A predictive model of days from infection to discharge in patients with healthcare-associated urinary tract infections: a structural equation modelling approach.

Background: Length of stay (LOS) in hospital is an important component of describing how costs change in relation to healthcare-associated infection and this variable underpins models used to evaluate cost. It this therefore imperative that estimations of LOS associated with infections are performed accurately.

Aim: To test the relationships between the size of hospital, age, and patient comorbidity on days from admission to infection and days from infection to discharge in patients with a healthcare-associated urinary tract infection (HAUTI), using structural equation modelling (SEM).

Methods: A non-current cohort study in eight hospitals in New South Wales, Australia. All patients admitted to the hospital for >48 h and who acquired a HAUTI were included.

Findings: From the 162,503 eligible patient admissions, 2821 (1.73%) acquired a HAUTI. SEM showed that the proposed model had acceptable fit indices for the combined sample (GFI = 1.00; AGFI = 1.00; NFI = 1.00; CFI = 1.00; RMSEA = 0.000). The main findings showed that age of patient had a direct association with days from admission to infection and with days from infection to discharge. Patient comorbidity had direct links to the variables days from admission to infection and days from infection to discharge. Multi-group analysis indicated that the age of male patients was more influential on the factor days from admission to infection when compared to female patients. Furthermore, the number of comorbidities was significantly more influential on days from admission to infection in male patients than in female patients.

Conclusion: As the first published study to use SEM to explore a healthcare-associated infection and the predictors of days from infection to discharge in hospital, we can confirm that accounting for the timing of infection during hospitalization is important and that patient comorbidity influences the timing of infection.