Don't quite follow the calculations. If 5 weeks of Apiguard produced 100 per day, that would be 3,500 mites. Following with Apivar saw 500 plus 16 days of 200, that's 3,700 and small numbers left. If there was an initial higher drop that you have not included, then 60% but the figures you quote run under 50%.
The manufacturers of Apiguard themselves claim 93% average on
http://www.vita-europe.com at /products/Apiguard/#Apiguardaprovensuccess charting 88-97% in different trials.
Other studies have found lower success rates, a quick google trawl shows 56% here on
http://www.apimondiafoundation.org at /foundation/files/141.pdf an average which is much closer to MM's experience with a single hive.
To return to ksjs's request "if anybody does know whether mite count during treatment can be used as any sort of a population indicator I'd be interested to hear":
There is evidence that a full Apiguard treatment within the manufacturers guidelines could drop anywhere from below 50% to 97% of the Varroa. The practical reading of this asking around my local BKA is that you have to treat for 4 weeks at least to cover all the mites lurking in sealed brood. If the drop has not tailed off by the fourth week, you might still have as many mites again remaining in the hive.
If the drop was very low anyway it's not a problem but if you have thousands over four weeks and the drop has not tailed off then your bees still have a mite problem. Leaving the Apiguard remains in for a couple of weeks longer might help, but it's not a complete remedy. You do have to consider further and varied treatment such as alternative treatments now or repeated oxalic treatments as soon as they are broodless.
That's pretty much as I'd see it until a better strategy comes along.