Care seeking for childhood illnesses in rural Mtwara, south-east Tanzania: a mixed methods study

Abstract Background Care seeking was assessed in preparation for a study of the health impact of novel design houses in rural Mtwara, Tanzania. Methods A total of 578 residents of 60 villages participated in this mixed-methods study from April to August 2020. Among them, 550 participated in a healthcare-seeking survey, 17 in in-depth interviews and 28 in key informant interviews. Results The decision to seek care was based on symptom severity (95.4% [370]). Caregivers first visited non-allopathic healthcare providers or were treated at home, which led to delays in seeking care at healthcare facilities. More than one-third (36.0% [140]) of respondents took >12 h seeking care at healthcare facilities. The majority (73.0% [282]) visited healthcare facilities, whereas around one-fifth (21.0% [80]) sought care at drug stores. Treatment costs deterred respondents from visiting healthcare facilities (61.4% [338]). Only 10 (3.6%) of the households surveyed reported that they were covered by health insurance. Conclusions Quality of care, related to institutional factors, impacts timely care seeking for childhood illnesses in Mtwara, Tanzania. Ensuring accessibility of facilities is therefore not sufficient.


Introduction
In sub-Saharan Africa, children aged < 5 y (i.e.'under-five') remain highly vulnerable to infectious diseases.Each year, malaria is responsible for about 247 million clinical cases and 619 000 malaria-related deaths occur worldwide. 1Diarrhoea accounts for > 90% of under-five mortality in low-and middle-income countries. 2Acute lower respiratory trac t infec tions contribute to approximately one-third of under-five mortality in sub-Saharan Africa. 3lthough the United Nations' Sustainable Development Goal 3 advocates for access to healthcare services for all, in sub-Saharan Africa, 17% of the population lives ≥2 h away from public health facilities. 4 -6Inequities in access to healthcare in sub-Saharan Africa are exacerbated by the mismatch between a high disease burden and the lack of healthcare facilities in rural areas. 6actors influencing decisions to seek care for childhood illnesses include personal factors, such as child age, sex, knowledge of caregiver and illness perceptions 4 , 7 ; interpersonal and community factors, such as perceived quality of healthcare, influence from families, friends and social networks regarding healthcare; and institutional factors, such as access and affordability of healthcare. 7Seeking care is a complex process and, because of issues related to access and preference, caregivers may sometimes first visit non-allopathic providers or opt for home treatment, 8 resulting in potential delays in seeking care. 9 , 10n preparation for the 'Star Homes' project, a randomised controlled trial of a novel design house intervention, a baseline assessment of healthcare-seeking behaviour for childhood illnesses, was conducted.The Star Homes project aims to evaluate the protective efficacy of a new house design targeting malaria, respiratory trac t infec tions and diarrhoeal diseases in rural Tanzania. 11A total of 550 households from 60 villages are participating in this study, of which 110 have been allocated novel design houses, whereas the remaining 440 households reside in traditional wattle/daub houses as the control group. 11A baseline assessment of healthcare-seeking behaviour was needed to ensure the comprehensive capture of all disease episodes during the trial, particularly given the limited knowledge about healthcare seeking for childhood illnesses in Mtwara.
The main objective of this study was to explore the healthcareseeking behaviour for childhood illnesses.

Study location
Mtwara, located in south-eastern Tanzania, is a major cashewproducing area.More than 90% of its residents are involved in cashew farming.Subsistence farming and small-scale fishing are also important economic activities. 12Mtwara has a population of 1 634 947 across nine administrative district councils; this study was conducted in Mtwara District Council, which has a population of about 158 504 residing in 110 villages. 13There are 38 government healthcare facilities in Mtwara District Council. 14

Study design
This study utilised a mixed-methods approach, involving a questionnaire-based survey, interviews and focus group discussions (FGDs).

Recruitment of study participants
A total of 578 residents of 60 villages participated in the study from April to August 2020.Among them, 550 participated in a healthcare-seeking survey, 17 in in-depth interviews (IDIs) and 28 in key informant interviews (KIIs).The number of interviews was determined by the point of data saturation.Parents/guardians of children aged < 13 y participated in 14 FGDs each consisting of six to 10 participants (Table 1 ).

