I think you do not quite understand the statistical properties of the evolutionary mechanisms. The ravaging sickle cell disease you describe is the form patients develop if they carry two copies of the gene, which happens to one in four children of pairs where both parents carry one copy (it is not relevant to think about pairs where at least one parent carries two copies, these individuals usually do not reach reproductive age and are commonly infertile if they do).

Assume that 10% of a population carry one copy of the gene, and a carrier has a child with a random partner.

Then 10% of the children will be from partners also carrying the gene, with 2.5% inheriting no copy, 5% inheriting one copy, and 2.5% inheriting two. The other 90% will will be with partners without the gene, with 45% inheriting one copy, and 45% no copy. Absent other illnesses and medical care, 97.5% of the children will survive.

For a parent without the gene, 5% of the children will inherit a copy of the gene from the other partner, and 95% will inherit none. Absent other illnesses, 100% will survive and a person without the gene will on average produce slightly more viable offspring, slowly driving the gene out of the population.

I don't have the actual numbers, but assume that malaria kills 10% of the children without the gene, but only 5% of those with one copy.

Now the expected percentage of viable offspring for a carrier is 0.9 * (2.5 + 45) + 0.95 * (5 + 45) = 90.25, as it is for a non-carrier, since 0.9 * 95 + 0.95 * 5 = 90.25, too.

So with these hypothetical numbers, the partial malaria resistance does cancel the negative effects of the sickle cell gene and the 10% prevalence of the gene is stable. If the prevalence goes down, the gene starts to confer a benefit to the carrier (as the chance of meeting a partner that is also a carrier goes down, and fewer of the children of non-carriers will be carriers), and if it rises, it becomes a liability.

Other numbers will move the equilibrium point, but as long as the gene confers even the slightest advantage, it is expected to stay in the population at a rate that makes it sufficiently improbable for two carriers to meet and have children. (If the rate is low enough it may still be eliminated by pure chance.)

But aren't you describing a statistical side effect? It's not as though this mutation imparts a 100% chance of having sickle cell anemia. If 87.5% (carriers and non carriers) survive significantly longer due to malaria resistance and 12.5% lack that extended lifespan due to anemia, isn't it still an advantage overall?