Dengue Transfusion Risk Model

Lyle R. Petersen, MD, MPHBrad Biggerstaff, PhD

Division of Vector-Borne DiseasesCenters for Disease Control and Prevention

Blood Products Advisory Committee MeetingDecember 14, 2010

Agenda

• Arbovirus risk model• Dengue transfusion transmission risk in non-

endemic areas• Dengue transfusion transmission risk in Puerto

Rico, an endemic area of the United States• Limitations • Conclusions

General Arbovirus Transfusion Risk Model

Risk of transmission = Pd x Tr x Rs

Where

Pd = Prevalence donors viremic

Tr = Transmission rate from viremic donors to susceptible recipients

RS = Proportion of recipients susceptible to infection

Assumptions

• Pd = Prevalence donors viremic– Symptomatic persons don’t donate– Donors are similar to the population-at-large with

respect to mosquito exposure and infection risk• Tr = Transmission rate to susceptible recipients– 100%• RS = Proportion of recipients susceptible to infection– For dengue, impossible to determine under most

circumstances, so assume 100%

General Arbovirus Risk Model

• Risk derives from viremic persons who – remain asymptomatic and donate or– who donate before becoming symptomatic

• Transfusion transmission risk = Prevalence of viremic and asymptomatic persons in population-at-large

• Prevalence = incidence X duration of viremia

General Arbovirus Risk Model

Risk = (I x Pa ) x Da + (I x Ps) x Ds

Where I = Incidence of infection in population-at-large

Pa = Proportion infections that are asymptomatic

Da = Duration of viremia for asymptomatic persons

PS = Proportion of infections that are symptomatic

Ds = Duration of viremia before symptoms develop

Dengue Risk Model Variables

• Highly variable and uncertain– Incidence of infection

• Ranges from <1% to >30% per year in endemic areas

– Proportion of infections that are symptomatic• Varies by infecting strain and previous exposure to

heterologous dengue viruses• In adults in endemic areas, three studies range from 25-

50%

Dengue Risk Model Variables

• Less variable and more certain– Duration of viremia

• Approximately 5 days after onset of symptoms• Very short before onset of symptoms, thought to be

one day or less• Unknown for asymptomatic persons, but presumably

similar to that of symptomatic persons• Duration viremia < duration NAT positivity

Predicted Dengue Transfusion Risk,Key West, FL

• First dengue outbreak in Florida since 1934*• Population 23,262• First case onset July 19, 2009;

– 27 infections identified through October 19• Household-based serosurvey in conducted Sept 23-27

– Old-Town area– 4.9% determined to have recent infection

• Transfusion risk model assumptions– 4.9% population infected in 70 days (July 19-Sept 27)

• Incidence = 7 per 10,000 per day– 33% infections symptomatic– 1-day viremia before symptoms develop– 6-day total period of viremia

* MMWR 2010; 59: 577.

Predicted Dengue Transfusion Risk,Key West, FL

– Estimated average transfusion risk from Key West donors during outbreak period = 18.7/10,000

– Risk may be overestimated• 4.9% infection cumulative incidence in 70 days may not

apply to all of Key West• Outbreak duration may be longer than identified first case

– Compares to estimated transfusion risk of 0.5/10,000 during a 2004 outbreak in Cairns, Australia*

• Estimated 0.19% cumulative infection incidence in 196 days

* Transfusion 2009; 49:1482

Puerto Rico Dengue Transfusion Risk Modeling Method

• Statistical resampling approach*, which allows– Temporal estimates of risk

• Estimated viremia prevalences in population-at-large on any given day based on observed illnesses

– Uncertainty of model parameters• Random variation in model parameters within

proscribed limits to quantify uncertainty • 500 replicates• 95% confidence intervals around risk estimates

* Transfusion 2003; 43: 1007

Model Assumptions: Puerto Rico

• Proportion infections that are symptomatic • 33% (range 25-50%)

• Duration of viremia for asymptomatic persons• 6 days (range 5-7)

• Duration of viremia before symptoms develop• 1 day (range 0.5-1.5)

• Underreporting of clinical cases – 15X (range 10-20)

Estimated Dengue Transfusion Risk per 10,000 donations,

Puerto Rico, 1995-2010*

• Average risk 6.9 (4.6-9.2) • Maximum risk 51.1 (43.9-60.0)

Estimated Dengue Transfusion Risk; Puerto Rico, 1995-2010*

Year

Ris

k (p

er

10

,00

0)

19

95

19

96

19

97

19

98

19

99

20

00

20

01

20

02

20

03

20

04

20

05

20

06

20

07

20

08

20

09

20

10

06

1218

2430

3642

4854

6066

7278

84

Actual NAT Yield versus Predicted Number of Viremic Donors; Puerto

Rico, July –Dec, 2007

• NAT yield American Red Cross, Puerto Rico– 29 TMA + of 15,325 donors (18.9 per 10,000)

• Predicted number viremic donors– 29 (13-50) of 15,325 donors

Limitations• Inference of true infection incidence inferred

from disease surveillance data– Underreporting estimates uncertain– Unapparent:apparent infection ratio uncertain and

not constant over time

• Transfusion risk model only considered risk of viremic donations– True transmission rate less because of background

immunity in recipients (immeasurable in most circumstances)

Conclusions

• Variation in dengue transfusion risk mostly depends on infection incidence and proportion infections symptomatic – Both difficult to measure and variable

• Transfusion risk model plausible– Predicted risk and observed NAT yield same in

2007 in Puerto Rico

Conclusions• Estimated transfusion risk in Key West donors during 2009

outbreak– 18.7/10,000 – Model uncertainties regarding underlying incidence in

population-at-large and duration of outbreak

• Estimated annual transfusion risk Puerto Rico, 1995-2010– Average daily risk 6.9 (4.6-9.2)/10,000– Highly seasonal and variable by year– Maximum daily risk 51.1 (43.9-60.0)/10,000– Average predicted daily transfusion dengue risk over a 15-year

period similar to predicted WNV risk in 6 high-incidence states during peak of 2002 epidemics (2.1-4.7/10,000)*

* Transfusion 2003; 43:1007