(such as a mental status exam) or very few cognitive tests. Sampling of a broad range of cognitive functions is critical to understanding hypertension-cognition relations.
Investigations of hypertension and cognition typically control for numerous confounding variables by statistical adjustment (covariance), matching procedures, and/or study exclusions. Control variables often include age, education, alcohol consumption, anxiety, and depression, and they sometimes include smoking status, occupational status, race/ethnicity, socioeconomic status, and (if relevant) antihypertensive medications. Particularly in case-control studies, hypertensives are commonly either unmedicated or are removed from antihypertensive medication prior to the study. Individuals with medical, neurological, or psychiatric comorbidities are generally excluded from case-control studies. However, because of the resultant exclusion of hypertensives with major end-organ damage, the impact of hypertension on cognition may be underestimated, particularly among older adults.
Longitudinal studies of hypertension and cognition do not always control for comorbidities such as diabetes mellitus and coronary heart disease. This is an important consideration, because hypertension is highly prevalent among individuals having certain medical or psychiatric comorbidities (e.g, depression, diabetes mellitus). Furthermore, hypertension may bear relatively stronger or weaker relations to cognition in the presence of more severe cardiovascular or metabolic diseases. For example, Elias et al. (1997) have found synergistic effects of hypertension and noninsulin-dependent diabetes mellitus with respect to diminished cognitive function. However, Phillips and Mate-Kole (1997) did not find hypertension to be a predictor of cognitive performance in patients with peripheral vascular disease. In this group of patients, more potent manifestations of cardiovascular disease may have overshadowed any effects of hypertension.
Longitudinal investigations also have to contend with problems related to study attrition. It is often the least healthy or least motivated individuals who drop out of ongoing studies. Several available statistical methods for analyzing longitudinal datasets, such as two-stage growth curve analysis and survival analysis, will take such attrition into consideration (Collins and Horn, 1991; Dwyer and Feinleib, 1992; McCardle et al., 1991)
Although hypertensives generally should not be characterized as clinically impaired on cognitive tests (Elias et al., 1987), the impact of hypertension on cognition can be considered clinically significant at an individual level and significant at the population level. In this regard, although a full range of effect sizes is apparent (from d < 0.1 to d > 1.0), numerous case-control studies have found that hypertensive-normotensive differences in cognitive