Health-endangering everyday settings and practices in a rural segregated Roma settlement in Slovakia: A descriptive summary from an exploratory longitudinal case study

The study comprised four methodologically distinct phases (see Fig. 1). It combined ethnography (phases 2 and 4) [51, 52] with methods used in applied medical-anthropological surveying (phases 1 and 3) [53, 54]. First, a socio-graphic survey of several localities was carried out in order to select a single segregated place. Next, ethnographic methods were used in the selected place to gain close personal access to and primary data regarding the setup and possible social root-causes of the local everyday health-endangering settings and practices. Consequently, systematic interviewing was undertaken to increase local representativeness and the systematic breadth of the collected material. During the last phase, local people’s reflections of preliminary interpretations and additional material regarding long-term shifts in local health-endangering settings and practices were constructed through follow-up communication.

The fieldwork was carried out by the corresponding author. Acquired data consisted of field notes on direct participant observations and records of personal elicitations focusing on both the setup and possible social root-causes of local everyday health-endangering settings and practices. To gain data specifically and exhaustively regarding all aspects considered to be health-endangering according to contemporary biomedical theory, throughout all phases of the study an encyclopaedic practitioner’s handbook covering both clinical and public-health knowledge was being used to guide observations and elicitations [55].

The south-central region of Slovakia was picked because of its historically high proportion of segregated Roma residents [56]. The single settlement used in this study, selected based on the socio-graphic survey, had a growing population of approximately 260 people (230 in 2004, 300 in 2014) – all self-declared Roma and speaking Romani as their mother tongue – compared to a declining population of approximately 530 non-Roma living in the rest of the village (580 in 2004, 470 in 2014). In 2004, approximately half of the settlement’s inhabitants were children under 15 years old, and only 5 people were older than 60. For a concise overview of the recent history and variability of segregated Roma Settlements in Slovakia, see Scheffel [57] and Musinka et al. [56].

The sampling is detailed in Fig. 1 and Fig. 2. In the first ethnographic phase, most data obtained in the settlement was related to approximately 90 people belonging to one of the 3 then largest local fajti, i.e. specific transient kinship formations roughly overlapping with unilateral extended families [58]. The systematic interviewing visited a sample of 10 households out of the settlement’s total 48. The sample was representative according to the households’ local socioeconomic position (SEP), level of prestige, and affiliations to fajti. In the selected households, 28 informants were elicited, with 22 of them being adult women. Several other people participated in shorter sequences of the interviewing. Locally, men were considered less competent regarding health-related issues both by themselves and by women, and most of them also showed less interest in discussing health spontaneously. None of the people approached refused to participate in the interviewing. The closing follow-up observations and elicitations were limited to approximately 15 locals personally closest to the corresponding author.

The socio-graphic field survey phase consisted of personal inspection of all existing 11 rural segregated localities in the surveyed area. Local demographics, infrastructure, history, socioeconomic differences and inter-ethnic relations were assessed using brief questionnaires (see Appendix). The particular settlement was selected randomly from the majority of examined localities that exhibited non-extreme characteristics in all surveyed respects: middle-sized enclaves older than five decades, physically and socially segregated, with substandard public infrastructure and apparent internal socioeconomic gradients.

The systematic interviewing was carried out using a bilingual template of implicit topics covered by several hundreds of questions in local Romani dialect, with particular questions focusing on elicitation of the locals’ perspectives on the local setup and the social root-causes of particular local health-endangering settings and practices. The direction and sequence of questions for each particular topic were identical and analogous to the logic of elicitation proposed by Arthur Kleinman and widely used in clinically applied medical anthropology to construct ‘illness explanatory models’ [59, 60]. Particular topics were adopted from the practitioner’s guide [55]. Wording was prepared with a close local informant. Interviewing differed considerably between households in detail and depth within a length range from several hours to several days per interview. The corresponding author’s rather specific position of a friendly outsider and a supposed health expert allowed him to interview adult women intimately despite being an unrelated adult male – i.e. exceptional, according to strict local gender norms. Answers to questions were recorded in writing with a focus on capturing parts considered directly relevant to the particular questions and the specific Romani expressions used. Stratifications of households used in the sampling of households (see Fig. 1 and Fig. 2 for details) represented the consensus of several local informants regarding the particular households’ SEP (barvaľipe = affluence; associated with possession of amenities), level of prestige (level of ascribed gizda = snobbery), and affiliations to extended families (fajti).

