Care Functions within the Kinship Network: Explaining Care Arrangements and Female Health Choices in Post-Disaster Nepal

Abstract This article investigates the relationship between kinship-based care and the phenomenon of women turning down the offer of free treatment in a hospital. As care is perceived and enacted very differently all over the world, the article aims to outline an approach to researching care and care-related health decisions that represents actors’ perspectives. It does so by comparing care-responsibilities, marriage and residence patterns in two ethnic communities. Data was gathered during six months of qualitative research in several health camps after the 2015 earthquakes in Nepal. The results show that what I call the ‘chain of care-responsibility’ (the persons taken into consideration as primary or alternative caregivers) diverges widely between ethnic Tamangs and Chhetris of the research area. The challenge of finding persons to fulfil care functions and to replace working power may lead to substantial changes in household formation including travelling kin, polygamy, or divorce. Under otherwise similar conditions, such considerations of care and replacement place Chhetri women at structural disadvantage in accepting free medical treatment as compared to Tamang living in ‘traditional’ clustered settlements. These findings suggest that referral programmes would profit from partnering with local communities to develop practical solutions to questions of care and replacement.


Introduction
The Himalayan State of Nepal was hit by two successive earthquakes in April and May 2015. They led to great damage to the health infrastructure and disrupted health services in all 14 affected districts (Government of Nepal [GoN], 2015, pp. 13-14). Health services and infrastructure remained interrupted even months later (World Health Organization, 2015) leaving not only old, disaster-related injuries, but any kind of medical condition without the possibility for treatment. In an attempt to fill this gap, large-scale health camp projects were conducted by international and national development and relief organisations. As the health camps were not equipped for surgery, cases in need of such care were offered referral either to the district result of female status, power structures and domination (Fine, 2005, p. 146;Furuta & Salway, 2006, p. 17; for examples of these approaches see Allendorf, 2007;or Juran & Trivedi, 2015).
The data discussed in this article will add a different angle to these frameworks and interpretations. To do so, it follows a more relational approach to gender and health that seeks to do justice to agential, dynamic and situational issues (Butler, 1990;Lamb, 2000;West & Zimmerman, 1987). This growing body of literature is highlighting the potential of including intersectional aspects of gender to broaden the understanding on social determinants of health (Springer, Hankivsky, & Bates, 2012). An intersectional perspective insists that sociological characteristics of an individual and their ramifications (such as ethnicity, gender, age and racism, sexism, and so forth) never operate in isolation and independently, but rather intersect, work simultaneously and transform each other. These authors challenge the 'assumptions that gender [ … ] is always the primary axis through which to understand specific health experiences, conditions or even access to health services' (Springer et al., 2012(Springer et al., , p. 1665).
2.2. Concept(s) of care 2.2.1. The multiple dimensions of care. The literature engaging with care covers an extremely wide range of approaches and disciplines, dealing with issues and activities from informal child care to professional medical services (Alber & Drotbohm, 2015, pp. 2-3); from body care (Lidola, 2015) to overseas remittances (Drotbohm, 2015;Gamburd, 2000). It frequently touches on different aspects of a range of topics at the same time, such as gender, work, ethnicity, affect, or mobility regimes and control (Alber & Drotbohm, 2015, pp. 2-3;Constable, 2020;Johnson & Lindquist, 2020). At its core, care goes far beyond the mere economic or practical aspects. It is both resource and relational practice (Buch, 2015: 279). It may be thought of as a product of social relations, a commodity on the labour market or part of a wider, institutionalised social service or state policy (Barg, Keddem, Shiekman Cohen, & Henderson, 2014, p. 180). New approaches to kinship have dissolved kinship as something given either by 'nature' or by 'biology' and have argued that the heart of kin ties is marked by care (Carsten, 2000).
2.2.2. The multi-layered and ambivalent nature of care. The theoretical groundwork on care has been strongly influenced by its inherent connotations of term and practice (Thelen, 2014(Thelen, , pp. 37-38, 2015. In its everyday Western understanding the English term care is closely connected to the ideas of love and altruism (Fine, 2007, p. 8;Glenn, 2010, p. 10;Thelen, 2014, p. 23) and perceived as reflecting central human values of support and reciprocity (Thelen, 2015, p. 140). As several authors note, this has led to a normative and highly moralised discourse (both public and academic) on the social practice of care (Alber & Drotbohm, 2015, pp. 3-4;Drotbohm, 2015;Thelen, 2015).
At the same time care is a word that is pretending more comparability than there actually is. It is what Rodney Needham (1972Needham ( , pp. 124-125, 1983) with reference to Ludwig Wittgenstein (1958) terms an 'odd-job' word. Like 'belief' or 'marriage', 'care' presumes similarity but effectively obscures both conceptual and contextual differences. This polysemic character of care (Buch, 2015, p. 287), with its multiple meanings and interpretations, has ramifications for both theory and methodology. Thus 'care' is a fluid concept that defies a unifying definition. In fact, scholars still struggle to extract the 'overall similarity' (Needham, 1983, pp. 58-59) which is needed to define a concept (see also Thelen, 2014, p. 25). With differing theoretical approaches and resulting shifts in perspectives (for example, from care as relational practice to care as a resource), the concept as a whole remains fluctuating and vague. Without analytical clarity and preciseness however, care is primarily a notion (compare Needham (1972, p. 61) on 'belief'). With regional and cultural variations of care practice and understanding any observer is at risk to be caught up in their own perspective and their own understanding of care.
