A comprehensive framework to estimate the frequency, duration, and risk factors for diagnostic delays using...

Area of Science:

• Health Informatics
• Epidemiology
• Biostatistics

Background:

• Diagnostic delays are poorly understood across many diseases and healthcare settings.
• Current methods for identifying delays are often resource-intensive and lack broad applicability.
• Real-world data sources offer potential for improved identification and study of diagnostic delays.

Purpose of the Study:

• To develop and validate a comprehensive framework for estimating missed diagnostic opportunities and delays using longitudinal real-world data.
• To model the disease-diagnostic and data-generating processes.
• To apply the framework to specific diseases like tuberculosis, acute myocardial infarction, and stroke.

Main Methods:

• A conceptual model of disease diagnosis and data generation was established.
• A bootstrapping method was developed to estimate the frequency and duration of diagnostic delays.
• Three distinct bootstrapping algorithms were implemented and compared using administrative data.

Main Results:

• Analysis of IBM MarketScan data (2001-2017) for tuberculosis, acute myocardial infarction (AMI), and stroke.
• Estimated missed diagnostic opportunities: 63.9-82.3% for tuberculosis, 16.0-21.3% for AMI, and 6.9-8.3% for stroke.
• Estimated diagnostic delay durations: 34.3-44.5 days for tuberculosis, 6.7-8.2 days for AMI, and 6.7-7.6 days for stroke.

Conclusions:

• The proposed framework is adaptable for studying diagnostic delays across various diseases using longitudinal administrative data.
• The choice of simulation algorithm can influence delay estimates, necessitating careful statistical consideration.
• The approach provides a flexible method for disease-specific customization and future research into diagnostic pathways.