To summarise the study findings on the local setup of health-related settings and practices, we coded and analysed selected study data as described in detail below. To ensure comprehensiveness and convenient intelligibility of the summary, especially for public health practitioners, we based the analysis on the latest World Health Organization (WHO) classification of known health-exposures, defined under the notion of ‘intermediate social determinants of health’. The classification comes from a widely used theoretical framework, the WHO Framework on action for social determinants of health [45], compatible with our theoretical premises regarding the social root-causes of health-inequalities cited in the Background.

We coded and merged all types of data from all sources and phases of the research as follows. We first coded any sequences of field notes considered relevant regarding the local setup of health-endangering settings and practices as such. Field notes from the initial ethnographic phase and from the follow-up ethnographic communication were coded manually, while transcripts of audio recordings from the late interviews were coded using the MAXQDA® software. To these sequences of text, as well as to those parts of the systematic interviews explicitly covering analogous themes, we then ascribed further hierarchical codes to distinguish data sequences relevant for particular domains of exposures and for their core elements, as defined in the guiding WHO source [45]. In parallel to all these sequences of text, we also assigned codes denoting their relevance regarding the following variables: SEP (codes for ‘rich’, ‘common’ and ‘poor’ households), level of prestige (codes for ‘snobby’, ‘normal’ and ‘squalid’ households), affiliations to dominant fajti (codes for families ‘A’, ‘B’, ‘C’ and ‘smaller fajti’), and time period (codes for ‘first three years’, ‘mid-period’ and ‘last three years’). The same sub-coding was also applied to the selected relevant parts of the systematic interviews. As the levels of SEP and levels of prestige factually equalled – e.g. the households ascribed highest socioeconomic position were also ascribed the highest level of prestige – upon coding we eventually merged these two variables into one entitled ‘social level’. This variable had three levels – high, medium and low – each indicating the levels of both SEP and prestige ascribed to particular households (see also Figs. 1 and 2).

We then performed qualitative content analysis, combining all coded data from the field notes and the relevant data from the systematic interviews, i.e. on health-endangering settings and practices. As a method for content analysis we used recurrent abstraction [61]. This means that we repeatedly read and in steps summarized all text sequences on the endangering settings and practices that the locals faced and on how they engaged with them, regardless of their original source. Upon summarizing, we focused mainly on capturing the variability and dominant trends in local health-endangering settings and practices. For each intermediary determinant, we first created descriptive summaries regarding its particular core elements for particular social strata. We then cross-compared these summaries for estimations of major differences according to social level. To assess variability with respect to distinct age, gender and family-affiliation groups as well as with respect to time periods, notes were taken during the process of re-reading and then summarized for each domain of social determinants separately.

Based on the above-described analysis, we report on the local setup of health-endangering settings and practices across the following intermediary social determinants of health, as defined in the WHO source [45]: material circumstances, psychosocial factors, health-related behaviours, social cohesion and health-system interactions. For each, we present dominant local trends regarding particular core elements. To these observations, we add notes on related variability according to social level, age, gender and research periods (no variations were found across fajti). We used bold text to point to particular core elements of the discussed intermediary determinants. To support the thus constructed findings with original data, we include illustrative quotes by local consultants. The quotes were selected based on two criteria: 1) they compactly illustrate our findings regarding particular exposures in the locals’ own wording, and/or 2) they compactly illustrate on what kind of utterances we base our observations, suggestive of the locals’ racialized ethnically framed reasoning (See also Additional file 1: Fieldwork visual reference for illustrative photographs).

The majority of houses in the settlement were built illegally and maintained in a provisory way using unsuitable materials such as industrial landfills waste. Most houses sheltered several separate households in improvised extensions of the original buildings. Typical internal housing conditions included over-heated and damp air (locals’ comfort zone was in the upper 20s°C and most houses were not ventilated regularly), cold walls (no insulation) and overcrowding (rooms sleeping up to five people were common).

Most households lacked basic household infrastructure. Only two households had indoor running water. For both heating and cooking, raw wood, illegally harvested from surrounding forests, was being burnt in second-hand iron stoves. Most households were connected to electricity, but many only through illegal extensions via other households. Several households used reclaimed car batteries instead. Only one nuclear family possessed a bathroom. Everybody used self-built outdoor dry toilets, with most children up to six years old defecating in public spaces. Basic amenities, such as refrigerators, washing machines, audio-visual entertainment equipment and cars, were popular and common in the settlement but usually limited to second-hand items which did not work and were not used as intended by the manufacturers. ‘Strong’ hi-fi equipment and cars in particular were praised and preferred as ‘Gypsy’ features.

