Citation

Golomb BA. Ann. Intern Med. 1998; 128(6): 478-487.

Affiliation

University of California, Los Angeles, USA.

Copyright

(Copyright © 1998, American College of Physicians)

DOI

unavailable

PMID

9499332

Abstract

PURPOSE: To determine whether the seeming relation between low or lowered cholesterol levels and violence is consistent with causality according to Hill's criteria and whether construct validity is supported by convergence of findings across different types of studies. DATA SOURCES: Search of the MEDLINE database for English-language articles published between 1965 and 1995 was supplemented by searches of the PsycINFO and Current Contents databases and bibliographies of relevant articles. STUDY SELECTION: Peer-reviewed observational and experimental articles and meta-analyses that presented original research; related cholesterol levels to behaviorally defined violence; and, if experimental, had single-factor (lipid-only) intervention. DATA EXTRACTION: Studies were grouped according to type. Data on the relation of violence to cholesterol levels from each study were recorded. DATA SYNTHESIS: Observational studies (including cohort, case-control, and cross-sectional studies) consistently showed increased violent death and violent behaviors in persons with low cholesterol levels. Some meta-analyses of randomized trials found excess violent deaths in men without heart disease who were randomly assigned to receive cholesterol-lowering therapy. Experimental studies showed increased violent behaviors in monkeys assigned to low-cholesterol diets. Human and animal research indicates that low or lowered cholesterol levels may reduce central serotonin activity, which in turn is causally linked to violent behaviors. Many trials support a significant relation between low or lowered cholesterol levels and violence (P < 0.001). CONCLUSIONS: A significant association between low or lowered cholesterol levels and violence is found across many types of studies. Data on this association conform to Hill's criteria for a causal association. Concerns about increased risk for violent outcomes should figure in risk-benefit analyses for cholesterol screening and treatment.

Language: en