hesim (version 0.2.3)

Psm: N-state partitioned survival model

Description

Simulate outcomes from an N-state partitioned survival model.

Usage

Psm

Format

R6Class object.

Fields

survival_models

The survival models used to predict survival curves. Must be an object of class PsmCurves.

utility_model

The model used to predict utility by health state. Must be an object of class StateVals.

cost_models

The models used to predict costs by health state. Must be a list of objects of class StateVals, where each element of the list represents a different cost category.

n_states

Number of states in the partitioned survival model.

t_

A numeric vector of times at which survival curves were predicted. Determined by the argument t in sim_curves.

survival_

Survival curves generated using sim_curves.

stateprobs_

Health state probabilities as a function of time generated using sim_stateprobs.

costs_

Total (discounted) costs by type generated using sim_costs.

qalys_

Total (discounted) quality-adjusted life-years (QALYs) generated using sim_qalys.

Methods

new(survival_models, utility_model = NULL, cost_models = NULL)

Constructor for the class. Note that the number of health states, n_states, is set equal to the number of survival models plus one.

sim_survival(t)

Simulate survival curves as a function of time. Equivalent to the member function survival in PsmCurves.

• t: A numeric vector of times. The first element must be 0.

sim_stateprobs()

Simulate the probability of being in each of N health states using the survival curves generated from sim_curves.

sim_qalys(dr = .03)

Simulate (discounted) QALYs over the times selected in t associated with each health state based on the state probabilities calculated using sim_stateprobs. See "Details".

• dr: Discount rate to apply to QALYs. May be a vector in which case QALYs are calculated for each element in dr.

sim_costs(dr = .03)

Simulate (discounted) costs for each cost type over the times selected in t associated with each health state based on the state probabilities calculated using sim_stateprobs. See "Details".

• dr: Discount rate to apply to costs. May be a vector in which case costs for each cost type are calculated for each element in dr.

,

summarize()

Summarize costs and QALYs so that cost-effectiveness analysis can be performed. See summarize_ce.

check()

Input validation for class. Checks that fields are the correct type.

Details

Discounted costs and QALYs are calculated by integrating the "weighted" probability of being in each state. Weights are a function of the discount factor and the state value predicted using either the cost or QALY model. Mathematically, discounted costs and QALYs in health state \(s\) are calculated as,

$$\int_0^T w_h e^{-rt} P_h(t) dt $$,

where for health state \(h\) and time t, \(w_h\) is the predicted cost or QALY weight, \(r\) is the discount rate, and \(P_h(t)\) is the probability of being in a given health state. The integral is calculated numerically using the composite trapezoid rule from the points in t_.

Examples

# NOT RUN {
library("flexsurv")

# Simulation data
strategies <- data.frame(strategy_id = c(1, 2, 3))
patients <- data.frame(patient_id = seq(1, 3),
                          age = c(45, 50, 60),
                          female = c(0, 0, 1))
states <- data.frame(state_id =  seq(1, 3),
                        state_name = paste0("state", seq(1, 3)))
hesim_dat <- hesim_data(strategies = strategies,
                        patients = patients,
                        states = states)
n_samples <- 3

# Survival models
surv_est_data <- psm4_exdata$survival
fit1 <- flexsurv::flexsurvreg(Surv(endpoint1_time, endpoint1_status) ~ age,
                              data = surv_est_data, dist = "exp")
fit2 <- flexsurv::flexsurvreg(Surv(endpoint2_time, endpoint2_status) ~ age,
                              data = surv_est_data, dist = "exp")
fit3 <- flexsurv::flexsurvreg(Surv(endpoint3_time, endpoint3_status) ~ age,
                              data = surv_est_data, dist = "exp")
fits <- flexsurvreg_list(fit1, fit2, fit3)

surv_input_data <- expand(hesim_dat, by = c("strategies", "patients"))
psm_curves <- create_PsmCurves(fits, input_data = surv_input_data,
                               bootstrap = TRUE, est_data = surv_est_data, 
                               n = n_samples)

# Cost model(s)
cost_input_data <- expand(hesim_dat, by = c("strategies", "patients", "states"))
fit_costs_medical <- stats::lm(costs ~ female + state_name, 
                               data = psm4_exdata$costs$medical)
psm_costs_medical <- create_StateVals(fit_costs_medical, 
                                      input_data = cost_input_data, 
                                      n = n_samples)

# Utility model
utility_tbl <- stateval_tbl(tbl = data.frame(state_id = states$state_id,
                                             min = psm4_exdata$utility$lower,
                                             max = psm4_exdata$utility$upper),
                            dist = "unif",
                            hesim_data = hesim_dat)
psm_utility <- create_StateVals(utility_tbl, n = n_samples)

# Partitioned survival decision model
psm <- Psm$new(survival_models = psm_curves,
               utility_model = psm_utility,
               cost_models = list(medical = psm_costs_medical))
psm$sim_survival(t = seq(0, 5, .05))
psm$sim_stateprobs()
psm$sim_costs(dr = .03)
head(psm$costs_)
head(psm$sim_qalys(dr = .03)$qalys_)
# }