Study site

This study was situated within the Chikhwawa district, in the southern region of Malawi. It was nested within a community-based trial investigating the programmatic implementation of ACTs in children below 5 years of age (Clinical Trial ID No: NCT010380632). Malaria is the leading cause of morbidity and mortality in children under five and pregnant women in Malawi, with an estimated six million cases annually [7]. The 2005 integrated household survey found that 52% of the population live beneath the upper poverty line, classed as poor and 22% live beneath the lower poverty line, considered ultra-poor [18]. The southern region holds the largest proportion (60%) of individuals living beneath the poverty line. The Malawi National Malaria Indicator Survey 2010 found that 39% of children under five years had had fever in the previous two weeks, with comparable results from the Chikhwawa district (37%) [7]. The majority of households in the Chikhwawa district depend on subsistence farming. This study was conducted in the Chikwawa district because of the high incidence of fever in children and the opportunities arising from the parent trial.

The Malawi health system is divided into three levels: a limited number of referral hospitals in major urban centres (tertiary level); district hospitals (secondary level); health centres and clinics (primary level). Within the Chikhwawa district, basic care is also provided at the community level by Community health workers (CHWs), known locally as Health Surveillance Assistants, who deliver treatment from health posts and conduct health education campaigns. First-line treatment of malaria switched from sulfadoxine-pyrimethamine to ACTs (artemether-lumefantrine; locally referred to as LA) in 2007. At the time of this study, ACTs were not provided by community health workers but could only be accessed in health centres.

Study design

This qualitative study was conducted between September 2010 and February 2011 and included a series of Focus Group Discussions (FGDs) and Critical Incident Interviews (CIIs) (Table 1). Four local fieldworkers conducted the qualitative data collection in the local language (predominantly Chichewa, although Chisena was used in some villages). 12 FGDs were conducted with male and female community members in two rounds in two different subsets of the population. Separate FGDs were held with groups of younger and older women and men. Two male fieldworkers conducted FGDs with male participants and two female fieldworkers conducted FGDs with female participants. Critical Incident Interviews (CIIs) were conducted with primary caregivers of children who had recently experienced a febrile episode. 22 eligible individuals were initially identified and CIIs were conducted with each of them. At this point no new themes or ideas appeared to be arising and so no further CIIs were conducted. All the CIIs were conducted with female participants by one of the female fieldworkers.

Download:

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

Table 1. Qualitative methods.

https://doi.org/10.1371/journal.pone.0125439.t001

Training and Quality assurance

Fieldworkers undertook a series of training sessions in qualitative research methods, with a specific focus on conducting CIIs and FGDs. Training sessions also explored issues around treatment-seeking for childhood fever. Voice recordings of each IDI were reviewed on return from the field; translated transcripts of the pilot FGDs were discussed in detail, highlighting specific areas of good practice and those requiring further development.

Quality checking of transcribed data was conducted by a second transcriber who reviewed 10% of the transcripts against the audio file. The quality assurance of translation was conducted through back-translation by a second translator. Initially small sections were back-translated with further back-translation if discrepancies were found. The translators and transcribers met with the PI regularly to review transcripts and to discuss any issues arising.

Data analysis

The team met after each qualitative data collection to review transcripts and discuss emerging themes. Data analysis was based on a framework approach [25]. The CIIs and FGDs were recorded, transcribed and translated into English then imported into the qualitative analysis software QSR NVivo 9 for management and analysis. All transcripts were read through and summaries written. A broad index was developed during coding of the first two FGDs and CIIs (by VE in discussion with AK and RT), based on themes emerging from the data. The index had the following main categories: Accessibility; Illness Typology; Diagnosis; Disease Outcome; Finance; Interactions; Treatment type; Narrative; People; Perception; Recognition; Roles and responsibilities; Service Quality; Signs and Symptoms; Steps in treatment-seeking pathway. The index was then applied to code all data and iteratively refined throughout coding (VE and AK under guidance from RT). Data were charted by summarising and comparing data coded under different indexes and across different categories of participants, and these charts were used to develop descriptive analyses and explanations (VE, RT and ER).

Ethics statement

Ethical approval was obtained from the College of Medicine Research and Ethics Committee, Malawi (P.04/09/783 and P.10/08/707) and the Liverpool School of Tropical Medicine (09.07). Additionally permission was granted from village leaders. Sensitisation campaigns were conducted as part of the parent study to inform the communities about all aspects of the trial and affiliated studies. Fully informed written consent was obtained from all participants in the local language.

