Table 4

Association between patient/clinical characteristics and consumption of a delayed antibiotic in 150 patients who consulted with 66 clinicians^a with LRTI and were issued with a delayed prescription

Variable	Odds ratio	95% CI	P-value
Patient characteristics
Age (decades)	1.15	0.86 to 1.54	0.343
Sex: Female (Ref) Male	0.55	0.22 to 1.38	0.204
No comorbidities (Ref)
At least one comorbidity	0.55	0.17 to 1.82	0.326
Employed (Ref)
Not employed	1.00	0.39 to 2.54	1.000

Patient illness information
Clinician rated symptom severity at consultation^b	1.13	0.78 to 1.64	0.545
Patient rated symptom severity on day 1^b	0.99	0.72 to 1.35	0.938
Waited <7 days before consulting (Ref)
Waited ≥7 days before consulting	0.86	0.35 to 2.13	0.746

Clinician working diagnosis
Lower respiratory tract infection (Ref)
Asthma/COPD-related illness	0.35	0.05 to 2.60	0.308
Upper respiratory tract infection	0.35	0.12 to 1.00	0.050
Non-specific respiratory tract infection	0.21	0.05 to 0.97	0.046
Viral/other illness	0.11	0.02 to 0.53	0.005

Duration of advised delay
<7 days (Ref)
≥7 days	0.38	0.11 to 1.29	0.120

Patient behaviour
Patient did not want antibiotics (Ref)
Patient wanted antibiotics	2.51	1.06 to 5.90	0.035

Clinician perception ‘This patient wanted me to prescribe antibiotics for them’
Clinician does not agree (Ref)
Clinician agrees	0.38	0.15 to 1.00	0.050

Clinician perception: ‘Antibiotics will help this patient get better quicker’
Clinician does not agree (Ref)
Clinician agrees	1.30	0.52 to 3.22	0.573

COPD = chronic obstructive pulmonary disease.
↵a The clinician-level intracluster correlation coefficient (using the standard π2/3 estimator) was 0.10.
↵b OR for a 10 percentage point increase