Slide 1

Report
Taking the first steps
Xpert MTB/RIF Implementation in
public sector in South Africa:
Lessons Learned
Wendy Stevens
Molecular Medicine and Haematology
University of the Witwatersrand & NHLS
Acknowledgments to:
Health Economics and Epidemiology Research Office
HE RO
2
Wits Health Consortium
University of the Witwatersrand
1
GeneXpert Technology (Cepheid)
GX48 (Infinity)
GX16
GX4
16
64
255
throughput/ 8hr day
FiND , 2010
•Automated
•Real-time PCR
•Rapid (2 hours)
•Cartridge based
Result
•Positive/negative
TB
•Resistance
yes/no to
Rifampicin
•Low contamination
risk
Boehme,C et al
NEJM 2010
Disease Burden in South Africa
• 20% worlds reported HIV‐associated TB cases and 2nd
largest reported numbers of MDR
• 70%-80% TB suspects infected with HIV
• Overall TB rates 980/100,000
– Mining populations 2500/100,000
– Correctional Services 4500/100,0000
• Increasingly smear negative (8-10% positivity) and extrapulmonary TB(16%)
• WHO Strong Recommendation: “The new automated DNA
test for TB should be used as the initial diagnostic test in
individuals suspected of MDR-TB or HIV/TB” (i.e. all SA TB
suspects)
4
NHLS TB Laboratory Facilities: 2010/2011
N=244
• 4.7 million
smears
• 1 million
cultures
• 90 000 LPA
5
Phase 1 rollout
High burden, TB Intensified Case Finding
campaign districts
•Limited Pilot in all 9 provinces
•Selection: volumes, district
selected
•25 sites, 30 instruments
•20 GX4, 9 GX16, 1 GX48
•Placement by world TB day:
March 24th
•11% national coverage based
on 2010 smears/2.0
2 smears at diagnosis to be replaced by
one Xpert MTB/RIF (Phased approach)
(microscopy centre based)
Where should Xpert be placed
within TB diagnostic algorithm?
7
Methodology: March-June 2011
• Site needs assessment: 25 sites
– Hoods, space, network points, power, A/C, HR, checklist developed
• Training
– 80 laboratory technologists : intensive 2 day centralised training
– -microscopists currently first cadre
– SOP driven
• LIMS interfacing (pilot)
– A Lab-Track LIS interface was developed to automatically report: Lab
number, cartridge number, TB detected/not, RIF detected/not.
• A verification program (“fit for purpose”) for placement
and calibration of each module
– [MOPE147]
• Development of implementation plan,
8
budget and National TB Costing Model (NTCM)
54 NHLS staff members trained prior to world TB day
National Xpert MTB Results
(cumulative March to June)
N = 50 093
ICF
MTB
MTB not
detected detected
Test
failure
Total
%
Positive
ICF
2218
12 762
744
15 724
14.11%
NonICF
6373
26 725
1271
34 369
18.54%
Total
8591
39 487
2015
50 093
17.15%
%
Total
17.15%
78.83%
4.02%
100%
10
National Xpert RIF results:
March-June 2011
N = 8591 (MTB detected); 630 RIF Resistance
ICF
No
Statu Indeterminate
Resistant
result
s
Sensitive Total %
ICF
15
78
195
1930
2218 8.79
NonICF
57
57
435
5824
6373 6.83
Total
72
135
630
7754
8591 7.33
% 11
0.84
1.57
7.33%
90.26%
100% 7.33
Geographical Variation
Province
Eastern Cape
Free State
Gauteng
Kwazulu-Natal
Limpopo
Mpumalanga
North West
Northern
Cape
Western Cape
Total
MTB
Detected
MTB Not
Detected
Test Failure
Total
632
3141
148
3921
523
2701
1
3225
683
3528
94
4305
3941
14490
788
19219
515
4142
62
4719
879
4515
557
5951
527
2867
72
3466
868
4049
292
5209
23
54
1
78
8591
39487
2015
50093
% MTB
Positive
% RIF
16.12
7.12
16.22
5.93
15.87
7.32
20.51
7.13
10.91
8.16
14.77
8.08
15.20
9.30
16.66
7.03
29.49*
17.15 %
-
7.33 %
TB GeneXpert Positivity: eThekwini
District in KZN
eThekwini GeneXpert Positivity Data
Date period: March 2011 to 9 June 2011
YEAR
MONTH MTB Detected MTB Not Detected Test Unsuccessful
3
470
1455
214
4
1568
5647
646
2011
5
847
3179
490
6
232
1013
55
Grand Total
3 117
11 294
1 405
% of Total
19.71
71.41
8.88
Total
2 139
7 861
4 516
1 300
15 816
100
% MTB Detected
21.97
19.95
18.76
17.85
19.71
Average smear positive rates for same period
2010 and 2011: 8%-9%
Challenges
Lessons Learned
Challenges and
Lessons
learned
Time to get
consensus, ideally before
implementation
Algorithm development
Need to build in flexibility
Changes: TB guidelines, request forms, training etc, resistance reporting
Training
Site needs assessment
At least 2 days, several individuals at each site
Better on site,
Include GLP, safety, computer literacy
Focus on sample preparation
Clinician training critical
Workflow issues problematic on large instruments
Regulatory issues
Costing implementation &
modelling future costs
Numerous sources for input
Need to model future
Opportunity for costing and reviewing current TB service
Error rates
3-4%: error codes: 5011 (73%), 5006/7 (16%)(insufficient vol), 2008 (10%)
EQA program
Verification program : DCS
Frequency? Per module?