Interpretations of febrile illness

Interpretation of febrile illness were explored in detail during FGDs, however the findings were supported by discussions in CIIs. In the local language, the phrase ‘kutentha thupi’, or ‘hot body’ was used to describe fever. A number of illnesses were said to cause this. The most frequently mentioned were malungo, mauka (also masungu), nyankhwa and (m)tsempho. The symptoms of these illnesses were described as similar, despite considerable difference in perceived causes.

Responses to childhood fever

The analysis of the qualitative data led to the development of an overview of the treatment-seeking process in three phases: Phase 1—The assessment phase; Phase 2—Seeking care outside the home; Phase 3—Evaluation of treatment response (Fig 1).

Download:

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

Fig 1. Diagrammatic representation of the treatment-seeking process.

https://doi.org/10.1371/journal.pone.0125439.g001

Phase 1: Assessment period

The usual first step after recognition of illness was to assess its severity. Participants frequently differentiated between mild forms of illness, which could be treated using home remedies, and more serious forms requiring hospital assistance:

Yes, the extent of the sickness. If he is very sick, we rush to the hospital because a child can be sick and be able to play. Yet another child would be too sick to eat or play; he just sleeps. It won’t help to buy drugs from a shop; we’d better take the child to the hospital.

(Younger women’s FGD, NTH, 22/01/11)

In most cases illnesses were described as initially mild; caregivers observed the child to assess whether the illness became more severe or resolved. Lack of perceived severity was the most common reason why CII participants whose child had fever and had tested positive for malaria by the MIS team had not attended a health facility; in most cases the fact that the child was still playing was taken as a sign that illness did not require hospital intervention.

Immediate recognition of an illness as serious was rare; however caregivers stressed that in situations where an illness was recognised as serious they would go straight to the health facility. The process of assessing illness severity usually involved a ‘wait and see’ approach, most often combined with some form of action, such as using purchased drugs or bathing the child in an attempt to cool them. When symptoms fail to resolve or worsen, hospital care was sought:

We try to see, as I said before, whether or not it is real malungo. When he starts in the evening, we wait till morning to see how he will spend the night and in the morning when we see that the fever is still there, we will try some small drugs like children’s cafenol or panado so that if it is ordinary, simple fever, then the drugs will work and the fever will cease. When we try the children’s cafenol and the fever does not stop it or it increases, then we think of taking the child to the hospital.

(Men’s FGD, FFH, 07/10/10)

As described in the above quote, milder, resolving fevers were not necessarily categorised as a particular ‘disease’, but were often referred to as ‘simple fever’. In most instances, lack of recovery after the initial period of assessment was described alongside symptoms indicative of malungo, and this combination was considered evidence that malaria was the cause, prompting attendance at the hospital. Caregivers explained that it usually took between two and five days, sometimes longer, to recognise their child has malungo. Several CII participants who had not attended a health facility said they had planned to do so, but had not yet done so when the MIS team arrived. This period of waiting meant that the illness could be quite serious by the time they attended hospital. Other reasons for not attending a health facility until the illness was serious included the difficulties with recognising illness until it had progressed, and rapid development of illness such that it became serious before the caregiver had a chance to respond.

External constraints also influenced decisions by caregivers. Shop-bought drugs were sometimes used while caregivers established illness severity and the need for other care. However at times shop-bought drugs were used because of difficulties in accessing health care, rather than strategically to assess illness severity. A number of references were made in both FGDs and CIIs to using shop-bought drugs at night-time to control the fever until the child could be taken to a health facility. Caregivers did not always appear satisfied with the use of shop-bought drugs; comments about the use of expired shop-bought drugs were made in male FGDs in NTH and FFH villages:

Respondent 1: Especially at the grocer’s in our villages, we are just given drugs which perhaps expired long ago, and sometimes these drugs are kept in hot places, not good for drugs.

Interviewer: So we just buy?

All: Yes!!

Respondent 2: We just buy them. By the time we go to the hospital, it means we have failed. Let’s be honest, before we reach the hospital, we try all ways available and when we fail, it’s when we go to the hospital.

(Men’s FGD, FFH, 07/10/10)

Phase 2: Seeking care outside the home

If symptoms did not resolve after the assessment period, individuals moved to Phase 2, meaning that individuals had to decide which source of care to attend outside the home. All of the symptoms associated with malungo were said to prompt caregivers to seek formal healthcare (public or private health facility, or CHW).

