Predictive scoring models for persistent gram-negative bacteremia that reduce the need for follow-up blood cultures: a retrospective observational cohort study

Table 3 Independent risk factors and assigned scores used to build the predictive scoring model for positive follow-up blood culture in gram-negative bacteraemia, according to eradicable and non-eradicable sources of infection

Eradicable source of infection
	Beta-coefficient	Odds ratio (95% CI)	p-value	Assigned score
ESBL-producing microorganism infection	1.001	2.720 (1.179–6.271)	0.019	+ 1
CRBSI	1.374	3.95 (1.522–10.255)	0.005	+ 1
Unfavourable treatment response^a	0.802	2.229 (1.262–3.937)	0.006	+ 1
qSOFA score ≥ 2 on the day of FUBC	0.864	2.371 (1.034–5.438)	0.041	+ 1
Effective antibiotics administration before the day of FUBC	−1.007	0.365 (0.164–0.814)	0.014	−1
Adequate source control before the day of FUBC	−1.983	0.138 (0.064–0.294)	0.000	−2
Non-eradicable source of infection
ESRD on HD	1.406	4.081 (1.331–12.515)	0.014	+ 1
Unfavourable treatment response^a	0.802	2.229 (1.262–3.937)	0.006	+ 1
Effective antibiotics administration before the day of FUBC	−2.015	0.133 (0.069–0.258)	0.000	−2

CI Confidence interval, ESBL Extended-spectrum beta-lactamase, CRBSI Catheter-related bloodstream infection, qSOFA Quick sequential organ failure assessment, FUBC Follow-up blood culture, ESRD End-stage renal disease, HD Haemodialysis
^aUnfavourable treatment response was defined as positivity for at least 2 variables among the presence of fever, aggravated leucocytosis, and no decrease of C-reactive protein on the day of FUBC

ISSN: 1471-2334