ORAL ABSTRACTS: INFECTIOUS DISEASE
Chronic hepatitis C (CHC) imposes considerable clinical and economic burden. Direct-acting antiviral drugs show improved sustained virologic response rates (SVR) compared to standard treatments, but associated high drug costs are an ongoing debate in Germany. We systematically evaluated long-term clinical effectiveness and cost effectiveness of different antiviral treatments for treatment-naïve and -experienced patients with genotype 1 (GT1) CHC with or without compensated cirrhosis in Germany.
We developed and applied a Markov state-transition model for the German health care context to evaluate different antiviral treatment regimens: ombitasvir/paritaprevir/ritonavir+dasabuvir± ribavirin (OBV/PTV/r+DSV±R), peginterferon+R (PR), telaprevir+PR (TVR+PR), boceprevir+PR (BOC+PR), sofosbuvir+PR (SOF+PR), simeprevir+PR (SIM+PR), SOF+ledipasvir (SOF/LDV). The model considers German epidemiological, health-related quality-of-life (HRQoL) and economic data from literature and databases. Antiviral treatment was modelled according to German drug labels. Treatment-related SVR, adverse events, discontinuation and HRQoL changes were based on international clinical trials. We adopted the perspective of the patient community insured by the German Statutory Sickness Funds. Cost and effects were discounted at 3% per year. Outcomes included reduction in lifetime risk for decompensated cirrhosis (DCC), hepatocellular carcinoma (HCC) and liver transplantation (LT), lifetime costs, quality-adjusted life years (QALY), and incremental cost-effectiveness ratios (ICER; in Euro/QALY). We applied the IQWiG efficiency frontier approach to assess and visualize incremental cost effectiveness. Comprehensive sensitivity analyses were performed.
Result(s): In the base-case analysis, OBV/PTV/r+DSV±R achieved the highest effectiveness based on patient-relevant outcomes including lifetime risk reductions of 91% (DCC), 82% (HCC) and 85% (LT) in treatment-naïve and 91%, 75%, 81% in treatment-experienced GT1 CHC patients when compared to no treatment. Based on the efficiency frontier, OBV/PTV/r+DSV±R was cost effective with an ICER of 26,423 Euro/QALY (vs. SIM+P+R as the next best non-dominated strategy) in treatment-naïve, and 16,893 Euro/QALY (vs. no treatment) in treatment-experienced GT1 CHC-patients. Results remain robust in most sensitivity analyses. The parameters SVR, discount rate, progression to more advanced liver disease and the relative utility for being HCV-RNA positive had the greatest influence on the ICER.
Based on our analyses, antiviral treatment with OBV/PTV/r+DSV±R achieves the highest benefit in terms of life expectancy, QALY and reduction in DCC, HCC, and LT. Based on our modeling analyses, OBV/PTV/r+DSV±R is considered cost effective for the treatment of patients with GT1 CHC with or without compensated cirrhosis in Germany.
Clostridium difficile infection (CDI) is a serious recurrent diarrheal infection and is associated with increases in patient healthcare resource utilization because of both extended hospitalization and re-hospitalization. Recurrent CDI (rCDI) patients are significantly more likely to have a hospital readmission and spend increased time in inpatient settings. Clinical trials MODIFY I & II demonstrated that bezlotoxumab along with standard of care antibiotics (SoC) significantly reduced the incidence of rCDI compared to SOC alone (placebo) among patients with initial or rCDI. In a post hoc analysis, we used pooled data from the MODIFY trials to estimate the cumulative hospitalized days (all settings and intensive care units (ICU)) summed over the 84-day trial follow-up period.
We used hospital admission and discharge dates to generate daily patient record of inpatient hospitalization on any day from day of infusion through 84 days post-infusion. We used these data to estimate the probability of spending a day in the hospital using a logistic model. Lin’s (2000) method was applied where the daily average probability of spending a day in hospital among those non-censored was weighed by the probability of survival to obtain estimate of cumulative hospitalized days, adjusted for survival and censoring. Treatment effects were obtained using recycled predictions for the overall sample and for subgroups identified based on trial protocol and risk of rCDI. Standard errors, p-values, and 95% confidence intervals (CI) were obtained using 1000 bootstrap replicates. We repeated the analysis using patient admission to ICU to estimate cumulative ICU-hospitalized days (a cloglog model provided a better fit for the probability model).
