Competing Risk Modeling Of HIV Progression

Background and objectiveThe HIV/AIDS is a serious infectious disease,representing an important public health issue worldwide.The introduction of Highly Active antiretroviral treatment(HAART)in 1996 has largely extended the survival time of HIV/AIDS patients and improved their life quality.In 2003,the Chinese government started to implement the "Four Free and One Care" policy and "National Free Antiretroviral Treatment Program".Great achievements have been made nationwide since then.However,few studies have examined the pattern of disease progression and survival among the HIV/AIDS patients under the universal access to the HAART in Shandong Province.Information on disease progression of HIV/AIDS patient in an HIV/AIDS cohort is of obvious importance to develop optimal treatment and intervention guidelines,to monitor and predict the progress of the HIV/AIDS epidemic,and to plan health services during the time of HAART.In the analysis of HIV cohort in order to study the HIV/AIDS progression,competing risks often exist,which could preclude the presence of our interested.Few previous studies about the HIV progression had pay attention to the competing events during follow-up.The occurrences of AIDS un-related death were often treated as censoring,and standard Kaplan-Meier method were used to estimate the cumulative incidence of our interested events,which may lead to bias.Therefore,proper statistical method must be applied to handle the problem of competing events.In the last three decades different methods for analyzing the survival data in the presence of competing risks have been introduced,provide us effective tool to study the disease progression of HIV/AIDS.The aims of this study are to estimate the cumulative risk of progressing to AIDS and death after HIV diagnosis and to identify determinants for disease progression using the competing risk model.Methods:1 A longitudinal cohort dataset was established based on the surveillance and follow-up data of all the HIV infectors reported to Shandong provincial CDC from 1992 to July 2013.A total of 4650 subjects were included in the analysis.Based on this large dataset,3 sub-cohort were selected to assess the disease progression of HIV infectors in Shandong province:1)non-AIDS HIV progressing to AIDS cohort.;2)AIDS progressing to death cohort;3)HIV infection to death cohort in which the presence of AIDS related death as interested events.In all the three cohort,the occurrence of AIDS unrelated death were defined as competing events,and the absence of both interested events and competing events during the follow-up period was considered as censor.2 The probability of disease progression were estimated by the cumulative incidence function(CIF).Then the differences of cumulative incidence function and standard K-M method for estimating the cumulative incidence of ADIS related death after HIV diagnosis were compared.3 Both cause-specific and sub-distribution hazard regression were used to explore influencing factors and to make predictions of HIV/AIDS disease progression.And the difference in the interpretation of parameters obtained from these two methods were comparedResult:1.The cumulative probability of progressing to AIDS for HIV infectors were 5.7%,18.2%,33.9%and 47.9%in the end of 1 year,3 years,5 years and 10 years after HIV diagnosis respectively.While for the AIDS patients,the cumulative incidence of AIDS related death were 18.4%,23.3%,24.9%and 27.1%in 1 year,3 years,5 years and 10 years after AIDS diagnosis.The overall probability of death for all the HIV patients(including those have concurrent AIDS)in 1 year,3 years,5 years and 10 years after first diagnosis of HIV positive was 8.4%,11.8%,14.8%and 19.3%.By comparing the estimation by cumulative incidence function with Kaplan-Meier method,the results showed increased differences between two methods for cumulative incidence of AIDS related death estimates with extended follow-up time and that Kaplan-Meier method would overestimate the incidence of AIDS related death.2.Both the results of sub-distribution and cause-specific regression model showed that,the patients with higher CD4+T-cells count and lower age at diagnosis had lower risk of progressing to AIDS and AIDS related death after HIV diagnosis as well as progressing to AIDS related death after AIDS diagnosis.The highly active antiretroviral therapy(HAART)delayed progression to AIDS related death from HIV diagnosis and reduced the risk of death after AIDS diagnosis.While sex and transmission category have no effect on the disease progression and survival of HIV/AIDS patients.Conclusions1.Determinants of HIV/AIDS progression and survival include baseline CD4 count,age at diagnosis,HAART and education level.Effective interventions should target those at excess risk of disease progression as well as those with higher risk of AIDS unrelated death.2.Competing risk model is more comprehensive in the analysis of HIV/AIDS cohort to study the disease progression.