The effect of pre-existing cardiovascular disease on cancer treatment receipt

This directed acyclic graph is used to inform the estimation of the total causal effect of pre-existing cardiovascular disease on cancer treatment receipt using population-based cancer registry records in Northern Ireland.

Northern Ireland Cancer Registry (NICR) has collected cancer data for NI residents since 1993, using pathology reports, hospital records, and death registrations. Cancer Information Officers verify diagnoses following international standards, and can also use additional data from radiotherapy, chemotherapy systems and multidisciplinary team discussions records.

All patients aged 18 or older who were diagnosed with cancer, excluding non-melanoma skin cancer, for the first time in NI between 2009 and 2019 were included in the study. Using unique patient identifiers, the NICR securely and confidentially linked cancer records with other relevant centralised databases, including hospital admissions, systemic anticancer therapies, radiotherapy data, and prescriptions.

The primary outcome of interest was the receipt of cancer treatments, including chemotherapy, hormone therapy, radiotherapy, and surgery. NICR have identified chemotherapy, hormone therapy and radiotherapy events from the Regional Information System for Oncology & Haematology (RISOH), Enhanced Prescribing Database (EPD) and regional radiotherapy information system (ARIA), respectively. Surgery data has been determined using OPCS Classification of Interventions and Procedures codes from the Patient Administration System (PAS).

The presence of a cardiovascular disease (CVD) diagnosis before or on the date of cancer diagnosis was the key exposure of interest. The definition of CVD included congestive heart failure (CHF), myocardial infarction, other ischaemic heart diseases (categorised together as angina), atrial fibrillation (AF), other cardiac arrhythmias, valvular disease, pulmonary circulation disorder, peripheral vascular disease (PVD), embolism and thrombosis, myopericarditis, and cardiac arrest. Data on cardiovascular diseases were derived from the diagnoses (including primary, subsidiary, and up to five secondary diagnoses) recorded at any hospital admission between 2006 and 2019 using ICD-10 codes as previously reported.

Deprivation was measured using the NI Multiple Deprivation Measure and categorised by quintiles. Rurality consisted of three categories: urban, mixed, and rural. Both deprivation and rurality were at the area level (ie. Super Output Area), based on where the patient resided at the time of cancer diagnosis. Other comorbidities included cerebrovascular disease, chronic pulmonary disease, hypertension, diabetes, liver disease, renal disease, peptic ulcer, anaemia, neurodegenerative disorders, and rheumatoid disorders. Similar to cardiovascular diseases, other comorbidities were identified from the diagnoses at hospital admissions in 2006-2019 using ICD-10 codes.

More details are available at https://osf.io/vgzq5