The Catholic Church doesn't take you off the roles even after you've left, unless you do something drastic like deny your own baptism or die, making their growth statistics suspect at best.
"Roman Catholics, the largest U.S. church with a reported 69 million members, start counting baptized infants as members and often don’t remove people until they die. Most membership surveys don’t actually count who’s in the pews on Sunday. To be disenrolled, Catholics must write a bishop to ask that their baptisms be revoked..."
....it is possible, for example, to be born Catholic, married Methodist, die Lutheran and still be listed as a member of the 1 billion-member Roman Catholic Church....
"...The Catholic understanding of membership is that a person becomes a member upon baptism and remains a member for life," Gautier said. "Whether you show up at church or not is not what determines whether you're a member."
-- from the thread When It Comes to Church Membership Numbers, the Devil's in the Details
Catholics are leaving the faith at four times the rate that newcomers are joining. "Religious change is not simply a function of retention; it's a function of recruitment. It's both sides of the ledger," explains the Pew Forum on Religion and Public Life's Greg Smith. "In no other religious groups we looked at did we see this high a ratio people leaving versus joining."See also Roman Catholics total 64 million in U.S. ["counting Catholics is really more art than science"]
.... from the thread Does the American Catholic Church Have a Numbers Problem?
The methodology of a survey like this is tricky in every case; it's not a "Catholic thing."
It's not a case of "not taking you off the roles [I think you mean "rolls"] even after you left," it's a case of "how do you know when they've left, and what is 'left,' anyway?".
(They've left if they go to the Assemblies of God fulltime, but have they left if they come to Mass on Christmas and Easter? What if they never come to Mass but still call themselves "Catholic"?)
Some Catholic parishes & dioceses take "parish censuses" regularly, specifically polling people in the pews on Sunday. They would have an accurate idea of how many people are regularly attending members. But others don't do that. A parish could also count contributing members (those who use envelopes or checks), but that wouldn't count the noncontributors who still attend.
One number that is accurate is the number of new members added in a year, because that's just the number of infant baptisms plus the number of adult conversions. So it's safe to say that the percentage increase described in this article is correct or on the low side. What's harder to know is what the baseline number is against which that percentage increase is computed.