Patients and methods

Children (less than 15 years of age) in Wales with tuberculosis who were diagnosed between 1986 and 1992 inclusive were identified retrospectively from statutory notifications, laboratory data held by the Mycobacterium Reference Unit of the Public Health Laboratory Service, Cardiff, and by written inquiry to all paediatricians and chest doctors in Wales. After obtaining written consent from the relevant consultant, patient records and radiographs were retrieved and a proforma completed for each patient. Data were sought on: age at diagnosis; sex; ethnic origin; mode of presentation; basis of diagnosis; sputum smear and tissue culture results; BCG immunisation status; and details of treatment received. Presenting radiographs were reviewed by a consultant radiologist (HA). Abnormal chest radiographs were classified as (a) cavitary tuberculosis, (b) non-cavitary consistent with tuberculosis, (c) non-cavitary unusual for tuberculosis, and (d) other abnormalities. Chest radiographs showing single or multiple cavities with patchy or nodular infiltrates, with or without calcification, were classed as cavitary tuberculosis. Opacities (linear or confluent homogenous shadowing, or both), mainly in the upper zone, with or without calcification, hilar lymph nodes, and/or pleural effusion were classed as non-cavitary consistent with tuberculosis. Characteristics and distributions of shadowing other than those described (for example, lobar consolidation) were grouped as non-cavitary unusual for tuberculosis. All other abnormalities on the chest radiograph (for example, mediastinal widening) were classed as other.

Patients were classified as confirmed, probable, or possible tuberculosis, or tuberculosis chemoprophylaxis (see ). A positive tuberculin test was defined as an induration of 5 mm or greater (transverse diameter to 0.1 ml of 1:1000 tuberculin PPD solution by intradermal injection read at 48–72 hours) for a Mantoux test, or a grade 2 reaction or greater (read at 7 days) for a Heaf test.11

Appropriateness of management was assessed by the following criteria based on the British Thoracic Society recommendations10: prompt case notification; joint management by a paediatrician and chest doctor; and treatment with a recommended chemotherapy or chemoprophylaxis regimen (table1).

Results

EPIDEMIOLOGY AND CLINICAL FEATURES

Sixty three patients were identified, of whom 48 met the case definition for tuberculosis (eight confirmed, 34 probable, and six possible) and 10 for tuberculosis chemoprophylaxis. Of the remainder, three had opportunist mycobacterial infection and two had insufficient evidence to substantiate a diagnosis of tuberculosis. Thirty one patients presented with symptoms, but the remaining 17 (35%) patients were detected by tuberculosis contact screening. None of the patients with tuberculosis was detected as a result of the routine school BCG immunisation programme, although one 13 year old boy (whose father was later found to have pulmonary tuberculosis) was referred by the school programme for tuberculosis chemoprophylaxis.

The mean annual tuberculosis incidence was 1.2/100 000, with a peak annual incidence of 1.7/100 000 in 1991 and 1992. At the time of the 1991 census, Wales had a population of 2.9 million, of whom 193 000 were aged 0–4 years, 187 000 were aged 5–9 years, and 178 000 were aged 10–14 years. Of these, 97.5% were white, 1% originated from the Indian subcontinent, 0.1% were Afro-Caribbean, and 1.4% were of other ethnic origins (black African, black other, Chinese, other Asian). Table 2 gives the demographic characteristics and clinical presentations of the patients.

Pulmonary disease was the most common presentation. Of the 10 patients examined for acid fast bacilli by sputum smear, however, only two were positive. Other specimens examined bacteriologically included gastric aspirates (four), lymph nodes (six), early morning urine (three), cough swab (one), and cerebrospinal fluid (three). Of these, six were positive on culture. In 24 patients no culture was performed. Tuberculin testing was performed in 44 (92%) of the 48 patients. Twenty four (50%) patients had a Mantoux test and 22 (46%) had a Heaf test, including two patients who had both tests.

Documentation of the tuberculin test results was poor. The strength of tuberculin used was recorded for only nine Mantoux tests, and the diameter of induration was precisely measured in only seven of the 19 positive Mantoux tests, the remaining results being reported as “weakly” or “strongly” positive. The grade of reaction was recorded in only 12 of the 19 positive Heaf tests. BCG immunisation status was not recorded in 25% of patients. Two (20%) patients of ethnic minority origin and four (11%) white patients had been immunised with BCG, the latter after contact with patients with tuberculosis.

PATIENT MANAGEMENT

The patients were mostly managed by paediatricians alone (69%) or chest doctors alone (25%), but rarely by both (6%). Paediatricians were more likely to undertake microbiological investigations (17 of 33 (52%) patients) and to perform Mantoux tests (19 of 33 (58%) patients) than chest doctors (one of 12 (8%) patients and four of 12 (33%) patients, respectively). A similar proportion of symptomatic patients (73%) and patients detected by contact tracing (67%) were managed by paediatricians. Seventeen (35%) patients with tuberculosis were not notified, but there was no difference in the proportion notified by paediatricians or chest doctors. Similarly, the likelihood of notification was not influenced by the source of the patient (acute illness or contact tracing), the site of disease, or the geographical area.