Inclusion criteria
A participant had to be a household head recruited in the Star Homes project, aged ≥18 y and had to provide written informed consent.KIIs included selected influential community members, such as community leaders, community health workers and medical officers at local health facilities.For the FGDs, any parent/guardian from the Star Homes project who was taking care of a resident child aged < 13 y was invited.

Data collection
The questionnaire was developed to collect sociodemographic information of the participants and cues of action from a child's

Data management and analysis
Quantitative data were analysed using Stata software version 14 (Stata Corp, College Station, TX, USA) where categorical variables were analysed as proportions.Comparison tables were used to examine the differences in proportions where a χ 2 test was used to assess the statistical significance of the differences.Medians and ranges were used to summarise continuous variables.Multivariable analysis was conducted to identify factors independently associated with healthcare-seeking behaviour.Statistical significance was ascertained when p < 0.05.The 95% CIs are reported where appropriate.Qualitative analysis was conducted using ATLAS.tiversion 9 software ( http://www.atlasti.com).Data were coded using a predeveloped codebook.The codebook was developed by the site investigators together with the research assistants.Data were coded by the research assistants familiar with the local sociocultural context.Coding was followed by thematic analysis, supported by extracting relevant quotations.

Diagnosis of (febrile) illness
About 89.3% (491) of household heads recalled one of their child's last fever episodes.It was generally detected by placing a hand on the forehead of the child (73.1% [359]) to see if they felt hot.A small minority had access to thermometers and could measure axial temperature (5.9% [29]).The diagnosis was made mostly by parents (72.1% [354]) and they did not seek advice from their friends/family regarding the illnesses (86.2% [423]) (Table 3 ).
Participants in the FGDs and IDIs recognised the severity of illnesses in children by assessing the progression of symptoms and observing their behaviour.Parents perceived a child as seriously ill when the child exhibited signs such as unusual tiredness, reduced playfulness, visible weakness and elevated body temperature.Parents would ask children who could communicate about their symptoms.Malaria was considered the most common problem in the community.Diseases such as diarrhoea and respiratory trac t infec tions were equally reported as major health problems: The symptoms to help [assess] the child's illness has become severe are: the child having no desire to play, refusing to eat and other conditions, such as not being happy (FGD-0802, 112-129).

I recognise the severity of the illness because I am the head of the family. […] [W]
hen he is sick, he will be sleeping all the time, and when you ask, he would say that the body, head, legs and other parts of the body are hurting, and they are weak.Therefore, I recognise that the illness is severe because when children are well, they go out and play (IDI-082001, 50-62).
The most common illnesses include malaria, cough and respiratory diseases and diarrhoea.These are the three most common illnesses that bothers people in this area (KII-HC08007, 26-36).

Care seeking
Four-fifths (79.0%[388]) of participants sought medical care when their children fell sick; 72.7% (282) went to healthcare facilities.The decision to seek medical care was based on severity of illness (95.4% [370]), which echoed the care seeking for older people, of whom 60% (330) were sent to the hospital only when they had severe illnesses (Table 4 ).Timeliness in seeking care was assessed by the number of hours that had elapsed between identifying the child's illness and When the children get this problem, most of the time, we take them to the dispensary, for example here in Magomeni our nearest dispensary is at Tangazo (IDI-042010, 62-63).
Figure 1 Words crowd from all interviews and FGDs (Mtwara District Council, April-August 2020).In this word crowd, the words in large font are those that were most used during the interviews.This is an indication that children were most likely to be brought to the health facilities (dispensary) when they became ill.

They go to the dispensary, talk to the doctor and the doctor will direct them in terms of what to do (KII-CH08009, 288-307).
There was no significant association of health-seeking behaviour with gender, education level or household expenditure among participants (Table 5 ).
Although participants described initially seeking care at healthcare facilities, insights from IDIs, KIIs and FGDs revealed that participants opted to treat at home using traditional herbs or purchase medication from drug stores before eventually seeking care at dispensaries: When my children get sick, first, I go to the pharmacy, and if the pharmacy provider tells me to go to the dispensary, then I will go (IDI-041011, 262-266).

