Active TB case finding in the Western Pacific Region Health Organization, Western Pacific Regional Office Active TB case finding in the Western Pacific Region experience and policy

World Health Organization, Western Pacific Regional Office

Active TB case finding

in the Western Pacific Regionexperience and policy direction

Nobu Nishikiori

Medical Officer, Stop TB

WHO Western Pacific Regional Office


Contents

• Importance of ACF for TB control in the Region

• Experience of ACF in the Region

• ACF targeting tool – to guide TB REACH project

formulation

• Suggestions for ACF guidelines

– Risk-by-Risk approach

– Consideration for tailored care and support


Increasing importance of ACF

• Case detection stagnating in most

of the countries in the Region

• TB concentrates among high risk

populations

• Emerging challenges

– Migrants

– Urban poor

– Emerging risk factors for TB

Aging, tobacco, diabetes

• Low diagnostic sensitivity

• Infectious patients with minor

symptoms may not seek care

0

1000

2000

3000

4000

5000

0

10

20

30

40

50

1998199920002001200220032004200520062007

Nu

mb

er

of

mig

ran

t TB

cas

es

TB C

ase

No

tifi

cati

on

(pe

r 1

00

00

0)

TB among migrants and local residents,Shanghai, 1998 - 2007

Number of TB cases among internal migrants

TB notification rate among local residents


Migrant TB burden in MalaysiaNumber and % of migrant among all TB cases, by State, 2008

Number of

migrant TB cases

959

Sabah355

Selangor

134

Johor

375

Kuala

Lumpur

Myanmar

Thailand

Cambodia

Viet NamPhilippines

IndonesiaIndonesia


Experience in ACF (1)

Prisons and confined settings

Entry screening

• e.g. Mongolia

– „double entry screening‟ with fluorography film

– Significant reduction of TB burden

Periodic (or project-based) screening

• e.g. Viet Nam

– X-ray Microscopy + Culture

– Prevalence 1560 (S+) 2537 (C+) per 100 000 (NNS 40)

• e.g. Cambodia

– X-ray Microscopy

– Prevalence 950 (S+) per 100 000 (NNS 105)

„Enhanced‟ case finding

• Philippines

– Peer cough monitoring – cough surveillance

Reduction of TB burden among

prisoners in Mongolia

• Decline in TB

CNR

Prison vs national

– 18 times in 2001

– 10 times in 2005

– 5 times in 2008

• Improvement of

entry TB screening

on detention (461)

and on allocation

(401)

• Decline among

sentenced

prisoners (Red)

Joint Ministerial Order:Screening policy established

•100% entry screening with GF support•Equipment upgraded

Global Fund support

Continuous improvement of prison conditions


Experience in ACF (2)

TB contact investigation

Routine CI established and reported

– e.g. China, Mongolia, Malaysia and other IBCs

– Yield tends to vary…problem in implementation level ?

Low sensitivity of the procedures?

Other environmental determinants?

Policy established but weak implementation and reporting

– e.g. Many HBCs

– Too labor intensive for over-stretched health systems

CI combined with community-based ACF in Cambodia

– Unique strategy in Cambodia

– Can be very cost-effective


Community ACF strategies

in Cambodia (2005 – 2010)

• Define target areas with high TB case load

• Identify smear positive index cases from TB register and:

Strategy 1 (Adult contacts):

– House-to-house symptom screening + microscopy sessions

Strategy 2 (Adult contacts):

– House-to-house invitation + X-ray & microscopy sessions

Strategy 3 (childhood contacts):

– House-to-house invitation + PPD sessions


Community-based ACF in Cambodia:

A step-wise approach

Target

CommunityTB suspects

Step 3: Community

volunteers conduct

house-to-house

visit and invite

contacts and TB

suspects to ACF

sessions

Step 4:

ACF team (stay

1-2 weeks)

screen all TB

suspects with

mobile X-ray

Abnormal

chest X-rayMicroscopy

Step 5:

Three sputum

smear

microscopy for

diagnosis

Step 1:

Geographical

targeting based on

TB case load and

socio-economic

status

Step 2: Local

advocacy meeting

Step 6:

Treatment

follow-up by

local health

workers and

volunteers


Rationale for

“Retrospective” CI

• Routine contact investigation in

HBCs

– Should be done as much as possible

– But…difficult to implement fully

• Contacts have increased risk of active

TB disease for several years

– One time CI might miss many cases

– Cumulating cases for 1-2 years can

be cost-effective

• Contacts share same environmental

risks with their index

Lee, M. S., et al. 2008. "Early and late tuberculosis risks among

close contacts in Hong Kong." Int J Tuberc Lung Dis 12(3).

Morán-Mendoza, O. et al., 2010. Risk factors for developing

tuberculosis: a 12-year follow-up of contacts of tuberculosis cases.

