Background

The growing burden of chronic diseases (CDs), such as hypertension and diabetes, in low and middle-income countries, is an area of increasing concern [1–7]. This health transition is an additional strain on health systems already over-burdened with infectious disease management [5]. Efforts to understand people’s perceptions of these conditions have given rise to a large number of disease-specific studies aimed at informing health education and management approaches in Africa. This large body of work covers, for example, studies on the lay understanding of hypertension [8–13], stroke [14–21], diabetes [22–25] asthma [26–28] and epilepsy [29–39]. Some have suggested that lessons should be learned from the management of HIV in mobilising health services to manage the different CDs [40], particularly given the assertions that, due to the life-prolonging effect of anti-retroviral therapy, HIV should be considered a CD [41–43] and a condition that it is intimately linked to non-communicable diseases [44, 45]. While studies have shown that health system utilization is low for CDs other than HIV [46], it remains unclear how much of this usage of health facilities is related to differences in perceptions about these conditions or other factors, such as access to reliable and effective health services.

We conducted qualitative and quantitative research in southern Uganda and north-west Tanzania in order to collect background information for the future design of an intervention package aiming to improve services for the prevention and treatment CDs, including the formulation of educational messages to influence health seeking behaviour and service uptake. In this paper we describe the findings from that research on the knowledge and perceptions of CDs identified from rural and urban communities in the two countries. We collected data on the following conditions: hypertension, epilepsy, HIV, diabetes, cardiac failure, asthma and chronic obstructive pulmonary diseases. We did not investigate perceptions of and risk factors for other chronic conditions such as cancers, because the overall project had an a-priori focus on selected common non-communicable diseases. HIV infection was included due to its high prevalence, and its similarity with CDs with regards to service needs.

Methods

Our complementary studies comprised: a quantitative population survey involving a random sample of community members, a quantitative health facility survey in the study communities, in-depth interviews (IDIs) in a subgroup of population survey participants and focus group discussions (FGDs) with men and women from the study areas. The purpose of this combination of studies was to triangulate findings from the different approaches; but also to generate additional knowledge beyond that which could be gathered using just one of these approaches. The objectives of the population survey were to determine CD prevalence, the prevalence of risk factors associated with CDs and to document self-reported treatment-seeking behaviour. An initial set of 12 FGDs was conducted just before the population survey commenced to establish the local terminology for CDs in order to inform the population survey questionnaire design and to develop the topic guide for the IDIs. A second more extensive set of 12 FGDs was conducted after the population survey to develop deeper insights into the thinking of members of the population on CDs and their views on the most appropriate resource for treatment. The objective of the IDIs was to explore such beliefs at the individual level, taking any personal or family disease experience into account. These interviews were conducted two to four months after the population survey was completed in each community where the interviews were to take place, Finally, the objectives of the facility study were to determine service readiness and utilisation of health facilities for these CDs and thus to possibly verify shortcomings reported by the public.

The methodologies of the population survey and the health facility study have been described elsewhere [46–48]. Briefly, for the population survey, a total of 2011 adults (≥18 years) were enrolled in Uganda and Tanzania, using a stratified, multistage sampling design with five strata in each country (one municipality, two district towns, two rural areas). Consenting adults were interviewed using an adapted version of the WHO STEPS questionnaire (WHO 2008) about their disease history and treatment seeking behaviours with regards to selected non communicable diseases and HIV infection, after which they were physically examined to determine the prevalence of these conditions and of CD risk factors.

For the health facility survey, a quantitative cross-sectional study of a representative sample of public and not-for-profit health facilities in urban and rural Uganda and Tanzania, including a total of eight hospitals, and 20 middle and 24 small sized health facilities, was carried out using an adapted version of the WHO SARA instrument (WHO 2013). We conducted structured interviews with facility managers, inspected resources and administered self-completed questionnaires to health workers. Study outcomes related to service readiness and provision regarding selected chronic diseases. Health facilities were located within the same geographical strata in which the population survey and the qualitative studies were conducted.

For the initial set of focus group discussions, we invited six groups of men and six groups of women from urban and rural settings, outside the main study area, in both Uganda and Tanzania. To obtain participants for FGDs the Tanzania social science team liaised with street leaders which are the lowest administrative authority in Tanzania and comprise individuals with usually profound knowledge of the neighbourhood. The street leaders were asked to identify people with the following key attributes namely: residents of study area, people who were knowledgeable about CDs based on their personal experience with CDs, including those who had suffered or cared for people with CDs, or who knew people with CDs. Similarly, the initial set of FGD participant names in Uganda was obtained through local council leaders, who again represent the lowest administrative unit. The Ugandan social science team asked the local council leader to identify the people who were residents in their areas. All the participants for the FGDs were to be persons aged 18 years or above. For IDIs, 116 participants were selected through simple random sampling from the population survey participants, and this sample included 46 individuals with and 70 without CDs. Few of the affected participants had more than one condition. The random sampling aimed to ensure the different study areas and the CDs under investigation were represented in the selection. Five epileptic patients (two from Uganda and three from Tanzania) were purposively sampled for inclusion after the random sample failed to include any epileptic cases. In order to limit interviewer bias, the interviewers were blinded about the health status of the sample composition and of individual participants.

