Bulletin of the World Health Organization

Management of children exposed to Mycobacterium tuberculosis: a public health evaluation in West Java, Indonesia

Merrin E Rutherford a, Rovina Ruslami b, Melissa Anselmo a, Bachti Alisjahbana b, Neti Yulianti c, Hedy Sampurno c, Reinout van Crevel d & Philip C Hill a

a. Centre for International Health, Faculty of Medicine, University of Otago, PO Box 913, Dunedin, New Zealand.
b. Health Research Unit, University of Padjadjaran, Bandung, Indonesia.
c. Bandung Community Lung Clinic, Bandung, Indonesia.
d. Department of Internal Medicine, Radboud University, Nijmegen, Netherlands.

Correspondence to Merrin E Rutherford (e-mail: merrin.rutherford@gmail.com).

(Submitted: 31 January 2013 – Revised version received: 25 June 2013 – Accepted: 18 July 2013 – Published online: 10 September 2013.)

Bulletin of the World Health Organization 2013;91:932-941A. doi: http://dx.doi.org/10.2471/BLT.13.118414

Introduction

Current efforts to control childhood tuberculosis are failing, with over 100 000 children dying from the disease globally each year.1 Children under 5 years of age who are in contact with a patient with infectious tuberculosis are at an especially high risk of Mycobacterium tuberculosis infection and early progression to tuberculosis disease, which is characterized by the presence of symptoms.2 However, disease progression can be halted using preventive therapy, which has a reported efficacy of up to 93%.3 The World Health Organization (WHO) recommends that children who come into contact with an individual with infectious tuberculosis undergo child contact management, which includes screening for tuberculosis disease and, for those younger than  5 years, 6 months of isoniazid preventive therapy, even if disease is ruled out.4 This strategy can greatly reduce childhood tuberculosis, yet it is rarely practised in endemic settings.

Although earlier research has identified barriers to the success of child contact management programmes, a focus on single barriers has hindered the development of comprehensive programmes.59 Previously we presented a public health evaluation framework that involved situational, gap and options analyses and that could be used to identify problem areas and to develop appropriate multi-targeted solutions.10 In this paper we present the findings of the first two stages of a public health evaluation carried out using this framework in Bandung, West Java, Indonesia. Indonesia has the fifth highest tuberculosis case load in the world11 and reports indicate that a substantial proportion of tuberculosis patients in Java (i.e. 11 to 27%) are children.12 We hypothesized that there are widespread gaps between actual and ideal performance across a range of child contact management system parameters.

Methods

The study was conducted between April 2009 and February 2012 at a community lung clinic in Bandung that diagnoses approximately 1500 adults with pulmonary tuberculosis annually. Of these adults, 50% test positive on sputum smear analysis. The clinic has sputum smear and mycobacterial culture facilities, a pharmacy and a paediatric clinic. Screening of household contacts of sputum-smear-positive tuberculosis cases is encouraged but is not subsidized for children.

Data were obtained from administrative records, staff and adult tuberculosis patients and their households. Informed consent was gained from all participants. Ethical approval was given by the Lower South Regional Ethics Committee, New Zealand, and the ethics committee of the Medical Faculty, Padjadjaran University, Bandung.

Performance parameters and indicators

We developed a number of parameters for assessing the performance of the child contact management programme at the clinic, each of which was associated with a performance indicator (Table 1). The targets adopted for each indicator were derived from established programme indicator targets,13,17,18 the scientific literature14,15,19 and expert opinion.

Indicators of screening compliance and of the initiation of isoniazid preventive therapy in children (Table 1) were assessed in a cohort of consecutively diagnosed, sputum-smear-positive tuberculosis patients who had been living for the previous 3 months in the same house as at least one child aged 15 years or younger (cohort 1). Participants were informed about child contact management. A paediatric nurse used a standardized form to record each child’s attendance at the paediatric clinic in the 3 months after diagnosis of the index tuberculosis case and these records were cross-checked with the child’s clinical files. Information was also recorded on the outcomes of any diagnostic procedures and on the initiation of isoniazid preventive therapy in children younger than 5 years who were not diagnosed with tuberculosis.

