BACKGROUND : Tuberculosis control programs rely on accurate collection of routine surveillance data to inform
program decisions including resource allocation and specific interventions. The electronic TB register (ETR.Net) is
dependent on accurate data transcription from both paperbased clinical records and registers at the facilities to
report treatment outcome data. The study describes the quality of reporting of TB treatment outcomes from
facilities in the Ehlanzeni District, Mpumalanga Province.
METHODS : A descriptive crossectional study of primary healthcare facilities in the district for the period 1
January – 31 December 2010 was performed.
New smear positive TB cure rate data was obtained from the ETR.Net followed by verification of paperbased
clinical records, both TB folders and the TB register, of 20% of all new smear positive cases across the district
for correct reporting to the ETR.Net. Facilities were grouped according to high (>70%) and low cure rates
(≤ 70%) as well as high (> 20%) and low (≤ 20%) error proportions in reporting. Kappa statistic was used to
determine agreement between paperbased record, TB register and ETR.Net.
RESULTS : Of the100 facilities (951 patient clinical records), 51(51%) had high cure rates and high error
proportions, 14(14%) had a high cure rate and low error proportion whereas 30(30%) had low cure rates
and high error proportions and five (5%) had a low cure rate with low error proportion. Fair agreement was
observed (Kappa = 0.33) overall and between registers. Of the 473 patient clinical records which indicated
cured, 383(81%) was correctly captured onto the ETR.Net, whereas 51(10.8%) was incorrectly captured and
39(8.2%) was not captured at all. Over reporting of treatment success of 12% occurred on the ETR.Net.
CONCLUSIONS : The high error proportion in reporting onto the ETR.Net could result in a false sense of
improvement in the TB control programme in the Ehlanzeni district.