Judging by the questions that soccertese quite rightly asks, I think that I gave too little explanation of what was going on in Scenario 3. I'm assuming that:
- the person has taken 100mg doses of levodopa at 0700 and 1000 and these have worked;
- at 1300 the person goes into an expected "off" as the previous dose (1000) wears off as normal;
- at 1300 the person unknowingly eats a substantial amount of protein, which delays for 1 hour any passing of levodopa into the brain, and ultimately reduces the bioavailability by 30%;
- this is modelled by replacing the 1300 dose with one at 1400, with only 70% of the dose;
- the patient will have to make a decision as to what to do as he doesn't feel the 1300 dose kicking in, I'm assuming that he will give this one hour's grace before writing it off as a lost dose, he then takes the rescue dose;
- this is modelled as an 10mg apomorphine dose at 1400.

It is important not to take Scenario 3 as the only possible case. There are many other scenarios to explore: what if the rescue dose is taken at 1330?; what if the 1300 dose is completely lost?

It is of the nature of mathematical models that they only show part of reality, but not all of it. What I've developed here is a pharmacokinetic (what the body does to the drug) model, when what we are really interested in are the pharmacodynamics (what the drug does to the body) of the situation.

soccertese writes:

'isn't it just slightly amazing that after 40 or more years since c/l came on the market we are still discussing such basic problems? there should have been a "cure" by now.'

I completely agree. I'm now 13 years post diagnosis, and so I'm unlikely to benefit from any new drugs. But what I can do is to get more out of existing drugs. I think that mathematical models have a part to play in that process.

John