The results for cumulative hospitalized days summed over an 84-day trial period, adjusted for censoring and survival, are provided in Table 1. Mean cumulative hospitalized days in the placebo arm (14.2 days) was greater than that for the bezlotoxumab arm (12.1 days). Overall, the mean difference between treatment groups was 2 days (CI: 0.3, 3.7). The mean difference was greatest for patients with clinically severe CDI (3.4 days). The difference in mean cumulative ICU-hospitalized days in the placebo vs. bezlotoxumab arm was 0.27 days (CI: -0.32, 0.87) about 13.4% of the difference in mean cumulative hospitalized days.
Bezlotoxumab is associated with reduction in total number of days spent in a hospital.
Method(s): We study `critical vaccination coverages', which are vaccination allocations that result in R=1. We show that for this threshold the introduction of the disease in a population does not result in an outbreak. In a population with interacting subpopulations there are many different critical vaccination coverages. To find the most efficient one, we define the following optimization problem: minimize the required amount of vaccines to obtain R=1. We prove that this optimization problem is equivalent to the problem of maximizing the proportion of susceptibles that escape infection during an epidemic (i.e., maximizing the herd effect). This herd effect is directly related to the final size of an outbreak.
Result(s): We propose an efficient general algorithm based on Perron-Frobenius theory to solve these optimization problems. We study two special cases that provide further insight into these optimization problems: the case of separable mixing and the case of n=2 populations. The case of separable mixing is often studied and assumes that upon transmission from one individual to another the two individuals involved influence transmission independently. For these two special cases we are able to characterize the optimal solution completely. The algorithm for separable mixing provides especially interesting insights: we show that vaccinating according to a very simple priority ordering based on population size and disease parameters results in the optimal allocation. We illustrate our approach to find the optimal allocation in a case study for pre-pandemic vaccination in the initial phase of an impending influenza pandemic.
Conclusion(s): The results of the case study show that using the optimal allocation determined with our solution methods can increase the herd effect by 9 to 26% compared to proposed allocations in literature. Equivalently, our optimal allocation is able to significantly reduce the required vaccine stockpile to attain a reproduction ratio of one.
Method(s): We used an age-structured dynamic transmission model to estimate the burden of rotavirus disease from 2015 to 2019, with and without vaccination. We assumed vaccine efficacy to be 93% against severe infection and 60% against symptomatic rotavirus infection. Using real-time immunization data, we modeled an increase in vaccination coverage after the program start from 77% in 2015 to 91% during 2016-2019. We evaluated a two-dose vaccination program with Rotarix® at a baseline cost of €24 per dose and a three-dose program with Rotateq® at a cost of €20 per dose. We developed an economic model to calculate costs and health effects of vaccination during the study period and estimated cost-effectiveness of the program in 2019 from a healthcare and societal perspective. We measured cost-effectiveness as cost per quality adjusted life year (QALY) using a threshold of €73,444 per QALY. Parameter uncertainty was accounted for using Monte Carlo methods. The estimates of vaccine price, disease burden, vaccine coverage, medical costs, production losses, QALY losses, and discount rate used in the model were varied in sensitivity analyses.
Result(s): During the first five years following vaccine introduction, rotavirus-associated deaths and hospitalizations are expected to be reduced by 67%, primary care consultations by 63%, and home care episodes by 57%. Vaccination was cost-effective from the healthcare perspective resulting in a cost of €1,379 per QALY for Rotarix® and €25,215 per QALY for Rotateq®. From the societal perspective, vaccination was found to be cost-saving resulting in €332,451 per QALY gained for Rotarix® and €308,615 per QALY gained for Rotateq®. Vaccination from a societal perspective remained cost-saving when productivity losses were reduced by 50%. Overall, vaccination resulted in an 82% reduction in rotavirus hospital costs, mainly due to avoided inpatient hospitalizations. A similar reduction (82%) was predicted for primary care costs. The cost-effectiveness of the program was most sensitive to changes in the estimated vaccine costs and QALY losses.
Conclusion(s): Ongoing rotavirus vaccination in Norway is cost-effective from both healthcare and societal perspectives, and is expected to reduce the burden of rotavirus disease considerably.
