Transcript ISPHC-PresentationJean-GillesQuinnCamilleri

A cost-effectiveness evaluation of
preventive interventions for HIV-TB in
Sub-Saharan Africa (Tanzania):
Relevance for neurological infections
Lucie Jean-Gilles
Casey Quinn
Corinne Camilleri-Ferrante
Outline





Background on HIV-Tuberculosis (TB) in SubSaharan Africa and neurological implications
Prevention & treatments for HIV-TB
Cost-effectiveness study for HIV-TB in Tanzania
Implication of results and relevance to neuro-TB
Conclusion and future directions
Background

Burden of TB closely linked to HIV epidemic in
SSA (80% of global co-infection cases) HIV
increases risk for Mycobacterium tuberculosis
infection and for latent TB progression into
active disease (30-50 fold higher)
Microscopic view of T. Bacilli
Estimated global incidence of TB (2007)
http://www.who.int/tb/publications/global_report/2009/pdf/chapter1.pdf
Estimated global distribution of HIV prevalence in new TB
cases (2007)
http://www.who.int/tb/publications/global_report/2009/pdf/chapter1.pdf
Background

HIV-infected TB (prevalence 30-75% of ~7%
Tanzania HIV prevalence) = high annual
death incidence (↑ death risk <200 CD4+
lymphocytes/mm3) and ↑ opportunistic
infections  TB accelerates HIV progression

Although 1/3 TB deaths are HIV associated
and 7% mortality rate in 1.4 million HIV
population of Tanzania in 2009, central
nervous system TB kills & disables more than
any other form of TB
Neurological implications

Meningeal TB most lethal (~30% of cases) and
disabling despite anti-TB chemotherapy common and
serious clinical problem in high HIV-TB prevalent SSA

Diagnosis & treatment of meningeal TB difficult HIV
co-infection increases management difficulties

Worldwide use of highly active anti-retroviral therapy
(HAART) (↓ opportunistic infections)  significantly
reduced incidence of neurological complications in
AIDS patients and TB/HIV/AIDS progression and
mortality
Meningeal Tuberculosis
T.S. Ramachandran (2008)Tuberculous Meningitis. Medscape
Neurological implications (cont.)

Survival improvement and decrease in HIV-TB by
HAART can be complicated by immune reconstitution
inflammatory syndrome (IRIS) (transient worsening of
disease i. e. neurological symptoms due to meningitis)
and drug toxicity when:
1) <200 CD4+ lymphocytes/mm3 (high HIV viral load)
2) HAART initiated within 2 months of anti-TB Rx
regimen

However, HAART overall clinical and mortality benefits
on HIV and HIV-TB outweighs adverse effects
HAART & Anti-TB therapy in Tanzania

Anti-TB drugs widely available (isoniazid, rifampicin,
pyrazinamide) in Tanzania

High unmet need only 3-6% of HIV +ve population have
access to HAART in Tanzania; no specific treatment for
neurological HIV-TB
Contributing factors to problem:
 Systematic: high cost, lack of availability, distribution logistics,
int’l programme sustainability and health care system support

Priority setting: Lack of cost-effectiveness evidence of
HAART on HIV-TB strong economic case needed for
prioritisation of TB preventive interventions in HIV population
Are HAART & Anti-TB therapies worth
prioritising for HIV-TB in Tanzania?
Aims of Study

Cost-effectiveness of HAART as an adjunctive
therapy to, or following, anti-TB regimen for
HIV-TB from health care system perspective in
low-resource setting Tanzania

Relevance of results for HIV-TB neurological
intervention measures
Methodology

Decision analysis model using transition probabilities
sourced from Schiffer & Sterling (2007) study:
“Timing of Antiretroviral Therapy Initiation in Tuberculosis
Patients with AIDS: A Decision Analysis”

Data obtained from primary source studies where patients
received standard rifamycin-based anti-TB therapy for 6
months with or without HAART

Analysis ran under 3 conditions:
1)
Early HAART + Anti-TB
Deferred HAART + Anti-TB
No HAART (Anti-TB only)
2)
3)
Methodology
Methodology
Methodology
Pro ®  Probabilistic micro-simulation
for parameter sampling distributions (transition
probabilities & costs)
• TreeAge
• Markov modelling- Monte Carlo and MonteCarlo PSA micro-simulations (n= 1000)
Results
Cost ($)
No HAART
Early HAART
Deferred HAART
Life-Years
Cost/LY ($)
361
0.720
502
13
0.026
19
1,687
0.906
1,864
181
0.042
201
1,198
0.862
1,390
109
0.034
129
Results
Kernel Densities of Costs using Monte-Carlo Simulation
Results
Kernel Densities of Life-Years using Mont-Carlo Simulation
Results
Scatter plots of Costs and Life-Years using Monte-Carlo Simulation
Results
Cost per life-
Incremental Cost
Cost
Life-year
year
361
0.720
501
Deferred HAART
1,198
0.862
1,389
5,888
Early HAART
1,687
0.906
1,862
11,095
No HAART
Incremental Cost per Life
- Year =
Cost
HAART
per Life-Year
 Cost
-
NoHAART
LY HAART  LY NoHAART
Results
Mean
Std. Dev.
Lower 95%
Upper 95%
C(def) - C(none)
837
110
621
1,052
LY(def) - LY(none)
0.142
0.043
0.058
0.226
C(early) - C(none)
1,326
182
969
1,684
LY(early) - LY(none)
0.186
0.050
0.088
0.284
C(early) - C(def)
490
208
LY(early) - LY(def)
0.044
0.054
-0.062
0.151
Incr. C/E (Def vs. None)
6,716
4,980
-3,045
16,477
Incr. C/E (Early vs. None)
7,738
2,828
2,196
13,280
Incr. C/E (Early vs. Def)
-9,655
442,866
83
-877,672
897
858,362
Results
Cost-Effectiveness Acceptability Curves of Treatment Options
Discussion

More sensitivity analysis to be undertaken for this model:
Difference between Early HAART and Deferred HAART (life-years) 
not statistically significant and may influence results.
More consideration for transition probabilities and effect to be
sensitive to our definitions (i.e. death due to drug toxicity).


This analysis only considered pulmonary TB in HIV patients, however
extra-pulmonary TB-related neurological conditions (i.e. meningeal
TB) in particular are of interest in this study. Next steps are:
- To infer rates of meningeal TB from these results
- To analyze a more complicated decision model or Markov model that
accommodates meningeal TB.
Inclusion of meningeal TB, its costs and mortality/morbidity in
Tanzania  conducting a cost-effectiveness analysis will provide
essential information for priority setting of this disease burden in LowIncome Countries like Tanzania.
Conclusion

Although previous analysis found that Early HAART was
preferred to Deferred HAART for HIV +ve TB patients
incremental effect of Early HAART on mortality due to
disease (HIV and TB) and drug toxicity is not statistically
significant.

Early HAART does not appear to be cost-effective
relative to Deferred HAART.

However, adjunctive treatment with HAART (early &
deferred) are effective and cost-effective relative to
treatment only using standard TB therapy.

Expanding this HIV-TB model to meningeal TB in the
Tanzanian setting will help determine cost-effective
strategies to help significantly reduce risk of TB-related
morbidity and mortality in its HIV population.
Thank you!