Validation of haematology peripheral blood slide review rules at various tiers of diagnostic laboratories in South Africa

Abstract

Automated analysers in the haematology laboratory have vastly improved since their inception in the mid-20th Century. The capability of these analysers to provide information exceeding a numerical output have made them allies in the clinical laboratory field. These analysers generate suspect flags and additional differential parameters with enhanced sensitivity and specificity which can assist in reducing manual peripheral blood smear (PBS) review rates and increasing laboratory productivity. In 2005, the staffing pressures faced by clinical laboratories prompted the International Consensus Group (ICGH) for Haematology Review to establish a set of criteria (consensus rules) to guide manual smear review of automated full blood counts (FBC) and white blood cell differential counts (Diff) in order to reduce unnecessary smear review rates. The usefulness of these rules is dependent on the patient population, laboratory criteria for PBS review and the specifications of the analyser. The aim of this study was to apply ICGH rules to FBC samples (with and without Diffs) from tertiary, regional and district diagnostic laboratory tiers, to evaluate their efficiency, and to optimise their performance in the South African context. In this study, the ICGH rules were compared to the individual representative laboratory standard operating procedures (SOP) for smear review in 600 samples from laboratories representing the tertiary, regional and district health tiers. Manual PBS review was the reference method. The rules were analysed for sensitivity, specificity and efficiency. False negative (FN) and false positive (FP) rates were also recorded for the rule set collectively, as well as for individual parameter and morphology flags. In order to optimise the ICGH rules for our setting, samples with FP flagging were then more closely analysed to determine if exclusion or modification of the flags triggered in these samples would be beneficial. Smear review rates were substantially reduced on implementation of the ICGH rules in the tertiary laboratory as compared to the current laboratory SOP (82% vs 70%; p=0.022), were not iv significantly altered in the regional laboratory (72% vs 80%; p = 0.198), while review rates increased on implementation of the rules in the district laboratory (67.1% vs 79.3%; p <0.0001). FN rates were reduced in all three sites, with the greatest reduction occurring in the district laboratory (dropping from 32% to 5%; p < 0.0001). Efficiency rates of the rule set were similar to those reported in the literature in the tertiary laboratory (78.1%), but were poorer in the regional (74.5%) and district laboratories (68.6%). Leukopenia and thrombocytopenia were the most common parameter flags triggered in the tertiary hospital (20.7% and 25% of samples respectively), while anaemia and microcytosis were most prevalent in the regional facility (22.5% and 17.5% of samples respectively). The most common abnormal parameter in the district laboratory was leucopenia (WCC<4 x 109/L) (16.4%). The analysers from the different tiers triggered morphology flags variably. The findings of this study indicate the need for optimisation of the rules. Recommendations were made for each tier individually, each of which could reduce smear review rates further without increasing the clinically significant FN rates. In the tertiary laboratory, request for a Diff alone should not be a criterion for smear review and the white cell threshold for smear review should be reduced from 4 x 109/L to 2 x 109/L. Recommendations for the regional laboratory include modification of the WCC threshold to that of <2 x 109/L and exclusion of the platelet clumping (PCL) rule when the platelet count is normal/increased. In the district laboratory the nucleated red blood cell (NRBC) flag and PCL flag in samples with normal/high platelet counts should be excluded.