Low case notification rates of childhood tuberculosis in southern Ethiopia

Background Childhood tuberculosis (TB) is a public health concern causing considerable mortality. However, control of childhood TB receives little attention. The control efforts could be inadequate because of challenges associated with difficulties in diagnosing the disease in children. Understanding the burden of the disease among children is important to assess the ongoing transmission of the disease in a community and improving TB control efforts. This study was carried out to assess TB case notification rates (CNRs) and treatment outcomes in children aged less than 15 years over a ten-year period. Methods Data were collected from unit TB registers from all health facilities providing TB treatment in the Sidama Zone in Ethiopia. We analysed the CNRs and treatment outcomes by age category, gender, and place of residence. We used logistic regression analysis to identify factors associated with treatment outcomes and to control for confounding. Results A total of 4,656 cases of children less than 15 years of age were notified as diagnosed and treated for TB, constituting 13 % of all notified TB cases in the study area. The mean CNRs per 100,000 children less than 15 years were 30 for all new cases of TB, 28 for rural cases, 67 for urban cases, 28 in boys, and 32 in girls. The proportions of treatment success were 82 % for new and 77 % for retreatment cases for the entire study period and increased to 93 % for new cases in 2012 (X2trend, P < 0.001). Children less than five years old had a lower treatment success [adjusted odds ratio (AOR) 0.64 (95 % CI, 0.52-0.80)] and higher deaths [AOR 2 (95 % CI, 1.27–3.12)]. The proportion of children who died during treatment among children in the less than 2-year-old age group was three times higher than children in the 2 year and above age groups [AOR 3.34 (95 % CI, 1.92–5.82)]. Conclusion The CNRs of childhood TB were low in Sidama. Children less than 5 years old had a higher proportion of deaths. Efforts need to be made to improve the diagnosis and treatment of TB among children. Electronic supplementary material The online version of this article (doi:10.1186/s12887-015-0461-1) contains supplementary material, which is available to authorized users.


Background
Childhood tuberculosis (TB) is a public health concern causing considerable mortality [1][2][3]. In 2013, approximately 550,000 new cases of TB and 80,000 deaths occurred in children who were HIV-negative [2]. However, childhood TB has been addressed inadequately in resource constrained settings due to challenges associated with difficulties in diagnosis and treatment of the disease [3,4]. Lack of trained human resources to deliver the services, high cost related to training and to provision of services, and limited diagnostic facilities are some of the challenges for childhood TB diagnosis and control [5]. On the other hand, low case notification rates (CNRs) of TB in children could be because of underdiagnoses, underreporting, and due to the lower priority that the health systems give for children with TB compared to adults [6,7]. In areas with poor access to diagnostic facilities and skilled personnel, children could suffer from the disease even without being diagnosed, or they are diagnosed late. Late diagnosis can contribute to considerable childhood deaths and to the transmission of the disease.
In Ethiopia, the National TB Control Program decentralized and expanded the directly observed treatment short course (DOTS) strategy. However, the existing TB control programs may have limitations in addressing childhood TB. The national TB prevalence survey of Ethiopia did not report the estimates of TB burden in children [8]. There is also limited information on age disaggregated case notification and treatment outcomes in routine surveillance system.
The burden of TB in children can serve as an indicator to assess effectiveness of TB control programs, and one of the indications for the ongoing transmission of the disease in a community [9]. Therefore, generating information from surveillance data is useful for understanding the burden of the disease in a community and such information could provide essential evidence for improving tuberculosis control [6,7]. The objective of this study is to assess case notification rates and treatment outcomes of childhood TB over a ten-year period in the Sidama Zone in southern Ethiopia.

Study setting
The study was conducted in the Sidama Zone in southern Ethiopia, which is one of the most densely populated areas in Ethiopia with a population of over 3.4 million. Ninety-two per cent of the population live in the rural areas, and agriculture is the major livelihood of the community. Administratively, the Sidama Zone is divided into 19 woredas (districts) and two towns. There are 524 rural and 39 urban kebeles (the lowest administrative unit for about 5,000 people or 1000 households on average). In 2012, 114 health facilities provided TB treatment and 81 health facilities carried out sputum-smear examination. Ninety-five percent of the districts in the study area had at least one antiretroviral treatment (ART) centre. The demographic and health survey of Ethiopia estimated the adult HIV prevalence in the southern region of Ethiopia, where the present study was conducted, to be approximately 1 % [10]. In 2011, a community based intervention was implemented aimed at improving TB case finding and treatment outcomes [11]. In 2012, there were only two health facilities with radiographic services.