2.3. Care as a function in (culture-specific) networks of social security An analysis of care decision-making needs to take into account that social roles shape the expectations of receiving and providing care (Alber & Drotbohm, 2015, pp. 2, 10) and thus largely influence who cares for whom and how. On the one hand, the rights and obligations of care vary between gender roles (Alber & Drotbohm, 2015, pp. 2, 10). Thus some individuals may be considered eligible or designated for certain tasks while others are not. On the other hand, these social roles also change over the life course as they correlate with different locally specific constructions of life stages (Alber & Drotbohm, 2015, pp. 2, 10). Consequently, whenever individuals change their position within the kinship network, for example through marriage or childbearing, their responsibilities and obligations may change. At the same time, new members to the family may constitute new resources for providing care whenever it is needed. The social positions and corresponding roles are forming a network of social security.
To analyse the resulting care arrangements, I conceptualise care as a 'function' which needs to be fulfilled to solve the situation of need. The definition of function in the 2019 Oxford English Dictionary (OED Online, 2019) includes: (1a) A person's role [ … ]; (2a) An activity or mode of operation that is proper or natural to a person or thing. Care combines these meanings: Persons are considered as either possessing a role as a caregiver for a particular person or not, and activities around care are bound to certain social positions and roles. This also applies to activities indirectly associated with care arrangements, like replacing lost working power. As Rosie Read and Tatjana Thelen stated, 'care as a dimension of social security [ … ] consists of practices that address socially constructed needs that have a giving and a receiving side' (Read & Thelen, 2007, p. 7). Although the resulting care relations may be situational and contextualised, supplemented through non-kin relations, or impeded by hostility, they are mapped on culture-specific patterns of dealing with perceived need. As will become evident below, these properties of the polysemic concept of care will allow us to analyse structural differences of the networks in which women are situated. In line with the so-called 'new kinship studies' with its focus on processuality rather than rigidity (for example, Carsten, 2000), this approach can address both influences of kinship structure on care, and influences of care on kinship. It will help to explain not only kinship-based care choices but also health choices in which care-taking roles are considered.

Field site and methodology
The empirical data for this article was collected during six months of fieldwork in Nepal from October 2015 to April 2016. Funding was provided by the German Academic Exchange Service. I adopted the role of an independent but associated researcher in a large-scale health camp (HC) project in Dhading and Nuwakot District. Eight national medical teams conducted 93 HCs and reached a total of 50,436 patients. Each camp was scheduled for three consecutive days of treatment in which the teams provided medical consultation, pharmaceuticals, and selected laboratory services ( Figure 1). I had direct access to patients, their families and social surrounding, local leaders, NGO and hospital staff, as well as to other actors in the humanitarian field. Informed consent was obtained with the limitations for informal conversations and participant observation mentioned in the ASA Ethical Guidelines for good research practice. The research was approved by the University of T€ ubingen's Research Ethics Committee (Department of Social and Cultural Anthropology).

Participant observation
I visited five HCs and their partner hospitals. Subsequently, I joined one of the medical teams for the entire duration of three HCs in a Tamang-dominated area referred to as Ankhu Khola Valley (Toffin, 1976) in the remote north of Dhading District, as well as in Goganpani, a multi-Care functions within the kinship network 555 ethnic area in the south of the district with mainly Bahun-Chhetri and Tamang families. Situated within 2 h walking distance from the main road between Kathmandu and Pokhara, Goganpani is part of the so-called hill region of Nepal. It is not considered to be particularly remote. Ankhu Khola Valley in turn could only be reached by a journey of several days from Kathmandu, including two days of walking. I returned to these villages after the HC project was finished in February and March 2016, to spent three weeks in Goganpani and another two weeks in Ankhu Khola Valley. Living with local families enabled me to observe daily life, health-seeking behaviour, the cooperation with other community members and the neighbourhood. It also provided insight into intra-familial interactions such as care work and bedside manner, activities according to gender and age, workload distribution and cooperation, and decision-making processes. These observations allowed me to confirm and substantiate what I had learned from the interviews.

