Identifying important comorbidity among cancer populations using administrative data: Prevalence and impact on survival

Aims: Our study sought to optimize the identification and investigate the impact of comorbidity in cancer patients using routinely collected hospitalization data. Methods: We undertook an iterative process of classification of important clinical conditions involving evaluation of relevant literature and consultation with clinicians. Patients diagnosed with colon, rectal, breast, ovarian, uterine, stomach, liver, renal or bladder cancers (n=14096) between 2006 and 2008 were identified from the New Zealand Cancer Registry. Conditions were identified using data on diagnoses from hospital admissions for 5years prior to cancer diagnosis. Patients were followed up until end of 2009 using routine mortality data. Prevalence estimates for each condition by site were calculated. All-cause mortality impact of common conditions was investigated using Cox regression models adjusted for age and stage at diagnosis. Results: Patients with liver and stomach cancers tended to have higher comorbidity and those with breast cancer, lower comorbidity than other cancer patients. Of the 50 conditions, the most common were hypertension (prevalence 8.0-20.9%), cardiac conditions (2.1-13.5%) and diabetes with (2.3-13.3%) and without (2.9-12.9%) complications. Comorbidity was associated with higher all-cause mortality but the impact varied by condition and across cancer site, with impact less for cancers with poor prognoses. Conditions most consistently associated with adverse outcomes across all cancer sites were renal disease, coagulopathies and congestive heart failure. Conclusion: Comorbidity is highly prevalent in cancer populations, but prevalence and impact of conditions differ markedly by cancer type.