Approaches to Estimating Indirect Costs in Healthcare: Motivations for Choice

When performing health economic evaluations all costs and expected benefits (in terms of clinical effectiveness, utility, monetary benefits) should be taken into account. Costs are direct and indirect: concerning the latter, two main methods have been developed. The Human Capital Approach (HCA) considers the gross salary in the days of absence from work due to a disease; the method based on the Friction Costs (FC), instead, considers equally the value of productivity, but the losses are limited to the period of illness when it is necessary to replace the absent worker. This paper carries out a review of the literature contributions for European countries and aims at identifying common trends within geographical areas; estimation of indirect costs in different European areas may reflect the real cost of health services across countries and schematize the conditions under which a certain method should be preferred. Overall, cost analyses applying HCA are the most frequent, while studies based on FC are more common in the Netherlands, where this method has been developed: the reasons to apply HCA or FC may depend on the institutional context, for example the flexibility in the job market, and the epidemiologic environment.


Introduction
Among the objectives of the economic evaluation in healthcare there is the estimation of the costs associated with the health programs and technologies that represent an essential step in the choice and application of methods with different characteristics.
Let us consider these two aspects.

Relevance of cost estimation in health economics analyses
The information relevant to the decision-maker regard, on the one hand, the estimate of the resources needed to implement a given program and, on the other hand, the analysis of the economic consequences produced by that program, both from the point of view of the National Health Service (NHS), where this is present, and of the welfare system.
The resources needed to implement a health program can be divided into three categories: first, there are the direct costs, i.e. the resources associated with medical treatments (such as drugs, medical examinations, support therapies, laboratory tests, possible hospitalizations, etc.); non-health direct costs relate to the resources used by patients and family members (these ones include the cost of social services, domestic assistance, transport costs, assistance provided by caregivers outside the working hours): these are mainly the so-called out-of-pocket expenses that are sustained directly by the patient; finally, the indirect costs are related to the impact that the disease has on the social sphere of the patient (for example, the working days lost for medical treatment the working days lost by patients' family members, as well as the share of social security expenditure borne by the public sector).
A cost assessment including indirect costs can efficiently drive health policy planning (Puddu et al., 2016;Pederzoli and Gandini, 2008) and allow a better allocation of resources (Koopmanschap and Rutten, 1994).
Three fundamental components have been identified as the elements that contribute to determine the indirect costs (van Roijen et al., 1996): 1) absences from work (which have to be paid); 2) reduced productivity at work; 3) production not remunerated, due, for example, to the lower possibility of carrying out the usual activities at home.
All these components depend on the state of illness.
The choice of a suitable methodology to calculate all kind of costs, included indirect costs, is an issue open to different solutions: in fact, there are no established guidelines and precise criteria universally recognised.
Productivity costs are often omitted from economic assessments, despite their impact, often considerable, in terms of cost-effectiveness, Brouwer, 2014, Krol et al., 2011). When a worker cannot work due to illness, production decreases, giving rise to immediately quantifiable losses from an economic point of view (Knies et al., 2010).
The contributions that, in the economic literature, have carried out estimates of these losses usually employ two main techniques: the best known is the Human Capital Approach (HCA); more recently, it has been implemented the method based on the calculation of the Friction Costs (FC).
Among the two methodologies, which one should be preferred?
The present analysis compares HCA and FC: it has been carried out for European countries, distinguishing geographical areas; the survey has considered the studies carried out in the last decade.
The present analysis has been conducted to highlight how studies in the medical-economic literature have been oriented to carry out an evaluation of indirect costs. Applying the same research methodology in each country and in each reality examined may not be appropriate: the institutional context is different, the epidemiological scenario that occurs in different European regions is different.
Even if the prevailing analysis methodology in one geographical area was, for example, that based on CF, this choice may not be the one to be preferred in another area, characterized by a different legislative and institutional context.
Overall, the important indication that can be drawn from such a review, therefore, concerns: -the appraisal of the prevailing methodology according to different pathologies; -the identification of the methodology to be preferred in various institutional settings.

Characteristics of evaluation methods
In general, both methods estimate the value of resources in monetary terms (represented by the salary value): while the HCA considers the gross salary in the days of absence from work due to the disease, the method based on FC considers the value of productivity losses, limited, however, to the period of illness when it is necessary to replace the worker absent due to the disease Sculpher, 2001). The loss of productivity is thus calculated for the period necessary to restore production levels within an organization van Ineveld, 1992). Since it is assumed that it is possible to replace workers by drawing on labor reserves (Koopmanschap et al., 1995), it has been argued that FC method generates more realistic estimates of lost productivity than HCA.