Indicators of the accuracy of tuberculosis disease diagnosis were assessed in a second cohort of consecutively diagnosed, sputum-smear-positive tuberculosis patients (cohort 2). Patients had at least one child contact younger than 10 years. This age limit was selected to enrich the sample of children younger than 5 years. The tuberculosis patients were invited to bring the children for screening and, to encourage attendance, both screening and transportation costs were reimbursed. Screening included symptom evaluation and, in keeping with clinic policy, a tuberculin skin test (2TU PPD RT23, Biofarma, Bandung, Indonesia) that was read after 48 to 72 hours. Children who tested positive (i.e. induration ≥ 10 mm) underwent chest radiography and the radiograph was interpreted by the clinic’s paediatrician. Depending on clinical and radiological findings, antituberculosis medication was prescribed in accordance with the paediatrician’s recommendations. These children were also evaluated using the Indonesian Paediatric Scoring System, which is the clinic’s primary screening and diagnostic tool; it takes into account details of the tuberculosis case contact, symptoms, tuberculin skin test results and chest radiograph findings.16 The children’s chest radiographs were also evaluated by an external paediatric tuberculosis expert and classified according to whether or not they indicated the presence of tuberculosis disease. Two diagnostic accuracy indicators were employed: agreement between the attending paediatrician and the external paediatrician and the proportion of children with a latent tuberculosis infection (i.e. a positive tuberculin skin test result but no clinical or radiological evidence of tuberculosis disease) who were diagnosed with disease using the Indonesian Paediatric Scoring System.

The indicator of adherence to isoniazid preventive therapy was assessed in children younger than 5 years from cohort 2 who were eligible for therapy according to WHO recommendations.4 Prescriptions for therapy were provided exclusively at the clinic and patients had to pay for the medications themselves, as per local practice. The number of prescriptions collected for each child was recorded over a 6-month period. The adherence indicator was the proportion of children for whom three or more prescriptions were collected.

Indicators of primary caregivers’ knowledge and acceptance of tuberculosis disease screening and of isoniazid preventive therapy were assessed by conducting in-depth, unstructured interviews with consenting participants in their homes. Participants were selected to represent primary caregivers with child contacts who complied (n = 5) or did not comply (n = 5) with screening and those whose child contacts did (n = 5) or did not adhere (n = 7) to isoniazid preventive therapy. The proportion of primary caregivers who understood the need for child assessment was used as an indicator of screening knowledge. The absence of barriers to screening and preventive therapy was used an indicator of acceptance.

Indicators of medical staff’s knowledge of and attitude towards child contact management were assessed using a prepiloted, self-administered questionnaire. The questions, which were adapted from an existing health-care worker assessment tool,20 assessed staff’s attitudes and evaluated their knowledge of the provision of information on screening, current child contact management policy, the tuberculosis diagnostic process and treatment options. The proportion of staff who answered at least 75% of questions on knowledge and attitude correctly was used as an indicator of these parameters. Indicators of medical staff’s acceptance of tuberculosis disease screening and of isoniazid preventive therapy were assessed using a semi-structured interview. Staff were selected to represent a cross-section of doctors (n = 5) and nurses (n = 5) at the clinic. The absence of barriers to screening and preventive therapy was used as an indicator of acceptance.

To obtain an indication of the cost of child contact management, we gathered data on the costs associated with routine screening for all child contacts who presented to the clinic. Costs included a consultation fee, a registration fee and payments for the tuberculin skin test and the chest radiograph. For households in cohort 1, we asked index cases about the household’s monthly income and the cost of travelling to the clinic with their children. The total cost of screening was the cost of transport plus the cost of routine screening. It was expressed as a percentage of the household’s monthly income. The cost indicator was the proportion of households whose screening costs exceeded 10% of monthly household income. This measure has been used previously as a threshold for catastrophic expenditure on health.15

Indicators of the availability and quality of medications were evaluated by determining how many days the clinic was without a supply of isoniazid suitable for children in a 3-month period and by examining the credentials of the clinic’s isoniazid supplier.

