South Africa has quickly and successfully scaled up the
first phase of implementation of the new diagnostic test for tuberculosis, the
GeneXpert MTB/RIF assay, as the first-line TB test in 25 sites, achieving two
to threefold higher rates of TB and drug-resistant TB diagnoses in many of the
facilities, according to a late breaker presentation at IAS2011 in Rome.
However, switching over from relying on smear microscopy-based diagnoses to
using the new test presents a host of challenges.
“It changes everything we do,” said Professor Wendy Stevens
of the University
of Witwatersrand and the
National Health Laboratory Service, who added that, despite the expected substantial increase in people put
on earlier treatment, the cost of rapid full implementation has the programme
“Our initial algorithm is an expensive one, which may well
have to be modified as confidence in technology and data emerges,” she said.
TB in South Africa
The rates of TB, HIV-related TB and MDR-TB in South Africa
are among the highest in the world — with case detection and cure rates that
are below WHO targets. This failure to diagnose and effectively treat TB leads
to the alarming overall rate of TB, around 980 cases per 100,000, much of which is fuelled by the
HIV epidemic. 70%-80% of the
country’s TB suspects are infected with HIV, generating one-fifth of world’s reported HIV-associated TB cases.
Even if case detection were at recommended levels, the
currently used first line diagnostic tests are
notoriously outdated and insensitive, especially for detecting TB in smear-negative individuals, and for detecting extrapulmonary TB. Both smear negative and extrapulmonary TB
are much more common in people living with HIV.
Patients seeking a diagnosis for TB and treatment have to
make multiple clinic before they are finally diagnosed
and access treatment. Culture takes weeks or months and cannot be not performed
on all suspects. While a WHO-recommended algorithm exists to empirically
manage suspected smear-negative TB, it relies on cough as a gatekeeper symptom,
and thus may not be sensitive enough to capture many people living with HIV,
and requires a return clinic visit before an empirical diagnosis can be made.
diagnostic techniques have been under review over the last few years. The
GeneXpert MTB/RIF assay has seemed one of the most promising technologies — a
self-contained TB diagnosis system that automates sample processing on sputum and
provides real time PCR results within about two hours.
test also screens for the presence of resistance to rifampicin, which is used
as a surrogate for MDR-TB, and should therefore speed the time to diagnosing
drug-resistant TB as well. As previously reported, the test requires minimal
specimen handling and could be theoretically be performed by trained nurses.
after a review of the new test, the WHO gave it strong recommendation: “The new
automated DNA test for TB [GeneXpert] should be used as the initial diagnostic
test in individuals suspected of MDR-TB or HIV/TB.”
Africa already has a well
resourced TB diagnostic infrastructure,
according to Prof. Stevens. It has 244 microscopy centres that last year
performed over 4 million smears, and 16 culture labs that did 1 million
cultures and thousands of line probe assays.
“Despite this we are not making the diagnosis adequately in most of our
patients. Our smear sensitivity rates in
HIV are about 35-40%, and delay in the results from culture frequently lead to
loss to follow-up,” she said.
Scaling up Gene Xpert
Africa began to implement a pilot roll-out
of Gene Xpert in March 2011, after Health Minister Aaron Motsoaledi ordered a
radical change in the country’s policy on TB diagnosis. He instructed the NHLS
to get the first pilot sites up and running within six weeks.
The selection of the pilot sites was based upon burden, with
30 instruments installed at 25 of the highest burden districts in each
province. The initial rollout was at microscopy centres, which generally
provide about 11% of national coverage for TB diagnosis based upon smears.
“The decision at the time was to replace smears with one
GeneXpert test for diagnosis. What has happened subsequently is that a lot of
work has gone into the development of an algorithm,” said Prof. Stevens. The
plan for the near future is that all patients will get the GeneXpert test up
front. If the test fails (due to an error, such as lack of an adequate
specimen), the test is repeated. If the TB result is positive, they are
diagnosed with TB. If the test shows rifampicin
resistance as well, the results must be confirmed by TB culture and drug
sensitivity testing (DST).
One of the problems with this algorithm is that requiring only one GeneXpert test could
miss a high percentage of people living with HIV who have smear-negative TB.
