Secondary Logo

Journal Logo


Millership, Sally E. MD*; Anderson, Charlotte MSc; Cummins, Amelia J. MB BS*; Bracebridge, Samantha MB BS; Abubakar, Ibrahim PhD

The Pediatric Infectious Disease Journal: October 2009 - Volume 28 - Issue 10 - p 915-916
doi: 10.1097/INF.0b013e3181a3d289
Brief Reports

We undertook a systematic literature and database review for reports of nosocomial exposures of infants less than 24 months of age to tuberculosis. We found 7 instances of transmission among 4867 babies in 26 reports (19 published, 7 from a national database for reporting such events).

From the *Essex Health Protection Unit, Health Protection Agency, Witham, Essex; †Tuberculosis Section, Department of Respiratory Diseases, Centre for Infections, Health Protection Agency, London; and ‡Regional Epidemiology Unit, Health Protection Agency, Institute of Public Health, University Forvie Site, Cambridge, United Kingdom.

Accepted for publication March 4, 2009.

Address for correspondence: Sally E. Millership, MD, Essex Health Protection Unit, Health Protection Agency, 8 Collingwood Road, Witham, Essex CM8 2TT. E-mail:

The risk to infants in hospital nurseries of nosocomial infection with tuberculosis is an important issue owing to the greater likelihood of severe disease and high mortality in this age group.1,2 The risk of infection and disease for infants born to infected mothers is very high,3 but this may not be the case for more casual contact. There are guidelines on managing neonates who have had close contact with people with infectious tuberculosis4 but there are none for more casual contact such as occurs in hospitals. We therefore conducted a systematic review to examine the risk of transmission in these circumstances.

Back to Top | Article Outline


We conducted a search for incident reports from a national database and from the literature. Data on incidents of tuberculosis exposure in health care and community settings have been held in a passively collected national database, the Tuberculosis Incident and Outbreak Surveillance (TBIOS) system, since 2004. We examined records submitted to TBIOS between 2004 and 2006. Incidents were defined as events in which potential exposure to tuberculosis in infants younger than 24 months occurred in a hospital or other health care setting.

We systematically searched Medline, Embase, and other electronic databases up to October 2008 with the following terms: Tuberculosis, Mycobacterium tuberculosis, Health Personnel, Disease Transmission, Professional-to-Patient, Infant, Newborn, Nurseries, and Hospital. In addition a search with Google scholar using the terms “neonatal tuberculosis,” “newborn tuberculosis,” and “infant tuberculosis” was undertaken.

Two reviewers independently reviewed all identified titles, abstracts and reports. The full text of all articles considered relevant by either reviewer was then assessed, including a search of all potentially relevant references not already identified. Studies were independently selected for inclusion. All studies reporting the results of an incident in which a health care worker, mother or infant was diagnosed with tuberculosis, and neonates/infants up to the age of 24 months exposed to the potential risk of infection in a hospital or other health care setting were included. Studies describing a case report with no subsequent investigation and non English language studies in which there was no English translation were excluded. Data were independently extracted in a standard form. Disagreements were resolved by consensus.

From each report we collected core data comprising the study setting, the year of study, the category of index case (HCW, mother or neonate), the infectivity of the index case (whether sputum smear positive or negative), the duration of infectivity of the index case, the definition of exposure (criteria for screening), the number of infants exposed, the number of infants screened, the screening method, the number of cases of latent and active disease identified, the number of infants started on chemoprophylaxis, and the further follow-up and results. No formal assessment of study quality was undertaken.

The total number of cases of latent infection and active disease among those exposed was calculated for each incident. Sub group analysis was undertaken by the type of index case (whether HCW, mother or infant), their sputum smear status and the setting in which exposure occurred. Owing to the limited number of cases identified meta analysis was not undertaken. Statistical tests were χ2 or Fisher exact test as appropriate.

Back to Top | Article Outline


Forty-seven potentially relevant articles were identified of which 15 were not considered further after examination of titles and abstracts. Thirty-two full text studies were reviewed and 19 included in the analysis. A summary of the characteristics of the population exposed, the management and outcome of the incidents, and excluded studies are available from the authors. In most situations, all exposed infants were given chemoprophylaxis (with isoniazid, apart from contacts of a multidrug resistant case). Only 3 of 19 investigations reported adverse effects and very few individuals were affected. From the TBIOS dataset and local Health Protection Unit records we identified 24 reports. However, despite attempts to fill missing fields by contacting the relevant Units we could only complete the core data set on 7 incidents (Table 1).