The timeframe considered differs for both approaches: while productivity losses may extend from the short term to some decades in the case of HCA, when the FCs are considered, they refer to a shorter period, although the time needed to train a new worker is not known in advance. The worker who need to be replaced may, however, increase his/her productivity when back at work, to compensate for the initial loss due to the state of illness. Further, if the sick worker carries out the work he/she has not been able to do when he returns to work (or his work is temporarily carried out by any colleague) it is possible that his/her absence does not result in any loss of production.
In the absence of an approach that is strictly preferred to the other, a solution is that one of choosing the perspective that suits the current political context, the regulatory framework, or using both methods and comparing the results achieved.
A review of the criteria that should be observed in calculating indirect costs has been carried out by Gianino et al. (2009), who, discussing about characteristics of FC, outline some key elements that must be considered. Among these ones, there are: the frequency of friction periods, determined by the recurrence and duration of the absence from work; the length of the friction period; the relationship between the loss of productivity and the time spent working.
Summarising, advantages and pitfalls of both methods can be observed in Table 1: HCA is the most dating method and has been developed in the context of labor economics studies.
Over the last twenty years, it has often been applied to estimate indirect costs and productivity losses.
The FC method, on the other hand, was initially developed by health economics scholars in the Netherlands and in the Anglo-Saxon countries; its main characteristics have been described in the works by Koopmanschap et al., 1995;Koopmanschap and van Ineveld, 1992;Liljas, 1998;Sculpher, 2001. Despite the advantages of the FC method that is applied mainly in countries characterized by a highly flexible labor market, few studies have employed this method to carry out economic evaluations in other countries (for example, Kigozi et al., 2017). This may be due to the absence of reliable data regarding the frictional period; there are proposals drawn from contributions in the literature (Koopmanschap et al., 1995;Pritchard and Sculpher, 2000). The duration of the frictional period suggested by Pearce et al. (2015) varies from 10 weeks to 13 weeks for manual occupations.
Likewise, the probability of replacement by a person who was previously unemployed may differ depending on the type of employment 1 .
In the next section it will be carried out an analysis of the economic literature on this topic looking at European geographical areas.

The analysis of economic literature by European geographical areas
The diversity of the types of intervention and the methodologies followed can be seen by examining some of the studies that have discussed this topic, the objective of the analysis, the pathology analyzed, and the results obtained.
The literature contributions have been selected according to the following procedure: first, research of the studies was carried out through Pubmed.org, including, the following keywords: -[human capital approach + indirect costs + country], substituting to "country", each European country examined. From an initial search, where "Europe" has been added as third keyword, 324 studies were retrieved.
-[friction cost + indirect costs + country]. From this other search, instead, considering as third keyword "Europe", 28 studies were retrieved.
The studies that have been identified are largely related to case studies: several observational studies have been conducted as part of international research projects, some contributions concern the two methodologies of analysis and present both the results obtained through the application of HCA and the FC-based approach.
The European geographical areas, for which the results obtained from the pharmacoeconomic literature are described, are those ones identified in the study by Gianino et al. (2009) (Andrews et al., 2008;Adams et al., 2002;Roper, 1988;Beaumont, 1978) have addressed the problem related to the quantification of the period of absence from work: however, these studies do not provide the necessary information to correlate the frictional period to the productivity costs. The differences between types of employees in the labor market are ignored and this circumstance may lead to an imprecise assessment of productivity costs Koopmanschap et al., 1995).
The common elements that may be found in the various studies are: -the use of prevalence data; -the circumstance that there are more frequently top-down rather than bottom-up analyses; -effectiveness data are collected through interviews and patient surveys; -the costs that are considered are health costs, related to the costs of treatment, hospitalization, drugs and medical devices; non-health costs are usually indirect costs and concern absenteeism, presenteeism and loss of productivity because of the morbid condition; -health policy considerations relate to the costs of care to provide for patients; -it is always difficult to quantify the intangible costs that are also borne by patients and their caregivers. Table 2 summarizes the studies reviewed and the indications about the possible inclusion of indirect costs in health economics studies.

The literature contributions for Eastern Europe.
The contributions found for Eastern Europe and carried out in the last five years are not many, but they allow us to draw indications regarding the inclusion of costs in the economic evaluation of indirect costs.
