Skip to main content

Table 4 Diagnostic performance for malaria VA using the EIR-mortality relationship as gold standard

From: Infant and child mortality in relation to malaria transmission in KEMRI/CDC HDSS, Western Kenya: validation of verbal autopsy

 

Neo-natal

Post-neonatal

Child (1–4)

Overall

Estimates of numbers of deaths

 a: Malaria by VA, exposure attributable

7.7

141.1

243.1

391.9

 b: Malaria by VA, not attributable

0.3

172.9

70.9

244.1

 c: Other causes by VA, exposure attributable

64.3

436.3

605.0

1105.6

 d: Other causes by VA, not attributable

251.7

586.7

527.0

1365.4

Derived quantities

 Overall all-cause mortality rate (deaths/1000 person-years) (from Table 1)

296.6

140.4

51.6

80.4

 % of deaths assigned malaria as cause in VA (a + b)/(a + b + c + d) (%)

2.5

23.5

21.7

20.5

 % of deaths attributable to malaria exposure (a + c)/(a + b + c + d) (%)

26.3

43.2

63.9

48.2

Malaria specific mortality rate estimated from VAs (deaths/1000 person-years)

7.3

33.0

11.2

16.5

 Malaria exposure attributable mortality rate (deaths/1000 person-years)

78.0

60.7

33.0

38.8

 Sensitivity of VA (a/(a + c))

0.11

0.24

0.29

0.26

 Specificity of VA (d/(d + b))

1.00

0.77

0.88

0.85

 Positive predictive value of VA (PPV) (a/(a + b))

0.96

0.45

0.77

0.62

 Negative predictive value of VA (NPV) (d/(c + d))

0.80

0.57

0.47

0.55

  1. For consistency across age groups, rates are expressed as deaths per 1000 person-years, rather than relative to numbers of live births (which is the usual convention for deaths in the first year of life)