As shown in Table 1 only 7 instances of transmission of M. tuberculosis to infants were identified. The overall proportion infected was 0.14% among 4867 infants who were screened. The index case was a health care worker in 14 incidents and 0.09% babies (4/4310) were infected, a mother in 7 incidents and 1.28% (3/235) of babies were infected whereas none of 239 babies exposed to 4 infected infants were infected. The differences were statistically significant (P < 0.01). The yield was greater among contacts of sputum smear positive cases at 7 of 2568 babies screened (0.27%) than sputum smear negative culture positive pulmonary cases (none of 2299 babies screened) (P < 0.01). Six of 3390 babies (0.18%) screened in nurseries, 1 of 1462 babies (0.07%) in neonatal intensive or special care units and 1 of 15 (6.67%) in pediatric intensive care were infected (P = 0.02).

Back to Top | Article Outline


This review of the published literature and national data shows that despite screening over 4800 infants only 7 were found to be infected. It included 2 publications, which described the search for a source case after identifying tuberculous disease in an infant rather than follow up of infants exposed to a known source.5,6 The highest yield was among contacts of mothers with tuberculous disease, and lowest when the index case was an infant. Smear positive cases were more likely to transmit infection. Although the numbers were small transmission risk appeared greatest in pediatric intensive care.

United Kingdom guidelines4 include recommendations for neonates who are close contacts of smear positive tuberculosis cases. There are, however, no guidelines on the management of casual contacts such as those in hospital settings, the role of Bacillus of Calmette and Guerin (which is now given to infants at higher risk of tuberculosis at birth), or for those exposed to sputum smear negative tuberculosis.

There are a number of problems with the available data, and overall the evidence base is poor. The criteria for defining a contact varied greatly among incidents from any exposure to an index case with no defined risk period, to a minimum exposure period of 12 weeks before the index case was diagnosed. There was no clear pattern of disease detection in contacts to suggest which period was most appropriate.

It is clear from our investigation of incidents reported to TBIOS that reporting and follow up is far from complete. It is likely that, even if our literature search identified every published incident, there are many others remaining unpublished. Although publication bias usually results in reporting of positive findings, many institutions in these circumstances may be reluctant to report a potentially embarrassing problem.

We would recommend that more effort is focused on prospectively identifying and collecting core data from these incidents and in following up individuals, particularly neonates, identified as part of a contact tracing exercise. Further research should include the systematic follow up of the cohort of contacts, and assess the predictive value of screening tests in this population. Systematic follow up of exposed infants would also provide evidence of the effectiveness of regimens for the treatment of latent infection and the adverse effects of chemoprophylaxis.

Back to Top | Article Outline


1. Comstock GW, Livesay VT, Woolpert SF. The prognosis of a positive tuberculin reaction in childhood and adolescence. Am J Epidemiol. 1974;99:131–138.
2. Abubakar I, Laundy MT, French CE, et al. Epidemiology and treatment outcome of childhood tuberculosis in England and Wales: 1999–2006. Arch Dis Child. June 2008. Available at: Accessed 30 Oct, 2008.
3. Tuberculosis. In: Pickering LK, Baker CJ, Long SS, et al, eds. Red Book: 2006 Report of the Committee on Infectious Diseases. 27th ed. Elk Grove Village, IL: American Academy of Pediatrics; 2006;678–698.
4. The National Institute for Health and Clinical Excellence, Tuberculosis [NICE website]. March 2006. Available at: Accessed Oct 30, 2008.
5. Kim KI, Lee JW, Park JH, et al. Pulmonary tuberculosis in five young infants with nursery exposure: clinical, radiographic and CT findings. Pediatr Radiol. 1998;28:836–840.
6. Steiner P, Rao M, Victoria MS, et al. Miliary tuberculosis in two infants after nursery exposure: epidemiological, clinical and laboratory findings. Am Rev Resp Dis. 1976;113:267–271.

tuberculosis; infants; contact tracing; nosocomial exposure

© 2009 Lippincott Williams & Wilkins, Inc.