In this study, we investigated the economic burden associated with hospitalisation for acute care of older adults, by evaluating health care related and non-medical OOPE in two care settings: a GEMU and IMW at IMSS. In contrast to estimating health care system costs it rather quantifies the costs that an individual patient, and their families or primary caregivers pay when faced with such events, through overall OOPE. It also investigates indirect costs by caregivers when they have to stop or reduce work hours, or hours spent in other income-generating activities such as informal and self-employment, due to their care giving responsibilities. As stated previously, these results add to a previous report in which geriatric outcomes were more favourable in the GEMU compared to IMW . Moreover, it takes into account indirect costs of acute care of elderly, rather than analysing only direct health costs; this opens a new approach to cost effectiveness analyses using a more holistic picture of the actual costs and the potential benefit of a GEMU . The results of our analyses suggest that overall OOPE by caregivers of a patient at IMW are significantly higher than for those caring a patient at GEMU with figures at IMW that are at least double of those at GEMU. This is the case for transportation costs as well as for food, non-medical OOPE. Mediators of these differences are related to the early detection of high-risk elderly in the GEMU and a closer approach with the caregivers, which in turn are more satisfied and confident and with less perceived need to remain in the hospital due to the coordination between health professionals and caregivers in GEMU settings [13, 25].
Even when overall OOPE are common when a family member is hospitalized, policy makers need to understand the economic burden these expenses represent for the household; future research is needed to approach these issues. Individual social costs faced by patients and their families due to hospitalization or other health care use are seldom taken into account, and when services or health care reforms are proposed, usually the attention is centred in the costs incurred by the health care delivery institutions, and little on the costs incurred by patients, families and primary caregivers, including health care related and non-medical OOPE, as well as indirect costs .
Taking into account that for the year 2011 the daily minimum wage in Mexico City was $5.12 US dollars  it is clear that non-medical OOPE can represent high expenditures, with care givers spending more than the daily minimum wage, in particular those caring for a patient at the IMW. The results show the high economic impact non-medical OOPE can have to households in Mexico, and in particular households with older adults.
With respect to the income reported by these care givers, comparing their average daily expenditures reported with their average daily reported income, OOPE represent approximately 35% and 19% of daily income for care givers at IMW and GEMU respectively. Even when analyses of catastrophic expenditures usually include OOPE with respect to full household expenditures, we can assume that these older adult households do not have many additional income sources, and thus these representing considerable expenditures.
Within the economic evaluation literature, the fact that bias may arise in using recall methods and self-reported data to gather expenditure information has been widely discussed. While some studies favour the use of diaries for the patients and their caregivers to fill out detailed information on their expenditures, and others have relied on direct recall of expenditures, there is no consistency in the results, and therefore no consensus on the optimal method. For example, a recent review by the World Health Organization  to investigate current evidence on measurement errors in self-reported household expenditure and health expenditure finds inconsistent results in the estimation of total expenditure by reporting method, number of questions or items included, the use of bounded or unbounded interviews, the type of goods the studies included, among other issues. Some of the results of the review show how even when diaries are considered to retrieve more accurate data, at least one fourth were filled at moment of pickup and therefore based on recall, they tended to produce loss of interest in their completion as time from event passed. Regarding the results, while some diaries produced lower expenditures versus face to face interviews, others report higher estimates in a diaries compared to surveys. Given that we do not have similar studies in the country to compare our results, it is hard to assert the exact accuracy of the self-reported data. Due to the highly assumed responsibility of caring, it is likely that carers oversee reporting small payments or related errands, and these may underestimate the current results. In addition, in calculating productivity loss, income loss is taken into account but we had no information on the total time the care giver spent taking up these caring activities, and therefore we could not measure other opportunity costs like loss of time forgone in leisure activities, carrying out other care-related errands, or specific care giver burden. These again, could yield an under estimation of the full costs of informal care giving for older adults in Mexico and their opportunity costs.
On the other hand, when assessing productivity losses, several methods can be used such as the willingness-to-pay approach; the human capital approach and the friction cost approach, being the latter the two most frequently used methods. In this context, the human capital method takes the caregiver’s perspective into account and takes into account all hours not worked and income forgone due to care giving and related activities (transportation, buying devices, etc.), taking into account potential and actual losses. The friction cost approach method on the other hand, aims at estimating actual production lost or costs forgone until that worker is replaced. There is no consensus on which of these methods is best with the human capital model often noted as overestimating actual productivity loss and the friction cost method being difficult to implement given its requirement of detailed data on labour market conditions [28, 29].
