Azuonwu Obioma1,*, Ihua, Nnenna1, Adebayo-Olajide Testimonies2, Chikanka Onwuli Donatus3, Reuben E4, Orabueze, Ifeoma Celestina5, Azuonwu Goodluck6, Poplong Natasha Angel7, Tamuno-Boma8, Odinga, Omaegbu Ebere Elizabeth9, John-Amadi Victory Sorbari10, Ekwuozor, Ikem Kris Eloka10, Wokem Gloria Ngozika1, Vetty Agala11
1Department of Medical Laboratory Science, Medical Bacteriology/Virology/Parasitology Unit, Rivers State University, Nkpolu–Oroworukwo, Port Harcourt, Rivers State, Nigeria.
2 Department of Microbiology and Biotechnology, Caleb University, Lagos, Nigeria
3Department of Medical Laboratory Science, Chemical Pathology Unit, Rivers State University, Nkpolu–Oroworukwo, Port Harcourt, Rivers State, Nigeria
4Department of Human Physiology, College of Medicine, Rivers State University, Nkpolu–Oroworukwo, Port Harcourt, Rivers State, Nigeria
5Faculty of Pharmacy, Department of Pharmacovigilance, University of Lagos, Nigeria
6Department of Nursing, University of Port Harcourt, Choba, Nigeria
7 Departments of Biological Science, Faculty of Science, Federal University of Kashere, Gombe State, Nigeria
8Department of Biochemistry, Rivers State University, Nkpolu–Oroworukwo, Port Harcourt, Rivers State, Nigeria
9Department of Computer Science, Rivers State University, Nkpolu–Oroworukwo, Port Harcourt, Rivers State, Nigeria
10Department of Animal and Environmental Biology, Rivers State University, Nkpolu–Oroworukwo, Port Harcourt, Rivers State, Nigeria
11Department of Community Medicine, University of Port Harcourt, Choba Nigeria
*Corresponding Author: Azuonwu Obioma, Rivers State University, Port Harcourt, Nkpolu–Oroworukwo, Port Harcourt, Rivers State, Nigeria; Email: [email protected]
Received Date: March 9, 2023
Publication Date: April 26, 2023
Citation: Obioma A, et al. (2023). Application of Mathematical Model-Latent Class Model in Methodological Evaluation of Diagnostic Algorithms and Imperfect Reference Standard of Selected Index Test Techniques in Parasitology. Clin Res. 4(1):6.
Copyright: Obioma A, et al. © (2023).
Background: Disease diagnosis cannot be made with certainty thus, choosing the best diagnostic strategy is basic for understanding patient management outcomes. This requires substantiation of the comparative performance of diagnostic algorithms. The use of a single index test in parasitic detection has been invalid and had also proven unacceptable among critical professionals. The aim of this study anchors on Methodological Evaluation of Diagnostic Algorithms and Imperfect Composite Reference Standard of Selected Index Test Techniques in Parasitology using the application of mathematical models-Latent class model. It will also compare the diagnostic performance of three index test techniques in the detection of parasites, using Extrapolated-composite Imperfect Reference Standard and Bayesian Latent Class Model. Study Protocol: This study was carried out in Rivers State, Nigeria. Laboratory investigation of the index test techniques for direct microscopy, Brine microscopy and Diethyl Ether Microscopy followed the routine parasitological methods with a sample size of eighty. The imperfect reference (gold) standard was extrapolated from a combination of the three index test techniques. All tests were categorically analysed as binary outcomes (positive or negative). Statistical analysis was performed using SPSS version 21 to test for inter-rater agreement and other concordance indices. Sensitivity, Specificity, Positive Predictive Value, Negative Predictive Value, Prevalence, Likelihood Ratio Positive and Negative, False Discovery Rate, False Omission Rate, and Diagnostic Odd Ratio, Kappa, Kendall's Coefficient of Concordance, average Spearman Correlation and Cochran Q were the test statistics used in this study. An alpha level of 5% was set for decision. Also, Bayesian latent class Model was performed with Modelling of Infectious Disease Centre (MICE) Model Code MODEL103. Results: For detection rate, Direct Microscopy was the least while Diethyl Ether Microscopy was the highest. Strong concordance was observed showing good inter-rater agreement. The study generally recorded low sensitivity irrespective of the technique or model used. Composite reference standard did not differ statistically (p>0.05) from the Latent Class Model only for sensitivity, others showed marked variation (p<0.05). Conclusion: This current study has been able to bear out the significance of LCM as a useful tool.
Keywords: Diagnostic, Algorithms, Imperfect, Reference, Standard, Parasitology, Methodological, Evaluation, Index, Test, Techniques, Extrapolated, Bayesian, Latent, Class, Model