In general, most participants had low knowledge of TB. A majority self-treated before seeking care from the formal health sector, and most sought care at public health facilities. A diagnosis of TB was confirmed at public facilities in all but two of the patients. However, only half of patients visiting these public facilities and one of the twelve patients who visited private facilities for the first time were diagnosed with TB.
Understanding of TB aetiology
Many TB patients started with relatively non-specific symptoms, partially explaining the observed delay in seeking care or difficulties in obtaining appropriate treatment among the study participants visiting health facilities. Many participants also held ideas about TB, most of which were inconsistent with the fact that bacteria cause TB. Client explanations of aetiology of TB in this study suggest the presence of popular and folk beliefs . Some patients mentioned sharing utensils, food, and water as the cause of their TB; others attributed TB to hard physical work or exposure to cold or smoke. Still others believed that TB is inherited. Similar findings have been reported in a previous study  and elsewhere . These findings are of concern, since participants being newly enrolled in a district treatment programme should have been clearly informed about the aetiology of TB at the start of their treatment.
Health care seeking
Early detection and treatment of tuberculosis is critical to controlling the disease . Although cough and fever were common symptoms at onset, these symptoms alone did not always prompt patients to seek medical treatment early. Symptoms in the early stages of TB are not very specific and may be attributed to self-limiting illnesses, such as viral infections, and only when symptoms become worse or persist will the person consult a health service. Symptom misinterpretation has been associated with patient delay in other studies internationally, where they are attributed to external causes such as overwork or exposure to cold [23, 24]. This delay in health-seeking behaviour is likely to have increased the risk of morbidity, mortality, and transmission of TB to contacts.
Personal and community knowledge of TB and interpretation of health beliefs influences attitudes and health-seeking behaviour significantly [25–27]. Self-treatment, involving a variety of home remedies and traditional and modern drugs, is the first step in the health-seeking behaviour process. This observation is consistent with previous studies, and is linked with the perception of the seriousness of the symptoms and the label the patient attaches to his/her condition [9, 11, 28]. Persons with untreated sputum smear positive TB can infect 10 to 14 others in a year . There is a need for interventions that encourage symptomatic individuals to seek medical help early.
Missed diagnoses at health facilities
Lack of money for diagnostic tests and low suspicion of TB at health facilities caused further delay in obtaining correct treatment, once a person decided to seek care at a health facility. Only half of patients who visited public health facilities and only one among those who visited a private facility for the first time were diagnosed with TB. Though we did not investigate this, an extensive review of previous studies show that most private hospitals serving the urban and rural poor in the developing world are ill-equipped and their staff unqualified, hence the low suspicion for TB among patients who visited these facilities . This confirms findings from a previous study in Kenya that showed that health units failed to investigate chronic coughs in a certain proportion of TB suspects . Even though a good proportion of patients visited providers in the private sector, there was a marked decrease in the number of patients seeking care at these facilities. For instance, at attempt 1, 12 out of the 31 sought diagnosis at a private facility; attempt 2, only 3 out of 31 went to a private facility; attempt 3, zero; and attempt 4, only 2 (Figure 2). While this may suggest a loss of confidence in these facilities by patients, more investigation is needed to explain this shift in health care seeking.
There are many consequences of missing the diagnosis of tuberculosis, and this raises several programme and policy issues . For the patient, misdiagnosis and faulty treatment leads to loss of scarce time and money in the search for treatment, and may increase the duration of illness and the possibility of death. For public health officials, misdiagnosis causes an underestimate in the rate of incident TB, and increases the duration of infectivity. Interventions that could improve the likelihood of TB diagnosis at health facilities may include implementation of standard screening procedures, additional training of health care workers, education of patients (so that they expect and request diagnostic testing for TB when appropriate), and better access to and reduced costs for diagnostic tests.
Very few people felt that their having TB had affected their relationships with friends and family, and if it did, it appeared to elicit more support. This finding is consistent with other studies that have documented social stigma of TB to be more extra-familial than intra-familial . This is important because social support provided by family often plays a pivotal role in promoting early TB diagnosis and adherence to treatment [11, 34]. However, a small proportion of our participants expressed how they were treated as though they had HIV due to their cough and mass wasting.
The association between HIV and TB could extend existing TB stigma, as observed elsewhere [35, 36]. Our study also suggests that stigma related to HIV infection may reduce TB test uptake among TB suspects. Stigma is linked to concealment of symptoms, treatment default, isolation from support networks, decreased self-esteem, self-perception, and self-care . Health education should therefore aim at reducing tuberculosis-related stigma. The key here is the Provider Initiated Testing and Counseling programme, which is now part of the integrated service delivery in Kenya. This programme promotes the awareness that TB is not always associated with HIV , and that it can be cured in both persons with and without HIV.
Potential reasons for defaulting on treatment
One of the objectives of our study was to assess the potential reasons for defaulting on treatment. We found that our study participants could decide to interrupt TB treatment because of lack of knowledge on the duration and the importance of completing the full treatment course, improvement in symptoms, drug stock-outs at the health facility, side effects of TB medications, and lack of social support. These findings are consistent with other studies [39, 40]. Patient narratives seemed to suggest lack of communication and lack of patient involvement in the treatment process, leaving them poorly equipped to take an active role in managing their own health. They are, therefore, poorly prepared to make informed decisions about their treatment . These are factors that have been found to be associated with high rates of treatment default . An improvement in these aspects of TB treatment is crucial in encouraging patients to continue with treatment for the full duration of the regimen.
This was a sociological study involving in-depth interviews. For this reason, the sample size of our investigations was small, with only 31 respondents selected by purposive sampling method which could have introduced bias into the results. We tried to reduce recall bias by including only patients who were diagnosed no more than four weeks before the interview. Undetected "cases" were not included, and have been addressed in our recent study that actively identified TB cases in a TB prevalence survey in the same study area (AH van't Hoog, BJ Marston, JG Ayisi, JA Agaya, O Muhenje, LO Odeny, J Hongo, KF Laserson, MW Borgdorff: Risk Factors for TB Case Finding in a high-HIV population: a Comparison of Prevalent and Self-reported TB Cases, submitted). Our outcome measure of delay in seeking care was self-reported, and no attempt was made to verify the patient reports. We did not collect HIV status of the interviewed participants, hence a possibility of misclassification of non-TB symptoms as those of TB, with consequent unrealistically long periods of delay (several years). Our findings may only be applied to the factors studied, for this study did not intend to assess other important aspects of TB control, such as perceptions and attitudes of health care providers, the community, and suspects, nor did it assess the quality of the TB control programme.