Table 15 presents the point estimates for the social EPV model parameters and assumptions. The following sections discuss the basis for these estimates in further detail.
-
3.6.1 Real Annual Social Rate of Discount
-
In accordance with the Office of Management and Budget (OMB) guidelines for economic analysis, we use a real social discount rate of 3 percent in the analysis. The social discount rate is assumed to have a triangular probability distribution with a lower bound of 1 percent and an upper bound of 7 percent in the sensitivity analysis.
Table 15: Social EPV Model Parameters and Assumptions (Point Estimates)
Model Parameter/Assumption ABOM ABSSSI CABP CIAI CUTI HABP/VABP Real Annual Social Rate of Discount 3% % of Patients Not Responding to Existing Drugs 20% % Increase in Duration in Patients Not Responding to Existing Drugs 50% Loss in Quality of Life, Acute 0.11 0.36 0.15 0.5 0.27 0.17 Duration (days) 10 6 4 10 4 8.5 Loss in Quality of Life, Convalescence 0.04 0.36 0.1 0.15 N/A N/A Duration (days) 20 18 5 12 N/A N/A Lost QALYs per illness 0.0049 0.0239 0.0038 0.0023 0.0030 0.0040 Total Number of Cases per Year (unadjusted for population growth) 13,200,000 726,000 1,170,000 72,000 1,083,000 272,600 Mortality Parameters Deaths 0 1,923 51,683 14,554 36,900 81,779 Lost QALYs for Patients that Die 0 26,167 572,741 243,987 319,913 1,848,212 VSL per Patient N/A $5,623,708 $5,301,924 $5,585,504 $4,953,688 $4,770,000 Morbidity Parameters Number of Patients that Survive 13,200,000 724,397 1,118,000 57,489 1,045,986 190,818 Lost QALYs for Patients that Survive 65,248 17,336 4,295 1,632 2,432 756 WTP (VSLY*Lost QALYs) per patient $1,124 $8,749 $1,113 $12,717 $758 $1,149 N/A = Not applicable
QALY = Quality-adjusted-life-year
VSL = Value of a statistical life
VSLY = Value of a statistical life year
WTP = Willingness-to-pay
The figures in the table are rounded for presentation purposes.
-
-
3.6.2 Percentage of Patients not Responding to Existing Commonly Used Antibacterial Drugs
-
The percentage of patients that do not respond to existing commonly used antibacterials is difficult to estimate on a nationwide as well as on an indication basis. Antimicrobial resistance varies widely by hospital and geographic region, depending on local resistance patterns and standard prescribing practices.
Published estimates of indication-specific antimicrobial resistance are scarce. Evans, et al. (2007) find that out of 604 surgical admissions treated for at least one Gram-negative rod (GNR) infection in a university hospital surgical intensive care unit and ward, 137 (23 percent) were due to infections with GNR resistant to at least one major class of antibacterial drugs (rGNR). In a later study, Roberts, et al. (2009) report that in a sample of 1,391 patients in a Chicago area hospital, 188 (13.5 percent) had antibacterial drug-resistant infections.11 According to a 2009 report by the European Center for Disease Prevention and Control (ECDC) and the European Medicines Agency (EMEA), resistance to antibacterial drugs is high among Gram-positive and Gram-negative bacteria that cause serious infections in humans and reaches 25 percent or more in several EU Member States.12
One expert interviewed for the study estimated that roughly a third of all hospital-acquired infections are resistant to standard antibacterial drugs but that resistance is increasing more slowly in the outpatient setting. The expert further speculated that if approximately 30 percent of infections are resistant to antibacterial drugs in hospitals, the rate of resistance in the outpatient settings might range from 10 percent to 15 percent. A large pharmaceutical executive noted that resistance to commonly used antibacterial drugs currently ranges from 20 percent to 25 percent according to internal research conducted by his company.
Table 16 summarizes the antimicrobial resistance data available for the model estimate. As can be observed from the table, reported estimates of antimicrobial resistance are highly varied. Based on the collective body of evidence available, we use 20 percent as the average percentage of patients not responding to existing commonly used antibacterial drugs in the U.S., independent of type of indication. For sensitivity analysis, we assume that the parameter follows a uniform probability distribution with a lower limit of 10 percent and an upper limit of 25 percent.
