What does this study add to previous knowledge?
Qualitative analysis of 13 SSI and 3 FGDs with PHC staff suggest that childhood TB is rarely suspected and not perceived to constitute an important health problem in the study area. While this finding could reflect a genuine low burden of childhood TB in the population, it conflicts with previous data, which indicates a rising burden of total TB with evidence of children contributing significantly to the total TB caseload in high-endemic communities [
1,
2,
22]. We find it more likely that our finding reflects an inadequate awareness of childhood TB in the study setting. Most of the staff in our study had not received specific education or guidelines with regards to childhood TB, although such guidance is essential to create awareness and ensure implementation of expected practices. Nor did the staff receive feedback from the hospital subsequent to referral of suspected childhood TB cases, leaving them unaware of the burden of TB disease suffered by children. Retrospective studies support this finding, showing that reports of children diagnosed with TB at referral hospitals do not always get back to the primary care clinics [
22,
23].
Childhood TB cannot be viewed as a single disease entity and the natural history of TB, as it is described in the pre-chemotherapy literature, comprises a wide spectrum of manifestations [
24]. Our study revealed that PHC staff had a uniform expectation that children with TB present with a long history of disease, classic and obvious symptoms of intra-thoracic TB combined with apparent risk factors. TB was rarely considered as a differential diagnosis, despite the fact that TB can mimic many common childhood diseases including pneumonia, generalized bacterial and viral infections, malnutrition and HIV. We argue that this indicates a missed opportunity for childhood TB case detection, as childhood TB is not suspected in the early stages of the disease where symptoms are less severe and more general. Moreover, our findings show that TB will rarely be considered in children with unknown exposure to TB, or in more complicated situations with haematogenous spread and extra-pulmonary TB.
The PHC staff identified the most important needs for improving identification of children with TB at PHC level to be training, better tools and appropriate guidelines. Our findings show that the health staff were well informed about the core symptoms and risk factors for childhood TB, but not the extent of the childhood TB burden and wide spectrum of clinical presentations. We support efforts to train health staff, but strongly argue that the focus of training should be in accordance with the responsibilities of PHC level health staff, i.e. to suspect and refer cases. Based on our findings, training should aim at increasing awareness of the burden of childhood TB backed up by the best available epidemiological information, and expand the knowledge of clinical signs and symptoms of the disease with emphasis on early symptoms and timely referral of these children.
Our study also identified a missed opportunity for continuous learning and motivation of PHC staff through strengthening of supervision and feedback systems. The staff did not feel that the present supervision of their work was adequate or supportive. They were further demotivated by not receiving feedback about their work or the outcome of patients referred to hospital. Another study from Tanzania describes absence of referral feedback to PHC level as a general constraint to quality health service delivery and motivation of PHC health staff [
25]. TB registers and treatment cards were not available at the privately or NGO managed health facilities included in this study, which indicates a serious breach in the chain of providing referral feedback to these health care providers.
Training, positive and more systematic supervision as well as structured feedback systems have previously been emphasized as important areas to develop health workers' clinical skills, increase job satisfaction and reduce diagnostic delay [
7,
25,
26]. However, a number of studies that evaluate IMCI have demonstrated difficulties of implementing effective training and supervision, even though IMCI is relatively well institutionalized in national health systems [
27‐
29].
Better diagnostic modalities have been the target of much research [
1,
3,
4]. In response to the diagnostic challenges, a variety of clinical scoring systems have been developed to assess child TB cases. These clinical scoring systems are meant to provide health staff in resource poor settings with a rational, stepwise and feasible aid for childhood TB screening, but are often criticized for lacking standard symptom definitions and adequate validation [
30‐
32]. While the PHC staff in our study requested simple tools and guidelines to identify child TB suspects, the clinical score chart included in the Tanzanian NTLP Manual, was unknown to the staff in our study. Moreover, the score chart is partly based on x-ray and other investigation tools, which are unavailable to the PHC staff and hence not adapted to their setting and responsibilities. An assessment report of the management of childhood TB in Tanzania confirms these findings [
33]. It is evident that clinical decision support systems must be targeted the level of use and only be based on applicable tools. We argue that tools and guidelines made available at PHC level must reflect the responsibilities and capacities of the PHC staff and take into account the current challenges reported by PHC staff e.g. limited knowledge of childhood TB, limited access to diagnostic tools, inadequate staffing and stigma associated with TB. We recommend an assessment of various strategies for clinical decision support at PHC level e.g. by integrating childhood TB into the existing IMCI guidelines with due attention to evidence from Tanzania and South Africa of limited adherence to IMCI guidelines [
34,
35].
Low referral rate of paediatric cases from rural facilities to hospital has previously been shown in Tanzania [
34,
36]. In our study, referral of child TB suspects was a rare event and supposedly compromised by patient incurred costs associated with referral to hospital e.g. transportation fees and time lost from work. A study from Ethiopia confirmed that costs related to TB diagnosis incurred by patients and their families represent a significant proportion of their monthly income [
37]. They conclude that health providers' low level of TB suspicion in adult cases results in several consultations at various health facilities before referral and diagnosis, increasing cost of TB diagnosis significantly. This underpins the need for strengthening the PHC staff's skills in early suspicion of childhood TB on the basis of moderate symptoms and appropriate referral of suspected TB cases, while limiting unnecessary referral and overburdening of the National TB programme with unfounded TB suspects.
Contact tracing may be a powerful strategy for improving case detection, and early identification of children with TB [
38,
39]. This is recognized by WHO which recommends active tracing of young children in close contact with a source case of pulmonary TB as a desired element in the identification of new childhood TB cases [
9]. In acknowledgement of the limited resources available for such initiatives, WHO has suggested that a prophylaxis register, keeping track of contacts, should be added within the structures of the country NTLP. The PHC staff in our study setting suggested a similar strategy, but only when confronted with the issue directly, and there were no on-going initiatives with regards to contact tracing. Given the many obstacles previously identified in implementing active case finding and preventive treatment [
39,
40], we find it important to formulate a specific and detailed strategy of contact tracing within the NTLP.
Limitations to the study
This study represents the views and experiences of PHC staff from one specific geographical area in Tanzania, which limits the scope for drawing general conclusions. However, the findings are characterized by in-depth and rich information with a range of different perspectives from typical health staff working at Governmental, NGO and private PHC facilities. This makes us confident that our findings can be translated to settings comparable to our study setting. We realise that the researchers' professional status as medical doctors was likely to yield answers that focused on the biomedical aspects of childhood TB rather than the processes of case identification. It is moreover possible that the informants may have avoided exposure of their challenges and sought to produce what they considered as correct answers in order to please the researches. This may explain why it proved difficult to elicit answers beyond standard answers concerning the weaknesses and challenges in case identification, particularly in the SSIs. We found the FGDs more useful in exploring these more sensitive and pertinent issues.
The use of an interpreter can influence the quality of information gained. This was sought reduced by preparing the interpreter before data collection to ensure that he had knowledge about the purpose of the study, his role in the study and the information that interview questions aim to elicit.
To add further quality to the findings, our study could have benefited from direct observation of the health staff at their work site to triangulate between what was said about a situation and what can otherwise be known or experienced concerning this experience. The time commitment required to do this was not possible in this study, but observation of participants is recommended for follow-up studies.