Skip to main content
. 2018 Dec 5;13:147. doi: 10.1186/s13012-018-0834-6

Table 4.

Effect of the intervention on Rx cost, HSU cost, and total cost (per patient)

Variable No. of patients (EDs) Mean (SE) Incremental cost, raw (95%CI) Incremental cost, adjusted (95%CI)
Rx Control Rx Control
NET
 Rx Cost
  Base-case 893 (14) 1050 (17) $139.0 (1.6) $0.8 (0.0) $138.20 (135,141)
  No URG 893 (14) 1050 (17) $139.0 (1.6) $0.8 (0.0) $138.20 (135,141)
 HSU Cost
  Base-case 893 (14) 1050 (17) $825.9 (44) $777.4 (33) $48.52 (− 57, 154) $23.86 (− 106,153)
  No URG 893 (14) 1050 (17) $354.0 (42) $315.7 (31) $38.35 (− 62, 139) -$1.08 (− 125,122)
 Total Cost
  Base-case 893 (14) 1050 (17) $964.9 (44) $778.2 (33) $186.72 (81, 292) $169.89 (43, 297)
  No URG 893 (14) 1050 (17) $493.0 (42) $316.5 (31) $176.55 (76, 277) $159.96 (39, 281)
NET-Plus, including post-NET HSU
 Rx Cost
  Base-case 126 (10) 218 (14) $143.8 (4.6) $0.8 (0.0) $142.96 (136,150)
  No URG 126 (10) 218 (14) $143.8 (4.6) $0.8 (0.0) $142.96 (136,150)
 HSU Cost
  Base-case 126 (10) 218 (14) $980.8 (235) $738.5 (72) $242.38 (− 156,640) $341.78 (− 58, 742)
  No URG 126 (10) 218 (14) $526.6 (233) $276.5 (66) $250.06 (− 137,638) $338.54 (− 56, 733)
 Total Cost
  Base-case 126 (10) 218 (14) $1124.6 (235) $739.3 (72) $385.34 (− 12, 783) $505.06 (96, 915)
  No URG 126 (10) 218 (14) $670.3 (233) $277.3 (66) $393.02 (− 6, 780) $543.49 (116, 971)

†Incremental cost, raw = unconditional difference in cost per patient due to exposure to the intervention from two sample t test with equal variances

‡Incremental cost, adjusted = difference in cost due to exposure to the intervention; adjusted for the following minimisation factors and pre-specified confounders: age, sex, out_of_hours, rurality, mTBI protocol, ED participation in NET-Plus, and annual_presentation_rate. Estimates derived from margins, dydx(i.study_group) after xtgee, family(gamma) link(log) corr(exchangeable) vce(robust) to account for within-cluster correlation structure and yielding cluster-robust standard errors even if the correlation structure is misspecified