I’ll give you an imaginative but not so far-fledged example.
Divorced father of two from the US, visiting his Israeli kids and ex. He gets the kids to stay over but also gets COVID-19. For two days, he can’t breathe. He takes Paracetamol and feels a bit better.
He sent his kids back to his ex. Within a day, she loses her sense of smell and taste. The next day, she has terrible chest pain but that subsides.
The kids were a bit cranky. They got sick and transported the virus to their mother.
Within a couple of days, they’re all fine again. And probably immune.
They never called a doctor. The father was afraid to be blamed. Also, the mother doesn’t want to be accused in the future in a court case that she was sick with the kids around.
So these are people who know that they went through it but it was too short and too mild to make an issue of it. (And in my example, also felt that they had a reason to be secretive about it.) These stories come on top of people who really had no symptoms at all.
We will only know how many people got infected and immune already once we have an anti-body test (to see if someone recently has developed immunity for this coronavirus) that is at random and mandatory done at a larger scale. There barely exists such a test, let alone that people could be obligated to cooperate being tested for the statistics.
On the one hand, if many of the infected hardly or not at all feel sick, its deadliness could be less than feared. On the other hand, if many die from the virus at home (UK, US) and are not counted, it’s worse than assumed.
From dictatorships (most countries) no numbers can be trusted.
For sure, when the poorer countries’ numbers will come in, the pandemic’s deadliness will dramatically rise above the present 6.8% of the infected. (Which in itself is probably half the real percentage because the mortality rate still relates to the much lower infection rate of days to weeks before.)
The above is a separate issue from the tests for who is sick and infectious at the moment. They are needed to prevent outbreaks from spreading. And that testing is already possible but has its own logistic problems.
There are ongoing worries that getting over the infection might not be for good. That the virus might still be hiding and can reappear. Early worries that this virus likely very easily can mutate seem confirmed now. Add to this the newest finding that the virus may attack the immune system.
So, at least for now, we are still relatively in the dark. But physicians don’t raise their hands in despair when they don’t know something. These are simply some of the many unknown variables that they work with.