Data analysis

On the basis of previous experience, we estimated that, for cohort 1, 25% (95% confidence interval, CI: 20.8–29.2) of 400 child contacts would attend screening, that 80 (80%; 95% CI: 71.2–88.8) of those attending would be eligible for isoniazid preventive therapy and that 70 (87.5%; 95% CI: 80.2–94.8) would commence treatment. For cohort 2, we estimated that 150 of 300 child contacts would be younger than 5 years, that 120 (80%) of the 150 would be commenced on isoniazid preventive therapy, and that three or more prescriptions would be collected for 90 (75%; 95% CI: 67.3–82.8) of the children receiving isoniazid.

All quantitative data were double-entered into Microsoft Access databases (Microsoft Corporation, Redmond, United States of America) and verified. Summary statistics with 95% CIs were presented where appropriate. Data from all investigations were used to evaluate performance indicators for each parameter. Each indicator was compared against a desired target to derive a value for the gap in the performance of the child contact management system. The gap was expressed as the absolute difference between the result observed for each indicator and its corresponding target. Interobserver agreement on interpretation of the chest radiographs was assessed using the kappa statistic. All quantitative data analyses were conducted using Stata version 11.0 (StataCorp. LP, College Station, USA). All qualitative data were digitally recorded, transcribed and translated to English before being analysed using thematic coding.

Results

Of 410 eligible tuberculosis patients, 242 were recruited to cohort 1 and interviewed. These patients had 437 child contacts who were eligible for screening (Fig. 1, available at: http://www.who.int/bulletin/volumes/91/12/13-118414). The median age of the tuberculosis patients was 31 years and 52% were male. For child contacts, the median age was 7 years and 53% were male. Overall, 34 of the 437 (7.8%) child contacts returned to the study clinic for screening within 3 months of the adult patient’s diagnosis, which gave a gap of 82% between the observed and target performance (Table 1). Sixteen of the 34 (47%) screened children received antituberculosis medication, while 6 of 15 children (40%) younger than 5 years who were eligible for isoniazid preventive therapy actually received it. This resulted in a gap of 50% between the observed and target performance for the initiation of preventive therapy.

Fig. 1. Participants recruited to assess disease screening of child contacts of adult tuberculosis patients, community lung clinic, Bandung, Indonesia, 2009–2012
Fig. 1. Participants recruited to assess disease screening of child contacts of adult tuberculosis patients, community lung clinic, Bandung, Indonesia, 2009–2012

Of 329 eligible tuberculosis patients, 210 were recruited to cohort 2 and interviewed. They had 318 child contacts who were eligible and recruited for screening (Fig. 2, available at: http://www.who.int/bulletin/volumes/91/12/13-118414). The median age of the tuberculosis patients was 31 years and 47% were male. For child contacts, the median age was 5 years and 49% were male. Ninety-two of the 318 (29%) children had symptoms suggestive of tuberculosis disease and 159 (59%) tested positive on the tuberculin skin test. Of these children who tested positive, 102 (64%) had chest radiographs that were interpreted as showing tuberculosis disease by the clinic’s paediatrician. The chest radiographs of 118 child contacts were also evaluated by the external paediatrician: tuberculosis was identified in 59% and 4% of these radiographs by the clinic’s paediatrician and the external paediatrician, respectively. The kappa value for interobserver agreement was 6.0%, which gave a gap of 64% for this indicator (Table 1). The proportion of children with a latent tuberculosis infection who were diagnosed with tuberculosis disease using the Indonesian Paediatric Scoring System was 100% (i.e. 41 of 41 children), which gave a gap of 100% for this indicator.

Fig. 2. Participants recruited to assess adherence to isoniazid preventive therapy in children younger than 5 years in contact with adult tuberculosis patients, community lung clinic, Bandung, Indonesia, 2009–2012
Fig. 2. Participants recruited to assess adherence to isoniazid preventive therapy in children younger than 5 years in contact with adult tuberculosis patients, community lung clinic, Bandung, Indonesia, 2009–2012

In total, 82 child contacts younger than 5 years from cohort 2 were prescribed isoniazid preventive therapy. Three or more prescriptions were collected from the clinic for 21 of the 82 (25.6%, 95% CI: 16.5–35.5). This resulted in a gap of 54% between the observed and target performance for adherence to isoniazid preventive therapy.