Interviews
I conducted a total of approximately 60 interviews, both formal, semi-structured interviews (N ¼ 31) and informal interviews (N > 30). While the former followed various flexible interview guides, the latter were open, in-depth discussions on the topic. In line with common practice in qualitative research, I used purposive, at most representative sampling until a point of information redundancy and saturation was reached. The main interview partners were participants of the HCs, health post staff, Female Community Health Volunteers (FCHV), doctors, community members, NGO staff, caregivers and referral cases. Additional phone interviews (N ¼ 6) three months after the HC with some of the women who were offered referral served to complement previous inquiries. Special attention was given to the following topics: 1. Care-responsibilities, both actual (Who provides care in this specific case? Who provides care in other families?) and hypothetical (Who would care for you if you needed care? Who would care for a woman in other families?) 2. Details of care behaviour including activities undertaken by whom and why 3. Consequences of the illness/injury for both the nursing case and the household 4. Reasons for or against healthcare seeking as stated in interviews or reported via health personnel 5. Household decision-making around healthcare seeking 6. Medical history and health behaviour 7. Household formation and distance to kin The diversity of interview partners allowed to compare multiple perspectives and to combine different levels of information on these topics. Interviews were conducted in English, Nepali, or Tamang. A translator was used when the conversation was held in a language other than English.

Scope, limits, and validation of data
Ethnographic data was analysed in a combination of inductive reasoning and strong engagement with literature both before and after data collection. Discussion and feedback with research assistants, community members, and NGO personnel were used to reduce bias and to preclude misinterpretation (Hampshire, Panter-Brick, Kilpatrick, & Casiday, 2009). To achieve saturation and to ensure the validity of my data, the principles of triangulation of different methods (interviews, participant observation) and crosschecking (with different informants and sources) were applied (Fusch & Ness, 2015). While the qualitative nature of this research does not provide numerical results, it provides rich insight into the localised care practices of and for women. Care patterns concerning men, the quality of care, or general healthcare decisions were not the focus of the fieldwork.

Ethnographic background
To understand both the health decisions of these women and possible care arrangements, it is vital to take a closer look at the ethnographic background of the area. Nepal is a multi-ethnic and multi-religious country. Different communities often do not only differ in terms of religious Care functions within the kinship network 557 belief, language and customs, but also in their social organisation, gendered role allocation and kinship structures. This article focusses on the social and kinship structure of Chhetris and Tamangs. I encourage the reader to delve into the life worlds described below and to develop an awareness for the different options and varying logics underlying care arrangements world-wide.

Chhetris
The Chhetris are high-caste Hindus who belong to the dominant Parbatiya population of Nepal (Gellner, 2007(Gellner, , p. 1823. Chhetri households usually consist of a lineal joint family (a couple and their married sons) with a tendency to fragmentation into nuclear family units. Lynn Bennett's (1983) classic monograph has described the life of high-caste women in great detail. Today, decades later, the norms and values my interlocutors referred to still resonate with her descriptions.
Within the household, activities and responsibilities tend to be clearly allocated according to sex and position in the family hierarchy. While men are expected to provide the cash and many of them seek employment outside the villagereturning on weekends or holidayswomen take over most of the household-and agriculture-related day-to-day activities (see also Regmi, 2016, p. 283). The hardest and largest share of work is usually done by the youngest daughterin-law (DIL), who is also expected to serve her mother-in-law (MIL) in everything (see also Bennett, 1983, pp. 23-28).
For a Chhetri woman her marriage constitutes a profound change in their life. After the wedding, she usually leaves her natal house (Np. maita) to live in the house of her husband's family (Np. ghar, lit. house). Following an exogamous marriage pattern, she is likely to marry outside her village and gotra (clan). A dominant discourse in society portrays the marriage as a complete break with a woman's natal lineage. From the moment of marriage, she lives in distance to her natal house and visits only in exceptional occasions. She is expected to contribute her fertility and labour to the affinal patrilineal group, who in turn are responsible for her well-being (see also Bennett, 1983, pp. 165-166, 244-46, 328;Weiss, 1999, p. 250).

Tamangs
The Tamangs are the largest Tibeto-Burman-speaking group in Nepal (Fricke, 2007, pp. 29-30;Holmberg, 2005, p. 17). Although Tamangs mainly practice Tibetan Buddhism mixed with Hindu and shamanic elements, there is an increasing number of conversions to Christianity (Fricke, 2007, p. 31;Ripert, 2014, pp. 51, 54). There are considerable differences between different Tamang groups in the country. My field observations focussed on so-called North-western Tamangs who live in in Upper Ankhu Khola Valley.
The economy in the area is dominated by subsistence agriculture and increasingly remittances from labour migration. Generally, all who live or have once lived in the same household constitute a potential labour resource (Fricke, 2007, pp. 130, 158, 166). Traditionally the youngest male child will stay with his parents and eventually inherit the house, while the other sons and their wives will leave the common hearth after some years and build new houses nearby. The resulting cluster of agnates living next to each other provides rich ground for inter-household cooperation (Fricke, 2007;Holmberg, 1983, p. 46;March, 1997, p. 152).
