Evaluating Firefly's Impact

IMG_5489.JPG

In commercial markets, profit or return on invested capital offers a relatively straightforward benchmark for performance.  Design that Matters solves problems in global health for the poor in developing countries.  As a nonprofit, we focus on creating social impact in areas not served by market-based solutions.  Engineers argue that you cannot control what you cannot measure.  If we aren’t going to use profit as a benchmark for DtM’s return on investment, what can we measure to assess our results compared to our expectations, and as compared to other potential investments in the social sector?  In other words, how might we validate our hypothesis that design really matters? 

It’s tempting to evaluate products intended for developing countries in terms of purchase price.  If an existing medical device that you might find in an American hospital costs $10,000, an equivalent device that costs just $500 might sound like a better fit for a rural hospital in a poor country.  This approach is called value engineering, but the apparent virtue can be misleading.  Saving money only makes sense if you can still achieve the desired social outcome.  The key is in evaluating total program cost per unit of social outcome.

In global health, the standard metric for the burden of a particular illness is the DALY, the disability-adjusted life year.  To calculate DALYs for a particular health condition, you start with an estimate of life expectancy for the study population.  DALYs combines the impact of morbidity and mortality into a single statistic, combining the number of years an affected person would be disabled by the disease (YLD or years lost to disability) with the time lost due to premature death (YLL or years of life lost).  

The World Health Organization provides detailed assessments for the burden of different diseases, among different populations, in terms of DALYs.  We have found estimates that ten percent of newborns with jaundice will, without treatment, develop lifelong disability, including hearing loss and brain damage, or even die.  Untreated jaundice therefore creates an expected burden of 6 DALYs: 60 disability-adjusted life years times a 10% probability rate.

As designers and innovators, we can benchmark some new intervention in terms of cost per DALY.  If some reference "bad design" costs $100/DALY, what can we say about the cost of an improved design?  

Accounting for "total cost of outcome" in $/DALY for a product like Firefly must include factors like:

  • Cost to buy: R&D cost (the marginal cost associated with developing Firefly as opposed to sticking with overhead phototherapy, amortized over the production volume), manufacturing (BOM, labor), distribution (shipping costs, import duties)

  • Cost to own: user-training, consumables, maintenance

  • Cost of failure: expected expenses associated with alternative or second-line treatment (actual cost multiplied by the probability of failure, for example the cost of an exchange transfusion)

  • Externalities: parents' lost wages due to prolonged hospital stay; opportunity cost (slower treatment time means reduced patient capacity and more patients referred to other hospitals); the risk of trading problems (jaundiced newborn contracts an MRSA infection in a difficult-to-clean device, or pneumonia while sharing a bed with three other infants under conventional overhead PT)

To elaborate on one externality example, consider the healthcare system cost of treating a newborn, and the savings that might result from a device that reduces the total treatment time (and thus the length of a newborn’s stay in the hospital).

A conventional overhead LED phototherapy device takes three days to treat newborn jaundice.  During that time, the hospital has to pay for staff time and consumables related to the newborn’s stay.  We estimate these healthcare system costs to be $15 per day, meaning a total health care system cost of $45 for each jaundiced baby.  A small hospital with a single phototherapy device that treats 100 jaundiced newborns a year will spend $4,500 on health care system costs.

Firefly is designed to provide double the light intensity of conventional overhead phototherapy devices.  Early results from the Firefly clinical trial showed a reduction in treatment time from an average of 72 hours to 52 hours of phototherapy per patient.  In other words, Firefly was allowing many newborns to go home with their parents almost full day early.  That’s a health care systems savings of $11 per baby, or $1,100 per year for a typical small hospital.

Consider a $1,500 Firefly device compared to a $500 single-sided LED phototherapy device.  Both devices will treat 500 newborns over five years of use.  

  • With a single-sided LED phototherapy device, a hospital will incur $45 in healthcare system costs per newborn treated, or $22,500 over five years.  The total cost is the $500 purchase price plus the $22,500 healthcare system cost, or $23,000.  Divided by the 500 newborns treated over five years, you get $46 per newborn treated.  Assuming treating jaundice prevents 6 DALYs, this is a total cost of $7.70/DALY.

  • With Firefly, a hospital will incur $34 per newborn treated, or $17,000 dollars over five years.  The total cost is the $1,500 Firefly purchase price plus the $17K healthcare system cost, or $18,500, a total cost reduction of $4,500 compared to overhead LED phototherapy.  Divided by the 500 newborns treated over five years, you get $37 per newborn treated.  Again assuming treating jaundice prevents 6 DALYs, this is a total cost of $6.10/DALY, or 20% reduction in total cost/DALY.

The key point is that healthcare system costs over the life of the device can be much more significant than the device’s purchase price.  We’ve argued that in social impact design, “cheap has a floor but better has no ceiling.”  The purchase price of Firefly may be three times as much as a value-engineered phototherapy device, but the overall savings from a better device are enormous.  We see savings of a similar magnitude from the way Firefly’s single-infant bassinet reduces the incidence of hospital-acquired infection, and the way Firefly’s higher-intensity lights reduce the need for exchange transfusions (an expensive and dangerous second-line treatment for severe jaundice).  


Firefly Program Cost

Program Stage Budget
User and Market Research $100,000
Product Design, Engineering & Testing $900,000
Tooling, Initial Production & Distribution (260 units) $1,400,000
Volume Production & Distribution (800 units) $1,300,000
TOTAL $3,700,000

Firefly Cost = $10/DALY

Compare to EPI vaccine at $7/DALY  (World Bank DCP 2006 est.)

Key assumptions:

  • 530K newborns treated by 1,060 Firefly units over 5 years
  • $1,700/unit for Firefly manufacture, delivery, training
  • $45/newborn for health system cost of treatment
  • 10% kernicterus rate for untreated jaundice

Firefly vs Conventional Overhead PT

Benefit Weekly Savings
Early patient release $23
Reduced newborn cross-infection $11
Averted exchange transfusion $23
TOTAL HEALTH SYSTEM SAVINGS $57

Firefly Break-Even = 7 months
Firefly Lifetime ROI = 8x

Key assumptions:

  • Firefly double-sided PT reduces treatment time 25%
  • Typical hospital-acquired infection rate of 25%
  • Infected newborns extend hospital stay 10 days @ $15/day
  • Firefly newborn isolation reduces cross-infection by 50%
  • Typical hospital: 30 exchange transfusion/yr @ $200/ea
  • Firefly averts 90% of transfusion vs 70% with conventional PT

Firefly vs Conventional Overhead PT

Line Item Firefly Overhead PT Firefly Savings
Device purchase, delivery and user-training $3,500 $2,500 ($1,000)
Cost of basic jaundice therapy $17,550 $23,400 $5,850
Cost of hospital-acquired infection $2,925 $5,850 $2,925
Cost of exchange transfusion (second-line therapy) $3,000 $9,000 $6,000
TOTAL HEALTH SYSTEM COST $26,975 $40,750 $13,775

After the treatment of 500 patients over five years, Firefly purchase price will be a fraction of the total healthcare cost.

Firefly reduces total healthcare costs by 34%


For more information, see the Firefly project page.