Table 2.

Univariate Cox Regression Analysis of NCCN Guideline Performance

	SEER-Medicare (Derivation Cohort)		VHA (External Validation Cohort)
Proposed Risk Factor	HRa	P Valuea	HRa	P Valuea
Individual
Prior VTE	3.20	<.01	3.85	<.01
Obesity	0.86	.63	1.26	.24
Central venous catheter or pacemaker	0.75	.68	0.34	.28
Cardiac disease (CHF, MI, CA)	1.00	.98	0.92	.75
Chronic renal disease	1.14	.38	1.09	.75
Diabetes	1.14	.38	0.76	.23
Acute infection	1.42	.20	1.60	.64
Immobilization	1.24	.71	NE	N/A
General surgery or anesthesia	1.82	.12	2.16	.07
Trauma	0.96	.83	1.30	.39
Erythropoietin	0.89	.61	1.14	.58
Blood clotting disorders	0.96	.86	1.67	.31
Myeloma-related
Diagnosis of myeloma per seb	N/A	N/A	N/A	N/A
Hyperviscosityc	N/A	N/A	N/A	N/A
Therapy-related
High-dose dexamethasone, ≥480 mg	1.09	.55	1.89	<.01
Doxorubicin	NE	N/A	NE	N/A
Multiagent (cytotoxic) chemotherapy	2.05	.48	0.91	.77
Risk stratificationd	HR (high/low) = 1.21 (P=.17)		HR (high/low) = 1.41 (P=.07)
	c-index = 0.51 (95% CI, 0.48–0.54)		c-index = 0.54 (95% CI, 0.50–0.59)

Abbreviations: CA, cardiac arrhythmia; CHF, congestive heart failure; HR, hazard ratio; MI, myocardial infarction; N/A, not applicable; NE, no event; VHA, Veterans Health Administration; VTE, venous thromboembolism.

^aHRs and P values were estimated from univariate Cox regression analyses because risk factors from the guidelines were not derived statistically.

^bDiagnosis of myeloma per se was not assigned any point because all patients had a myeloma diagnosis.

^cHyperviscosity was a clinical definition that could not be defined accurately using claims data.

^d“High-risk” was defined as ≥2 risk factors in the individual or myeloma-related categories or any factor in the therapy-related category according to the NCCN Guidelines. The HR and Harrell’s c-index were directly estimated from the regression of the final risk groups.