Usefulness of screening large numbers of contacts for tuberculosis: questionnaire based review

Introduction

In two separate incidents of tuberculosis that we investigated, 732 contacts were identified and 524 screened. As no further cases of tuberculosis were detected, we began to question the value of screening and conducted a questionnaire survey of colleagues in England and Wales to determine their experience.

Subjects,_methods,_and_results

In April 1994 we sent a questionnaire to consultants in communicable disease control and medical officers of environmental health in 155 districts in England and Wales. The questionnaire asked for the number of new cases of tuberculosis found in the preceding three years in incidents in which more than 100 contacts had been screened.

Altogether 123 replies (79%) were returned, with 56 incidents reported. West Midlands and North Western regions reported the largest numbers of incidents (eight and six respectively) while South Western and Northern regions reported none. Altogether, 14 267 were identified (mean 255 per incident). One incident had 1500 contacts and four other incidents had between 500 and 1000 contacts. In 53 incidents, 11 720 contacts were screened; no details were given for the three other incidents.

Forty four cases of tuberculosis were found in 18 of the 56 incidents (table) giving a detection rate of 0.375%. A further 106 (0.9%) contacts received chemoprophylaxis. The development of tuberculosis in contacts, in the 39 incidents with details available, was significantly correlated with the proportion of contacts who had positive results on tuberculin testing (Spearman rank correlation coefficient rs=0.416, P=0.008).

Comment

The screening of 100 or more contacts of an index case was not unusual—46% of districts reported at least one such incident.

The detection rate of 0.375% for tuberculosis was much lower than the historical detection rate of 1% for several reasons.1 Firstly, it may be related to inappropriate screening—usually only close contacts of patients with disease that is smear positive need screening.1

Secondly, some cases of tuberculosis in contacts may have been missed because of inadequate follow up. Most disease, however, is found at the first follow up after screening.2

Thirdly, the screening test may be inappropriate today. The tuberculin test may be a poor screening tool for tuberculosis,3 although in our survey the development of tuberculosis in contacts was correlated with the proportion who had positive results on tuberculin testing.

Fourthly, tuberculosis may now be less infectious in close contacts because of changes in the host, the organism, or the environment.

Finally, the cases of tuberculosis may have been unconnected with the index case and an incidental finding in a susceptible group. This would require clarification with molecular techniques.

Whatever reason for the low yield we question the continued use of large scale screening. Close contacts should be carefully defined, and only they should be screened. The considerable pressure to screen large numbers of casual contacts should be resisted.4 Veen has suggested screening the closest contacts and then extending this to more contacts, depending on the yield of tuberculosis in the initial screening.5 In addition to screening genuine close contacts, other contacts and groups and people at high risk could be provided with information about tuberculosis to enable them to look for symptoms of the disease in themselves and others and to take appropriate action.

If screening of large numbers of contacts for tuberculosis continues, health authorities and providers must be aware of the implications. Local arrangements for the screening and management of contacts will need to be agreed by several departments and organisations. More information will be needed on the closeness to an index case of contacts uncovered by screening. Furthermore, techniques of molecular epidemiology should be applied to determine whether contacts with tuberculosis are connected with or incidental to an index case.