Comparing laboratory surveillance with the notifiable disease surveillance system in South Africa

Abstract

Purpose: A functioning surveillance system is essential for the successful response to communicable diseases. The purpose of the study was to compare laboratory surveillance with the notifiable disease surveillance system (NDSS) in South Africa, as part of a broader evaluation of the NDSS. Methods & Materials: A retrospective record review was conducted to analyse the attributes of data quality (percentage completeness of data), stability (reliability of the system in providing a diagnosis), representativeness (percentage of provinces represented in the system), sensitivity (proportion of cases detected by the system) and positive predictive value (PPV - proportion of reported cases that actually have the disease) of the national NDSS. We compared the data from the two systems on three tracer notifiable diseases- measles, meningococcal meningitis and typhoid- for 2013. A record review form extracted information on disease diagnosed or notified and patient demographics. All records with inconclusive diagnostic information were excluded from the study sample. We used the Wilcoxon and the Chi-square tests to compare the systems. All tests were conducted at 5% significance level. Results: For all three tracers, fewer cases were notified than laboratory confirmed. The median completeness for the laboratory was higher than notified cases for both measles (63% versus 47%, p<0.001) and meningococcal meningitis (63% versus 57%, p<0.001) but there was no significant difference for typhoid (63% versus 60%, p=0.0818). Stability was higher in the laboratory (100% for all three diseases) compared to the notifiable diseases for measles (24%, p<0.001), meningitis (74%, p<0.001), and typhoid (36%, p<0.001). Representativeness was also higher in the laboratory (all 100%) compared to the notifiable cases where we found 67% for measles (p=0.058), 56% for meningitis (p=0.023), and 44% for typhoid (p=0.009). We found the sensitivity of the NDSS to be 38%, 81% and 50% and the PPV to be 4%, 35% and 41% for measles, meningococcal meningitis, and typhoid respectively. Conclusion: Compared to laboratory surveillance, the NDSS performed poorly on most system attributes. The revitalisation of the NDSS in South Africa should address the completeness of data, as well as the stability, representativeness, sensitivity and PPV of the system.