Comparing strategies for United States veterans' mortality ascertainment

Table 1 Potential Bias Associated with Alternative Strategies For Veterans' Mortality Ascertainment *

	BIRLS Only Strategy			Death Certificate Only Strategy
	Cases identified by BIRLS	Additional cases identified by death certificates	P-value	Cases identified by death certificates	Additional cases identified by BIRLS	P-value
Number of cases	6,387	2,426		8,124	689
Age (years)	70.86	71.15	0.8253	70.95	70.79	0.6891
Gender
Male	98	97	0.2733	98	98	0.8662
Race / Ethnicity
White	57	54		58	31	<0.001
Black	12	8		11	6
Hispanic	5	5		5	1
Other	2	2		2	1
Missing	24	31	<0.001	23	61
Marital Status
Married	46	49		47	45	0.0028
Single	16	13		15	16
Divorced	23	24		23	23
Widowed	11	12		11	11
Missing	4	2	0.0002	3	6
State of Residence
California	92	99		99	38
Non-California	8	1	<0.001	1	62	<0.001
Diagnosis
Cancer	35	32	0.0426	35	17	<0.001
CHF	22	19	0.0175	22	7	<0.001
COPD	28	24	0.0002	28	11	<0.001
ESLD	6	4	0.0327	6	3	0.0010
ESRD	3	1	<0.0001	3	0	0.001
Dementia	11	11	0.9656	11	3	<0.001
HIV	1	1	0.1586	1	0	0.0621
Complexity Index
0	35	37		32	69
1	35	39		38	22
2	22	18		22	6
3	7	5		7	2
4	1	1	<0.0001	1	0	<0.001
Site of Utilization
Any inpatient
Any long term	45	29	<0.0001	42	21	<0.001
care	20	12	<0.0001	19	7	<0.001
Any outpatient	95	96	0.0225	95	96	0.1218

*Findings are expressed as proportions unless otherwise identified. P-values reflect Wilcoxon two-sided probabilities for continuous variables and chi-square for categorical variables. Categorical tests reflect tests for differences including missing.