For the second set of focus group discussions, participants were also selected from the population survey participants, maintaining the stratification of the population survey into rural areas, district towns and the municipality. Twelve groups divided on the basis of sex were convened in each country. Participants already selected for in-depth interviews were excluded.

The IDIs and FGDs were conducted using standardised topic guides. For Uganda, these guides were translated into Luganda while for Tanzania they were translated into Kiswahili. These are the languages spoken and understood by the majority of the people in the study areas.

Contact information of eligible IDI participants was obtained from the database established by the population survey. Community leaders together with population survey interviewers acting as tracers helped social scientists to identify the physical address (streets and households) of selected participants. They were contacted, and appointments for IDIs or FGDs given to consenting participants.

To facilitate coding and analysis of data, all audio files with IDIs and FGDs were transcribed and translated from Kiswahili or Luganda to English. The translated texts were thematically coded using NVivo 9 Software (QSR International, Doncaster Victoria, Australia) to manage the coding. The coded thematic segments were compared and clustered to produce patterns and determine consistency or contradiction of responses. To complement qualitative data we used the population survey data set to extract variables on knowledge, perceptions, attitudes and treatment seeking for common CDs to facilitate coding. Fig 1, below, illustrates the way in which the different parts of the study data collection fitted together.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Flowchart of interconnected qualitative and quantitative studies.

https://doi.org/10.1371/journal.pone.0142194.g001

This paper focuses on the results from the qualitative studies. Results from the population and health facility surveys are reported in this paper when they confirmed, amended or contradicted qualitative results about public perceptions on CDs. The overall results of these surveys were published elsewhere [46–48].

Results

Local terminology used for different CDs

In both countries we found locally used terms to refer to different CDs. People used combinations of biomedical and local terminologies/concepts interchangeably to refer to a specific CD depending on the context. People living in rural and urban settings in Tanzania often used differing terminologies for the same condition, whereas in Uganda there were no marked differences. The majority of the terms were simply descriptive of symptoms or a metaphor for the condition (sometimes described by a phrase rather than a single word or term). In some cases biomedical terminologies were locally appropriated or adapted to refer to a CD. For example, hypertension was referred to as `pulesa’ (derived from ‘pulse’) in Uganda, and ‘BP’ in Tanzania. Table 1 summarises the terminologies and phrases used for different chronic conditions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Local terminology used in Tanzania and Uganda for chronic conditions.

https://doi.org/10.1371/journal.pone.0142194.t001

People often used metaphors and euphemisms to refer to certain CDs—seemingly to avoid mentioning a stigmatised condition directly. Some of the terminologies used to refer to CDs were idioms or popular local terms that were introduced at a time when there was no formal vocabulary or direct translation of a certain CD. For example, in Uganda HIV/AIDS was sometimes referred using a local term “kayovu” or “kayola”. These were the local names for a banana pest which led to the destruction of many banana plantations in the areas around Lake Victoria in the 1990s when HIV infection became widely noticed. HIV/AIDS was the condition for which more alternative terms were given than for other CDs.

Discussion

Our study illustrates how what people know and think about CDs can affect how they act. As observed in other studies, we found that the understanding of CDs drew on information and terminology gained from the biomedical sphere [8, 14]. Perceptions about causes, aetiologies and preferred treatment for common CDs were informed by multiple determinants. Traditional norms and community beliefs about the causes of CDs influenced individual choices of care and management of CDs. It is, however, apparent that although knowledge of treatment options is a necessary precursor for accessing care, other external influences including access to the services, and beliefs about treatment efficacy, influenced individual treatment seeking pathways. The understanding of most chronic conditions involved a combination of biomedical information, gleaned from clinic visits, and local informants and beliefs drawn from information shared in the community and the family [49]. However, our findings demonstrate that beyond the experience and beliefs of individuals, the societal and cultural context is important in shaping commonly shared beliefs and this understanding helps to explain the eclecticism which is practiced when treatment seeking choices are made [50–54].