A sing’anga (traditional healer—asing’anga plural) was perceived to be the only place to receive treatment for illness caused by witchcraft, tsempho, nyankhwa, and usually mauka. Asing’anga were thought to be the only ones capable of identifying witchcraft as the cause. Despite this, most FGD participants stated that if the illness did not resolve after the assessment period, then care would be sought from a formal health facility initially. Reasons for doing this rather than visiting a traditional healer included the potential for malaria to progress rapidly leading to death; difficulty in recognising traditional causes of illness; and the importance of biomedical diagnosis:

But first, it is important to rush to the hospital before going to the traditional healer because the traditional healer may tell lies since he doesn’t have any machine for diagnosis, he will just give you lots of medicine and that may create many problems. That’s why it is really important to first of all rush to the hospital for diagnosis.

(Men's FGD, NTH, 06/10/10)

However, participants described situations where it was possible to recognise that an illness required assistance from the sing’anga in the first instance, rather than from a health facility. The main example of this was when the child suffers fever and the mother ‘an itch’ at the same time (mauka and nyankhwa).

Phase 3: Evaluation of treatment response

Caregivers relied on lack of resolution, or increased severity of symptoms as indicators of treatment failure and described reduced fever, increased strength, eating and playing as signs of recovery. The treatment-seeking process was not linear and caregivers described how after attending a source of care outside the home, they would then assess whether treatment had worked.

Some members in all FGDs indicated they usually returned to the health facility before trying other sources. These participants emphasised that they might make several trips to the hospital before searching out other methods. This was supported by the findings of CIIs: three participants had made return visits to the hospital. A minority of FGD participants mentioned switching from public to private health facilities if treatment was unsuccessful. Reasons for returning to the hospital rather than seeking care from elsewhere included: getting another opinion; hospital treatment being free; showing others that you did all you could; getting another drug; and convincing doctors of the severity of the illness. A minority of FGD participants suggested that they would switch to a traditional healer rather than make a return visit. Reasons for not returning to the health facility included: receiving a negative malaria test at the first visit, which some considered to indicate a traditional illness; fear of receiving the same drug (LA) again; and no improvement upon completion of medication, which was felt to indicate a traditional cause. Occasionally, participants described visiting traditional healers in-between health facility visits. Repeat visits were also described for shops and traditional healers.

Attendance at a second source of care was not necessarily a sign of treatment failure. CII and FGD participants explained that they were frequently advised to obtain shop-bought paracetamol to take alongside LA. In some cases individuals outlined their belief that LA cannot work properly unless combined with paracetamol.

Individuals combined observed symptoms with treatment responses in establishing their final perceived causes of illness, through a process of reinterpretation, or trial and error. Illness cause was recognised retrospectively by successful treatment. When hospital treatment was unsuccessful, individuals began to doubt malaria as the cause. Recognition of mauka was an example of this reinterpretation process frequently described in FGDs and also in one CII. If the mother had symptoms, the next stage was to treat her, either with biomedical or traditional treatment. Treatment failure was described as being particularly important in recognising witchcraft and tsempho, as these illnesses were not considered to be readily identifiable during the initial stages of illness. Treatment failure leads to the suspicion that somebody is to blame for the child's illness, either through their unfaithfulness (a parent) or through engaging in witchcraft (usually a relative). However successful treatment by the traditional healer was the ultimate proof of cause. Witchcraft and tsempho were not described in CIIs, perhaps because these were considered later in the course of the illness and the CIIs only captured recent illness.

The response to non-resolving fever sometimes involved re-assessing the cause of illness and switching providers. However caregivers did not necessarily switch providers just because their perception of the cause had changed. The order of treatment options was influenced by attempts to minimise costs, as well as which source of care was perceived to be most suitable. At times a source of care was used because they had given up on the previous one, rather than because they had reinterpreted the illness. The following quote demonstrates how perceived lack of treatment options at health facilities results in individuals’ pursuing alternative options:

We go again and when we are given LA again and there is no change then we make plans to see other people to help us. At the hospital we are told that the only available drug is LA, so, as beggars, we do not have anything to say.

(Men’s FGD, NTH, 17/01/11)

Some Individuals combined traditional and biomedical sources of care and did not categorise all illnesses as only having a traditional or biomedical cause, illustrated by the following quote:

When the mother has mauka removed, what remains is just ordinary malaria that is treated using LA and if the child doesn’t recover, all the mother does, is to go to the hospital for change of treatment or to get another dosage.