The prevailing criterion for the estimation of costs is HCA, rather than FC. The pathologies that have called for investigation about costs, in this geographic area, concerned mainly metabolic diseases (Crohn's disease or ulcerative colitis): while information on productivity regarding the performance at work was collected, mainly, through questionnaires, the disability rates for the whole population derive from national databases (see, for example, Mandel et al., 2014;Holko et al., 2016, Kawalec, 2017. Some studies (as Łyszczarz and Nojszewska, 2017, in a study carried out for Poland and related to oncology diseases) underline the importance of taking into account, together with the health and social costs, the economic implications of cancer both for the patient involved and for the system.
HCA is employed to evaluate all pathologies that implies a personal involvement by the patient, for example, migraine (Lublóy, 2019), or neurological conditions as epilepsy (Jędrzejczak et al., 2021).
Overall, indirect costs, due to reduced productivity at work, are strongly influenced by the disease 2 .
Some studies do not examine specific pathologies: the study by Lasocka et al. (2013), aims to estimating annual productivity losses due to smoking by looking at a social perspective and comparing the results obtained for Poland with similar research carried out in other countries (Germany, Sweden and USA) through the method of HCA.
The estimation of the full costs of illness depends on the method chosen for the analysis. In a study concerning asthma, indirect costs, determined looking at the perspective of the third payer amounted to € 4,287.6 per patient applying HCA, while they decrease to € 1,457.2 using the FC method (Jahnz- Różyk K et al., 2015).

Conclusions reached for Eastern
Europe are comparable to all OECD countries. An aspect that should be considered concerns the financing of health expenditure; for example, in Poland 70% of health expenditure is financed through universal contributions and increases in the same way as citizens' wages increase.

The literature contributions for Western Europe.
In the countries of Western Europe (Austria, Belgium, France, Germany, the Netherlands, Switzerland), most studies aimed at estimating indirect costs concerns the HCA method, while none applies FC, although such methodology has been developed in the Netherlands: this tendency may derive from a more conservative attitude by researchers, who prefer a consolidated method like HCA, rather than a more recent method like FC.
The majority of costs are attributable to the indirect costs resulting from lost productivity. It has been seen that estimates of costs obtained with FC method are 56% lower than those estimated with HCA (see Prast et al., 2013, who examines the costs of endometriosis, whose results can be compared with Simoens at al (2007), who applies the FC method obtaining lower indirect costs).
For Belgium, the studies identified are not numerous, and the attention of the literature is directed, rather than to specific pathologies, to assess implications for costs determined by lifestyles (for example, smoking, use of drugs, etc.). For example, Lievens et al. (2017) examine the economic burden for both individuals and society due to the use of both legal and illegal substances (including alcohol, tobacco, and illicit drugs): the use of illegal substances determines costs of law enforcement, loss of productivity and reduced quality of life. The studies concerning social aspects should adopt a wider perspective, and this justifies the application of HCA, that is not temporarily limited to the period of absence from work but considers a longer time horizon than FC.
It has been seen how, in Western Europe countries, the analyses dealing with economic evaluations focusing on indirect costs are rather limited. This could be explained by considering also the ISPOR indications which, precisely for Belgium, state that, when carrying out economic evaluation analyses, "the reference case should only include the costs of direct health care. These include the costs directly related to the treatment of the disease as well as the direct costs of health care related to the disease in the years of life gained. Direct costs outside the health sector, productivity costs and healthcare costs associated with unrelated diseases should not be included in the reference case, but can be reported as a separate analysis" (https:// www. ispor.org/PEguidelines/countrydet.asp?c=3&t=1).
When there is not a clear indication, as in France, because the pathologies for which economic evaluations considering the indirect costs were carried out are different (see, for example, Fautrel et al., 2007;Serrier et al., 2014;Chevreul et al., 2015), according to the criteria established by the Haute Autoritè de la Santè (2012) the evaluation method to assess the indirect costs, has to be chosen by the researcher and adequate motivations must be provided.
In Germany, most of the literature recommends including productivity losses in calculating healthcare costs, and, therefore, indirect costs. In the German context, the inability to perform any working activity is associated with partial loss of income: consequently, since productivity losses due to the inability of work should be presented among costs (IQWIG, 2009), HCA for the estimation of indirect costs would be the preferred method (Bommer et al., 2017, Huebner et al., 2017. Other examples not related to specific diseases but affecting the costs of care are described in studies using HCA. For example, although patients with injuries from an accident represent a significant medical and socio-economic burden in the German healthcare system, there are few data describing costs in the period between the accident and the professional reintegration (Anders et al., 2013), so that the analysis can only be considered as a starting point for further studies. This might be a framework where the application of FC could be, indeed, evalutated.
