Background
Tuberculosis (TB) remains an important challenge for global health. The World Health Organization (WHO) estimated that in 2017 there were about 10 million incident cases of TB worldwide, and 10% of these were in children under 15 years old [
1]. This is probably an underestimation, as childhood TB often remains undetected due to diagnostic challenges [
2,
3]. The burden of this disease in children reflects its prevalence in adults [
3].
Management of childhood TB poses challenges to low- and middle-income countries. Morbidity and mortality are high enough to make this one of the three primary infectious diseases related to poverty [
4‐
6]. In 2017, about 233,000 children died from TB worldwide, which translates to 13% of total deaths in children [
1].
A 2018 WHO report ranked Ethiopia 14th worldwide with regard to the global TB burden; 117,705 cases were reported in 2017, with an estimated incidence rate of 164 cases per 100,000 population per year. Of the reported cases in 2017, about 20,000 were in children under 15 years [
1].
Given the natural history of TB, age remains the most important variable influencing the risk of disease progression following a primary infection with
Mycobacterium tuberculosis. Disease manifestation patterns show clear associations with age at the time of primary infection. Immune-compromised children also carry an increased risk for TB [
6], so it seems prudent to categorize all children under 2 years of age and immune-compromised children of any age as being at high-risk for this disease [
7].
Although several studies have been published about childhood TB in Ethiopia, few have taken place in a rural environment. Moreover, they do not generally compare age groups in terms of epidemiology, disease manifestations, diagnosis or outcome [
8‐
12], even though this information could be very useful for designing effective treatment strategies.
Ethiopia’s National Tuberculosis and Leprosy Control Programme (NTLCP), launched in 1992, helped standardize the registry of TB cases following WHO guidelines [
13,
14]. This study describes the main characteristics of childhood TB in a rural hospital in Southern Ethiopia over 18 years, according to NTLCP registry data. We also compare epidemiological, clinical and outcome data in children aged under 5 versus those aged 5 to 14 years.
Discussion
More than a third (36.3%) of the total recorded cases of TB in our study population occurred in children—a higher proportion than reported elsewhere in Ethiopia [
9,
11,
17], in the national estimate (16.1%) [
18], and in other areas of sub-Saharan Africa [
19]. Several factors may have contributed to these results. As our study took place in a private hospital, it offered several services, such as X-ray diagnosis and cheaper treatment, that may not have been present in public facilities. At the same time, the hospital has a referral malnutrition clinic, where children with malnutrition are admitted for a minimum fee, and these children may have TB. Some overdiagnosis of child TB cases may have also been possible.
Given the scant data available on childhood TB (and the dominance of high-income settings where studies exist), in 2007 WHO called for additional research to characterize the global epidemiology of the disease [
20].
Nearly half (45.5%) of the total childhood TB cases were in children under five, which is consistent with results from a previous study in Tanzania [
21] but higher than in other studies in Ethiopia and other low- and middle-income countries, where the proportion of TB cases in this group is estimated at 11 to 23% [
8,
9,
22]. Overall, the heterogeneous estimates reflect limitations in available data both at a country and regional level.
Childhood TB is mostly due to recent transmission, so its burden provides a good proxy measure for TB prevalence in the community [
19]. Therefore, the high rates of childhood TB that we observed suggest that the community bears a high burden of untreated disease. Moreover, 41.6% of Ethiopia’s population is aged under 15 years [
23], and this is also relevant to the burden.
There has been a decrease in prevalence of tuberculosis in Ethiopia over the past 15 years [
1]. In our study, the annual case count also decreased over the study period, especially in the final 4 years, as the number of patients transferred to local health centers (and whose cases were therefore registered there instead of in the hospital) increased as these centers were newly constructed in the area. We consider that most of the reduction in numbers is due to a shift in patients receiving treatment in new community centers rather than in the hospital.
Most cases (98.2%) were new TB rather than previously treated TB. This proportion was slightly higher than in other studies of childhood TB (92%) [
8,
9]. The younger age group presented with more new cases, which is logical, as for the most part such children are too young to have already been treated for the disease. In general, previously treated TB and transfer cases in our study were infrequent.
PTB was the most common type of TB in our children, which is consistent with the literature [
18]. In our study, microbiological diagnosis (only acid-fast bacillus smear positivity, as bacteriological confirmation by culture was not done) was very low, as in other studies in Africa [
21]. Just 12.5% of the PTB cases were smear positive, which is also in line with previous reports in Ethiopia [
9,
11]. Most of the patients with smear-positive PTB were aged 5–14 years. Children develop mainly primary PTB presenting with lymph nodes and areas of parenchymal infiltration, and they tend to have low bacillary load and immunological changes, which explains this result [
9]. Moreover, young children rarely produce sputum samples for smear microscopy, so they are generally diagnosed based on clinical examination and chest X-ray [
24]. This was also true in our study, in line with NTLCP/WHO guidelines on diagnosing children [
9]. The probability of smear positive sputum is positively correlated with the child’s age [
8,
9,
24,
25].
