To conduct an economic evaluation of directly observed treatment (DOT) and conventionally delivered treatment for the management of new cases of tuberculosis in adults.
Community based directly observed treatment, which has been implemented in the Hlabisa district of South Africa since 1991, was compared with a conventional approach to tuberculosis treatment widely used in Africa. Each was assessed in terms of cost, cost effectiveness, and feasibility of implementation within existing resource constraints.
Hlabisa Health District, South Africa.
Adult patients with new cases of tuberculosis on smear testing; the number of cases increased from 20 per month to over 100 from 1991 to 1996.
Main outcome measures:
Cost of case management in 1996, cost effectiveness in terms of the cost per case cured, and bed requirements in comparison with bed availability for the 1990, 1993, and 1996 caseload. Costs are expressed in US dollars at values for 1996.
Directly observed treatment was 2.8 times cheaper overall than conventional treatment ($740.90 compared with $2047.70) to deliver. Directly observed treatment worked out 2.4-4.2 times more cost effective, costing $890.50 per patient cured compared with either $2095.60 (best case) or $3700.40 (worst case) for conventional treatment. The 1996 caseload of tuberculosis required 47 beds to be dedicated to tuberculosis to implement directly observed treatment, whereas conventionally delivered treatment would have required 160 beds; the current number of beds fur tuberculosis treatment in Hlabisa is fixed at 56.
Because of the reduced stay in hospital, directly observed treatment is cheaper, more cost effective, and more feasible than conventional treatment in managing tuberculosis in Hlabisa, given the existing hospital bed capacity and the escalating caseload due to the HIV/AIDS epidemic. Such results may hold elsewhere, and wherever conventional tuberculosis management is practised a switch to directly observed treatment will increase hospital capacity to cope with a growing caseload.