Often, it is the kind of disease that suggests an assessment of costs across time (as it happens in the study by Wolf at al., 2010, dedicated to a therapy program concerning HIV-positive patients, that concludes that the administration of abacavir in HIV-positive patients reduces the risk of hypersensitivity reaction to the drug, and determines potential cost savings, measured with HCA).
The study is not limited to examine the period needed to restore work skills, as in Dodel et al. (2010), for example, who assess the health burden of patients with Gilles de la Tourette syndrome (GTS) in Germany during a 3-month observation period, a time span necessary to identify the rational allocation of resources.
While HCA is a widely accepted method for measuring the loss of production for the society caused by the disease, as it evaluates the output generated by a person, measured in market prices, this method is often criticized as it overestimates the "real" costs of the disease: in fact, it assumes a perfect labor market and an immediate substitutability of the workers.
Other conditions, as stroke, for example, absorb many resources. While studies on the costs of acute stroke treatment have been conducted in Europe, a thorough analysis of direct and indirect long-term costs is lacking.
Cancer also causes a high economic burden. Reis et al. (2006) compare the direct, indirect and social costs of the disease for Hodgkin's disease, non-Hodgkin's lymphoma, plasmacytoma and chronic lymphatic lymphoma. Direct, indirect, and social costs were calculated using the HCA method.
Lung cancer shows the main incidence of all cancers among men and an increasing incidence among women. The study by Weissflog et al. (2001) evaluates the economic burden of lung cancer in Germany and identifies the main cost factors. In a retrospective analysis, direct and indirect costs based on data from government institutions and the pharmaceutical industry were calculated: the HCA method was applied for the calculation of indirect costs.
Some studies examine mental illnesses. Konnopka et al. (2009) perform a systematic review of the literature on the cost of the disease for schizophrenia in Germany. Indirect costs are mainly due to early retirement or unemployment and the HCA method is used to calculate them. Krauth et al. (2000) evaluate the indirect costs of an outpatient rehabilitation program for the Hanover medical school. In the economic assessment, the cost of time is expressed by the loss and reduction of working time, time for housework and leisure time. To estimate the actual loss of working time which results in a loss of production, it is possible to apply the HCA and FC methods. The time lost due to housework can be estimated both considering the production of goods and services, and the opportunity cost of an equivalent working hour.
In Switzerland, the inclusion of indirect costs is subject to the relevance of the study; the studies carried out present particular cases and often evaluate general programs rather than the impact of specific pathologies.
The study by Pugliatti et al. (2007) estimates the cost of epilepsy in Europe (25 EU countries plus Iceland, Norway and Switzerland). The selected approach is based on a bottom-up cost estimate that employs HCA for the calculation of indirect costs.
Another study considers the weight and burden of flu, which varies with age and the patient's state of health (Szucs, 1999). The disease imposes a significant burden on all individuals, but hospitalization and treatment occur more frequently in high-risk patients (the elderly and those with certain co-morbidities). For the measurement of indirect costs, it is preferred the HCA method. Other intangible costs associated with flu include reduced productivity and adverse effects on the quality of life of patients and their families. Although vaccines are widely effective, the greatest potential benefits are for high-risk patients; for them, as well as for caregivers and health professionals, vaccination is highly recommended.
Finally, Sagmeister et al. (1998) estimate the cost of premature mortality due to coronary artery disease for the working age population, using HCA. Indirect costs are around 25% of the total costs generated by the disease. From the study it is possible to observe a reduction in costs from one year to the other, corresponding to a lower loss of productivity.
Instead, the studies that apply the method based on FCs for the estimation of indirect costs are less numerous.
The diseases examined are mainly rheumatic diseases, as well as oftalmic diseases and vaccination programs. Huscher et al. (2015) estimate changes in direct and indirect costs caused by patients with rheumatoid arthritis in Germany between 2002 and 2011. Costs were updated and indirect costs were calculated using HCA and FC. There has been a significant increase in direct costs, attributable to the increase in the prescription of biological agents. However, such increase in expenses for medical treatment is accompanied by a reduction in hospital expenses and indirect costs.
