dc.description.abstract | Background: Researchers in medical sciences often tend to prefer Cox semi-parametric instead
of parametric models for survival analysis because of fewer assumptions but under certain
circumstances, parametric models give more precise estimates. The objective of this study was to
compare two survival regression methods Cox Regression and Parametric Models - in patients
with HIV/AIDS at Karuri Health Centre.
Methods: It was a retrospectively study where a total of 248 subjects were sampled .The period
of study was from 1st January 2008 through 31st December 2012.Gender, age at initiation of
HAART, baseline weight (kg) ,substance abuse (smoking, alcohol or any other drug), CD4 cell
count (cells/mm-jtaken at the beginning of the study, regimen type (TDF +3TC+EFV or NVP
and AZT+3TC+NVP or EFV),history of drug adherence, whether the patient is on tuberculosis
treatment or not and the world health clinical staging at the initiation of HAART were the
predictor variables. Cox Proportional Hazard model a semi-parametric model and Accelerated
Failure Time a parametric model (Weibull, Exponential and lognormal form) were performed
with the Akaike Information Criterion (AlC) been used to compare the efficiency of models.
Results: The mortality rate was high in subjects who abused drug, those who did not adhere to
treatment as prescribed, those who had tuberculosis and those who had low cd4 cell count and
low BMI at the beginning of treatment. Using Cox proportional hazard model, covariates that
significantly influence the survival of lIIVIAIDS infected patients were gender, history of drug
abuse and poor treatment adherence
Conclusion: Based on AlC, there was no major variability between the three parametric models
as witnessed with Cox Proportional Hazard model.Among the parametric models Wiebull is the
best model in multivariate analysis. The Weibull AFT model appears to be an appropriate AFT
model according to AlC compared to other AFT models, although it is only slightly better than
exponential or log-normal model. Cox PH model fair poorly compared to other parametric
models. | en_US |
dc.description.department | a
Department of Psychiatry, University of Nairobi, ; bDepartment of Mental Health, School of Medicine,
Moi University, Eldoret, Kenya | |