Only five (10%) patients were managed according to British Thoracic Society guidelines. All of these were patients with pulmonary tuberculosis treated with a two drug (rifampicin and isoniazid) regimen (table 3). Eighteen (38%) patients were undertreated either because they were given an inappropriate drug combination or because the duration of treatment was inadequate. This included 11 patients unnecessarily treated with ethambutol. Many patients were over treated; 23% of patients with pulmonary disease were treated for more than nine months and 8% of non-pulmonary patients were treated for more than six months. The longest duration of treatment was 32 months in a patient with tuberculous meningitis. Four patients were inappropriately treated with a quadruple drug regimen (two with pulmonary disease, one with lymph node disease, and one with tuberculous meningitis). Twenty five (52%) patients received unnecessary treatment with pyridoxine. Three patients required steroids, two for tuberculous meningitis and one for phlyctenular conjunctivitis. Surgery was performed in 10 patients, mostly for the removal of lymph nodes. No death occurred. All 10 patients treated by chemoprophylaxis received two drugs (isoniazid and rifampicin); nine were over treated (more than six months of treatment) and one was under treated (two months only). No patient required a change of treatment as a result of an adverse drug reaction.

Paediatricians were less likely than chest doctors to under treat patients with tuberculosis (eight of 33 v nine of 12; odds ratio 0.11 (95% confidence interval 0.02 to 0.59, p < 0.01)), although nine of the 10 patients lost to follow up were under the care of a paediatrician. There was no significant difference with respect to the proportion of over treated patients or the inappropriate use of ethambutol.

Discussion

This study found that the management of children with tuberculosis in Wales falls short of the recommendations of the British Thoracic Society.10 There was an undernotification of cases, infrequent shared care between paediatricians and chest doctors, and inappropriate treatment, both with respect to the choice of drugs and the duration of treatment. Childhood tuberculosis remains uncommon in Wales. There is evidence, however, of an increase in childhood tuberculosis, particularly in urban areas and among new immigrants and refugees.4 5 This, in turn, implies a high rate of recent transmission of tuberculosis. Diagnostic awareness and familiarity with recommendations on the management of tuberculosis in children are therefore particularly important.

Undernotification of tuberculosis is well recognised and estimated to be around 20–30%.12 13 In this study, there was a 35% undernotification of tuberculosis in children. Childhood tuberculosis is almost invariably the result of recent infection from an adult with tuberculosis. Undernotification in children thus has significant implications for tuberculosis control as it may lead to an infective case of tuberculosis in an adult remaining unrecognised. Many cases of childhood tuberculosis could probably be prevented if case detection of adults and subsequent contact tracing were carried out more promptly and efficiently.14 15

The study also found that less than a quarter of patients with tuberculosis from ethnic minority groups had been immunised with BCG, indicating further lost opportunities for prevention. As far as we could ascertain, the primary failure was with the neonatal BCG programme and not with the immigration screening procedure. Similarly poor rates of BCG immunisation have been noted in other studies of children with tuberculosis.8 16 17 Neonatal BCG immunisation is recommended in the UK for all infants born to immigrants from countries with a high prevalence of tuberculosis.11 BCG immunisation coverage is not routinely monitored in the UK, however, unlike the remainder of the childhood immunisation programme, and at least one study has shown considerable room for improvement.18 19

The tuberculin skin test is an important diagnostic test for tuberculosis in children. Our study found that tuberculin tests were often not properly conducted (the Mantoux test reaction was only accurately measured in 40% of tests) or the results were not clearly recorded. Isolation of mycobacteria from children is difficult as they often do not produce sputum and the sensitivity of diagnostic tests is low because the number of organisms that cause the disease in children is generally much lower than in adults. Nevertheless, it is desirable to seek culture confirmation of tuberculosis if at all possible. Attempts at isolating mycobacteria were made in only half of the patients in this study, of whom a third were culture positive.

Although there is little research on tuberculosis chemotherapy in children, the same basic principles as adults are considered to apply. There is, in fact, considerable consensus on the treatment of childhood tuberculosis and British,10 American,20 21and Canadian22 recommendations on chemotherapy for pulmonary tuberculosis in children are now identical. In this study, one third of the patients were over- treated, most commonly with regard to the duration of treatment or the number of drugs against tuberculosis used. None of the chest doctors used the triple drug regimen of isoniazid, rifampicin, and pyrazinamide. Thirteen patients received ethambutol, although this is only recommended in exceptional circumstances (table 1) because of the difficulties of detecting ocular side effects in children. Pyridoxine (to prevent the peripheral neuropathy associated with isoniazid) was also used in many patients, although this is only routinely recommended for patients at high risk such as those with diabetes, alcoholics, and those with chronic renal failure.10

No published British guidelines on the management of childhood tuberculosis were available before 1990. It is possible that the issue of guidelines may have led to an improvement in practice during the second half of our study. The findings suggest, however, that the British Thoracic Society recommendations on the management of childhood tuberculosis should be more widely known, particularly among paediatricians. To monitor and improve the management of childhood tuberculosis in Wales, we have disseminated the results of this study widely among paediatricians, chest doctors, and specialist tuberculosis nurses. New British Thoracic Society guidelines have been published on the control and prevention of tuberculosis23 and on notification.24 Guidelines for treating children will be published soon. Tuberculosis is now included in the Welsh paediatric surveillance scheme, which is based on monthly returns on a small range of uncommon childhood disorders made by all paediatricians, and a similar scheme has been established with chest doctors. This will enable us to identify any improvements in current practice that occur as a result.

Shared care between paediatricians and chest doctors should ensure that adequate support is available from the full chest clinic team throughout the duration of the child’s treatment. This would help overcome many of the deficiencies identified by our study and should improve the management of tuberculosis in children.