Many Tamangs of Ankhu Khola Valley follow positive marriage rules which prescribe bilateral cross-cousin marriage; that is for a man either the father's sister's daughter or the mother's brother's daughter; and for a women her father's sister's son or her mother's brother's son (Fricke, 2007, p. 31). Despite patrilineal descent and patrilocal residence, the marriage of a woman results in a 'dual affiliation' (Holmberg, 2005, p. 78) which is characterised by continuing involvement and (labour) obligations in both lineages and households (Holmberg, 2005, p. 78;March, 1997March, , p. 153, 2002 Figure 2).
It is important to keep in mind that social organisation is neither rigid nor uniform. There are differences between regions and households. Love-marriage is on the rise in many areas of Nepal (Ahearn, 2004) and ever-increasing labour migration and growing education expectation brings about rapid transformations of village life (March, 2018). Nevertheless, my data shows how different patterns of marriage, residence, and domestic organisation lead to fundamentally different preconditions in planning care constellations.

Care functions: a comparison of Chhetris in Goganpani and Tamangs in Ankhu Khola Valley
To elucidate care decisions, I address the perspectives of the actors themselves on possibilities and expectations regarding care arrangements. 'Care' in this case designates the practices considered necessary by my interlocutors to meet the practical, everyday needs of a nursing case: such as preparing food, assisting with bodily hygiene, treating wounds, and washing clothes and dishes.
5.1.1. The Chhetri chain of care-responsibility. It is very rare that a Chhetri husband nurses his wife and takes over her activities in the research area. Instead, these care aspects are expected to be fulfilled by other women. Even though subject to individual negotiation, it is the wife who is considered responsible for the well-being of her husband and family, both in spiritual concerns and in everyday life. A reversal of this marital relationship by a husband effectively serving his wife instead of being served, taking over her functions, and performing female activities Care functions within the kinship network 559 would usually expose him to ridicule. Similar constraints affect other senior male household members.
There are men who overcome the societal framework conditions, such as the husband of K. Thapa: a couple of years ago, the 36-year-old Chhetri woman was diagnosed with uterine prolapse and had to undergo surgery. Infections and complications after the surgery left her unable to fulfil her daily activities for almost a year. During that time, she was cared for by her husband, who also undertook all her tasks in the household. Yet, to hide it from the neighbours, he only did it in the evenings or at night. His wife explained his behaviour by societal norms: 'the neighbours think that only women should do housework, everyone thinks that. They gossip a lot'. Her account is typical and correspondingly this option is usually not taken into consideration when planning care arrangements.
According to the statements of my interlocutors in Goganpani, it is generally the DIL who is responsible for nursing other household members in Chhetri households. In the rare cases that there are several DILs in one household, they may care for each other.
Whenever there is no DIL available or she is the one in need of care, the care function falls on the closest female household members which is either an unmarried sister-in-law or the MIL. Most MILs, however, would hesitate to care for a DIL, as it is largely incompatible with their social roles. This is vividly reflected in the following statements of three Chhetri women: Sometimes a mother-in-law cares for her daughter-in-law. But usually she fears to lose her social standing if she cares too much. (S. Budhathoki) A mother-in-law might not dare to touch her daughter-in-law. (S. Karki) A mother-in-law will never touch the legs and feet of a daughter-in-law. She will not give massages. She thinks she would be cursed if she did. Both, society and the mother-in-law herself think that she should be highly respected. (D. Katwal) An open wound which needs treatment for example, or a DIL during and after childbirth would combine the pollutant nature of bodily substances in Hindu thought, the DIL's lower standing in hierarchy, and the principally harming potential of sexuality and pollution associated with a young woman in her reproductive years (for more details on pollution and hierarchy see Bennett, 1983;Kondos, 2004, pp. 4-5;Lamb, 2000, p. 185). Instead, her own mother or a sister are considered the ideal caregiver for a DIL. The prevalent practice of village exogamy in Chhetri marriage, however, implicates considerable distances between a woman's affinal home and her natal house. Consanguineous caregivers would either have to travel and stay for the time of recovery, to commute between the two places, or to bring the person in need with them. Yet all these options face substantial impediments. It is not considered appropriate for consanguineous kin of the DIL to stay in the affinal house of the DIL. Their lower stand as bride-giver in a hypergamous marriage and the hesitation to take anything from ritually higher persons (Bennett, 1983, pp. 144-145, 149) would strongly contradict the occupation of a guest role in the house where the daughter is married. Moreover, high-caste Hindu women avoid travelling alone as it would affect their reputation and that of the family (see also Lamb, 2000, p. 191). Finally, potential caregivers have duties in their own families which would either have to be omitted or given to someone else. The distance between the different households complicates the issue. Neighbours are said to help in an emergency but are not included in the planning of long-term care arrangements.
Any care-dependent DIL whose children are not yet old enough to provide care to her is consequently in the most difficult position in the household in terms of finding a caregiver. It is quite usual for a DIL to return home for childbirth for about a month so that consanguineous women may take over postnatal care (including washing her clothes and dishes which are considered especially polluted). Yet, apart from these birth-related visits and a few brief visits for ritual obligationslike the two important festivals for siblings: bhai tika and tij)her proper place is seen in her affinal home only (see also Bennett, 1983, pp. 244-245;Weiss, 1999, p. 244).