Using other sample types, such as gastric aspirate, results in a higher diagnostic yield than spontaneous sputum in children. The Xpert MTB/RIF assay in sputum and gastric aspirates may identify
M. tuberculosis in a greater percentage of cases in this group [
25‐
27]. Thus, the gradual introduction of Xpert MTB/RIF in gastric aspirate and induced sputum samples could help to improve diagnosis in children under 15 in rural settings [
26,
27].
EPTB made up a third of the childhood TB cases, which is a lower proportion than in the studies by Hailu et al. [
9] and Tilahun and Gebre-Selassie [
8], where EPTB accounted for about half. There were more cases of EPTB in the older age group compared to the younger children (45.8% versus 17.5%), which is consistent with Hailu et al’s [
9] results in Ethiopia and Mtabho et al’s [
22] observations in Tanzania. In these studies, as in ours, a patient with both PTB and EPTB was classified as having PTB.
On the other hand, there was a slightly higher proportion of TB meningitis in children under five (2.6% versus 1.6% of total TB); it is well known that this type of TB is more frequent in children under three [
27].
Abdominal TB appeared less frequently in younger children (4.9%) compared to older ones (13.8%). Although this variant is rare in children, most reported cases are in low- and middle-income countries with a considerable TB burden. Consistent with our results, most cases of abdominal TB are in patients aged over five [
28,
29]. Diagnosis is extremely challenging; our patients were diagnosed by abdominal ultrasound, ascitic fluid analysis (lymphocytosis) and clinical suggestion because microbiological confirmation was not possible. Only a few cases had histopathological diagnosis.
Overall TB/HIV co-infection in our series showed a prevalence of 2.3%, much lower than elsewhere in Ethiopia, where it ranges from 10 to 28.2% [
8,
9,
30‐
34]; in other African countries, where it ranges from 14 to 50% [
19,
31,
33]; and in Southeast Asia [
22,
35]. The low proportion of our patients tested for HIV precludes any conclusion on why the prevalence was so low; however, we can cautiously speculate that it is because HIV is less prevalent in rural populations than in other areas of Ethiopia [
36,
37].
In our study, 39.8% of the children were admitted to hospital. Some patients were hospitalized for reasons other than TB, such as severe acute malnutrition, and were only then diagnosed with TB. Hospitalization was more frequent in the under-five group.
TB programs in sub-Saharan Africa rarely assess treatment outcomes in children [
18]. The WHO considers an overall treatment success rate of 85% to indicate good quality TB case management [
38]. Our success rate—63.5%—was low compared to that achieved elsewhere in Ethiopia (~ 85%) [
8,
9,
11] and Africa [
22]. If we assume that the patients transferred to other TB clinics (16.3%) fully completed the course of treatment, though, the success rate of 80% is closer to the WHO target [
38].
Indeed, the transfer rate was relatively high in our series [
8,
9,
11] because patients beginning treatment during hospitalization in Gambo General are registered in their local health center upon discharge, and they complete therapy there [
39]. Due to limitations in these registry data, we could not determine the final treatment outcome in these patients [
39].
In our series, the proportion of patients who successfully completed treatment was lower in children under 5 than in those aged 5–14 (56.6% versus 69.2%), but the transfer rate was higher (19.4% versus 13.7%). The rate of treatment loss to follow-up was 13%—higher than in other studies in Ethiopia with more urban and periurban settings [
8,
9] and in Tanzania [
21]. However, our loss to follow-up rate was marginally lower than the 15% reported by Adejumo et al. [
19] in Nigeria.
With regard to the mortality rate, the 4.1% we observed is within the range reported in other studies of childhood TB in Ethiopia [
8,
9,
11,
33]. Elsewhere in Africa, mortality has higher: 10.5% in Botswana [
40], 10.6% in Tanzania [
21] and 17% in Malawi [
41].
Young children, particularly those under two years of age, are at higher risk of dying from TB because their immune system is still immature [
9,
17,
27,
42]. We did not see a difference in fatal outcomes in our series based on age, although this result may be confounded by the lack of treatment outcome data for transfer patients.
In terms of treatment outcome predictors, age group was the only variable that was independently associated with treatment success. This outcome was significantly higher in 5–14-year-olds compared to the under-5 group. Other studies have reported similar findings [
8,
9,
18,
43].
In contrast, other series have reported significant associations between treatment failure and HIV co-infection, unknown serostatus and malnutrition [
2,
9,
39,
44].
The findings of this study must be interpreted in light of its limitations. First, it is common to find both PTB and EPTB in the same pediatric patient, but in this study the variable was dichotomized (EPTB only in the absence of PTB), following WHO definitions. In children, this classification can obscure very important information about serious cases of TB, for instance if the child is diagnosed as having PTB but also has TB meningitis, which could be much more life threatening. Secondly, we were unable to collect sociological data and only limited data about HIV infection, which might have affected the risk of a poor outcome. Finally, Gambo General Hospital is a private hospital, and patients were often referred from other health facilities for diagnosis and admission but then transferred to their local health center following the intensive treatment phase. Thus, we might have underestimated treatment failure and mortality rates. Additional limitations are the retrospective and hospital-based nature of the study, which could cause bias. Moreover, there also seem to be recent changes in health system in the area, which could depress estimates of prevalence.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.