The link/graph has a paywall. But I can engage with the key question. Yours is a very common misconception.
The quoted statistic says that 60% of all women who get pregnant give birth before the 39th week. That statistic probably includes a lot of pregnancies that deliver well before that, too (before 36 weeks, even). So, by making it to 36 weeks, your chances of making it to/beyond 39 weeks could be higher than that of the whole population of pregnant women. This is a conditional probability.
Let's take an extreme example that considers similar ideas. Imagine you're in a racing car, racing against other drivers; there are 20 of you in total. The aim of the race is to drive as far as you can (time doesn't matter). You're told just before the start that 16 (80%) of the cars only have enough gas in the tank to make it one mile. 3 of the cars have enough gas to make it 10 miles. And the final car has enough gas to make it 20 miles.
At the start, you'd be right to say that you have a 5% chance of making it to 12 miles. But if we asked you later in the race and found that you had already driven 5 miles absolutely fine... I'd say your chances of making it 12 miles is considerably higher than our first guess. At this point, we know for sure that you have more gas than just 1 mile's worth. So either you're one of the 3 that can only make it 10 miles OR you're the who can make 20 miles. So now there's a 25% chance you'll make it to 12 miles (or more). If I asked again when you had driven 11 miles, I'd say you had a 100% chance of making it to 12 miles.
Do you see the connection?