The need to embrace the local understandings of disease causation in order to offer treatment options has been the subject of several other studies [9, 11, 25, 38, 53, 55–57], our findings show that this behaviour cuts across all the chronic conditions examined. While the sharing of new information about chronic conditions has yielded results in terms of the information that people now possess, that information should not obscure an appreciation for the existence of beliefs and tried and tested practices that may affect treatment seeking, and therefore influence the design and implementation of interventions. For example, health care providers need to be aware that in keeping with findings from studies elsewhere in Africa, we found that epilepsy was a stigmatised condition and, in addition, could not be treated by biomedical remedies. It was thought by some to be caused by witchcraft or evil spirits which affected treatment choices [39]. And, although HIV had been the target of many anti-stigma campaigns, our findings and the findings of other studies in East Africa show that it remains a stigmatised condition [58–60], which continues to influence treatment access and the sustained use of treatment.

There were several areas of similarity in the findings from the two countries regarding perceived causation, symptoms and preferred treatments for common CDs. In both places the terminologies used in clinical settings have been appropriated for use outside the clinic. It was therefore not surprising that biomedical terms were mimicked in local languages, referring to hypertension as ‘BP’ and ‘pulesa’ and diabetes as ‘kisukari’ and “sukaali’ respectively. Susan Reynolds-Whyte [57] documented similar naming practices in her research in Uganda and more broadly in Africa.

As already noted, in line with the context dependent eclecticism described above, awareness of CDs was not always accompanied by biomedical service utilisation. There was a wide usage of alternative care options both independently of biomedical care and in combination with biomedical care. Different approaches to care were influenced by many factors, as we have shown, including understandings of the cause of the condition as well as the availability of care at local health facilities. Indeed, other findings from the wider research project of which this study is a part show that most CD services, except for HIV infection, were concentrated at higher level health facilities, which were not always readily accessible to participants [46, 48, 61]. A recent study from Uganda which looked at health care utilisation for all conditions, carried out in one of the same districts as our study reported a high rate of the use of alternative care including the private purchase of drugs from pharmacies and drug shops and use of herbal remedies although all respondents were aware of the existence of local government health services [62]. Drug stock outs, costs of services and long distances to facilities, were mentioned as barriers to the access to health-facility based care, in keeping with our own findings.

The inequalities in access to health care because services are not available at local health centres and the services that are available are poor, has been highlighted by many recent commentators. For example, in their recent `viewpoint’ on the quality of care during the Ebola outbreak in West Africa, Boozary, Farmer and Jha [63] observe that `when people receive care that is unsafe or ineffective or they are not treated with respect, it is little surprise they avoid further care’ (p.E2). This structural violence of inequality in access to health care [64], inherent in health systems where reliable and trusted services are not available to the socially and economically marginalised [65], has a profound impact on health seeking behaviours; this rather than knowledge of disease aetiology greatly influences access to care. This contextual understanding of public health facilities and services, both in terms of access and quality of care, and the social and economic situation of those who may need care is essential in order to understand how people view care services and may utilise different sources of care [66–68]. Barriers to access may be compounded for non-communicable diseases, which can often be seen as being associated with modifiable life-style factors [69], if survival is viewed as a person’s life-style choice (to lose weight or stop drinking alcohol, for example). As Marsland and Prince [70] observe `access to choice depends to a large extent on the means people have at their disposal and the environments in which they live’ (p. 458). The underutilisation of health services for chronic conditions needs to be seen in this context. Knowledge is not enough; trusted and appropriate services need to be accessible and reliable.

A strength of our study was that we were able to compare information from qualitative work, conducted through FGDs and IDIs, with findings from quantitative investigations: epidemiological and health facility surveys in the same population and about the same time. This allowed us to triangulate and cross-check information across the different studies. Another strength is that all these studies were in two different East African countries, using standardised research protocols. Our observations were strikingly similar across the two populations. The importance of understanding how people not only view the causes of these conditions, but also hold perceptions of the efficacy of different treatment options can guide the design of interventions to improve health care for CDs in East Africa and elsewhere. A first step towards this is raising the awareness of, and sensitivity to, different understandings of causality of CDs of the health facility staff, so they make fewer assumptions about what their patients may believe about the treatment they are offered.

In summary, local beliefs and perceptions as well as importantly, the availability of services, for CDs represent a barrier for people in our study areas accessing health accessing care. Efforts to improve this situation should focus on the quality of existing health services for CDs in order to improve the quality of services as well as the trust among the population towards those services, while also taking account of community as well as patients’ beliefs about CDs.

Supporting Information

Acknowledgments

This research was funded by a grant from the UK Government Medical Research Council (MRC UK; grant id-nu 99430) provided with support from the Department for International Development (DFID) under the MRC/DFID Concordat agreement. We are grateful to the research teams for collecting the data and to all the participants for giving both their time and information to this study. We are grateful to Paula Munderi and Kathy Baisley for their input to this paper. We acknowledge the permission, support and encouragement given by the Ministry of Health in both countries.