LAUSR

Background: Previous studies of associations between home visits by general practitioners and end-of-life care for cancer patients have been subject to confounding. Aim: To analyse associations between general practitioners’ propensity to pay home visits and the likelihood of hospitalisation and dying out of hospital among their cancer patients. Design: A national register cohort study with an ecological exposure. Standardised incidence rates of general practitioner home visits were calculated as a measure for propensity. Practices were grouped into propensity quartiles. Associations between propensity groups and end-of-life outcomes for cancer patients aged 40 or above were calculated. Setting/participants: Danish general practitioners and citizens aged 40 or above were included from 2003 to 2012. Results: We included 2670 practices with 2,518,091 listed patients (18,364,679 person-years); of whom 116,677 died from cancer. General practitioners were grouped into quartiles based on the general practitioners’ propensity to pay home visits, which varied 6.6-fold between quartiles. Cancer patients in Group 4 (highest propensity) were less hospitalised than patients in Group 1 (lowest propensity): odds ratio: 1.13 (95% confidence interval: 1.08; 1.17) for ⩽3 bed-days and odds ratio: 0.95 (0.91–0.99) for ⩾20 bed-days. Group 4 patients were more likely to die out of hospital (odds ratio: 1.20 (1.16; 1.24)) than Group 1 patients. Conclusion: We found a dose–response-like association between general practitioners’ higher propensity to pay home visit and their patients’ likelihood of less end-of-life hospitalisation and more often dying out of hospital.