Apart from the electrical network, community infrastructure consisted of one asphalt road connecting the settlement to the nearby village, several dirt roads and three outdoor sources of cold potable water (see Fig. 2). For liquid-waste disposal, households with running water used improvised drainages out to public spaces; others used the surroundings of their houses. Solid waste was being burnt in public places, disposed of at improvised landfills on the outskirt of the settlement or in open industrial containers provided and occasionally emptied by the municipality.

Related direct health-risks included: regular health and safety incidents within households (e.g. roof implosions, leakages, fires, window breakages); unhealthy household climates; frequent electricity outages and occasional injuries from improper handling of equipment; lack of personal-hygiene means, the presence of parasites (e.g. lice, fleas) and frequent intestinal infections; contamination of public space by urine, faeces and smoke, the presence of rodents; constant ergonomic strain. Related social and economic tolls included: stigmatization outside the settlement (due to e.g. smell, parasites, dirty clothes, outworn equipment) and frequent relatively high-cost breakages (due to e.g. high-input and short working life of the outworn equipment).

The highest-ranked households occupied old, legally built former peasant houses, possessed amenities in better condition (e.g. registered cars, chainsaws, DVD-players), and paid more attention to tidiness outside their households (e.g. possession of a waste bin, children using dry toilets).

Over the 10-year research span, some material aspects improved, especially for those holding higher social positions. Several highest-ranked households insulated their houses; two bought new washing machines; most families originally in the possession of a car (some medium and all high social level households) kept renewing their second-hand fleets; several high and medium social level families purchased reclaimed computers, and even the poorest families were able to start using newer models of second-hand mobile phones. The lowest-ranked households experienced little or no improvements, with some of them experiencing further decline (e.g. moving from a deteriorated self-built wooden shelter to a smaller reclaimed wagon).

For illustrative quotes related to the above section of Results, see Table 1.

In terms of longer periods of stress, the locals appeared to suffer the most from their nuclear family members’ detachment due to hospitalization, work trips or incarceration, and an effort was made to prevent such scenarios. Another long-term stressor was the terminal stages of terminal diseases in related elderly. The locals perceived the following in particular as the most frequent incidental stressors: a total lack of funds (i.e. fluctuating periods of literally no cash and no subsistence opportunities for the coming several weeks), visits outside the settlement beyond the nearby village (approximately weekly trips to the local administrative centre) and incidents of physical violence (approximately monthly involvement in fights within the settlement, mostly due to jealousy or conflicts of interests amongst fajti, culminating during celebrations after welfare payments).

Perception of own low SEPs (both within the settlement and in comparison, to local non-Roma) did not appear to cause the locals any stress. The low SEPs nevertheless remained causally linked to the frequency of a total lack of funds, which was considered an incidental stressor. Common experiences of racism and discrimination (e.g. preferential treatment of non-Roma in GP waiting rooms or withdrawals of employment opportunities upon arrival in person) were considered unjust but ‘normal’ and not stressful unless overtly offensive.

We identified the following local strategies to prevent and cope with stress. These were solidarity within fajti in cases of food shortages, serious health issues and violent incidents; solidarity beyond fajti where children were at serious risk (e.g. rides to the hospital) or in conflicts with non-Roma (e.g. side-taking in fights); travelling only in groups pretending ostentatious confidence (e.g. loudness, strict speech-tones); avoidance of persons and institutional venues with a racist track-record; and drawing self-assurance from adherence to social norms framed in ethnic terms (‘Gypsy’ / ‘Romani’) constructed and praised mostly in contrast or in direct opposition to alleged non-Roma norms. The above strategies were not being interpreted by the locals themselves as such.

The local ethnically framed social counter-norms included e.g. specific work ethic (stressing the importance of the Roma ability to improvise in contrast to the non-Roma ability to bear drudgery), a specific kinship ethic (stressing e.g. different concepts of discretion and openness between generations), aesthetic style (e.g. relatively expressive ‘Gypsy’ clothing patterns and festive verbal communication), etc. The appropriateness of the locals’ adherence to the particular norms was often argued by them using racialized arguments quoting specific related ‘natural’ Roma capacities embodied via their ‘blood’, ‘brains’, ‘bodies’, ‘genes’, etc. The same or analogous reasoning and argumentation was commonly used in local discussions (i.e. also beyond the discussions with the researcher) of most local practices (for examples from other domains see other sections of Results).

