TITLE:
Generalized Estimating Equations for Repeated Measures Logistic Regression in Mosquito Dose-Response
AUTHORS:
Gabriel Otieno, Gichihu A. Waititu, Daisy Salifu
KEYWORDS:
Dose-Response; GEE; Lethal Time; Probit Analysis; Repeated Measures
JOURNAL NAME:
Open Journal of Statistics,
Vol.3 No.5,
October
9,
2013
ABSTRACT: Dose-response studies in arthropod research usually involve
observing and collecting successive information at different
times on the same group of insects exposed to different concentrations of
stimulus. When the same measure is collected repeatedly over time, the data become
correlated and Probit Analysis technique which is the standard method in
analyzing bioassay experiments data cannot be used. Lethal time is estimated
when the speed of kill is of interest since mortality varies over time. We
evaluate a complementary approach, repeated measures logistic regression using
Generalized Estimating Equations (GEE), for lethal time determination in
mosquito dose response. Mortality data from anopheles larva exposed to 3
botanical extracts (B,C,E) at 4 concentration levels: 500 mg/ml, 250 mg/ml, 50
mg/ml and 12.5 mg/ml were used. The result shows the estimated LT50 values with concentration 500 mg/ml being the most virulent chemical for
extract B (LT50 = 10.3 hrs), C (LT50 = 7.2 hrs) and E (LT50 = 10.3 hrs). The
least virulent chemical was concentration 12.5 mg/ml for extract B (LT50 = 52.1
hrs), C (LT50 = 70.7 hrs) and E (LT50 = 55.0
hrs). We conclude that repeated measures of logistic regression via GEE can be
used as a tool to estimate LT50 more effectively in repeated
measures of arthropod data.