Int J Tuberc Lung Dis, 14(9),

3mon Y1 Y2 Y3 Y4 Y5


Results (2005-2010)

How much yield we can get?

TB cases diagnosed:

(among all participants

attended ACF sessions)

All TB: 6% to 12%

NNS: 8 ~ 17

Smear +ve: 2% to 3.5%

NNS: 28 ~ 50

* A systematic review (Morrison

et al, 2008) reported pooled yields

of 4.5% and 2.3% for all and

confirmed TB respectively.


Was it cost-effective?

• Cost data was available for the

sessions in 2010

• The strategy is highly cost

effective

– Cost per case identified

Diagnostic cost: $21 per case

Overall cost: $113 per case

(logistics and operations cost)

(c.f. TB REACH criteria $350 per

case diagnosed and successfully

treated)

Cost per case identified

Diagnostic cost 18$ 18$ 16$ 25$ 41$ 21$

Overall cost 91$ 86$ 86$ 129$ 223$ 113$

Proportion of TB cases identified among all TB suspects

attended in outreach active case finding sessions, 2010.


The upgraded strategy for TB REACH

Project 2011

• Better area targeting using:

– Socio-economic indicators

(poverty, health access, etc)

– TB case load

• Further increase the yield

– Screening: Symptom + X-ray

for all contacts

– Diagnostic: Xpert MTB/RIF

– Target adult and childhood

contacts together

(Poor) Poverty/access index (Better-off)

TB

no

tifica

tio

n r

ate

(p

er

10

0 0

00

)

Targeted districts

○ Non-targeted districts


A tool for ACF targeting and strategy selection

Guidance was needed to support formulating ACF projects (for TB REACH)

An electric tool for ACF targeting developed

What factors determine the yield and cost-effectiveness of ACF?

1. TB prevalence among the target• Higher prevalence higher yield

2. Diagnostic algorithms• More comprehensive screening higher cost & yield

3. Targeting approaches (not yet included)• Scattered target higher „opportunity‟ cost

Increased TB case detection

Prisoners

TB contacts

Mal-nourished

Diabetics


1. TB Prevalence among the target

• Number needed to screen (NNS) shoot

up as TB prevalence goes down

• Roughly, ACF is not feasible for a

target < 0.5% prevalence (if X-ray for

all)

• However, NNS alone cannot guide

whether we should target or not

– Diagnostic cost significantly varies

between dx algorithms

– Operational and logistics cost also

different for each target/context

• How to find populations with >

0.5% prevalence?

Tentative cost effective zone


2. Diagnostic algorithms and yields

Model algorithms

1. Symptom screening microscopy

(routine programme model)

low cost & low yield

2. Symptom microscopy + x-ray

3. X-ray + symptom microscopy

4. X-ray + symptom microscopy + culture

(prevalence survey model)

high-cost & high-yield

5. X-ray + symptom Xpert MTB/RIF

(prevalence survey model with Xpert)

high cost model with Xpert

6. Symptom X-ray Xpert MTB/RIF

low cost model with Xpert

Estimating yields

• How many TB cases detected for

each algorithm roughly defined by

suspect/yield profile as below

example

• Important to note that:

– Initial symptom screening substantially

decrease a yield

– Low sensitivity of microscopy

Suspects (additionally) identified by X-ray

Smear- positive TB11%

Culture-positive Smear-negative TB

49%

Suspects identified by symptom screening

Smear- positive TB19%

Culture-positive Smear-negative TB

21%

Fig. An example of yield profile based on the

prevalence survey, Cambodia 2002.


An example of the tool outputs:

Diagnostic cost per case detected

Prevalence > 2%

• Cost effective for all strategies

including prevalence survey

models (culture or Xpert)

Prevalence 1-2%

• X-ray screening (strategy 3)

may be still cost-effective

• Culture probably feasible but

requires careful planning

Prevalence <1.0%

• Up to strategy 1 &2 acceptable

(i.e. routine procedure)

* Cut-off of USD 200 are arbitrary. TB REACH

criteria employ USD 350 per case detected and

successfully treated.

200

0

200

400

600

Strategy 1 Strategy 2 Strategy 3 Strategy 4 Strategy 5 Strategy 6

Dia

gn

osti

c c

ost

per

case d

ete

cte

d (

US

D)

Cost-effectiveness of TB screening by diagnostic algorithm, Cambodia

RR = 1.0 (0.69% -baseline)

RR = 1.5 (1.04%)

RR = 2.0 (1.39%)

RR = 3.0 (2.08%)

RR = 5.8 (4.00%)

RR = 8.7 (6.00%)

Cost-effectiveness target (200)

