Table 5

The prevalence and diagnostic characteristics of proxy indicators and clinical algorithms for identifying HIV infection in severely malnourished children

Prevalence

OR

Sensitivity

Specificity

PPV†

NPV‡

PLR$

NLR&

(%)

(95%CI)

(%)

(%)

Individual proxy indicators, symptoms and signs

From widow headed household

1.4

30.0(7.4–121.6)

19

99.2

44.4

97.4

23.8

0.8

Looked after by grand-mother

4.1

4.1(1.1–14.6)

13.6

96.3

10.7

97.1

3.7

0.9

Orphan (one or both parents dead)

5.0

10.8(4.1–28.5)

31.8

95.8

20

97.7

7.6

0.7

From female headed household

7.1

5.8(2.1–17.2)

28.6

93.6

12.8

97.6

4.5

0.8

Symptoms and signs used in algorithms

Child with tuberculosis

1.4

9.2(1.8–47.7)

10

98.8

20

97.4

8.3

0.9

Minor muco-cutaneous manifestations

7.2

4.6(1.6–13.2)

25

93.3

10

97.6

3.7

0.8

Variables associated with HIV in the present study

Death of the father

3.1

15.6(4.7–49.9)

27.3

97.6

27.3

97.6

11.4

0.7

Age < 12 months or age > 59 months

13.6

3.7(1.4–9.5)

35

85.2

8.8

97.5

2.4

0.8

MUAC < 110 mm

29.6

3.5(1.9–10.2)

61.1

71.3

5.9

98.4

2.1

0.5

Absence of oedema

17.1

2.9(1.1–7.7)

36.4

83.7

8.3

93

2.2

0.8

Axillary nodes enlargement

4.8

6.7(1.7–26.4)

23.1

95.7

13.6

97.7

5.4

0.8

Algorithm and combinations

South-African IMCI modified algorithm for paediatric HIV diagnosis

20.0

94.5

8.7

97.8

3.6

0.8

Original IMCI algorithm

9.1

96.7

8.0

97.1

2.8

0.9

Action Against Hunger algorithm

60.0

62.1

3.5

98.3

1.6

0.6

Presence one or more proxy indicators and MUAC < 110 mm

95.5

54.5

7.1

99.7

2.1

0.1

† PPV = positive predictive value; ‡ NPV = Negative predictive value; § PLR = likelihood ratio for a positive test; & NLR = likelihood ratio for a negative test

Bahwere et al. BMC Infectious Diseases 2008 8:106   doi:10.1186/1471-2334-8-106