A Systematic Review of Kidney Transplantation Decision Modelling Studies.

BACKGROUND: Genome-based precision medicine strategies promise to minimize premature graft loss after renal transplantation, through precision approaches to immune compatibility matching between kidney donors and recipients. The potential adoption of this technology calls for important changes to clinical management processes and allocation policy. Such potential policy change decisions may be supported by decision models from health economics, comparative effectiveness research and operations management. OBJECTIVE: We used a systematic approach to identify and extract information about models published in the kidney transplantation literature and provide an overview of the status of our collective model-based knowledge about the kidney transplant process. METHODS: Database searches were conducted in MEDLINE, Embase, Web of Science and other sources, for reviews and primary studies. We reviewed all English-language papers that presented a model that could be a tool to support decision making in kidney transplantation. Data were extracted on the clinical context and modelling methods used. RESULTS: A total of 144 studies were included, most of which focused on a single component of the transplantation process, such as immunosuppressive therapy or donor-recipient matching and organ allocation policies. Pre- and post-transplant processes have rarely been modelled together. CONCLUSION: A whole-disease modelling approach is preferred to inform precision medicine policy, given its potential upstream implementation in the treatment pathway. This requires consideration of pre- and post-transplant natural history, risk factors for allograft dysfunction and failure, and other post-transplant outcomes. Our call is for greater collaboration across disciplines and whole-disease modelling approaches to more accurately simulate complex policy decisions about the integration of precision medicine tools in kidney transplantation.

Cost-effectiveness Analysis of Vascular Access Referral Policies in CKD.

BACKGROUND: The optimal timing of vascular access referral for patients with chronic kidney disease who may need hemodialysis (HD) is a pressing question in nephrology. Current referral policies have not been rigorously compared with respect to costs and benefits and do not consider patient-specific factors such as age. STUDY DESIGN: Monte Carlo simulation model. SETTING & POPULATION: Patients with chronic kidney disease, referred to a multidisciplinary kidney clinic in a universal health care system. MODEL, PERSPECTIVE, & TIMEFRAME: Cost-effectiveness analysis, payer perspective, lifetime horizon. INTERVENTION: The following vascular access referral policies are considered: central venous catheter (CVC) only, arteriovenous fistula (AVF) or graft (AVG) referral upon HD initiation, AVF (or AVG) referral when HD is forecast to begin within 12 (or 3 for AVG) months, AVF (or AVG) referral when estimated glomerular filtration rate is <15 (or <10 for AVG) mL/min/1.73m2. OUTCOMES: Incremental cost-effectiveness ratios (ICERs, in 2014 US dollars per quality-adjusted life-year [QALY] gained). RESULTS: The ICER of AVF (AVG) referral within 12 (3) months of forecasted HD initiation, compared to using only a CVC, is ∼$105k/QALY ($101k/QALY) at a population level (HD costs included). Pre-HD AVF or AVG referral dominates delaying referral until HD initiation. The ICER of pre-HD referral increases with patient age. Results are most sensitive to erythropoietin costs, ongoing HD costs, and patients' utilities for HD. When ongoing HD costs are excluded from the analysis, pre-HD AVF dominates both pre-HD AVG and CVC-only policies. LIMITATIONS: Literature-based estimates for HD, AVF, and AVG utilities are limited. CONCLUSIONS: The cost-effectiveness of vascular access referral is largely driven by the annual costs of HD, erythropoietin costs, and access-specific utilities. Further research is needed in the field of dialysis-related quality of life to inform decision making regarding vascular access referral.

Forecasting the effect of physician assistants in a pediatric ED.

BACKGROUND: Most pediatric ED visits are for nonemergent problems. Physician assistants are well trained to manage these patients; however, their effect on patient flow in a pediatric ED is unknown. OBJECTIVES: To compare the effect on key pediatric ED efficiency indicators of extending physician coverage versus adding PAs with equivalent incremental costs. METHODS: We used discrete event simulation modeling to compare the effect of additional physician coverage versus adding PAs on wait time, length of stay (LOS), and patients leaving without being seen. RESULTS: Simulation of extended physician coverage reduced wait times, LOS, and rates of leaving without being seen across acuity levels. Adding PAs reduced wait times and LOS for high-acuity visits, and slightly increased the LOS for low-acuity visits. CONCLUSIONS: With restricted autonomy, PAs mainly benefitted the high-acuity patients. Increasing the level of PA autonomy was critical in broadening the effect of PAs to all acuity levels.

Timing of arteriovenous fistula creation in patients With CKD: a decision analysis.

