Dr. Nash regrets that I have "inexplicably ignored the greater tragedy" that the National Cholesterol Education Panel's guidelines are not scrupulously followed. I agree there exist undertreated groups (for example, compliant middle-aged men who are at high risk for cardiovascular disease and have no history of psychiatric disease or alcohol abuse receive overall mortality benefit with statins and should definitely be treated), but this is irrelevant to the issue of whether other selected populations may receive no mortality benefit or accessory risk for violence. Dr. Nash charges that the meta-analyses were not prospective trials; however, meta-analyses never are. His characterization of trials included in these meta-analyses, such as the Helsinki trial and the Lipid Research Clinics Coronary Primary Prevention trial, as not randomized or prospective is erroneous.

Although recent statin trials have not shown increased violence despite marked reductions in cholesterol levels, these findings do not extinguish the issue. These trials exclude persons in whom an effect may most likely be seen: those at increased prior risk for violence (such as those with a history of alcohol abuse or psychiatric illness or those who are noncompliant with treatment). Moreover, cholesterol may serve as a marker for a fatty acid (or some other factor) that more proximally relates to violence and that may be variably affected by different cholesterol-lowering treatments [1].

Dr. Goldstein declares that the true connection is a hunger-aggression link. However, some meta-analyses have shown significantly increased rates of violent deaths in persons assigned to drug treatment despite a diet identical to that received by the placebo group. Moreover, primates show increased violence with a cholesterol-reducing diet despite ad libitum feeding with no weight loss. Hunger may indeed induce curmudgeonliness, perhaps through alterations in serotonin levels [2], but some cholesterol-lowering strategies are linked to low serotonin levels or violence without the intermediate of hunger.

Faustman and colleagues observe that their study, which I included in my paper for the sake of completeness, did not support a relation between cholesterol and serotonin. I agree; however, their design enforced high variability in both cholesterol (single measure) and serotonin (psychiatric convenience sample, with variable past use of serotonergic drugs potentially adding noise to serotonin measures). A study without these design flaws did show a significant association [3] that was in conformance with experimental findings in primates [4, 5].

In response to Dr. Weinberger, I offer my own precis in iambic heptameter:

"The time has come to spur debate, to terminate the silence

Enshrouding low or lowered blood cholesterol and violence,

Can thoughtful new approaches make the true relation clear?

With which designs of study will the findings reappear?

Do risks of harm confine themselves to he (or she) who's treated

Or will some injure others when an argument is heated?

While step II diets help your heart, do they engender thugs-

Or is the risk exclusively with certain kinds of drugs?

Is there some novel membrane need, some crucial fatty acid

That switches the susceptible to rancorous from placid?

And speaking of susceptible, who will, indeed, succumb?

Are those depressed at higher risk? Or those who favor rum?

Can blood tests, questionnaires or news of past proclivity,

Predict the likelihood of drug-spawned impulsivity?

In light of the recurrent issues, yet with us this season,

To ask such questions shouldn't still be deemed an act of treason.

The struthious ignore these matters: Hail the status quo!

But maybe-maybe-careful study is the way to go."