ICF commentary

New York TAG core group

• Systems– Lab resources – how available are culture facilities?– Human resources – how many staff are needed?

• Contamination and “sub-clinical TB” – are all +ve cultures due to tuberculosis?

• Predictive values not sensitivity and specificity• What is the objective?

– Morbidity– Mortality– Transmission

• We do not live in a Platonic world!

How many samples can the lab handle? Do we need to screen?

0

2

4

6

8

10

12

14

16

18

20

Kain lo

CD4

Kim1 C

3

Kim1 S

x

Moh

d C2

Moh

d Sx

Shah

C

Shah

Sx

Kain C

Kain C

2

Kain C

3

Kain S

x

Kim2 C

2

Kim2 S

x

Kain h

iCD4

Day C2

Day Sx

ZAMSTAR C

ZAMSTAR C3

0%

2%

4%

6%

8%

10%

12%

NNCulture

Prevalence

To exclude TB we need high NPV and we want as many as possible to benefit from IPT

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

80% 90% 100%NPV

Pro

po

rtio

n t

rea

ted

Cough

Cough >3/52

Cough >2/52

Symptoms

+ Hi CD4

X Lo CD4

Sensitivity and specificity of algorithms, stratified by CD4 count (courteousy of Cain CROI 2008)

Algorithm CD4 < 250 CD4 >250

Sens Spec NPV Benefit Sens Spec NPV Benefit

Day 97 31 98.9% 28.2 70 39 97.7% 38.7

Mohammed 92 51 98.3% 46.7 48 67 97.7% 66.6

Kimerling 92 35 97.5% 32.3 67 54 98.1% 53.4

Demissie 95 37 98.5% 33.8 81 47 98.8% 46.2

Pre-IPT 92 51 98.3% 46.7 52 64 97.7% 63.5

Cough/fever/wt. loss

97 27 98.8% 24.6 81 37 98.4% 36.5

Possible recommendations

• Context is vital, and it may be hard to make one recommendation

• For IPT screening, the main issue is NPV, which must be reported

• For ICF, there is a trade-off between lab workload and PPV, which determines how many samples need to be sent to find one case

Possible recommendations• In late HIV disease(ARV clinics; home base care

settings, CD4<250), the prevalence of culture positivity is 7-10% - we don’t need to screen – everyone needs a culture

• In earlier HIV (VCT centres; PMTCT; STI clinics; community based screening) prevalence is lower; absence of any cough has a high NPV and allows most people to benefit from IPT; adding other symptoms to the screen improves the NPV a little (but not much), but reduces the number who benefit from IPT substantially.

• Culturing everyone with a cough will overload fragile systems; adding additional symptoms or using cough for >2-3/52 reduces the lab workload but inevitably misses some cases.

• People with symptoms clearly need appropriate referral….

IPT for people infected with HIV

• Early HIV infection

• Late HIV infection

No symptoms Symptoms TB

No symptoms Symptoms TB

IPT for people infected with HIV

• Early HIV infection

• Late HIV infection

No symptoms Short Cough Longer cough +/- others

Give IPT Follow upSputum sample

Sputum sampleFollow up

No symptoms Short cough Longer cough +/- others

Ongoing care TB found

Benefit from IPTTB

found

H monotherapy

TBOngoing

careNo symptoms