11 In the study, an infection was defined as antibacterial resistant if the implicating organism(s) fell into one of four subgroups: (1) methicillin-resistant Staphylococcus aureus, (2) vancomycin-resistant enterococci, (3) Escherichia coli resistant to fluoroquinolones or third-generation cephalosporins or Klebsiella species resistant to third-generation cephalosporins (AREK), and (4) amikacin- or imipenem-resistant Enterobacter, Pseudomonas, or Acinetobacter species (AIR).
12 In the study, the antibacterial resistant gram-positive bacteria included Staphylococcus aureus, Enterococcus spp. (e.g., Enterococcus faecium), and Streptococcus pneumoniae. The gram-negative bacteria included Enterobacteriaceae (Escherichia coli and Klebsiella spp.), and Non-fermentative Gram-negative bacteria (Pseudomonas aeruginosa).
-
-
3.6.3 Percentage Increase in Disease Duration for Patients Not Responding to Existing Commonly Used Antibacterial Drugs
-
There are no publicly available estimates of how long each of the different types of illnesses last in patients who do not respond to existing commonly used antibacterial drugs. Thus, for analysis purposes, we assume that those patients who do not respond to existing drugs have an average duration of illness 50 percent longer than those who do respond. For sensitivity analysis purposes, we further assume that the parameter follows a uniform distribution with a lower bound of 25 percent and an upper bound of 100 percent.
In the analysis, we further assume that 1) all those not responding to existing commonly used antibacterial drugs respond to the new drug and 2) their duration of illness is reduced to the average of those responding to existing drugs.13 Combining these assumptions, we then calculate that a new antibacterial will reduce the total social burden of illness by about 9 percent. This estimate is highly uncertain, as we do not know the actual improvement in patient response to a hypothetical new antibacterial drug. It should also be noted that this does not imply that the new antibacterial drug will avert 9 percent of deaths attributable to the different types of indications studied here. Given data limitations, our analysis cannot distinguish between avoided mortality and morbidity cases due to the new antibacterial drug. We only are able to compute overall reductions in the total social burden of illness due to these new drugs in monetary terms.
Table 16: Reported Estimates of Antimicrobial Resistance
Source % of Patients Resistant to Commonly Used Antibacterial Drugs Evans, et al., 2007 [a] 23.0% Roberts, et al., 2009 [b] 13.5% ECDC/EMEA Joint Technical Report, 2009 [c] Methicillin-resistant S. aureus (MRSA) 25.0% Vancomycin-resistant Enterococcus faecium 8.0% Penicillin-resistant S. pneumonia 4.0% Third-generation cephalosporin-resistant E. coli 9.0% Third-generation cephalosporin-resistant K. pneumoniae 20.0% Carbapenem-resistant P. aeruginosa 19.0% Expert 1 Inpatient 30.0% Outpatient 10.0% - 15.0% Expert 2 20.0% - 25.0% [a] Based on a sample of 604 surgical admissions treated for at least one Gram-negative rod (GNR) infection [b] Based on a sample of 1,391 patients in a Chicago area hospital
[c] Based on European Antimicrobial Resistance Surveillance System (EARSS) for EU Member States, Iceland
and Norway for each year during the period 2002–2007
13 This is a simplifying assumption that likely leads to an overestimation of social benefits. In reality, there will be time lost for the patient due to being on the wrong drug initially.
-
-
3.6.4 Value of a Statistical Life (VSL)
-
To calculate VSL, we first took the value of a statistical life reported in 2000 dollars by age group from Aldy & Viscusi (2008). Next, it was necessary to adjust the VSL values by age group to capture changes in real income as well as prices from 2000 to 2011. Current data from the U.S. Bureau of Economic Analysis (BEA) show that the real personal income per capita was $28,888 in the year 2000 and $32,635 in 2012 (both in 2005 dollars), yielding a growth rate of 13 percent over this span of time. Moreover, Hammitt & Robinson (2011) report that U.S. regulatory agencies generally assume that a 1.0 percent change in real income over time will result in a 40 to 60 percent change in the VSL. Using the midpoint of this range (50 percent), we inflated the reported VSL values by age group by 1.065 (= 1 + [0.5 × 0.13]) to account for changes in real income from 2000 to 2011. To adjust the VSL values for price changes, we used the general consumer price index-based inflation calculator (available on the Bureau of Labor Statistics website) that shows an average price increase of approximately 31 percent over the same time period. We then calculated the age-specific VSLY and apply it to the estimated number of years of life lost for each condition.
-
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®