Nevertheless, if she has a severe medical condition a return of the DIL to her maita may be the only option to obtain comprehensive care. It is a situation which frequently leads to divorce and remarriage of the husband. 5.1.2. The Tamang chain of care-responsibility. The care situation among Tamangs as compared to Chhetri families is in many respects a different one. The threshold for men to care for their wife and to take over her activities as long as she is not able to fulfil them is considerably lower. Despite a trend for increasing segregation and hierarchy in Tamang gender relations (March, 1997, p. 168), femaleness is not as separated from the male sphere as in Chhetri families: 'Tamang couples are not only husband and wife, they are friends also', a Chhetri woman explained. While her statement is not to be taken as an indicator for marital quality, it still points towards the profound conceptual differences as compared to the ideal Hindu marriage (see also Holmberg, 1983;March, 2018). Tamang affinal men and women are neither subject to high-caste respect-avoidance behaviour rules, nor does the hierarchic principle entail the same inhibition in terms of daily interaction and communication.
There is an overall tendency in the research area to prefer female over male caregivers, but there are also households in which the son or grandson undertakes nursing activities including those which involve physical contact such as bathing the patient or changing clothes. Often care activities are shared between various persons, with the exact allocation of tasks according to personal attachment, capacity, and availability. In the case of a senior married woman for example, any person who lives in the household will be directly responsible for contributing to her well-being. 'If a mother is ill, all will care', explained 23-year-old Y. Tamang who lives with his wife in his parent's house.
Usually, the youngest son will inherit the house and thereby obtain the responsibility for the well-being of the household, while his wife is considered responsible to support himespecially, in providing care for the parents in old age. If they are not able to fulfil these duties for any reason (labour migration, poor financial condition, or physical impairment), an older son and his wife will be expected to step in; either by returning to the house or by taking the nursing case with them.
A daughter's responsibility to provide care for members of her natal house (dim) equally continues beyond marriage and moving to live with her husband. A care-relationship between a mother and daughter is allocated a special quality and she is likely to supplement other potential caregivers.
A daughter-in-law cares, but never as well as a real daughter. She is the daughter of another mother. I am also more attached to my mother than to my mother-in-law. And if my mother needs care, I will also need to help. (A. Tamang,24) An important implication of Tamang cross-cousin marriage is the spatial closeness of the different descent groups. In Upper Ankhu Khola Valley the affinally linked households are often located in the same village. And even outside the bounded Upper Ankhu Khola Valley, they are almost never further than a few hours' walking distance apart (Holmberg, 2005, p. 78). Distance between nursing case and potential caregiver is thus not the same problem as among Chhetris. This is especially so, because women commute freely between their natal and marital homes and frequent visits to their family of origin are not only tolerated but an institutionalised and central mode of interaction between two patriclans. The resulting flexibility in movement allows consanguineal women to care for each other even after they are already married.
From the perspective of a care-dependent DIL cross-cousin marriage entails affinal relationships that are built on premarital familiarity. A woman is not only related to her MIL through her husband, but her MIL is simultaneously her paternal aunt or her maternal uncle's wife. This twofold kin-linkage is reflected in a mutually adjuvant relationship between the two women. They share domestic responsibilities and care may be generally provided in both Care functions within the kinship network 561 directions. Thereby they appear from a Chhetri perspective 'almost like mother and daughter' (D. Katwal, Chhetri nurse).
Nevertheless, the preferred care constellation for a DIL includes the own mother or a sister who will be expected to take over a considerable part of care-responsibility. The spatial closeness of the woman's natal home would enable them to come for frequent visits. The same applies to other kin whoin case there is persisting need in lack of carer or working powerare expected to play their part.
In general, the overall pattern of responsibility follows a principle of consanguineal closeness: it is the own daughter and son (including the DIL) for a senior woman, and the own mother or sister for a DIL. Brothers are also responsible to support sisters in case they return to their natal home. The more distant the degree of kinship the less pronounced and precise the allocation of responsibility. Instead they give way to more general statements such as 'clan-sisters'. Yet, beyond the role of the persons directly associated with her natal homethe mother, father, sisters, brothers and their wivesthe statements of interview partners concerning responsible caregivers differ, leaving the impression that care-responsibilities of marital kin and neighbours are strongly related to spatial closeness and availability and personal relationships.
It is possible that the nursing case goes to her natal home and returns when recovered. During this time, her domestic tasks and responsibilities will be covered by other members of the household or wider kinship network. Even though divorce and remarriage (of both spouses) is a frequent phenomenon in Tamang families, I was assured that it is not a common reaction to injury or illness. If the woman is still childless and unlikely to recover, however, her husband may marry a second wife.

