High burden of pulmonary tuberculosis and missed opportunity to initiate treatment among children in Kampala, Uganda

Background There is uncertainty about the actual burden of childhood TB in Uganda, but underestimation is acknowledged. We aimed at determining prevalence, factors associated with PTB among children attending PHC facilities in Kampala. Methods This was a cross-sectional study of 255 children, with presumed TB, attending six health facilities in Kampala, Uganda, in March 2015. Socio-demographic, clinical, and laboratory data were collected using a questionnaire. TB was diagnosed using “Desk Guide” algorithms. Sputum based on ZN/FM and/or Gene-Xpert. Logistic regression was used to assess associations with outcomes. Results Overall, prevalence of PTB 13.7 % (2.6 – 24.8). Among HIV-positive, the prevalence of PTB was 41.7%, while among malnourished children, 21.7% and contacts, 89.3%. The factors that influenced PTB included: tobacco smoker at home (OR = 1.6, 95 % CI: 1.07 – 6.86), stunting (OR = 2.2, 95 % CI: 1.01 – 4.15). Only 5.3% of the smear-negative TB children and 81.3% of the smear-positive children were initiated on treatment within a month of diagnosis. Conclusion Clinical TB among children is underdiagnosed and undertreated. There is a need for more sensitive and specific diagnostic tests, need ways to disseminate and promote uptake of standardized clinical algorithms. Also, contact TB tracing should be strengthened so that such cases can be actively detected even at community level.


Introduction
Although curable, tuberculosis (TB) continues to cause high morbidity and mortality among children below 15 years 1-3 . In 2019 alone, there was an estimated 1.2 million children suffering from TB, and a total of over 23,000 children died from TB 4 . Despite this high burden, child-hood TB has been neglected [5][6][7][8] in Uganda, one of the 30 countries identified by WHO as having the highest burdens of TB 4,9,10 . Although the majority of childhood pulmonary tuberculosis (PTB) is smear-negative 11,12 , in Uganda, its diagnosis is mainly based on sputum AFB smear microscopy or GeneXpert MTB/RIF with limited clinical diagnosis 13,14 . Children under 15 represent less than 10% of AFB smear-positive patients notified to the national TB program 10,14 . Clinical algorithms for diagnosing of pediatric TB are not widespread and their impact has not been assessed 15 . Therefore, the burden of disease in children is underestimated 16,17 . Our study aimed to determine the burden and predictors of TB, and how a standardized approach to clinical diagnosis (using the Union Pediatric TB Desk Guide) might impact case notification and treatment rates among older children in Uganda.

Study sites and setting
A cross-sectional study was conducted in six primary health facilities within Kampala district. Kampala is the capital city and is in central Uganda. The city has several slum areas served by these clinics. The health directorate of Kampala Capital City Authority (KCCA) provides health services to the population through 8 health facilities. These facilities receive patients directly from the community and offer outpatient and maternity inpatient services. Diagnosis of pulmonary TB is based on sputum analysis using ZN, FM, or GeneXpert MTB/RIF (Xpert) and follows the guidelines laid out in the National TB and Leprosy Program 18 . Occasionally, a clinical diagnosis of PTB is made based on clinical and radiological findings for the participants who had negative sputum or were unable to expectorate 18 . Routinely, the nurses usually treat TB infected children at the TB clinics, and only complicated cases are referred to clinicians.

Sample size and participants
Based on childhood TB prevalence of 10.7% from the national TB notification rates for 2014, a target power of 80% and a precision of 0.05, we estimated a sample size of 147 children with presumed TB. We applied a design effect of 2 to yield 294 participants to account for clustering within health centers. The study consecutively enrolled children 5 to 15 years who attended outpatient departments at the KCCA clinics between February and April 2015 and had clinical features suggesting pulmonary TB, as elaborated in the intensified case finding guidelines 19 . The total number of children who attended the outpatient department for the study period was obtained by reviewing the facility records. We excluded very sick children who warranted immediate referral and those below 5 years of age because clinicians routinely refer them out of the facility for further evaluation because they are usually unable to expectorate. Participants were screened from the OPD using the Intensified TB case finding form. All the eligible presumptive TB patients were then enrolled in the study.

Study exposure variables and measurements
After pre-testing, an interviewer administered a semi-structured questionnaire to capture age, sex, nationality, religion, tribe, participant and caretaker education level, and nationality. House-hold characteristics including monthly income, persons at home, number of rooms and ventilation, energy source for cooking and cooking area. Clinical history including BCG vaccination, contact with a confirmed TB patient, IPT, ART status, chronic cough, haemoptysis, weight loss, frequent evening fever and drenching night sweats.

HIV diagnosis
Clinics performed routine HIV counselling and testing using whole blood assays according to a parallel rapid testing algorithm recommended by the Uganda Ministry of Health 20 .

Anthropometric measurements
Weight was determined using calibrated SECCA scales; and height was measured using a non-stretching meter that was placed and calibrated against a vertical wall. Stunting, underweight, and low BMI were defined as Z-score <-2SD for height-for-age, weight-for-age, and BMI-forage respectively. These were calculated using ENA® (Emergency Nutritional Assessment) and AnthroPlus® developed by WHO. The children's nutritional status was categorized into good or bad basing on whether they had stunting and wasting or not.

