This requires the availability of a market substitute for the non-market commodity, which is assumed to be almost perfect, e.g. same quality. Informal care time is e.g. valued at the wage rate or market price of a professional caregiver. The value can differ per task: e.g. housework valued at the market wage of a housekeeper and personal care valued at the market wage of a nurse.2 The measurement of time spent on informal care, instead of the time forgone, is the cornerstone of the application of the proxy good method. Two important measurement methods are the diary and recall method. Because a diary, the gold standard (Juster & Stafford, 1991, p. 473), is time-consuming for respondents, a recall method is often applied. Respondents are retrospectively asked how much time they spent on different care tasks during a certain time period. To ensure comparison of results between different populations, it is important to standardize the concept of informal care, e.g. due to the development of a standard list of care tasks. If one defines only additional housework as part of caregiving, as would be reasonable, it is crucial to distinguish ‘‘normal’’ housework from additional housework due to caregiving. This is troublesome when the caregiver and care recipient live together or if care has been provided for several years already (as is often the case in RA). 2
One can debate whether this should be the gross wage (the real opportunity costs to society) or the net wage (the wage rate for which the professional is willing to sacrifice leisure).
Informal care for care recipients with stroke The stroke data was collected as a supplement to the EDISSE study (Huijsman et al., 2001). Care recipients with stroke were included at hospital admission and followed for a period of 6 months. They were asked whether or not they received informal care. If they received informal care, we asked them to indicate their primary caregiver and caregiver was asked to fill in a written survey. If the caregiver was not present, a survey was left behind which could be posted at a later date. About 597 stroke care recipients were included in the EDISSE study. Over 181 care recipients died in the period up to 6 months after stroke and 336 of the care recipients were discharged. A total of 255 caregivers completed the survey. Due to item non-response, especially on care time questions, we used a net sample of 218 stroke-caregivers. Table 1 provides the descriptive statistics of caregivers and their care recipients. It shows that the majority of caregivers are females mostly caring for their partners and 42% have a paid job. Stroke care recipients have mean EQ-5D scores of 0.47. Informal care for care recipients with RA The data for the RA part of this study were collected as a supplement of the RA+ study, a panel study on health and health care utilization among people with RA (Jacobi, Triemstra, Rupp, Dinant, & Van den Bos, 2001; Jacobi et al., 2003). In the 2001 wave, 365 of 683 care recipients indicated they received informal care. We approached the 365 receiving informal care and asked them to pass our mail survey on to their primary informal caregiver. We also asked care recipients to complete a mail survey and included a question for the 318 care recipients without informal care asking if they currently received informal care. If so, we asked them to
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
838
Table 1 Characteristics informal caregivers and their care recipients Characteristics
Stroke (n between brackets)
RA (n between brackets)
Informal caregivers Age (mean) Female (%) Partner (%)
58.5 (218) 61.6 (218) 54.2 (218)
62.4 (147) 23.7 (147) 93.2 (147)
Education (%) Primary school Lower vocational Medium vocational Higher vocational University Duration of care (mean)
11.8 (218) 44.1 (218) 27.2 (218) 12.7 (218) 4.2 (218) n/a
13.3 45.1 28.3 9.1 4.2 11.5
24.2 (218) 22.4 (218)
17.7 (147) 58.5 (147)
Occupation (%)a Housework Disability pension/ retired Paid job Other Monthly income (mean)b Care recipients Age (mean) Female (%)
41.8 (218) 11.6 (218) 1635.03 (67)
72.0 (218) 57.5 (218)
Education (%) Primary school Lower vocational Medium vocational Higher vocational University
n/a
Occupation (%) Housework Disability pension/ retired Paid job EQ-5D (mean) EQ-VAS (mean) Professional care (%) Waiting list professional care (%) Other informal care (%)
n/a
(147) (147) (147) (147) (147) (147)
36.7 (147) 4.2 (147) 1506.55 (114)
62.1 (147) 83.9 (147) 20.7 52.6 16.4 6.7 3.6
(147) (147) (147) (147) (147)
