CBER Presentation

Printable Version

Risks versus benefits related to the possible implementation of a malaria blood-screening test

FDA Workshop on Testing for Malarial Infections in Blood Donors
July 12, 2006

Steve Anderson
Office of Biostatistics & Epidemiology
FDA-Center for Biologics Evaluation and Research

Probabilistic Modeling Risk/Benefit of New Donor Populations

Current U.S. policy includes deferral of:

Travelers - malaria endemic countries in last year
Immigrants - from malaria endemic countries < 3 yrs
Donors that had malaria - asymptomatic < 3yrs

Goal: Use probabilistic model to evaluate potential risks / benefits and uncertainties of:

Current policy
Universal NAT Testing Scenario
Universal Antibody Testing Scenario

Probabilistic Modeling

Rather than single numbers or "point estimates"

Employs statistical distributions for INPUT PARAMETERS - represents uncertainty of data

Monte Carlo method chooses a value from each distribution as the "single number" for ONE iteration and generates OUTPUT as distributions

Model is run thousands or millions of iterations and single "aggregate" OUTPUT distributions reflecting uncertainty and variability are generated

Uncertainty

Arises from lack of or limited data for an input parameter(s)
Assumptions used in model - add to uncertainty

Lack of information or data for estimating -

Self deferral for travelers to / immigrants from malaria areas,
effectiveness malaria deferrals,
Donation rates of travelers / immigrants,
NAT test sensitivity,
Antibody test sensitivity, etc.

Uncertainty represented as confidence intervals about mean estimated outcomes

Malaria Risk in the United States

1,325 reported cases of Malaria identified in the U.S. in 2004 (CDC, MMWR 2006)

All but 4 cases imported

~ 50% cases were Plasmodium falciparum

Transfusion transmitted malaria (TTM) rate is low

~ 0.25 cases per million units collected

Possible Risks (Costs) and Benefits of Malaria testing of blood

Risks (Costs)

Additional malaria units, transfusion transmitted malaria (TTM), etc.
Costs of testing entire supply (>14 million units / yr)
Costs of re-testing units
Loss of blood donors and blood units
Costs of recruiting donors

Benefits

Number of additional donors gained
Detection of additional malaria units from non-deferred donors

Overview of Model Components

Estimation Size of Donor Pool

INPUT DATA:

~ 8 - 9 million Total Annual number blood donors
~ 27.4 million US travelers to malaria countries
~ 382,000 Immigrants from malaria countries
~ 60 % Population qualified to donate
5 % Donation rate general population
1.7 Annual donations per donor per yr
~ 14 million Total number blood donations per yr

OUTPUTS:

> 880,000 Donors per year travel to malaria country
> 730,000 Donors - self defer for malaria risk
> 150,000 Donors - deferred by questionnaire

Estimation of malaria infection prevalence potential new donor groups

INPUT DATA :

95 - 99% Effectiveness of Questionnaire screen
(effectively lowers malaria prevalence in donors)

OUTPUTS:

~ 42 Potential mean malaria donors per year*
~ 71 Potential mean malaria donations per yr*
~ 3 Malaria units - not deferred per yr

*Most are removed by donor screening

Testing Scenarios: Universal Nucleic Acid Test (NAT)

Test all donations using NAT

Travelers (< 1yr) and Immigrants (< 3yr) to Malaria endemic countries

Assumed there was a one month window period (WP) - donors with malaria not detected

All other donors

Assumed no window period

Test Sensitivity assumed 99% - 100% sensitive

Testing Scenarios: Universal Antibody testing

Travelers (≤ 3 months) to Malaria countries

Assumed a 3 month WP - test may not detect malaria

Travelers (> 3 months) to Malaria countries

Test sensitivity - assumed to vary by species

Immigrants (≤ 3yr) to Malaria countries

Assumed no WP

All other donors

Assumed no WP

Travelers (> 3 months) to Malaria countries

Adjust test sensitivities for (>3 mo) traveler population by occurrence of species in geographic regions traveled

(1) Assumed Test Sensitivity:

P. falciparum 94% - 99.5%
P. vivax 75% - 100%
Others 50% - 75%

(2) Occurrence of species in travelers by region

	Pf	Pv	Other
Africa	82%	10%	7%
Asia	11%	83%	6%
Americas	36%	57%	6%
Others	10%	76%	14%
All regions	63%	30%	7%

Results: potential risks and benefits of alternative screening methods

	Risks (5th, 95th perc)				Benefits (5th, 95th perc)
CurrentPolicy	Blood units lost	Donors removed	Malaria units - not removed	Costs of screening	Malaria units removed	Potential donors gained
Self deferred	1,276,000	729,000	na	Assumed low Costs for recruiting	58 (48-79)	na
Questionnaire deferred	207,000	150,000	3 (1 - 5)		9 (3 - 18)	na
Total: Self + Questionnaire	1,483,000	879,000	3 (1 - 5)		67 (48 - 90)	na

Blood units collected per year in US = ~ 14 million

Results: potential risks and benefits of alternative screening methods

		Risks (5th, 95th perc)				Benefits (5th, 95th)
	Blood units collected	Blood units lost	Donors removed	Malaria units - not removed	Costs of screening	Malaria units removed	Potential donors gained
Current	~ 14 million	1,483,000	879,000	3 (1 - 5)	Assumed low Costs for recruiting	67 (48 - 90)	na
NAT testing	15,761,616	66 (46 - 87) (benefit)	40 (30 - 51) (benefit)	5 (2 - 9)	Costs > 14 million tests Re-testing of units	66 (46 - 87)	~ 880,000
Antibody testing	15,760,264	1,418 (954 - 1912) (benefit)	890 (600- 1200) (benefit)	10 (4 - 16)	Costs >14 million tests Re-testing of units	61 (43 - 81)	~ 880,000

Key Uncertainties

Overall there is uncertainty for many of model inputs

Would expect Malaria prevalence in donors with travel history (< yr) or immigrant - Malaria countries to be leading contributor to uncertainty

Variability in malaria species by region over time

Sensitivity of test that would by used

Conclusions from Malaria model

Current policy - many donors (~ 150,000) deferred or ~ 880,000 donors if include self-deferrals

Antibody testing - fewer donors deferred(~1,400)

NAT testing - even fewer deferred (66)

However, testing has significant costs associated with testing / re-testing >14 million units / yr

But, testing scenario there may be a net gain of ~ 880,000 donors

Need further exploration of costs of each option

Testing
Re-testing
Recruitment of donors

Validate assumptions (with data) on test sensitivities

Peer review of Model

Assumptions, data used, etc.

Acknowledgements

Hong Yang - CBER/OBE
Sanjai Kumar - CBER/OBRR
OBRR staff