Abstract
BACKGROUND: In Australia, as in many high income countries, there has been a movement to improve out-of-hospital care. If primary care improvements can yield appropriately lower hospital use, this would improve productive efficiency. This is especially important among 'high cost users', a small group of patients accounting for disproportionately high hospitalisation costs. This study aimed to assess the association between regularity of general practitioner (GP) care and 'high use' hospitalisation. METHODS: This retrospective, cohort study used linked administrative and survey data from the 45 and Up Study, conducted in New South Wales, Australia. The exposure was regularity of GP care between 1 July 2005 and 30 June 2009, categorised by quintile (lowest to highest). Outcomes were 'high use' of hospitalisation (defined as ≥3 and ≥ 5 admissions within 12 months), extended length of stay (LOS, ≥30 days), a combined metric (≥3 hospitalisations in a 12 month period where ≥1 hospitalisation was ≥30 days) and 30-day readmission between 1 July 2009 and 31 December 2017. Associations were assessed using multivariable logistic regression. Potential for outcome prevention in a hypothetical scenario where all individuals attain the highest GP regularity was estimated via the population attributable fraction (PAF). RESULTS: Of 253,500 eligible participants, 15% had ≥3 and 7% had ≥5 hospitalisations in a 12-month period. Five percent of the cohort had a hospitalisation lasting ≥30 days and 25% had a readmission within 30 days. Compared with lowest regularity, highest regularity was associated with between 6% (p < 0.001) and 11% (p = 0.027) lower odds of 'high use'. There was a 7-8% reduction in odds for all regularity levels above 'low' regularity for LOS ≥30 days. Otherwise, there was no clear sequential reduction in 'high use' with increasing regularity. The PAF associated with a move to highest regularity ranged from 0.05 to 0.13. The number of individuals who could have had an outcome prevented was estimated to be between 269 and 2784, depending on outcome. CONCLUSIONS: High GP regularity is associated with a decreased likelihood of 'high use' hospitalisation, though for most outcomes there was not an apparent linear association with regularity.