Outcome variables
Microbiological diagnosis of TB followed Uganda NTLP guidelines 21 , which follow WHO recommendations on sputum smear microscopy and Xpert 22 . It was based on having the spot, morning or both sputum smears positive for by microscopy or Xpert. Sputum examination analysis was done at the KCCA health facility laboratories, however two of the facilities lacked Xpert facilities. These laboratories participate in an external quality assessment program overseen by the Uganda National TB Reference Laboratory. A clinical TB diagnosis was retrospectively made from the data collected basing on the "Desk-Guide for Diagnosis and Management of Tuberculosis in Children" guidelines 18 : having history of contact and clinical features suggestive of PTB. After a month, TB registers were reviewed in each of the facilities to determine the number of the children diagnosed in the study that were initiated on anti-TB medicine.

Data management and analysis
Data was double-entered Epidata, validated and exported to STATA version 12.0 for analysis, adjusted for clustering at the level of the health facility. Continuous variables were summarized using measures of central tendency. Categorical variables were summarized using frequencies, proportions, and percentages. Bivariate logistic regression analysis was performed and all variables with a P-value of ≤ 0.2 were entered in a multivariate logistic regression model to assess their association with the outcomes. Interaction between the variables was assessed using the Chunk test. Confounding was assessed using a difference of ≥ 10% between the crude and adjusted odds ratio for the variables.

Ethical Considerations
Institutional approval for this study was obtained from the Makerere University School of Medicine Research and Ethics Committee. Written informed consent was obtained from the mothers/ caregivers. Additional assent was sought from children 8 years and above. Permission was sought from the KCCA directorate of public health to conduct the study within the KCCA health facilities.

Results
There were a total of 1975 children aged 5 to 14 years who attended the outpatient department during the study period. Of these, 255 had features suggestive of Tuberculosis. The patient flowchart shows details in figure 1.

Discussion
In this study we determined the burden and predictors of TB, and how a standardized approach to clinical diagnosis (using the Union Pediatric TB Desk Guide) could impact case notification and treatment among older children in Uganda. We found that 13.7 % of the children we enrolled into our study (with presumed TB) had clinical or smear positive TB. The burden of TB was higher among the malnourished, HIV positive and children with history of contact with an adult TB patient. Previous studies have reported lower prevalence of TB among children 23 2, 9 24 . However, the lower prevalence was not among children with presumed TB. Kampala, the location for our study, is over-crowded yet crowding has been shown to increase risk of TB infection 25, 26 . The lower prevalence in national notification figures could be due to poor recording and reporting of routine pediatric TB cases in the national TB registries which has a lot of missing information and mainly bases on smear results 10,27 . These findings may imply that the burden of childhood TB has been undermined. Our findings showed that more than half of the children with TB were missed by the routine diagnostic practices, more so children with smear-negative TB. The high rate of missed TB diagnosis can be attributed to the difficulty in diagnosis of TB, yet in peripheral facilities patient management including diagnosis and treatment is routinely done by lower medical cadres including clinical officers and nurses. Also, the clinical algorithms for guiding the diagnosis of TB are not routinely used during the management of these children. Lack of capacity to diagnose TB had been previously highlighted as a barrier to diagnosis of TB in Uganda 27 . Furthermore, there is a significant loss to follow-up of patients with TB. In our study, only 40% of children diagnosed with TB being started on treatment within a month following diagnosis. Household overcrowding, indoor air pollution and malnutrition were significantly associated with TB spread among children in our study. Malnutrition impairs the body's immunity making the child more susceptible to TB 25 on the other hand, a child who develops tuberculosis is more likely to develop malnutrition 5 . With high rates of malnutrition among the children in Uganda, TB is likely to continue being a major cause of morbidity. Indoor air pollution has been associated with development of PTB since pollutants impair mucosal integrity and lower surface immunity 5,25 . This is in line with the findings from our study since having a tobacco smoker in the household was associated with a 1.6 increase in the odds of developing PTB 5 . Our study had several strengths. Frist, while many studies among children report one form of TB, in our study we report both laboratories confirmed and clinically diagnosed TB. In addition, we evaluated TB diagnosis and management using a standard guideline. Moreover, we did our evaluation retrospectively when the actual patient diagnosis was already done. This way, our presence did not influence how clinicians diagnose the patients. The retrospective evaluation enabled us to evaluate how well TB is diagnosed but also gave us time to determine the proportion of children diagnosed with TB that were started on treatment. Our study enrolled participants from six public health facilities, which makes our sample representative of the general population. Also, we employed robust analysis methods and controlled for several factors that could confound the association between TB and our main predictors. However, our study had some limitations. Children below five years were not included in the study. This is a special population with regards to TB. Additionally, the study used hospital participants who were already presumed to have TB; therefore, they were more likely to have TB than the general population. This population is more likely to over-estimate the burden of TB. These shortcomings could have introduced a selection bias. For some of the characteristics like monthly income, and BCG vaccination, the researcher had to rely on self-report. This may have introduced an information bias in the study. In the study sputum cultures, the gold standard for TB diagnosis, were not done. Therefore, it is possible that some of the TB cases were missed.

Conclusions
Clinical TB among older children is underdiagnosed and undertreated. There is pending availability of more sensitive and specific diagnostic tests, need ways to disseminate and promote uptake of standardized clinical algorithms. There is a need to strengthen contact TB tracing so that such cases can be actively detected even at community level. Larger studies, employing gold standard diagnostic means is recommended to further study the pediatric TB burden, especially among contacts to TB patients, children below 5 years, and the malnourished children.