older than the stroke caregivers and more than 90% are the partners of the care recipients. Less RA caregivers have a paid job compared to those caring for people with stroke. The duration of providing care for RA care recipients is 11.5 years, while the disease duration was 13.0 years. This indicates that providing care often starts soon after the diagnosis of RA. For stroke we assume that providing care starts directly after the diagnosis.3 The EQ-5D scores of the RA care recipients are with 0.48 similar to the stroke care recipients. Care recipients’ partners were predominantly male in the case of RA compared with mainly females in case of stroke. The latter is quite common (caregivers are often females) and the former reflects the relatively high prevalence of RA in females. The valuation methods Opportunity costs method We asked informal caregivers retrospectively what types and amount of time of paid work, unpaid work and/or leisure they gave up in order to be able to provide care. The validity of these questions increases if they are asked soon after the caregiving episode started. This makes a comparison between stroke and RA without a clear starting point particularly interesting. As we anticipated the absence of a clear starting point the case of RA, we also asked some additional questions. These questions pertain to the best alternative time use, not retrospectively, but in terms of which activity the respondents would prefer to spend time on as well as how many hours a week if they could reduce their caregiving time. These questions were phrased as follows: Suppose, you do not have to spend time on caring anymore: Would you prefer to spend this time on paid, unpaid work or leisure and how many hours per week per category?
38.9 (147) 51.8 (147)
0.47 (218) n/a n/a n/a n/a
15.4 0.48 55.61 26.1 5.9
(147) (147) (147) (147) (147)
68.0 (147)
a In case of RA, the percentages add up to over 100% due to the fact that some respondents reported different occupations. b Note that in case of stroke the net monthly income is private income, while in case of RA the net monthly income is family income.
hand a mail survey over to their primary informal caregiver. Table 1 also shows the characteristics of the net sample of 147 RA caregivers. RA caregivers are slightly
Proxy good method There are no instruments that distinguish between ‘‘normal’’ housework and additional housework due to caregiving. Existing instruments, e.g. caregiver activities time survey (CATS) (Clipp & Moore, 1995), caregiver activity survey (CAS) (Davis et al., 1997) and resource utilization in Dementia (RUD) (Wimo, von Strauss, Nordberg, Sassi, & Johansson, 2002), seem to neglect this difference. We asked caregivers to report whether, and if so, how much time they spent on a list of 16 activities (see Table 3 for the complete list) in the week preceding the interview. We distinguished between (1) 3
Information about co-morbidities is lacking, so we have to assume that the provision of informal care is due to stroke or RA.
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
household activities of daily living (HDL), (2) activities of daily living (ADL), and (3) instrumental activities of daily living (IADL). We also added a question about travel time due to caregiving, because this might be an important aspect of total time use.4 In order to derive the time spent on different care tasks, we asked caregivers two different types of questions. First: ‘‘How much time did you spend on assistance with y’’? These questions pertained to support activities, e.g., assisting the care recipient in visiting the toilet or with mobility outside. Given their nature, answers to these questions would inevitably indicate time spent on caregiving (correct responses assumed for the moment). Secondly, we asked them questions like: ‘‘How much time did you spend on y’’? These questions mainly pertained to time spent on different types of housework. Because our study focussed on time investment in the context of informal care, we were sceptical about the respondents’ ability to separate ‘‘normal’’ from additional housework, particularly for caregivers living with the care recipients. But also measuring additional housework for caregivers not living with care recipients may be troublesome e.g. separating shopping time due to caregiving from shopping for their own purposes. Comparing the indicated time spent on housework of caregivers with that of the general population could indicate whether or not caregivers are expected to have indicated ‘additional time’ spent on these tasks or rather ‘normal time’. Information on time allocation of the Dutch general public was derived from the Dutch TimeAllocation Survey 2000 (TBO’2000) (SCP, 2000). The data were collected using time budget diaries. Respondents (n ¼ 1813) reported their time spent on a broad range of activities including HDL tasks every 15 min per day. With the results from TBO’ 2000, we forecast the expected time spent on several HDL activities in our two populations corrected for age and gender to get an idea of the correctness of the answers provided in our sample. Finally, one could argue that the time invested in caregiving should equal the total time forgone due to providing care. A possible difference may be due to the fact that one of the two methods is easier to complete. It is also possible that this is due to the neglect of joint production in one way of questioning or because the figures do not adequately reflect that certain household tasks (e.g. house maintenance) are sacrificed to perform more urgent household tasks (cleaning or cooking). Possible differences between the two populations may 4
Some questions had answer categories in minutes per day, while others had answer categories in hours per week, depending on the expected weekly time investment.
