Skip to main content

Table 1 Multivariable analysis of center-level factors potentially associated with changes in the cumulative incidence of invasive candidiasis in European ICUs

From: Incidence and outcome of invasive candidiasis in intensive care units (ICUs) in Europe: results of the EUCANDICU project

Center-level variables* Number of IC episodes Number of ICU admissions Cumulative incidence (IC episodes/1000 ICU admissions) CIR (95% CI) p
Year of study      0.313
 2015 271 40,642 6.67 Ref  
 2016 299 40,003 7.47 1.18 (0.85–1.63)  
Type of ICU      0.073
 Medical (n = 5) 149 7828 19.03 Ref  
 Surgical (n = 3) 51 29,087 1.75 0.10 (0.01–0.76)§  
 Mixed (medical plus surgical, n = 15) 370 43,730 8.46 0.40 (0.10–1.63)  
  1. The sample size was of 46 observations (2 for each of the 23 participating ICUs, one in 2015 and one in 2016). Non-independence was accounted by adding center as random effect. The model also included an interaction term (year of study × type of ICU), with p for interaction 0.761. Results of the main model (including both candidemia and IAC) were confirmed in a subgroup analysis including only patients with candidemia and not IAC (n = 422), suggesting that the observed increased cumulative incidence of IC in medical ICU vs. surgical ICU was mainly due to candidemia and not IAC: year of study (CIR 1.09, 95% CI 0.77–1.55, p 0.619), type of ward (p 0.005, with CIR 0.03 for surgical vs. medical, 95% CI 0.00–0.31, and CIR 0.34 for mixed vs. medical, 95% CI 0.07–1.55), year of study × type of ward (p for interaction 0.782), center as random intercept (standard deviation of the random effect = 1.410; model β0 = − 3.937). Results of the subgroup analysis of patients with IAC (n = 148) were as follows: year of study (CIR 1.77, 95% CI 0.79–4.21, p 0.177), type of ward (p 0.798, with CIR 0.94 for surgical vs. medical, 95% CI 0.07–12.51, and CIR 0.82 for mixed vs. medical, 95% CI 0.12–5.37), year of study × type of ward (p for interaction 0.290), center as random intercept (standard deviation of the random effect = 1.565; model β0 = − 6.285). Stratified cumulative incidences for countries in the entire study period was as follows: Italy (2 medical and 7 mixed ICUs), 89.62 episodes per 1000 ICU admissions (range 1.20–114.21); France (2 surgical, 1 medical, and 1 mixed ICUs), 11.85 episodes per 1000 ICU admissions (range 0.62–27.63); Greece (1 medical and 1 mixed ICUs), 30.79 per 1000 ICU admissions (range 7.50–45.73); Belgium (1 medical ICU), 9.28 episodes per 1000 ICU admissions; Czech Republic (1 surgical ICU), 0.90 per 1000 ICU admissions; Germany (1 mixed ICU), 42.43 episodes per 1000 hospital admissions; Ireland (1 mixed ICU), 5.63 episodes per 1000 ICU admissions; Portugal (1 mixed ICU), 9.33 episodes per 1000 ICU admissions; Spain (1 mixed ICU), 10.46 episodes per 1000 ICU admissions; The Netherlands (1 mixed ICU), 2.29 episodes per 1000 ICU admissions; UK (1 mixed ICU), 41.67 episodes per 1000 ICU admissions
  2. CI confidence intervals, CIR cumulative incidence ratio, IC invasive candidiasis, ICU intensive care unit, IQR interquartile range
  3. *The model also includes center as a random intercept (standard deviation of the random effect = 1.293; model β0 = − 3.763)
  4. §p = 0.022 for the subgroup comparison surgical vs. medical