Analytical Framework for Examining the Value of Antibacterial Products. 3.5 Expected Present Value (EPV) Framework for Evaluating Social Returns


In addition to estimating the private expected net present value (ENPV) of a new antibacterial drug to the sponsor, we also need to consider the expected net present value (ENPV) of the new antibacterial drug to society as a whole.  The social ENPV considers the societal costs and benefits of these drugs. From the perspective of economic efficiency, providing incentives to private individuals to develop new antibacterial drugs makes sense if the social ENPV of these drugs is positive, but the private ENPV is insufficient for sponsors to produce these drugs.

The methodology we employed for evaluating social ENPV for each of the six indications involved the following steps:

Step 1 – Estimate the Value of the New Antibacterial Drug to the Individual.  We estimate the burden of experiencing an illness to the individual for each indication we study.  We consider two potential cases for each illness:

  • Morbidity – the individual becomes sick, then returns to full health, and
  • Mortality – the individual dies as a result of the disease

To measure the burden of morbidity, we use Quality Adjusted Life Years (QALYs).  QALYs measure the equivalent number of years of life in perfect health lost as a result of the illness, and are widely considered to provide some measure of a patient’s lost “utility” or preference due to illness (although economists might argue that it is not derived from a well-defined utility function).  Intuitively, the QALY weight is a preference ranking bounded by 0 and 1 that reflects a person’s state of health (1 signifies perfect health, and 0 indicates health equivalent to being dead).7  We estimate two key data elements: 1) the QALY weight, and 2) the average duration of the illness.  Tufts Medical Center maintains a searchable online “Cost-Effectiveness Analysis Registry” that includes QALY weights (Cost-Effectiveness Analysis Registry); which is where we obtained weights to use for each indication.  We then adjusted these weights by period of illness using the following equation, which represents the reduction in QALYs multiplied by the percent of a year during which the illness is experienced:8, 9

Lost QALYs per Patient

Average duration of illness was obtained from the literature on each indication as described in further detail in the following sections below.

The data elements required to calculate lost QALYs for patients who die are: 1) the age at which a patient dies due to the infection, and 2) the typical life expectancy for the age at which the patient dies.  In the case of death, the QALY weight is zero, so the loss of QALYs for a single patient for a single year of life would be 1.0, and the period over which the QALYs were lost would be the years of life lost.10

Step 2 – Estimate Annual Societal Burden by Extrapolating the Individual Burden.  To calculate societal burden, we estimated the total number of illnesses and deaths associated with each of the indications in the U.S. per year.  In most cases, the total number of cases per year was available or derived from the literature on each indication.  Using population estimates from the U.S. Census Bureau, we converted the total number of episodes (by age group, where possible) to rates and applied those rates to 2011 population figures to obtain a total number of cases for the year 2011 for all indications (U.S. Census Bureau, 2008).  Additionally, mortality rates for each indication were either derived from the clinical literature or calculated using the Compressed Mortality File (from Centers for Disease Control and Prevention, National Center for Health Statistics, available online from CDC WONDER).

Step 3 – Monetize Societal Burden of Illness.  We monetize our estimates of QALYs lost by using the value of a statistical life year (VSLY).  The VSLY is based on the value of a statistical life (VSL).  The VSL is a measure of how much consumers are willing to pay for a small reduction in their probability of dying.  This small amount is then aggregated over the probabilities to give the VSL.  For example, if consumers are willing to pay $400 to reduce their risk of dying by one in 20,000, then the VSL is $8,000,000 (= $400 × 20,0000).  We then extrapolate the VSLY from the VSL by amortizing it at a 3 percent real discount rate over the remaining years of expected life.  Thus, we applied the value of that year of life to the duration of the illness and the loss of utility from that illness to place a monetary value on the lost QALYs.  This approach is clearly a mere approximation of WTP and has been criticized in the past for using a constant estimate of VSL instead of allowing VSL to vary over a person’s life span.  Although still controversial, in the absence of direct, valid estimates of WTP, this approach provides a usable alternative.

Step 4 – Calculate Net Present Value (NPV) of the Total Societal Burden of Disease for the Projected Useful Life of the New Antibacterial (i.e., 20 Years).  Using the annual monetized societal burden of disease computed in Step 3, we estimated the 20-year burden by adjusting for population growth and using a 3 percent social discount rate.

Step 5 – Estimate Reduction in Total Societal Burden of Disease due to the New Antibacterial Drug.  From a societal perspective, social benefits accrue only if the new drug offers a therapeutic benefit over existing drugs in terms of reductions in morbidity or mortality.  Therefore, we assumed that the new antibacterial drug must somehow be an improvement over existing drugs: patients get better faster, and/or patients that are resistant to existing drugs will not be resistant to the new drug.

To derive this benefit, we used estimates of 1) percentage of patients that are not effectively treated by existing drugs, and 2) percentage increase in disease duration in those patients that do not respond to existing therapies compared to those that do respond.  Combining these assumptions, we then calculated the percentage reduction in the total social burden of illness for the new antibacterial (see Sections 3.6.2 and 3.6.3 below).

Step 6 – Calculate Social EPV at the Model Reference Point (i.e., Start of Pre-clinical Phase).  The value computed in Step 5 represents the social NPV at the point of product launch (i.e., the social value that corresponds to the uppermost end node on the decision tree depicted in Figure 2).  To be able to compare private ENPV to social EPV, we rolled back the social NPV using the respective success/failure probabilities for each decision tree branch all the way back to the model reference point.

7 It is possible to conceive of cases in which the health state of a living person is zero, or even negative, however the intuitive explanation provided here is adequate for our current purposes.

8 The key assumption here is that the quality of the patient’s life is reduced by the value of the QALY weight.  For example, a patient suffering from a preexisting condition might have an initial health state value that is less than 1; however, as long as the illness further reduces the patient’s health state by the value of the QALY weight, the calculation is still correct.  It is plausible that the combined effect of two or more health conditions will reduce the patient’s quality of life by less than the sum of the individual QALY weights, in which case this calculation will overestimate the reduction in QALYs; however we did not have data to support such adjustments.

9 Assuming an initial health value of 1 will overestimate benefits as most individuals are not in full health.

10 This method is an overestimate of the QALYs lost due to death because absent the illness that caused death, the individual would not have lived for the rest of his or her life in a state of perfect health.


View full report


"rpt_antibacterials.pdf" (pdf, 1.43Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®