In a previous study, Huscher et al. (2006) had already estimated and compared the direct and indirect costs of rheumatoid arthritis, ankylosing spondylitis, psoriatic arthritis and systemic lupus erythematosus, evaluating the impact of certain variables (gender and severity of the patients' conditions) on the various cost items, applying, for the calculation of indirect costs, HCA and FCs.
With the FC method the estimated figures were lower, with a strong effect of the functional state on the total costs.
Overall, it can be argued that the indirect costs of chronic diseases are receiving ever increasing attention because recent diagnostic and therapeutic improvements have significantly augmented treatment costs, and it is therefore crucial to pursue cost savings. However, the methods of valuing indirect cost components still differ widely, leading to heterogeneous data that does not allow for direct comparison. This is the consideration advanced, for example, by Merkesdal et al. (2002) that, in a prospective study, compare the indirect costs calculated with the HCA and FC method, providing recommendations to improve the comparability of such costs.
The Netherlands represents the context in which the FC methodology has been developed and in which the studies applying this method are more numerous. The interest of scholars is in assessing the impact of indirect costs on the labor market: this is why productivity losses receive higher attention.
In Politiek et al. (2016), a systematic review of the literature is carried out considering both HCA and FC. The study assesses the costs of eczema in the hand, the burden of which is considerable, estimating indirect costs that account up to 70% of total costs and are mainly due to work absenteeism. women's domestic injuries). However, regardless of the method used, more than 80% of indirect costs are the result of injuries among males, mainly caused by a high frequency of accidents at work and sports injuries among young males. The application of the FC method confirms injuries as a source of production losses compared to other diseases.
All the other studies identified focus on a broad discussion on the advantages of each estimation method. The debate dates to the early 1990s, with the study by Koopmamschap and van Ineveld (1992), who warned that many researchers, when assessing health care, have doubts on the usefulness of estimates of indirect costs of disease in setting priorities in health care.
The study seeks to respond to the criticism related to the definition of the concept of indirect costs, that are meant as the value of loss of production due to the disease, by first proposing a new method for estimating indirect costs: the authors conclude that the method based on FCs for estimating indirect costs is promising but needs further development 3 . The consequences of the disease in unemployed people must be incorporated into the analysis; the relationship between internal labor reserves and disease costs should also be further investigated.
While, in the early 1990s, the HCA was commonly used to estimate indirect costs, a growing number of authors questioned its validity. Koopmansch and Rutten (1993) illustrate the relevance of indirect cost estimates for health policy and stress various important issues and disputes concerning indirect costs (such as the reduction of productivity without absence from work, the influence of unemployment on loss of production, the relationship between health effects and indirect costs).
Indirect costs are relevant to health policy, provided that their estimates reflect the actual changes in production due to the disease, including production by unpaid labor.
In economic evaluations, indirect costs, according to the studies carried out in the Netherlands, should preferably be presented separately from direct costs, health effects and other results. Koopmansch et al. (1995) describe the FC-based approach for estimating the indirect costs and provide some evidence: for the Netherlands, short-term FC in 1990 amounts to 1.5-2.5% of national net income, depending on the extent to which short-term absence from work leads to losses and production costs. The macroeconomic consequences of absence from work and disability in the medium term reduce the national net income by a further 0.8%. These estimates are significantly lower than estimates based on the traditional HCA approach, but, perhaps, reflect better the economic impact of the disease and should orient the choice of the analyst for FC methods.
In the subsequent study by Koopmanschap and Rutten (1996), a practical guide is provided for the quantification and exploitation of indirect disease costs, both at the aggregate level of general cost of disease studies, and in an economic assessment of precise health programs 4 .
Among the studies that have been identified and which apply FCs, there are Reurings et al. (2010), which evaluates the latest developments in colon surgery. The primary outcome parameters are the two main cost factors: direct medical costs (calculated through the actual cost) and indirect nonmedical costs (estimated through the FC method). Postma et al. (1999) had already proposed a single case study, evaluating the cost effectiveness of the Chlamydia trachomatis (CT) screening of young women who go to primary care physicians. Data on the need for health care for CT complications and its costs are estimated through FCs. Hutubessy et al. (1999) estimate the indirect costs of back pain in 1991 in the Netherlands based on both HCA and FC. The results of the study showed that the estimated short-term indirect costs with HCA were more than three times higher than the indirect costs estimated by the FC-based method (respectively, $ 4.6 billion vs. $ 1.5 billions). The lowest estimate of indirect costs in the use of the FC method is mainly due to the fact that actual production losses were assessed during a relatively short friction period. In contrast to HCA, long-term absenteeism and disability do not induce additional costs when the FC method is applied. Policies aimed at reducing the indirect costs of back pain are increasingly focused on the development and evaluation of interventions immediately after the onset of the disease, taking also into account the possibility of reintegrating chronic patients into the workforce.