Also, if you don't have money you can decide to take leaves that cure the disease. For example, neem leaves (mwarubaini): you boil [them] and leave it for some time, then later you drink it and use it to treat the body (IDI-082001, 82-86).
This kind of health-seeking behaviour was seen by some as contributing to delays in seeking care, promoting the progression of illnesses: They don't bring the child to the hospital early; they bring the child to the hospital when the illness is severe.However, there are few who understand and bring the children early to the hospital, but most of them bring them when the condition of the illness is worse (KII-CH08013, 33-33).
But, most of the time, we usually don't take the child to the hospital on the first day we notice that s/he is unwell.We just buy 'Sheladol' (a painkiller) for them.We take the child to the hospital on the second day after his/her condition worsens (FGD-0814, 77-114).
Traditional medicine or healers were sought for illnesses perceived as untreatable at the hospital.When family members were repeatedly brought to the hospital with a complaint and it was not cured, the community members perceived that the disease could only be treated by a traditional healer: There are diseases, [and] even if you take the child to the hospital, the child will not get any treatment.For example, when you went to the market and suddenly the leg starts hurting and swelling then, my friend you cannot go to the hospital you must go to spiritual or traditional healers, so that you can get treatment.It is true when you go there, they will tell you that someone [bewitched you] on the way and when they give you medicine for three to four days, the legs will improve (IDI-082001, 122-125).
Participants were concerned about affordability of healthcare (61.4% [338]).Study participants made cash payments at the healthcare facilities, with the overall treatment cost being < 10 000Tsh (US$4.3)for those who sought care (51.4% [145]), equivalent to one-quarter of their monthly expenditure.Around one in five respondents (20.6% [58]) paid 11 000 (US$4.8)-50 000Tsh (US$21.7),while 21.6% (60) did not incur any cost because healthcare was provided free for children aged < 5 y; a very small percentage incurred > 50 000Tsh (US$21.7)for treatment.Only 10 (3.6%) households who utilised healthcare had health insurance at the time of the survey (Table 4 ): years and above, you need the average of nine thousand shillings to cover the costs of all the services that you need at [the] dispensary (FGD-0804, 178-182).

Most of us are not financially well-off enough to pay for health insurance, but a few people are using health insurance. In our dispensary they have outlined that, from six
Yes, health insurance.Those who have money, they pay for the insurance and get an insurance card, which can be used by six people in the household (IDI-81007, 121-122) .
Respondents described how children aged < 5 y were not required to pay for healthcare services.When children aged ≥5 y or adults became ill then the cost of treatment was perceived to be a major deterrent to seeking healthcare: You pay depending on the patient because, below five years [of age], the patient is treated for free.But, for six years and above, a test for malaria costs 2000 shillings; seeing the doctor is 1000 shillings; and testing for a urinary tract infection is 4000 shillings.Therefore, you must pay for any lab test (FGD-0804, 153-163).
The clinical officer in one of the health facilities explained how they categorised patients.Treatment was free for underfive children, TB and HIV patients and pregnant women.He further explained that the medicines were free for special conditions, such as epilepsy and asthma.Those who did not have health insurance had to pay for health services.There were two types of health insurance available, National Health Insurance Fund and Improved Community Health Fund.Although healthcare for under-five children was provided free, because of stockouts, sometimes parents were obliged to buy medicines from drug stores.
Respondents described the distance to the healthcare facility as a major determinant of health-seeking behaviour.Travel time and expenses influenced care seeking, yet health facilities were reported to be mostly within 1 h on foot from respondents' homes (78.3% [138]): …with motorcycle, it is half an hour to 40 minutes and the fare is 7000 to 10 000TSh for a return trip (KII-CH04004, 268-309).
Perceived quality of care influenced healthcare seeking.Respondents were concerned with the number of staff at facilities and they described how insufficient staff numbers had implications for the care delivered.Healthcare workers acknowledged deficits in staffing and medical supplies necessary for providing quality care: You might find a nurse is attending a malaria patient and at the same time there is a woman who is due to deliver.It is very difficult (FGD-0804, 411-422).
For example, in our dispensary, it would be better if they improve by increasing medical supplies because most of the time we are told that there is no medicine (FGD-0415, 688-713).
Apart from understaffing and a shortage of medical supplies, undesirable interactions between clients and service providers were said to affect healthcare utilisation.Some service providers were alleged to be untrustworthy because they asked for bribes from patients.Lack of confidence on the part of either party led to underutilisation of healthcare: A child won't get tested without cash.If you don't give 2000Tsh, a child would be sent back home, or they will collect her/his registration/clinic card until you pay.The healthcare provider will ignore you if you don't give her/him something.This should be looked at (FGD-814, 423-436).