BACKGROUND: The optimal time for arteriovenous fistula (AVF) referral is uncertain. Improving the timeliness of referral may reduce central venous catheter (CVC) use. STUDY DESIGN: Monte Carlo simulation model. SETTING & POPULATION: Patients with chronic kidney disease (CKD) followed up in a multidisciplinary clinic, overall and stratified by age. MODEL, PERSPECTIVE, & TIMEFRAME: Decision analysis, patient, patient's lifetime. INTERVENTION: AVF referral, using 1 of 2 strategies: refer when hemodialysis is anticipated to begin within a certain time frame or refer when estimated glomerular filtration rate (eGFR) drops below a certain threshold. OUTCOMES: A range of values for each strategy are compared to each other with respect to incident vascular access type (AVF or CVC), percentage of patients with an unnecessary AVF creation, and life expectancy after dialysis therapy initiation. RESULTS: A 15-month referral time frame gave 34% with incident CVCs, 14% with unnecessary AVFs, and a life expectancy of 1,751 days. Time frames of 12-18 months performed similarly. Referral at eGFR of 20 mL/min/1.73 m(2) gave 38% with incident CVCs, 20% with unnecessary AVFs, and life expectancy of 1,742 days. Using an eGFR threshold of 15 mL/min/1.73 m(2), 10% had an unnecessary AVF. Policy performance was affected by CKD progression rate and age. For fast progressors (ΔeGFR = -7mL/min/1.73 m(2) per year), referral at eGFR of 25 mL/min/1.73 m(2) achieved a similar incident CVC percentage (~40%) as referral at 15 mL/min/1.73 m(2) in slower progressors (ΔeGFR = -2.78 mL/min/1.73 m(2) per year). For patients aged 70-80 and 80-90 years, time frames of 15-18 months yielded 16%-22% with unnecessary AVFs (vs 9%-11% in 50- to 60-year-olds); an eGFR threshold strategy of 20 mL/min/1.73 m(2) yielded 24% unnecessary AVFs in 80- to 90-year-olds versus 16% in 50- to 60-year-olds. LIMITATIONS: Our model does not consider patients with nonlinear CKD progression or acute kidney injury. We did not include arteriovenous grafts or consider cost or quality of life. CONCLUSIONS: In general, AVF referral within about 12 months of the estimated time to dialysis performed best among time frame strategies, and referral at eGFR < 15-20 mL/min/1.73 m(2) performed best among threshold strategies. The timing of referral should also be guided by the individual rate of CKD progression. Elderly patients with CKD could be referred later to reduce the risk of creating an AVF that is never used.

Using operations research to plan improvement of the transport of critically ill patients.

OBJECTIVE: Operations research is the application of mathematical modeling, statistical analysis, and mathematical optimization to understand and improve processes in organizations. The objective of this study was to illustrate how the methods of operations research can be used to identify opportunities to reduce the absolute value and variability of interfacility transport intervals for critically ill patients. METHODS: After linking data from two patient transport organizations in British Columbia, Canada, for all critical care transports during the calendar year 2006, the steps for transfer of critically ill patients were tabulated into a series of time intervals. Statistical modeling, root-cause analysis, Monte Carlo simulation, and sensitivity analysis were used to test the effect of changes in component intervals on overall duration and variation of transport times. Based on quality improvement principles, we focused on reducing the 75th percentile and standard deviation of these intervals. RESULTS: We analyzed a total of 3808 ground and air transports. Constraining time spent by transport personnel at sending and receiving hospitals was projected to reduce the total time taken by 33 minutes with as much as a 20% reduction in standard deviation of these transport intervals in 75% of ground transfers. Enforcing a policy of requiring acceptance of patients who have life- or limb-threatening conditions or organ failure was projected to reduce the standard deviation of air transport time by 63 minutes and the standard deviation of ground transport time by 68 minutes. CONCLUSIONS: Based on findings from our analyses, we developed recommendations for technology renovation, personnel training, system improvement, and policy enforcement. Use of the tools of operations research identifies opportunities for improvement in a complex system of critical care transport.

State-space size considerations for disease-progression models.

Markov models of disease progression are widely used to model transitions in patients' health state over time. Usually, patients' health status may be classified according to a set of ordered health states. Modelers lump together similar health states into a finite and usually small, number of health states that form the basis of a Markov chain disease-progression model. This increases the number of observations used to estimate each parameter in the transition probability matrix. However, lumping together observably distinct health states also obscures distinctions among them and may reduce the predictive power of the model. Moreover, as we demonstrate, precision in estimating the model parameters does not necessarily improve as the number of states in the model declines. This paper explores the tradeoff between lumping error introduced by grouping distinct health states and sampling error that arises when there are insufficient patient data to precisely estimate the transition probability matrix.

