Excess length of stay and readmission following hospital-acquired bacteraemia: a population-based cohort study applying a multi-state model approach

Abstract

ObjectivesPopulation-based estimates of excess length of stay after hospital-acquired bacteraemia (HAB) are few and prone to time-dependent bias. We investigated the excess length of stay and readmission after HAB. MethodsThis population-based cohort study included the North Denmark Region adult population hospitalized for ≥48 hours, from 2006 to 2018. Using a multi-state model with 45 days of follow-up, we estimated adjusted hazard ratios (aHRs) for end of stay and discharge alive. The excess length of stay was defined as the difference in residual length of stay between infected and uninfected patients, estimated using a non-parametric approach with HAB as time-dependent exposure. Confounder effects were estimated using pseudo-value regression. Readmission after HAB was investigated using the Cox regression. ResultsWe identified 3457 episodes of HAB in 484 291 admissions in 205 962 unique patients. Following HAB, excess length of stay was 6.6 days (95% CI, 6.2–7.1 days) compared with patients at risk. HAB was associated with decreased probability of end of hospital stay (aHR, 0.60; 95% CI, 0.57–0.62) driven by the decreased hazard for discharge alive; the aHRs ranged from 0.30 (95% CI, 0.23–0.40) for bacteraemia stemming from ‘heart and vascular’ source to 0.72 (95% CI, 0.69–0.82) for the ‘urinary tract’. Despite increased post-discharge mortality (aHR, 2.76; 95% CI, 2.38–3.21), HAB was associated with readmission (aHR, 1.42; 95% CI, 1.31–1.53). ConclusionHAB was associated with considerably excess length of hospital stay compared with hospitalized patients without bacteraemia. Among patients discharged alive, HAB was associated with increased readmission rates.