How accurate are diagnostic procedures for tuberculosis (TB)? A study by The Netherlands' Royal Tropical Institute and the Kenya Medical Research Institute compares the results of laboratory tests with routine diagnostic procedure in a Kenyan chest clinic. The study identified the main weaknesses in the procedure and proposed improvements.

TB programmes in many sub-Saharan countries are under strain due to the enormous increase in registered patients in recent years. Because of this, the time allowed for counselling patients has been reduced. Also, personnel make diagnostic decisions having devoted less time to reading laboratory smears or rely too much on chest x-rays (CXR). Further complications arise because the high prevalence of HIV in the area often leads health personnel to suspect TB even when symptoms are uncommon. The high levels of work associated with the increase of patients have also resulted in mistakes in diagnosis. Researchers carried out a study to compare the results of laboratory tests with the routine diagnostic procedure at the Rhodes Chest Clinic (RCC) in Kenya.

TB suspects at RCC underwent a routine diagnostic procedure. This procedure consisted of reading three sputum smears in those who had TB-associated symptoms for more than three weeks. When the smears turned out negative, a further diagnostic test - a CXR - was performed. The diagnostic results of this procedure were compared to a Löwenstein-Jensen culture; a laboratory test that was used as the standard to which the routine procedure was compared. Researchers established the following:

• The combination of sputum smears and CXR is a highly sensitive method for diagnosing TB. In all, 92 per cent of those who were infected with TB were detected by this method.
• However, this procedure over-diagnosed TB in 23 per cent of cases and failed to detect eight per cent of cases of TB.
• The reasons for missing real cases while over-diagnosing took place mainly in cases where the sputum smear was negative and physicians relied on CXR to make the diagnosis.
• The study argues that CXR-based diagnosis depends on the reading skills of the physician and on the quality of the picture. Since carrying out a Löwenstein-Jensen culture for the entire in Kenyan population is impossible, better levels of detection will have to be based on improving CXR reading skills.

In conclusion, this study showed that the routine diagnostic procedure at RCC is highly sensitive. However, this left eight per cent of those who were infected undiagnosed and it over-diagnosed many more. Improvements in the procedure might be made by:

• narrowing the criteria for CXR diagnosis
• improving the quality of CXR reading in the clinic
• improving the counselling offered to patients
• asking patients to return to the clinic if their symptoms persist in order to ensure that no infected cases are missed.

Source(s):
‘A comprehensive study of the efficiency of the routine pulmonary tuberculosis diagnostic process in Nairobi’, International Journal of Tuberculosis and Lung Disease 7(2): 1-4, by M.R.A. van Cleeff et al., 2003
‘Under-diagnosis of smear-positive pulmonary tuberculosis in Nairobi, Kenya’, International Journal of Tuberculosis and Lung Disease 5(4): 360-363, by M. Hawken et al., 2001

Funded by: The Netherlands Ministry of Developing Cooperation

id21 Research Highlight: 4 March 2003

Further Information:
Paul R. Klatser
KIT Biomedical Research
Royal Tropical Institute
Meibergdreef 39
Amsterdam 1105 AZ
The Netherlands

Tel: +31 20 566 5440
Fax: +31 20 697 1841
Contact the contributor: p.klatser@kit.nl

Royal Tropical Institute, the Netherlands

Kenya Medical Research Institute

Other related links:
'TB decided… Patient choice of treatment strategies in South Africa'

'Time lapse - delays in TB diagnosis in Zambia'

'Coughing up for TB tests - cost-effective diagnosis in Zambia'

See id21's collection of links relevant to infectious diseases.