Recently, in Dutch studies emerges a greater openness to innovative methods, but also a greater attention to the consequences on the labor market. One indication is that the implementation of measures of economic and health policy should be considered within the regulatory sphere.

The literature contributions for Northern European (United Kingdom, Finland and Sweden).
In Northern European countries (including, in this group, UK and Scandinavian countries), HCA still results the prevalent approach. In UK, significant studies have evaluated pathologies as cancer: for example, Round et al. (2016)  One of the weaknesses of the HCA-based approach is that it could underestimate indirect costs; costs, in fact, are estimated taking into account market prices and very low values is attributed to expenses done by retired older people or other frail people (as in the work by Mangalore and Knapp, 2007, which conducted different estimates for categories of not working subjects as people living in families, institutions, prisons and homeless people). The costs considered include health and social care, informal care, private expenses, lost productivity, criminal justice services and other public expenditure, such as the social security system, leeading to an estimated burden of indirect costs almost double comparing to the direct costs.
In a study presenting both methods of estimation, Liu et al. (2002) assess the economic burden of coronary heart disease in the UK by using both direct and indirect costs. After having estimated indirect costs using the HCA, the productivity losses were adjusted for FC: the article presents some considerations on the friction period to consider for the analysis.
However, in spite of some isolated works applying FC, the prevailing approach for the United Kingdom is based on human capital (HCA), without a consolidated tradition of literature based on FC.
Looking at Scandinavian countries, in Finland, the economic assessment studies are not many and look at individual case studies (as Haikonen et al., 2015, who examines the indirect economic burden of deaths caused by fires in Finland in the period 2000-2010: the HCA-based method is the main tool used to estimate productivity losses, and the potentially lost lives are also estimated). Puolakka et al. (2009), on the other hand, prefer the FC-based approach. In a study aimed at estimating the loss of productivity due to rheumatoid arthritis, the loss of productivity was calculated with both HCA and FC.
Instead, a greater attention to the economic evaluation studies is reserved in Sweden: here, the number of studies, which apply both methods for estimating indirect costs, is among the most copious in Europe.
Some studies are aimed at suggesting health policy indications: Ernstsson et al. (2016), observing studies on multiple sclerosis (MS), a disease whose prevalence in Scandinavian countries is higher comparing to other European countries, argue that studies centered on the cost of the illness are fundamental for planning adequate interventions.
These considerations concern rheumatological diseases, as rheumatoid arthritis, in the study by Hallart et al (2014), who use HCA to calculate indirect costs. The results can be used to analyze possible improvements in care, as in Persson et al. (2012), which estimate the economic burden related to the treatment of stroke. Estimates of indirect costs associated with breast cancer are based on HCA, although it is suggested to use the FC method as an alternative method: in a comparison of the two methods, it has been seen how FC produces smaller estimates of indirect costs (Lidgren et al., 2007).
The assessment of the pathology itself concerns a condition as depression, which is among the most common causes of disability and is associated with substantial reductions in individual quality of life.
In a study by Sobocki et al. (2007), it is measured both the direct cost of providing healthcare to patients, and indirect costs, measured through HCA approach.
HCA and FC can be compared looking at the results obtained: some case studies looking at costs of informal care in the context of home care (Andersson et al., 2002) include both HCA and FC approaches. While estimating indirect costs with FC, the amount is only 18% of total costs, HCA approach shows an amount of 44% of indirect costs. However, the main indication that comes from the study, is that the cost of informal care in evaluating home care programs is often underestimated, due to the exclusion of indirect costs.
Another case, which sees the application of the FC, is treated by von Thiele Schwartz and Hasson (2012), whose objective is to investigate the effects of physical exercise during working hours and the effect on the related direct and indirect costs. Johannesson and Karlsson (1997) propose the FC method for the estimation of indirect costs, although they recognize that this method is based on hypotheses not supported by the neoclassical economic theory. An important point concerns the fact that the FC-based method for indirect costs should also be applied in estimating direct costs, with the consequence of substantially reducing the costs of health care programs. In a study that is a corollary of those already conducted by the same authors, Koopmanschap and Rutten (1994) analyze the impact of the inclusion of indirect costs of the disease in the economic evaluations of eight health care programs. The impact of indirect costs varies widely between health programs: in particular, indirect costs tend to play an important role if health programs produce effects in the short term, if the absence from work is significantly affected, and if a significant part of the reference population is considered in the analysis.
