TY - JOUR T1 - Understanding the consequences of GP referral thresholds: taking the instrumental approach JF - BMJ Quality & Safety JO - BMJ Qual Saf SP - 309 LP - 311 DO - 10.1136/bmjqs-2022-015557 VL - 32 IS - 6 AU - Jen Lewis AU - Christopher Burton Y1 - 2023/06/01 UR - http://qualitysafety.bmj.com/content/32/6/309.abstract N2 - Trade-offs between patient safety and efficient use of healthcare services occur in clinical decisions across all forms of healthcare. In the case of acutely unwell older patients, decisions about referral to hospital involve trade-offs between the safety associated with inpatient hospital treatment and the burden on both the patient and health system associated with hospital admission. In many healthcare systems, these decisions are largely made by general practitioners (GPs), often without first-hand knowledge of the patient, especially when presentation is in an out-of-hours setting. This raises three questions: how much do practitioners vary in their decisions? is this variation systematic (ie, after adjusting for patient and context, do some practitioners have a greater or lesser tendency to refer than others?)? and are those who make fewer referrals making better decisions (ie, admitting those who will benefit, keeping at home those who will not)?In this issue of BMJ Quality & Safety, Svedahl and colleagues address these questions within a large, routinely collected dataset from Norway using an instrumental variables (IV) analysis.1 The authors used IV analysis as they aimed to delineate the causal relationship, rather than simply to show an association, between referral by out-of-hours GPs and older patients’ subsequent use of healthcare services and mortality up to 6 months. This is important because the relationship between referral and mortality depends on both the patients’ initial condition (sicker patients are more likely to be admitted, thus introducing confounding by indication) and the treatment (following referral) they receive.Nearly 500 000 patients aged 65 years or older were included in the study. While all patients were included, the nature of the analysis (explained further) facilitates a focus on those patients whose referrals could be attributed to their GP’s ‘referral threshold’ or tendency to refer more or fewer patients. For these referred patients, … ER -