839
also have to do with the starting point, which is clear in stroke while it is lacking in RA or with a relatively long caregiving history. Results from the two valuation methods Opportunity costs—measuring time forgone We distinguished three types of time forgone in order to be able to provide care: paid and unpaid work and leisure. Table 2 presents types and time forgone in both populations. Firstly, it shows the percentage of respondents involved in the three different categories before they became a caregiver, just like the mean hours per week in column 3. Next, it gives the percentage of respondents that gave up paid work, unpaid work or leisure in order to provide informal care. Column five provides the total number of weekly hours forgone per activity. The results show that 16.5% of stroke caregivers with paid work reduced their time spent on paid work, resulting in a mean reduction of 3 h paid work per week. RA caregivers with paid work indicated a slightly lower amount of paid work forgone, that is, 2.2 h/week, whereas the amount of unpaid work forgone was higher (1.2 versus 0.6 h/week). In the case of stroke we collected no information about the amount of leisure before the informal care episode started, because of expected difficulties in the measurement of leisure time. Based on the stroke experience, we redesigned the survey (just a question with one table involving a complete overview of the different sources and amount of time forgone instead of complicated routing via many different questions) and included the question on the amount of leisure before providing care. A majority of RA caregivers indicated they spent time on leisure before the caregiving episode started. Assuming that all stroke caregivers enjoyed leisure before becoming a caregiver, they indicated more often than the RA group that they had given up leisure in order to provide informal care (37.6% versus 25.2%). Moreover, the average number of hours of leisure forgone was higher in stroke (8.8 versus 6.1). In total, stroke caregivers indicated higher opportunity costs than RA caregivers: 12.4 versus 9.5 h/ week. It is worth noting that Table 2 indicates increasing item non-response when asking about numbers of hours forgone, i.e. n ¼ 30 in case of RA. As mentioned above it could be difficult for caregivers to indicate the amount and sources of time forgone if a clear starting point is lacking as in RA, because the caregiving episode may have started many years ago, e.g. before retirement. Moreover, the number of care tasks provided as well as the time spent on caregiving may slowly increase, without clear starting points. Such increases may go unnoticed. This makes it expectedly difficult for the RA respondents to indicate time spent on other activities forgone. We tested for this hypothesis
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
840
Table 2a Informal caregivers’ opportunity costs of time Percentage respondents performing activity before caregiving episode
Mean hours a week before (n between brackets)
Percentage respondents having forgone activity due to informal care
Mean hours a week forgone (n between brackets)
Stroke Paid work Unpaid work Leisure Total (218)
44.5 13.3 n/a 47.2
27.7 (103) 1.7 (103) n/a 29.4 (103)
16.5 9.7 37.6a 67.7
3.0 0.6 8.8 12.4
(70) (70) (70) (70)
RA Paid work Unpaid work Leisure Total (147)
36.7 25.2 60.5 72.8
15.9 (107) 2.5 (107) 7.2 (107) 25.6 (107)
8.4 9.3 25.2 28.0
2.2 1.2 6.1 9.5
(30) (30) (30) (30)
Sub samples and activities (n between brackets)
a
Percentage based on total group (n ¼ 218).