Although primary caregivers had adequate knowledge of child contact screening, the majority were unaware of the existence of isoniazid preventive therapy (Table 1). Interviews with primary caregivers revealed four barriers to the acceptance of screening: (i) difficulty accessing the clinic for initial patient diagnosis; (ii) problems with health-care workers and their appreciation of the effect of the disease; (iii) inappropriate health behaviour; and (iv) difficulty accessing screening related to transport, cost and the time needed (Table 2). In addition, interviews with caregivers whose children received isoniazid preventive therapy revealed four barriers to the acceptance of therapy: (i) difficulty accessing treatment, such as travel costs and time; (ii) medication issues, such as the ease of administration and experience of side effects; (iii) patients’ experience of the disease and health services, such as waiting times at the clinic; and (iv) patients’ knowledge and beliefs, such as a reluctance to treat healthy children. Details have been reported previously.21

The questionnaire on clinical staff’s knowledge of and attitudes towards child contact management was completed by 22 of 25 (88%) nurses approached and 12 of 15 (80%) doctors. The results are shown in Table 3. On knowledge, 50% of doctors and 18% of nurses answered at least 75% of questions correctly, which gave a gap of 46% between the observed and target performance (Table 1). On attitudes, 50% of doctors and 41% of nurses answered at least 75% of questions correctly, resulting in a gap of 31%. All staff agreed that child contacts should be screened but only 29% agreed that disease-free child contacts younger than 5 years should receive isoniazid preventive therapy. The development of multidrug-resistant tuberculosis due to isoniazid preventive therapy was a major concern.

Two main barriers to staff’s acceptance of screening were identified: doubts about the workability of the child contact management programme and doubts about the clinic’s capability (Table 4). Barriers to staff’s acceptance of isoniazid preventive therapy related to: (i) staff’s knowledge of treatment; (ii) compliance with treatment guidelines; and (iii) confusion about who is responsible for prescribing therapy.

The fixed costs of screening were 30 000 Indonesian rupiah (approximately 3 United States dollars, US$) per child for registration and consultation, US$ 3 for the tuberculin skin test and US$ 4 for chest radiography, which was performed if the tuberculin test was positive. The monthly household income was known for 149 of the 242 households in cohort 1 (62%). The median cost of screening was 9.8% (range: 0.8–80) of monthly household income and the cost exceeded 10% of that income for 50% of households.

The pharmaceutical company that supplied isoniazid to the clinic complied with good manufacturing practices and, in the 3 months preceding our evaluation, the pharmacy did not run out of isoniazid suitable for children.

Discussion

To our knowledge, this is the first comprehensive evaluation of a programme for managing the child contacts of tuberculosis patients. Using a public health framework, we identified and quantified gaps between current practice and desired targets. Moreover, our approach illustrated the importance of simultaneously evaluating all barriers. Although we observed considerable gaps in all measures of programme performance, all appear amenable to intervention.

That 92% of eligible children did not come for screening is of concern. Similar poor responses have been found elsewhere. In Malawi, screening compliance rates below 10% have been reported.5,8 In South Africa and Thailand, reported rates were under  3% and 52%, respectively.22,23 In contrast, the compliance rate in India for symptom-based screening conducted at the child’s home was 67%.24 Screening conducted in children’s homes using symptom-based screening, which is recommended by WHO,4 would be expected to improve the performance of child contact management systems.

Since childhood tuberculosis is rarely confirmed bacteriologically, chest radiographs are important for diagnosing disease.2527 However, their sensitivity and specificity in the diagnosis of tuberculosis are reportedly low.25,27 Furthermore, poor agreement between readers is common: in South Africa, the reported agreement on hilar lymphadenopathy between different observer pairs ranged from 5 to 55%27 and a similar study reported a kappa of 33%.28 In another study from the country, the proportion of children’s chest radiographs judged positive for tuberculosis by three reviewers ranged from 11 to 51%.29 While it is reasonable that the use of chest radiographs for diagnosis in child contact management programmes should be reviewed,25,30 there is, perhaps, as great a need for standardizing the training of readers.