Discussion
In rural Mtwara, Tanzania, seeking care for childhood illnesses depended on personal factors, such as the perceived severity of illness, and benefits of early treatment, as well as institutional factors, such as accessibility and the cost of health services. 15t the community level, healthcare seeking was influenced by sociocultural dynamics, such as beliefs related to specific illnesses, including perceived severity and the perceived competence of healthcare services. 4Respondents generally sought care for illnesses with severe signs and symptoms. 16Seeking care at health facilities was, however, often delayed until symptoms were severe. 15Lack of urgency has also been established as an independent factor affecting treatment-seeking behaviour. 17nstitutional factors, such as the availability, accessibility, affordability and acceptability of healthcare, had a significant impact on healthcare seeking.These findings align well with the WHO's description of how institutional factors affect the coverage and quality of care. 18Availability refers to the existence of obtainable health services and medications. 19Most of the health facilities in this study were within 1-h walking distance, which suggests reasonable coverage, 20 although this might not necessarily mean good access and quality care.Understaffing and stock-outs influenced care seeking, regardless of accessibility. 21lthough respondents recognised the need to seek care when a child displayed symptoms, uncertainty and inconsistency of services in facilities, particularly anticipated long waits and medicines stockouts, affected their responses.As a result, parents may anticipate and thus seek treatment at private drug shops, ultimately undermining the achievement of universal health coverage, even when risk pooling through health insurance is introduced.A recent review also highlighted the challenges in accessing healthcare for community members, suggesting the need to enhance community-based health services in the rural communities of Tanzania. 22are seeking was affected by the availability of funds to pay for transport and healthcare costs at facilities.Indeed, in many settings, a lack of access to cash to cover expenses often affects treatment-seeking behaviour. 23Most respondents explained the absence of health insurance and attributed this to financial cosntraints.Even if they had health insurance, they had to make some cash payments because of occasional health facility stockouts, which forced them to purchase medication from private drug stores. 24Out-of-pocket payments for healthcare are anathema to the idea of universal coverage of healthcare and significantly affect the sustainable development goals. 25 , 26Tanzania established a low-cost health insurance system (Community Health Fund) in 2001 to provide the poor rural community with access to a prepaid system of healthcare.This health insurance scheme costs 30 000Tsh (US$13) per year for a family of six people.By 2018, 2.1 million households and 12.6 million beneficiaries were enrolled. 27Nonetheless, the implementation of this health insurance scheme has faced challenges and its uptake is still limited.Respondents regarded enrollment in the Community Health Fund as expensive and unaffordable. 21

Strengths and limitations
Most of the data in this study were collected from the household heads and we did not triangulate the information with the health facilities, drug stores and traditional healers.This potentially gives a one-sided perspective on care seeking.The study may have incurred recall and social desirability biases.Aligning with the original analysis plan, we did not model the patterns of delayed care seeking, including factors and their magnitude, in affecting outcomes for specific children.Future research can incorporate regression models to explore factors affecting care-seeking behaviour.

Conclusion
Improving access to healthcare facilities and ensuring the availability of high-quality services is a global priority.This study of care seeking for childhood illnesses in Mtwara, Tanzania, highlighted the relevance of institutional factors along with other influences.Hence, increasing access without improving quality of services may not promote greater utilisation of health facilities.In this setting, future research is needed to assess the impact of health insurance schemes on care quality and care seeking.Furthermore, data collection took place before the Star Homes S. Mshamu et al.   project and follow-up research should assess the impact of its implementation on care seeking in participating communities.
Authors' contributions: All authors read and approved the final manuscript for this submission.

Table 1 .
Summary of qualitative data collection methods and a description of respondents (Mtwara District Council, April-August 2020)

Table 5 .
Action taken vs gender, relationship with children, education level, monthly household expenditure (Mtwara District Council, April-August 2020)