Future studies have to be generated in order to explore non-medical OOPE with respect to total household expenditures and investigate possible catastrophic expenditures. They also need to take into account the perspective of the carers and full information on all opportunity costs incurred by taking up care giving activities and the burden it may be generating, such as negative health effects.
In addition, given the economic burden on caregivers with patients at the IMW, the difference in indirect costs between the two types of services is considerable. This figure is also relevant given that IMSS has only one GEMU. Thus, costs to caregivers could be expected to be as high in these other units. From the results of the model estimations we can observe the relevance that type of service has in determining type of OOPE and level of expenditures. This data could also be useful for policy makers that face a growing demand of older population and their specific needs in order to include health services models that respond in a comprehensive way to clinical, economic and social aspects.
Studies in Mexico as in other countries [14, 30, 31] have showed that the intensity and frequency of care needed by older adults is undertaken by their partners, children, friends, or other siblings, all of them non-remunerated. Notwithstanding, health system institutions and studies of health care systems usually ignore that informal care represents an essential piece of health care and as such should be taken into account. In light of the accelerated increase in the ageing population in the country, policy makers need to urgently address the issue of unpaid, informal care for older adults. It is necessary to recognise its role and that it is thoroughly investigated at the patient, family-household, and societal level and most importantly, its impact on all these angles.
In this sample, the vast majority of caregivers were women who probably assume many other competing responsibilities. Thus, opportunity costs of informal care should also be systematically researched and the results used in the planning of strategies to support these caregivers .
It is interesting to see that even when older adults in the study were admitted to hospital and provided the acute health care needed; the vast majority required an informal caregiver to provide additional support. In the future, studies on informal care giving should include in-depth qualitative studies in order to investigate further about the care the institutions are asking them to provide, by type of activity, the frequency, and intensity of care required.
In planning for new services or restructuring care for older adults in Mexico, institutions should also look into social values, beliefs, and economic consequences around having a family member or a loved one hospitalized. Although it has been noted how it is culturally expected to have family around the patient all the time, further studies should investigate how much having informal care in addition to all health care received responds to cultural values and how much it responds to institutions transferring the burden and the costs to the household. Health care planners and decision makers should be aware of this economic burden; the opportunity costs involved, and generate strategies to alleviate it .
Given the results using this data, further studies should investigate the differences in care provided for older adults at an IMW and GEMU in order to search for the best alternative and a more efficient way to provide acute care for this population group. Although one of the reasons that care is taken up by the family has to do with cultural traditions, an increasing cause has to do with hospitals discharging patients sooner and transferring health care activities to a primary caregiver at home and this should be further investigated . One policy implication for Mexico is that in rationing decisions in health, many criteria like cost-effectiveness, equity and feasibility concerns play a major role. Relevant criteria have to be taken into account in order to make the best decision. In this report, an evaluation of health expenditures in a specific health care services setting, the GEMU, is added to a clinical evaluation previously reported . It is hoped that both criteria will be considered in future planning and policy decisions, and in setting priorities of care. Furthermore, additional research looking for strong evidence has to be run.
In the context of a rapidly ageing population, the results of this study add evidence to the fact that national level policies should recognise the value of informal care and how it is expected to change in the future. There is a great need to generate policies that get together health and social care in order to provide long-term care services for older adults in order to provide them with optimal care, reduce expenditures, hospital admissions and readmissions, and ultimately, relieve the families from an important part of this burden of care.
Finally, there are some limitations to the data on carers used in this study. Additional information on their occupation, education level, the type of care activity provided, and the total number of hours caring would be desirable in order to generate further, more detailed analyses. Although we did not have additional information on these characteristics, one could expect these carers, mostly women, to have many other competing responsibilities such as domestic activities, caring for children and grandchildren, work outside home, among others, with scarce or no support to aide them in these difficult tasks. Moreover, some authors assume that an economic burden is also represented by domestic lost work by these women . Qualitative information through focus groups and in-depth interviews in order to obtain more information on patient accompaniment values would have been desirable and should be included in future studies in order to have more complete information for the planning and decision making process regarding acute care for older adults.