There are several problems with the "stats" cited by Stossel, in addition to the good point you raised. Each of these bullet items suggests reasons for believing the "88%" figure does not apply to typical addict situations.

1) Was this a "study" or instead a "monitoring", more like a "medical parole". Stossel doesn't say, but his use of the term "track" instead of "studied" implies the latter. If so, then monitoring would reduce, possibly significantly, the incidence of relapse. The vets would be receiving periodic contact from an authority figure who may also provide support.

2) What happened after the 3 years were up? Did the "tracking" action merely delay the incidence of relapse? I have no stats, but relatives/friends who have spent years in the VA system tell me that relapse has been very common.

3) How do the demographics of heroin-addicted Vietnam vets compare to modern heroin addicts? It's highly likely that the average vet had a far better family/community support system than the average addict today.

4) Related to (1), there is also the Heisenburg Uncertainty Principle. In a famous study about 80 years ago, researchers found that employees became more productive in all control groups (increased lighting in work area, same, decreased lighting). What they found was that the mere act of being studied (watched) changed the subjects behavior. What steps were taken in the Vietnam vet "tracking" to make sure this didn't happen? Most likely none, since this probably wasn't a "study", and the goals of the monitors were probably to prevent relapses.

In sum, this was a typical Stossel quote. Like Jeff Jacoby, he carefully selects statistics taken completely out of context to support his point. Makes you wonder just how bad their positions really are that they must resort to these tactics in order to convince people.