HTB South

High mortality and poor record keeping for TB in-patients in Kwazulu-Natal Hospital

Nathan Geffen, TAC

A poster by Cudahy and colleagues examined mortality and record-keeping of TB in-patients in Edendale Hospital in Kwazulu-Natal.  [1]

The researchers conducted a cross-sectional audit of patient charts in medical wards for a 4 week period in early 2009. These records were compared with the hospital TB register. Patients who were prescribed TB treatment and lived longer than 24 hours from admission were included for analysis. They identified 79 such patients. All were on first-line TB medication, but some were retreatment cases (ie they also received streptomycin). There were 40 pulmonary and 39 extra-pulmonary TB cases.  HIV status was known for 68 (86%) of patients, of whom 58 were HIV-positive and 43 had CD4 counts < 200 cells/mm3, but only 13 patients were on ART. The mean CD4 count for HIV-positive patients was 95.

Of the 79 TB patients, 24 died as inpatients. None were recorded as deaths in the hospital register submitted to the district TB control programme. Five were recorded as transferred to a local clinic for follow-up in the hospital register. The remaining 19 were unrecorded.  16 had TB as the recorded cause of death on their death certificates.

Two of the 79 left hospital against medical advice, but they were unrecorded in the TB register.

Five of the 79 were transferred to a TB hospital. Four of them were unrecorded in the TB register and one was recorded, incorrectly, as transferred to a local clinic for follow-up.

The remaining 48 were discharged to home to be followed up at their local clinics. Only 66% of them reported to their local clinic within 30 days. The register recorded only 36 patients transferred to local clinic for follow-up. The other 12 were unrecorded.

The authors note that a similar audit was conducted seven years ago with similar results. However, the poor recording of deaths is a new finding. One positive finding is that in the previous audit 30% of patients did not have an HIV test result, while now it was only 14%.

The authors conclude that “these results add to the conclusion reached elsewhere that patients, often HIV-positive, present late for care directly to hospitals and sick enough to require admission. This has left medical wards overcrowded and overworked leading to the poor outcomes seen in this audit. More resources need to be directed to TB care at hospital sites and better systems need to be implemented to track patient outcomes once discharged.”


These are disturbing findings and action by the Kwazulu-Natal Provincial Department of Health is necessary to rectify the discord between hospital and patient records. While the authors’ conclusion is accurate, the death reporting system worked better seven years ago. It is possible that years of mismanagement of health in Kwazulu-Natal, has led to worsening systems.

Nevertheless, it is unlikely that the findings about record keeping of TB mortality at Edendale are exceptional. Nor are the problems likely to be peculiar to South Africa. This has implications not only for the accuracy of statistics kept for the Millenium Development Goals as pointed out by the study’s authors, but also for Statistics South Africa’s mortality data.

The accuracy of TB and HIV data supplied by the National Department of Health are ultimately dependent on the quality of data at health facility level. Given these findings about TB statistics at one of the larger Kwazulu-Natal health facilities, it raises questions about the quality of data of HIV patients generally and consequently the quality of the statistics provided by the National Department of Health on numbers of people on ART.

Elsewhere in this issue and in several previous HTB-South issues we have described high mortality in cohorts of drug-resistant patients. But this small study demonstrates very high mortality in HIV-positive hospitalised TB patients generally. It underscores the importance of early TB screening and ensuring HIV-positive people with TB are prioritised for ART. It is disturbing that only 13 out of 43 patients with AIDS were on ART, though it is possible that all or some of those were still in the intensive phase of TB treatment and consequently not eligible for ART if their CD4 counts were about 50 cells/mm3.

Ref: Cudahy P et al. 2009. TB Case Outcomes at a Large Public Sector Hospital in Kwa-Zulu Natal, South Africa. 40th World Lung Conference, Cancun, December 2009.

Links to other websites are current at date of posting but not maintained.