Table 2b Alternative question posed to RA informal caregivers Activities (n between brackets)
Percentage respondents indicating they would spend freed time on activity
Mean hours a week (n between brackets)
Paid work Unpaid work Leisure Total (147)
6.8 9.5 61.9 68.0
0.6 1.0 5.7 7.3
(100) (100) (100) (100)
by means of a simple correlation coefficient between respondents indicating both their opportunity costs of time and the number of years they had provided care. The relatively strong correlation coefficient of 0.50 supports this hypothesis. Because we expected difficulties in indicating their opportunity costs of time, we also asked RA caregivers which activities they would spend their time on if they no longer had to fulfil care tasks. Table 2b shows the results. Most caregivers preferred to spend freed time on leisure. The percentages of caregivers indicating that they preferred to spend freed time on paid work or unpaid work resemble those in Table 2a. For leisure however, there is a substantial difference compared to the results reported in Table 2a. In terms of hours per week, both methods yield similar results for unpaid work and leisure. However, the amount of paid work is in the alternative question (Table 2b) lower compared with Table 2a. This might be related to the fact that some caregivers have retired within the long time
interval of 11.5 years between the start of providing informal care and date of survey completion. Therefore, although the alternative method may be useful in a context of long term care and slowly progressive diseases, the validity of provided answers remains to be established. Proxy good method—measuring time investment In applying the proxy good method, time investments on different care tasks needs to be assessed. We distinguished HDL, ADL and IADL tasks and travel time. Table 3 presents the results both in percentages of caregivers performing specific activities as well as the number of hours per week they spent on these activities. Most caregivers performed HDL tasks, and the time spent on these tasks was relatively large. The majority of caregivers were also involved in IADL tasks, yet the time involved in these tasks was substantially lower compared to HDL. About one third of the caregivers were involved in ADL tasks, which required around 2 h/ week. More stroke than RA caregivers had to travel, probably because more RA caregivers were partners of the care recipients. The overall percentage of caregivers performing tasks was similar in both groups, but the RA caregivers provided about 7 h more care per week. This difference is mainly related to the performance of HDL tasks, probably because more RA caregivers are men who indicate HDL more often as informal care compared to women. The large amount of time invested in HDL tasks may reflect the fact that ‘‘normal’’ HDL tasks are not fully separated from additional HDL tasks. Therefore, we compared the time allocation of the caregivers to that of the general public, using the results from the TBO’ 2000
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
841
Table 3 Informal care time in mean hours a week Strokea
Preparation of food and drinks Time investments in shopping, groceries etc. Housecleaning Washing, ironing or sowing Caring for and playing with own children Chores, gardening, maintenance HDL (Total)
Mean hours a week (n between brackets)
56.8 85.3 62.5 62.9 4.8 43.8
3.3 4.6 2.3 0.2 2.4 1.3
RAa
Mean hours a week (n between brackets)
76.7 87.2 86.0 43.0 11.3 74.2
5.0 6.3 2.0 0.9 4.2 2.9
89.0
14.0
94.1
21.2
18.2 8.5 18.6 19.2
0.6 0.1 0.4 0.5
34.9 6.7 17.7 10.3
1.3 0.1 0.5 0.2
ADL (Total)
32.9
1.6
37.3
2.1
Aiding the patient in travelling outside the house Aiding the patient with visiting and in excursions Aiding the patient in contacting health care suppliers Aiding patient in organising home adaptations, etc. Aiding patient in financial matters (insurance, rent)
38.5 45.5 59.7 34.4 55.1
0.6 0.9 0.6 0.2 0.6
38.8 51.4 60.5 23.2 39.7
1.1 1.2 0.9 0.1 0.4
IADL (Total) Travelling to and from care recipient Total
76.9 42.4 94.5
Aiding Aiding Aiding Aiding
a
patient patient patient patient
with personal care in visiting the toilet moving around within the house with eating and drinking
2.9 1.7 20.2 (218)
68.6 8.8 96.1
3.7 0.3 27.4 (147)
Percentage indicating that they spend time on the activity.