Clostridium difficile infection (CDI), a form of infectious diarrhea that can recur repeatedly after treatment, is associated with considerable morbidity, mortality and healthcare resource utilization. There were 453,000 new CDI episodes resulting in 29,000 deaths in 2011 in the U.S. Bezlotoxumab is a novel antitoxin agent that, when used in conjunction with SoC antibiotic therapy, prevents recurrent CDI (rCDI), leading to sustained clinical response. The purpose of this research is to model and evaluate the cost-effectiveness of Bezlotoxumab+SoC compared with placebo+SoC patients with CDI in the U.S.
We developed a computer-based Markov health state transition model to simulate the natural history of CDI (Figure 1). In the model, we followed patients with CDI from infection until death and evaluated the costs and effectiveness of bezlotoxumab+SoC compared with Placebo+SoC using a third-party payer perspective. To evaluate cost-effectiveness in different patient population, we conducted our analysis for the entire clinical trial population (subgroup 1), and for CDI patients at higher risk of rCDI—age 65 years and above and having a history of CDI (subgroup 2). Recurrence rates after infusion for bezlotoxumab and placebo were taken directly from the pooled MODIFY I & II phase III clinical trials' efficacy data. Other transition probabilities and costs of rCDI were obtained from the literature. We projected rCDI averted, discounted age-weighted quality-adjusted life years (QALYs), and threshold prices at which Bezlotoxumab would be cost-effective at the $100,000/QALY threshold.
The model predicted that treating patients with
bezlotoxumab+SoC will reduce the combined incidence of first, second, and third
CDI recurrences after infusion by 16.4% and 39.4% in subgroup 1 and subgroup 2,
respectively. This resulted in 0.16 and 0.28 incremental discounted
age-weighted QALYs gained per-patient for subgroup 1 and subgroup 2,
respectively. The threshold price at which bezlotoxumab is cost-effective at
the $100,000/QALY threshold is $17,188 and $30,118 for subgroup 1 and subgroup
2, respectively. Key influential parameters include CDI-specific mortality,
cost of a rCDI episode, and underlying recurrence rate.
Based on the Markov model, bezlotoxumab has the potential to reduce the disease burden associated with CDI in a cost-effective manner, by reducing the incidence of rCDI.
Figure 1: Natural History of CDI
Budget Impact of Fidaxomicin compared with Vancomycin for the treatment of Clostridium difficile associated diarrhea in the United States
Clostridium difficile–associated diarrhea (CDAD) is the leading cause of nosocomial diarrhea. In 2011, there were 453,000 new CDAD episodes resulting in 29,000 deaths in the U.S. Fidaxomicin is an antibacterial drug indicated for treatment of CDAD. Compared with vancomycin, it has a comparable clinical response and superior sustained clinical response, defined as clinical response at the end of treatment, and survival without proven or suspected CDAD recurrence through 25 days beyond the end of treatment. We present a model to evaluate the budget impact of fidaxomicin compared with vancomycin patients with CDAD in the U.S.
We developed a decision-analytic model to simulate the natural history of CDAD among a cohort of 1000 patients. In the model, we followed patients with CDAD for a period of 25 days post-treatment and evaluated the annual budget impact of fidaxomicin compared with vancomycin using a hospital payer perspective. The probabilities of clinical response, sustained clinical response, and death were obtained from the fidaxomicin product insert. CDAD recurrence through 25 days beyond the end of treatment was calculated as persons who did not have a sustained clinical response, death, or a clinical failure. Wholesale acquisition cost of vancomycin was obtained from the analysource.com database. Cost estimates of CDAD were obtained from the literature and indexed to 2015 cost figures using CPI medical inflation data. We then projected clinical response, sustained clinical response, and the threshold price at which fidaxomicin would be cost-neutral.
The model predicted that among the 1000 patients treated, 880 patients achieved clinical response in the fidaxomicin arm compared with 865 patients in the vancomycin arm (Figure 1). 709 patients obtained sustained clinical response through 25 days post-treatment in the fidaxomicin arm, compared with 570 patients in the vancomycin arm. The threshold price below which fidaxomicin was cost-saving was $202.
Based on the decision-analytic model, fidaxomicin has the potential to be cost-saving from a hospital payer perspective.
Figure 1: Estimated number of patients in various health states at 25 days post-treatment