Care, replacement, and constrained healthcare decisions
As mentioned above, many women in the HCs rejected the offer for referral. Several months after the earthquake itself, the HCs saw very few injuries directly related to the earthquake. Instead, most serious cases were related to female reproductive health and thus more women than men were offered referral. No overall data on rejected offers or the percentage of male versus female referral cases is available. But the HC personnel uniformly agreed on the difficulties in convincing women to accept the offer (as compared to men). In the camps I accompanied, most women rejected. As will become apparent below, these health questions are ultimately practical questions of care functions and arrangements, in which kinship structures and care functions intersect with questions of organising daily livelihood.
In general, decisions concerning care constellations for these women connote three central aspects: (1) The nursing case will need a suitable caregiver, who willdepending on circumstancesprovide care at home, come over for visits, or accompany her to the hospital for further treatment. As the nursing staff in Nepali hospitals do not undertake all needed activities, this person would also provide care in the hospital itself. (2) For the time of bodily impairment or absence due to healthcare seeking, she will be unavailable to fulfil the duties allocated to her social position. (3) The situation will also bind the working capacity of the carer.
In consequence, any care-intensive health decision involving a longer stay in a hospital, especially surgery, is bound up with the following questions (1) Who may provide care in general?
(2) Is this person available? (3) Is there available replacement? Whether the women considered the referral to a hospital as an option or not was almost exclusively related to how they could answer these questions, either explicitly, by stating that they had no one to come along, or implicitly, by referring to the difficulties of managing time and workload.
A case in point is the story of M. Thapa, a 55-year-old Chhetri woman. She was diagnosed in one of the HCs with third degree uterine prolapse, a condition in which the cervix of the uterus protrudes from the vagina. In a phone interview three months later, she was still unsure whether to take advantage of the referral offer or not. It was difficult for her to find a solution for the time she would be absent. Even though her 22-year-old son who still lived at home offered to help as much as he could, she insisted that both he and her husband were too busy with earning money and farming and were not the right ones to provide care. The only person who would be suitable was her oldest daughter. But being already married and living a severalhour drive by bus away as result of her exogamous marriage, she was considered to be part of her new household and not taken into account to fulfil care functions in this case. 'As you know', Ms Thapa said, 'it is difficult for a [Chhetri] woman who is already married to return to her maita [Np. mother's house] for a longer time'.
Conversely, the perceived possibilities for finding a carer and replacement may have a direct positive impact on healthcare seeking, as the following statement of a 67-year-old Tamang woman demonstrates: The decision was easy. Both I and my husband needed surgery. My [married] daughter came with us for three days to the hospital and then helped us at home. After four days, we did not need any more help Her narration reflects a care constellation which would be very unlikely for Chhetris. It is, in fact, the very constellation which would have significantly simplified the case of M. Thapa, above. Typically, the chain of care-responsibility in Chhetri cases, and thus the precondition to accept referral, was dependent on the availability of a DIL or an unmarried woman of the patrilineage. A 60-year-old Chhetri woman, for example, who had accepted, explained how her DIL accompanied her to the hospital, did all housework and cared for her 'like a daughter' (while her own daughter was expected to stay with her respective in-laws). Another Chhetri woman (whose daughters were equally all married) accepted the offer, as her husband's unmarried sister was available to accompany her.
The HC personnel noticed substantive differences in the ways in which the different communities could respond in such situations. One of the coordinators explained the perspective of Chhetri women who were offered referral to a hospital as follows: It is so hard to convince them [the Chhetri women]. They have work at home. They tell me: 'Who will feed the cattle? Who will do all the housework?' And even harder for them is to find a second person who can leave daily work [to care] for an uncertain amount of time. And they are afraid that they won't be able to work afterwards but will need rest.
Tamang women equally had to balance workload, time, and availability. Yet in line with the typical chain of care-responsibility and modes of social organisation, their cases demonstrated a tendency to greater flexibility in finding solutions. The coordinator of another HC explained: During the harvest season, they prefer not to go; [But] if they have to, because of the severity, then their family members from the maternal side, and also from the husband's side come and help.
[ … ] Their relatives [live] in the same or near-by village, it is much easier.
Similar flexibility applies to the gender of escorting relatives. The same HC coordinator gave the example of a Tamang women who came to the health camp after a miscarriage: She was brought by five or six men, including the husband and other male relatives. In Chhetri [households] men do not come in that way [ … ] and women are accompanied by elderly women or the daughter-in-law.
'Definitely, it is much easier for Tamang women,' she concluded. My point here is not to claim a per se higher likelihood of Tamang women to seek healthcare. The clear tendencies in dealing with care situations, get complicated by intersectional mechanisms, and influencing factors may interact differently with different contexts and individual circumstances. Varying levels of medical knowledge, shyness vis-a-vis medical teams, and fear of surgery in both communities may superimpose on the effect of care networks in some settings, but not in others. Financial status is likely to affect the pressure of earning and dependence on human labour. Changing marriage practices, such as love marriages between classmates, can lead to closer distances to caring kin in Chhetri cases, while systemic advantages may get lost if Tamang social constellations get altered, for example through death or outmigration (compare Kathryn March (2018) on effects of migration on Tamang gender relations). One of the Tamang women explained how she had to change her mind about her uterine prolapse surgery after her husband (and primary caregiver) had suddenly died. She moved to the city to live with her son who worked all day while the DIL ran a small restaurant. The woman recounted how all three of them had too much work to make ends meet and that she had decided to decline the treatment. Pressured by their financial situation and without the care network of the village, the family did not find a solution.
