Pseudo and conditional score approach to joint analysis of current count and current status data

Area of Science:

• Biostatistics
• Epidemiology
• Survival Analysis

Background:

• Recurrent and nonrecurrent events are common in health studies.
• Existing methods often analyze these event types separately.
• Interval censorship (current count/status data) presents unique analytical challenges.

Purpose of the Study:

• To develop a joint statistical model for analyzing recurrent and nonrecurrent event processes simultaneously.
• To account for the association between recurrent and nonrecurrent events within subjects using shared frailty.
• To apply the developed method to identify risk factors for fracture and osteoporosis in elders.

Main Methods:

• A shared frailty model links recurrent (nonhomogeneous Poisson process) and nonrecurrent (semiparametric transformation model) events.
• The frailty distribution is left unspecified.
• Estimation employs pseudo-likelihood and conditional score techniques.
• Semiparametric transformation models are used for mean and survival functions.

Main Results:

• Consistent estimators for regression parameters and baseline functions are achieved.
• Asymptotic normality and closed-form variance are derived for regression parameter estimators.
• The method was successfully applied to fracture-osteoporosis survey data.

Conclusions:

• The proposed joint analysis effectively models recurrent and nonrecurrent events under interval censorship.
• The shared frailty approach successfully captures the association between linked event processes.
• This method aids in identifying risk factors for complex health outcomes like fracture and osteoporosis.