study described above. We predicted the time spent on different HDL tasks by Tobit-regression, because the time spent on an activity is constrained to be greater than zero (Scott Long, 1997). See Appendix A for the exact estimation results. Table 4 shows that in both samples total reported time spent on HDL is greater than the predicted HDL time. The total differences are significant. However, in the case of housecleaning, the reported time is less than the forecasted time. This suggests that caregivers re-allocate their time spent on housework due to their caregiving responsibilities: sacrificing around 2 h of housecleaning in order to be able to spend time on other activities. Comparing the two methods—measurement Looking at measurement of time, the proxy good and the opportunity costs methods yield different results. The average weekly time spent on caregiving is 20.2 h in the case of stroke and 27.4 h in the case of RA using the proxy good method (Table 3). The opportunity costs method yields substantially lower estimates: 12.4 and 9.5 h, respectively (Table 2a). The alternative opportunity costs used in the RA population leads to even lower time estimates (7.3 h). These substantial differences
demand more research in terms of validation of the measurement methods. Another important aspect of the comparison is the number of respondents that complete the questions. As can be derived from Tables 2 and 3, the measurement questions related to the proxy good method appear to perform much better than those related to the opportunity costs method. Money value using the opportunity costs method In the opportunity costs method, we use the caregiver’s hourly wage to value the provided care. According to Eq. (2), we also use hourly wages as shadow prices for unpaid work and leisure. Table 5 shows the wage rates, while those for RA are somewhat overestimated, because they represent hourly household income. The subgroup with available information about income from paid work is small (see Table 1). Combined with item non-response on the opportunity costs method time questions, this results in very small n’s (stroke: n ¼ 23; and RA: n ¼ 7) making weekly average costs for stroke of 204.64 euro and for RA of 49.18 euro. The opportunity costs method is often applied by combining
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
842
Table 4 Housework part of informal care Variable
Stroke forecast normal
Preparation of food and drinks Time investments in shopping, groceries etc. Housecleaning Washing, ironing or sowing Caring for and playing with own children Chores, gardening, maintenance Total N
Reported additional
Significance
RA forecast normal
Reported additional
Significance
1.9
3.3
o0.0001
2.5
5.0
o0.0001
0.8
2.4
0.0009
1.0
4.3
o0.0001
7.3 0.2 2.3
4.6 2.3 0.2
o0.0001 o0.0001 o0.0001
8.1 0.5 2.0
6.2 1.8 1.0
0.0071 o0.0001 o0.0001
0.5
1.0
0.0211
0.4
3.0
o0.0001
8.3 218
13.8
o0.0001
10.5 147
21.3
o0.0001
Table 5 Opportunity costs and proxy good methods compareda
Opportunity costs method
Proxy good method
a
Stroke (n between brackets)
RA (n between brackets)
Per hour Per week Sign.
Per hour Per week Sign.
17.34 (60) Diff. 0.00 Proxy good time with housework 17.34 (60) Diff. 0.90 Without housework 18.24 (218) Diff. 4.73 With housework 13.51 (218)
204.64 10.64 (23) (37) b 0.00 131.56 336.20 10.64 (59) (37) 216.40 o 0.0001 9.60 119.80 20.24 (n ¼ 218) (147) 119.44 o 0.0001 8.05 239.24 12.19 (218) (147)
49.18 (7) 129.66 178.84 (37) 25.33 153.51 (147) 181.25 334.76 (147)
b
o 0.0001 o 0.0001
In euro. No statistical test because of the low n.
b
time input rather than activities forgone with an hourly wage rate. If we adopt that approach, e.g. if wage rates were combined with the time investment as indicated in Table 3, costs per week would amount to 336.20 euro for stroke and 178.84 euro for RA.5 Money value using the proxy good method The proxy good method requires a close market substitute as the value of informal care. In The Netherlands professional home care seems to be a good proxy. 5
Leaving out HDL activities (if one considers these to result in an overestimation of time investment) would change the results to 72.38 euro for stroke and 27.66 euro for RA.
