Egypt has the highest prevalence of hepatitis C virus (HCV) in the world (1) . HCV was estimated to affect 10% of 15-59 year olds in 2015, therefore making it one the country’s leading public health concerns (2). As such, choosing an affordable and effective treatment option is a priority in allocating Egypt’s healthcare budget.
HCV infection occurs via blood contact and vertical transmission. In Egypt, the high rates are attributed to a mass parenteral anti-schistosomiasis campaign that used unsterilized and shared syringes and needles (1). The most common HCV genotype in Egypt is genotype 4, accounting for 85% of HCV cases (3).
HCV can be treated using antiviral medications, with the goal of achieving a sustained virological response (SVR) following treatment. Historically, pegylated interferon and ribavirin (PIR) has been the primary treatment option, but its success in achieving SVR is only ~40% for those with genotype 4 (Table 1) (2). New, more costly direct acting antiviral (DAA) medications, such as Ledipasvir and Sofosbuvir (LS) are more effective across all genotypes, achieving ~90% rates of SVR, and carry fewer side effects (2). The Egyptian government has been negotiating a 90% discount for LS, reducing the treatment cost from 400,000 Egyptian pounds (EGP) to 40,000 EGP (Table 1). Therefore, the government has requested a rapid cost effectiveness analysis (CEA) of LS, using PIR as the relevant comparator.
CEAs are the most common type of economic evaluation used in the healthcare sector as they permit comparisons of two treatment options to a single effect, such as quality adjusted life years (QALYs) gained (4). CEAs identify health interventions that offer large enough health gains, relative to costs, to warrant adoption (5). Results of CEAs are expressed using an incremental cost effectiveness ratio (ICER), which is the ratio of the difference in costs to the difference in effects between two treatments (4).
A Markov model was developed using Microsoft Excel to estimate the cost effectiveness of LS compared to PIR. Markov models are useful for chronic conditions like HCV, as they allow for the evaluation of several health states over a specified time. This is beneficial as it reflects disease progression, and takes into account that some variables are time dependent, such as all-cause mortality (4).
Figure 1: Markov Model Schematic
The Markov model simulates a hypothetical cohort of individuals with HCV through a lifespan, dividing it into equal periods (cycles) (6) (Figure 1). In each cycle, there is a risk (defined by transition probabilities) that an individual will enter the next progressive health state, or death (from disease or all-cause mortality, based on age) (4).
The model was simulated using 2,000 people, divided equally between PIR and LS treatments at cycle 0. All people entered the model as having genotype 4, in the moderate health state, at age 40. In cycle 0, people are treated with either PIR or LS for 6 months, and either achieved SVR or remained in the moderate state. The transition probabilities associated with progressing through disease states, and the all cause age specific mortality rates, determined how people moved through disease states, and to death, in each annual cycle.
All input parameters are presented in Table 1. Data used for the model was provided by the Egyptian government. Treatment regimen effectiveness was derived from non-randomized comparisons across different RCTs and non-randomized studies.
It was assumed that the transition probabilities for the health states were the same for each treatment. People who achieved SVR remained in that health state, unless they transitioned to death through all-cause mortality. People in SVR were assumed to incur no further costs. Hepatocellular carcinoma (HCC) and decompensated cirrhosis were jointly termed as liver failure (LF), and represented a single health state. It was assumed that if people achieve SVR, they cannot be re-infected with HCV, and if they do not achieve SVR, there were no further treatment options. Similar levels of adherence were assumed over both treatment groups. It was also assumed that everyone progresses sequentially through the disease stages, and that the disease process is not reversible.
A lifetime time horizon was chosen for this study, as this allows for consideration of all potential costs and benefits of the treatment options (7). Lifetime was taken to be the natural length of life in the population, and the Egyptian life expectancy of 71 years (32 cycles) was therefore used, as recommended by Gates reference case guidelines (8), (9).
A health services perspective was chosen, capturing all healthcare costs and outcomes. This perspective is recommended by the Gates reference case guidelines, and costs aside from direct healthcare costs were not available (8). Table 1 outlines the costs provided for the health states, including costs associated with treatments and care for each of the states and costs of the treatments (under the assumption that the government negotiates at 90% discount on LS). The model assumed similar monitoring costs between the two treatments, and there were therefore not included in the model.
QALYs were the chosen outcome measure for this model, and life years were also calculated. QALYs are a generic measure of disease burden and are calculated by adjusting the time spent in a disease state by the utility weight of that state (Table 1) (10). This is an appropriate outcome measure since individuals with HCV have lower quality of life (QoL) compared with healthy individuals. Also, by incorporating utility weights, the model considers how different treatment options both extend life, and the quality of that extended life (4).
