Clinical characterisation of patients diagnosed with cancer following emergency self-referral.

[BACKGROUND] Despite their frequency and potential impact on prognosis, cancers diagnosed via self-referral to the emergency department are poorly documented. We conducted a detailed analysis of cancer patients diagnosed following emergency self-referral and compared them with those diagnosed following emergency referral from primary care. Given the challenges associated with measuring intervals in the emergency self-referral pathway, we also aimed to provide a definition of the diagnostic interval for these cancers.

[METHODS] A retrospective observational analysis was performed on patients diagnosed with 13 cancers, either following emergency self-referral or emergency referral from primary care. We analysed demographics, tumour stage, clinical data (including 28 presenting symptoms categorised by body systems), and diagnostic intervals by cancer site, then testing for differences between pathways.

[RESULTS] Out of 3624 patients, 37 % were diagnosed following emergency self-referral and 63 % via emergency referral from primary care. Emergency self-referrals were associated with a higher likelihood of being diagnosed with cancers manifesting with localising symptoms (e.g., breast and endometrial cancer), whereas the likelihood of being diagnosed with cancers featuring nonspecific symptoms and abdominal pain (e.g., pancreatic and ovarian cancer) was higher among patients referred from primary care. Diagnostic intervals in self-referred patients were half as long as those in patients referred from primary care, with most significant differences for pancreatic cancer (28 [95 % CI -34 to -23] days shorter, respectively).

[CONCLUSION] These findings enrich the best available evidence on cancer diagnosis through emergency self-referral and showed that, compared with the emergency referral pathway from primary care, these patients had a significantly increased likelihood of presenting with symptoms that are strongly predictive of cancer. Since the starting point for the diagnostic interval in these patients is their emergency presentation, comparing it with that of those referred from primary care as an emergency is likely to result in biased data.