What’s missing in Unscheduled Care Big Data? A free-text analysis of General Practice Out-of-Hours (GPOOH) Attendances by People who Die from Cancer.
Access to care General Practice Out-of-Hours (GPOOH) is key to supporting patients with advance cancer in living, and dying, well in the community. Unfortunately, the clinical content in GPOOH data is poorly coded; over half of all consultations for people who die from cancer are coded as ‘other’ or having missing coding. Consequently, very little is known about why people with advanced cancer access GPOOH.
This free-text analysis was part of a wider retrospective cohort study involving all 2,443 people who died from cancer in NHS Tayside from 03/2013-06/2015. This cohort was defined using General Registry Office (GRO) data to identify all people with cancer in position 1 on heir death certificates, and was cross-linked to demography and cancer registry datasets using Community Health Index (CHI) numbers. Clinical information for all contacts with GPOOH in patients’ last year of life was obtained. For the 2,443 people in the cohort this represented 5,749 GPOOH attendances. The free text were cleaned and anonymised through the Health Informatics Centre (HIC) and stored in the virtual SafeHaven platform. This free-text analysis used a randomly-generated sample of 575 attendances, representing 10% of the cohort’s GPOOH attendances. All consultations were read, analysed and coded into ‘main reason for presentation’, presence of palliative care symptoms, purpose of consultation (cancer-related, pain-related or palliative-care-related), and evidence of Anticipatory Care Planning (Do Not Attempt CPR forms, Just in Case medication prescribing, presence of electronic palliative care summaies). Half (n=288 attendances) were double-coded by two reviewers. A Kappa analysis was performed on all allocated variables. Only variables scoring as having moderate-to-perfect inter-rater agreement were included in the final analysis.
Palliative care and pain accounted for half of the assigned presenting complaints. More than half of GPOOH attendances recorded at least one key or additional palliative care symptom. Breathlessness, nausea and vomiting and cough were the commonest key palliative care symptoms recorded in GPOOH consultation free texts. Agitation, oedema and weight loss were the commonest additional palliative care symptoms recorded in consultation free text. The underlying reason for consultation was cancer-related in over seventy percent attendances, pain-related over sixty percent of attendances and palliative care-related in over forty percent of attendances. Anticipatory care planning paperwork and medication were poorly recorded in the notes and often absent. There was a significant association between presence of eKIS summaries, DNACPR forms and JIC medication.
This is the first free-text analysis of GPOOH consultations in patients who die from cancer, and fills a knowledge gap in determining why people with advanced cancer use GPOOH. It identified a number of anticipatory care planning and prescribing factors which are associated with increased GPOOH use, and which could be targeted in order to minimise avoidable GOOH use by cancer decedents.