QALY utility values for Egypt were not available, so values from UK patients were used (7). Though it is assumed that there were no side effects (disutility) from receiving treatment, this is acknowledged as a potential limitation of the study, given that LS is known to have fewer side effects than PIR (2).
Discounting was done to costs and outcomes to reflect society’s positive rate of time preference, meaning that people prefer to receive benefits sooner than later, and that they would rather incur costs later than sooner (4). A discounting rate of 3.5% per year was used as per Egyptian guidelines (7).
One way sensitivity analysis
A one way deterministic sensitivity analysis (DSA) was performed for all point estimate parameters- disease state transition probabilities, QoL weights and costs, and treatment effectiveness and costs. 95% confidence intervals (CIs) were used as upper and lower bounds to test how sensitive the ICER was to the given parameter, as recommended by Drummond et al. (see Table 1 for 95% CIs) (10). All-cause mortality was not included, as these estimates are based on large numbers from the general population, and because standard errors (SEs) were not provided.
To address methodological uncertainty, time horizon was extended to 61 and 80 cycles (reflecting old age and until everyone in the model had died), and discounting rates were varied between 2% and 6%, as per Egyptian reporting guidelines (7). Additionally, the uncertainty regarding the discount was analyzed, using a cost of 400,000 EGP for LS. Alternative treatment assumptions were tested in two further analyses, one being treating with PIR and LS in every cycle, and one being treating everyone with PIR in every cycle, but only once with LS.
Probabilistic sensitivity analysis (PSA)
To address the uncertainty of parameters simultaneously, a Monte Carlo PSA was conducted. Parameters were assigned appropriate distributions, using beta distributions for probabilities (transition states and treatment effectiveness) as these are bounded by 0 and 1, lognormal distributions for QoL weights as these are bounded by 1 but could be less than 0, and gamma distributions for costs as these cannot be negative and are not bounded by 1 (4) (Table 1 contains distributions and probabilistic values). SEs were used to transform the parameter to the assigned distribution. Probabilistic values were drawn at random from each of the parameter distributions, and 1000 simulations were run.
The results of the outcomes and costs for each drug treatment are provided in Table 2. The ICER for the base case scenario was 20,868 EGP/QALY.
A Tornado diagram (Figure 2) demonstrates the impact that the input parameters had on the overall ICER. As shown, the QoL weights for SVR and moderate, and the cost of LS were the parameters the model was most sensitive to. To explain, if the lower range for the CI for QoL for SVR is used, the resultant ICER increases to 111,868 EGP. This is useful as it may point to whether focusing research in this area to obtain a more certain estimate of QoL of SVR would be of value in reducing the uncertainty in the model. The model was most robust to the following parameters: transition probabilities for cirrhosis to LF and from LF to death, costs of being in moderate and LF, and the QoL of LF state.
Figure 2 Tornado
The model was also tested to see if it was sensitive to alternative assumptions. When time horizon was extended, the ICER decreased (Table 2). This demonstrates the intuitive thought that, should a longer time horizon be chosen, the ICER will fall, as more benefits reaped from the initial costs will be captured. However, one must be cautious to use these smaller ICERs to inform decision making, as it is unlikely to be the case that a person will survive to age 119 (when everyone in the model dies, or theoretically, when all costs and outcomes of the treatments are accounted for). Therefore, using this as a time horizon is misleading, even if it captures more outcomes from LS than using life expectancy.
Alternative assumptions regarding treatment frequency demonstrated higher ICERs than the base case scenario, meaning that, for the analyses conducted, the current idea of treating patients once only in cycle 0 is the most cost effective treatment option. This is because the additional costs associated with treating patients every cycle is not found to reap significant outcome benefits compared to the base case scenario.
Probabilistic sensitivity analysis
After running 1000 simulations using stochastic variables, the resultant average ICER was 21,433 EGP (near to the base case ICER of 20,868 EGP). The simulations of incremental cost and QALYs were plotted on a cost effectiveness plane (Figure 3). The clustering of points in the north-east quadrant indicates that, for the majority of simulations, LS is more effective and more costly compared with PIR. The dots in the other planes reveal that this was not always the case, indicating that there is some uncertainty regarding whether LS is cost effective.