The GeneXpert test is able to detect TB in many — though not all — smear-negative
people living with HIV, but more can be detected by performing a second or a
third test — which of course, increases the cost of diagnosis.
NHLS had a very
limited time to perform the activities needed to support the rollout. Site
assessments had be done at all 25 sites, with a checklist developed to looking
for network points, power, space, air conditioning, and adequate human
resources. Standard operating procedures had to be developed and 80 laboratory
technologists had to be trained (54 were trained by World TB Day) with an
intensive two day centralised training — so far, these have largely been
microscopists who had previously been doing first line diagnosis.
All instruments have been interfaced with the laboratory
information management system (LMIS). A Lab-Track LIS interface was developed
to automatically report the lab number, instrument, cartridge number whether or
not TB was detected, and whether RIF resistance was detected or not.
“This is so we can automatically collect data in real time,”
said Prof Stevens. . “We have also developed an external quality assurance
programme using dried culture spots. We quantitated TB bacteria by flow, placed
it on a dried culture spot, and every module that gets placed in the field is
tested in this way.”
The next steps were the development of the further
implementation plan, the budget plan and a national TB Costing Model (NTCM).
All the sites were launched by World TB day, and most were
operating at full capacity by April.
Cumulatively, between March and June, 50,093 people were screened with the test
— yielding a TB diagnosis in 8591 cases.
“We have identified
a positivity rate of 17.15%, which is huge,”
said Prof. Stevens. The test failure rate was 4.02% (these tests must be
re-run). “For rifampicin resistance, in the same population, we have picked up
7.3% (n=630), which is literally triple what has been identified in the
population previously,” she said.
Indeed, if these rates are accurate it would suggest that
around 30,000 to 40,000 cases of the 400-500,000 notified TB cases in South
Africa are MDR-TB — which would have profound implications for South Africa’s
TB programme. However, she added that the test is known to produce false
positives for rifampicin resistance, so those findings should be seen as preliminary.
Indeed, in a newly published study by Lawn et al, Gene
Xpert correctly identified rifampicin resistance in four cases of
multidrug-resistant TB but incorrectly identified resistance in three other
patients whose disease was subsequently confirmed to be drug sensitive by gene
sequencing. This means that the test is highly sensitive, but has a specificity
of 94.1%; and a positive predictive value of 57%. Even at that predictive value
however, the burden of MDR-TB in South Africa could be substantially
greater than previously thought.
There was significant geographical variation in rates of
M.TB positivity, and the rates of rifampicin
resistance to a lesser extent. Excluding the Western Cape (which does not seem
to have reported many test results), KwaZulu Natal has the highest rate of TB
diagnoses (20.51%) ranging down to 10.91% in Limpopo; while the Northwest
Province reported the highest rate of rifampicin resistance (9.3%).
Focusing just on the eThekwini District (Durban), the GeneXpert more than doubled TB
diagnoses to 19.71%, compared to 8-9% for the same period in 2010.
“We’ve learned that the algorithm development is complex and
time is needed to get consensus,” said Prof. Stevens. Switching over to
diagnostic platforms meant sweeping changes had to be made to the TB
guidelines, request forms, training and resistance reporting.
“Training needs to focus on items like sample preparation
and good laboratory practice. We’re also running into regulatory issues such as
who can do the test, when scope of work is well-defined in South Africa,”
The error rate of about 3-4% is a cause for concern because
it adds to the cost of implementation and delays diagnosis. Some of these
errors are well characterised, such as one resulting from putting an
insufficient volume of sputum into the cartridge. But the most common error
code being seen appears to be unique to South Africa and is being
investigated at the current time. “Electricity is also critical for the good
running of the instrument,” said Prof. Stevens. “As is good temperature
implementation, cost modelling and anticipated impact
But the biggest challenge could be the cost, which could
clearly affect the pace to full implementation.
There are two timelines for phased implementation on the
table, a fast-scale up scenario, with full coverage with GeneXpert by December
2012 and a slower scale-up scenario, with full coverage by September 2013.
“The minister would prefer the first one,” said Prof.