Despite the multiplicity of influencing factors, all of these ethnographic examples above show that the decision to accept or refuse the offer for referral is ultimately condensed to a very practical problem. It is a decision for which different actors can think of a different set of options; for which the actors' 'horizons of choice' (Boholm, Henning, & Krzyworzeka, 2013) differ widely. Understanding such systemic differences in dealing with perceived need, and supporting women and household heads in finding practical solutions, should be an important component in project design.

Care functions and kinship structure
A central feature of the tending-barriers or solutions which are discussed here is that they are not necessarily a reflection of personal feelings. A Hindu MIL will probably not perform all aspects of care to her DIL, but she may still care about her, and pay for another carer. Although tension and rivalry between MIL and DIL are quite frequent, even a MIL who feels close to her DIL and responsible for her well-being may rather pay a woman of lower standing than fulfilling the tasks herself. Such avoidance behaviour vividly shows the incompatibility of certain social positions (MIL) with certain tasks (nursing a DIL). We also had the example of a caring son, who was nevertheless not considered eligible to care for his mother in the hospital. On the other hand, certain positions (for example the daughter or DIL) are designated to perform care following their role expectations. These observations underline the importance to strictly distinguish between caring for someone and caring about someone (Ungerson, 1983, p. 31)between activities and services of support, and the emotional and affective dimension (Alber & Drotbohm, 2015, p. 2).
As indicated above, care may be conceptualised in various ways: as a resource (Buch, 2015, p. 279) either scarce or abundant, as relational practice (Buch, 2015, p. 279) which may be provided or resented, as a commodity (Barg et al., 2014, p. 18) to be exchanged, or product of loving relationships (Barg et al., 2014, p. 18). All these approaches have their justification and may grasp certain aspects. Yet, seeing care as a function, and to acknowledge (non-judgementally) that some persons will be considered eligible or even designated to fulfil this function while others are not, opens new perspectives on obstacles and means for the provision of care. It shifts the focus from care as a social practice and matter of emotions, compassion and helpfulness, to the network in which the person in need is situated, from the ethics of care to the options for care. After all, if certain aspects of care can or will only be performed by certain persons, these persons become a pivotal resource in the planning of care constellations (Figure 3).
In assessing the eligibility of caregivers, two aspects proved to be particularly prominent: role expectation on the one hand and consanguineous kin on the other (Alber & Drotbohm, 2015;Weiss, 1999). An important difference between different kinship systems discussed above is the access they provide to the latter.
Cross-cousin marriage and village endogamy allow Tamang women access to members of both their natal and affinal lineage, and major caregiversthe DIL, but also daughters, sisters, and mothersmay commute freely between the houses, even if they are already married. The formal lineage conversion of Chhetri women in turn, restricts the number of possible carers in the first instance on the eligible women within the household and female nursing cases would be expected to be cared for by a DIL or a daughter who is still living there. Severe nursing cases in their early years of marriage may fall back on their natal kin. This involves considerable effort and is for temporary care arrangements, apart from childbirth, usually not taken into consideration. Yet it is a vivid demonstration of how eligible caregivers may affect the locality of care. Interestingly enough, this role of the maita is in stark contrast to the dominant discourse in society that portrays marriage as a complete break with the natal lineage (Lamb, 2000, p. 208;Raheja & Gold, 1996, pp. 18-19). In this respect, care reflects the continuing bonds that connect a woman to her natal kin.