Figure 3 Cost Effectiveness Plane
The two lines on the cost effectiveness plane indicate two potential cost effectiveness thresholds (CETs) for Egypt (see discussion for rationale). For each of the CETs, the points below the line are cost effective, and the points above the line are not (note that more dots fall under the solid line with the higher CET than the dashed line).
A cost effectiveness acceptability curve (CEAC) shows the probability that LS is cost effective for a range of CETs (Figure 4). As seen by the solid line, at a CET of 52,120 EGP, there is a 0.84 probability that LS cost effective. The CET shown with the dashed line demonstrates that LS is less likely to be cost effective at that CET.
Figure 4 Cost effectiveness curve
This report was commissioned to determine if treatment with LS is cost effective in Egypt for patients with chronic HCV infection. To answer this, a cost effectiveness threshold (CET) needs to be delineated. In the literature, multiple methods have been suggested to determine CETs, such as the WHO-CHOICE recommendation for low and middle income countries (LMICs) of 1-3x the GDP per capita (11). Woods et al. suggested that this CET is too high, and use opportunity costs to estimate a CET range for Egypt (5). In this report, the midpoint values of the threshold ranges from the WHO and Woods et al. methodologies to estimate CETs (52,120 EGP and 32,941 EGP, respectively) are used to illustrate the likelihood that LS is cost effective in the Egyptian context (see appendix for calculation methodology). However, it is important to note that it is up to the Egyptian government to determine what they are willing to pay for to adopt LS, and that the thresholds used in this report, while informed by literature, are illustrative. Using both the WHO and Woods et al. thresholds, based on the PSA, LS is found to be cost effective 84% and 70% of the time, respectively. This suggests that introducing LS as the primary HCV treatment in Egypt could be cost effective.
There are limitations to using Markov models, such as the assumption that transition probabilities are constant over time (6). The treatment effectiveness data and transition probabilities were from one study-ideally data from a meta-analysis of RCTs would be used (7). Using utility weights derived from UK patients with HCV may be problematic, and ideally, utility would be derived from the direct use of a EQ-5D questionnaire taken in Egypt (7).
Using a health services perspective was justified because of the audience of report and the data made available to conduct the evaluation. An additional analysis considering relevant non health costs and outcomes would be useful to decision makers (8).
This model was influenced by numerous assumptions. First, alternative treatment regimens were not considered, such as consideration other DAAs, or drug combinations (3). Second, not including decompensated cirrhosis and HCC as distinct health states limits the validity of the model, for one, because it inaccurately assumes that transition to death for both of these conditions would be the same (3). Assuming there were no further treatment options if SVR is not achieved is not realistic-liver transplants are used in Egypt to treat patients with LF, and not including them is a limitation (12). Assuming that people who achieve SVR cannot be re-infected is not clinically accurate, and incorporating age specific annual incidence rates of HCV into the model could account for this (12). Additionally, assuming equal treatment lengths between PIR and LS is contestable, as LS treatment length is usually much shorter than PIR (1),(13). Since PIR is known to have more side effects than LS, assuming similar adherence between the groups is problematic. Accounting for PIR side effects in the model by incorporating disutility weights would have resulted in a smaller ICER, increasing LS’s likely cost effectiveness.
Despite high prevalence of genotype 4, not examining treatment effectiveness in different groups is a limitation. Heterogeneity could be accounted for by disaggregating the population into subgroups (for example, by genotype, virus onset, region, public vs private insurance coverage) and conducting a stratified analysis to highlight differences in cost effectiveness between subgroups (7). This could improve generalizability, which is currently limited, as significant regional differences in HCV prevalence, healthcare practices and treatment accessibility do exist (7),(1). These challenges are not unique to this evaluation, however, as acknowledged by Elsisi et al.: “HTA implementation in Egypt, however, is significantly challenged by the diversity and heterogeneity of the health care system, limited tradition for national treatment guidelines, and limited availability of epidemiological, health outcomes, and cost data“ (7)326.
A Markov model was used to determine if implementing LS as treatment for patients with HCV in Egypt is likely to be cost effective, compared to existing treatment (PIR). There were several assumptions made when constructing the model, and the potential limitations of those were acknowledged and discussed. The results suggest that LS might be cost effective in treatment of moderate HCV patients. However, this is dependent on other considerations, including what Egypt determines their CET to be, whether a discount for LS will be negotiated, as well as what other public health priorities the government has chosen to focus its resources on. Further research on things like treatment effectiveness and utility weights would improve the quality of the study.