(Younger women’s FGD, NTH, 22/01/11)

Other examples included the belief that children may require hospital treatment for anaemia after receiving traditional treatment for tsempho: and the suggestion that biomedical researchers should do more to develop cures for illnesses with traditional names (male FGD).

There was disagreement among the participants about the appropriate treatment of mauka; some argued it could only be treated by a traditional healer, while others stated that it could be treated at the hospital. The following quote demonstrates the blurring of boundaries between biomedical and traditional theories, and how healthcare workers’ interpretation and management of childhood fever may further complicate matters:

…at the completion of the dosage, we find that there is no improvement. Usually, we go back to the hospital and explain that the medicine has not been effective. The doctors then tell us that there is nothing wrong with the child and usually they ask if the mother doesn’t feel itching in the private parts and they suggest a certain sickness other than mauka and so they administer some pills to be applied in the private parts.

(Younger women’s FGD, NTH, 06/10/11)

This support for traditional beliefs by some biomedical personnel was also shown by references to biomedical staff referring individuals to traditional healers.

Treatment failure did not always lead to reconsidering the cause of illness. Participants explained that illnesses caused by witchcraft (ufiti) may not be successfully treated by a traditional healer if the traditional healer is a fake, or if the spirits decide the child will not recover. In this way ufiti, in particular, may provide a kind of ‘fall-back’ explanation for unsuccessful treatment.

Caregivers also had to decide when further visits are no longer necessary. In a number of FGDs individuals stated that if they had been to the hospital with no success and then the traditional healer had failed, then they would give up:

Interviewer: So when you go to the hospital for the first time and it doesn’t help, you go again and it fails again-

Respondent: Then we go to the traditional healer.

Interviewer: You go to the traditional healer, and when he fails?

Respondent: We become stranded, we really don’t have a solution.

Interviewer: Where do you go?

Respondent: Whatever happens, we give in (Laughter).

(Men’s FGD, FFH, 07/10/10)

A few FGD and CII participants described turning to prayer or depending on God, either during treatment, or following failure of both biomedical and traditional sources of care. These references seem in contrast to the idea of ‘giving-up’; rather depending on God was considered an active process that gives hope:

No. I wanted to just add that we do not just give up, but we commit everything in the hands of God, whether or not we sinned against him. We expect anything to happen.

(Men’s FGD, NTH, 17/01/11)

Although severe illness was not investigated in detail it was notable that despite probing into treatment failure, there were few references to death. Mauka, tsempho and ufiti were described as potentially leading to death and in two of the FGDs (both with younger women) individuals stated that death may result from attempting to treat these illnesses at the hospital. However delays caused by attending a traditional healer were described in FGDs and CIIs as potentially leading to death or severe illness if the child had severe malaria:

Whereas the witchdoctor will tell you to cut incisions all over the body of the child; in fact you just delay yourself and if the child has malaria in the brain, you might as well lose the child

(CII, attended a health facility, FFH, 06/01/11)

Study limitations

The qualitative data collection focused on gathering women's views since women in this context tend to conduct the majority of child-rearing duties and were considered to have greater first-hand experience of the treatment-seeking process. Men in similar settings have been seen to be important decision makers within households [5,45,46] and therefore FGDs were also conducted with men. However the more limited data collection among this group may mean that male views and experiences are not as clearly expressed or represented. Additional sampling biases were present: FGD groups were selected by Community Advisory Groups set up as part of the parent study, and individuals selected may have represented more elite members of the community. However selection of villages for FGDs ensured both remote, more deprived communities were represented alongside the more central, wealthier villages and levels of wealth were known to be fairly homogenous within these sub-groups. Selection of individuals for CIIs did not include any children admitted as inpatients at the time of the MIS and therefore experiences of more severe illnesses may not have been fully captured. Selection of CII participants also focused on those whose child had recent probable malaria so experiences of non-malarial febrile illnesses may not have been fully represented.

Since this study was undertaken, Community Health Workers in hard-to-reach areas of the Chikhwawa district have increased the services that they provide and ACTs are now available at rural health posts. These are provided without diagnostic support. This could clearly have a potential benefit in reducing the waiting time during the assessment period and lead to more rapid treatment of malaria. However, as many childhood fevers are not due to malaria, it is unclear how a failure to respond to treatment will alter the population’s perception of formal health care and contribute to promoting initial alternative sources of care.