Association between Diagnostic History and Cancer Incidence within 5 Years: A Real-world Observational Analysis

Area of Science:

• Epidemiology
• Medical Informatics
• Public Health

Background:

• Cancer development is influenced by numerous pre-existing health conditions (PHCs).
• Quantifying these associations using large-scale electronic health record (EHR) data is crucial but challenging.
• Previous analyses often lack the granularity to assess cancer risk across diverse patient subgroups.

Purpose of the Study:

• To evaluate associations between pre-existing health conditions (PHCs) and subsequent cancer diagnoses.
• To analyze cancer risk within a fixed five-year window using real-world EHR data.
• To stratify findings by demographic factors like gender, race, and area deprivation index (ADI).

Main Methods:

• Utilized a large EHR dataset (8,283,236 records; 1,460,738 patients) from 2013-2022.
• Applied epidemiological, statistical, and artificial intelligence methods.
• Conducted analyses on the full dataset and stratified subgroups for overall cancer and 20 specific cancer types.

Main Results:

• Identified 221 PHCs across nine ICD-10 chapters linked to increased cancer risk (RR > 1, BH-FDR-adjusted p < 0.05).
• Key associated PHCs include systemic sclerosis, blood type, HIV, vitamin D deficiency, and diabetes.
• Discovered age-, gender-, race-, and ADI-specific high-risk PHCs, with notable links in metabolic, genitourinary, skin, and circulatory disease chapters.

Conclusions:

• Established significant associations between numerous PHCs and subsequent cancer diagnoses.
• Highlighted the importance of considering demographic factors (gender, race, ADI) in cancer risk assessment.
• Acknowledged limitations including potential confounding from unmeasured socioeconomic/environmental factors and lack of tobacco data.