Finally, Liljas (1998) describes the elements to be included in the calculation of indirect costs. The method proposed and based on FC excludes many aspects of the indirect cost components: for this reason, it should not be recommended in place of the HCA approach, if the economic evaluations aim at taking into account the patient's needs.

The literature contributions for Southern Europe. The Mediterranean area (Italy, Spain and Portugal).
The indication resulting from the guidelines developed for health economics studies is for the inclusion in any case of the indirect costs in the economic evaluation analyses. There is a preference for the HCA approach, confirmed by the largest number of studies carried out.
Examining the studies carried out for Italy, the recent work of Turchetti et al. (2017) estimate the average annual social cost per patient with chronic kidney disease for stages 4 and 5 pre-dialysis and cost components in Italy. Productivity losses for patients and caregivers were assessed using HCA. Marcellusi et al. (2015b) calculate the burden due to respiratory allergies; despite the economic and social burden are consistent, there are no precise data on the costs associated with the management of allergic respiratory diseases in Italy. The study performs a systematic review of the literature to identify both the cost per case and the number of affected patients, applying an incidence-based estimation method. The direct costs were estimated by multiplying the admissions, the drugs and the management costs deriving from the literature with the Italian epidemiological data. Indirect costs have been calculated on the basis of lost productivity estimated with HCA. This is the first study in which both direct costs (incurred by the health system) and indirect costs (incurred by the society) were taken into consideration to calculate the overall burden associated with respiratory allergies and their co-morbidities. The study can be considered an effective tool for public decision-makers to correctly understand the economic aspects involved in the management and treatment of diseases caused by respiratory allergies in Italy.
Again, Marcellusi et al. (2015a) estimate the impact of the hepatitis C virus and the economic burden of the disease, developing a probabilistic cost model. Indirect costs have been calculated on the basis of lost productivity based on HCA. The relevance of the problem can be understood considering that Italy is one of the European countries with the greatest number of people suffering from chronic HCV infection, the main cause of cirrhosis, HCC and liver diseases. These diseases induced by HCV generate high costs for the Italian health system as well as high indirect costs. Marcellusi et al. (2016) try to interpret available information on epidemiology of diabetes mellitus by estimating the average annual cost incurred by the health services and the society. Direct medical costs (drugs, hospitalizations, monitoring and adverse events) and indirect costs (absenteeism and early retirement) were estimated. The study has been the first to be conducted in Italy to assess the direct and indirect cost of diabetes (through HCA) with a probabilistic prevalence approach. The difficulty in finding the relevant information confirms that the real burden of diabetes is underestimated, especially as regards indirect costs. Mennini et al. (2014) assessed the indirect costs and the impact of bipolar disorder with HCA. The study stresses the relevance of indirect costs of this pathology, which affects mainly employed people.
The objective of the study by Garattini et al. (2000) is to evaluate the indirect costs in pharmacoeconomic studies in Italy and the attitude of Italian pharmacists towards indirect costs. A literature review was conducted, focusing on studies about economic evaluation of drugs. Eighteen studies were available for review. Although the methods used to calculate the loss-of-production value due to morbidity were all based on HCA, there was great variability between studies in practical methods. The parameters used to evaluate production losses also vary widely: of the 25 respondents, 20 considered it important to include indirect costs in pharmacoeconomic studies; 56% of respondents said that health authorities should request indirect cost assessments, that, as stressed in several Italian studies, show a misalignment regarding the methods used.
Another study carried out with the HCA method (Leardini et al., 2004) estimates the impact due to osteoarthritis of the knee, which is one of the main causes of disability due to rheumatic diseases, and require extensive use of health resources. The study involved 29 rheumatology institutes. Productivity losses to patients and caregivers and informal care were considered as indirect costs. The study confirms that the direct and indirect costs attributable to osteoarthritis of the knee are substantial. The losses incurred by workers, assessed with HCA, were measured in terms of wage variations (including the tax burden charged to employers) under the assumption that income reflects productivity.
