Managing demand for the direct antiglobulin test with a big data-derived predictive equation

The direct antiglobulin test (DAT) is a key diagnostic tool in evaluating autoimmune haemolytic anaemia. However, indiscriminately ordering this test, together with certain methodological limitations, can compromise the efficiency of the clinical laboratory. This study aimed to develop and validate a predictive equation to identify negative results, optimising the use of DAT while maintaining the quality of care.

Through the laboratory information system (LIS), 1155 data were obtained from patients requesting DAT. A multiple logistic regression analysis was performed based on magnitudes related to haemolytic anaemia to obtain the best predictive model. The predictive equation obtained was: p = 1/(1 + e^-z) where 'P' represents the probability that the DAT is positive and 'z' the equation with the variables included in the model. Subsequently, its diagnostic efficiency was evaluated using a receiver operating characteristic curve. Finally, the equation was validated using a new cohort of data (N = 164).

The 'z' value obtained from the best predictive equation was: $z=-2.884\hspace{0.17em}-\left(0.373\text{\hspace{0.17em}x\hspace{0.17em}Haptoglobin}\right)+\left(0.312\text{\hspace{0.17em}x\hspace{0.17em}\%Ret}\right).$ For the selected threshold, the equation demonstrated a sensitivity of 81.6%, a negative predictive value of 95.8%, and an area under the curve [95% confidence interval] of 0.812 [0.760-0.864]. According to the proposed equation, the performance of 61.6% of DAT would be reduced.

The proposed equation has an excellent predictive ability for negative DATs. Its simple integration into the LIS confirms its applicability in routine clinical laboratory practice, providing an effective screening tool for optimising DAT demand and managing resources efficiently.