Appendix: CET Calculations
For consistency, all data used were from 2015
WHO-CHOICE Method: 1-3x GDP/capita
Egyptian GDP/capita 2015: 26,060 EGP (14)
Threshold range: 26,060 EGP-78,180 EGP
Midpoint: 52,120 EGP
Woods et. al Method: Opportunity Costs
Here, the USD, PPP adjusted range provided by Woods et al. (2015) was used, and converted to EGP using 2015 USD to EGP currency exchange rates, since the initial range was already PPP adjusted.
CET range provided by Woods et al.: 1,745-6,669 USD, PPP adjusted (5)
USD to EGP exchange rate for December 2015: 7.83 (15)
Converted threshold range: 13,663-52,218 EGP
Midpoint: 32,941 EGP
- Gomaa A, Allam N, Elsharkway A, El Kassas M, Waked I. Hepatitis C infection in Egypt: prevalence, impact and management strategies. Hepatic Medicine : Evidence and Research. 2017;9:17-25.
- Kandeel A, Genedy M, El-Refai S, Funk AL, Fontanet A, Talaat M. The prevalence of hepatitis C virus infection in Egypt 2015: implications for future policy on prevention and treatment. Liver Int. 2017;37(1):45-53.
- Elgharably A, Gomaa AI, Crossey MME, Norsworthy PJ, Waked I, Taylor-Robinson SD. Hepatitis C in Egypt – past, present, and future. International Journal of General Medicine. 2017;10:1-6.
- Fox-Rusby J, Cairns J. Economic Evaluation. Maidenhead: Open University Press; 2005.
- Woods B, Revill P, Sculpher M, Claxton K. Country-Level Cost-Effectiveness Thresholds: Initial Estimates and the Need for Further Research. Value Health. 2016;19(8):929-35.
- Benbassat J, Baumal R. The time horizons of formal decision analyses. QJM: An International Journal of Medicine. 2007;100(6):383-8.
- Elsisi GH, Kaló Z, Eldessouki R, Elmahdawy MD, Saad A, Ragab S, et al. Recommendations for Reporting Pharmacoeconomic Evaluations in Egypt. Value in Health Regional Issues. 2013;2(2):319-27.
- Claxton KP, Revill P, Sculpher M, Wilkinson T, Cairns J, Briggs A. The Gates Reference Case for Economic Evaluation: The Bill and Melinda Gates Foundation; 2014.
- WHO. Egypt: WHO statistical profile. January 2015.
- Drummond MF, Schulpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programs. 4th ed. Oxford: Oxford University Press; 2015.
- Robinson LA, Hammitt JK, Chang AY, Resch S. Understanding and improving the one and three times GDP per capita cost-effectiveness thresholds. Health Policy and Planning. 2017;32(1):141-5.
- Kim DD, Hutton DW, Raouf AA, Salama M, Hablas A, Seifeldin IA, et al. Cost-effectiveness model for hepatitis C screening and treatment: Implications for Egypt and other countries with high prevalence. Glob Public Health. 2015;10(3):296-317.
- Palumbo E. Pegylated Interferon and Ribavirin Treatment for Hepatitis C Virus Infection. Therapeutic Advances in Chronic Disease. 2011;2(1):39-45.
- Bank W. GDP per capita (current LCU) 2015 [Available from: https://data.worldbank.org/indicator/NY.GDP.PCAP.CN?end=2015&start=1960.
- XE. XE Currency Charts: USD to EGP 2015 [Available from: http://www.xe.com/currencycharts/?from=USD&to=EGP&view=5Y.
Cite This Work
To export a reference to this article please select a referencing stye below:
Related ServicesView all
Related ContentAll Tags
Content relating to: "Pharmacology"
Pharmacology involves the study of drugs and how they affect the body. A pharmacologist contributes to drug development by researching and testing how the body reacts to medication, and whether the medication can have a positive impact on the body in terms of fighting illness and disease.
Action of Corilagin on Hyperglycemia, Hyperlipidemia and Oxidative Damage in Streptozotocin-induced Diabetic Rats
ABSTRACT: Diabetes mellitus is a world’s largest endocrine disease involving metabolic disorders of carbohydrate, protein and fat. This study was undertaken to investigate the anti-diabetic activ...
Saliva Therapeutic Drug Monitoring
Introduction Therapeutic drug monitoring (TDM) is defined as: “measuring serum concentrations of a drug in a single or multiple time points in a biological matrix after a dosage” (1). TDM is usef...
DMCA / Removal Request
If you are the original writer of this dissertation and no longer wish to have your work published on the UKDiss.com website then please: