Table 4.

Association between benzodiazepines and mortality in patients initiating hemodialysis (n=69,368) between 2013 and 2014 stratified by age, sex, race, and opioid codispensing

Exposure	No. of Deaths	Follow-Up Time, person-yr	Unadjusted aHR (95% CI)	Adjusted aHR (95% CI)	P Value for Interaction
Any benzodiazepine					0.001
No opioid/no benzodiazepine	2556	182,800	Reference	Reference
No opioid/benzodiazepine	485	41,273	1.32 (1.20 to 1.44)	1.22 (1.13 to 1.32)
Opioid/no benzodiazepine	2071	141,526	Reference	Reference
Opioid/benzodiazepine	190	8105	2.35 (2.04 to 2.71)	1.66 (1.46 to 1.90)
Short-acting benzodiazepine					<0.001
No opioid/no benzodiazepine	2514	174,636	Reference	Reference
No opioid/benzodiazepine	420	31,186	1.50 (1.36 to 1.65)	1.34 (1.23 to 1.45)
Opioid/no benzodiazepine	2094	143,450	Reference	Reference
Opioid/benzodiazepine	167	6181	2.74 (2.35 to 3.19)	1.90 (1.65 to 2.18)
Long-acting benzodiazepine					0.72
No opioid/no benzodiazepine	2305	157,870	Reference	Reference
No opioid/benzodiazepine	67	10,280	0.74 (0.58 to 0.94)	0.83 (0.69 to 1.00)
Opioid/no benzodiazepine	2238	147,590	Reference	Reference
Opioid/benzodiazepine	23	2041	1.11 (0.74 to 1.67)	0.89 (0.64 to 1.24)

Use of benzodiazepines and other medications was treated as time varying, and all models were adjusted. The results below are from three separate models; all models were adjusted for age, sex, race, prescription (antidepressants and CNS depressants), and comorbidities. CNS depressants included sedatives, muscle relaxants, and antipsychotics. Comorbidities included diabetes mellitus, cardiovascular disease, peripheral vascular disease, hypertension, COPD, smoking history, cancer, drug abuse, inability to ambulate, institutionalized, and obesity.