The tariff of a professional for HDL tasks is approximately 8.53 euro/h, for ADL tasks and for IADL tasks 32.67 euro/h. Using these figures, the cost estimates can be derived, as shown in Table 5. The hourly wage rate differs between stroke and RA because the combination of different types of tasks is different for the two groups. Again, two estimates are shown, one with and one without HDL time investment resulting in statistically significant differences in favor of estimates with housework. Comparing the two methods—valuation When we compare both methods, it is clear that important differences arise, both between as well as
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
within methods when using different ways of application. Some of the differences relate back to the differences in measured time investment (measurement according to Eqs. (1) or (2)). Other differences relate to distinct hourly values of professional care (housework is cheaper compared to personal care) or even to the use of hourly private (in the case of stroke) versus hourly household income (in the case of RA). It is also worth noting that we only value the informal caregivers’ opportunity costs of time with the opportunity costs method. But in comparing the opportunity costs method with the proxy good method one should also take into account other caregiver opportunity costs like financial outlays. This is because these costs are included in the tariffs of professional caregivers as used with the proxy good method. See Netten (1990) for an overview of other caregiver opportunity costs. In sum, weekly cost estimates range from 204.64 to 336.20 euro for stroke caregivers and 49.18 to 334.76 euro for RA caregivers.
Discussion This paper discusses, applies and compares two often recommended methods to value informal care in economic evaluations: the opportunity costs and proxy good methods. Valuing informal care firstly implies a valid measurement of the amount and sources of time forgone in order to be able to provide care (opportunity costs method), or measurement of the amount of time invested in informal care (proxy good method). Secondly, it implies economic valuation: determining valid shadow prices per hour of provided informal care. Results show that the two methods do not differ very much with respect to the valuation step. Differences between the methods are quite small or explained by the relatively low prices of housework in the case of the proxy good method. The measurement step, however, seems to be more problematic and crucial, as the opportunity costs and proxy good methods yield quite different results. On an average the total weekly time spent on caregiving is 20.2 h in the case of stroke and 27.4 h in the case of RA using the proxy good method, whereas the opportunity costs method yields much lower estimates (12.4 and 9.5 h, respectively). The measurement questions related to the proxy good method appear to perform better than those related to the opportunity costs method, at least from a response point of view. This does not necessarily imply that the answers are valid. For example, Van den Berg & Spauwen (Accepted) compared the results of a retrospective way of questioning similar to those proposed in the proxy good method using a diary (within subject comparison). They concluded that a retrospective way of questioning involves an overestimation of the provided care. A comparison of our results with national time allocation
843
data revealed that there might be a re-allocation of caregivers’ time spent on housework due to caregiving responsibilities, sacrificing around 2 h of house cleaning in order to be able to spend time on other activities. Future research could focus on measurement of care using diaries with a focus on informal care (between subject comparisons). In general, clarification of terms used in the survey measuring time seems to be crucial. Our analysis seems to indicate that the terms ‘unpaid work’ and ‘leisure’ in the opportunity costs method were not entirely clear for all respondents. This may be because the 60% of RA caregivers indicating that they spent time on leisure before the caregiving episode started is strikingly low (Table 2a). One would expect a percentage close to 100%, as almost everybody enjoys leisure now and then. The same holds for unpaid work because one would expect that almost everybody performs some unpaid work (in and around the house). For the measurement of time we recommend the addition of an open answer category in order to give respondents the chance to indicate other possible opportunity costs. This holds also, but to a lower extent, for the proxy good method. Giving respondents the opportunity to indicate other informal care tasks could provide additional insights. In applying the proxy good method, one could also use the market prices of house workers. On the one hand this would be a better proxy because the market for house workers is not as heavily regulated as the health care market. On the other hand the quality of home keepers could be less than the quality of the professional caregivers due to education and training. So, using the salaries of house workers could underestimate the value of informal care. There is also an institutional argument in favor of the professional caregiver. In The Netherlands if no