Introduction

Tuberculosis (TB) remains a major public health problem and is the leading cause of morbidity and mortality worldwide, despite progress that includes reversing upward trends in TB incidence [1, 2]. In 2013, the African region, home to 12% of the global population, accounted for 29% of cases, coming second only to South Asia with 56%. In high HIV prevalence settings such as the southern African region, the vulnerability of people living with HIV (PLWH) to rapid progression to TB following infection makes early diagnosis and treatment especially important. According to the 2014 global TB report, 13% of people who developed TB in 2014 were infected with HIV [1]. HIV has dramatically reversed the downward trend of TB in sub-Saharan Africa in the recent past [3–5]. The African region accounts for 80% of HIV-positive TB cases and TB fatalities among HIV infected people [1]. In Malawi specifically, case notification has increased fivefold between 1984 and 2006 [6], trailing the accelerated growth in HIV burden on the continent.

There is currently a steady but slow decline in TB incidence, prevalence and mortality within the African region[1]. Among factors that hamper TB control efforts are weak health systems, whose effects include poor infection control and prevention practices, inadequate supply chain management leading to stock-out of diagnostic and treatment essentials, and low quality of care as decided by patients [7–9]. In addition, End TB, the global TB control strategy currently in use in many countries including Malawi, relies primarily on individuals presenting at a health care facility rather than undertaking active case finding in the community where transmission occurs.

In turn, at the level of the community, delayed treatment-seeking for symptoms suggestive of TB arises from inappropriate health care-seeking habits, which are shaped by socio-economic factors such as poverty, the financial burden of diagnosis and treatment on patients, and associations made between TB and HIV [7–9]. Moreover, there is a gendered pattern in TB epidemiology; men in Africa generally have a higher burden of both undiagnosed and diagnosed TB [10–14]. This may reflect the interplay between higher true incidence of disease in men, and their lesser engagement with primary health care (PHC) and longer delay in seeking care for most illnesses including for TB [15–19]. In Africa, poverty also seems a more critical barrier to treatment seeking than gender [7, 20–28], compared to many parts of Asia where, to a large extent, gender disparities explain women’s longer diagnostic delay including at facilities [29–31].

Stemming the spread of TB requires action at different levels, including facilitating early case detection through interventions that take into account determinants of health services utilisation at community and health care provision levels. It is critical to understand health care systems problems, as well as how TB disease symptoms are recognised and acted on [32–34]. This paper draws on perspectives generated from ordinary community members, TB patients, undiagnosed chronic coughers, and health care workers to understand the interaction of PHC facilities with patients and the community. The perspectives emerged within a larger qualitative study that sought to understand why men remain undiagnosed with TB in the community generally, but more specifically when having symptoms suggestive of TB [33]. During the study, it emerged health systems challenges were a major concern that likely affected men and women’s ability and willingness to engage with formal health care services. This paper highlights the health systems challenges described, and the implications for health care seeking by men and women, and conceptually links these by applying the notion of human agency.

Methods

Table 1 summarises the data gathering techniques and the study participants and their characteristics [33]. Data were collected over 12 months from March 2011, in three high-density and low-income suburbs of urban Blantyre in Malawi. Given limited understanding around the study topic, purposive sampling [36] was employed. The aim was to include participants best placed to provide information, and to vary sources of data to allow exploring different dimensions and generating a comprehensive picture of the study issue, as well as enhancing truthfulness and completeness [37] through participant and data techniques triangulation [38, 39].

