This study describes the characteristics of children with TB in Malaysia from 2013 to 2017 using the recommended NTP surveillance database. TB incidence among children is monitored according to age group, as younger children constitute a high-risk group for TB disease complications . The standard age disaggregation for WHO TB reporting among children is as follows: 0–4 years and 5–14 years . Other researchers have divided age groups into three categories: 0–4 years, 5–9 years, and 10–14 years [12, 14]. A few studies in countries with a low TB burden have used different cut-off points to define paediatric TB, such as being under 18 years [16, 26] or 0–16 years ; in such cases, inter-country comparison was unsuitable.
The proportion of registered TB cases among children in Malaysia was less than 5% compared to the total registered cases. The ratio of TB cases was much lower among the age group 0–4 years than among the older group. Regarding these findings, the national database is still far from the WHO benchmarks for a higher-quality database for children with TB disease . The most cases were in children aged 10–14 years, followed by those aged 0–4 years. In particular, the youngest age group has a 30 to 40% risk of developing PTB disease and a 10 to 20% of developing TB meningitis or miliary TB following Mtb infection. The risk of PTB disease was 2% for children aged 5–9 years and between 10 and 20% for those aged 10–14 years . Younger children have immature immune systems, while adolescents have greater social contact, leading to a greater risk of TB infection . The number of registered TB cases was low for children aged 0–4 years because there were more missing cases, even though Malaysia does not have a high TB burden. A known challenge for cases among children is accurately diagnosing TB because of the non-specific symptoms and paucibacillary. A study conducted among countries with a high TB burden found a low number of cases among the 0–4-year age group compared to the older age group, which indicates the possibility of missing cases among the younger age groups [11, 12, 30].
The proportion of registered TB cases among children in Malaysia can be broken down by the main ethnicities in Malaysia, including non-Malaysian citizens, who accounted for 13.5% of cases. Among immigrants, a higher proportion of TB patients originated from neighbouring countries such as the Philippines, Indonesia and Myanmar, countries with a high TB burden . Similar observations have been made among registered TB cases in countries with a low TB burden, such as Israel, Australia and Canada, where a substantial proportion of TB cases were immigrants from countries with a high TB burden [15, 16, 31].
The overall rate of treatment success among children in Malaysia was 90.1%, which achieved the WHO target of 90%, except for children under 5 years old. The Malaysian goal of a 95% treatment success rate in 2020 seems unachievable . The rates of treatment success in other countries varied because of the wide range of data source used, either from an NTP database or review of hospital-based records. Hospital-based data [30, 32] usually reflect a small number of TB cases compared to those from the NTP database . Moreover, hospital-based studies often showed a higher rate of severe TB disease  and thus a lower proportion of treatment success. Two kinds of databases are useful to monitor the outcomes of TB, whether success or unsuccessful treatment [9, 33, 34]. The MyTB version 2.1 database includes a large number of variables, such as place of treatment, BCG scars, treatment regime, directly observed therapy (DOT) for the intensive phase, and DOT supervisor. However, this database needs improvement regarding the documentation of household income because there are many missing values. Nevertheless, nine independent variables were selected as determinants of treatment success among children in Malaysia.
The age of children independently determines their likelihood of treatment success when the analysis includes a continuous variable. The findings differed from previous studies in which the analysis employed a categorical variable [9, 14]. However, the conclusions of this study and other studies were similar in that the older age group had a higher likelihood of treatment success than the 0–4-year age group [14, 35]. In this study, the categorical variable of age exhibited a significant interaction with other determinants, such as TB disease diagnosis and BCG scars, which led to biased estimates.
In terms of citizenship as a determinant of treatment success, Malaysian citizens had a higher likelihood of treatment success than non-Malaysians. This finding was in line with the national TB policy stating that TB treatment is free of charge. TB treatment in Malaysia utilizes a holistic approach including case detection, a DOT programme and TB counselling, effective follow-up and a BCG vaccination policy for newborns . On the other hand, the non-Malaysian group included children from countries with a high TB burden and low BCG coverage. Although TB treatment is free in Malaysia, the initial evaluation procedure before the confirmation of TB disease is expensive. The associated financial burden may prevent non-Malaysian families from visiting the clinic at an early stage of health changes.