In Portugal, studies based on FC are more recent over time and focus mainly on specific diseases. an older population in calculating the impact on early retirement due to self-reported rheumatic diseases. The study estimates the indirect costs and working years lost, employing a national database to calculate productivity values by gender, age and region, using HCA. The working years lost were estimated as the difference between the current age of each participant and the relative retirement age, while the potential years of working life lost were given by the difference between the age of the individual and the actual retirement age.
The same authors (Laires et al., 2015) evaluate the physical disabilities caused by rheumatic diseases, which can lead to a rapid exit from work and generate indirect costs to society. The analysis was based on prevalence through a bottom-up approach: health and sociodemographic data were collected for all people aged between 50 and 64, while an official national database was used to estimate productivity values by gender, age group and region, again using HCA. The early exit from work attributable to this condition is about 0.4% of national GDP. For this reason, public health concerns and the economic impact highlight the need to invest in health and social protection policies for patients with rheumatic diseases.
Finally, a neurological disease is the object of the analysis by Reese et al. (2011). Health and economic assessments of Parkinson's disease for southern European countries are limited: in this study, carried out between 2004 and 2005, costs were assessed from the societal perspective using health and economic questionnaires. The HCA was employed to estimate the indirect costs, and the quality of life related to health was assessed using the EQ-5D and outline the extent to which assistance from the family and other relatives play an important role. In general, the costs were lower than those reported in other Western countries: overall, it is stressed that the economic burden of Parkinson's disease in Portugal is considerable, and the study emphasizes that further research, that include indirect costs, is needed to describe effective health models and to guide health policy decisions.

In
Spain The latest studies identified refer to the economic burden of specific conditions, such as headache (Badia et al., 2004); the direct costs (due to drugs, health care, specialist, etc.) and indirect costs (working days lost and work reduction) were calculated using the prevalence approach. The HCA was used to calculate indirect costs. As in many other developed countries, migraine determines a significant economic burden in Spain, particularly in terms of lost productivity. The recommendation from the study is that developing strategies aimed at reducing indirect costs may lead to the reduction of productivity losses.
A similar study (Badia et al 2002) is based on the economic impact in terms of direct and indirect costs of mental health in the Canary Islands (Spain). The disease cost method was used, and HCA method has been applied to estimate indirect costs, that were related to premature death, short-term illness and permanent disability. The high socio-economic cost of mental health helps to quantify the size of the problem, to set priorities and allow a more transparent debate on this topic.
The last works that have been identified for this review deal with assessing the economic impact of road accidents in Spain. Bastida et al. (2004) use the disease cost approach for their analysis. While direct costs refer to healthcare costs, insurance administration costs, and material damage costs to vehicles, indirect costs have been calculated using the HCA-based approach.
The high socio-economic cost of road accidents indicates the need for the various administrations in Spain to implement preventive measures.
Also the study by López et al. (2001) assesses the economic impact in terms of direct and indirect costs of road accidents in the Canary Islands (Spain) in 1997 using HCA. Total indirect costs amounted to 18% of total costs due mainly to premature death and absenteeism from work. Although this study adopts a conservative approach, omitting the costs associated with pain and suffering, permanent disability and home-based care from the family, the frequency of road accidents and their consequences clearly show the need, for different Canary Islands administrations, to implement preventive measures.

Conclusions
The examination of the existing literature on indirect costs and methods of their calculation has been carried out with the objective to identify the method most likely to represent indirect costs. Health policy considerations that can be developed concern the costs for health assistance to plan for the patients: however, it is difficult to quantify the intangible costs incurred by patients themselves and their caregivers, as there is no consensus on the method to apply for the estimation of such indirect costs.
To sum up, the economic evaluation of health programs is now a strongly consolidated practice in all industrialized countries and, specifically, in European countries. From the evidence reported in this survey, it emerged that the literature is now vast and robust as regards the assessment of indirect costs and, especially, the estimation of the lost productivity, for different health conditions.
With the exception of the Netherlands and some other scattered studies, the prevalence of use of HCA compared to FC is unequivocal. From the methodological point of view, HCA is more robust since it has its roots in general economic theory and allow a more comprehensive analysis. Then, HCA is certainly preferable to FC for those countries where the labor market is characterized by rigid regulations.
Hence, in carrying out economic evaluations, HCA may be eventually accompanied, as emerged from some recent studies, by a cost analysis based on FC, so to confirm the results achieved.
"The authors certify that they have the right to deposit the contribution with MPRA".