Bayesian Methods in Health Economics

Health economic evaluation in practice 203

Notice that in this case it is straightforward to check that

S



s



=1

λ

(t)

1s



= λ

(t)

11

+ λ

(t)

12

+ λ

(t)

13

=[(1− π

tox

t

)(1 − β

tr

t

)] + [(1 − π

tox

t

)β

tr

t

]+[π

tox

t

]

=1,

i.e. the transitions out of the state Stable sum to 1. In fact, because this

condition needs to always hold, it would be sufficient to compute all but one

of the required transition probabilities and derive the remaining one setting

it at 1 minus the sum of all the others.

Using a similar reasoning, consider a patient who is in the state Response

and does not experience toxicity. If they have a positive response, which hap-

pens with probability π

res

t

, then they will remain in that state and thus we

can set λ

(t)

22

=(1−π

tox

t

)π

res

t

. On the other hand, there ar