informal caregiver was available, the care recipient would get professional care as a consequence of his/her social insurance and therefore would not need to hire a house worker. It is conspicuous if we compare our results with that of Blackwell, O’Shea, Moane, and Murray (1992) where our caregivers reported lower opportunity costs of paid work forgone despite the fact that their caregivers are older. They found that on an average 24% of informal care provided was at the cost of paid work, 37% at the cost of unpaid work and 32% at the cost of leisure. Blackwell et al. (1992) also added a category of voluntary activity forgone and found that 7% was at the cost of voluntary activity. Moreover, their average amount of time spent on providing informal care was around twice as much compared to our estimates (50.5 h a week versus 20.2 and 27.4 for stroke and RA, respectively). Our paper adds to this literature by estimating the opportunity costs of caring directly from informal caregivers involved in the caregiving episode instead of indirectly through the general population as
ARTICLE IN PRESS 844
B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845
Blackwell et al. (1992) did. They justified their approach by assuming that care recipients were not able to indicate informal caregivers’ opportunity costs of care and they only interviewed care recipients. We interviewed both care recipients and their informal caregivers. Timmermans (2003) found that caregivers (a sample of the general population in The Netherlands) on an average worked 9.9 h a week less due to providing care (Timmermans, 2003). This is much higher compared to our findings. However, Timmermans (2003) also included respondents who stated that they rejected additional paid work due to their caregiving responsibilities. Informal caregivers who indicated that they had given up paid work reported an average of 9.1 h/week, while caregivers with a paid job reported 3.6 h/week less paid work due to the provision of care. Future research could collect more details about caregivers’ ability to choose freely changing their hours in the labor market. Labor market regulations often restrict people in choosing their optimal hours of labor supply, but all kinds of institutions like care leave facilities support caregivers’ choices. Restrictions on caregivers’ choices could have implications for the valuation of informal care using marginal wages. This
would provide an extra illustration of the complexity of the application of the opportunity costs method. In order to assure the incorporation of informal care in economic evaluations of health care, it would be useful to develop more precise guidelines for the use of both methods instead of just the recommendation to apply one of them. A consistent application is necessary in order to ensure comparison in results between different studies that provide economic valuations of informal care and also between different economic evaluations.
Acknowledgments The stroke part of this paper was presented at the Third World Conference of the international Health Economics Association (iHEA) 2001 in York. We would like to thank the editor and two anonymous referees for useful comments on an earlier version of the paper and Liz Chinchen for editing the English. We acknowledge The Netherlands Organization for Health Research and Development (ZON-MW) (Grant no. 945-10-044) for their funding.
Appendix A Table Tobit-regression to predict time spend on six different housework tasksa Independent variables
Eat and drink
Cleaning
Washing
Playing children
Shopping
Chores, gardening
Intercept Age Dummy sex (1 ¼ male) Pseudo R2 N
0.44 0.04 1.13 0.11 1813
0.41 0.01 0.98 0.13 1813
0.28 0.01 0.80 0.22 1813
0.69 0.04 1.25 0.04 1813
0.60 0.01 0.35 0.06 1813
1.69 0.03 1.49 0.05 1813
a
All estimates are statistically significant at the 5% significance level: po0:0001.
References Blackwell, J., O’Shea, E., Moane, G., & Murray, P. (1992). Care provision and cost measurement: Dependent elderly people at home and in geriatric hospitals. Dublin: Economic and Social Research Institute. Clipp, E. C., & Moore, M. J. (1995). Caregiver time use: An outcome measure in clinical trial research on Alzheimer’s disease. Clinical Pharmacology & Therapeutics, 58(2), 228–236. Davis, K. L., Marin, D. B., Kane, R., Patrick, D., Peskind, E. R., Raskind, M. A., & Puder, K. L. (1997). The caregiver activity survey (CAS): Development and validation of a new measure for caregivers of persons with Alzheimer’s disease. International Journal Geriatric Psychiatry, 12(10), 978–988.
Drummond, M. F., Mohide, E. A., Tew, M., Streiner, D. L., Pringle, D. M., & Gilbert, J. R. (1991). Economic evaluation of a support program for caregivers of demented elderly. International Journal Technology of Assessment in Health Care, 7(2), 209–219. Drummond, M. F., O’Brien, B. J., Stoddart, G. L., & Torrance, G. W. (1997). Methods for the economic evaluation of health care programmes. Oxford: Oxford University Press. Huijsman, R., Klazinga, N. S., Scholte op Reimer, W. J. M., Van Wijngaarden, J. D. H., Van Exel, N. J. A., Van PutteBoon, C., Presholt, F. T., Koopmanschap, M. A., & Niessen, L. W. (2001). Beroerte, beroering en borging in de keten: Resultaten van de edisse-studie van drie regionale experimenten met stroke service. Den Haag: ZON-MW.
