Disease prevalence estimations based on contact registrations in general practice

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Estimating chronic disease prevalence from general practice data is crucial for public health.
• Traditional methods adjusted for observation length may lack precision.
• Limited time data presents challenges for accurate prevalence estimation.

Purpose of the Study:

• To develop and evaluate novel methods for estimating chronic disease prevalence using limited-duration general practice contact registration data.
• To improve the accuracy of prevalence rate estimation beyond simple adjustments for observation period length.

Main Methods:

• Proposed three new prevalence rate estimators utilizing patient first contact time, joint patient-contact numbers, and patient sets in distinct time intervals.
• Developed estimators for both homogeneous and parameterized heterogeneous patient populations.
• Cross-validation performed using data from the Dutch 'Study on Chronic Conditions' for five chronic diseases.

Main Results:

• The first two estimators demonstrated effectiveness for diseases with structured visiting patterns (hypertension, diabetes mellitus) under specific assumptions.
• These methods performed well assuming a time-constant contact rate and homogeneous patient populations.
• Underestimations were observed for diseases with less structured visiting patterns (ischaemic heart disease, respiratory diseases, osteoarthritis).

Conclusions:

• New methods offer improved chronic disease prevalence estimation from limited general practice data, particularly for diseases with predictable patient contact behaviors.
• The choice of estimator should consider the specific disease's visiting patterns and population characteristics.
• Further refinement may be needed for diseases with complex or irregular patient contact behaviors.