Participation in activity and risk for incident delirium

Abstract

Objectives: To examine the mediating role between educational attainment and risk for incidence delirium of activity participation and to examine the contribution of participation in specific activities to the development of delirium.

Design: Prospective cohort study.

Setting: Urban teaching hospital in New Haven, Connecticut.

Participants: Participants were drawn from two prospective cohort studies of 779 newly hospitalized patients aged 70 and older without dementia.

Measurements: The main outcome was delirium, measured using the full Confusion Assessment Method (CAM) algorithm, which consisted of acute onset and fluctuating course, inattention, and disorganized thinking or altered level of consciousness, as rated by trained clinical interviewers.

Results: Bivariable results indicated a significant relationship between education and the development of delirium (odds ratio (OR)=0.92, 95% confidence interval (CI)=0.88-0.97) and between activity and delirium (OR= 0.60, 95% CI=0.46-0.79). In multivariable analysis, activity mediated the relationship between education and risk for delirium. Considering each activity separately, multivariable logistic regression analysis showed that regular exercise significantly lowered the risk for developing delirium (OR=0.76, 95% CI=0.60-0.96).

Conclusion: In older persons without dementia, activity participation before hospitalization is a mediator between education and incidence of delirium. Specifically, it was found that participation in regular exercise was found to be significantly protective against delirium.

Figure 1

This is a depiction of the proposed reserve model through the education-activity-delirium pathway using a path diagram for structural equation modeling . This schematic conceptual representation shows how we simultaneously examine the latent variable of activity, depicted by an oval, and observed variables of education, age, sex, and comorbidity in predicting delirium, depicted by rectangles. The indicators for activity are also observedvariables, depicted by the rectangle labeled “activity involvement.” Paths labeled “f”indicate the factor measurement model of latent activity as measured by observed activity items. Paths labeled “c,” “b,” and “d” are direct relationships of delirium on education, activity, and other background variables modeled, in respective order, using logistic regression. Path “a” is a linear regression of activity on educational attainment (included in Model 5). Mediation of the activity and delirium relationship is determined by comparing the effect of delirium regressed on education (path “c”) with and without path “a,” which represents activity regressed on education.