Download:

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

Table 1. Data collection techniques, study participants and sample size.

https://doi.org/10.1371/journal.pone.0154103.t001

In-depth interviews (IDIs) (n = 20; female = 14) were conducted with patients who had been diagnosed with TB in the month preceding their being contacted, to understand patients’ personal experiences of symptoms and care seeking. Patients were identified through TB registers and TB cards from two primary health centres, and traced using an approach similar to the one routinely used to follow-up TB patients; specifically, through their physical addresses or cell-phone numbers. Additional IDIs (n = 20; female = 8) were carried out with un-investigated chronic coughers to explore meanings given to symptoms suggestive of TB, and the associated care-seeking practices. Participants were identified through cluster enumeration from a TB/HIV epidemiology cluster randomised trial (CRT) also conducted in the study catchment area. The CRT investigated, among other things, the presence of cough with duration of three weeks or more within enumerated households [40–42].

Focus group discussions (FGD) were held with 74 members of the community (n = 8; Male only = 3; Female only = 3, Mixed sex = 2) to investigate general beliefs regarding health care seeking, TB and chronic cough, and normative views of gender roles. Potential participants were approached in their respective homes, informed about the study and invited to take part in the group discussions. Two FGDs with 20 health care providers were conducted to gain an understanding of the operations of the health delivery system and its interaction with patients and the community. Participants were identified at health service centres and comprised different cadres including nurse midwife technicians, nurses, health surveillance assistants, TB officers and clinical officers.

In short, the research process, which is described in greater detail in a related prior paper [43] was aligned to grounded theory as propounded by Charmaz [44]. Data were collected by two graduate level social scientists led by MK both of whom are fluent in Chichewa, with oversight from JC. None of the participants who were contacted and were eligible directly declined to participate; a sizeable number were excluded for not meeting inclusion criteria, while five TB patients could not be traced based on credentials that they had supplied to facilities. Fifteen ‘chronic coughers’ were excluded because they did not have a chronic cough at the time of our visit. All 96 potential participants invited to community FGDs accepted; however, 22 did not show up, as did two for the health care providers’ FGDs.

Data were recorded, and transcribed and translated verbatim by trained personnel, and then checked for accuracy by MK and intelligibility by JC. The transcripts were analysed using NVivo 9 data analysis software, with text being coded, retrieved and categorised and reconnected iteratively to generate and consolidate thematic categories. The coding scheme was developed both from the study aims and from themes that were emerging during the emergent analysis [37]. Coding was iteratively repeated until themes were saturated and refuting data had been taken into account. Health workers’ perspectives were examined to provide a validating perspective to the community and patient accounts. Parallel coding and analysis were done by MK and JC. Agency was then applied as an analytic lens to further make sense of the emerging themes.

The study was approved by College of Medicine Research Ethics Committee in Malawi and Human Sciences Research Council Research Ethics Committee in South Africa. Informed consent was obtained from all study participants, and the research team members observed and maintained anonymity and confidentiality of the study participants during the entire research process.

Findings

Themes emerging from the analysis pointed to barriers to treatment that fell into two distinct yet interconnected domains, namely community as well as PHC system levels. Participants described myriad health systems challenges, among them equipment and drug shortages, and poor patient-health provider interaction, that led to general distrust of and lack of faith in the health system. Another prominent theme revolved around perceived unprofessional conduct by health care providers, including being generally disrespectful of patients and failing to uphold confidentiality of sensitive information. Views of community members and patients were confirmed by health care workers. The themes and perspectives are presented accompanied with verbatim quotes carrying limited participant identifying information.

Discussion

Malawi has made commendable progress in the health sector by including PHC facilities within the hierarchy of health delivery in order to ensure free and accessible health care to the majority poor. Our findings point nevertheless to salient health system problems, as is the case in most resource-constrained settings [45, 46]. On the one hand is a frustrated health care workforce overburdened by workload, constrained by unavailability of essential resources, and disillusioned with their remuneration patterns, in ways well reported in the literature as interfering with optimum discharge of duties [45, 47]. Unsurprisingly, their frustrations were often vented on patients. While PHC staff admitted to unprofessional conduct and knew that it was improper, they did not consider some of their actions as unreasonable given the circumstances.