ARTICLE IN PRESS B. van den Berg et al. / Social Science & Medicine 62 (2006) 835–845 Jacobi, C. E., Triemstra, M., Rupp, I., Dinant, H. J., & Van den Bos, G. A. (2001). Health care utilization among rheumatoid arthritis patients referred to a rheumatology center: Unequal needs, unequal care? Arthritis & Rheumatism, 45, 324–330. Jacobi, C. E., Van den Berg, B., Boshuizen, H. C., Rupp, I., Dinant, H. J., & Van den Bos, G. A. (2003). Dimensionspecific burden of caregiving among partners of rheumatoid arthritis patients. Rheumatology, 42, 1–8. Juster, F. T., & Stafford, F. P. (1991). The allocation of time: Empirical findings, behavioral models, and problems of measurement. Journal of Economic Literature, XXIX, 471–522. Kooreman, P., & Wunderink, S. (1996). The economics of household behaviour. Houndmills: MacMillan Press. Luce, B. R., Wanning, W. G., Siegel, J. E., & Lipscomb, J. (1996). Estimating costs in cost-effectiveness analysis. In M. R. Gold, J. E. Siegel, L. B. Russell, & M. C. Weinstein (Eds.), Cost-effectiveness in health and medicine (pp. 176–213). New York: Oxford University Press. McDaid, D. (2001). Estimating the costs of informal care for people with alzheimers’s disease: Methodological and practical challenges. International Journal of Geriatric Psychiatry, 16, 400–405. Mohide, E. A., Torrance, G. W., Streiner, D. L., Pringle, D. M., & Gilbert, R. (1988). Measuring the wellbeing of family caregivers using the time trade-off technique. Journal of Clinical Epidemiology, 41, 475–482. Netten, A. (1990). An approach to costing informal care. Canterbury: University of Kent. Norton, E. C. (2000). Long-term care. In A. J. Culyer, & J. P. Newhouse (Eds.), Handbook of health economics (pp. 955–994). Amsterdam: Elsevier. Posnett, J., & Jan, S. (1996). Indirect cost in economic evaluation: The opportunity cost of unpaid inputs. Health Economics, 5(1), 13–23.
845
Scott Long, J. (1997). Regression models for categorical and limited dependent variables. SAGE Publications: Londen. SCP. (2000). Tijdsbestedingsonderzoek 2000—TBO’2000. Amsterdam: Steinmetz-archief. Stone, P. W., Chapman, R. H., Sandberg, E. A., Liljas, B., & Neumann, P. J. (2000). Measuring costs in cost-utility analyses. Variations in the literature. International Journal of Technology Assessment in Health Care, 16(1), 111–124. Timmermans, J.M. (Ed.). (2003). Informal care: On the help from and at informal caregivers. Den Haag: SCP [in Dutch]. Van den Berg, B., Al, M. J., Brouwer, W. B. F., Van Exel, J. A. J., & Koopmanschap, M. A. (2005). Economic valuation of informal care: The conjoint measurement method applied to informal caregiving. Social Science & Medicine, 61, 1342–1355. Van den Berg, B., Bleichrodt, H., & Eeckhoudt, L. (2005). The economic value of informal care: A study of informal caregivers’ and patients’ willingness to pay and willingness to accept for informal care. Health Economics, 14(4), 363–376. Van den Berg, B., Brouwer, W. B. F., & Koopmanschap, M. A. (2004). Economic valuation of informal care: An overview of methods and applications. The European Journal of Health Economics, 5(1), 36–45. Van den Berg, B., Brouwer, W. B. F., Van Exel, J. A. J., & Koopmanschap, M. A. (2005). Economic valuation of informal care: The contingent valuation method applied to informal caregiving. Health Economics, 14(2), 169–183. Van den Berg, B., Spauwen, P. Measurement of informal care: An empirical study into the valid measurement of time spent on informal caregiving. Health Economics, accepted for publication. Wimo, A., von Strauss, E., Nordberg, G., Sassi, F., & Johansson, L. (2002). Time spent on informal and formal care giving for persons with dementia in Sweden. Health Policy, 61(3), 255–268.