A simple and robust statistical framework for planning, analysing and interpreting faecal egg count reduction test (FECRT) studies

TY - ABST

T1 - A simple and robust statistical framework for planning, analysing and interpreting faecal egg count reduction test (FECRT) studies

AU - Denwood, M.J.

AU - McKendrick, I.J.

AU - Matthews, L.

AU - Prentice, J.C.

AU - Reid, Stuart W.J.

AU - Innocent, G.T.

N1 - Conference code: 26

PY - 2017

Y1 - 2017

N2 - Introduction. There is an urgent need for a method of analysing FECRT data that is computationally simple and statistically robust. A method for evaluating the statistical power of a proposed FECRT study would also greatly enhance the current guidelines. Methods. A novel statistical framework has been developed that evaluates observed FECRT data against two null hypotheses: (1) the observed efficacy is consistent with the expected efficacy, and (2) the observed efficacy is inferior to the expected efficacy. The method requires only four simple summary statistics of the observed data. Power calculations can also be obtained for any proposed study design and given set of parameter values. For illustration, the power to detect a significant deviation from 95% efficacy is estimated for a population efficacy of 90% and parameter values typical to parasitology. Results. Simulation studies reveal that the notional type 1 error rate of the new statistical test is accurate. Power calculations demonstrate a power of only 65% with a sample size of 20 treatment and control animals, which increases to 69% with 40 control animals or 79% with 40 treatment animals. Discussion. The method proposed is simple and robust, and can analyse paired and unpaired data with a range of study designs, including pooling and repeated observations. The facility for power calculations gives crucial insight into the design of more efficient FECRT studies, and suggests that the priority should be to collect more samples from the treatment group. The software has been made freely available as an online tool.

AB - Introduction. There is an urgent need for a method of analysing FECRT data that is computationally simple and statistically robust. A method for evaluating the statistical power of a proposed FECRT study would also greatly enhance the current guidelines. Methods. A novel statistical framework has been developed that evaluates observed FECRT data against two null hypotheses: (1) the observed efficacy is consistent with the expected efficacy, and (2) the observed efficacy is inferior to the expected efficacy. The method requires only four simple summary statistics of the observed data. Power calculations can also be obtained for any proposed study design and given set of parameter values. For illustration, the power to detect a significant deviation from 95% efficacy is estimated for a population efficacy of 90% and parameter values typical to parasitology. Results. Simulation studies reveal that the notional type 1 error rate of the new statistical test is accurate. Power calculations demonstrate a power of only 65% with a sample size of 20 treatment and control animals, which increases to 69% with 40 control animals or 79% with 40 treatment animals. Discussion. The method proposed is simple and robust, and can analyse paired and unpaired data with a range of study designs, including pooling and repeated observations. The facility for power calculations gives crucial insight into the design of more efficient FECRT studies, and suggests that the priority should be to collect more samples from the treatment group. The software has been made freely available as an online tool.

M3 - Conference abstract for conference

T2 - Conference of the World Association for the Advancement of Veterinary Parasitology

Y2 - 4 September 2017 through 8 September 2017

ER -