Objective: Methods that include the time-varying nature of healthcare-associated infections (HAIs) avoid biases when estimating increased risk of death and excess length of stay. We determined the excess mortality risk and length of stay associated with HAIs among inpatients in Singapore using a multistate model that accommodates the timing of key events.
Design: Analysis of existing prospective cohort study data.
Setting: Seven public acute-care hospitals in Singapore.
Patients: Inpatients reviewed in a HAI point-prevalence survey (PPS) conducted between June 2015 and February 2016.
Methods: We modeled each patient's admission over time using 4 states: susceptible with no HAI, infected, died, and discharged alive. We estimated the excess mortality risk and length of stay associated with HAIs, with adjustment for the baseline characteristics between the groups for mortality risk.
Results: We included 4,428 patients, of whom 469 had ≥1 HAI. Using a multistate model, the expected excess length of stay due to any HAI was 1.68 days (95% confidence interval [CI], 1.15-2.21 days). Surgical site infections were associated with the longest excess length of stay of 4.68 days (95% CI, 2.60-6.76 days). After adjusting for baseline differences, HAIs were associated with increased hazards of in-hospital mortality (adjusted hazard ratio [aHR], 1.32; 95% CI, 1.09-1.65) and decreased hazards in being discharged (aHR, 0.75; 95% CI, 0.67-0.84).
Conclusions: HAIs are associated with increased length of hospital stay and mortality in hospitalized patients. Avoiding nosocomial infections can improve patient outcomes and free valuable bed days.