Need for negative controls for larger analysers?
Electricity, temperature,
waste disposal, cartridge
storage
UPS, A/C (if>30C)
Cartridges fairly bulky (2-28C)
National Phased Implementation
PHASES| PILOT | FULL PILOT|HIGH CASE| GF XPERT | CONTROL | DISTRICTS| ALL LABS
FAST SCALE-UP | July 2011 | Dec 2011 | Sept 2011 | Mar 2011 | Dec 2011 | Dec 2012
SLOW SCALE-UP | July 2011 | Dec 2011 | Sept 2011 | Mar 2012 | Mar 2013 | Sept 2013
 FAST SCALE-UP scenario: Full coverage by December 2012 (Ministerial mandate)
 SLOW SCALE-UP scenario: Full coverage by September 2013
Model for instrument placement
(Fast scale-up, 10% growth in suspects)
2011/12
Province
EC
FS
GP
KN
LP
MP
NC
NW
WC
TOTAL
2012/13
GX4 GX16 GX48 GX4 GX16
4
1
3
6
3
2
1
12
5
13
11
4
5
2
3
1
10
1
18
7
3
2
1
1
4
65 GX4,
GX48
2013/14
GX4
14
2
3
14
36
20
7
1
11
7
169 GX16, 4GX48
GX16
Tests/
day at
GX48 full
capacity
2,720
496
1,552
2,944
1,056
544
192
656
1,088
11,248
Initiated at current microscopy centres, volumes based on adjusted smear
per patient , throughput of analysers.
CAPITAL : $21 M
Recurrent cost
Cost per MTB/RIF test (including hidden costs)
Cost item
Cartridge
Calibration
Staff
Consumables
Waste disposal
Transport and logistics
Training and QA
Overheads
Total
Cost
R 161.45
R 4.47
R 18.77
R 5.02
R 1.92
R 15.33
R 3.83
R 19.17
R 229.96
% of total
70%
2%
8%
2%
1%
7%
2%
8%
100%
Cost will vary: dependent on implementation rate, exchange
global volumes, negotiation, freight
Modelled Average per test cost across all scenarios
• 2011/12 to 2013/14: R 216.30 $ 26-36
• 2014/15 to 2016/17: R 189.85
National TB Cost Model
•
To estimate implementation costs for NHLS lab network
•
To inform national-level budget requirements (2011-2017)
•
To estimate the incremental national health service cost of replacing
the existing pulmonary TB diagnostic algorithm with a new algorithm
incorporating Xpert MTB/RIF molecular technology, under routine care
conditions and at costs incurred by the government (Excel-based population
level decision model) (HER0)
•
Built into Rollout BMGF study: cluster randomised trial design (phase 3a
and b) : to verify modelling and assess impact ( Aurum Institute)
Programme cost:
Total and per case cost in 2013 [2011 USD]
(Fast scale-up, 10% growth , SA at 50% of global volume, purchase)
Scenario
Annual cost
Cost per
suspect
Cost per case
1) Cost of diagnosis only
Baseline
$ 105 M
$ 45
$ 312
Xpert scenario
$ 160 M
$ 69
$ 367
$ 55 M
$ 24
$ 54
+53%
+53%
+17%
Difference to Baseline
% change
2) Cost of diagnosis and outpatient treatment
Baseline
$ 280 M
$ 121
$ 835
Xpert scenario
$ 399 M
$ 172
$ 912
Difference to Baseline
$ 118 M
$ 51
$ 77
42%
9%
% change
+42%
Conclusions I
• Pilot demonstrated feasibility of implementation
• Significantly increased early detection of MTB
• Significantly increased screening for potential MDR
cases
• Significant changes to National TB program envisaged
• Facilitating HIV/TB integration at laboratory, clinic and
programmatic level
• Expensive algorithm which may well have to be modified
as confidence in technology and data emerges
Infinity Installation in Prince
Msheyni in KZN: truly a team
effort
Acknowledgements
•
•
•
•
•
•
•
•
•
•
•
NHLS NPP program
NDoH: Drs Mametje, Pillay, Mvusi, Barron
NTBRL: Drs Erasmus and Coetzee
CHAI SA
HERO team, G. Meyer –Rath, K. Bistline
Right to care: Ian Sanne
MM&H: Prof Scott, N. Gous, B. Cunningham
USAID South Africa
CDC for funding and support
FIND
Aurum Institute

similar documents