Across local social strata, the only variability in any of the above concerned that higher-ranked people experienced more stress regarding financial difficulties projected over longer periods (e.g. related to pension plans). Moreover, some practices necessarily associated with the maintenance of a higher SEP, such as compliance with rules in education, long-term employment, etc., were generally considered and ridiculed locally as being ‘too non-Roma-like’ (gadjikano). This posed an extra psychosocial dilemma for locals of high social level. In turn, lower-ranked people were generally considered and respected as more ‘true Roma’ (prave Roma). Over the last five years of the study period, a general lack of means of subsistence seemed to become more severe and continuous (constantly decreasing seasonal, unqualified and unregistered employment-opportunities; decreasing welfare payments; increasing formal requirements for employment; increased indebtedness by commercial lenders). This shift was also quoted as the sole reason for alleged local dramatic increase in the use of prescription tranquilizers and antidepressants in the settlement.

For illustrative quotes related to the above section of Results, see Table 2.

Similar age and gender patterns were present with respect to alcohol consumption. Above the lowest social strata, where binge drinking was somewhat more frequent (several times per month), daily drinking was only moderate for most of the month (a pint of beer now and then) with the exception of two alcohol-dependent persons. In most households, binge-drinking took place solely following monthly welfare-payments and at anniversary celebrations. Even here, however, drinking until loss of basic social skills (more common amongst non-Roma men in the nearby village) was being discouraged as inappropriate. Such celebrations included loud reproduced ‘Gypsy music’ and intense dancing. During the research period there were no other cases of drug abuse observed.

As regards consanguinity, people typically tried to form couples across geographical distance, yet preferred partners from already related fajti. Several first-cousin marriages were thus tolerated, silently or stressing the families’ distant residency. High and medium social level adult women often opted for intrauterine devices after conceiving several children. Condoms were not in use in the settlement.

With the exception of winter months, when watching TV for several hours per day was common, most children and adults within the settlement engaged in strenuous physical activity throughout most of their days. Children were playing outdoors, men were busy with taking care of wood, repairs of all kinds (houses, cars, other amenities), and women with repetitive cleaning of households, washing of clothes, cooking and carrying around of children. In addition, most people frequently organized gathering and hunting hikes in small collectives for both additional subsistence and entertainment (collecting scrapped metal, picking mushrooms, harvesting corn or potatoes, catching crawfish). Given the lack of standard amenities for most activities in most households, all of the described physical activity was relatively un-ergonomic and often dangerous. With increasing social status, people could afford and readily invested more into amenities that made the respective activities more comfortable (e.g. purchasing second-hand chainsaws).

For illustrative quotes related to the above section of Results, see Table 3.

In most respects, bonding social capital within the settlement was restricted to fajti comprising of several dozens of people. Despite a simultaneous public performance of distance according to the ascribed levels of prestige (hierarchies running within particular fajti), within these local kinship networks solidarity was absolute: from sharing of or donating supplies to nuclear families in need, through constant reciprocity of small favours, to side-taking in violent conflicts. Across local fajti boundaries, relationships were mostly competitive and rogue (constant mutual monitoring, blackwashing, public provocations, ostentatious ignorance, feuds, etc.) with the exception of life-threatening incidents involving children or elderly or in conflicts with non-Roma outside the settlement. Both of these bonding-capital networks provided an important welfare safety-net with respect to frequent health-related crises (e.g. preventing hunger, and assistance with severe injuries).

Framed in kinship (within fajti) and ethnic terms (towards the non-Roma), the above described social norms on one hand encouraged and enabled the building of strong local-bonding networks and on the other contributed to the isolation of particular fajti from each other and of the whole community from the non-Roma. The second restriction made it difficult for the locals to accumulate any bridging and linking social capital through engaging in reciprocity with outsiders. The lowest-ranked families possessed almost no relationships with non-Roma (except for friendships with socially excluded non-Roma), while even the highest-ranked individuals managed to maintain only several informal personal ties with local non-Roma from the village and only exceptionally in various local offices (e.g. long-term friendships or regular barters actively hidden from public by the non-Roma). The lack of bridging and linking capital meant the only resources the communities could rely on with respect to health were resources owned by the particular local fajti.