TB diagnosis and treatment definition
We used the National TB control guideline of Ethiopia for diagnosis and treatment, case definition, and treatment outcomes [12].
Smear-positive pulmonary TB (PTB) is diagnosed with at least two positive initial sputum smears for Acid Fast Bacilli (AFB) by direct microscopy, or one positive smear for AFB by direct microscopy and culture positive for Mycobacterium tuberculosis or one positive smear for AFB by direct microscopy and radiographic abnormalities consistent with active TB as determined by a clinician. The laboratory keeps all positive and negative slides for external quality assurance. Quality assurance is performed regularly at the regional laboratory, and feedback is given to a reporting health facility. Previous studies reported a high specificity and good agreement of sputum-smear microscopy for AFB between peripheral and reference laboratories [11,13].
Smear-negative pulmonary TB (PTB) is diagnosed when the patient presents with symptoms suggestive of TB, has at least three initial smear examinations negative for AFB, no response to antibiotics, repeat smearnegative and radiological abnormalities consistent with pulmonary TB, as well as a clinician's decision.
Extra pulmonary TB (EPTB) is diagnosed by one culture-positive specimen from an extra pulmonary site or histopathological evidence from a biopsy, which is based on strong clinical evidence consistent with active EPTB by a clinician's decision. However, most health facilities diagnose the disease based on a clinician's decision because there are inadequate laboratory facilities for sputum culture or histopathology. The body sites for EPTB are lymph nodes, intestine, bone, kidney, central nervous system, and other organs.
A short-course treatment regimen is given in two phases with fixed-dose combination first-line anti-TB drugs [12]. The intensive phase treatment lasts for two months with ethambutol (E), isoniazid (H), rifampicin (R) and pyrazinamide (Z) followed by a continuation phase of six-month with ethambutol and isoniazid. Since 2011, the continuation phase has lasted for four months, and uses isoniazid and rifampicin.
"Cured" is smear-positive TB patient who is sputum smear-negative at, or one 'month' prior to the completion of treatment and on at least one previous occasion.
"Treatment completed" is a TB patient who completed treatment but for whom smear results are not available one month prior to the completion of treatment.
"Treatment success" is patients who was declared "cured" or "treatment completed"

Data collection
The data were collected from August 2012 to February 2013 from all DOTS providing health facilities during 2003 to 2012. We collected unit TB registers from all health facilities in the study area and included the following variables: address of the patient, name of the health facility, age, sex, smear result, TB category, TB classification, intensive phase treatment drugs, year of treatment, date of treatment started, last date of treatment, and treatment outcomes. The treatment results included treatment completed, cured, defaulted (lost-tofollow up), died, transferred out, treatment failure, and unknown. We could not collect the data of nutritional status and HIV status of the children because the data were not recorded in the unit TB registers. Address of patients consisted of actual district and kebele of the patient at the time of diagnosis. Data entry personnel were university graduates with experience of data management. Training was given to data collectors and the data were double entered. We cross checked the data for the number of cases by year, facilities and districts as well as the correctness and consistency of information. The data were entered using Microsoft Access and exported to IBM SPSS (Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp).
We supervised the data entry for completeness and consistency on regular basis. To ensure data quality we checked consistency and correctness of entered information with information in the TB registers. We did an exploratory analysis, looked for errors and corrected them from the registers. Number of cases and patient information entered by each year and health facility were checked with information in unit TB register page by page and year of treatment to ensure data completeness and accuracy.

Statistical analysis
Case notification and treatment outcomes were computed for smear positive PTB, smear negative PTB and EPTB cases. We obtained each year's population data for the study area from Central Statistical Agency of Ethiopia (CSA) [14] and computed case notification rates using under 15 years population for different years by age category as a denominator and notified cases as a numerator. We used IBM SPSS 20 for data analysis. Multivariate logistic regression analysis was performed to determine independent risk factors associated with treatment outcomes and to control for confounding.

Ethical consideration
We obtained ethical clearance from the Regional Health Bureau of southern Ethiopia, in addition to a letter of support from the Sidama Zone Department of Health, to obtain information from all districts and health facilities. Personal identifiers of the cases were coded prior to analysis and medical records were kept in a secure place to help maintain the confidentiality of clinical information of cases.

Results
A total of 4,656 cases of children less than 15 years of age were notified as diagnosed and treated for tuberculosis from 2003 through 2012, constituting about 13 % of cases of all age groups notified in the study area. Fifty-two percent (2,434 cases) of the cases were girls and 48 % (2,220 cases) were boys. Fifteen percent (719 cases) of the cases were less than five years old and 2,433 cases (53 %) were in the 10-14year-age group. Of the cases, 4,087 (88 %) were from rural areas, 1,956 cases (42 %) were smear-positive PTB, and 1,308 (28 %) were smear-negative PTB cases ( Table 1). The proportion of smear-positive PTB in the younger age group (less than 5 years) was lower than the proportions among 5-9 and 10-14 year-age groups ( Table 2).

Case notification rates
The mean CNRs for new cases of TB of all forms were 30 per 100,000 children, 67 in urban and 28 in rural areas, and 32 in girls and 28 in boys per 100,000 children. No decline was observed in childhood TB over a ten-year study period ( Fig. 1 and Table 3). The mean CNR of smear-positive PTB was about 13 per 100,000 children and increased from 11 in 2003 to 16 per 100,000 in 2012. The mean CNRs of smear-positive PTB were 14 for girls and 11 for boys per 100,000 children (Table 4).

