Classifying perinatal mortality using verbal autopsy: is there a role for nonphysicians?

Table 2 Comparison of neonatal underlying cause reported by physician consensus (PC) to underlying cause of death reported by community coordinator (CC) for select causes of early neonatal death (n = 118)

Underlying COD, as determined by CC	Underlying COD, as determined by PC (reference standard)		Measures²
	Identified as COD	Not identified as COD	SE	SP	PPV	NPV	CSMF_PC (95% CI)	1-CSMF_PC	CSMF_CC (95% CI)	1-CSMF_CC	RD (95% CI)	Kappa (95 CI)
Preterm/LBW	20	98	0.70	0.83	0.45	0.93	0.17 (0.09, 0.25)	0.83	0.26 (0.16, 0.37)	0.74	0.55 (0.08,1.04)	0.43 (0.24, 0.62)
COD	14 (70.0)	17 (17.3)
Not COD	6 (30.0)	81 (82.7)
Infection	52	66	0.37	0.94	0.83	0.65	0.44 (0.29, 0.59)	0.56	0.19 (0.11, 0.28)	0.81	0.56 (0.40, 0.70)	0.32 (0.17, 0.48)
COD	19 (36.5)	4 (6.1)
Not COD	33 (63.5)	62 (93.9)
Asphyxia	31	87	0.42	0.90	0.59	0.81	0.26 (0.16, 0.37)	0.74	0.19 (0.10, 0.27)	0.81	0.29 (0.03, 0.54)	0.35 (0.15, 0.54)
COD	13 (41.9)	9 (10.3)
Not COD	18 (58.1)	78 (89.7)
Fetal Trauma	0	118	--	0.94	0.00	1.00	--	--	0.06 (0.01, 0.11)	0.94	--
COD	0 (0.0)	7 (5.9)
Not COD	0 (0.0)	111 (94.1)
Tetanus	5	113	0.80	0.98	0.67	0.99	0.04 (0.00, 0.08)	0.96	0.05 (0.01, 0.09)	0.95	0.20 (0.00,2.00)	0.71 0.41, 1.00)
COD	4 (80.0)	2 (1.8)
Not COD	1 (20.0)	111 (98.2)
Unknown/ no cause	4	114	0.50	0.91	0.17	0.98	0.03 (0.00, 0.07)	0.97	0.10 (0.04, 0.16)	0.90	2.00 (0.27, 11.0)	0.21 (-0.07, 0.49)
COD	2 (50.0)	10 (8.8)
Not COD	2 (50.0)	104 (91.2)
Other¹	6	112	0.50	0.88	0.18	0.97	0.05 (0.01, 0.09)	0.95	0.14 (0.07, 0.22)	0.86	1.83 (0.38,7.33)	0.20 (-0.04, 0.44)
COD	3 (50.0)	14 (12.5)
Not COD	3 (50.0)	98 (87.5)

¹ Other includes congenital malformation, birth trauma, and neonatal accident.
² Definitions of measures are provided in the methods section. Measures are as follows: sensitivity (SE), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), cause-specific mortality fraction (CSMF) of physician consensus (PC) and community coordinator (CC), and absolute relative difference of the CSMF (RD). RD was calculated using the actual numeric values rather than the rounded CSMFs reported in the table.

ISSN: 1478-7954