For illustrative quotes related to the above section of Results, see Table 4.

Within the settlement, access to healthcare services differed along the social gradient. As elsewhere in Slovakia, services availability (the technical potential to deal with local health needs), accessibility (relative geographical distance) and accommodation (ability to meet the technical constraints of potential clients) met contemporary EU standards (i.e. at least since a major health reform introduced in 2006). Several general practitioners were available within a radius of some kilometres. There was a local hospital 15 km away with an emergency entry-point operating non-stop, and emergency-rescue teams were typically able to reach the settlement within 15 minutes of being called, as required by law. In households of lower social status, however, people experienced greater affordability difficulties. Most of the time, their members were incapable of paying for transportation and service-related complementary fees (e.g. purchase of prescription medications not fully covered by the national insurance plan). Only the highest-ranked people were able to visit the appropriate providers whenever they felt they should, though with periods when even they could not. Others would recourse to improvised home-healing or sometimes to inappropriate use of emergency services (e.g. emergency calls in cases of uncomfortable long-term stomach pain).

Actual use of healthcare services within the described affordability constraints was as follows. Regarding light transient diseases (bežne nasvaľipna), all local people used available services and took medication whenever possible exactly as recommended. Regarding serious chronic and terminal diseases (phare nasvaľipna), consistent following of clinical recommendations over longer periods was generally an exception. Analogous general non-compliance occurred also regarding preventative and recovery recommendations. Such non-compliance regarding serious chronic and terminal diseases contrasted with several local strong inclinations. Everybody in the settlement feared pain and death, often spontaneously describing such emotions as ‘naturally’ more intense in comparison with the non-Roma. All locals claimed interest in and made an effort to learn their diagnoses and related medical recommendations, especially regarding chronic and terminal diseases specifically. And all locals trusted local medical practitioners, especially in terms of the functionality of their medical know-how (often quoting also a supposed non-Roma ‘naturally’ superior capacity to ‘deal with complicated matters’). This contrast was considered understandable based on its congruency with local understanding of ‘proper (Roma) life’, framed in ethnic and often racialized terms. E.g. the locals would spontaneously quote a ‘natural Roma incapacity’ for long-term attentiveness to one’s health. Despite evocations of ‘nature’ and biology (‘Gypsy blood’, ‘Gypsy brains’, ‘Gypsy genes’) in such claims, however, in practice such ethnically framed norms applied mostly to the adult population and did not apply to children (at all) and elderly (as strictly). In the cases of the latter, healthcare was being utilized without any normative restrictions.

People of higher social status tended to opt for use of healthcare services for lighter symptoms – some of which the lower-ranked people did not regard as health issues – and to engage in short-term attempts to follow medical recommendations also regarding severe diseases. Among people of lower social status, knowledge and interest in knowledge of medical theory and recommendations was much rarer. Instead, people were using either explanations and therapeutic procedures improvised de novo within nuclear families (e.g. treatment of syphilis with petrol) or therapeutic procedures analogous to those used among local non-Roma (e.g. folk herbal-medicine).

For illustrative quotes related to the above section of Results, see Table 5.

The main strength of our study was its mixed-methods approach. The preliminary socio-graphic survey enabled identification of a relatively typical locality. The use of ethnographic methods enabled intimate access to local everyday settings and local people. The systematic interviewing across several local stratifications allowed accounting for local variability as well as for topical omissions in the previous less-systematic phase. The follow-up communication enabled detection of major changes in the observed phenomena over time and additional reflections of preliminary interpretations by local core informants. Choosing the WHO classification of social determinants of health as a template for analysis and reporting enabled direct public-health significance of our summary according to contemporary health-inequality theories.

Our use of the research-design and our reporting also had some limitations. First, the field-work and most of the analyses were performed by a single researcher, limiting the potential for inter-personal corroboration. Second, although preceded by a careful selection of a place typical for the geographical area, the research examined only a single settlement. Third, it was impossible to remain personally embedded in the settlement for the whole course of the research. Lastly, the explicit focus of our summary on the description of local settings and practices limited the space for presentation of authentic reasoning of the people being described. While all these limitations appear to be well out-weighed by the good match of our findings with other literature, any generalizations of the clues our paper presents should only be attempted with due caution.