Symptom

microscopy

Symptom

microscopy

+ X-ray

Symptom

+ X-ray

microscopy

Symptom

+ X-ray

microscopy

+ culture

Symptom

+ X-ray

Xpert

Symptom

X-ray

Xpert

General

population

Urban

slum

Smokers

Diabetics

Contacts

Prisoners

Contacts Prisoners

General

population

Urban

slum

Smokers Diabetics


General observations from the tool outputs

• Conservative algorithm can be acceptable even NNS is high

• The higher the prevalence

the more extensive approach high yield

However, we have a dilemma…

– Very high risk groups tend to be small

– Lower risk groups larger size and difficult to target

So the key is to find a high risk target with a good pop size

Risk-by-Risk – combining multiple risks – might be a way to manipulate a risk

profile and a target size

e.g. Geographical targeting x TB contacts (Cambodian Retro CI),

Deported migrants x detention history (another TB REACH project in

Cambodia)


Narrowing down the target: Risk x Risk approach

(Elderly x diabetics) (Elderly x smokers)

Current smoker:

annual incidence

735 per 100 000

Ex-smoker:

annual incidence

427 per 100 000

Never smoked:

annual incidence

174 per 100 000

Cumulative hazards for active TB by smoking status,

among a cohort of clients (>65yrs) registered with an

elderly health service in Hong Kong

Leung, et al. 2004. "Smoking and tuberculosis among the elderly in

Hong Kong." Am J Respir Crit Care Med 170(9): 1027-1033.

Cumulative hazards for active TB by diabetic status,

among a cohort of clients (>65yrs) registered with an

elderly health service in Hong Kong

Leung, et al. 2008. "Diabetic control and risk of tuberculosis: a cohort

study." Am J Epidemiol 167(12): 1486-1494.

HbA1c >= 7%:

annual incidence

422 per 100 000

No diabetes:

annual incidence

214 per 100 000


Neighborhood factor

analysis for geo-targeting

• Neighourhood factor analyses using

socio-economic characteristics have a

potential to guide geo-targeting

• Risk micro-stratification to identify

target area/population

• Risk x Risk approach

e.g. poor neighborhood x malnourished

e.g. densely populated area x contact

investigation

Quezon City

Manila

Taguig

Paranaque

Valenzuela

Las Pinas

Muntinlupa

Pasig City

Makati City

Kalookan City

Marikina

Malabon

Pasay City

Navotas

Mandaluyong

San Juan

Pateros

Obando

Barangay-wise population density, Metro Manila

Barangay

Population density

0.000000 - 199438000.000000

199438000.000001 - 392535008.000000

392535008.000001 - 651995008.000000

651995008.000001 - 1044950016.000000

1044950016.000000 - 16246800384.000000

±


Entire

gro

up

HIV

Sm

okers

Maln

ourished

TB

conta

ct

his

tory

Alc

oholic

s

Dia

bete

s

Eld

erly

Pre

vio

us T

B

Slum dwellers x x x x x x x x

HIV x

Smokers

Prisoners x

Migrants x x x x

Malnourished x

TB contacts x x x x x x x

Alcoholics

Miners x

Diabetes x x x x x x

Elderly x x x x x x

previous TB

Target population (venue)

Risk factors (to be included in suspect definition)

Risk-by-Risk Table


Targeted case finding approaches require

specific measures to provide care

• High risk populations requires a

tailored service delivery mechanism

– e.g. Migrants / urban poor

• highly mobile / high default and transfer

• Social and financial insecurity

– e.g. Prisons

• High co-morbidity including HIV (what if

ARV is not available?)

• Transfer and referral system (release

screening?)

• For guidelines:

– Cases successfully treated is important

outcome (not only case finding)

– Is there any group for which ACF should not

be conducted unless the specific support

mechanism is not ensured (other than routine

DOTS)?

91%

37%

13%

11%

39%

0%

20%

40%

60%

80%

100%

Permanent Residents (N=5915)

Migrants(N=2423)

Treatment outcome of new smear positive TB by residence status, Beijing, 1997-2002

Transferred

Defaulted

Died

Failed

Completed

Cured

Cases among

migrants

(Up to 35%)

Zhang, L.X. et al., 2006.

Int J Tuberc Lung Dis,

10(9).


Summary

• ACF is potentially a very important TB control strategy in the Region

• Experience shows some positive outcomes in the Region (though limited)

• An interactive tool can facilitate country level targeting, strategy selection

and planning

• Risk-by-risk approach can help increase the TB risk and narrowing down

the size of the target population (concentration and selection)

– “Does combining more than two risk factors to identify the target population increase the

yield and cost-effectiveness of TB screening?”

• Targeted ACF requires extensive consideration for treatment and care

delivery strategies

– “Does a tailored care mechanism for a specific high risk group (migrant, prisoners, etc)

improve TB treatment outcome compare to the routine DOTS programme”

Active TB case finding in the Western Pacific Region Health Organization, Western Pacific Regional Office Active TB case finding in the Western Pacific Region experience and policy

Documents