Table 3.

Association between frequent emergency department use and patient-, medication-, system- and opioid-related characteristics among patients who filled an opioid prescription before their first post-discharge emergency department visit (n = 475)

Variable	Adjusted HR1 (95% CI)
Healthcare use characteristics
Emergency department (ED) visits
0	Reference
≥1	1.73 (1.08–2.75)
Ambulatory visits
<15	Reference
≥15	1.54 (1.06–2.34)
Chemotherapy
No	Reference
Yes	2.05 (1.40–3.00)
Comorbidities
Renal disease
No	Reference
Yes	0.46 (0.20–1.26)
Heart disease
No	Reference
Yes	1.95 (1.27–2.97)
Diabetes
No	Reference
Yes	1.75 (1.14–2.69)
Lung cancer
No	Reference
Yes	1.71 (1.07–2.74)
Medication use
Active opioid prescription at admission
No	Ref
Yes	0.83 (0.52–1.32)
In-hospital characteristics
Antidepressant use
No	Ref
Yes	1.19 (0.76–1.87)
Type of surgery
No surgery	Reference
Cardiac	0.18 (0.06–0.49)
Gastrointestinal	0.35 (0.14–0.93)
Thoracic	0.42 (0.26–0.70)
Unrelated	0.18 (0.04–0.80)
Admission to the ICU
No	Reference
Yes	2.20 (0.99–4.88)
Discharge prescription for an opioid medication
No	Reference
Yes	1.53 (0.95–2.47)

HR = hazard ratio, CI = confidence interval, LOS = length of hospital stay

¹Adjusted for all the variables in the table. Covariate selection was based on backward selection, using a p-value of 0.1 for variable inclusion, as well as a number of characteristics with low p-values. The final number of variables selection was based on the events-per-variable rule, where the total number of variables in the final model did not fall beyond an event-to-variable ratio of 10. Statistically significant findings are bolded. In total, there were 92 patients (19.4%) who had died before having the opportunity to become frequent emergency department users.