Diabetes Care, Vol 17, Issue 11 1273-1280, Copyright © 1994 by American Diabetes Association
Structural equation modeling of symptoms, awareness and fear of hypoglycemia, and personality in patients with insulin-treated diabetes
DA Hepburn, IJ Deary, KM MacLeod and BM Frier
Department of Diabetes, Royal Infirmary, Edinburgh, Scotland.
OBJECTIVE--To assess the possible influence of personality on self-reported
awareness, symptoms, and fear of hypoglycemia and also to identify the
relationship among these self-reported measures using formal structural
equation modeling. RESEARCH DESIGN AND METHODS--A structured questionnaire,
which included questions about sociodemographic details, awareness of the
onset of hypoglycemia, and a list of symptoms of hypoglycemia, was
completed by 305 consecutive insulin-treated diabetic patients attending
the diabetic clinic at the Royal Infirmary of Edinburgh. They also
completed the Hypoglycemia Fear Survey (HFS), and personality was assessed
using the short form of the shortened Eysenck Personality
Questionnaire-Revised (EPQ-R). Formal structural equation modeling was
performed using the following variables: awareness, autonomic symptoms,
neuroglycopenic symptoms, severe hypoglycemic episodes in the last year,
worry and behavior (from the HFS), and extroversion and neuroticism (from
the short EPQ-R). This allowed a model to be constructed that expressed the
putative causal associations among the variables that could be tested
statistically. RESULTS--Of the 302 patients who had experienced
hypoglycemia, 111 (37%) reported reduced awareness, and these patients
scored higher on the worry subscale of the HFS (reduced awareness: 41 +/-
12 vs. normal awareness: 34 +/- 12, P < 0.001). The patients with
reduced awareness scored higher for neuroticism than did the patients with
normal awareness (reduced awareness: 6.1 +/- 3.4 vs. normal awareness: 4.9
+/- 3.3, P < 0.01) and scored lower for extroversion (reduced awareness:
5.8 +/- 3.7 vs. normal awareness: 7.1 +/- 3.7, P < 0.01). In the
structural equation modeling exercise, neuroticism was a significant
putative determinant of many of the other variables.
CONCLUSIONS--Personality was the major determinant of the variance that
could be accounted for in this study and influenced self-reported symptoms,
awareness, and fear of hypoglycemia. Personality factors may, therefore,
influence self-reports from patients, particularly when soft measures, such
as symptoms, are assessed and even when using validated clinical
questionnaires. This finding stresses the importance of using additional
evidence, such as reports from relatives, to substantiate reports from
patients of loss of hypoglycemia awareness.