(individual) Should the causalities already incurred in a war be taken into account by a government in deciding whether it is in the national interest to continue the war? This is obviously not a trivial question. And it is a much more difficult question than you might at first suppose, especially for a government depending on popular support.
Your real consideration is will winning be worth the future cost of the casualties. However, future costs may include past casualties. To explain, consider public opinion which will force an administration out of office if casualties climb above 1000. If 800 casualties have already occurred, the marginal cost of 200 casualties is much higher than the first 800 casualties. Strictly speaking, the administration is only considering the cost of the next 200 casualties. But they have to know and consider that there were already 800 casualties in order to calculate the marginal cost of 200 casualties.
Also, one other consideration that includes past casualties is something I learned in estimating project costs. If your previous estimates were off by a certain amount, you don't ever get that back. I estimate a project consisting of two parts will take 2 months, 1 month on each part. If part 1 takes 2 months, it's likely that part 2 will be off by a similar amount. In war, you don't have any clear knowledge of future casualties. You have an estimate. If your estimate to pacify half of a country is that it will take 500 casualties, but instead you lose 800. It's likely the original estimate for pacifying the other half of the country is underestimated as well.