TB was diagnosed in children in Malaysia primarily by sputum-negative PTB, followed by sputum-positive PTB and EPTB. The pattern of TB diagnosis among children was similar to that in previous studies [9, 14]. Studies from hospital-based settings report a higher percentage of EPTB than PTB and a higher number of more severe TB types, such as TB meningitis [33, 36]. Younger age groups have difficulty producing sputum samples; hence, sputum examination results rarely become positive for younger groups compared with older age groups . Children with sputum-positive PTB and EPTB have a low likelihood of treatment success; in fact, children in EPTB groups may present with TB meningitis, a severe type of TB disease . Children with sputum-positive PTB presented with a higher load of Mtb in their bodies, which triggered an excessive immune reaction that contributed to the probability of unsuccessful treatment among adolescents [9, 14, 29]. In addition, chest X-ray findings were associated with treatment success; children with chest X-ray results of advanced lesion were observed to have a low success rate compared with the rate for children whose chest X-ray showed no lesion or minimal lesion. Previous studies conducted in Ethiopia and Pakistan excluded chest X-ray results due to the availability of variables in their NTP database [9, 12, 14]. Recent local studies found an association of chest X-ray findings with unfavourable outcomes; however, these studies were conducted among all age groups [38, 39].
TB disease is associated with HIV co-infection because HIV alters the human immune system. HIV co-infection accounted for only 1.4% of the total TB cases among children in this study. Malaysia still has a low occurrence of HIV co-infection, but those who had HIV co-infection had low odds for TB treatment success. This finding was similar to the observations in other countries with a higher prevalence of HIV co-infection [14, 30, 35].
BCG vaccination provides protection from TB meningitis and miliary TB among the age group 0–4 years and some protection from PTB until the age of 14 years . Considering this fact, 86.5% of our registered TB cases among children had BCG scars. Children without BCG scars were more likely to be non-Malaysian citizens (304/479, 63.5%) than Malaysian citizens (174/3071, 5.7%). Children with BCG scars had a higher likelihood of treatment success than those without BCG scars in this study, possibly because they have been protected from severe TB. However, previous research in China did not find a significant association between BCG scars and TB treatment success with the utilization of hospital databases . Many other studies in Ethiopia and Pakistan did not include the presence of BCG scars as a determinant of treatment success due to data availability in their NTP system [9, 12, 14, 35].
Place of treatment was another predictor of TB treatment success Children who were treated at a public hospital were less likely to succeed than those treated in a public clinic. Thus, cases in the hospital were more severe and difficult to manage. A local study using a Malaysian cohort in 2012 observed that a patient treated in a public hospital or clinic had a higher likelihood of unfavourable outcomes than those treated in a private facility; however, that study included all age groups . Studies conducted in countries with a high TB burden did not include the place of treatment in their model; hence, comparisons cannot be made. Health facilities providing TB treatment should ensure that their services are patient-friendly to promote treatment success. In addition, early detection during active case evaluation, especially among younger age groups, may reduce the number of severe TB cases that require hospital management.
The place of residence for children showed a non-significant result in the final model. Thus, the country’s health services did not differ in terms of the accessibility and quality of TB care between urban and rural areas. A study conducted in Pakistan, with a high TB burden, observed that children from a rural area had higher odds of unsuccessful treatment , possibly because children in urban areas have greater access to health facilities, are better educated and are more confident taking modern medicine.
Among the socio-demographic determinants of treatment success or unsuccessful treatment outcomes, no association with children’s sex was found in previous studies [9, 11]. Similar findings were reported in our research, concluding that TB treatment success among children did not differ by sex.
Limitations and strength of the study
The study used secondary data that depended upon the availability of relevant parameters. A few variables could not be included because of the large amount of missing data and lack of valid definitions, such as those for household income, number of household dependents and education level. Thus, many predictors, such as economic status, parental education level and BMI status of the child, cannot be assessed in this study. Second, this study used a cross-sectional design; therefore, a causal relationship between the dependent and independent variables could not be established. The main strength of this study was the utilization of 5 years of data on TB case registration among children in Malaysia. The sample size in this research was adequate to generate statistically meaningful findings. Although secondary data analysis might have led to more missing values, only 1.6% of the values in the selected dataset were missing. The use of secondary data depends upon the percentage of missing data and the standard definition in the manual.