Traditionally, it’s been stated (and believed) that there is a U-shaped relationship between mortality and cholesterol level, with both very high and very low levels of cholesterol associated with increased death rates.
We (presumably) understand why high cholesterol levels are associated with higher rates of death, i.e. high cholesterol levels are associated with atherosclerosis and thus with things such as heart attacks and strokes.
Very low cholesterol levels have for many years also been observed to be associated with higher death rates (as has, for example, very low body weight). The explanation for these observations has generally been attributed to the fact that people with cancer and other serious illnesses are often malnourished and that leads to lowering of the cholesterol levels (and body weight). Obviously, the higher death rates would seem to be a result of the underlying serious illness and not the low cholesterol. As mentioned, the same relationship has been observed for low body weight and mortality. With respect to the explanation of that relationship, the notion of underlying serious illness has been put forth just as it has for low cholesterol and mortality. However, with respect to low body weight and higher death rates, there is probably also a contribution (and confounding) by cigaretter smokers. People who smoke tend to be leaner than those who don’t. And, of course, smoking IS associated with higher death rates and thus likely explains some of the relationship.
More recent studies have indicated that the above U-shaped relationship may not, in fact, be valid, or at least may not be valid in all populations. For one, the use of statins has now placed more healthy people than in the past in the ‘very low cholesterol’ range. Such individuals would not be expected to have high rates of death. In addition, the prevalence of vegetarianism has increased in some populations, as has the adherence to a generally lower cholesterol-inducing diet. The net effect in the population is, again, to shift the average cholesterol level and place some totally healthy people into the “very low cholesterol” group. And, once more, their death rates would not be expected to be higher than those with higher cholesterol levels.
Despite the new data, the older observations are the ones that still inform insurance companies in terms of actuarial survival. If the newer data is valid, the insurance companies WILL catch up but it will take decades.
(I should also point out that the older studies, in particular, often did not distinguish among the various types of cholesterol which, as you know, have opposite effects on mortality. This sort of omission obviously adds to the potential for the older studies to be misleading.)