Page 34 - JSOM Winter 2025

predictor of discharge status dead, P=.04, OR 8.27 (95% CI The HEMS EMS service studied has started carrying fibrino-
1.15–59.0), and those meeting code red criteria had an 8.27 gen and one of its clinical indications for administration during
times greater odds of death. The model failed to converge for transport is SI >1.4. Due to uncertainty and to provide greater
CRCpre, when CRCpre was removed from the analysis com- clarity, further exploratory investigation was conducted. The
parable results were produced, CRC P=.03, OR 7.8 (95% CI levels initially dropped from EDSI and EDSBP were collapsed
1.2–50.4). into the next category (EDSBP categories, 0–74, 75–90, and
>90; EDSI categories, <1, 1–1.49, and ≥1.5).
INR-ED predicted from CRC, CRCpost, and SBP returned
a significant model χ² =27.0, P<.001. McFadden R =0.44. With the collapsed categories and ITED included, a signifi-
2
5
However, the model failed to converge for all variables and cant model was produced χ² =17.4, P=.004. McFadden R =
2
5
none of the predictor variables were significant. 0.21= .16. A significant effect was found for EDSI where those
with an EDSI ≥1.5 had 14.088.93 times greater odds of having
A follow-up analysis conducted with INR-ED treated as a con- fibrinogen <1.6g/L than those with an EDSI <1, P=.003006,
tinuous variable using multivariate linear regression produced OR 14.08 (95% CI 2.4–82.2). When compared with the EDSI
a significant overall model, P=.01, with CRCpost as a signifi- 1–1.49 group, those with an EDSI ≥1.5 had 96.9 times greater
cant predictor of INR level, P=.012, standardized β=0.71(95% odds of having fibrinogen <1.6g/L, P=.02, OR 96.9 (1.6 –
CI 0.16–1.25). The mean INR level for the CRCpost NO code 61.43 – 36.1). No effect was observed between the EDSI <1
red remains was 1.16 (SD 0.17), median 1.1g/L, while for YES and EDSI 1–1.49 groups, P=.669 OR 1.4 1.4 (95% CI 0.38–
code red remains was 1.55 (SD 0.62), median 1.4g/L. 5.28). No effect was observed for EDSBP.

The prediction of MHP-ED from the primary predictors When ITED was dropped from the analysis the same overall
returned a non-significant model, χ² =8.57, P=.5. McFad- effects remained χ² =12.5, P=.014. McFadden R =0.16, EDSI
2
4
9
den R =0.1. For surgery/angiography predicted from CRC, ≥1.5 vs. <1; P=.003006, OR 13.9 (95% CI 2.4–81.67) and
2
CRCpost, CRCpre, and SBP a non-significant model was re- EDSI ≥ 1.5 versus 1–1.49, P=.01402, OR 6.98 (1.6 – 613 –
turned χ² =2.81, P=.83. 36.4), indicating no effect of ITED.
6
The model for the prediction of 14-day mortality from the pri- Collapsing the categories to include the sickest patients indi-
mary predictor variables was uninterpretable due to a failure cates there may be an effect of EDSI on fibrinogen levels. This
to converge. This is a product of the small sample size and an exploratory portion of the analysis points to the need for fur-
inadequate number of cases per variable. ther research with larger samples to include a larger propor-
tion of the sickest patients (EDSI >2) to further refine at what
Tertiary Analysis point EDSI may have an effect on fibrinogen levels, helping to
The effect of CRCED, EDSI, EDSBP, and ITED on the first fi- inform practice.
brinogen level in the ED was examined. The initial model failed
to converge as EDSBP did have not enough cases in the first A linear regression was performed with age and first fibrino-
level (SBP 0/PEA-59, n=2); EDSI did not have enough cases in gen level treated as continuous variables. A significant overall
the last level (SI >2, n=2), and EDSBP was predominated by model was produced, P=.013, R =0.09. Age was a significant
2
Level 4 (SBP >90, n=47, 70.1%) leading to collinearity with predictor of fibrinogen level, P=.013, standardized β=0.31
CRCED A, specifically Code Red not met (SBP >90). As such, (95% CI 0.07–5.6). With age treated as a categorical predic-
EDSBP Level 1, EDSI Level 4, and CRED were dropped from tor variable (geriatric vs. non-geriatric), age was no longer a
the analysis. CRCED was examined separately to determine if significant predictor, P=.115, standardized β=0.54 (95% CI
there was an effect for CRCED. –0.13 to 1.22). No significant results were found when fi-
brinogen was treated as categorical, P=.12, OR 0.97 (95% CI
A non-significant model was returned χ² =7.13, P=.21. Mc- 0.95–1.01), or both fibrinogen and age were treated as cate-
5
Fadden R =0.09, indicating poor model fit. There were no ef- gorical, P=.1, OR 0.17 (95% CI 0.02–1.4).
2
fects for any of the predictor variables. When CRCED was
examined separately from EDSI, EDSBP, and ITED, the effect PMID: 41402981; DOI: 10.55460/T0ZU-OA0G
of code red remaining at ED arrival on fibrinogen <1.6g/L
was nearly significant, χ² =3.41, P=.065. McFadden R =0.04,
2
1
CRCED P=.06, OR 2.8 (95% CI 0.93–8.5).

32 | JSOM Volume 25, Edition 4 / Winter 2025