One important finding of our study is that inappropriate use of the Indonesian Paediatric Scoring System resulted in disease being misdiagnosed in children with latent tuberculosis infections. Misdiagnosis was due to the high scores awarded for being a contact of a tuberculosis patient and for a positive result on the tuberculin skin test. Another Indonesian study found that 82% of child contacts with latent tuberculosis infections were diagnosed with tuberculosis disease using the scoring system.31 The performance of other scoring systems is variable: a comparison of nine systems in 1445 children found diagnostic yields ranging from 7 to 89%.29

Rates for the initiation of isoniazid preventive therapy in eligible children ranging from 1.3 to 26% have been reported in settings where tuberculosis is endemic.5,23,32,33 Such low rates are due to poor screening compliance. When the initiation of therapy was evaluated in children who had been screened, the reported rate was 50 to 84%.5,24,32,33 Similarly, in our study, it was 40%. However, as in our study, adherence to therapy was poor in settings where the disease is endemic (range: 15 to 76%).6,7,3437

In our study, poor staff knowledge of isoniazid preventive therapy and poor compliance with guidelines were found to be barriers to the initiation of therapy. Similar barriers have been reported in other countries: in India, health-care workers reported that unclear guidelines on child contact management inhibited their ability to implement therapy;38 in Thailand, doctors were reluctant to initiate therapy due to concerns about isoniazid toxicity and resistance;22 in Australia, doctors actively advised patients not to use isoniazid preventive therapy;34 in the United States of America, medical graduates did not believe isoniazid preventive therapy was protective against disease progression;39 and, in Malawi, health-care workers felt that screening child contacts by chest radiography was not worthwhile.33

As in other countries, we found that primary caregivers’ acceptance of the child contact management programme was hindered by barriers such as limited knowledge of isoniazid preventive therapy, difficulty accessing screening and treatment and reluctance to treat asymptomatic children.7,22,33,34,40

There is a lack of research on the cost of child contact management to households. However, the cost of tuberculosis disease often exceeds the 10% of monthly household income considered catastrophic41 and a figure as high as 89% of the household’s annual income has been reported.42 We found that the cost of screening child contacts alone exceeded the 10% threshold for many households in Bandung. A symptom-based approach to screening4 could reduce this cost considerably.

The continuous supply and consistent quality of antituberculosis drugs are key elements of the DOTS strategy.17 At the study clinic, the availability and quality of medications were adequate. In contrast, the quality and, to a lesser extent, the availability of antituberculosis drugs have been reported to be poor in other settings.43,44

The study has a number of limitations. Since it was carried out at a single institution, its external validity is a concern. Moreover, the characteristics of the child contacts recruited to cohort 1 may have differed from those of child contacts recruited to cohort 2, whose costs were reimbursed. It was not possible to document reasons for non-attendance for screening or poor adherence to therapy, although we did collect information to assess risk factors for these outcomes, which will be reported elsewhere. Some of our findings are based on data from few participants (e.g. few staff were interviewed). Also, since we used qualitative methods to identify barriers to child contact management, we could not quantify the effect of these barriers. Our investigation of the diagnostic accuracy of chest radiography in child contacts was hindered by the lack of microbiological confirmation of tuberculosis disease. Another limitation is that the medication audit was conducted retrospectively. Further, we judged medication quality on the basis of the quality of the supplies; however, medications could have been degraded by problems with storage at the clinic, such as lengthy storage.45 The only indirect costs for child contacts’ households we considered were for transportation to screening; a diagnosis of tuberculosis and its treatment would be associated with other indirect costs.46 Although testing for the human immunodeficiency virus was not carried out, infection is rare in the general population and in tuberculosis patients in Indonesia.4749

In conclusion, we used a public health framework to identify substantial gaps in the performance of a programme for managing the child contacts of tuberculosis patients. The next step is to consider ways of reducing these gaps. The different options available for addressing each gap could be combined using a set of weighted criteria to devise a new child contact management programme. Our public health framework could then be used to evaluate this programme and any further changes made to it. Since problems with child contact management are not unique to our study clinic or to Indonesia, the adoption of a similar evaluation framework could be useful in other settings.


Competing interests:

None declared.

References

Share