Bivariable Modelling | Multivariable Modelling* | ||||
OR (CI) | P value | OR (CI) | P value | ||
30-day mortality | |||||
Continuous variables (ORs and CIs are per +1 score) | |||||
Females (n=142): | GWTG-HF | 1.06 (0.99 to 1.13) | 0.12 | 1.05 (0.98 to 1.12) | 0.18 |
IMRS (30 day) | 1.15 (0.98 to 1.35) | 0.09 | 1.13 (0.96 to 1.34) | 0.14 | |
Males (n=158): | GWTG-HF | 1.09 (1.01 to 1.18) | 0.020 | 1.10 (1.02 to 1.19) | 0.011 |
IMRS (30 day) | 1.27 (1.06 to 1.52) | 0.008 | 1.25 (1.04 to 1.49) | 0.015 | |
Categorical variables (combined females and males, n=300) | |||||
GWTG-HF: | Tertile 1 | (referent) | ----- | (referent) | |
Tertile 2 | 0.62 (0.19 to 1.97) | 0.41 | 0.83 (0.25 to 2.77) | 0.76 | |
Tertile 3 | 2.18 (0.84 to 5.68) | 0.11 | 2.62 (0.96 to 7.12) | 0.06 | |
IMRS† (30 days): | Low risk | (referent) | ----- | (referent) | |
Moderate risk | 3.16 (0.85 to 11.78) | 0.09 | 2.66 (0.70 to 10.13) | 0.15 | |
High risk | 8.25 (2.19 to 31.09) | 0.002 | 6.69 (1.75 to 25.60) | 0.005 | |
1-year mortality | |||||
Continuous variables (ORs and CIs are per +1 score) | |||||
Females (n=142): | GWTG-HF | 1.03 (0.98 to 1.08) | 0.29 | 1.02 (0.97 to 1.08) | 0.38 |
IMRS (1 year) | 1.11 (0.98 to 1.27) | 0.11 | 1.11 (0.97 to 1.26) | 0.13 | |
Males (n=158): | GWTG-HF | 1.08 (1.02 to 1.14) | 0.006 | 1.08 (1.02 to 1.14) | 0.005 |
IMRS (1 year) | 1.33 (1.14 to 1.55) | <0.001 | 1.31 (1.12 to 1.53) | <0.001 | |
Categorical variables (Combined females and males, n=300) | |||||
GWTG-HF: | Tertile 1 | (Referent) | ----- | (Referent) | |
Tertile 2 | 1.34 (0.65 to 2.77) | 0.43 | 1.51 (0.72 to 3.19) | 0.28 | |
Tertile 3 | 2.17 (1.08 to 4.37) | 0.029 | 2.27 (1.12 to 4.63) | 0.023 | |
IMRS† (1 year): | Low risk | (Referent) | ----- | (Referent) | ----- |
Moderate risk | 2.70 (0.60 to 12.18) | 0.20 | 2.66 (0.59 to 12.08) | 0.20 | |
High risk | 5.98 (1.32 to 27.17) | 0.021 | 5.43 (1.19 to 24.84) | 0.029 |
All of these analyses only evaluated n=300 patients (or sex-specific subsets) for whom GWTG-HF was available.
*Study covariables considered for entry into multivariable models were: sex, race (African American, American Indian/Alaska Native, Asian, Native Hawaiian/Pacific Islander, White, unknown), admit source (another hospital, clinic, emergency department, non-healthcare facility, skilled nursing facility, other health facility), insurance type (private/HMO, Medicaid, Medicare, Medicare advantage, none), prior heart failure (HF) diagnosis, anaemia, atrial fibrillation, atrial flutter, chronic obstructive pulmonary disease/asthma, coronary artery disease (CAD), depression, hyperlipidaemia, hypertension, insulin-dependent diabetes, non-insulin-dependent diabetes, dialysis, pacemaker, peripheral vascular disease (PVD), prior coronary artery bypass grafting, prior myocardial infarction, prior percutaneous coronary intervention, rales, renal insufficiency, sleep disordered breathing, smoking history, stroke, valvular heart disease, EKG QRS duration, ACE inhibitor, aldosterone antagonist, beta-blocker, bumex, demadex, edecrin, factor Xa inhibitor, hydralazine nitrate, lasix, loop diuretics, metolazone, and type of heart failure diagnosis (primary diagnosis without CAD, primary diagnosis with CAD, or secondary diagnosis). Multivariable models entered the following covariables in the final models (in addition to the two risk score variables): for 30-day Mortality: PVD and beta-blocker; for 1-year mortality: insulin-dependent diabetes, hyperlipidaemia, history of smoking, and beta-blocker.
†IMRS values were categorised into low risk, moderate risk, and high risk based on criteria from 2009 for 30-day and 1-year risk scores.12 Once the categories were assigned based on sex-specific and time frame-specific criteria, the data for females and males could be combined because the criteria for assigning thresholds of risk were the same for both sexes although the numeric distribution of the scores and the actual thresholds were different.