Comparison with other packages

Comparison with heemod

The heemod package is designed to simulate more general models but let us perform the same analysis to compare. First, we need to define the transition matrix. This is essentially the same as above except because heemod simulates over a predefined number of iterations or cycles we need to add a final absorbing sink state (i.e. \(p_{ii} = 1\)) with no associated costs.

library(heemod)

mat_base <- define_transition(
  state_names = as.character(1:8),
  0, 0.2, 0.8,0,0,0,0,0, 
  0,0,0, 0.2, 0.8,0,0,0,  
  0,0,0,0,0, 0.2, 0.8,0,
  0,0,0,0,0,0,0,1,
  0,0,0,0,0,0,0,1,
  0,0,0,0,0,0,0,1,
  0,0,0,0,0,0,0,1,
  0,0,0,0,0,0,0,1
)

Define cost at each node:

state_1 <- define_state(
  cost_total = 0,
  qaly = 0)

state_2 <- define_state(
  cost_total = 10,
  qaly = 1)

state_3 <- define_state(
  cost_total = 1,
  qaly = 1)

state_4 <- define_state(
  cost_total = 10,
  qaly = 1)

state_5 <- define_state(
  cost_total = 1,
  qaly = 1)

state_6 <- define_state(
  cost_total = 10,
  qaly = 1)

state_7 <- define_state(
  cost_total = 1,
  qaly = 1)

state_8 <- define_state(
  cost_total = 0,
  qaly = 0)

strat_base <- define_strategy(
  transition = mat_base,
  "1" = state_1,
  "2" = state_2,
  "3" = state_3,
  "4" = state_4,
  "5" = state_5,
  "6" = state_6,
  "7" = state_7,
  "8" = state_8
)

Finally, we get the expected cost £5600 for 100 cycles which is equivalent to £5.6 in our previous analysis.

run_model(
  strat_base,
  cycles = 100,
  cost = cost_total,
  effect = qaly)
#> No named model -> generating names.
#> 1 strategy run for 100 cycles.
#> 
#> Initial state counts:
#> 
#> 1 = 1000L
#> 2 = 0L
#> 3 = 0L
#> 4 = 0L
#> 5 = 0L
#> 6 = 0L
#> 7 = 0L
#> 8 = 0L
#> 
#> Counting method: 'life-table'.
#> 
#>  
#> 
#> Counting method: 'beginning'.
#> 
#>  
#> 
#> Counting method: 'end'.
#> 
#> Values:
#> 
#>   cost_total qaly
#> I       5600 2000

Nathan Green

2022-08-17

Comparison with heemod