Transcript Costs and cost effectiveness of training traditional birth

COSTS AND COST EFFECTIVENESS OF
TRAINING TRADITIONAL BIRTH
ATTENDANTS TO REDUCE NEONATAL
MORTALITY IN THE LUFWANYAMA
NEONATAL SURVIVAL STUDY
Christopher J. Gill MD MS
Center for Global Health and Development
Department of International Health
Boston University School of Public Health
ICIUM 2012, Antalya Turkey
Background
• Problem: Neonatal Mortality accounts for ~40% of
‘Under Five’ deaths
• 75% due to perinatal conditions: birth asphyxia, hypothermia
and sepsis
• In areas with limited access to health services, Traditional
Birth Attendants are a common source of basic obstetrical
care
• Response: The Lufwanyama Neonatal Survival
Project (LUNESP) assessed the effectiveness of
training TBAs in skills targeting birth asphyxia,
hypothermia and sepsis.
• Question: What is the cost effectiveness of this
strategy?
Setting and Methods
• Setting: Lufwanyama, Zambia
• Low population density: 6.4 persons/square
kilometer
• High neonatal mortality: ~40/1000 live births
• Methods: Cluster randomized and
controlled effectiveness trial
• 120 TBAs randomized to intervention/control
• Control TBAs continued standard of care
• Intervention TBAs trained in two skill sets:
• Neonatal resuscitation protocol
• Antibiotics with Facilitated Referral
• Primary endpoint: mortality by day 28 among
live-born infants
• Endpoints captured on ~3500 deliveries
(97.9% of total enrolled)
Lufwanyama facts:
• 12 health
posts/centers
• No physicians
• No hospitals
Results of main study
Death Rate on Day of Delivery:
19.9/1000 births (control) vs.
7.8/1000 births (intervention)
RR = 0.4, 95% CI 0.19-0.83
Results from main trial
• 1 death averted per 56
deliveries attended
• Relative risk reduction
0.55
(95% CI 0.33 to 0.90)
• Absolute risk reduction
of 18 deaths / 1000 live
births
Key Question: But is it cost effective?
Cost effectiveness analysis
• Costs and effectiveness data taken directly from the trial
• Cost effectiveness assessed from three perspectives
1. Financial – actual costs incurred during LUNESP
2. Economic – factors in additional costs from a societal perspective
3. 10-year forecasted economic analysis – models the cost
effectiveness of the LUNESP interventions if applied
programmatically
• All costs adjusted for inflation, expressed in constant dollars
• Discount rate of 3%
• Key Outcomes:
1. Cost per delivery attended
2. Cost effectiveness: per life saved
3. Cost effectiveness: per DALY averted
Results: Costs
Fixed and variable costs for the LUNESP interventions (2011 US$)
Cost item
Total fixed costs
Total variable costs
Total costs
Cost per program year
Cost per birth
Variable costs per birth
Financial
2006-08
Cost
% total
103,963
13,986
117,949
51,101
58.6
7.4
88.1
11.8
100.0
Economic
2006-08
Cost
% total
111,864
15,771
127,635
55,571
63.7
8.3
87.6
12.4
100.0
Projected Economic
2011-20
Cost
% total
214,792 81.6
48,321 18.4
263,114 100.0
27,533
34.1
6.3
Assumed main features of
LUNESP except 100% task
shifting:
1. TBA training
2. Program management
Results: Cost effectiveness
Multivariate sensitivity analyses
Base
case
High
impact scenario
Conservative
scenario
Cost per death avoided
2003
694
2993
90% probability value
3811
1241
6468
80
28
119
90% probability value
152
49
258
Cost per birth attended
30.0
10.5
33.0
90% probability value
44.2
13.5
58.5
Cost per DALY averted
Parameters varied in Monte Carlo:
1. Effect size
2. Average No. deliveries/month/TBA
3. Training workshop logistic costs
4. Costs Monitoring and Supervision
Conclusions
WHO classification of cost effectiveness of interventions:
• ‘Cost effective’ if a DALY averted is less than three times per capita GDP
• ‘Highly cost effective’ if less than per capita GDP
• Zambia’s 2010 per capital GDP was 1500 dollars
LUNESP’s interventions were ‘highly cost effective’ - even
under most conservative assumptions
• Intervention will be maximally cost effective in settings
where TBAs are busier, and where local ownership of
program is complete.
• This approach can be recommended as high value for
money.
Acknowledgements
Our team
• Lora Sabin
• David Hamer
• Anna B Knapp
• Nicholas Guerina
• Grace Mazala
• Joshua Kasimba
• William MacLeod
Our Funders
• USAID
• NIH/NIAID
• AAP
• UNICEF
backups
Overview of LUNESP
study design
STUDY OVERVIEW
12
Data collector
assessments
Intervention
TBAs:
Trained in AFR
+ NRP
Deliveries
Live
births
Week
One
Death
Stillbirths
Randomization
of TBAs
Week
Four
Death
Verbal autopsies
Death
Stillbirths
Death
Control TBAs:
Existing
standard of
care
Deliveries
Live
births
Week
One
Week
Four
Statistical
analysis
Assumptions for scenario analyses
Parameter
Base case
High Impact
Conservative
# of TBAs trained
together
60
80
60
Time for refresher
workshop
2
1
2
Annual trainings
needed
3
3
4
1.29
3.34
1.21
17.9/1000 live
births
17.9/1000 live
births
13.4/1000 live
births (25% drop)
Monthly
Every other month
Monthly
Births/TBA/Month
Effect size of
intervention
Monitoring
Results: Costs for the three models
One-way Sensitivity analyses: key drivers of CE
Multivariate sensitivity analyses