Then on the other hand is the materially deprived patient, and whose expectations about services offered at PHC centres are still barely met. The patient’s reaction includes taking conscious, though not entirely voluntary, decisions to procrastinate or completely defer seeking health care, or to purchase drugs from unlicensed vendors [32]. In instances where patients do engage with formal health care, they lack the confidence that they are receiving dedicated or even correct service. From a human agency theoretical perspective, therefore, community narratives of poor patient interaction with PHC services are an important indicator of how patients order and schematize previous and current experiences to generate an impression of expected future experiences.

In patients’ eyes, problems encountered are a common feature of PHC services. Communities and patients take cues, even on the basis of isolated negative incidents, to re-affirm their general perceptions of health services, with enduring effects on subsequent treatment seeking behaviour [48, 49]. The capacity to remember, select and apply schemas of action developed through past interaction with PHC facilities, and reference to known routines, patterns and traditions [35] enables patients and communities to generate their own scripts about the nature of service that may be anticipated. The patient's memory reproduces, sustains and amplifies the anticipation that past patterns will recur in future encounters with PHC services. This means patients with specific medical problems, for example TB-suggestive symptoms, may arrive at a course of action that may be neither desirable nor externally rational, primarily because they may apply selective focus to a limited set of cues—usually negative ones—to inform decisions.

The influence of past experience with health care facilities on patient responses to a particular health problem is reported in studies from resource-constrained settings globally [50, 51]. What we wish to highlight here is that this response must be seen as intrinsically relational. It becomes centred on engagement and disengagement of patient with the PHC facility, and simultaneous deliberation with other community members to identify pragmatic possibilities and gain support for decisions that might challenge normative patterns of action [35]. In the process, patients reflect and consider new possibilities and courses of action through imaginative engagement about likely outcomes of such actions. For example, patients were seen to distance themselves from seeking formal health care when they perceived that such action would negatively impact their social identity, for example through possible breach of their privacy and confidentiality, which would expose them to damaging stigma associated with TB or HIV. This pattern has also been reported elsewhere [34, 52], and again is associated with delay in seeking treatment.

In conclusion, PHC systems, which act as a fundamental entry point to formal health care, have salient problems that seem to undermine the quality of health care services provided. This paper has framed treatment seeking, including for symptoms suggestive of TB, within the interplay of the health delivery system challenges and the agency of patients coping with emergent illnesses vis-à-vis the health-system challenges. Negative experiences in PHC facilities sustain “ripple effect” on the subsequent treatment-seeking patterns of patients and community members at large. Individuals who have symptoms of TB can thus be pushed into remaining u-investigated in the community for prolonged periods before fully completing a diagnosis working through formal health care channels. Dealing with TB in TB- and HIV-endemic but resource poor settings requires understanding extant challenges in health care systems, and how these shape treatment seeking behaviour for individual patients. An ‘integrative health care’ model at PHC level is indicated, that employs an interdisciplinary team approach that pursues mutual respect and a vision of health care that permits practitioners and patients to contribute their particular knowledge within a collaborative plan of care [53]. It is worth noting, however, that these findings pertain to a phase of immense economic difficulty in Malawi, which may raise questions about the applicability of the findings in different economic scenarios. Having said this, Malawi is a known low-income country, and besides, for a qualitative study of this nature whereby findings pertain to a specific case context, it is deemed possible to transfer the findings to other settings that are inherently similar.

Acknowledgments

The authors thank the participants; the leadership and authorities of Blantyre for permission to carry out the study; the District Medical Officer and health managers for permission to access health facilities and records; Daniel Mwale, the research assistant; Professor Rob Heyderman, former Director of MLW; the data processing and entire support staff at Malawi Liverpool Wellcome Trust. This study was mainly supported via a Wellcome Trust Training Fellowship award to Jeremiah Chikovore, with additional support coming from a Wellcome Trust Senior Research Fellowship award to Professor Liz Corbett, and a Helse-Noord Initiative grant to Moses Kumwenda.