The first phase of implementation is already completed. The
second phase will be to fully capacitate the high burden districts. Phase 3a/3b
will be part of a Gates funded, cluster-based randomised trial to analyse cost
effectiveness and impact on patient outcome.
But the capital expenditures for the equipment alone with be
significant, at around (US) $21 million for 238 machines (65 G4, 169 G16 and 4 G48s)..
“This is a huge capital outlay, and we are looking at this
with great anxiety,” said Prof. Stevens.
This costing model adjusts for a 10% increase in TB suspects
coming in for screening — but this estimate may be low given the intensified
case finding campaign and routine TB screening of people coming through the HIV
counselling and testing campaign.
The cost of each test itself, or rather the cartridge, is
currently 161.45 Rand, but there are a number of hidden costs including staff
costs, the cost of calibrating the equipment, consumables, waste disposal (the
cartridges are bulky), transport and logistics, training and quality assurance
and overhead. Costs will also vary dependent on implementation rate, exchange rates, global
volumes, negotiations, and freight costs.
When hidden costs are included, the model projects that the
actual cost of performing a test will be R 216.30 (around US $36) between 2011
to 2014, decreasing to 189.85 Rand ($26) per test from 2014 to 2017.
The NHLH has also
developed a National TB Cost Model to estimate the
incremental costs to the national health service of switching to the new
diagnostic platform, under routine care conditions and at costs incurred by the
“We think that the programme costs per suspect will increase
by about 53%, and that the cost per case, for diagnosis alone will increase by
17%. Having said that, we will probably identify 30% more cases, 76% more
MDR-TB and by 2013, if everything goes to plan, initiate 39% more people on
treatment,” said Prof Stevens.
The increase in initiation of treatment is only one of the
benefits however. According to a study by Boehme et al, the test
will lead to dramatic reductions in the time to initiating treatment from 56 days
(39–81) currently to 5 days (2–8). Currently, only 46% of
patients are diagnosed by their second visit to the clinic, and another 40%
only by the third visit. In contrast, at full GeneXpert coverage, 83% of
patients will be diagnosed by their second visit and 89% by visit three.
Meanwhile, there should be reductions in the
use of other TB diagnostics. AFB microscopy — a time-consuming task — will
still be necessary for monitoring treatment response, but the number of smears
from 4.1 million per year currently to 1.5 million
per year. Use of culture should also decrease by about 30%.
A leap into the
Although switching to use of GeneXpert for first-line
diagnosis is a radical step, so far it has gone well, and one cannot fault the
South African National Health Laboratory Services for lack of preparation.
“Significant changes to the national TB programme are
envisaged,” she said. “What we have already identified is that it is beginning
to facilitate [TB/HIV] integration at both the laboratory, clinic and
programmatic level because people are being forced to analyse their budgets and
look at putting clinics together.”
“I think it is a brave bold step, and we are hoping that by
the end of full Xpert implementation, the bulk of our patients
will be diagnosed by Xpert, and the bulk of our patients will be diagnosed
before 5 days. We are completely aware of the implications this will have on
our treatment programme.”
However, it should be pointed out that the current algorithm
is relatively conservative. Other presentations at the conference (and in a
series of articles just published in PLoS Medicine) argue for much broader use of
GeneXpert, with repeated testing to diagnose more smear negative TB in people
living with HIV. However, this would increase financial outlays significantly —
and not everyone is convinced that the expenditures are the best use of limited
Stevens W. Taking the first steps: Xpert MTB/RIF
Implementation in public sector in South Africa: lessons learned.
International AIDS Society Conference on HIV Pathogenesis, Treatment and
Scott L et al. Dried culture spots for Xpert MTB/RIF
external quality assessment. International AIDS Society Conference on HIV Pathogenesis, Treatment and
2011 Abstract MOPE147.
Lawn SD et al. Screening
for HIV-associated tuberculosis and rifampicin resistance before antiretroviral
therapy using the Xpert MTB/RIF assay: a prospective study. PLoS Med 8(7):
et al. Feasibility, diagnostic accuracy,
and effectiveness of decentralised use of the Xpert MTB/RIF test for diagnosis
of tuberculosis and multidrug resistance: a multicentre implementation study.
The Lancet, 377(9776): 1495 - 1505, 2011.