Treatment evolution and new standards of care: implications for cost-effectiveness analysis.

BACKGROUND: Traditional approaches to cost-effectiveness analysis have not considered the downstream possibility of a new standard of care coming out of the research and development pipeline. However, the treatment landscape for patients may change significantly over the course of their lifetimes. OBJECTIVE: To present a Markov modeling framework that incorporates the possibility of treatment evolution into the incremental cost-effectiveness ratio (ICER) that compares treatments available at the present time. DESIGN: . Markov model evaluated by matrix algebra. Measurements. The author evaluates the difference between the new and traditional ICER calculations for patients with chronic diseases facing a lifetime of treatment. RESULTS: The bias of the traditional ICER calculation may be substantial, with further testing revealing that it may be either positive or negative depending on the model parameters. The author also performs probabilistic sensitivity analyses with respect to the possible timing of a new treatment discovery and notes the increase in the magnitude of the bias when the new treatment is likely to appear sooner rather than later. Limitations. The modeling framework is intended as a proof of concept and therefore makes simplifying assumptions such as time stationarity of model parameters and consideration of a single new drug discovery. CONCLUSIONS: For diseases with a more active research and development pipeline, the possibility of a new treatment paradigm may be at least as important to consider in sensitivity analysis as other parameters that are often considered.

The use of discrete-event simulation modelling to improve radiation therapy planning processes.

BACKGROUND AND PURPOSE: The planning portion of the radiation therapy treatment process at the British Columbia Cancer Agency is efficient but nevertheless contains room for improvement. The purpose of this study is to show how a discrete-event simulation (DES) model can be used to represent this complex process and to suggest improvements that may reduce the planning time and ultimately reduce overall waiting times. MATERIALS AND METHODS: A simulation model of the radiation therapy (RT) planning process was constructed using the Arena simulation software, representing the complexities of the system. Several types of inputs feed into the model; these inputs come from historical data, a staff survey, and interviews with planners. RESULTS: The simulation model was validated against historical data and then used to test various scenarios to identify and quantify potential improvements to the RT planning process. CONCLUSIONS: Simulation modelling is an attractive tool for describing complex systems, and can be used to identify improvements to the processes involved. It is possible to use this technique in the area of radiation therapy planning with the intent of reducing process times and subsequent delays for patient treatment. In this particular system, reducing the variability and length of oncologist-related delays contributes most to improving the planning time.

Influence of alternative thresholds for initiating HIV treatment on quality-adjusted life expectancy: a decision model.

BACKGROUND: The optimal threshold for initiating HIV treatment is unclear. OBJECTIVE: To compare different thresholds for initiating HIV treatment. DESIGN: A validated computer simulation was used to weigh important harms from earlier initiation of antiretroviral therapy (toxicity, side effects, and resistance accumulation) against important benefits (decreased HIV-related mortality). DATA SOURCES: Veterans Aging Cohort Study (5742 HIV-infected patients and 11 484 matched uninfected controls) and published reports. TARGET POPULATION: Individuals with newly diagnosed chronic HIV infection and varying viral loads (10,000, 30,000, 100,000, and 300,000 copies/mL) and ages (30, 40, and 50 years). TIME HORIZON: Unlimited. PERSPECTIVE: Societal. INTERVENTION: Alternative thresholds for initiating antiretroviral therapy (CD4 counts of 200, 350, and 500 cells/mm3). OUTCOME MEASURES: Life-years and quality-adjusted life-years (QALYs). RESULTS OF BASE-CASE ANALYSIS: Although the simulation was biased against earlier treatment initiation because it used an upper-bound assumption for therapy-related toxicity, earlier treatment increased life expectancy and QALYs at age 30 years regardless of viral load (life expectancies with CD4 initiation thresholds of 500, 350, and 200 cells/mm3 were 18.2 years, 17.6 years, and 17.2 years, respectively, for a viral load of 10,000 copies/mL and 17.3 years, 15.9 years, and 14.5 years, respectively, for a viral load of 300,000 copies/mL), and increased life expectancies at age 40 years if viral loads were greater than 30 000 copies/mL (life expectancies were 12.5 years, 12.0 years, and 11.4 years, respectively, for a viral load of 300,000 copies/mL). RESULTS OF SENSITIVITY ANALYSIS: Findings favoring early treatment were generally robust. LIMITATIONS: Results favoring later treatment may not be valid. The findings may not be generalizable to women. CONCLUSION: This simulation suggests that earlier initiation of combination antiretroviral therapy is often favored compared with current recommendations.

Estimating the rate of accumulating drug resistance mutations in the HIV genome.