Yet care does not only reflect 'relatedness' (Carsten, 2000) and perceived modes of belonging, the need for care may equally cause divorce and the dissolution of kin ties. This sort of 'de-kinning' (Howell, 2006) and subsequent remarriage of the husband which can be rather observed in Chhetri than in Tamang casesjust like the occurrence of polygamy in Tamang casesmay be interpreted in various ways. On the one hand, an outside perspective may perceive it as disconcerting that a person in need is 'simply' replaced or taken somewhere else. Cultural background shapes the notion of care (Barg et al., 2014, pp. 177, 180). And if, for example, care is understood as something done by relatives, an observer's specific notion of relatives and their responsibility mayprecipitatelylead to an interpretation of such cases as negligence. On the other hand, a perspective which takes the structure of the kinship network and the positions and possibilities for care constellations and work replacement therein into account provides an alternative source for interpretation. The unavailability of a woman in case of long-term sickness or severe bodily impairment creates a gap, not only in household economy, but also in both her reproductive role and responsibility to care for other household members. A second marriage with a healthy wife constitutes a solution. This is especially obvious in the case of Chhetri women where the often rather small number of household members and available kin are exacerbating factors. At the same time, care needs to be negotiated among those who can provide it (Drotbohm, 2015, p. 110). This negotiation of providing care may result in the movement of carers. It may cause a Chhetri woman to call upon her son's wife if he moved out, or Tamang women to commute between households. But it may also cause the shift of the nursing case to the natal home. The maita is a place where a Chhetri woman can be taken care of when social roles in her affinal home are too incompatible, and where she moreover feels comfortable among her consanguineous relatives. Tamang women equally prefer consanguineous taking over this position. It is a reordering guided by the chain of care-responsibility and may thereby lead to substantial changes in household formation. Seen in this way the shift of a nursing case to her natal home, the divorce of a Chhetri woman and the remarriage of her former husband, as well as possible polygamy in Tamang families display culture and context specific coping strategiesfor a need for care on the one hand and its inherent call for replacement on the other. The comparison of Chhetri and Tamang families showed that the network of possible caregivers, resulting from the combination of their eligibility and availability (influenced by factors such as social role, position, capacity, locality), results in fundamentally different key prerequisites between different kinship systems in this respect: it constitutes the number of possible caregivers, and the possibilities for care arrangements. In this manner it also shapes the conditions on which central questions considering care need to be answered: such as who may care, and where? From this perspective care is first and foremost a question of (network) structurea question of the number of eligible persons, locality of those persons and resulting flexibility in care constellations.
A life course perspective broadens this observation on different kinship systems, as it shows how the constitution of structure varies during different phases of a woman's life. But it also reaffirms the systemic differences between marriage patterns, which may situate women either in the periphery (like a young Chhetri wife in patrilocal village exogamy) or the centre of the network of eligible caregivers (like a Tamang woman in a traditional cross-cousin marriage).
With ongoing social change, regional variations of care functions, changing household chores and distances between kin when households move to the city, there is a huge potential for future research on chains of care-responsibility in different regions of Nepal and world-wide.

Concluding remarks: perceived options for practical decisions
When women in a setting like post-earthquake Nepal turn down the offer of free treatment, the established literature on gender and health offers various approaches to explain their decision: from socio-cultural and awareness-related factors hampering female freedom of movement or the demand for healthcare, to financial, geographical, or institutional obstacles to accessing healthcare. Yet, as this article shows, any approach which assumes people act as mere individuals (educated/non-educated, poor/rich, powerful/powerless) is going to be highly misleading. People tend to act and think within the frameworks they learned as part of a group, including certain unavoidable obligations and entitlements. By comparing the traditional kinship patterns of Chhetris and Tamangs, I have highlighted the capacity of this group membership to form and restrict a set of perceived options that a person is likely contemplate in decision-making. Inspired by Boholm et al.'s (2013) concept of 'horizon of choice', I consequently argue that an important key to understand (care) behaviour, is to acknowledge that not all actors experience the same range of options to choose from. It is important to attempt to specify the mechanisms that constitute this perspective. One of these mechanisms was addressed in this article: It is the filtering effect of a culturally acquired lens through which actors see members of their social environmenta filter which includes certain persons as potential caregivers or work replacement while others are conceptually and categorically excluded. As demonstrated by the examples above, questions of care are not only inseparable from the wider organisation of social life, but can also have direct bearing on health decision-making.
On theoretical level, the discussion above showed that a way to counteract the analytical weakness of the term care is to work out those properties of care which are central to the research context and question. Seeing care as a function, and through the perspective of the carer and nursing case, opened the window for a structural analysis and thereby made it possible to investigate a much-neglected aspect of female access to care and healthcare.
I do not aim to depict kinship systems and normative role expectations as rigid and unshakeable determinants of human behaviour. Changing marriage practices and the growing numbers of love-marriage in Nepal show that there is always process and change in structure, and personal relations as well as character may inform the interpretation of normative standards. Nevertheless, care remains a question in which the perceived need for care and other pressing concerns like fulfilling one's duties and organising livelihood need to be negotiated among those who are considered eligible and available. In this way, structural preconditions resulting from socially learned care roles, culture-specific kinship organisation and distances to eligible kin come to play a central part.
With these insights into local realities, it became clear that neither the negligence of care nor obstacles to the female search for healthcare can be reduced to the fact of being a woman. Nor can these disadvantages be explained by power structures and levels of autonomy. Instead, it highlighted the complex interplay of contributing factors in shaping a range of perceived possibilities. Even if the female beneficiaries in the contexts discussed above had biomedical knowledge and full decision-making power, their ability to act on the demand of healthcare (Furuta & Salway, 2006) would still be bound within the constraints of organising daily livelihoods and would still be influenced by the perception of care and eligible caregivers. And while the ethnographic examples from post-disaster Nepal are very particular, the point I want to make is general: people considering major health interventions all over the worldbe it a single mother in London, a farmer in Bavaria whose children moved to the city, or a referral case in a primary healthcare centre of rural Haitineed to answer the questions of care and replacement. Health and referral programmes need to take the corresponding contexts and perspectives of their beneficiaries into account.