Treatment outcome
The proportions of treatment success were 82 % for new and 77 % for retreatment cases, and increased to 93 % in 2012 for new cases (X 2 trend, P < 0.001) ( Table 3). The proportion of cases lost-to-follow up declined from 20 % in 2003 to 1 % in 2012 (X 2 trend, P < 0.001). Children less than five years old had a lower treatment success [AOR 0.64 (95 % CI, 0.52-0.80)] while the treatment success was higher in the 10-14 year-age group compared to 5-9 year olds [AOR 1.60 (95 % CI, 1.30-1.86)]. A considerable improvement was observed in treatment success in 2011-2012 (Table 5).
The proportion of childhood TB cases who died was 3 % (140 cases), constituting 12 % of the 1,202 deaths from TB in all age groups notified during the study period. The proportion of cases who died while on treatment declined from 7 % in 2003 to 2 % in 2012 (X 2 trend, P < 0.001), and was higher among under five-year age group [AOR 2.00 (95 % CI, 1.27-3.12)], and less in EPTB cases [AOR 0.58 (95 % CI, 0.36-0.95)]. The proportion of children who died in the less than 2-year-age group was three times higher than in the 2 year and above age groups [AOR 3.34 (95 % CI, 1.92-5.82)].
The proportions of cases lost-to-follow up and who died while on treatment were less in the 10-14 year olds than the younger age groups. Notable improvements in treatment success, loss-to-follow up, and mortality were observed by years of treatment, and the highest improvement in treatment outcome was observed in 2011-2012 (Table 5). The year of treatment was not mentioned for 15 cases for all age groups

Discussion
We found low CNRs of childhood TB in Sidama. Our data show that the CNR in children was lower than other reports from developing countries [15,16]. We analysed the CNRs based on age specific risk for developing TB, and challenges associated with its diagnosis [17,18]. The low CNR among young children (less than 5 years old) in our finding could be explained by underreporting, difficulties in diagnosis of childhood TB due to distinct clinical features of the disease in young children [18], poor access to diagnostic facilities such as chest radiography and culture, and poor access to skilled health personnel to diagnose the disease in children. Instituting better diagnostic services such as   NA not applicable CNRs = the mean case notification rates over a ten-year period except for the year of treatment category interferon-gamma release assays, culture, and nucleic acid amplification tests could help improve diagnosis and treatment of childhood TB [19]. The higher CNRs among older children (10-14 years) in our data could be because the clinical features of the disease are often similar to adults, and there is a better yield of sputum-smear examination among older children [17,20]. The higher CNRs in urban areas than in rural areas could be explained by better access to TB diagnostic facilities, better education and awareness of the parents and early health seeking, and this finding was in agreement with a study from Tanzania [21]. Likewise, in an earlier study, we found a higher CNR of TB among adults in urban areas [22].
The proportion of childhood TB cases in our data (13 %) was lower than the national estimate and report [2,6]. Our finding was in agreement with a previous report from the study area (13 %) [23], and higher than the global report (6 %) [2]. Other studies also report childhood TB constitutes 10-20 % of the disease burden in endemic areas [17].
From Ethiopia, one study from an urban setting reported a lower proportion (6.6 %) [24], and another study from a rural hospital [25] reported a higher proportion (46 %) of childhood TB than our finding. The reported higher proportion could be because the study was conducted in hospital setting which could have better diagnostic setup for children and could be due to referral cases who come from high burden areas due to poor access to TB diagnosis and treatment facilities. The lower proportion of childhood TB for the former study [24] could be due to underreporting or due to the lower burden of the disease. Nevertheless, both studies did not report the CNRs of childhood TB per 100,000 children unlike our report. Thus, the differences in the proportion of childhood TB from both studies and our finding could be partly explained by differences in study settings because we conducted the study in both urban-rural settings, in all DOTS providing health facilities, and we linked TB cases to their true home address to compute the case notifications to avoid under-or over reporting [22].
In our study, we noted that more girls than boys had smear-positive TB in the 10-14-year age group. This could reflect, at least in part, that girls enter puberty earlier and therefore might develop adult type of PTB earlier than boys [19,26]. In our data, the proportion of TB cases in children less than 5-year old was lower than in children 5-9-year old. In high burden setting with a broad-based population pyramid, the number of cases in children less than 5 and 10-14-year age groups are often higher than in the 5-9-year age group [4,24]. The lower CNR in the younger age group in our study suggests TB in the less than 5-year age group could be underdiagnosed. The proportion of smear-positive PTB among all pulmonary TB cases was higher than other reports from Ethiopia [24] and elsewhere [21,27]. Evidence shows that children usually have approximately 10-15 % smear-positive TB [4]. The difference in our finding is most likely because of underdiagnosing of smearnegative cases, particularly in younger children. This suggests concerted efforts to improve the diagnoses of TB in the younger age group.
In our data, the proportions of children successfully treated were higher among older age groups than among younger children, as has been shown from other studies from Ethiopia [23,24] and elsewhere [28]. Nevertheless, there is a growing concern on TB treatments in children, which are not "child-friendly" and TB drugs for children may not be in the correct dose, and the reports suggest "simple and safe treatments designed for children" [29]. The proportion of cases who died during treatment among children constituted about 12 % of the total deaths of TB in all age groups during the study period. However, the overall case fatality during treatment