OBJECTIVE: HIV mutation accumulation has great implications for pharmacoeconomics and clinical care, yet scarcity of data has hindered its representation in decision analytic models. Our objective is to determine the accuracy with which mutation accumulation and other unmeasured parameters could be estimated during model calibration. METHODS: We used a second-order Monte Carlo simulation of HIV natural history that had been calibrated by varying two unmeasured parameters (mutation accrual rate and probability of adherence) to minimize differences between estimated and observed clinical outcomes (time to treatment failure and survival). We compared these estimated values first with only those results that had been already published at the time of model calibration, and second including results that were published after model calibration. RESULTS: The value for mutation accrual rate assigned during calibration was 0.014 mutations per month for antiretroviral-naïve patients, at the lower bound of the results for nine heterogeneous studies published at the time of calibration (pooled 95% confidence interval [CI] 0.014-0.039 mutations per month). In contrast, this estimate accurately anticipated results from 11 larger and more homogeneous studies published after calibration (pooled 95% CI for antiretroviral-naïve patients, 0.012-0.015 mutations per month). The value for probability of adherence assigned during calibration (75%) was also within the range of published results (pooled 95% CI 62-76%). CONCLUSION: Estimates for unobserved parameters derived during model calibration were not only within the range of clinical observations, but anticipated with accuracy clinical results that were not yet available. It may be feasible to use models to estimate unobserved parameters.

Increasing the efficiency of Monte Carlo cohort simulations with variance reduction techniques.

The authors discuss techniques for Monte Carlo (MC) cohort simulations that reduce the number of simulation replications required to achieve a given degree of precision for various output measures. Known as variance reduction techniques, they are often used in industrial engineering and operations research models, but they are seldom used in medical models. However, most MC cohort simulations are well suited to the implementation of these techniques. The authors discuss the cost of implementation versus the benefit of reduced replications.

Incorporating biological natural history in simulation models: empirical estimates of the progression of end-stage liver disease.

OBJECTIVE: To develop an empiric natural-history model that can predict quantitative changes in the laboratory values and clinical characteristics of patients with end-stage liver disease (ESLD), to be used to calibrate an individual microsimulation model. METHODS: The authors report the development of a stochastic model that uses cubic splines to interpolate between observed laboratory values over time in a cohort of 1997 patients with ESLD awaiting liver transplantation at the University of Pittsburgh Medical Center. The splines were recursively sampled to provide a stochastic, quantitative natural history of each candidate's disease. RESULTS: The model was able to simulate the types of erratic disease trajectories that occur in individual patients and was able to preserve the statistical properties of the natural history of ESLD in cohorts of real patients. Moreover, the model was able to predict pretransplant survival rates (87% at 1 year), which were statistically similar to rates observed in the authors' local cohort (92%). CONCLUSIONS: Cubic splines can be used to generate quantitative natural histories for individual patients with ESLD and may be useful for developing clinically robust microsimulation models of other diseases.

A clinically based discrete-event simulation of end-stage liver disease and the organ allocation process.

BACKGROUND: The optimal allocation of scarce donor livers is a contentious health care issue requiring careful analysis. The objective of this article was to design a biologically based discrete-event simulation to test proposed changes in allocation policies. METHODS: The authors used data from multiple sources to simulate end-stage liver disease and the complex allocation system. To validate the model, they compared simulation output with historical data. RESULTS: Simulation outcomes were within 1% to 2% of actual results for measures such as new candidates, donated livers, and transplants by year. The model overestimated the yearly size of the waiting list by 5% in the last year of the simulation and the total number of pretransplant deaths by 10%. CONCLUSION: The authors created a discrete-event simulation model that represents the biology of end-stage liver disease and the health care organization of transplantation in the United States.

A methodological framework for optimally reorganizing liver transplant regions.

BACKGROUND: The United States is divided currently into 11 transplant regions, which vary in area and number of organ procurement organizations (OPOs). Region size affects organ travel time and organ viability at transplant. PURPOSE: To develop a methodologic framework for determining optimal configurations of regions maximizing transplant allocation efficiency and geographic parity. METHODS: An integer program was designed to maximize a weighted combination of 2 objectives: 1) intraregional transplants, 2) geographic parity-maximizing the lowest intraregional transplant rate across all OPOs. Two classes of functions relating liver travel time to liver viability were also examined as part of the sensitivity analyses. RESULTS: Preliminary results indicate that reorganizing regions, while constraining their number to 11, resulted in up to 17 additional transplants/year depending on the travel-viability function; when not constrained, it resulted in up to 18/year of increase. CONCLUSION: Our analysis indicates that liver transplantation may benefit through region reorganization. The analytic method developed here should be applicable to other organs and sets of organs.